66229 South Asia Development Matters More and Better Jobs in South Asia More and Better Jobs in South Asia South Asia Development Matters More and Better Jobs in South Asia Reema Nayar, Pablo Gottret, Pradeep Mitra, Gordon Betcherman, Yue Man Lee, Indhira Santos, Mahesh Dahal, and Maheshwor Shrestha © 2012 International Bank for Reconstruction and Development / International Development Association or The World Bank 1818 H Street NW Washington DC 20433 Telephone: 202-473-1000 Internet: www.worldbank.org 1 2 3 4 14 13 12 11 This volume is a product of the staff of The World Bank with external contributions. The � ndings, interpretations, and conclusions expressed in this volume do not necessarily reflect the views of The World Bank, its Board of Executive Directors, or the governments they represent. The World Bank does not guarantee the accuracy of the data included in this work. The boundaries, colors, denominations, and other information shown on any map in this work do not imply any judg- ment on the part of The World Bank concerning the legal status of any territory or the endorsement or acceptance of such boundaries. Rights and Permissions The material in this work is subject to copyright. Because The World Bank encourages dissemination of its knowledge, this work may be reproduced, in whole or in part, for noncommercial purposes as long as full attribution to the work is given. For permission to reproduce any part of this work for commercial purposes, please send a request with complete information to the Copyright Clearance Center Inc., 222 Rosewood Drive, Danvers, MA 01923, USA; telephone: 978-750-8400; fax: 978-750-4470; Internet: www.copyright.com. All other queries on rights and licenses, including subsidiary rights, should be addressed to the Of�ce of the Publisher, The World Bank, 1818 H Street NW, Washington, DC 20433, USA; fax: 202-522-2422; e-mail: pubrights@worldbank.org. ISBN (paper): 978-0-8213-8912-6 ISBN (electronic): 978-0-8213-8913-3 DOI: 10.1596/978-0-8213-8912-6 Library of Congress Cataloging-in-Publication Data More and better jobs in South Asia / World Bank. p. cm.— (South asia development matters) Includes bibliographical references. ISBN 978-0-8213-8912-6 — ISBN 978-0-8213-8913-3 (electronic) 1. Labor market—South Asia. 2. Labor policy—South Asia. 3. Employment—South Asia. 4. South Asia—Economic policy. 5. South Asia—Economic conditions. I. World Bank. HD5812.57.A6M67 2011 331.120954—dc23 Cover photo: Ray Witlin, The World Bank Cover design: Bill Pragluski, Critical Stages, LLC Contents Foreword . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xvii Preface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xix Acknowledgments. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xxi Abbreviations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xxiii 1 Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 South Asia’s track record . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 Determinants of job quality and the employment challenge . . . . . . . . . . . . . . . . . . . . . . . . 9 Improving an inconducive business environment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18 Improving workers’ skills . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27 Reforming labor market institutions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32 Creating jobs in confl ict-affected areas . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40 Annex 1A Summary statistics on South Asian countries . . . . . . . . . . . . . . . . . . . . . . . . . 42 Annex 1B De� nition of key labor market terms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43 Annex 1C What is a “better� job, and which jobs are “better�? . . . . . . . . . . . . . . . . . . . 43 Notes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46 2 Growth and Job Quality in South Asia . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49 Economic growth in South Asia . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49 Sources of future growth . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54 The track record on employment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59 The urgency of reform . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68 v vi CONTENTS Annex 2A Methodology for decomposing growth . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69 Annex 2B Sources of average annual growth in output per worker . . . . . . . . . . . . . . . . . 70 Annex 2C Shares of agriculture, industry, and services in employment and GDP . . . . . . 72 Annex 2D Methodology and data sources for labor force projections . . . . . . . . . . . . . . . 72 Annex 2E Poverty rates and the number of working poor in South Asia . . . . . . . . . . . . . 75 Annex 2F Analysis of poverty and unemployment in India . . . . . . . . . . . . . . . . . . . . . . . 79 Notes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82 3 A Pro� le of South Asia at Work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85 Overview of employment and labor force participation in South Asia . . . . . . . . . . . . . . . . 86 The nature of employment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93 Where are the better jobs? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102 Who holds better jobs? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111 Annex 3A De� nitions and criteria used in pro� le of South Asia at work . . . . . . . . . . . 117 Annex 3B Regional employment patterns . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119 Notes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 120 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 122 4 What Is Preventing Firms from Creating More and Better Jobs? . . . . . . . . . . . . . . . . . . . 125 Methodological framework . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 125 Constraints in the urban formal sector . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 127 Constraints in the rural nonfarm and informal sectors. . . . . . . . . . . . . . . . . . . . . . . . . . . 140 Demand-side policy options . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 144 Constraints facing potential � rm entrants: Business entry regulations . . . . . . . . . . . . . . 153 Annex 4A Business environment constraints in high- and low-income states in India . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 155 Annex 4B Tax rates as a constraint to � rms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 155 Annex 4C Constraints facing nonbenchmark � rms . . . . . . . . . . . . . . . . . . . . . . . . . . . . 158 Annex 4D Access to � nance as a constraint to � rms . . . . . . . . . . . . . . . . . . . . . . . . . . . 159 Annex 4E Policy options for increasing access to � nance . . . . . . . . . . . . . . . . . . . . . . . . 162 Notes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 165 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 166 5 Opening the Door to Better Jobs by Improving Education and Skills . . . . . . . . . . . . . . . 171 Education and skills in South Asian labor markets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 172 Education and access to better jobs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 176 The education challenge . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 181 The next 20 years: Can South Asian countries improve the educational attainment of their labor forces? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 187 Addressing disadvantages before school: The role of early childhood development . . . . . 189 Primary and secondary education . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 192 Tertiary education and preemployment training systems . . . . . . . . . . . . . . . . . . . . . . . . . 199 On-the-job training . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 206 Annex 5A Additional tables and �gures on education and skills . . . . . . . . . . . . . . . . . . 210 Annex 5B Projections of the educational attainment of South Asia’s population and labor force . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 221 Notes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 222 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 224 6 The Role of Labor Market Regulations, Institutions, and Programs . . . . . . . . . . . . . . . . 229 Labor market institutions, policies, and programs in the formal sector . . . . . . . . . . . . . . 229 CONTENTS vii Labor market institutions, policies, and programs in the informal sector . . . . . . . . . . . . 246 Annex 6A Additional tables and �gures on labor market regulations and institutions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 262 Notes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 272 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 276 7 Creating Jobs in Confl ict-Affected Areas . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 281 Characteristics and intensity of armed confl ict in South Asia . . . . . . . . . . . . . . . . . . . . . . 282 Constraints to job creation in conflict-affected areas . . . . . . . . . . . . . . . . . . . . . . . . . . . . 284 Armed confl ict and labor markets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 285 Facilitating private sector job creation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 291 Education service delivery in confl ict situations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 299 Labor market policies and programs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 301 A jobs transition path in confl ict zones . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 308 Annex 7A De� nitions of high-confl ict and low-confl ict regions in selected South Asian countries . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 311 Annex 7B Labor market characteristics and educational attainment in high-confl ict and low-confl ict areas of selected South Asian countries . . . . . . . . . . . . . . 312 Notes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 316 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 318 Appendixes A Household surveys used . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 323 B Methodology used to analyze labor transitions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 325 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 326 Boxes 1.1 Increasing productivity in agriculture . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15 1.2 Options for reforming the power sector in South Asia . . . . . . . . . . . . . . . . . . . . . . . . . 24 1.3 The critical role of nutrition in early childhood development . . . . . . . . . . . . . . . . . . . 31 2.1 International migration in Nepal and its effects on poverty . . . . . . . . . . . . . . . . . . . . . 63 2D.1 Trends in female labor force participation in southeast and East Asian comparator countries . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74 3.1 Child labor in South Asia . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89 3.2 Composition of the labor force by employment status . . . . . . . . . . . . . . . . . . . . . . . . . 98 3.3 Determinants of informality . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101 3.4 Trends in India’s formal manufacturing sector, 1998–2007 . . . . . . . . . . . . . . . . . . . 109 4.1 Electricity challenges facing South Asia . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 131 4.2 Corruption in South Asia . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 135 4.3 Bringing light to rural consumers in Gujarat, India . . . . . . . . . . . . . . . . . . . . . . . . . . 147 4.4 Improving performance of state-owned power suppliers in Andhra Pradesh . . . . . . 148 4.5 Cutting red tape to reduce corruption . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 149 4.6 Legislative and administrative interventions to reduce corruption in East Asia . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 150 4.7 Public-private collaboration to implement reforms in the Cambodian garment industry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 152 4.8 Effects of easing business entry regulations on � rm entry, employment, and formalization. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 153 4E.1 Successful approaches to small and medium-size business banking . . . . . . . . . . . . . . 164 4E.2 Scaling up micro� nance institutions: The case of BRAC Bank . . . . . . . . . . . . . . . . . 164 viii CONTENTS 5.1 Recruiting teachers based on merit in Sindh, Pakistan . . . . . . . . . . . . . . . . . . . . . . . . 195 5.2 Teacher incentives schemes in Andhra Pradesh . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 196 5.3 The Reaching Out-of-School Children project in Bangladesh . . . . . . . . . . . . . . . . . . 198 5.4 Vocational education provided in the public school system . . . . . . . . . . . . . . . . . . . . 200 5.5 Learning from Australia’s systemic reforms of training and tertiary education . . . . . 202 5.6 Providing scholarships for tertiary education in Nepal . . . . . . . . . . . . . . . . . . . . . . . 206 5.7 Industry-government cooperation: The Penang Skills Development Centre . . . . . . . . 208 6.1 Severance reforms in Austria and Chile . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 244 6.2 Unemployment bene�t proposals for Sri Lanka . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 245 6.3 Nongovernment players in South Asia’s informal labor market . . . . . . . . . . . . . . . . . 247 6.4 Key features of India’s Mahatma Gandhi National Rural Employment Guarantee Act . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 249 6.5 Mexico’s proactive approach to supporting small and medium-size enterprises . . . . 255 6.6 Training informal workers: Kenya’s Jua Kali experience . . . . . . . . . . . . . . . . . . . . . . 256 6.7 Business training and � nancial support for self-employed women in Sri Lanka . . . . 259 7.1 Private sector solutions to the security constraint: Lessons from Afghanistan . . . . . 294 7.2 Community-led infrastructure provision: Afghanistan’s National Solidarity Program . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 296 7.3 Improving the land rights framework in India . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 298 7.4 Improving the regulatory framework in a postconflict situation: Lessons from Liberia . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 299 7.5 Improving schooling despite armed conflict through community schools in Nepal . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 300 7.6 Lessons from efforts to reintegrate Ugandan youth . . . . . . . . . . . . . . . . . . . . . . . . . . 303 7.7 Training and employing displaced people: The case of Asocolflores in Colombia . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 304 7.8 Implementing public works in a postconfl ict environment: Sri Lanka’s Northern Province Emergency Recovery Project . . . . . . . . . . . . . . . . . . . . . . . . . . . . 305 Figures 1.1 Annual growth in working-age population, employment, and labor force in selected South Asian countries . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 1.2 Distribution of employment by type in South Asia, by country . . . . . . . . . . . . . . . . . . . 4 1.3 Average annual increases in mean real wages in selected countries in South Asia . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5 1.4 Percentage of workers in households below the poverty line in selected South Asian countries, by employment status . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 1.5 Average number of months without work in the past year, casual laborers in India, by sector, 1999/2000–2009/10 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 1.6 Distribution of rural and urban workers in selected South Asian countries, by employment type . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7 1.7 Conditional probability of moving into and out of better jobs in rural India, by education and gender, 2004/05–2007/08 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8 1.8 Annual growth in GDP per capita, by region, 1960s–2000s . . . . . . . . . . . . . . . . . . . . . 9 1.9 Annual growth in GDP per capita in South Asia, by country, 1960s–2000s . . . . . . . . . 9 1.10 Sources of annual growth in labor productivity, by region, 1960–80 and 1980–2008. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10 1.11 Sources of annual growth in labor productivity in selected countries in South Asia, by country, 1960–80 and 1980–2008. . . . . . . . . . . . . . . . . . . . . . . . . . . . 11 CONTENTS ix 1.12 Sources of annual growth in total factor productivity in China, India, Pakistan, and Thailand, by sector and reallocation effects . . . . . . . . . . . . . . . . . . . . . 13 1.13 Median wage and value added per manufacturing worker in India, by � rm size and type, 2005 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13 1.14 Share of manufacturing employment in India, by � rm size and type, 1994–2005 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14 1.15 Ratio of working-age to nonworking-age population in South Asia, by country, 1960–2008. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16 1.16 Severity of constraints reported by South Asian benchmark � rm in the urban formal sector . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20 1.17 Cross-country comparisons of reported severity of electricity constraint and power outages for a benchmark � rm . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21 1.18 Percentage of � rms expected to give gifts to public of�cials, by type of interaction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22 1.19 Severity of constraints identi�ed by South Asian benchmark (nonexpanding) and expanding � rm in the urban formal sector . . . . . . . . . . . . . . . . . 22 1.20 Severity of constraints reported by micro benchmark � rm in urban and rural sectors of Bangladesh, Pakistan, and Sri Lanka . . . . . . . . . . . . . . . . . . . . . . . . . 26 1.21 Severity of constraints reported by micro benchmark � rm in India’s urban formal and informal sectors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26 1.22 Employers’ perceptions of skills of recently graduated engineers in India . . . . . . . . . . 28 1.23 Wage premiums in selected South Asian countries, by level of education . . . . . . . . . . 29 1.24 Share of South Asian labor force with no education, with international comparisons . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30 1.3.1 Percentage of children under �ve with malnutrition, by region and country . . . . . . . . 31 1.25 Employment protection indicators in selected countries . . . . . . . . . . . . . . . . . . . . . . . 33 1.26 Weeks of wages required to be paid in severance in regions, country income groups, and selected South Asian countries, by length of service . . . . . . . . . . . . . . . . . 34 1.27 Cross-country comparison of reported severity of the labor regulation constraint . . . 35 1.28 Proportion of country-years in armed confl ict, by region, 2000–08 . . . . . . . . . . . . . . 37 1.29 Severity of business environment constraints (average) reported by � rms in low-confl ict and high-confl ict areas of Afghanistan, 2008 . . . . . . . . . . . . . . . . . . . . . 38 1.30 Unemployment rates in the Northern and Eastern provinces of Sri Lanka, 1997–2001 and 2002–04 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39 1C.1 Percentage of workers in households below the poverty line in Bangladesh, India, and Nepal, by employment status . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44 1C.2 Ratio of rural nonfarm and urban wages to agricultural wages in Bangladesh, India, and Nepal . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45 2.1 Annual growth in GDP per capita, by region, 1960s–2000s . . . . . . . . . . . . . . . . . . . . 50 2.2 Annual growth in GDP per capita in South Asia, by country, 1960s–2000s . . . . . . . . 50 2.3 Sources of annual growth in labor productivity, by region, 1960–80 and 1980–2008. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51 2.4 Sources of annual growth in labor productivity in selected countries in South Asia, by country, 1960–80 and 1980–2008. . . . . . . . . . . . . . . . . . . . . . . . . . . . 52 2.5 Ratio of working-age to nonworking-age population in South Asia, by country, 1960–2008. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53 2.6 Sectoral shares of GDP and employment in selected countries in South Asia, 1980s–2008 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55 2.7 Shares of agriculture and services in employment and GDP in South Asian and comparator countries, 2008 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56 x CONTENTS 2.8 Sources of annual growth in total factor productivity in India and Pakistan, by sector and reallocation effects, 1980–2008 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57 2.9 Sources of annual growth in total factor productivity in China, India, Pakistan, and Thailand, by sector and reallocation effects . . . . . . . . . . . . . . . . . . . . . 58 2.10 Annual growth in working-age population, employment, and labor force in selected South Asian countries . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59 2.11 Average annual increases in mean real wages in selected countries in South Asia . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60 2.12 Percentage of workers in households below the poverty line in selected South Asian countries, by employment status . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61 2.13 Percentage of workers in households below the poverty line in India, by employment status and gender . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62 2.14 Average number of months without work in the past year, casual laborers in India, by sector, 1999–2010 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63 2.15 Distribution of per capita household expenditure in India and Nepal, by employment status . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64 2.16 Distribution of rural and urban workers in selected South Asian countries, by employment type . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65 2.17 Distribution of rural nonfarm workers in India, by employment type, 1983–2009/10 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66 2.18 Labor transitions in rural areas . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67 2.19 Probability of moving into or out of better jobs in rural Bangladesh, India, and Nepal . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67 3.1 Total employment in South Asia, by country, 2010 . . . . . . . . . . . . . . . . . . . . . . . . . . . 86 3.2 Employment rates in lower- and lower-middle-income countries . . . . . . . . . . . . . . . . . 87 3.3 Male and female employment rates in South Asia, by country . . . . . . . . . . . . . . . . . . 87 3.4 Trends in employment rates in South Asia, by country . . . . . . . . . . . . . . . . . . . . . . . . 88 3.1.1 Percentage of child workers attending school in South Asia, by age group and country . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90 3.5 Female labor force participation rates in South Asia, by age group and country . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91 3.6 Annual percentage increases in number of employed workers in South Asia, by sector and country. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94 3.7 Distribution of employment in South Asia, by sector and country . . . . . . . . . . . . . . . . 95 3.8 Percentage of rural workers in the nonfarm sector in South Asia, by country . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96 3.9 Percentage of rural workers in the nonfarm sector in China and India, 1983–2008 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97 3.10 Rural nonfarm sector employment in South Asia, by economic activity and country . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97 3.2.1 Distribution of per capita household expenditure in India and Nepal, by employment status . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98 3.11 Percentage of employment in South Asia classi�ed as informal, by country. . . . . . . . 100 3.12 Percentage of labor force not covered by pension schemes, by region . . . . . . . . . . . . 101 3.13 Percentage of workers in households below the poverty line in Bangladesh, India, and Nepal, by employment status . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103 3.14 Ratio of median rural nonfarm and urban wages to agricultural wages in selected South Asian countries . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103 3.15 Ratio of median industry and service sector wages to agricultural wages in selected South Asian countries . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104 CONTENTS xi 3.16 Average wage, value added, and capital per manufacturing worker in India, by � rm size and type, 2005 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106 3.17 Share of manufacturing employment in India, by � rm size and type, 1994–2005 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107 3.18 Share of manufacturing employment by � rm size in India and selected East Asian economies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 108 3.4.1 Employment in India’s formal manufacturing sector, by � rm size, type, and location, 1998–2007 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109 3.19 Average wage, value added, and capital per service sector worker in India, by � rm size and type, 2006 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110 3.20 Share of service sector employment in India, by � rm size and type, 2001 and 2006 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110 3.21 Percentage of workers with some education and percentage of workers with secondary education or above in South Asia, by employment type and country . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112 3.22 Percentage of rural workers in the rural nonfarm sector in South Asia, by gender and country . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113 3.23 Percentage of rural workers in the rural nonfarm sector in India, Nepal, and Pakistan, by gender and age cohort, 1999–2009/10 . . . . . . . . . . . . . . . . . . . . . . 114 3.24 Decomposition of wage gap between male and female workers in South Asia, by country . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115 3.25 Decomposition of wage gaps between nonethnic minority and ethnic minority workers in India, Nepal, and Sri Lanka . . . . . . . . . . . . . . . . . . . . . . . . . . . 116 3B.1 Regional variations in employment rate in South Asia, by country . . . . . . . . . . . . . . 120 4.1 Severity of constraints reported by South Asian benchmark � rm in the urban formal sector . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 128 4.2 Cross-country comparisons of reported severity of the electricity constraint . . . . . . 129 4.3 Cross-country comparisons of power outages . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 130 4.4 Cross-country comparison of reported severity of corruption constraint . . . . . . . . . 134 4.5 Cross-country comparison of bribe payments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 134 4.2.1 Percentage of � rms expected to give gifts to public of�cials, by type of interaction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 135 4.6 Cross-country comparison of reported severity of political instability constraint . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 136 4.7 Cross-country comparison of reported severity of tax administration constraint . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 137 4.8 Severity of constraints reported by South Asian benchmark (nonexpanding) and expanding � rm in the urban formal sector . . . . . . . . . . . . . . . . 138 4.9 Severity of constraints reported by South Asian benchmark (manufacturing) and service � rm in the urban formal sector . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 140 4.10 Severity of constraints reported by micro benchmark � rm in urban and rural sectors of Bangladesh, Pakistan, and Sri Lanka . . . . . . . . . . . . . . . . . . . . . . . . 141 4.11 Severity of constraints reported by micro benchmark � rm in India’s urban formal and informal sectors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 143 4A.1 Severity of constraints reported by benchmark � rm in urban formal sector in high- and low-income states in India . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 155 4B.1 Cross-country comparison of reported severity of tax rate constraint . . . . . . . . . . . . 156 4B.2 Tax revenue as a percentage of GDP in South Asia . . . . . . . . . . . . . . . . . . . . . . . . . . 157 4B.3 Highest marginal corporate tax rate in South Asian countries and selected comparator countries . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 157 xii CONTENTS 4C.1 Severity of constraints reported by benchmark � rm and � rm with 60 employees in the urban formal sector in Nepal and Sri Lanka . . . . . . . . . . . . . . . 158 4C.2 Severity of constraints reported by benchmark (nonexporting) and exporting � rm in the urban formal sector in South Asia and Bangladesh . . . . . . . . . 159 4D.1 Cross-country comparison of reported severity of access to � nance constraint . . . . . 160 4D.2 Percentage of � rms with credit line or loan from � nancial institution, by � rm size and region . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 161 5.1 Employers’ perceptions of skills of recently graduated engineers in India . . . . . . . . . 173 5.2 Evolution of skills content of urban wage workers in India, 1994–2010 . . . . . . . . . . 174 5.3 Wage premiums over elementary occupations in India, 1994–2010. . . . . . . . . . . . . . 175 5.4 Wage premiums in selected South Asian countries, by level of education . . . . . . . . . 175 5.5 Share of South Asian labor force with no education, with international comparisons . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 177 5.6 Predicted probability of working in rural nonfarm and urban regular wage jobs in selected South Asian countries, by level of education and gender . . . . . 178 5.7 Conditional probability of moving into and out of better jobs in rural India, by education and gender, 2005–08 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 180 5.8 Conditional probability of moving into and out of better jobs in urban India, by education and gender, 2005–08 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 181 5.9 Share of young cohorts with completed primary, secondary, and tertiary education in South Asia, by country . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 183 5.10 Mean years of education of 15–34 year olds in South Asia, by gender and country . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 184 5.11 Reading and arithmetic achievement in rural India and Pakistan, by class, 2010 . . . 186 5.12 Public expenditure on education as a share of GDP, in South Asia and other regions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 187 5.13 Educational attainment of the South Asian labor force in 2010 (estimated) and 2030 (projected), by country . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 188 5.14 Projected shares of prime-age (35–49) and oldest (50–64) labor force participants in South Asia with no education or only primary attainment in 2030, by country . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 189 5.15 Percentage of children under �ve with malnutrition, by region and country . . . . . . . 190 5.16 Share of primary and secondary enrollments in public and private institutions in selected South Asian countries . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 193 5.17 Percentage of graduates from public and private industrial training institutions employed in the organized sector in three states of India, 2002/03 . . . . . . . . . . . . . . 201 5.18 Share of preemployment training and tertiary education enrollment in public and private institutions in South Asia, by country. . . . . . . . . . . . . . . . . . . . . . 205 5.19 Percentage of � rms providing on-the-job training in regions and selected South Asian countries, by � rm size . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 207 5A.1 Educational attainment in the labor force in South Asia, by country. . . . . . . . . . . . . 212 5A.2 Share of labor force with at least primary, upper-secondary, and tertiary education, in South Asia and international comparators . . . . . . . . . . . . . . . . . . . . . . 213 5A.3 Gross and net enrollment rates in primary and secondary education, by region and country . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 215 5A.4 Enrollment in vocational education and training as a share of secondary enrollment, by region and country . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 216 5A.5 Gross enrollment rate in tertiary education, by region and country . . . . . . . . . . . . . 216 5A.6 Share of young cohorts with completed primary and lower-secondary education in South Asia, by gender and country . . . . . . . . . . . . . . . . . . . . . . . . . . . . 217 CONTENTS xiii 5A.7 Share of young cohorts with completed upper-secondary and tertiary education in South Asia, by gender and country . . . . . . . . . . . . . . . . . . . . . . . . . . . . 218 5A.8 Mean years of education of 15–34 year olds in selected South Asian countries, by caste/ethnicity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 219 5A.9 Share of 20–28 year olds in Afghanistan and Nepal with different years of education completed in Afghanistan (2008) and Nepal (2004), by gender and socioeconomic status . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 220 5A.10 Percentage of children under age 5 with stunting in selected South Asian countries, by socioeconomic status . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 220 6.1 Percentage of workforce not covered by formal pension scheme . . . . . . . . . . . . . . . . 231 6.2 Percentage of wage employees in India and Sri Lanka covered by social security, by type of worker . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 231 6.3 The continuum of employment protection legislation . . . . . . . . . . . . . . . . . . . . . . . . 234 6.4 Weeks of wages required to be paid in severance in regions, country income groups, and selected South Asian countries, by length of service. . . . . . . . . . 236 6.5 Employment protection indicators in selected countries . . . . . . . . . . . . . . . . . . . . . . 237 6.6 Job turnover rates and labor regulations in Indian states . . . . . . . . . . . . . . . . . . . . . . 238 6.7 Cross-country comparison of reported severity of the labor regulation constraint . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 239 6.8 Minimum wages as a proportion of median formal sector wages in Nepal, Pakistan, and Sri Lanka . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 240 6.9 Participation of women and members of scheduled castes and tribes in the Mahatma Gandhi National Rural Employment Guarantee Program, 2006/07–2010/11 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 251 6A.1 Trade union membership in India, 1987–2006 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 262 6A.2 Trade union membership in Pakistan, 1999–2008 . . . . . . . . . . . . . . . . . . . . . . . . . . . 262 6A.3 Trade union membership in Sri Lanka, 1987–2006 . . . . . . . . . . . . . . . . . . . . . . . . . . 263 6A.4 Percentage of unionized workers in India, by employment status and sector, 1994 and 2000 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 264 6A.5 Job creation and destruction flows in Sri Lanka and selected groups of countries . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 265 6A.6 Tax wedges in South Asian and international comparator countries . . . . . . . . . . . . . 266 6A.7 Ratio of minimum wage to median casual and formal sector wage, by states in India, 2008. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 267 7.1 Proportion of country-years in armed confl ict, by region, 2000–08 . . . . . . . . . . . . . 282 7.2 Top 15 countries in number of deaths from armed confl ict, 2008 . . . . . . . . . . . . . . 283 7.3 Effects of confl ict on demand for and supply of labor . . . . . . . . . . . . . . . . . . . . . . . . 284 7.4 Annual growth in GDP and number of battle deaths in India and Nepal, 2000–08 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 286 7.5 Unemployment rates in the Northern and Eastern provinces of Sri Lanka, 1997–2001 and 2002–04 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 286 7.6 Percentage of working-age population employed in high-confl ict and low-confl ict areas of India and Nepal . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 287 7.7 Percentage of working-age population employed in high-confl ict areas of Nepal, by gender, 1996–2004 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 287 7.8 Urban workers as share of all workers in India and Nepal. . . . . . . . . . . . . . . . . . . . . 289 7.9 Percentage of workforce employed in unpaid family labor in India, 2000–08 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 290 7.10 Percentage of working-age population with completed lower-secondary education in low- and high-confl ict areas of selected South Asian countries . . . . . . . 291 xiv CONTENTS 7.11 Severity of business environment constraints (average) reported by � rms in low-confl ict and high-confl ict areas of Afghanistan, 2008 . . . . . . . . . . . . . . . . . . . . 292 7.12 Sri Lanka infrastructure accessibility index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 295 7.5.1 Primary school enrollment rate in Nepal among children 6–10 years old, 1996–2004. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 300 7.13 Central government debt as a percentage of GDP in confl ict and nonconfl ict areas of South Asia and the world, 1990s and 2000s . . . . . . . . . . . . . . . . . . . . . . . . 307 Tables 1.1.1 Returns to agricultural growth from investments in public goods and subsidies in India, 1960s–90s. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15 1.1 Top �ve constraints reported by South Asian benchmark � rm in the urban formal sector, by country . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19 1.2 Top �ve constraints reported by South Asian benchmark (nonexpanding) and expanding � rm in the urban formal sector, by country . . . . . . . . . . . . . . . . . . . . . 23 1.3 Top �ve constraints reported by micro benchmark � rm in the urban and rural sectors of Bangladesh, Pakistan, and Sri Lanka . . . . . . . . . . . . . . . . . . . . . . . . . 25 1.4 Top �ve constraints reported by micro benchmark � rm in India’s urban formal and informal sectors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26 1A.1 Summary economic statistics of South Asian countries . . . . . . . . . . . . . . . . . . . . . . . . 42 1B.1 De� nitions of key labor market terms used in this book . . . . . . . . . . . . . . . . . . . . . . . 43 2.1 Labor productivity in South Asia and East Asia, by sector, 2008 . . . . . . . . . . . . . . . . 58 2.2 Decomposition of decline in worker poverty rates . . . . . . . . . . . . . . . . . . . . . . . . . . . 66 2.3 Correlations of country growth rates of per capita GDP across decades . . . . . . . . . . . 69 2B.1 Sources of average annual growth in output per worker, by region, 1960–2008. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70 2B.2 Sources of average annual growth in output per worker in South Asia, by country, 1960–2008. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71 2C.1 Regressions of shares of agriculture, industry, and services in employment and GDP, 2008 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72 2D.1.1 Female labor force participation in Bangladesh, India, and Pakistan and Asian comparator countries . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74 2E.1 Percentage of workers in households below the poverty line in Bangladesh, by employment type, 2000–10 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75 2E.2 Number of working poor in Bangladesh, by employment type, 2000–10 . . . . . . . . . . 76 2E.3 Percentage of workers in households below the poverty line in India, by employment type, 1983–2004/05 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77 2E.4 Number of working poor in India, by employment type, 1985–2005 . . . . . . . . . . . . . 78 2E.5 Percentage of workers in households below the poverty line in Nepal, by employment type, 1995/96 and 2003/04 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78 2E.6 Number of working poor in Nepal, by employment type, 1995–2005 . . . . . . . . . . . . 79 2F.1 Of�cial and authors’ estimated poverty rates for urban, rural, and all workers in India, 1983–2004/05 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80 3.1.1 Incidence of child labor in South Asia, by age group and country . . . . . . . . . . . . . . . . 89 3.1 Male and female labor force participation, employment, and unemployment rates in South Asia, by country . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90 3.2 Factors associated with participation of women in urban areas . . . . . . . . . . . . . . . . . . 92 3.3 Reasons why urban women in South Asia do not participate in the labor force, by country . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93 CONTENTS xv 3.4 Distribution of employment in South Asian countries, by location and sector . . . . . . . 94 3.5 Distribution of employment in South Asian countries, by type of employment . . . . . . 99 3.6 Distribution of formal and informal manufacturing � rms in India, by location and size, 2005. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105 3.7 Distribution of service � rms in India, by location and size, 2006 . . . . . . . . . . . . . . . 109 3.8 Average years of education in South Asian countries, by sector of employment. . . . . 111 3A.1 De� nition of employment and unemployment used based on national surveys . . . . . 117 3A.2 De� nition of formal and informal workers used based on national surveys . . . . . . . 118 4.1 Top �ve constraints reported by South Asian benchmark � rm in the urban formal sector, by country . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 129 4.1.1 Selected energy indicators in South Asia, by country . . . . . . . . . . . . . . . . . . . . . . . . 132 4.2 Electricity constraints faced by � rms, by developing region . . . . . . . . . . . . . . . . . . . 132 4.3 Top �ve constraints reported by South Asian benchmark (nonexpanding) and expanding � rm in the urban formal sector, by country . . . . . . . . . . . . . . . . . . . . 139 4.4 Top �ve constraints reported by micro benchmark � rm in the urban and rural sectors of Bangladesh, Pakistan, and Sri Lanka . . . . . . . . . . . . . . . . . . . . . . . . 141 4.5 Top �ve constraints reported by micro benchmark � rm in India’s urban formal and informal sectors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 143 4.6 Private investment in electricity in South Asia, by country, 1999–2000 and 2001–10. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 145 4.7 Market reforms in the power sector in selected countries in South Asia . . . . . . . . . . 145 4A.1 Top �ve constraints reported by benchmark � rm in the urban formal sector in high- and low-income states in India . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 155 5.1 Perceived demand for and de�ciency in skills in programming and software engineering in Sri Lanka . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 173 5.2 Routine and nonroutine skills categories . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 174 5.3 Main issues, interventions, and expected outcomes in early childhood development . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 191 5.4 Mechanisms for funding tertiary education in South and East Asia, by economy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 204 5A.1 Share of Indian labor force requiring high concentration of nonroutine cognitive analytical and interpersonal skills, 1994–2010 . . . . . . . . . . . . . . . . . . . . . 210 5A.2 Mean years of education of 15–34 year olds in South Asia, by gender and country . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 210 5A.3 Summary of randomized experiments on teacher incentives . . . . . . . . . . . . . . . . . . . 211 5B.1 Assumptions underlying scenarios used to project educational attainment of population in South Asia . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 221 5B.2 Educational attainment of the South Asian labor force in 2010 (estimated) and 2030 (projected) under various scenarios, by country . . . . . . . . . . . . . . . . . . . . . 222 6.1 Correlation coef�cients among union membership, social security coverage, and employment in � rms with 10 or more employees, India, 2010 . . . . . . . . . . . . . . 232 6.2 Percentage of workers with access to formal protection instruments in India and Sri Lanka, by worker characteristic . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 233 6.3 Year of rati�cation of International Labour Organization core conventions, Declaration on Fundamental Principles, and Rights at Work, by country . . . . . . . . . 233 6.4 Selected hiring and redundancy rules in South Asia, by country . . . . . . . . . . . . . . . . 235 6.5 Stylized characteristics of protecting workers versus protecting jobs . . . . . . . . . . . . . 242 6.6 Selected public works programs in South Asia, by country . . . . . . . . . . . . . . . . . . . . 248 6.7 Participation in the Mahatma Gandhi National Rural Employment Guarantee Program, by consumption quintile, 2010 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 250 xvi CONTENTS 6.8 Skill requirements in the informal sector . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 253 6.9 Coverage of micro� nance in South Asia, by country, 2009/10. . . . . . . . . . . . . . . . . . 256 6.10 Selected programs supporting self-employment and microenterprises in South Asia, by country . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 258 6A.1 Selected aspects of employment protection in India . . . . . . . . . . . . . . . . . . . . . . . . . . 269 6A.2 Job creation and destruction rates in large manufacturing � rms in India, by employment type, 2001–08 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 270 6A.3 Minimum wage policies in South Asia, by country . . . . . . . . . . . . . . . . . . . . . . . . . . 271 7.1 Major internal armed confl icts in South Asia since 1978 . . . . . . . . . . . . . . . . . . . . . . 283 7.2 Armed confl icts in South Asia, by confl ict stage, geographic scope, and number of casualties per thousand people . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 284 7.3 Labor market transitions in postconfl ict zones . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 308 7.4 Policy interventions in the initial postconfl ict stages . . . . . . . . . . . . . . . . . . . . . . . . . 310 7B.1 Labor market characteristics in high-confl ict and low-confl ict areas of Afghanistan, 2008 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 312 7B.2 Labor market characteristics in high-confl ict and low-confl ict areas of India, 2000 and 2008. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 312 7B.3 Labor market characteristics in high-confl ict and low-confl ict areas of Nepal, 1996 and 2004 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 313 7B.4 Labor market characteristics in high-confl ict and low-confl ict areas of Sri Lanka, 2004 and 2008 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 314 7B.5 Educational attainment in high-confl ict and low-confl ict areas of South Asia, by country . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 315 A.1 Labor force and living standards surveys used . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 324 Foreword I t is my great pleasure to introduce the institutions, development practitioners, and inaugural issue of South Asia Develop- the media. This series will serve as a vehicle ment Matters. More and Better Jobs in for voicing an in-depth synthesis of economic South Asia is timely because of its relevance and policy analysis on key development top- to the region’s 490 million young people ics for South Asia. who can make South Asia the region of Next year we will address the theme of the future. Despite having one of the low- inequality, the challenges faced by individ- est female participation rates in the devel- ual countries in South Asia, and the ways in oping world, South Asia will add at least which inequality can be addressed so that one million people to its labor force every countries can grow and develop with equity. month. The significant additions to the I hope that the knowledge gathered in the labor force could be a demographic divi- South Asia Development Matters series will dend or a curse, which is why this report bene�t the entire region and that it promotes is so important: the key to future peace debate and builds consensus of all those who and poverty reduction in South Asia is the care about stimulating development and creation of enough good-quality jobs in the eradicating poverty in South Asia. decades to come. Future generations will I would like to thank the team members thank this one for using this opportunity for their high-quality product. They were to create an environment for progressively able to tackle a very dif�cult issue in a way better jobs, which are the only sustainable that examined technical quality, with great pathway out of poverty. openness to views from different professional The South Asia Development Matters streams. This was a tough topic, and they series will be published annually under the have produced important recommendations supervision of Kalpana Kochhar, Chief Econ- to promote policy debate. omist for the South Asia Region. I hope the series will promote dialogue and debate with Isabel Guerrero all our partners, not only policy makers but Vice President, South Asia Region also civil society organizations, academic The World Bank xvii Preface M ore and Better Jobs in South Asia and unrest. Closer to home, labor is the most launches the series South Asia important, if not the only, asset of the poor. Development Mat ters. W hen South Asia, despite impressive growth selecting the �rst topic in the series, we had no and poverty reduction over the past two doubt that it should focus on jobs. South Asian decades, remains home to more than half a countries will add 1.0 million to 1.2 million billion poor people, large numbers of whom new entrants to the labor force every month have little or no education and suffer from for the next two decades and will contribute poor health. Part of the reason is that South about 40 percent of the total new entrants to Asia has some of the worst nutrition indi- the global working-age (15–64) population. It cators in the world. Research clearly shows is not surprising that we decided to focus our that a person’s cognitive development begins attention on the changes in policies necessary in the early years of life, long before formal to create a larger number and higher quality schooling begins. Nutrition and early child- of jobs. Identification and implementation hood development have a strong positive of these policies are central to South Asia’s relationship with educational achievement employment challenge of absorbing the grow- and signi�cant payoffs for lifetime learning ing number of entrants to the labor force at and labor market productivity. The chal- rising levels of productivity. lenge of generating more productive jobs is Recent global events have helped shine intensi�ed because most of the countries in a brighter light on this issue. According to the region are still in confl ict or have only the International Labour Organization, recently emerged from it and many people as many as 30 million people worldwide face serious problems related to access to lost their jobs as a result of the 2008 crisis. opportunities based on gender, caste, and Youth unemployment is especially high, and socioeconomic status. inequality has increased in many countries More and Better Jobs in South Asia around the world. As recent events during attempts to answer three questions: the “Arab Spring� in the Middle East and North Africa demonstrate, joblessness and • Has South Asia been creating an increas- inequality can trigger political instability ing quantity and quality of jobs? xix xx PREFACE • What are the determinants of the quality South Asia’s recent track record with regard of job creation and what is the employ- to the quantity and quality of job creation. ment challenge going forward? It traces the relationship of such job cre- • What demand- and supply-side bottle- ation mostly to overall economic growth and necks need to be eased to meet South attempts to answer what needs to be done Asia’s employment challenge in the face of to meet South Asia’s employment challenge. intensifying demographic pressure? Chapter 3 discusses the key features of labor markets in South Asia, including where Although details vary by country, overall the better jobs are, who holds them, and there is reason for cautious optimism. Over the implications for the employment chal- the past two decades, the region has created lenge ahead. Chapter 4 reviews the busi- more jobs—at a rate largely comparable to ness environment constraints affecting, in growth in the working-age population—and particular, those � rms that have expanded better jobs, in terms of higher pay for wage employment and discusses policy options for workers, lower poverty for self-employed overcoming the most binding business con- workers, and reduced risk of low and uncer- straints in South Asia. Chapter 5 analyzes the tain income for the most vulnerable group of dimensions of the education and skills chal- workers. Yet, there is no room for compla- lenge in the region and discusses policy pri- cency because the challenges are big. Not only orities for improving the quality and skills of do a larger number of jobs need to be created, graduates of education and training systems. but the jobs also need to be more produc- Chapter 6 reviews the role of labor market tive and make workers less vulnerable. Our policies and institutions in encouraging job study shows that creating a larger number creation and protecting workers in the formal of more productive jobs for a growing labor and informal economy and discusses possible force calls for a multisectoral reform agenda directions for labor market policies, including that includes improving access to electricity options to increase the access of informal sec- for �rms across all sectors in urban and rural tor workers to programs that help them man- settings, dealing decisively with issues of gov- age labor market shocks and improve their ernance and corruption, improving access to future earnings potential. Finally, chapter 7 land and transport links between town and reviews the key constraints to job creation country, improving nutrition in early child- and the policy priorities for creating more hood, equipping workers with skills relevant and better jobs in conflict-affected areas. for the world of work, and reorienting labor market regulations and programs to protect workers rather than jobs. Kalpana Kochhar This book is divided into seven chapters. Chief Economist, South Asia Region Chapter 1 is an overview. Chapter 2 reviews The World Bank Acknowledgments M ore and Better Jobs in South Asia pieces by Shaghil Ahmed, Harold Alderman, is the � rst in a series of fl agship Sudeshna Ghosh Banerjee, Hai Anh Dang, reports conceived and launched Puja Vasudeva Dutta, Madhur Gautam, by Isabel Guerrero, Regional Vice President Rana Hasan, Benjamin Herzberg, Ina of the South Asia Region. It is the product of Hoxha, Kalim Hyder, Karl Jandoc, Maria a collaborative effort by many professionals Jos, Samir KC, Somik Lall, Peter Lanjouw, and institutions from both inside and out- Norman Loayza, David McKenzie, Claudio side the World Bank. Monteneg ro, M ar tin Moreno, R in ku The report was prepared by a team led by Murgai, Denis Nikitin, Sheoli Pargal, Harry Reema Nayar and Pablo Gottret, under the Patrinos, Zhiheng Png, Shumaila Rifaqat, direction of Kalpana Kochhar, Chief Econ- Hiroshi Saeki, Mehnaz Safavian, Vibhor omist of the South Asia Region. The core Saxena, Claudia Ines Vasquez, Jessica Ville- team comprised Pradeep Mitra, Yue Man gas, Tomoko Wada, and Karar Zunaid. The Lee, Indhira Santos, Gordon Betcherman, team also benefited from papers prepared Mahesh Dahal, and Maheshwor Shrestha. by colleagues at various institutions in the Wendy Carlin, Amit Dar, Lakshmi Iyer, Toby region, including Nazneen Ahmed, Rushidan Linden, and Mark Schaffer made signi�cant Rahman, R. Shamsunnahar, and Mohamad contributions to speci�c chapters. Yunus (Bangladesh Institute of Development Many people provided written inputs and Studies); Farzana Munshi (BRAC University, contributions on various issues. Early think Dhaka); Bibek Debroy (Center for Policy pieces and background papers were prepared Research, New Delhi); Koushik Dutta (inde- by T. N. Srinivasan, Barry Bosworth, Peter pendent consultant); Ramani Gunatilaka B. Hazell, Derek Headey, Alejandro Nin (adjunct research fellow, Faculty of Business Pratt, Derek Byerlee, David Robalino, Ash- and Economics, Monash University); Amrita ish Narain, Ernest Sergenti, Pierella Paci, Dutta, Ann George, Dev Nathan, Preet David Margolis, Mario Di Filippo, Tanja Rustagi, Alakh Sharma, and Ravi Srivas- Lohmann, Tenzin Chhoeda, Mark Dutz, tava (Institute of Human Development, New Hong Tan, Stephen O’Connell, Lucia Madri- Delhi); and Nisha Arunatilake, Roshini gal, and Meera Mahadevan. The report also Jayaweera, and Anushaka Wijesinha (Insti- draws on speci�c contributions or analytical tute of Policy Studies, Colombo). xxi xxii ACKNOWLEDGMENTS The team bene�ted from advice and com- Karthik Muralidharan, Somil Nagpal, Has- ments from Eliana Cardoso and Andrew san Naqvi, Claudia Nassif, Naveed Naqvi, Steer (former Chief Economists of the South John Newman, Thomas O’Brien, Robert Asia Region), Martin Rama, Michal Rut- Palacios, Dilip Parajuli, Giovanna Prennushi, kowski, and Marcelo Selowsky. Arup Banerji, Jasmine Rajbhandary, Dhushyanth Raju, Emmanuel Jimenez, and Ana Revenga were Mansoora Rashid, Susan Razzaz, Silvia peer reviewers for the report. The team is Redaelli, Francis Rowe, Deepa Sankar, Tah- grateful for the contributions of participants seen Sayed, Hisanobu Shishido, Venkatesh at various panel discussions and brainstorm- Sundararaman, and T. G. Srinivasan. Many ing sessions. They include Janamitra Devan, individuals from a variety of research, pol- Tamar Manuelyan Atinc, Shanta Devarajan, icy, and academic institutions and interna- Ariel Fiszbein, Ernesto May, John Henry tional development agencies in Bangladesh, Stein, Roberto Zagha, Rachid Benmessaoud, India, Nepal, and Pakistan participated in Ellen Goldstein, Nicholas Kraft, Susan and provided extremely useful insights at Goldmark, Asli Demirgüç-Kunt, Kaushik consultation meetings held in the countries. Basu, Nadeem Haque, Michael Walton, Unfortunately, it is not possible to name them Dilip Mukherji, Siddiqur Osmani, Manish individually here. Sabharwal, Binayak Sen, and the late Suresh The report would have not been possible Tendulkar. The team gratefully acknowl- without the able assistance of Izabela Anna edges comments and assistance from Faizud- Chmielewska, Julie-Anne Graitge, Marjorie din Ahmed, D. H. C. Aturupane, Roshan Kingston, and Elfreda Vincent. The team Darshan Bajracharya, Dan Biller, Andreas also gratefully acknowledges � nancial sup- Blom, John Blomquist, Jose Roberto Calix, port from the Multi-Donor Trust Fund for Eliana Carranza, Anthony Cholst, Maria Trade and Development, the Poverty and Correia, Halil Dundar, Simeon Ehui, Social Impact Analysis (PSIA), and the World Ejaz Syed Ghani, Sangeeta Goyal, Mary Bank Research Committee. Aziz Gökdemir, Hallward-Driemeier, Zahid Hussain, Nalin Patricia Katayama, Andrés Meneses, Santi- Jena, Dean Mitchell Jolliffe, Sanjay Kathuria, ago Pombo-Bejarano, and Janice Tuten of the Ayesha Khan, Gladys Lopez-Acevedo, Eric World Bank’s Of�ce of the Publisher coor- David Manes, Nkosinathi Mbuya, Julie dinated the editing, design, production, and McLaughlin, Cem Mete, Hanid Mukhtar, printing of this book. Abbreviations ANTA Australian National Training Authority CDC community development council CPI Competitiveness Partnership Initiative CSSP Community School Support Program DDR disarmament, demobilization, and reintegration EPZ export processing zones FATA Federally Administered Tribal Areas FPD Financial and Private Sector Development GDP gross domestic product GET Global Education Trend GHTDP Great Himalaya Trail Development Programme G–PSF Government–Private Sector Forum GW gigawatts IFC International Finance Corporation ILO International Labour Organization MCTEE Ministerial Council for Tertiary Education and Employment MGNREGA Mahatma Gandhi National Rural Employment Guarantee Act MSME micro and small and medium-size enterprise MW megawatt NATO North Atlantic Treaty Organization NCEUS National Commission for Enterprises in the Unorganised Sector NGO nongovernmental organization NSDC National Skills Development Corporation NSP National Solidarity Program NTFP nontimber forest produce NWFP-KP North West Frontier Province (Khyber Pakhtunkhwa) OECD Organisation for Economic Co-operation and Development PAF Poverty Alleviation Fund PPP public-private partnership PPP purchasing power parity xxiii xxiv ABBRE VIATIONS PRI Panchayati Raj Institutions PSDC Penang Skills Development Centre SEZ special economic zone TEWA Termination of Employment of Workmen Act TFP total factor productivity UCDP/PRIO I Uppsala Conflict Data Program and Centre for the Study of Civil War at the International Peace Research Institute UNESCO United Nations Education, Scienti�c and Cultural Organization UNICEF United Nations Children’s Fund WHO World Health Organization Note: All dollar amounts are U.S. dollars (US$) unless otherwise indicated. CHAPTER 1 Overview Key Messages Message 1: South Asia has created many, years of high-performing East Asian econo- mostly better jobs. mies (excluding China). Going forward, rap- idly growing countries in South Asia need to • Job creation in South Asia averaged almost sustain and slow-growing countries to ignite 800,000 a month between 2000 and 2010. growth by easing constraints to physical and The rate of employment growth broadly human capital accumulation. Higher rates of tracked that of the working-age (15–64) pop- factor accumulation, alongside more typical ulation. Open unemployment is low. rates of TFP growth, which will vary according • Real wages rose for wage workers, and poverty to country circumstances, will allow the region declined for the self-employed as well as all to absorb new entrants to the labor force at types of wage workers. Wages and poverty are rapidly rising levels of labor productivity. the primary criteria for improved job quality • Aggregate TFP growth should also increase that guide the analysis in this book. A reduced through a faster reallocation of labor from risk of low and uncertain income for the most agriculture to industry and services, where TFP vulnerable group of workers is a secondary growth is higher. Reallocation across sectors criterion for improved job quality. It could be needs to be complemented by moving labor monitored only in India, where it is satis�ed. out of lower-productivity �rms in manufac- • The improvement in job quality has been turing and services, where the overwhelming associated with accelerating economic growth majority of South Asians who are employed in Bangladesh and India since the 1980s. In in these sectors work, into higher-productivity Nepal, where growth has been slow for several �rms within those sectors. Reallocation across decades, massive out-migration in response to and within sectors will require physical capital limited opportunities at home has improved accumulation (in electricity, for example, the labor market prospects for those who remain. lack of reliable supply of which is reported Workers’ remittances have reduced poverty by job-creating � rms as an obstacle to their across a wide swath of households. operation). It will also require investment in human capital to provide workers with the skills necessary to access better jobs. Message 2: The region faces an enormous • The “demographic transition�—the period employment challenge, but its demography can during which the number of workers grows work in support of the reforms needed to meet it. more rapidly than the number of dependents— • An estimated 1.0–1.2 million new entrants will can provide a tailwind in support of policy join the labor market every month over the next reform for the next three decades in much of few decades—an increase of 25–50 percent South Asia, as the resources saved from hav- over the average number of entrants between ing fewer dependents provides a “demographic 1990 and 2010. The employment challenge for dividend.� This dividend can be used for high- the region is to absorb these new entrants into priority physical and human capital invest- jobs at rising levels of productivity. ments necessary to absorb the growing number • Aggregate productivity growth in South Asia of entrants into the labor force at rising wages over the last three decades was driven by an and more productive self-employment. The div- extraordinary surge in the growth of total idend can be reaped, however, only if a policy factor productivity (TFP) (a combination of framework is in place that can channel the extra changes in the ef�ciency with which inputs are savings into priority investments (including, for used and changes in technology). Its contribu- example, an ef�ciently intermediating �nancial tion was larger than in the “miracle� growth sector and a business environment conducive to firms’ carrying out those investments). In equate access to land among their leading the absence of such a framework, productiv- constraints. Rural-based industry and ser- ity will grow slowly or remain stagnant, and vice �rms in Bangladesh, Pakistan, and Sri the dividend will go uncashed. The window of Lanka report as a top constraint inadequate demographic opportunity is expected to close transport, which inhibits their access to around 2040 for all countries except Sri Lanka, markets that would make them less depen- where it closed around 2005, and Afghani- dent on local demand. stan, where it will stay open beyond 2040. The • Agriculture will continue to be the largest demographic transition will eventually give employer in much of South Asia for the fore- way to old age dependency, as the share of the seeable future. Boosting TFP growth in the elderly in the population increases. sector through accelerated diversification • Continuance of high economic growth, which into cash crops and high-value activities has been an important driver of improved job will require investment in key public goods. quality, is not assured. Globally, correlations of Investment in agricultural research and country growth rates across decades are low. development has much higher returns than Thus policy reforms required to ease bottle- power, fertilizer, and credit subsidies. necks to improving job quality are needed, • Education reform and action before children irrespective of whether there is a demographic enter school are key. Poor nutrition in early dividend, in order to maintain and increase childhood, where South Asia has the weak- the pace of creation of better jobs, even in est indicators in the world, impairs cognitive lower growth environments. The window of development before children get to school, demographic opportunity lends urgency to the reducing the payoff from subsequent educa- agenda, since policies take time to bear fruit. tional investments. Policy makers must also strengthen the quality of learning at all levels to equip tomorrow’s workers, not only with Message 3: Creating more and better jobs for a academic and technical skills, but also with the growing labor force calls for a reform agenda behavioral, creative thinking, and problem- that cuts across sectors. solving skills employers increasingly demand. • Investing in reliable electricity supply is criti- • Moving away from protecting jobs to pro- cal. South Asian �rms of all types—rural and tecting workers is essential for formal sector urban, formal and informal—rate electricity as job creation in India, Nepal, and Sri Lanka. a top constraint to operations. Reported power Enterprise managers in the urban formal sec- outages are consistent with reported sever- tor report labor regulations as being a more ity: Afghanistan, Bangladesh, and Nepal have severe constraint to the operation of their busi- some of the highest reported outages in the ness than is the case for countries at their levels world. The gap between demand and supply of per capita income. High costs of dismissing of electricity is large. Reforms need to manage regular workers are, in effect, a tax on hiring the required expansion of capacity ef�ciently them. Reforms to encourage job creation in and improve the financial and commercial the formal economy should lower these costs, viability of the power utilities. They involve a which protect a minority of workers. These combination of investment and reform of gov- reforms must go hand in hand with reforms ernance in the sector—both are critical. that strengthen labor market institutions and • Formal urban �rms cite corruption in interac- programs that formal and informal sector tions with the state, especially in transactions workers can use to help them adjust to labor involving tax administration and utilities, as market shocks and improve their future earn- an important constraint to their operations. ings potential. Building incrementally on exist- • Informal urban �rms in India report inad- ing schemes is likely the best way forward. Overview 1 T his book investigates how more and quantity of jobs, the quality of jobs, and labor better jobs can be created in South mobility. Asia.1 It does so for two reasons. First, this region will contribute nearly 40 percent of the growth in the world’s working-age Job quantity (15–64) population over the next several Employment grew in South Asia over the decades. It is important to determine what past decade, broadly tracking growth in needs to be done to absorb them into employ- the working-age (15–64) population (�gure ment at rising levels of labor productivity. 1.1). Lack of safety nets precludes high rates Second, creating more productive jobs—with of open unemployment, which averaged jobs de�ned to include all wage work and self- a little over 3 percent in the region. Thus employment—is the most reliable route out of employment growth tends to broadly mirror poverty for a region that is home to more than growth in the labor force. As the propor- 40 percent of the world’s absolute poor.2 tion of the working-age population that is in The book addresses three major questions. the labor force changes slowly, the growth • Has South Asia been creating an increas- of the labor force tends to track that of the ing number of jobs and better jobs? working-age population. Together these • What determines the quality of job cre- observations imply that employment growth ation, and what is the employment chal- can be expected to broadly reflect that of the lenge going forward? working-age population. • What demand- and supply-side bottle- Among five of the larger countries in necks need to be eased to meet South the region, employment growth since 2000 Asia’s employment challenge in the face of was highest in Pakistan, followed by Nepal intensifying demographic pressure? and Bangladesh, India, and Sri Lanka. Total employment in South Asia (excluding Afghanistan and Bhutan) rose from 473 mil- South Asia’s track record lion in 2000 to 568 million in 2010, creating This section examines South Asia’s track an average of just under 800,000 new jobs record in creating jobs. It looks at the a month. 3 4 MORE AND BETTER JOBS IN SOUTH ASIA FIGURE 1.1 Annual growth in working-age population, employment, and labor force in selected South Asian countries 4 3 percent 2 3.6 3.6 3.3 3.3 3.0 2.5 2.6 2.8 2.3 2.2 2.2 2.3 1 1.0 1.2 1.0 0 Sri Lanka India Bangladesh Nepal Pakistan 2000–10 1985–2010 2000–10 1995–2010 2000–10 working-age population employment labor force Sources: Authors, based on data on working-age population from UN 2010 and data on employment and labor force from national labor force surveys. FIGURE 1.2 Distribution of employment by type in South Asia, by country 100 80 34 36 43 50 62 60 77 75 9 9 80 percent 8 13 40 2 32 57 55 17 0.5 22 1 4 20 2 14 10 14 21 17 21 9 8 0 07 tan 05 sh 20 tan 09 ndia 20 nka 20 ves /0 l 08 tan 07 pa 20 lade 20 Ne 8 6 07 0 08 04 8 9 20 nis u i 20 kis La ald /0 /0 /1 /0 I Bh Pa a i ng M Sr gh 20 Ba Af regular wage or salaried casual wage all wages self-employed (high end) self-employed (low end) Source: Authors, based on data from national labor force and household surveys. Note: The data for Maldives and Sri Lanka do not allow the separation of wage employment into regular wage or salaried workers and casual laborers. In all countries except Maldives and Sri Job quality Lanka, the largest share of the employed are the low-end self-employed (�gure 1.2). 3 South Asia has created better jobs, de� ned Nearly a third of workers in India and a primarily as those with higher wages for �fth of workers in Bangladesh and Pakistan wage workers and lower poverty levels for are casual laborers. Regular wage or sala- the self-employed and secondarily as jobs ried workers represent a �fth or less of total that reduce the risk of low and uncertain employment. income for the most vulnerable group of OVERVIEW 5 workers. By these measures, results have in the average number of months for been positive: which all casual laborers were without work despite looking for it (� gure 1.5). • Real wages for wage workers—both Thus, the secondary criterion for better casual and regular wage or salaried— jobs—that they should reduce the risk of grew 0.1–2.9 percent a year during vari- low and uncertain incomes for the most ous subperiods between 1983 and 2010 vulnerable—has been met in India. This for which comparisons can be made is not necessarily the case in other coun- (�gure 1.3). tries in South Asia. (For a discussion of • A higher proportion of self-employed these criteria and the way in which they workers (on whom information on earn- are used to rank jobs by quality, see ings is not available) are now in house- annex 1C.) holds above the national poverty line in Bangladesh, India, Nepal, Pakistan, and Notwithstanding the variation in wages Sri Lanka. This �gure is used as a proxy and poverty rates across employment types for improving job quality for this segment and their changes over time, there is a of the labor force (� gure 1.4).4 Increas- stable pattern of association between pov- ing proportions of casual and regular erty and the type of employment that has wage or salaried workers in Bangladesh, been maintained over time. Regular wage India, and Nepal and all wage workers or salaried workers have the highest wages in Pakistan and Sri Lanka are also now and lowest poverty rates; the self-employed in households that are above the poverty have higher poverty rates; and casual work- line. Indeed, poverty rates for all types of ers, especially in agriculture, have the low- workers during all time periods show a est wages and highest poverty rates (see decline when the data are disaggregated annex 1C). by location (rural or urban) or gender. The proportion of workers in different Thus, the primary criterion for better jobs employment types has remained largely is satis�ed. unchanged over time (� gure 1.6). At this • In India over the period 1999/2000 level of aggregation, better jobs have been through 2009/10, there was a decline created mainly as a result of increasing FIGURE 1.3 Average annual increases in mean real wages in selected countries in South Asia 4 2.9 2.8 percent 1.9 2.0 2 0.1 0 Bangladesh India Nepal Pakistan Sri Lanka 2002–05 1983–2010 1999–2008 2000–09 2000–08 Source: Authors, based on data from national labor force and household surveys. 6 MORE AND BETTER JOBS IN SOUTH ASIA FIGURE 1.4 Percentage of workers in households below the poverty line in selected South Asian countries, by employment status a. Bangladesh, 2000–10 b. India, 1983–93/94 and 1999/2000–2004/05 c. Nepal, 1995/96–2003/04 80 80 80 70 67 70 70 61 58 60 60 60 47 47 47 48 50 50 50 44 43 42 percent percent percent 38 38 39 40 40 40 33 31 28 30 39 29 38 28 30 30 30 30 22 21 29 29 28 27 20 24 20 26 20 21 18 18 10 10 15 10 12 7 0 0 0 2000 2005 2010 1983 1993/94 1999/2000 2004/05 1995/96 2003/04 URP URP MRP MRP all regular wage or salaried all self-employed all casual labor all workers d. Pakistan, 2001/02 to 2007/08 e. Sri Lanka, 1995/96–2006/07 80 80 70 70 60 60 50 50 percent 43 percent 40 40 30 30 25 26 20 23 20 24 13 15 10 10 11 0 0 2001/02 2007/08 1995/96 2006/07 all self-employed, all wages, and all wages all self-employed all workers all workers Source: Authors, based on data from national labor force and household surveys. Note: URP = uniform recall period (the period in which respondents were asked to recall all consumption items over the same recall period [for example, 7 days]). MRP = mixed recall period (the period need not be the same for all items, [for example, 7 days for some and 365 days for others]). Figures are for workers age 15–64. FIGURE 1.5 Average number of months without work in the past year, casual laborers in India, by sector, 1999/2000–2009/10 2.0 1.8 1.9 1.6 1.5 1.5 1.4 1.4 1.4 1.1 months 1.0 0.9 0.5 0 1999/2000 2004/05 2009/10 agriculture rural nonfarm urban Source: Authors, based on data from Indian labor force and household surveys. Note: Figures are for workers age 15–64 who were available for work during at least part of the month. OVERVIEW 7 FIGURE 1.6 Distribution of rural and urban workers in selected South Asian countries, by employment type a. Bangladesh, 2000–2010 b. India, 1983–2009/10 100 100 18 18 18 15 17 90 22 21 21 90 31 33 31 38 36 39 80 35 38 80 70 70 28 45 41 35 36 60 40 42 41 60 percent percent 50 50 40 49 44 40 61 61 47 60 57 54 30 30 51 20 43 43 20 43 41 41 40 43 10 17 10 14 15 8 7 8 8 8 0 0 00 05 10 00 05 10 83 4 0 5 0 83 4 0 5 0 –9 00 –0 –1 –9 00 –0 –1 20 20 20 20 20 20 19 19 –2 –2 93 04 09 93 04 09 19 99 20 20 19 99 20 20 rural urban 19 19 rural urban casual labor self-employed regular wage or salaried c. Nepal, 1995/96–2003/04 d. Pakistan, 1999/2000–2008/09 e. Sri Lanka, 2000–2008 100 100 100 15 13 12 90 18 17 16 16 19 20 19 31 33 80 80 80 45 44 70 60 60 60 61 46 45 45 percent percent 55 percent 50 70 71 79 83 71 40 40 40 69 67 30 55 56 20 20 20 35 36 36 27 26 10 12 14 13 5 4 0 0 0 6 4 6 4 0 8 9 0 8 9 00 08 00 08 /9 /0 /9 /0 00 –0 –0 00 –0 –0 20 20 20 20 95 03 95 03 –2 –2 07 08 07 08 19 20 19 20 99 20 20 99 20 20 rural urban 19 19 rural urban rural urban self-employed casual labor self-employed regular wage or salaried wage worker Source: Authors, based on data from national labor force and household surveys. a. Data from the Bangladesh Household Income and Expenditure Surveys (HIES) were used to calculate worker poverty rates. The share of workers by employment type in the HIES differs from the share in the Bangladesh labor force surveys. The difference is likely to be partly driven by how female employment is captured, with female participation rates in the HIES less than half those reported in the labor force survey. Therefore, the changes in the share of workers by type in Bangladesh from the HIES should be interpreted with caution. For example, between 2005 and 2010 the significant increase in the share of regular wage or salaried work in urban areas was driven largely by changes in the female urban workforce reported in the HIES 2005 and HIES 2010. b. Although there is variation in the shares of casual labor and self-employment in rural areas in India, there is no persistent increase or decline in the shares throughout the whole period (for example, the increase in casual labor between 2004/05 and 2009/10 mostly reversed the decline between 1999/2000 and 2004/05); the share of regular wage or salaried workers remained constant throughout the 25-year period. quality within jobs rather than reallocation offers better jobs than agriculture. Improve- of the labor force across employment ment in job quality has been associated with categories. increasing shares of industry and services Looking across broad sectors, wages in in employment, which includes a growing industry and services are higher than in agri- share of the rural nonfarm economy in rural culture. Thus, the rural nonfarm economy employment. 8 MORE AND BETTER JOBS IN SOUTH ASIA Labor transitions Rural workers who were in agriculture in the � rst period were more likely to make The broad constancy in the share of work- the transition to a better job—to nonfarm ers across employment types masks labor work—if they had secondary or higher levels mobility at the level of individual workers. of education. This higher mobility is typically Many rural workers in Bangladesh, India, greater for workers who completed upper- and Nepal (the three countries studied in the secondary education. Conversely, workers labor transition analysis in this book) have with less education were more likely to expe- moved from agriculture to the rural non- rience a transition in the opposite direction— farm economy and vice versa. from nonfarm work to agriculture. Workers Education is closely tied to labor mobil- with lower levels of education are more likely ity. Secondary and higher levels of education to lose better jobs than they are to secure increase the ability of workers to move out them, as shown in the higher levels of tran- of agriculture, casual wage jobs, and low- sition bars for lower levels of education in end self-employment to better jobs. Although the right-hand panel compared with the left- analysis was conducted for both rural and hand panel in �gure 1.7. Workers with higher urban Bangladesh, India, and Nepal, in the levels of education are more likely to move to interest of space only the results for rural better jobs than they are to lose them. India are shown (�gure 1.7). FIGURE 1.7 Conditional probability of moving into and out of better jobs in rural India, by education and gender, 2004/05–2007/08 a. Transition by men from b. Transition by men from agriculture to nonfarm jobs nonfarm jobs to agriculture 80 80 60 60 percent percent 40 40 20 20 0 0 tio no im e y da er nd er y tio no im e y nd er nd er y ar ar ar ar pr som pr som on low co pp co low co pp im rti im rti n y ry y n y y y ar ar ar ar ar se u se u te te pr pr a a uc uc c ed ed se se c. Transition by women from d. Transition by women from 80 agriculture to nonfarm jobs nonfarm jobs to agriculture 80 60 60 percent percent 40 40 20 20 0 0 io o im e y nd er nd er y io o im e y nd er nd er y ar ar ar ar pr som pr som n n co low co pp co low co pp im rti im rti n y y y n y y y ar ar ar ar ar ar se u se u te te pr pr at at uc uc ed ed se se Source: Authors, based on data from national labor force and household surveys. Note: The probability of transition is conditional on being in a specific type of employment in the first period (e.g., panel a shows the estimated probability that a rural male worker who was in agriculture in the first period is working in a nonfarm job in the second period). The probability differs by the level of education of the worker. Upper and lower bounds of the estimated probabilities are shown. The blue lines are drawn through the midpoints of the bounds. OVERVIEW 9 Determinants of job quality and second only to that of East Asia (�gure 1.8). the employment challenge But growth experiences have varied within South Asia (figure 1.9). Growth in GDP Improving job quality for most segments per capita accelerated, particularly since of the labor force can usually occur only in the 1980s, in Bangladesh and India. It stag- a growing economy. South Asia has seen nated in Nepal and was marked by volatility an acceleration of growth in gross domes- around a broadly declining trend over the last tic product (GDP) per capita over the three four decades in Pakistan. Sri Lanka witnessed decades since 1980; its growth has been an acceleration of growth over the last �ve FIGURE 1.8 Annual growth in GDP per capita, by region, 1960s–2000s 10 8.4 8 6.7 6.0 5.9 6 4.7 4.5 percent 3.8 4.0 3.7 4 2.8 2.7 3.1 2.8 3.0 2.5 2.3 2.5 2 1.6 1.2 1.4 0.5 0.1 0 –0.2 –0.4 –0.9 –2 1961–70 1971–80 1981–90 1991–2000 2001–10 East Asia and Pacific Latin America and the Caribbean Middle East and North Africa Sub-Saharan Africa South Asia Source: Authors, based on data from World Bank 2011c. FIGURE 1.9 Annual growth in GDP per capita in South Asia, by country, 1960s–2000s 9 8 7.9 7 6.6 6.6 6.3 6.2 6 5 4.8 4.9 4.6 4.5 4.3 percent 4.3 4.2 4.3 4 3.3 3.4 3 2.9 2.8 2.8 2.2 2.4 2.2 2.1 1.9 2 1.8 1.0 1.2 1.2 1.1 1 0.4 0.5 0 0.0 –1 1961–70 1971–80 1981–90 1991–2000 2001–10 Afghanistan Bangladesh Bhutan India Maldives Nepal Pakistan Sri Lanka Source: Authors, based on data from World Bank 2011c. Note: Growth in the earliest available decade for Afghanistan, Bhutan, and Maldives is not based on data for the entire decade because data for the entire decade were not available. Thus Afghanistan 2001–09 is based on 2003–09, Bhutan 1981–90 is based on 1982–90, and Maldives 1991–2000 is based on 1996–2000. 10 MORE AND BETTER JOBS IN SOUTH ASIA decades, except for a dip in the 1980s, and Growth in aggregate labor productivity it managed to avoid the slowdown or stagna- can be decomposed into two factors: tion of the 1970s that affected the other coun- • “Extensive� growth, comprising growth tries in the region. in physical capital per worker (capital deepening) and growth of human capital Sources of growth per worker (education) • “Intensive� growth, comprising growth in The marked acceleration in growth in South total factor productivity (TFP), a measure Asia has allowed better jobs to be created. of the efficiency with which inputs are Among industrial and all developing regions combined to produce output. except China, aggregate labor productivity (GDP per worker) grew fastest in South Asia, This decomposition indicates that growth at 3.7 percent a year, between 1980 and in TFP made a larger contribution to the 2008 (�gure 1.10). This performance repre- growth of aggregate labor productivity in sents a striking turnaround from the preced- South Asia during 1980 –2008 than did ing two decades (1960–80), when aggregate physical and human capital accumula- labor productivity in South Asia grew just tion (see �gure 1.10). In fact, the contribu- 1.6 percent a year—more slowly than any tion of TFP growth was higher than in the other region, including Sub-Saharan Africa. high-performing East Asian economies FIGURE 1.10 Sources of annual growth in labor productivity, by region, 1960–80 and 1980–2008 9 8 7 6 4.8 5 4 percent 1.3 1.0 3 0.4 0.6 1.1 2.0 0.4 1.3 1.4 2 0.5 1.3 1.6 0.3 0.4 0.4 0.8 1.1 0.4 0.5 2.6 2.9 0.3 0.3 0.4 0.4 2.5 0.5 1 0.3 1.8 0.2 1.0 1.3 1.3 1.2 0.5 0.9 0.8 0.9 0.9 0.5 0.9 0.1 0.4 0.5 0 –0.2 –0.7 –0.7 –1 –2 0 8 0 8 0 8 0 8 0 8 0 8 0 8 0 8 –8 00 –8 00 –8 00 –8 00 –8 00 –8 00 –8 00 –8 00 –2 –2 –2 –2 –2 –2 –2 –2 60 60 60 60 60 60 60 60 19 80 19 80 19 80 19 80 19 80 19 80 19 80 19 80 19 19 19 19 19 19 19 19 South Asia world industrial East Asia less China Latin America Middle East Sub-Saharan countries China Africa physical capital per worker education per worker total factor productivity Source: Bosworth 2010. OVERVIEW 11 FIGURE 1.11 Sources of annual growth in labor productivity in selected countries in South Asia, by country, 1960–80 and 1980–2008 5 4 2.6 3 0.5 0.2 1.2 percent 2 1.4 0.7 0.4 0.3 0.1 1.3 0.4 0.3 2.4 0.3 1 1.4 1.6 0.4 0.9 1.0 1.0 0.2 0.3 0.4 0 –0.3 –1 1960–80 1980–2008 1960–80 1980–2008 1960–80 1980–2008 1960–80 1980–2008 Bangladesh India Pakistan Sri Lanka physical capital per worker education per worker total factor productivity Source: Bosworth 2010. excluding China during their “miracle� of workers—estimated at at least a third of growth years.5 all working-age men—that contributed to The sources of growth varied across the reduction in labor supply and led to ris- countries (�gure 1.11). In Bangladesh, edu- ing real wages for those left behind (World cation accounted for a fifth of the growth Bank 2010). Declining poverty, which is in aggregate labor productivity. Growth of used as a proxy for improving job quality, TFP was more important in India, reflect- owes less to growth in Nepal than to the ing its increased exposure to external and inflow of worker remittances, estimated at internal competition brought about by trade nearly a quarter of GDP. liberalization and deregulation. Capital deep- ening played a signi�cant role in India and The employment challenge Pakistan. But whereas its contribution rose in India after 1980, it fell sharply in Pakistan, The pressure to create better jobs will inten- accounting for the relative importance of TFP sify very substantially over the next few growth there. Capital deepening was more decades. In its medium-fertility scenario, the important in Sri Lanka, where the share of United Nations projects that the region’s cur- investment in GDP nearly doubled following rent population of 1.65 billion will increase its “big bang� opening up in 1977. 25 percent by 2030 and 40 percent by 2050. Job quality has not been associated Given the region’s generally youthful popu- with accelerated economic growth every- lation, the working-age population is pro- where in the region. In Nepal, for example, jected to increase even more (35 percent by growth in per capita GDP remained at 2030 and 50 percent by 2050). about 2 percent a year during the last three Two scenarios reveal the job creation decades. It was the massive out-migration implications of these demographic changes. 12 MORE AND BETTER JOBS IN SOUTH ASIA In the first, there is no increase in female and human capital accumulation) and less labor force participation rates from current on the extraordinary growth of TFP seen levels. In this scenario, South Asia adds nearly in the last three decades.7 As the region has 1 million entrants a month to the labor force become more open to the international econ- between 2010 and 2030. The proportion- omy, it is importing better-quality capital ate increases are largest in countries with the and intermediate goods at world prices and youngest populations (Afghanistan, Nepal, using standard technology to produce goods Pakistan) and smallest in the single aging that are either sold domestically or exported country in the region (Sri Lanka). in competitive world markets. Inasmuch as Under the second scenario, female labor the technology is widely used internationally, participation rates increase 10 percentage the increases in TFP arising from it will be points by 2030 in Bangladesh, India, and limited to what is routine in global best prac- Pakistan, which together account for 95 per- tice. For a country such as India, which has cent of the region’s working-age population a large internal market, domestic sales could and have the lowest rates of female participa- lead to temporarily larger increases in TFP as tion (31 percent in Bangladesh, 30 percent in less competitive producers exit the market. India, and 22 percent in Pakistan). (This phe- But, even with acceleration in “second-gener- nomenon would be consistent with observed ation� structural reforms, TFP growth is not behavior in Indonesia, the Republic of Korea, likely to continue at the rates triggered by the Malaysia, and Thailand between 1960 and reforms of the 1990s. Hence, a key task for 2000.) Participation rates remain unchanged policy makers will be to create an improving in the rest of South Asia. Nearly 1.2 million enabling environment for factor accumula- new entrants a month join the labor force tion (physical capital deepening and human between 2010 and 2030, intensifying labor capital formation), which, alongside more market pressure in Bangladesh, India, and routine rates of TFP growth, can deliver ris- Pakistan. These projections imply a huge ing wages and declining poverty. increase over the just under 800,000 entrants Aggregate TFP growth could also be a month that joined the labor force between increased as a result of a faster reallocation of 1990 and 2010. labor out of low-productivity agriculture. The Can high economic growth, which has contribution of reallocation to TFP growth been the major driver of improving job quality has been substantially greater in East Asia in some South Asian countries, be expected than in South Asia (�gure 1.12). Reallocation to continue over the next few decades? The accounted for two-thirds of aggregate TFP historical evidence from around the world growth in Thailand between 1977 and 1996, shows on the contrary that growth rates are a period during which the share of agricul- highly unstable over time: the cross-decade ture in employment fell nearly a third. The rank correlation of growth rates per capita contribution of reallocation to TFP growth in for 94 countries across �ve decades is a mere China between 1978, when reforms started, 0.1–0.4—and correlations with time periods and 1993 was nearly one-third. During this more than two decades apart are typically period, the share of agriculture in employ- negligible.6 The rarity of sustained growth ment fell more than a �fth.8 In contrast, real- is underlined by the fact that since 1950, per location contributed 15 percent to aggregate capita GDP has grown at a rate of 7 percent TFP growth in Pakistan and 20 percent in or more—the rate required to double living India between 1980 and 2008, during which standards every 10 years—in only 13 coun- time the share of agriculture fell just under a tries, 9 of them in East Asia (World Bank �fth in both countries. 2008b). The creation of better jobs also requires Looking forward, productivity growth that labor be moved more rapidly not only in the region will need to rely more on fac- out of agriculture into industry and ser- tor accumulation (physical capital deepening vices but also out of lower-productivity into OVERVIEW 13 FIGURE 1.12 Sources of annual growth in total factor productivity in China, India, Pakistan, and Thailand, by sector and reallocation effects 4 0.2 0.3 3 1.0 0.5 percent 0.6 2 3.1 1.3 0.2 1.4 1 0.4 1.1 0.6 0.3 0.2 0.3 0.1 0.5 0.5 0.2 0.3 0 Pakistan Thailand India China China 1980–2008 1977–96 1980–2008 1978–93 1993–2004 reallocation services industry agriculture Sources: Authors, based on data from Bosworth 2005, 2010; Bosworth and Collins 2008. Note: The contribution of reallocation during a decade is calculated as aggregate TFP growth minus the sum over the three sectors of TFP growth weighted by the share of the sector in GDP at the beginning of the decade. FIGURE 1.13 Median wage and value added per manufacturing worker in India, by firm size and type, 2005 (percentage of median wages/value added per worker in formal firms with 200 or more employees) a. Median wage per worker b. Median value added per worker 100 100 90 90 80 80 70 70 70 66 64 60 60 60 55 53 percent percent 50 50 50 47 40 37 40 37 30 25 30 26 23 19 21 20 18 20 11 10 10 6 0 0 1–4 5–9 10–19 20–49 50–99 100–199 1–4 5–9 10–19 20–49 50–99 100–199 number of workers number of workers formal firms informal firms Sources: Authors, based on data on formal firms from the Annual Survey of Industries and data on informal firms from the National Sample Survey manufacturing surveys. Note: Formal firm with 200 or more employees = 100 percent. higher-productivity �rms within industry and firms employing one to four workers aver- services. Wage differentials between smaller age one-quarter the levels of �rms employing and larger � rms are particularly marked in more than 200 workers (�gure 1.13).9 India’s manufacturing sector, where both Output per worker and wages are also output and wages per worker in formal much lower in informal �rms than in formal 14 MORE AND BETTER JOBS IN SOUTH ASIA �rms within the same size class. In informal fewer than 50 people. As in manufacturing, firms with one to four workers, these mea- the size distribution of � rms did not change sures are just 25–50 percent of those of formal between 2001 and 2006. � rms the same size. The difference probably Notwithstanding its declining share of reflects both the higher capital intensity and employment, agriculture will continue to be the higher skill levels at larger firms versus the largest employer among the three broad smaller ones and at formal �rms versus infor- sectors in most of South Asia for some time. mal �rms of the same size. Firm-size produc- For this reason, it is important that agricul- tivity differentials exist in other countries, but tural productivity be increased to ensure that they are particularly high in India compared the quality of jobs be improved for workers with East Asia. in the sector (box 1.1). Although output per worker and wages at Accelerating the exit from agriculture to larger, formal � rms are higher, more than industry and services and enabling industrial 80 percent of employment in manufactur- and service sector � rms to expand, become ing in India is in micro �rms (�rms with 1–4 more productive, and thus pay higher wages workers) and small � rms (� rms with 5–49 requires urgent action on a number of fronts. workers), a situation that has persisted over The limited educational attainment of the time (�gure 1.14). In fact, half of employment labor force, inadequate infrastructure, and is in own-account manufacturing enterprises low capital intensity of most �rms imply that that do not hire any wage workers. The con- realizing higher TFP growth through the centration of employment in micro and small intersectoral and intrasectoral reallocation of firms is even higher in services, where 96 labor will require substantial investment in percent of workers are in � rms that employ human and physical capital. FIGURE 1.14 Share of manufacturing employment in India, by firm size and type, 1994–2005 100 90 80 70 60 percent 50 40 30 20 10 0 4 9 9 5 9 0– 9 9 0+ 4 9 9 50 9 10 –99 9 0+ 4 9 20 9 50 9 10 –99 9 0+ 1– 5– –1 –4 10 0–9 19 1– 5– –1 –4 19 1– 5– –1 –4 19 20 20 20 0– 0– 10 20 10 20 10 1994 2000 2005 informal directory manufacturing establishments (at least 1 hired worker and more than 6 workers in total) informal nondirectory manufacturing establishments (at least 1 hired worker but fewer than 6 workers in total) informal own-account manufacturing enterprises (no hired workers) formal Sources: Authors, based on data on formal firms from the Annual Survey of Industries and data on informal firms from the National Sample Survey manufacturing surveys. Note: The data show a small share (1 percent or less) of informal employment in the larger firms as well. OVERVIEW 15 BOX 1.1 Increasing productivity in agriculture The exit of workers from agriculture to industry well suited for small-scale production. Improved and services, whether rural or urban based, is an technology, together with a reduction of implicit important correlate of economic development. It is taxation, could boost the yields of such crops. nevertheless critical to ensure that TFP in agricul- Rising incomes, urbanization, and changing con- ture—the key driver of economic growth over the sumer preferences are creating strong demand for long haul— continues to grow. Increasing agricul- high-value commodities in most South Asian coun- tural TFP is important for two reasons. First, it can tries. The shift has increased incentives to diversify, to provide better jobs for workers who remain in the which farmers across the subcontinent are respond- sector. Second, it allows workers to transition more ing. Agricultural diversi�cation has proceeded most rapidly from agriculture to industry and services, rapidly for fruits and vegetables in Bangladesh, where TFP growth is higher. Bhutan, and Nepal; horticulture, � shing, and live- Notwithstanding South Asia’s transformation stock in India; and livestock in Pakistan. These from a food de�cit to a food surplus region, the pro- developments have occurred despite the disincentives ductivity of agriculture remains low. India and Paki- created by policies that favor food security crops (rice stan have improved their agricultural productivity and wheat), such as those in India, Pakistan, and Sri over the years; elsewhere in the region, improvements Lanka. A shift from cereal-based to high-value agri- began only in the 1990s, after decades of relative culture requires substantial farm-level investment, stagnation. There is some room to expand area under as well as greater exposure to risk. It is necessary to cultivation in selected rain-fed parts of the region (in widen access to � nancial and insurance services for Afghanistan, Bhutan, Nepal, and eastern Sri Lanka). many smallholders in order to enable them to partici- There are also some unexploited opportunities for pate in the high-value supply chains. expanding area through watershed development and Core public goods are particularly important in irrigation (in Afghanistan, Bhutan, India, Nepal, and agriculture. Only the public sector can invest in much Pakistan). But the bulk of future growth will have to research and development, because private inves- rely on boosting TFP growth, which has lagged inter- tors are not able to appropriate rents, except in a national best practice. few cases, such as hybrid seeds. Public investment in The key to accelerating TFP growth lies in diver- agriculture has been an important driver of growth sifying into cash crops (tea, sugarcane, cotton, and poverty reduction in India and can provide high spices, and rubber) and high-value activities. Cash returns to investment in South Asia. The highest crops have traditionally been important sources of returns to public spending during the 1970s through agricultural growth and employment in many parts the 1990s tended to be in research and development, of South Asia. They are more labor intensive than roads, education, and irrigation (box table 1.1.1). food staples (mechanization options are limited) and Although marginal returns have diminished over BOX TABLE 1.1.1 Returns to agricultural growth from investments in public goods and subsidies in India, 1960s–90s (percent) Public good 1960s 1970s 1980s 1990s Agricultural research and development 3.12 5.90 6.95 6.93 Road investment 8.79 3.80 3.03 3.17 Educational investment 5.97 7.80 3.88 1.53 Irrigation investment 2.65 2.10 3.61 1.41 Credit subsidies 3.86 1.68 5.20 0.89 Power subsidies 1.18 0.95 1.66 0.58 Fertilizer subsidies 2.41 3.03 0.88 0.53 Irrigation subsidies 2.24 1.22 2.38 — Sources: Fan, Gulati, and Thorat 2008; World Bank staff. Note: — = Not available. (continues next page) 16 MORE AND BETTER JOBS IN SOUTH ASIA BOX 1.1 Increasing productivity in agriculture (continued) time, they remain signi�cant. In contrast, returns to restrictions on marketing arrangements also constrain input subsidies (fertilizer, power, and credit) are gen- productivity growth in agriculture. There is thus a erally low. substantial agenda of institutional reform. Institutional weaknesses such as thin land markets, suboptimal water-use arrangements, and regulatory Sources: World Bank staff; Hazell et al. 2011. FIGURE 1.15 Ratio of working-age to nonworking-age population in South Asia, by country, 1960–2008 2.3 2.1 ratio of working-age population to 1.9 nonworking-age population 1.7 1.5 1.3 1.1 0.9 60 62 64 66 68 70 72 74 76 78 80 82 84 86 88 90 92 94 96 98 00 02 04 06 08 19 19 19 19 19 19 19 19 19 19 19 19 19 19 19 19 19 19 19 19 20 20 20 20 20 Afghanistan Bangladesh Bhutan India Maldives Nepal Pakistan Sri Lanka Source: Authors, based on data from World Bank 2011c. Cashing the demographic dividend Typically, there is an initial decline in this South Asia’s changing demographic pro� le ratio, reflecting a drop in the infant mortal- can help it meet the enormous employment ity rate that precedes a decline in the fertility challenge it faces.10 All South Asian coun- rate. The ratio subsequently increases as the tries are undergoing a process, known as the baby boom caused by the lagged decline in “demographic transition,� by which high fer- the fertility rate becomes part of the work- tility and mortality rates are replaced by low ing-age population. The resulting rise in the ones. A key indicator of where a country is share of the working-age to the nonworking- situated in the transition is the inverse depen- age population implies that there are fewer dency ratio, which is the ratio of the working- dependents to support (figure 1.15). The age population to the dependent population. resources saved as a result—the “demographic OVERVIEW 17 dividend�—can be used for high-priority way into the investments in physical and investments. Eventually, as the baby boom human capital that have the highest returns. cohort ages, the demographic transition gives Increased factor accumulation would then way to old age dependency. increase aggregate labor productivity, which Although the inverse dependency ratio would help absorb entrants into the labor has followed the same broad pattern in all market at rising wages and more remunera- of South Asia, there are country differences. tive self-employment. Afghanistan’s ratio started increasing only in Without policy reform, the demographic 2005. Bangladesh’s ratio rose sharply, catch- dividend cannot be increased or used to ing up with India’s in 2003 and exceeding it boost growth and living standards. In this thereafter. The improvement reflected, among event, entrants into the labor market will other things, a very rapid decline in fertility be absorbed at stagnant or slowly rising that was supported inter alia by the country’s levels of productivity, and the potential reproductive health program. Maldives saw of the demographic dividend will remain the fastest increase in the ratio, thanks to its untapped. Policies to improve the environ- plunging fertility rate. Pakistan’s ratio began ment for factor accumulation and raise a gentle climb in the 1980s and Nepal’s in the the quality of physical and human capital 1990s. formation are necessary to create better In its medium-fertility scenario, the jobs whether or not there is a demographic United Nations estimates that the inverse dividend. But the fact that, for most of the dependency ratio will peak for most South region, the window of demographic oppor- Asian countries around 2040. The excep- tunity will be open for only another three tions are Sri Lanka, where it occurred decades lends urgency to the need for policy around 2005, and Afghanistan, where it reform. will still be rising in 2040. Bangladesh, Because of volatile economic growth, Bhutan, India, and Maldives are already an uneven policy framework, and armed experiencing the demographic transition confl ict in a number of countries, there is and therefore have the potential to ben- only a broad correspondence between per e�t from it. Nepal and Pakistan, where the capita GDP growth and the demographic demographic transition started later, have transition. The acceleration in India’s eco- yet to see a dividend. nomic growth started in the 1970s, when its The demographic dividend grows when inverse dependency ratio started its climb the inverse dependency ratio rises more rap- (see figures 1.9 and 1.15). Bangladesh’s idly and peaks at a higher level. This will be acceleration began in the 1980s; during the case if the fertility decline occurs soon the middle of the decade, its ratio began after the decline in infant mortality and is to increase as well. With the exception of a rapid. Policies such as creating an effective slowdown in the 1980s, Sri Lanka has seen reproductive health program and expanding an acceleration of economic growth over female primary and secondary education, nearly �ve decades since the 1960s, when which reduces family size, can help bring its inverse dependency ratio started rising this about. rapidly. Pakistan, where economic growth The resources made available by the has been volatile around a broadly declining demographic dividend can be used for phys- trend across the decades, and Nepal, where ical capital deepening (electricity, transport) growth has been low and stagnant, have yet and human capital formation (education, to see a demographic dividend. With better skills) if the business environment is condu- policies, their growth performance could cive to making such investments. Policy also improve in the future as the demographic needs to help improve the quality of � nan- transition takes hold. cial intermediation, so that the increased Except in Nepal and, to a lesser extent private savings of households find their Bhutan, the female employment rate (the 18 MORE AND BETTER JOBS IN SOUTH ASIA ratio of female employment to the female demand for labor? How do the business working-age population) in South Asia is environments of individual South Asian among the lowest in the developing world— countries compare with the rest of the devel- primarily a reflection of the region’s low oping world? rates of female labor force participation. Enterprise surveys ask � rms to rate the Participation rates are particularly low in severity of inadequacies in the various ele- the three largest countries: Pakistan, where ments of the business environment for their almost four out of �ve women do not partic- ability to operate and expand their busi- ipate in the labor force, and Bangladesh and ness. These elements, which are external to India, where slightly more than two out of the � rm and resemble public goods, include every three do not participate. Nonpartici- regulation, physical infrastructure, the pation does not imply inactivity: household availability of skilled labor, macroeconomic duties were cited as the most important rea- conditions, the quality of the judiciary, and son for nonparticipation. A rising propor- crime and corruption. The question takes tion of working-age women in increasingly the form: “How much of an obstacle is X productive employment in the near future to the operation and growth of your busi- would provide a boost to growth in coun- ness?� The firm’s response regarding its tries such as Bangladesh and India, which severity—rated on a �ve-point scale, with 0 are going through the demographic transi- being no obstacle and 4 being a very severe tion. While the demographic transition is obstacle—is a measure of the marginal less advanced in Pakistan, the situation is reported cost imposed by the constraint on no less urgent there since the female partici- the operation and growth of its business. pation rate is the lowest in the region. (For These data can be interpreted as the differ- a comprehensive discussion of options to ence between the � rm’s pro�t in the hypo- improve economic opportunities for women, thetical situation in which the business see World Bank 2012.) environment poses a negligible obstacle to The employment challenge in South its operations and the � rm’s actual pro�t, Asia is one of improving job quality rather given the existing quality of the business than quantity, as job growth over long environment.11 periods tracks the growth of the working- age population. The challenge will be to Power, payments, and politics � nd better jobs for a workforce whose size will increase 25–50 percent in the coming The three most common binding con- decades. In the presence of policy reform, straints for medium-size urban formal the demographic transition can provide a � rms in South Asia are electricity, corrup- favorable tailwind in support of economic tion, and political instability (table 1.1 and growth and improving job quality. Policy figure 1.16).12 Although there are some will be needed, however, to address the variations, the top three constraints facing main demand- and supply-side constraints formal, urban firms are common to most to job creation, discussed in the next two countries. In every country except Bhutan sections respectively. and Maldives, electricity is one of the top two constraints; it is the top constraint in India and Sri Lanka. Except for Bhutan, political instability is among the top three Improving an inconducive constraints in all countries where it was business environment included in the survey instrument. In � ve What constrains the demand for labor in of the eight countries studied, corruption South Asia? What types of policy reform is among the four top constraints cited by would facilitate firm expansion and the urban formal sector � rms. OVERVIEW 19 TABLE 1.1 Top five constraints reported by South Asian benchmark firm in the urban formal sector, by country South Asia region Afghanistan Bangladesh Bhutan India Maldives Nepal Pakistan Sri Lanka Electricity 2 2 1 1 2 2 1 Political instability 1 1 2 n.a. n.a. 1 3 n.a. Corruption 3 3 3 2 3 4 Tax administration 4 5 5 3 1 Labor regulations 3 4 5 5 Inadequately educated labor 2 5 2 Access to land 4 4 1 Transport 1 3 Government policy uncertainty 5 4 2 Courts 4 5 Crime, theft, and disorder 5 5 Business licensing 4 Macro instability 3 Competition 4 Source: Authors, based on Carlin and Schaffer 2011b (from World Bank enterprise surveys). Note: A benchmark firm is a medium-size manufacturing firm with 30 employees that is domestically owned, does not export or import, is located in a large city, and did not expand employment in the preceding three years. n.a. = Not applicable (question was not asked). Analysis is based on pooled sample of enterprise surveys conducted between 2000 and 2010. Access to finance and tax rates constraints are excluded. Table 1.1 and �gure 1.16 show the sever- � rms within the urban formal sector across ity and ranking of constraints for a bench- countries in South Asia. mark firm in the urban formal sector. A benchmark firm is a medium-size manu- Electricity facturing firm with 30 employees that is In most South Asian countries, the cost domestically owned, does not export or imposed on firms by the electricity con- import, is located in a large city, and did straint is among the highest in the world; in not expand employment in the preceding Afghanistan, Bangladesh, and Nepal, it is three years. A comparison across coun- higher than in other countries at similar lev- tries and regions requires that these � rm els of per capita GDP (�gure 1.17). Moreover, characteristics, which are distributed dif- the severity of the constraint has increased ferently across countries, be controlled for. over time in India, Nepal, and Pakistan. For instance, if a country has a dominance The downward slope in the � gure implies of skill-intensive � rms, the answer to the that although � rms in richer countries can question on labor skills (“How much of an be expected to make greater demands on the obstacle are labor skills to the operation electricity grid, which would lead to rising and growth of your business?�) might be severity of complaints, richer countries can more important than it would be in coun- more than offset those demands in the pro- tries that do not have � rms requiring such vision of electricity, resulting in lower levels skills. That said, although the severity of of severity at higher incomes per capita. constraints differs between benchmark and The high frequency of power outages in nonbenchmark � rms (for example, � rms in South Asia is consistent with the reported the service sector), the ranking of the top severity of the electricity constraint. Indeed, constraints is very similar across types of Afghanistan, Bangladesh, and Nepal have 20 MORE AND BETTER JOBS IN SOUTH ASIA FIGURE 1.16 Severity of constraints reported by South Asian Corruption benchmark firm in the urban formal sector Corruption is among the top �ve constraints in �ve South Asian countries (see table 1.1). South Asian benchmark �rm Firms face high levels of corruption in a political instability 2.6 range of interactions with public of� cials, particularly for utilities and tax inspections electricity 2.2 (�gure 1.18). Government interactions that have the highest frequency of bribes vary by corruption 1.8 country (with the proportion of �rms report- tax administration 1.7 ing such payments in parentheses). government policy 1.4 • A fghanistan: Government contracts uncertainty (43 percent), electrical connections macro instability 1.3 (38 percent) • Bangladesh: Utilities (42–76 percent), tax competition 1.2 meetings (54 percent), import licenses crime, theft, and (51 percent) 1.2 disorder • India: Construction permits (67 percent), access to land 1.1 tax meetings (52 percent), operating licenses (52 percent), electrical connec- customs 1.1 tions (40 percent) inadequately • Pakistan: Electrical connections (71 per- 1.1 educated labor cent), water connections (62 percent), tax labor regulations 1.0 meetings (59 percent). transport 1.0 The high frequency of bribes faced in con- necting to power supply is another dimension business licensing 0.9 of the issue of access to electricity and may courts 0.9 be related to businesses having to compete to secure power (World Bank 2008a). More telecoms 0.7 than half of � rms in Bangladesh, India, and Pakistan are expected to pay bribes during tax 0 1 2 3 severity of constraint inspections. The tax systems in these countries are complex and create not only high costs of Source: Authors, based on Carlin and Schaffer 2011b (from World Bank enterprise surveys). compliance but also opportunities for corrup- Note: A benchmark firm is a medium-size manufacturing firm with 30 employees that is tion. (Chapter 4 compares the severity of cor- domestically owned, does not export or import, is located in a large city, and did not expand employment in the preceding three years. Analysis is based on pooled sample of enterprise ruption as an obstacle to doing business and surveys conducted between 2000 and 2010. The severity of constraint is rated by firms on a the prevalence of bribes in individual South 5-point scale, with 0 being no obstacle, 1 being a minor obstacle, 2 being a moderate obstacle, 3 being a major obstacle, and 4 being a very severe obstacle. Access to finance and tax rates Asian countries and countries outside the constraints are excluded. region at similar levels of per capita GDP.) Political instability The reported costs of political instabil- some of the highest reported outages in the ity are high in Afghanistan, Bangladesh, world, with virtually 100 percent of firms and Nepal; in all three countries, it is the experiencing them. Predictably, the use of most or second-most severe constraint (see generators to mitigate the effects of uncer- chapter 4 for details). These three countries tain power supply is higher in South Asia have some of the highest reported costs of than elsewhere, with 87 percent of � rms in political instability in the world. (Chapter 7 Afghanistan, 52 percent in Sri Lanka, and examines the costs imposed by armed con- 49 percent in India having generators. flict on � rms and workers.) OVERVIEW 21 FIGURE 1.17 Cross-country comparisons of reported severity of electricity constraint and power outages for a benchmark firm 4 a. Severity of electricity constraint Bangladesh Nepal 3 Afghanistan severity of constraint Pakistan 2 Sri Lanka India 1 Maldives Bhutan 0 6 7 8 9 10 11 log of per capita GDP in purchasing power parity dollars b. Power outages percent of �rms affected by at least one power outage a month Afghanistan Nepal 100 Bangladesh 80 60 India 40 Pakistan Bhutan 20 Maldives Sri Lanka 0 6 7 8 9 10 11 log of per capita GDP in purchasing power parity dollars Source: Carlin and Schaffer 2011b (based on World Bank enterprise surveys). Note: The cross-country regression line shows the relationship between the reported severity of the electricity constraint (panel a) and the percentage of firms experiencing more than one power outage per month (panel b) for a benchmark firm and the log of per capita GDP. The shaded area is the 95 percent confidence interval band around the regression line. Vertical bars show confidence intervals of 95 percent around the reported severity of the electricity constraint (panel a) and the percentage of firms experiencing more than one power outage per month (panel b) for countries in South Asia. Analysis is based on pooled sample of enterprise surveys conducted between 2000 and 2010. For further details, including why some observations in panel b are less than zero or are more than 100 percent, see notes to figures 4.2 and 4.3. 22 MORE AND BETTER JOBS IN SOUTH ASIA FIGURE 1.18 Percentage of firms expected to give gifts to public officials, by type of interaction 100 90 80 70 60 percent 50 40 30 20 10 0 operating import license construction electrical phone water meetings with government license permit connection connection connection tax officials contract Afghanistan Bangladesh Pakistan India South Asia world Source: Authors, based on data from World Bank enterprise surveys. Note: Figures show percent of firms in South Asian countries citing corruption as one of their top three constraints. FIGURE 1.19 Severity of constraints identified by South Asian benchmark (nonexpanding) and expanding firm in the urban formal sector Constraints facing job-creating firms in the urban formal sector telecoms 3.0 Job-creating firms, which are similar in political instability courts 2.5 all respects to the benchmark firm except electricity 2.0 business licensing that they expanded employment during 1.5 the preceding three years, report signifi- corruption transport cantly higher severity in 14 of the 16 busi- 1.0 0.5 ness constraints (�gure 1.19). The rankings tax administration 0.0 labor regulations are shown in Table 1.2. Job-creating � rms also report higher levels of mitigation activi- government policy ties, such as using generators as a response inadequately uncertainty educated labor to unreliable electricity supply and paying macro instability customs bribes to navigate a corrupt environment. Job-creating �rms in the urban formal sec- competition access to land tor perform well in other respects, too. They crime, theft, and disorder engage in research and development, intro- benchmark (nonexpanding) expanding duce new processes and products, sell to mul- tinational companies, offer in-� rm training, Source: Authors, based on Carlin and Schaffer 2011b (based on World Bank enterprise surveys). and have better-educated managers. The Note: Analysis is based on pooled sample of enterprise surveys conducted between 2000 and 2010. A point farther away from the origin indicates that the business constraint is considered more severe. increase in the cost of constraints for expand- The severity of constraint is rated by firms on a 5-point scale, with 0 being no obstacle, 1 being a ing � rms versus the benchmark is highest in minor obstacle, 2 being a moderate obstacle, 3 being a major obstacle, and 4 being a very severe obstacle. Survey does not make clear what firms mean by “competition.� Only statistically significant India and Pakistan. Job-creating �rms report differences in severity between the benchmark (nonexpanding) and the expanding firms are shown. lower costs in Afghanistan and Bangladesh. OVERVIEW 23 TABLE 1.2 Top five constraints reported by South Asian benchmark (nonexpanding) and expanding firm in the urban formal sector, by country South Asian region Benchmark firm Afghanistan Bangladesh Bhutan India Maldives Nepal Pakistan Sri Lanka Benchmark firm Benchmark firm Benchmark firm Benchmark firm Benchmark firm Benchmark firm Benchmark firm Benchmark firm Expanding firm Expanding firm Expanding firm Expanding firm Expanding firm Expanding firm Expanding firm Expanding firm Expanding firm Constraint Electricity 2 2 2 2 1 1 1 2 2 2 2 1 1 Political instability 1 1 1 1 2 2 n.a. n.a. 1 1 3 1 n.a. n.a. Corruption 3 3 3 4 3 3 2 1 3 4 4 4 1 Tax administration 4 4 5 5 5 5 3 3 1 4 Labor regulations 3 3 4 4 5 5 5 5 Inadequately educated labor 2 2 5 5 2 2 Access to land 4 5 4 4 1 1 Transport 1 1 3 3 Government policy uncertainty 5 5 4 3 2 2 Courts 4 5 5 Crime, theft, and disorder 5 3 5 3 Business licensing 4 4 Macro instability 5 3 3 Competition 4 4 Source: Authors, based on Carlin and Schaffer 2011b (based on World Bank enterprise surveys). Note: Analysis is based on pooled sample of enterprise surveys conducted between 2000 and 2010. n.a. = Not applicable (question was not asked). Access to finance and tax rates constraints are excluded. of workers to more productive sectors will Although the level of severity is different, require accumulation of physical capital. job-creating � rms rank constraints in much the same way that benchmark firms do, ranking electricity, corruption, and politi- Constraints facing rural firms cal stability as the top three constraints (see Improving the business environment can table 1.2).13 spur development of the rural nonfarm The severity of the electricity constraint economy, which accounts for an increasing facing urban formal �rms—as well as urban share of rural employment in many South informal � rms and rural nonfarm enterpris- Asian countries and, therefore, the creation es—prompts a discussion of the problems of better jobs within it. (See chapter 3 for a facing the sector and the policies and other discussion of the rural nonfarm economy.) initiatives being undertaken to address them Doing so requires an understanding of the (box 1.2). The need to make substantial constraints � rms in this sector face. investment in electricity is an example of the The severity of constraints reported by point made earlier that a rapid reallocation rural firms in Bangladesh, Pakistan, and 24 MORE AND BETTER JOBS IN SOUTH ASIA BOX 1.2 Options for reforming the power sector in South Asia South Asia is characterized by low levels of access, Improving the financial and commercial viability low consumption per capita, and wide demand- of the power sector supply gaps. Some 600 million people in the region Policy makers can choose from a range of options lack access to electricity—more than 40 percent of to improve the � nancial and commercial viability of the world total. Access rates range from 44 per- the power sector: cent of the population in Nepal to 77 percent in Sri Lanka. The average annual per capita consumption • Increase the level of tariffs to reflect the cost of for the region is 500 kilowatt hour, lower than any- supply, and rationalize tariffs to address cross- where else in the world except Africa. subsidization. Some countries have not revised Supply has not kept pace with demand, resulting tariffs in years. Others have made progress toward in shortages at peak times ranging from 1 gigawatt achieving cost-reflective tariffs, primarily to reduce (GW) in Bangladesh (13 percent) to 12 GW (10 per- �scal pressure. All countries offer “lifeline� rates to cent) in India. The toll on the economy is enormous: residential consumers to enable the poor to access in Pakistan, the cost of industrial load shedding at least a minimum quantity of electricity as well as is 400,000 lost jobs; in India, 17 percent of total nominally priced electricity to agriculture consumers capacity is based on expensive diesel generation. to support irrigation and food security. The burden Countries have responded through massive invest- of cross-subsidization falls on industrial and com- ment in expanding generation capacity. India added mercial consumers. Any tariff increase will need to 50 GW of capacity between 2006 and 2011 and ini- ensure that adequate safety nets are in place to min- tiated a series of “ultra mega� (4 GW) generation imize the impact on the poor. Innovative initiatives projects based on competitive bidding by independent such as the separate provision of heavy-duty agri- power producers. Bangladesh plans to develop 9.4 cultural feeders for agricultural needs and regular GW of new generation capacity by 2015. Bhutan has feeders for domestic and industrial purposes in the successfully established public-private partnerships Indian state of Gujarat has allowed transparency of for a large export-oriented hydropower project. agricultural consumption and, by providing reliable Signi�cant institutional reforms have also taken supply to both farmers and rural domestic consum- place since the 1990s. Most countries have unbun- ers, spurred the growth of rural productivity. dled their power sectors or corporatized previously • Reduce losses by improving collection, curbing vertically integrated power utilities. The sector has theft, and improving overall ef�ciency. India has been opened to private entry and greater competition initiated incentive schemes such as the Restructured in generation, transmission, and distribution, and new Accelerated Power Development Reform Program regulatory frameworks and independent regulatory (R-APDRP), which aims to limit losses to 15 bodies have been established. The degree of reform percent. varies across countries and across states in India. • Improve the capacity and independence of regula- Sector � nancial losses across the region are large, tory agencies to ensure transparency and account- resulting from the misalignment of tariffs, the high ability in tariff setting, which continues to be driven cost of power procurement, and high transmission by political exigencies. New initiatives in regula- and distribution losses. In India, the combined cash tion have been put in place, such as the implemen- loss of state-owned distribution companies is more tation of multiyear tariffs in India, which provide than $20 billion a year (total investment needs for certainty regarding the costs for which utilities can 2010–15 are estimated at $300 billion). The sector be held accountable and reduce day-to-day regula- de�cit in Pakistan is estimated at about $2 billion a tory interference. year (total investment needs for 2010–20 are esti- mated at $32 billion). Enhancing the business environment for private Several challenges need to be addressed to alle- investment in power sector viate power shortages and improve service delivery. Each is addressed briefly below. The generation sector has attracted substantial pri- vate interest, but obstacles remain in the form of (continues next page) OVERVIEW 25 BOX 1.2 (continued) procedural bottlenecks (for example, land acquisi- Bhutan to India and, to a limited extent, between tion, environmental and forest clearances, provision Nepal and India. of water for thermal plants); limited technical and � nancial capacity to implement large projects; and Improving the governance of utilities and the shortage of fuel (both domestic and imported) to strengthening institutional capacity ramp up capacity utilization. The region also needs Steps are being taken to develop strong boards and to improve the operating environment to attract pri- high-quality professional management. Doing so is vate players in transmission and distribution. necessary to transform an organizational culture of risk-averse top-down bureaucratic control to one Exploiting the significant potential of intraregional more suited to commercialization. Some states in energy trade India (such as Andhra Pradesh and West Bengal) have adopted technology initiatives, particularly in One of the most cost-effective options for alleviating metering, and accountability frameworks to improve shortages in the region is increasing intraregional sector performance. energy trade. Such trade has increased in recent years, particularly in the form of hydro exports from Source: World Bank staff. Sri Lanka is compared with the severity TABLE 1.3 Top five constraints reported by micro benchmark firm in reported by urban (formal) �rms in the same the urban and rural sectors of Bangladesh, Pakistan, and Sri Lanka countries.14 As rural � rms typically employ Bangladesh, Pakistan, Sri Lanka far fewer than 30 employees, the comparison is with a micro benchmark � rm—a bench- Constraint Urban, formal Rural mark �rm with 5 employees. The results are Electricity 1 2 as follows (table 1.3 and �gure 1.20): Political instability 2 3 Corruption 3 • Rural � rms report less severe constraints Macro instability 4 1 to their operations than urban firms. Access to land 5 This pattern is not unusual in develop- Transport 4 ing countries, where larger urban firms Source: Authors, based on Carlin and Schaffer 2011a (based on World Bank enterprise surveys). are typically more productive and, during Note: Analysis is based on a pooled sample of enterprise surveys conducted in Bangladesh, Pakistan, and Sri Lanka between 2000 and 2010. Access to finance and tax rates constraints are the course of their expansion, place more excluded. Crime, theft, and disorder as well as competition constraints were not asked in the rural demands on publicly provided services surveys. Macro instability was asked only in Bangladesh and political instability was asked only in Bangladesh and Pakistan. Only the top four constraints are shown for rural firms, as the remaining than rural � rms. constraints were not reported on average as obstacles. • Rural firms identify electricity as one of the most binding constraints to their operations. They report levels of power outages similar to those for urban � rms where less than 70 percent of enterprises and use generators more intensively than use electricity from the national grid. urban sector � rms. In Bangladesh, 73 per- • Unlike their urban counterparts, rural cent of nonmetropolitan nonfarm enter- firms cite transport as one of their top prises have an electricity connection, with four constraints. Firms in Bangladesh and 99 percent of them reporting power out- Sri Lanka complain about the poor con- ages. In Sri Lanka, shortages and unreli- ditions and inaccessibility of rural roads. ability of power are severe in rural areas, Poor transport limits access to larger, 26 MORE AND BETTER JOBS IN SOUTH ASIA FIGURE 1.20 Severity of constraints reported by micro urban markets, forcing rural enterprises benchmark firm in urban and rural sectors of Bangladesh, to sell predominantly to local customers, Pakistan, and Sri Lanka which limits the size of the market for their products and goods.15 Rehabilitat- telecoms ing and maintaining existing rural roads electricity 3.0 labor regulations as well as building new roads would open 2.5 up opportunities for rural � rms. political instability 2.0 customs 1.5 1.0 Constraints facing informal urban firms corruption transport 0.5 0.0 Enterprise surveys in the informal and for- inadequately mal urban sectors in India make it possible macro instability educated labor to compare the constraints encountered by informal firms with those facing formal access to land business licensing firms. The comparison is between formal government and informal urban benchmark firms. As tax administration policy uncertainty courts the median � rm in the informal urban sec- tor has fewer than 30 employees, the bench- micro benchmark (urban) micro benchmark (rural) mark � rm size is 5 rather than 30.16 There are both similarities and differences Source: Authors, based on Carlin and Schaffer 2011c (based on World Bank enterprise surveys). between formal and informal firms in the Note: A point farther away from the origin indicates that the business constraint is considered more ranking of business constraints (table 1.4 and severe. The severity of constraint is rated by firms on a 5-point scale, with 0 being no obstacle, 1 being a minor obstacle, 2 being a moderate obstacle, 3 being a major obstacle, and 4 being a �gure 1.21). Both types of �rms cite electric- very severe obstacle. Only statistically significant differences in reported severity between the ity as an important constraint. Informal �rms micro benchmark firms in urban and rural sectors are shown. For further details on the sample and constraints analyzed, see the note for table 1.3. are more likely to cite access to land and trans- port and less likely to cite corruption and tax administration.17 Informal �rms report access to �nance as the most severe constraint to the FIGURE 1.21 Severity of constraints reported by micro benchmark operations of their businesses. However, this firm in India’s urban formal and informal sectors �nding does not necessarily have implications courts electricity 3.0 telecoms 2.5 corruption macro instability TABLE 1.4 Top five constraints reported by micro 2.0 1.5 benchmark firm in India’s urban formal and informal tax administration 1.0 business licensing sectors 0.5 India 0.0 crime, theft, and disorder competition Formal Informal Electricity 1 1 access to land customs Corruption 2 labor regulations government policy Tax administration 3 inadequately uncertainty Crime, theft, and disorder 4 transport educated labor Access to land 5 3 micro benchmark (formal) micro benchmark (informal) Competition 2 Transport 4 Source: Authors, based on Carlin and Schaffer 2011a (based on World Bank enterprise surveys). Government policy uncertainty 5 Note: A point farther away from the origin indicates that the business constraint is considered more Source: Authors, based on Carlin and Schaffer 2011a (based on World Bank severe. The severity of constraint is rated by firms on a 5-point scale, with 0 being no obstacle, 1 enterprise surveys). being a minor obstacle, 2 being a moderate obstacle, 3 being a major obstacle, and 4 being a very Note: Analysis is based on a pooled sample of enterprise surveys severe obstacle. Only statistically significant differences in reported severity between the micro conducted in India between 2000 and 2010. Access to finance and tax benchmark firms in urban formal and urban informal sectors are shown. For further details on the rates constraints are excluded. Political instability was not asked in the sample and constraints analyzed, see the note for table 1.4. India surveys. OVERVIEW 27 for policy, as “access to �nance� is a dimen- technical skills needed for the job. Two out of sion of the business environment that, unlike three employers reported that most of these the judiciary or tax administration, is not a skills are “very� important but that they public good. The � rm’s response regarding were only somewhat satis�ed (at best) with the inadequacy of �nance could simply reflect the graduates’ skills (figure 1.22). (Similar the fact that some � rms do not have bank- concerns are echoed by Sri Lankan employ- able projects. Other indicators of access (for ers in a survey of the information technology example, the proportion of �rms using exter- workforce [Sri Lanka Information Commu- nal �nance) suggest that �nance may indeed nication Technology Association 2007].) The be an issue for micro and small �rms in some foundation for many of these skills is estab- countries. lished well before graduates enter the world The concern regarding access to land of work—in primary and secondary educa- expressed by urban informal firms may tion and indeed even earlier. reflect the impact of regulations that shape The wage premium has been rising for the operation of land markets in India. Den- higher levels of education in all countries, sity regulations in India, which limit the even as the supply of educated workers has ratio of floor space to plot area, lead cities to increased. Figure 1.23 presents trends in expand outward instead of upward. Together wage premiums for different levels of educa- with limited accessibility of public transport, tion in three South Asian countries between such expansion can make it more dif�cult for about 2000 and 2008 and a longer period for informal manufacturing units to be located India. The premiums reflect the differential where they should be—close to buyers and between the average earnings of a worker suppliers. Relaxing density regulations, with a particular level of educational attain- improving urban transport, and increasing ment and the average earnings of a worker the supply of property might help reduce the with the level of attainment just below (for severity of the land constraint for informal details, see chapter 5). The pattern over time urban �rms (World Bank 2011a). has been that the premium to lower levels of education has been falling while the premi- ums for upper-secondary and tertiary edu- Improving workers’ skills cation have been increasing. These changes Enterprise managers in the urban formal sec- have been taking place in a context in which tors in Bhutan, India, and Maldives report the educational attainment of the labor force inadequate skills of the labor force among has been rising. The pattern is thus consistent the top �ve constraints to the operation and with a situation in which the supply of work- growth of their �rms (see table 1.1). In Bhutan ers at lower levels of education is increasing and Maldives, inadequate skills of the labor faster than demand, whereas the demand for force are among the top two constraints; in workers with secondary or tertiary education India they rank 5th. Firms in Afghanistan is outpacing the increased supply. and Pakistan rate skills constraints among The heterogeneity of the region is reflected the least problematic. in variations across countries. India and Focused employer surveys in India and Sri Nepal have seen increases in premiums to Lanka highlight concerns with the skills of both upper-secondary and tertiary educa- tertiary education graduates. Employers hir- tion. Indeed, the wage premium for ter- ing fresh engineering graduates in India eval- tiary education more than doubled in India uated the degree of importance of a broad between 1999/2000 and 2009/10, despite a range of skills and their level of satisfaction large increase in the share of the labor force with recent hires. They rated behavioral skills with tertiary education. In Nepal, the larg- (teamwork, reliability, leadership, willingness est relative increases in wage premiums were to learn), creative thinking and problem- at the upper-secondary level, suggesting that solving skills, and specific knowledge and demand increases at this level were greater 28 MORE AND BETTER JOBS IN SOUTH ASIA FIGURE 1.22 Employers’ perceptions of skills of recently graduated engineers in India integrity reliability teamwork willingness to learn entrepreneurship self-discipline communication in english self-motivated flexibility leadership responsibility modern tools knowledge math/science/English creativity written communication reading technical skills experiments/data analysis verbal communication basic computer problem solving empathy system design contemporary issues advanced computer customer service 0 not at all not very somewhat very extremely satisfaction importance Source: Blom and Saeki 2011. than the small increases in supply. In con- people age 15–34 increased in all countries trast, in Pakistan and Sri Lanka, the wage between the late 1990s and early 2000s, premium increased for upper-secondary edu- but it is still low in most countries, ranging cation but decreased for tertiary education, from 2.5 years in Afghanistan to 7.1 years particularly in Sri Lanka. in India. (It is higher in Maldives [7.8 years] Despite significant progress in recent and Sri Lanka [10.2 years].) In many coun- years, the contrast between increasing tries, the picture is considerably worse for demand for higher levels of education and women (see panel a of �gure 1.24). the educational attainment of the labor Success in school is affected by what hap- force remains stark. Educational attainment pens to children before they enter school. In remains low, particularly in secondary and fact, the greatest payoffs to subsequent edu- tertiary education, with well over a quarter cational investments may well come from of the labor force in all countries except Sri addressing poor nutrition and other factors Lanka lacking any education at all (�gure in early childhood, before children enter for- 1.24). The average years of education of mal schooling (box 1.3). OVERVIEW 29 FIGURE 1.23 Wage premiums in selected South Asian countries, by level of education a. India and Sri Lanka 60 50 43 42 42 43 43 42 40 37 36 36 36 percent 31 32 33 33 30 30 26 26 20 19 21 18 20 17 14 16 15 10 13 10 0 1993/94 1999/2000 2004/05 2009/10 2000 2008 India Sri Lanka 60 b. Nepal and Pakistan 40 37 35 32 34 34 percent 32 28 26 25 24 22 23 24 22 20 14 8 10 11 13 9 0 1998/99 2007/08 1999/2000 2007/08 Nepal Pakistan incomplete primary complete primary complete lower-secondary complete higher-secondary complete tertiary Source: Authors, based on data from national labor force and household surveys. Note: The first bar for each country-year pair reflects the wage premium for even some primary education relative to no education; the last bar reflects the wage premium for completing tertiary relative to completing upper-secondary education. The education challenge facing South consider the following actions in the school Asia is broad. It includes improving nutri- subsector: tion and other factors in early childhood, increasing attainment from primary to sec- • Address information gaps by developing ondary and higher levels, ensuring equal national assessment systems that provide opportunity for all groups, and equipping reliable feedback on learning. graduates with the skills necessary to suc- • Improve capacity and accountability at the ceed in the world of work. Country priori- school level by devolving greater respon- ties will vary. In Afghanistan and Pakistan, sibility to schools while increasing their achieving universal primary education accountability to local stakeholders. remains a priority, and these countries still • Improve the quality and performance have significant gender disparities in pri- of teachers by engaging in transparent mary education. India, Maldives, and Sri recruitment and development of career Lanka are focusing on expanding upper- and pay systems that build capacity and secondary school. All countries are expand- provide incentives. ing tertiary education. A key priority for all South Asian countries is to improve the As more and more students enter higher quality of learning and skills of graduates levels of education, pressure to expand ter- at all levels. tiary education will intensify. Priorities to Improving the quality of learning in ensure a focus on the quality and relevance primary and secondary schools requires of skills of graduates of both tertiary institu- strengthening incentives and capacity in the tions and preemployment training systems school system. To do so, governments could include the following: 30 MORE AND BETTER JOBS IN SOUTH ASIA FIGURE 1.24 Share of South Asian labor force with no education, with international comparisons a. Female labor force 100 Afghanistan 2008 Bhutan 2003 Pakistan 2000 80 Nepal 1999 Bhutan 2007 Pakistan 2008 India 1994 India 2000 Pakistan 2009 Nepal 2008 India 2005 60 India 2008 Bangladesh 2002 India 2010 percent Bangladesh 2005 Maldives 1998 40 Maldives 2004 20 Sri Lanka 2000 Sri Lanka 2004 Sri Lanka 2008 0 Sri Lanka 2006 0 1,000 2,000 3,000 4,000 5,000 per capita GDP in purchasing power parity dollars b. Male labor force 100 80 Bhutan 2003 Afghanistan 2008 60 Bhutan 2007 percent Nepal 1999 Bangladesh 2002 Maldives 1998 Pakistan 2000 40 Bangladesh 2005 Pakistan 2008 Pakistan 2009 Nepal 2008 India 1994 India 2000 Maldives 2004 India 2005 20 India 2008 India2010 Sri Lanka 2006 Sri Lanka 2000 Sri Lanka 2008 0 Sri Lanka 2004 0 1,000 2,000 3,000 4,000 5,000 per capita GDP in purchasing power parity dollars Sources: Authors, based on data from World Bank 2011c and national labor force and household surveys. Note: The dark line shows the predicted values of the share of the labor force with no education by per capita GDP, based on a cross-country regression (excluding high-income countries). The shaded area is the 95 percent confidence interval band around the regression line. GDP = gross domestic product. OVERVIEW 31 BOX 1.3 The critical role of nutrition in early childhood development The � rst years of life— long before formal schooling only modestly. This means that high malnutrition begins—are a key period for building human capi- and micronutrient de�ciencies are likely important tal. The bene�ts of health and nutrition early on can contributors to developmental delays in low-income have effects that persist through life; damage from groups in South Asia; they may also be important childhood disease and malnutrition in terms of lost factors for overall cognitive development of the opportunity for learning can be dif� cult to undo. broader population. Income growth alone will not Low levels of cognitive development in early child- eliminate malnutrition and micronutrient de� cien- hood are often strongly correlated with low socio- cies; focused attention is needed. economic status and malnutrition. In addition to Most South Asian countries do not have inte- protein-energy malnutrition, micronutrient de�cien- grated policy frameworks for early childhood devel- cies, which often begin before birth, can impair cog- opment. Early childhood interventions—which nitive and motor development and therefore school include nutrition, hygiene, early cognitive stimula- outcomes. tion, and preschool programs—are among the most South Asia has some of the highest rates of cost-effective investments for improving the quality malnutrition in the world, as well as high levels of and ef� ciency of basic education, as well as labor anemia and iodine de�ciency. As measured by stunt- market success. ing, underweight, and wasting, it has the world’s India has the strongest enabling policy frame- highest prevalence of malnutrition in children under work in South Asia, with a foundation deriving from �ve (box � gure 1.3.1). a Indeed, malnutrition rates the constitution, among other sources. National are higher than in Sub-Saharan Africa. South Asia nutrition policies and programs in Bangladesh and also has high levels of anemia and iodine de�ciency. some aspects of public health campaigns related to In South Asia, as globally, rates of malnutrition, nutrition in Pakistan have contributed to improve- anemia, and iodine de�cits improve with wealth, but ments in nutrition in both countries, but neither BOX FIGURE 1.3.1 Percentage of children under five with malnutrition, by region and country 70 60 59 50 47 48 49 47 42 44 42 42 40 39 40 38 percent 32 31 33 32 30 26 23 25 21 20 19 21 19 20 17 15 13 14 12 13 14 12 10 11 10 9 9 78 5 4 2 6 0 ea d c Af and ld ca ka s an an h a l an sia pa ive i� di bb n es or fri an ut ist ist Ea Cari ica a hA ac n a In Ne lad ald W ric nA rth st Bh an iL k dP Pa ut No e Ea ng M Sr r gh ra th Ame So an Ba ha Af dl ia Sa id As tin e M b- La st Su stunting underweight wasting Source: Authors, based on World Bank 2011c. (continues next page) 32 MORE AND BETTER JOBS IN SOUTH ASIA BOX 1.3 The critical role of nutrition in early childhood development (continued) country has a national policy on early childhood mother-child interactions through home visits. The development. Lady Health Workers Programme in Pakistan—a Ver y few prog rams at scale seek to inte- community-based government preventive care grate early childhood interventions. Several pilot program—holds promise for promoting nutrition projects can serve as a laboratory for design- and child care through scaling up of carefully eval- ing cost-effective programs, but many lack care- uated cost-effective pilot designs. ful plans for evaluation. Potentially promising efforts are under way, however. For example, Source: Authors, based on Alderman 2011. pilots run by the International Center for Diar- a. Wasting, stunting, and underweight indicators refer to the proportion of rheal Disease Research (ICDDR) in Dhaka have children under five whose weight for height, height for age, and weight for age respectively are more than two standard deviations below the medians stronger evaluation designs and have proven the of an international reference population recognized by the World Health feasibility of promoting better parenting and Organization (WHO). • Provide information on the qualit y Beyond protecting the basic rights of of graduates of institutions and their workers, labor market institutions have employability, and strengthen quality an important role to play in regulating the assurance and accreditation. employment relationship, with potentially • Increase the role of the private sector in important implications for labor market provision and that of employers in the efficiency and social protection. Employ- management of public institutions. ment protection legislation covers the kinds • Increase the autonomy of public higher of contracts permitted and the conditions education institutions and improve incen- and procedures for termination. Restric- tives for improved performance, such as tions on nonpermanent hiring and employer those provided by moving from histori- dismissal rights can increase employment cally negotiated budgets to performance- security and provide protection to work- based approaches. ers from arbitrary dismissal by employ- • Increase contributions from students while ers. If excessively restrictive, however, they protecting students less able to pay. can discourage formal job creation, limit the ef� cient reallocation of labor, and fail to provide real protection, as employers Reforming labor market find ways around the rules. The evidence institutions suggests that, in some countries, notably Labor market policies in South Asia need to India and Sri Lanka, the ef�ciency costs of strike a balance between facilitating protec- employment protection legislation outweigh tion for the vast majority of workers, primarily the bene�ts in terms of worker protection in the informal sector, who are not covered by (see chapter 6). social protection instruments while enhancing Labor market policies in India, Nepal, their incentives for income generation. Efforts and Sri Lanka are oriented toward protect- are needed on two fronts: (a) reforming statu- ing jobs. India, for example, has employment tory regulations and institutions to encour- protection laws that are considerably tighter age job creation in the formal economy while than laws in Western countries and most protecting the fundamental rights of covered other major emerging economies (figure workers and (b) building on programs that 1.25). India, Nepal, and Sri Lanka require can help informal workers adjust to labor not only noti�cation but prior approval by market shocks and improve their productivity the state to lay off or retrench workers, indi- and future earnings potential. vidually or collectively. In Sri Lanka, prior FIGURE 1.25 Employment protection indicators in selected countries OECD employment protection benchmark score 4 3.5 3 0.3 2.5 2.19 1.6 0.6 1.6 0.9 1.2 1.1 1.6 2 1.3 1.1 0.0 0.9 0.8 0.3 0.5 1.5 1.35 0.7 0.9 0.6 1.6 0.0 0.4 0.5 0.3 0.0 0.2 1 0.88 0.1 0.1 0.5 0.5 0.3 0.6 0.3 1.8 1.7 1.5 0.1 0.4 0.5 0.1 0.0 1.4 0.5 1.1 1.2 1.0 1.0 1.1 1.2 0.5 0.6 0.6 0.9 0.7 0.7 0.4 0.5 0.6 0 0.1 a es a m d ile . ico l ey y ain g ca n il a sia k ep ga di an az ar ad in ur tio lan do at fri rk Ch ne ex In R Sp Ch tu nm Br bo rm n ra St Tu hA a ng a, Ca do M r Ze de Po m Ge De re d Ki ut In ite xe Ko Fe w So d Ne Lu Un te ian i Un ss Ru selected OECD countries selected non-OECD countries protection of permanent workers against individual dismissal (1) speci�c requirement for collective dismissals (2) regulation of temporary employment (3) OECD average individual (1) OECD average individual + collective (1+2) OECD average total (1+2+3) Source: OECD 2009. Note: OECD = Organisation for Economic Co-operation and Development. The OECD employment protection indicators cover three aspects of employment protection: individual dismissal of permanent workers, regulation of temporary employment, and specific requirements for collective dismissal. Each subindicator ranges from 0 to 6, with 0 the least restrictive and 6 the most restrictive. The overall indicator—the sum of the three subindicators, weighted at 5/12, 5/12, and 2/12—also ranges from 0 to 6. 33 34 MORE AND BETTER JOBS IN SOUTH ASIA FIGURE 1.26 Weeks of wages required to be paid in severance in regions, country income groups, and selected South Asian countries, by length of service a. International comparators by region 98 100 80 60 54 weeks 47 50 43 43 40 29 33 29 27 25 26 25 16 16 17 21 21 21 20 14 15 12 14 4 4 3 3 6 10 4 11 6 9 2 11 4 4 5 11 0 Af ran tin aci� ia be d Af and on lo at for sia an a l an h ka pa di es ib an P s an ist ist iti d tA hA In Ne a Am c an a en n ha ve per n lad ric ric ar ca rth st pm io ns De o tio an iL k an Eas Sa Pa ut No e Ea t e C eri ng tra Sr d o - sa gh So b- Ba an C i Af ic an dl Su id om Org M th La on Ec b. International comparators by country income group 98 100 80 60 50 54 weeks 47 43 43 40 32 29 25 26 21 21 21 25 20 16 20 17 20 15 12 11 10 6 9 11 5 3 5 4 4 2 4 4 0 a tri e tri e rie - nt le- sia an a l an h pa nt le nk di un m un m es ist ist ou idd ou idd hA In Ne es es s co inco co inco lad s La rie an k ec m ec m Pa ut ng Sri gh w- - m er- m r- So gh Ba co pe Af lo co w hi in up in lo 1 year 5 years 10 years Source: Holzmann and others 2011. approval is necessary only when the work- Enterprise managers in India, Nepal, and er’s written consent for layoff cannot be Sri Lanka report labor regulations to be an obtained. This is different from India, where important constraint to the operation and large �rms need prior approval from the gov- growth of their business. In contrast, formal ernment before they can dismiss an employee sector � rms in Afghanistan and Bangladesh even if the worker consents to the dismissal. rate labor regulations as among the least Severance pay, which typically requires problematic. There is some evidence that minimum tenure in the establishment and labor regulations become more costly to �rms increases with seniority, is also high in most as per capita GDP rises (�gure 1.27). India, South Asian countries, particularly in Sri Nepal, and Sri Lanka report higher levels of Lanka but also in Bangladesh, Nepal, and severity of this constraint than other countries Pakistan (�gure 1.26). at their level of development, however. When Most countries have reasonably flexible asked which labor regulations most affected contracting rules. Bangladesh, Nepal, and the operation of their businesses, nearly one Pakistan place limits on the use of fixed- in three �rms in India for whom labor regula- term contracts, however, and lack of clarity tions were perceived as a moderate or severe in India has led to widely varying interpre- constraint reported that restrictions on dis- tations, regulations, and practices across missal are a constraint to hiring. About one states. in four cited restrictions on casual work, and OVERVIEW 35 FIGURE 1.27 Cross-country comparison of reported severity of the labor regulation constraint 4 3 severity of constraint 2 Sri Lanka Nepal Maldives Bhutan India Pakistan 1 Bangladesh Afghanistan 0 0 6 7 8 9 10 11 log of per capita GDP in purchasing power parity dollars Source: Carlin and Schaffer 2011b (based on most recent World Bank enterprise surveys). Note: The cross-country regression line shows the relationship between the reported severity of the constraint for a benchmark firm and the log of per capita GDP. The shaded area is the 95 percent confidence interval band around the regression line. Vertical bars show confidence intervals of 95 percent around the reported severity of the constraint for countries in South Asia. Analysis is based on a pooled sample of enterprise surveys conducted between 2000 and 2010. one in five cited constraints on temporary lengthy and costly procedures for resolving work. disputes and grievances provide additional In addition to creating disincentives for incentives for noncompliance. expanding the formal sector, labor market South Asian countries would bene�t from rules in South Asia have failed to protect reorienting their labor market policies from workers for two main reasons. First, these “protecting jobs� to “protecting workers.� rules cover much less than 10 percent of the As the region modernizes, an approach that labor force in most countries and less than a moves away from protecting jobs through third even in Sri Lanka, which has the most strong job security laws to encouraging more formalized labor market. Second, the impact flexibility in the labor market, while provid- of labor market rules is weakened by non- ing workers with better tools to manage the compliance. As regulations become more fluctuations of the market, will lead to more costly, �rms increasingly employ strategies to job creation and offer more protection. circumvent them, reducing de facto protec- This approach requires two coordinated tion from labor market regulations. In India, strategies. The �rst involves realigning labor for example, medium-size and large �rms in market regulations and institutions to relax organized manufacturing adjust employment the procedures and costs associated with levels by hiring and terminating contract dismissals; extending the legality of nonper- workers. In Sri Lanka, where statutory sever- manent contracts; improving protection of ance rights are generous, workers often fail to fundamental worker rights; improving the bene�t from them because of nonpayment or ef�ciency of dispute resolution and the enforce- partial payment, particularly during periods ment of employment protection legislation; of economic distress. Limited options and and streamlining and clarifying regulations. 36 MORE AND BETTER JOBS IN SOUTH ASIA The second involves strengthening the tools contribute, coverage can be expanded only available to workers in both the formal and if governments are willing to allocate signif- informal sectors to help them manage labor icant subsidies. The level of subsidies would market shocks. These tools include income need to be determined not only in light of support in the event of unemployment and the mandate of the programs but also by active labor market programs, including cost- other calls on budgetary resources. It there- effective training and employment services. fore seems likely that protection will be built Over time, this strategy will create a more incrementally on numerous existing schemes favorable environment for formal sector job and adjusted in light of lessons learned dur- creation. Less restrictive regulations, espe- ing their implementation. cially pertaining to dismissal, could create Well-targeted and well-designed pro- incentives for formal sector job creation; ben- grams can help informal workers smooth efit certain groups, including women; and consumption and enhance their income- encourage compliance with the law. generating potential. Countries in the region In the long run, as the workforce becomes have a variety of training, public works, and better educated and productivity rises, the self-employment assistance programs, oper- formalization gains from such an institu- ated by the government as well as by pri- tional framework could be considerable. In vate and nongovernment sponsors. (Details the short run, however, this strategy would are in chapter 6.) The effectiveness of many benefit only some informal sector workers of these programs is not well understood who might be able to �nd formal sector jobs because of the lack of evaluation evidence. in a more favorable regulatory environment. However, if well targeted and efficiently The vast majority of workers would continue implemented, such programs can incor- to lack social security bene�ts and disability porate both a safety net perspective (help- and health coverage. ing workers manage income-related risks) One way to close the gap between pro- and an activation perspective (helping them tected and unprotected workers would be to improve their capacity to generate income). extend (statutory) social insurance to infor- A priority is to encourage evaluation and mal workers. India’s Rashtriya Swasthya expand programs that meet standards of Bima Yojna (RSBY) is designed to provide targeting and cost-effectiveness. hospitalization coverage to households below the poverty line. The new pension law in Maldives provides for matching pension con- Creating jobs in conflict-affected tributions for informal sector workers such as areas �shers, in order to encourage informal sector Measured by the proportion of country- participation in the pension scheme. years in conflict since 2000, South Asia Any such plan needs to take account of is the most conflict-affected major region � nancing, as well as the effect on the inci- in the world (�gure 1.28). Four of the top dence of informality itself. Employers may be 10 countries in terms of direct deaths from more likely to seek ways to opt out of formal armed conflict in 2008 were in South contributory systems if they know employees Asia (Afghanistan, Pakistan, India, and can access social protection in other ways; Sri Lanka). Ongoing confl icts affect all of workers themselves may prefer to remain Afghanistan and parts of Pakistan and, at a informal, depending on how programs are lower intensity, India. Nepal and Sri Lanka � nanced. Thus the potential for social insur- are in postconfl ict status. ance to be extended into the informal sector Confl ict affects the demand for and the will depend a great deal on � nancing and supply of labor. On the demand side, it affects the scope to which individuals are able to both the incentives and the ability of �rms to contribute. Given the high prevalence of invest in conflict-affected regions and create workers with very limited or no capacity to jobs. In addition to concerns about security, OVERVIEW 37 FIGURE 1.28 Proportion of country-years in armed conflict, by region, 2000–08 0.6 0.50 0.5 0.4 proportion 0.3 0.20 0.2 0.16 0.15 0.1 0.09 0.05 0.0 sia ca Af and ia ia be d ib an As As fri hA a an nA ric rth st ar ca l st ra ut No le Ea Ea e C ri nt ra th Ame So Ce ha d Sa id nd tin M b- ea La Su p ro Eu Source: UCDP/PRIO Armed Conflict Dataset version 4-2009 (http://www.prio.no/CSCW/Datasets/Armed-Conflict/UCDP-PRIO/Armed-Conflicts-Version-X-2009/). Note: Armed conflict refers to internal armed conflicts between the government of a state and one or more internal opposition groups that result in at least 25 battle-related deaths a year. disincentives to the operation of private �rms In Afghanistan, 56 percent of the working- may arise from inadequate infrastructure, age population was employed in low-conflict confusing regulations and the poor quality of and 68 percent in high-confl ict provinces. governance in areas affected by confl ict. On These numbers usually reflect higher female the supply side, confl ict affects the capacity labor force participation in confl ict-affected of the population to supply labor—because areas. of security concerns and, over time, disrup- Labor markets in conflict zones differ tions to education, increased mortality and from labor markets in nonconflict areas morbidity, and the loss of job-related skills in several other ways. First, except in Sri and training. At the same time, economic Lanka, jobs in high-confl ict areas are more need and the absence of key income earners likely to remain rural and based on agricul- could lead to an increase in labor supply, par- ture. The terrain in rural areas makes them ticularly among women. more favorable to rebellion (Collier and Armed conflict in South Asia seems to Sambanis 2005). Furthermore, the confl ict be associated with an increase (or a smaller itself delays the structural transformation of reduction) in the share of the working-age the economy. population that is economically active and Second, there is a lower likelihood of mov- employed. Between 1996 (preconfl ict) and ing to better jobs in conflict areas, as the 2004, for example, the proportion of the workforce is more likely to remain engaged working-age population in Nepal that was in unpaid family labor, partly because of the employed increased about 2 percentage higher concentration of employment in agri- points in low-confl ict areas and more than culture and related activities. The workforce 4 percentage points in high-confl ict areas. is also less likely to become regular wage or In India, employment rates in conflict-af- salaried employees because of the relative fected areas fell less than in peaceful areas. absence of employers. 38 MORE AND BETTER JOBS IN SOUTH ASIA Third, the workforce is less likely to be Governments can improve security in con- well educated—although confl ict in Nepal flict situations and help restore livelihoods by seems to have been associated with an implementing disarmament, demobilization, improvement in school access—and there- and reintegration (DDR) programs. Such fore less able to access better jobs were programs, which target excombatants, are they to become available. This outcome underway in Afghanistan and Nepal. They reflects both the negative impact of con- include three broad phases: fl ict on the demand for schooling and the • Disarmament (collecting and disposing of destruction of schools and complementary weapons) infrastructure. • Demobilization (disbanding military Even when it is already over, armed con- structures) flict remains a serious obstacle to job cre- • Reintegration (facilitating the return of ation in South Asia. Almost 60 percent of former combatants to civilian life, the �rms included in the enterprise surveys rank armed forces, or the police). political instability as a major or severe con- straint to doing business. In Afghanistan, In addition to providing a minimal level � rms located in areas where conflict is most of security, without which economic recov- violent report that they are more severely ery is virtually impossible, the public sector constrained than their counterparts in more has a potentially important role to play in peaceful areas with regard to infrastructure, creating jobs in the early stages of a post- the regulatory environment, security, and conflict situation. In addition to employ- skills (�gure 1.29). ment through DDR programs, well-designed FIGURE 1.29 Severity of business environment constraints (average) reported by firms in low-conflict and high-conflict areas of Afghanistan, 2008 political instability electricity crime, theft, and disorder corruption telecoms access to land business licensing transport tax administration courts competition inadequately educated labor customs labor regulations 0 1.0 2.0 3.0 4.0 severity of constraint high-conflict areas low-conflict areas Source: Authors, based on data from the 2008 Afghanistan enterprise survey. Note: High-conflict provinces are defined as provinces in which the number of deaths per 1,000 population caused by terrorist incidents in 2007 was greater than 0.1. By this definition, Baghlan, Ghazni, Farah, Helmand, Kandahar, Khost, Kunarha, Nimroz, Paktika, Paktya, Panjsher, Takhar, Urozgan, Wardak, and Zabul are high-conflict areas. Only differences that are statistically significant at at least the 5 percent in an ordinary least squares regression including firm size and industry fixed effects are shown. OVERVIEW 39 and implemented public works programs barriers to � rm entry and tackling at least that target rural areas and build or reha- the most blatant pockets of corruption bilitate community infrastructure can (World Bank 2011d). In addition, govern- play an important role. Training and self- ments could facilitate private sector activi- employment assistance programs are also ties by creating “safe economic zones� that important. provide the needed security, services, and Priority is often initially given to seg- infrastructure in a focused manner and pro- ments of the population that are particularly moting “resource corridors� in areas rich in vulnerable and sectors that have the poten- natural resources to better link them to the tial to absorb large numbers of workers. Ini- rest of the economy. Although much of the tially, labor market programs and policies in effort of job creation will initially be in agri- confl ict-affected areas need to target three culture and construction, the focus should types of populations with special needs: shift over time from low-skilled agricultural jobs to higher-productivity nonfarm jobs • Excombatants, who need to be integrated and from targeted programs to broad-based into the workforce and given incentives to job creation. refrain from violence The peace dividend for job creation is • At-risk youth and war victims (families potentially large. A striking example is the who lost members, people with physical reduction in unemployment in the Northern or mental disabilities, and households that and Eastern provinces of Sri Lanka during are extremely vulnerable as a result of the the 2002–04 cease� re (�gure 1.30). Unem- lack of a steady stream of income) ployment fell from 13.0 percent to 9.2 per- • People displaced by the confl ict, who may cent in the Northern Province and from wish to return to their homes but need 15.9 percent to 10.5 percent in the Eastern to be able to find jobs and feel secure Province, at a time when the national unem- there. ployment rate decreased from 8.8 percent to Labor market programs are important, 8.3 percent. but fiscal and capacity constraints will limit the potential for direct job creation by the public sector in confl ict environments. Both constraints are likely to be most severe FIGURE 1.30 Unemployment rates in the Northern and Eastern for nationwide confl icts. Thus the govern- provinces of Sri Lanka, 1997–2001 and 2002–04 ment is in a better position to implement 16 employment programs in Sri Lanka, where confl ict was localized, than in Afghanistan 14 or Nepal. International organizations and 12 foreign governments have key roles to play 10 percent in providing funding and building capac- 8 ity, particularly in cases of nationwide 6 confl ict. 4 Given resource constraints and the risks of the politicization of extended public sec- 2 tor involvement, policy makers need to take 0 ce ce de early steps to gradually make the private vin vin i nw ro ro sector a more signi�cant creator of jobs once tio nP nP na er r ste a minimum level of security is achieved. th Ea r No Improving the regulatory environment is important in this regard, but true institu- conflict years (1997–2001) ceasefire years (2002–04) tional transformation could take a genera- tion. Governments could begin by reducing Source: Annual Reports of the Central Bank of Sri Lanka. 40 MORE AND BETTER JOBS IN SOUTH ASIA Conclusion Asia—improved the well-being of workers in South Asia. Rising aggregate labor produc- This overview began by posing three ques- tivity has largely driven this growth. South tions. This section summarizes the answers Asia has seen the fastest growth in total fac- to those questions. tor productivity in the world over the last three decades. But not all countries in the Has South Asia been creating an region have enjoyed rapid growth. Some increasing number of jobs countries that had stagnant or slow growth and better jobs? experienced massive international out-migra- South Asia has created jobs at a rate that tion, which opened up job opportunities for broadly tracks the growth in the region’s those who remained behind. Together with working-age population. Indeed, the rank- a substantial inflow of workers’ remittances, ing among five of the larger countries in the tighter labor market has contributed to the region, in descending order of growth rising real wages and declining poverty. of employment (Pakistan, Nepal, Bangla- South Asia will have to accommodate desh, India, and Sri Lanka), coincides with 1.0–1.2 million new labor market entrants their ranking by growth of the working-age a month between 2010 and 2030—a 25–50 population. percent increase over the average number of But do workers have better jobs? This new entrants in 1990–2010. Finding jobs of book uses two primary measures (increases increasing quality for this massive number in real wages for casual laborers and regu- of new workers represents an enormous chal- lar wage or salaried earners and decreases lenge. With TFP growth likely to return to in poverty rates for the self-employed) and more typical rates for the region as a whole, one secondary measure (mitigation of the policy makers have to create incentives for risk of low and uncertain income arising physical capital deepening and human capital from lack of work for the most vulnerable formation. The relative contribution of these group of workers) to assess their well-being. factors to the growth of aggregate labor pro- Based on the primary criteria, South Asian ductivity will vary by country. workers are indeed better off than they were Faster reallocation of labor from agri- in the earliest period for which comparable culture to industry and services, where TFP data are available. Real wages have risen for growth is higher, can also raise aggregate wage workers, including both casual labor- TFP growth. Such a reallocation of workers ers and regular wage or salaried earners. will need physical capital accumulation (in, Among all groups of workers—rural and for example, electricity and transport) and urban workers, men and women—a higher investment in human capital. Reallocation proportion of the self-employed belong to across sectors needs to be complemented by households that are above the poverty line. moving labor out of low-productivity �rms in In India—the only country in which the manufacturing and services, where the bulk data permitted the secondary measure to of workers in these sectors are employed. be calculated—the risk of low and uncer- tain income arising from lack of work has What bottlenecks need to be eased to been mitigated over the last decade for the meet the employment challenge given most vulnerable segment of the labor force, intensifying demographic pressure? casual workers. Accommodating new entrants into the labor force at rising levels of productivity will What determines the quality of job require reforms to ease demand- and supply- creation, and what is the employment side bottlenecks to expanding employment. challenge going forward? Investing in electricity to ensure reliable Strong economic growth in the region since power supply is the most important and per- the 1980s—second only to that of East vasive reform. But the reform agenda is not OVERVIEW 41 only about investment. Improvements in the away from protecting jobs for a minority of regulatory framework and governance of the insiders toward protecting the vast majority sector must go hand in hand. Urban formal of workers in the informal sector who lack sector � rms consistently cite corruption in protection is essential. dealings with the state—in particular, utili- ties and tax administration—among the top A challenging, but feasible agenda constraints to their operations. The proposed reform agenda is challeng- Improving the quality of education is a key ing—but it is feasible, especially given the supply-side priority. Efforts to do so should resources that will be freed up over the start with interventions in early childhood to next three decades as a result of South improve nutritional status and prevent cogni- Asia’s demographic transition. Business as tive impairment before children get to school. usual will accommodate new entrants to Policies should focus on the quality of learn- the labor force, but doing so at stagnant ing at all levels once children enter the school or barely rising levels of productivity will system. Ensuring that graduates are equipped mean that the quality of employment will not only with speci�c knowledge and tech- be poor. The vicious circle of poverty and nical skills but also with the behavioral, low- productivity employment would then problem-solving, and creative thinking skills be drawn around another generation. This increasingly required in the world of work is the price of failure. It is a price that need is critical. Moving labor market institutions not be paid. 42 MORE AND BETTER JOBS IN SOUTH ASIA Annex 1A Summary statistics on South Asian countries TABLE 1A.1 Summary economic statistics of South Asian countries Statistic Afghanistan Bangladesh Bhutan India Maldives Nepal Pakistan Sri Lanka South Asia GDP per capita, 2009 (at 2005 purchasing power parity dollars) 879a 1,419 4,525 2,993 4,967 1,047 2,358 4,256 2,713 Gross capital formation (% of GDP) 25 24 54 36 53 30 19 25 33 Gross savings (% of GDP) — 39 — 35 35 38 22 24 34 Total population (millions) 30 162 1 1,155 0.3 29 170 20 1,568 Sector share of GDP (%) 1980 Agriculture — 32 44 36 — 62 30 28 35 Industry — 21 15 25 — 12 25 30 24 Manufacturing — 14 4 17 — 4 16 18 16 Services — 48 42 40 — 26 46 43 41 2008 Agriculture 32 19 19 18 6 34 20 13 18 Industry 26 29 46 29 18 17 27 29 29 Manufacturing 16 18 7 16 7 7 20 18 16 Services 42 53 35 54 76 50 53 57 54 Sector share of employment (%) Earliest year available in national labor force surveys — 2002/03 2003 1983 1998 1998/99 1999/2000 2000 Agriculture — 51 80 63 25 77 47 37 60 Industry — 14 3 16 25 10 19 25 16 Manufacturing — 10 1 12 15 6 12 17 11 Services — 35 18 21 49 13 34 38 24 Latest year available in national labor force surveys 2007/08 2005/06 2007 2007/08 2004 2007/08 2008/09 2008 Agriculture 59 47 68 53 17 73 43 31 52 Industry 13 15 7 20 27 11 21 27 20 Manufacturing 5 11 4 12 20 7 13 19 12 Services 29 38 24 26 55 16 36 42 28 Sources: Authors, based on data from Aggarwal 2010; ILO KILM and LABORSTA databases 2010; World Bank 2011c; and national labor force surveys. Note: Totals may sum to less than 100 percent because of employment in unknown or unclassifiable categories. Employment and GDP shares for 2008 are based on most recent year available. — = Not available. a. GDP per capita 2008. OVERVIEW 43 Annex 1B Definition of key labor market terms TABLE 1B.1 Definitions of key labor market terms used in this book Term Definition Employed Persons who worked during at least part of the reference period (typically the last seven days), regardless of whether employment was formal or informal, paid or unpaid. Reference period in Bangladesh, Bhutan, India, Nepal, Pakistan, and Sri Lanka was reference week (generally past seven days). Reference period in Afghanistan and Maldives was past month. Unemployed Persons who did not work in the reference period but actively sought work. Inactive Persons who were neither employed nor unemployed during the reference period. (nonparticipant in labor force) This category includes discouraged workers—people who left the labor force because they believed no jobs were available or did not know how to search. Regular wage or salaried workers Persons who receive regular wages or salary from a job in the public or private sector. These workers are usually on the payroll and usually earn leave and supplementary bene�ts. A signi�cant proportion of these workers are in the public sector (ranging from 27 percent in India to 67 percent in Afghanistan). Casual laborers Persons who are paid on a casual, daily, irregular, or piece-rate basis. These workers typically do not have access to formal instruments of social protection. In rural areas, this category includes landless agricultural workers as well as workers in rural-based industry and services, such as construction. Self-employed Employers, own-account workers, and unpaid family enterprise workers. Except in Maldives and Sri Lanka, this is the largest group of workers in South Asia, where the majority of people work as own-account or family enterprise workers. In rural areas, this category comprises largely farmers working their own land, although many self- employed workers also work in the rural nonfarm sector. High-end self-employed Employers in all occupations and own-account workers and unpaid family workers working as managers, professionals, technicians, and clerks. Low-end self-employed Own-account workers and unpaid family workers working as service workers, skilled agricultural workers, craftspeople, machine operators, and workers in elementary occupations. Source: Authors. Annex 1C What is a “better� job, and which jobs are “better�? Two main criteria are used to assess job qual- of stable employment and the associated vari- ity. The primary criterion is higher average ation in income and consumption are of con- earnings. For wage workers, earnings can be cern for casual wage workers, who are typi- assessed using information on average wages. cally the poorest segment of the labor force. This information is not available on the self- Data limitations in all countries except India employed, whose earnings are in the form of precluded a consistent analysis of this second- returns to both labor and capital. As these ary criterion. The primary criterion for better �gures are not available, poverty rates (the jobs thus guides most of this book. percentage of workers living in households Various additional dimensions of job qual- below the poverty line) are used as a proxy ity are often cited, including access to non- for job quality for this segment of the labor wage bene�ts, access to public social protec- force. Better jobs are thus those associated tion mechanisms, the ability to upgrade skills with higher (average) wage rates and lower and receive training on the job, and the pres- poverty rates. ence of a safe working environment. These The second criterion of job quality looks factors are strongly correlated with wages, beyond average income to its variability. Lack poverty, and job security. 44 MORE AND BETTER JOBS IN SOUTH ASIA Based on the criteria of average wages or workers reported being out of work for poverty rates and the risk of low and uncer- 0.9–1.4 months the previous year. In con- tain incomes, several observations can be trast, the self-employed reported 0.2–0.7 made about where the “better� jobs are in months out of work, and regular wage South Asia (�gures 1C.1 and 1C.2): or salaried workers reported virtually no time out of work. • Poverty rates are highest among casual • Among the casual labor force, workers wage workers and lowest among regular in the agricultural sector have the lowest wage or salaried workers. Within the rural average earnings and the highest poverty non-farm sector and in urban areas, regu- rates. Casual workers in rural-based indus- lar wage workers earn 23–59 percent more try and services (the rural nonfarm sector) than casual workers in the four countries— earn 10–50 percent more than casual work- Bangladesh, India, Nepal, and Pakistan— ers in agriculture, even though the skills where these comparisons can be made. In pro� les are broadly similar. Urban casual Bangladesh (with the exception of its rural workers earn up to 20–30 percent more non-farm sector), India, and Nepal, regu- than casual agricultural workers in Nepal lar wage or salaried workers have poverty and Pakistan. In India, the risk of uncer- rates that are just a third or less than those tain income arising from inability to � nd of casual workers. The poverty rates for work is also highest for agricultural casual the self-employed are typically between laborers, 49 percent of whom spent at least those of casual and regular wage workers. one month without work in the previous • In India and Nepal—the two countries for year, with an average of 1.4 months spent which data were available—casual labor is without work. In contrast, 40 percent of also associated with the least stability and rural nonfarm casual labor reported being regular wage work with the most stability. without work for at least 1 month the In India in 2009/10, for example, casual previous year, with an average of 1.1 FIGURE 1C.1 Percentage of workers in households below the poverty line in Bangladesh, India, and Nepal, by employment status a. Bangladesh, 2010 b. India, 2004/05 c. Nepal, 2003/04 agricultural casual 51 urban casual labor 45 agricultural casual labor 50 labor rural nonfarm casual agricultural casual rural nonfarm casual labor 45 31 43 labor labor urban casual labor rural nonfarm casual agricultural 44 25 31 labor self-employed rural nonfarm regular wage rural nonfarm or salaried worker 26 urban self-employed 24 17 self-employed rural nonfarm rural nonfarm self-employed 26 18 urban casual labor 16 self-employed agricultural agricultural rural nonfarm regular self-employed 22 17 10 self-employed wage or salaried worker urban regular wage urban regular wage or or salaried worker 16 14 urban self-employed 9 salaried worker rural nonfarm regular urban regular wage or urban self-employed 14 8 3 wage or salaried worker salaried worker 0 50 0 50 0 50 percent percent percent Source: Authors, based on data from national labor force and household surveys. Note: Figures are for workers age 15–64. Poverty rates for India are based on official poverty lines prevailing until 2010. Using the new official poverty lines for 2004/05 (revised in 2011) would increase poverty rates in rural areas, making the poverty rates of rural workers higher than those of urban workers for the same employment type. The hierarchy in terms of employment type would remain the same. OVERVIEW 45 FIGURE 1C.2 Ratio of rural nonfarm and urban wages to agricultural wages in Bangladesh, India, and Nepal Bangladesh, 2005/06 India, 2009/10 Nepal, 2007/08 agricultural casual agricultural casual agricultural casual 1 1 1 labor labor labor urban urban urban 1.0 1.0 1.3 casual labor casual labor casual labor rural nonfarm casual rural nonfarm casual rural nonfarm casual 1.1 1.3 1.5 labor labor labor urban regular wage urban regular wage urban regular wage 1.3 1.6 1.8 or salaried worker or salaried worker or salaried worker rural nonfarm regular rural nonfarm regular rural nonfarm regular 1.6 1.8 2.0 wage or salaried worker wage or salaried worker wage or salaried worker 0 1 2 0 1 2 0 1 2 Source: Authors, based on data from national labor force and household surveys. Note: Figures are median wages for latest year available. months spent without work. This risk is lower still for urban casual workers, 31 Notes percent of whom reported being without 1. See annex table 1A.1 for summary statistics work for at least one month, with an aver- on the eight South Asian countries age of 0.9 months spent without work. 2. The absolute poor live on less than $1.25 a • Wage differentials between sectors and day in 2005 purchasing power parity dollars. 3. See annex table 1B.1 for a de�nition of key employment types partly reflect educa- terms. Srinivasan (2010) provides a more tional attainment. Workers in indus- detailed account. try and services are more educated than 4. It is not possible to rule out reasons other than workers in agriculture. Regular wage or improving job quality for falling poverty rates salaried workers are more educated than in these households. Such factors include the casual workers. Almost all have some edu- flow of workers’ remittances in Nepal, which cation, and a signi�cant proportion have bring in close to a quarter of gross domestic secondary education or above. The self- product and are estimated to account for half employed are more educated than casual of the decline in national poverty rates, or an workers. Even after accounting for higher increase in hours worked by household mem- skills, however, the majority of industry bers in situations of low wage growth. 5. Young (1994, 1995) notes the dominance of and service jobs still pay more than casual physical and human capital accumulation jobs in the agricultural sector. compared with TFP growth in East Asia. Rodrik and Subramanian (2005) note the In summary, less desirable jobs are found in overwhelming importance of TFP growth in casual employment, with the most precarious India during 1980–99. and lowest-paying jobs held by agricultural 6. These data update the point made in Easterly casual workers. Self-employment is in the and others (1993). middle, with high-end self-employed work- 7. Bosworth, Collins, and Virmani (2007) ers having consumption pro�les and poverty emphasize this point in the context of India. rates closer to regular wage workers and low- 8. Reallocation contributed only 5 percent to end self-employed having pro�les and poverty TFP growth in China between 1993 and 2004 because of an extraordinarily high rate rates that are closer to those of casual labor- of within-sector TFP growth (more than 6 ers. “Better jobs� are held by regular wage or percent a year in industry). salaried workers in industry and services. 46 MORE AND BETTER JOBS IN SOUTH ASIA 9. The discussion of India’s manufacturing sec- India.� Policy Research Working Paper, World tor uses the terms establishments and �rms Bank, Washington, DC. interchangeably. Bloom, D., and D. Canning. 2008. “Global Demo- 10. The discussion draws on Bloom, Canning, graphic Change: Dimensions and Economic and Rosenberg (2011), which is based on an Signi�cance.� Population and Development analysis of the contribution of demographic Review 33: 17–51. change to economic growth at the global Bloom, D., D. Canning, and L. Rosenberg. level in Bloom and Canning (2008). 2011. “Demographic Change and Economic 11. Enterprise surveys provide information only Growth in South Asia.� Working Paper 67, about the constraints facing existing enter- Harvard School of Public Health Program prises; they are not useful in understanding on t he Globa l Demog raphy of Ag i ng, constraints perceived by potential �rms that Cambridge, MA did not enter in the �rst place. Bosworth, B. 2005. Economic Growth in Thai- 12. The question on political instability was not land: The Macroeconomic Context. Washing- asked in India and Sri Lanka. Tax rates and ton, DC: Brookings Institution. access to �nance constraints are excluded. ———. 2010. “Update of Bosworth and Collins 13. Only in Pakistan do the rankings by bench- 2003.� Unpublished paper, World Bank, mark and expanding �rms differ: for expand- Washington, DC. ing �rms corruption and political instability Bosworth, B., and S. Collins. 2003. “The Empir- are the top constraints. ics of Growth: An update.� Brookings Papers 14. The samples for the investment climate on Economic Activity 2: 113–206. assessments of rural �rms are pooled because ———. 2008. “Accounting for Growth: Com- of the limited sample size (500 �rms) in each paring China and India.� Journal of Economic country. Perspectives 22 (1): 45–66. 15. The importance of urban-rural links for the Bosworth, B., S. Collins, and A. Virmani. 2007. nonfarm economy has been discussed in “Sources of Growth in the Indian Economy.� other studies. The World Bank’s India pov- NBER Working Paper W1290, National erty assessment (2011b) �nds that that the Bureau of Economic Research, Cambridge, expansion of the nonfarm sector is more MA. http://www.nber.org/papers/w12901. closely linked to urban than to agricultural Carlin, W., and M. Schaffer. 2011a. “A Com- growth. This �nding is also con�rmed by a parison of Business Environment Constraints simple multivariate regression analysis using between Formal Sector Firms and Rural and census data from Nepal. Informal Sector Firms.� Background study 16. To facilitate comparison, �rms with more conducted for this book. than 20 employees were dropped from the ———. 2011b. “Which Elements of the Business sample of formal �rms. Environment Matter Most for Firms and How 17. Informal �rms also cite “competition� as a Do They Vary across Countries?� Background severe constraint. The survey does not make study conducted for this book. clear what �rms mean by competition. Central Bank of Sri Lanka. Various years. Annual Reports. Colombo. Collier, P., and N. Sambanis. 2005. Under- References standing Civil War: Evidence and Analysis. Washington, DC: World Bank. Aggarwal, S. 2010. “Labor Input and Its Com- Easterly, W., M. Kremer, L. Pritchett, and L. position: An Industry-Level Perspective.� Summers. 1993. “Good Policy or Good Luck? Paper presented at the Worldklems conference, Country Growth Performance and Temporary Cambridge, MA, August. Shocks.� Journal of Monetary Economics 32 Alderman, H. 2011. “Early Childhood Develop- (3): 459–83. ment and the Role of Pre-school.� Background Fan, S., A. Gulati, and S. Thorat. 2008. “Invest- paper prepared for Regional Quality of Educa- ment, Subsidies, and Pro-Poor Growth in tion Study, South Asia Region, Human Devel- Rural India.� Agricultural Economics 39 (2): opment Unit, World Bank, Washington, DC. 163–70. Blom, A., and H. Saeki. 2011. “Employability Gleditsch, N., P. Wallensteen, M. Eriksson, M. and Skill Set of Newly Graduated Engineers in Sollenberg and H. Strand. 2002. “Armed Con- OVERVIEW 47 fl ict 1946–2001: A New Dataset.� Journal of UN (United Nations). 2010. World Population Peace Research 39 (5): 615–37. Prospects: The 2010 Revision. New York. Harbom, L., and P. Wallensteen 2009. “Armed World Bank. 2008a. Bangladesh Investment Conflict, 1946 –2008.� Journal of Peace Climate Assessment. Washington, DC. Research 46 (4): 577–87. ———. 2008b. The Growth Report: Strategies Hazell, P., D. Headey, A. Pratt, and D. Byerlee. for Sustained Growth and Inclusive Develop- 2011. “Structural Imbalances and Farm Non- ment. Report of the Commission on Growth Farm Employment Prospects in Rural South and Development. Washington, DC. Asia.� Background study conducted for this ———. 2010. Large-Scale Migration and Remit- book. tances in Nepal: Issues, Challenges and Holzmann, R., Y. Pouget, M, Vodopivec, and Opportunities. Washington, DC. M. Weber. 2011. Severance Pay Programs ———. 2011a. India Urbanization Review. around the World: History, Rationale, Status, Washington, DC. and Reforms. Social Protection Discussion ———. 2011b. Perspectives on Poverty in India: Paper 62726, World Bank, Washington, DC. Stylized Facts from Survey Data. Washington, ILO (International Labour Organization). 2010. DC KILM and LABORSTA databases. Geneva. ———. 2011c. World Development Indicators. OECD (Organisation for Economic Co-operation Washington, DC. and Development). 2009. Indicators of ———. 2011d. World Development Report 2011: Employment Protection. Paris: OECD. Security and Development. Washington, Rodrik, D., and A. Subramanian. 2005. From DC. “Hindu Growth� to Productivity Surge: The ———. 2012. World Development Report 2012: Mystery of the Indian Growth Transition. IMF Gender Equality and Development. Washing- Staff Papers 52 (2): 193–228. ton, DC. Sri Lanka Information Communication Technol- Young, A. 1994. “Lessons from the East Asian ogy Association. 2007. Rising Demand: The NICs: A Contrarian View.� European Eco- Increasing Demand for IT Workers Spells a nomic Review 38 (3–4): 964–73. Challenging Opportunity for the IT Indus- ———. 1995. “The Tyranny of Numbers: Con- try. Colombo. fronting the Statistical Realities of the East Srinivasan, T. N. 2011. “The Utilization of Labor Asian Growth Experience.� Quarterly Journal in South Asia.� Background study conducted of Economics 110 (3): 641–80. for this book. CHAPTER 2 Growth and Job Quality in South Asia Questions and Findings Questions • Labor productivity growth since 1980 owes more to growth in total factor productiv- • What is South Asia’s recent track record with ity than to accumulation of physical and regard to the quantity and quality of job human capital, reflecting the region’s open- creation? ing up to the world economy and deregula- • What needs to be done to improve the qual- tion. Going forward, creating an enabling ity of jobs in the face of intensifying demo- framework for physical and human capital graphic pressure? accumulation to occur will be important. • Reallocation of workers across sectors has Findings played a comparatively modest role in total • Rapid growth in aggregate output per worker factor productivity growth in South Asia. in much of South Asia since 1980 has been Labor will need to be reallocated more rap- associated with rising real wages for casual idly, not only from agriculture to industry and labor and regular wage or salaried workers services but also from less productive to more and an increase in the proportion of the self- productive units within industry and services. employed above the poverty line (a proxy for Doing so will require investment in physical improved job quality). Larger shares of casual and human capital. workers and regular wage or salaried earners • Much of South Asia is going through the now also belong to households above the pov- demographic transition, where the number of erty line. workers is growing more rapidly than their • Rising real wages and declining poverty are dependents. The resources saved as a result of the primary criteria used to assess job qual- there being fewer dependents to support—the ity. By these measures, jobs improved for all demographic dividend—can be channeled into three types of workers. A secondary criterion high-priority investments, which can raise the is the reduction in the risk of low and uncer- productivity of the larger number of entrants tain incomes for the most vulnerable workers. into the labor force. But only if there is an Data limitations allowed this criterion to be enabling policy framework for doing so. monitored only in India, where it is satis�ed. • Continuance of high growth is not assured: • Large-scale out-migration in countries, where globally, correlations of country growth growth has been slow, has exerted upward rates across decades are low. Structural pressure on real wages, thus bene�ting work- reforms to ease demand- and supply-side ers who remain. bottlenecks to expanding employment are • Employment growth has broadly tracked needed, irrespective of whether there is a the growth of the working-age population, dividend, in order to maintain and improve creating just under 800,000 new jobs a the pace of creation of better jobs even in month between 2000 and 2010. A projected lower-growth environments. But the pros- 1.0–1.2 million entrants will enter the labor pect of reaping the demographic dividend, force every month for the next two decades. which will be available only for the next The employment challenge is to absorb them three decades, lends urgency to the need for at rapidly rising levels of productivity. reform. Growth and Job Quality in South Asia 2 T his chapter looks at the growth con- Growth in per capita GDP has accelerated, text in South Asia in which labor particularly since the 1980s, in Bangladesh market outcomes are embedded. The and India. Bhutan saw generally high growth �rst section decomposes growth in aggregate starting in the 1980s, albeit with some fluc- gross domestic product (GDP) per worker (or tuations. Maldives also enjoyed high growth, aggregate labor productivity) during the past although it experienced a deceleration three decades into the contributions of phys- between the 1990s and the first decade of ical and human capital accumulation and the 21st century. Per capita growth has been changes in total factor productivity (TFP). marked by volatility around a broadly declin- The second section explores how sources of ing trend since the 1980s in Pakistan and has growth may be different in the future. The stagnated in Nepal. Sri Lanka has witnessed third section examines South Asia’s track an acceleration of growth over the last �ve record regarding the number and quality decades, except for a dip in the 1980s, avoid- of jobs created. The last section argues for ing the slowdown or stagnation of the 1970s moving ahead quickly with reforms in order that affected the rest of the region. to absorb the rapidly growing number of new entrants to the labor market at rising levels of labor productivity even in situations Aggregate labor productivity growth of lower economic growth. Growth in aggregate output per worker, or aggregate labor productivity, may be decom- Economic growth posed into two factors:2 in South Asia • “Extensive� growth, comprising growth Improving job quality for most segments of in physical capital per worker (capital the labor force can usually occur only in a deepening) and growth of human capital growing economy. South Asia has seen an per worker (education) acceleration of growth over the three decades • “Intensive� growth, comprising growth since 1980 that is second only to that of East in TFP—a combination of changes in the Asia (figure 2.1).1 But growth experiences ef�ciency with which inputs are used and have varied within South Asia (�gure 2.2). changes in technology. 49 50 MORE AND BETTER JOBS IN SOUTH ASIA FIGURE 2.1 Annual growth in GDP per capita, by region, 1960s–2000s 10 8.4 8 6.7 6.0 5.9 6 4.7 4.5 percent 3.8 4.0 3.7 4 2.8 2.7 3.1 2.8 3.0 2.5 2.3 2.5 2 1.6 1.2 1.4 0.5 0.1 0 –0.2 –0.4 –0.9 –2 1961–70 1971–80 1981–90 1991–2000 2001–10 East Asia and Paci�c Latin America and the Caribbean Middle East and North Africa Sub-Saharan Africa South Asia Source: Authors, based on data from World Bank 2011c. FIGURE 2.2 Annual growth in GDP per capita in South Asia, by country, 1960s–2000s 9 8 7.9 7 6.6 6.6 6.3 6.2 6 5 4.8 4.9 4.6 4.5 4.3 percent 4.3 4.2 4.3 4 3.3 3.4 3 2.9 2.8 2.8 2.2 2.4 2.2 2.1 1.9 2 1.8 1.0 1.2 1.2 1.1 1 0.4 0.5 0 0.0 –1 1961–70 1971–80 1981–90 1991–2000 2001–10 Afghanistan Bangladesh Bhutan India Maldives Nepal Pakistan Sri Lanka Source: Authors, based on data from World Bank 2011c. Note: Growth in the earliest available decade for Afghanistan, Bhutan, and Maldives is not based on data for the entire decade because data for the entire decade were not available. Thus, Afghanistan 2001–10 is based on 2003–09, Bhutan 1981–90 is based on 1982–90, and Maldives 1991–2000 is based on 1996–2000. Growth in aggregate labor productivity in the growth of aggregate output per worker South Asia between 1980 and 2008 bene�ted in South Asia and its sources in relation to from rapid growth in TFP, in contrast to the growth in other regions. 3 Figure 2.4 shows 1960 –80 period, when extensive growth the sources of growth for Bangladesh, India, accounted for the bulk of growth in aggre- Pakistan, and Sri Lanka. Together they help gate labor productivity. Figure 2.3 presents illustrate the following points: FIGURE 2.3 Sources of annual growth in labor productivity, by region, 1960–80 and 1980–2008 9 8 7 6 4.8 5 4 percent 1.3 1.0 3 0.4 0.6 1.1 2.0 0.4 1.3 1.4 2 0.5 1.3 1.6 0.3 0.4 0.4 0.8 1.1 0.4 0.5 2.6 2.9 0.3 0.3 0.4 0.4 2.5 0.5 1 0.3 1.8 0.2 1.0 1.3 1.3 1.2 0.5 0.9 0.8 0.9 0.9 0.5 0.9 0.1 0.4 0.5 0 –0.2 –0.7 –0.7 –1 –2 0 8 0 8 0 8 0 8 0 8 0 8 0 8 0 8 –8 00 –8 00 –8 00 –8 00 –8 00 –8 00 –8 00 –8 00 –2 –2 –2 –2 –2 –2 –2 –2 60 60 60 60 60 60 60 60 19 80 19 80 19 80 19 80 19 80 19 80 19 80 19 80 19 19 19 19 19 19 19 19 South Asia world industrial East Asia less China Latin America Middle East Sub-Saharan countries China Africa physical capital per worker education per worker total factor productivity Source: Bosworth 2010. 51 52 MORE AND BETTER JOBS IN SOUTH ASIA FIGURE 2.4 Sources of annual growth in labor productivity in selected countries in South Asia, by country, 1960–80 and 1980–2008 5 4 2.6 3 0.5 0.2 1.2 percent 2 1.4 0.7 0.4 0.3 0.1 1.3 0.4 0.3 2.4 0.3 1 1.4 1.6 0.4 0.9 1.0 1.0 0.2 0.3 0.4 0 –0.3 –1 1960–80 1980–2008 1960–80 1980–2008 1960–80 1980–2008 1960–80 1980–2008 Bangladesh India Pakistan Sri Lanka physical capital per worker education per worker total factor productivity Source: Bosworth 2010. • Growth in aggregate labor productivity lowest among all developing and industrial in South Asia during 1980–2008, which regions, including Sub-Saharan Africa. averaged nearly 3.7 percent a year, was During 1980–2008, TFP growth accounted well above the world average and second for more than half of aggregate labor pro- only to that witnessed in China. However, ductivity growth in South Asia. Its contri- excluding 1997 and 1998, the years when bution ranged from just over a third in Ban- East Asia experienced a � nancial crisis, gladesh and Sri Lanka (a range comparable growth in aggregate output per worker to that in the high-performing East Asian in East Asia less China was higher than economies during 1960–96) to 50–60 per- in South Asia during 1980–96 and lower cent in India and Pakistan. The transforma- during 1999–2008. Productivity growth tion of the role of TFP growth is consistent varied within South Asia during 1980– with the picture of a region responding to 2008, when it ranged from 2 percent in improved policies that exposed it to greater Bangladesh to nearly 4.5 percent in India. internal and external competition.5 • TFP growth in South Asia—which aver- • Capital deepening was 2.5 times higher in aged about 2 percent a year over 1980– East Asia less China than in South Asia dur- 2008, 2.5 times the world average—was ing 1960–80; it was only about 40 percent second only to China’s and nearly twice higher during 1980–2008, when capital that of East Asia less China.4 This rapid deepening accounted for about 35 percent growth represented a striking turnaround of the growth in aggregate labor productiv- from the situation during 1960–80, when ity. The contribution of capital deepening virtually all growth in aggregate output per to labor productivity growth was about worker in South Asia was the result of fac- a third in India and Pakistan, more than tor accumulation and TFP growth was the 40 percent in Bangladesh, and more than GROWTH AND JOB QUALITY IN SOUTH ASIA 53 50 percent in Sri Lanka. The magnitude accelerating TFP growth, which was second of capital deepening in South Asia was 1.5 only to China. Although factor accumulation times the world average during this period. played a less prominent role than in the years • The difference in the contribution of edu- of rapid investment-led growth of East Asia cation in South Asia and in East Asia less China, both capital deepening and edu- less China steadily narrowed between cation were increasingly important sources 1980 and 2008, falling from a factor of of growth in India decade by decade over 2 during 1960–80 to a factor of 1.5 dur- 1980–2008. ing 1980–90 and about 1.3 during 1990– 2000 and 2000– 08. The magnitude of growth in education in South Asia was Demographic transition comparable to the world average during Almost all South Asian countries are expe- 1980 –2008. Within the region, educa- riencing a demographic transition—the tion accounted for 10 percent of growth process by which high fertility and mortal- in aggregate labor productivity in South ity rates are replaced by low ones. 6 A key Asia during 1980 –2008, ranging from indicator of where a country is situated in the low double digits in India, Pakistan, the transition is captured by the inverse and Sri Lanka to more than 20 percent in dependency ratio, the ratio of the working- Bangladesh. age (15– 64) population to the dependent In summary, rising aggregate labor pro- population (people under 15 and over 65) ductivity in South Asia owed a great deal to (�gure 2.5). Initially, the inverse dependency FIGURE 2.5 Ratio of working-age to nonworking-age population in South Asia, by country, 1960–2008 2.3 2.1 ratio of working-age population to 1.9 nonworking-age population 1.7 1.5 1.3 1.1 0.9 60 62 64 66 68 70 72 74 76 78 80 82 84 86 88 90 92 94 96 98 00 02 04 06 08 19 19 19 19 19 19 19 19 19 19 19 19 19 19 19 19 19 19 19 19 20 20 20 20 20 Afghanistan Bangladesh Bhutan India Maldives Nepal Pakistan Sri Lanka Source: Authors, based on data from UN 2008. 54 MORE AND BETTER JOBS IN SOUTH ASIA ratio decreases, because the infant mortal- • Potential demographic dividend coun- ity rate falls before the fertility rate. The tries: Bangladesh, Bhutan, India, and ratio subsequently rises, as the baby boom Maldives7 caused by the lagged decline in the fertility • Aging country: Sri Lanka. rate becomes part of the working-age popu- This classi�cation (used later in the chapter lation. The resulting rise in the share of the to project the numbers of entrants into South working-age to the nonworking-age popu- Asia’s labor markets in the coming decades) is lation means that there are fewer depen- chosen to reflect the following considerations. dents to support. The resources saved as a First, the demographic transition is over in result—the “demographic dividend�—can Sri Lanka. Second, with improved policies, be used for high-priority investments. Even- Bangladesh, Bhutan, India, and Maldives, tually, as the baby boom cohort ages, the which though growing rapidly, could bene�t demographic transition gives way to old-age yet more from the demographic dividend. dependency. Third, Nepal, where growth has been stag- Sri Lanka’s inverse dependency ratio nant, and Pakistan, where growth has been reached its peak around 2005. Since then it volatile around a broadly declining trend, has been declining, making it the only aging have yet to see a demographic dividend, while country in the region. Bangladesh’s ratio the demographic transition has barely begun shows a sharp increase since the mid-1980s, in Afghanistan. catching up with India’s in 2003 (the result, The resources made available by a demo- among other factors, of a very rapid decline graphic dividend can be used to deepen in fertility, which was supported by a repro- physical capital (for example by investing in ductive health program) and exceeding it electricity or transport infrastructure) as well thereafter. India’s inverse dependency ratio as human capital (by investment in educa- began to increase in the 1970s. Maldives saw tion and skills training). the fastest increase in the ratio, as a result But the realization of the dividend requires of its plunging fertility rate. In Bhutan, the a supportive policy framework, such as a inverse dependency ratio fluctuated, rising in � nancial sector that intermediates the addi- the mid-1970s and then falling through the tional savings effectively and a business envi- mid-1990s before rising sharply again. Turn- ronment that provides � rms with the incen- ing to countries with young populations, tives to make high-priority investments. Nepal’s ratio began to rise in the 1990s. Without policy reform, the demographic Pakistan’s ratio began a gentle climb in the dividend cannot be harnessed to productive 1980s. The inverse dependency ratio started ends. increasing in Afghanistan, the region’s most youthful country, only in 2005. In the medium-fertility scenario in the Sources of future growth United Nations’ population projections, the Looking forward, productivity growth in the ratio of the working-age to the nonworking- region will � rst need to rely more on factor age population in South Asia is expected to accumulation (physical capital deepening peak around 2040, except in Afghanistan, and human capital accumulation) and less where the ratio will still be increasing, and in on the extraordinary growth of TFP seen Sri Lanka, where it has already peaked. Thus in the last three decades. As the region has the demographic window of opportunity become more open to the global economy, will close after 2040 for most South Asian it is importing better-quality capital goods countries. and intermediate goods at world prices Trends in each country suggest a classi�ca- and using standard technology to pro- tion into three groups: duce goods that are sold domestically or • Young countries: Afghanistan, Nepal, and exported in competitive world markets. Inas- Pakistan much as the technology used is widely used GROWTH AND JOB QUALITY IN SOUTH ASIA 55 internationally, the increases in TFP arising employment by broad sector (agriculture, from exports will be limited to what is rou- industry, and services) have changed more tine in global best practice. For a country slowly than shares of value added in South such as India, which has a large domestic Asia. In 2008, India was an outlier in having market, domestic sales could lead to larger too large a share of workers in agriculture for increases in TFP as less competitive produc- its income level. Although the share of GDP ers exit the market. Even with acceleration provided by agriculture fell by almost half in “second-generation� structural reforms, between 1983 and 2008, the proportion of TFP growth, although still an important employment fell by only 20 percent. driver of long-run economic growth, is not The comparison of the shares of employ- likely to expand at the rates triggered by ment and GDP relative to average develop- the reforms of the 1990s. Hence, a key task ment experience is captured more formally by for policy makers is to create an improving comparing South Asian countries to a bench- enabling policy framework within which mark for market economies. The benchmark physical capital deepening and human capi- is derived by regressing the shares of employ- tal formation can take place. Such a frame- ment and GDP in each sector against per cap- work is needed to absorb the growing num- ita GDP, its square, and a measure of country ber of entrants into the labor force at rising size, represented by its land area, for nearly 55 productivity levels. industrial countries and emerging economies The transfer of underutilized labor from in 2008. Figure 2.7 shows the market econ- agriculture to the rest of the economy yields omy benchmarks for agriculture and services. reallocation-driven gains in TFP. The share The evolution of the benchmark is consistent of agriculture in employment in South Asia with the stylized facts in the development has generally fallen more slowly than its literature—namely, that as per capita income share in GDP (�gure 2.6). Indeed, shares of rises, the share of employment in agriculture FIGURE 2.6 Sectoral shares of GDP and employment in selected countries in South Asia, 1980s–2008 100 90 19 24 25 27 29 80 37 41 3 35 39 47 46 44 2 52 11 54 6 53 57 70 6 9 7 60 4 12 15 7 7 percent 11 9 12 50 7 9 13 12 11 17 7 19 40 14 13 16 16 11 59 67 30 18 54 20 48 16 53 46 20 45 18 31 34 30 33 28 10 19 17 20 13 0 GDP Emp GDP Emp GDP Emp GDP Emp GDP Emp GDP Emp GDP Emp GDP Emp 1984 2008 1983 2008 1980 2008 1981 2008 Bangladesh India Pakistan Sri Lanka services nonmanufacturing industry manufacturing agriculture Sources: Authors, based on data from ILO 2010; World Bank 2011c; and India National Sample Survey. Note: Emp = employment. 56 MORE AND BETTER JOBS IN SOUTH ASIA FIGURE 2.7 Shares of agriculture and services in employment and GDP in South Asian and comparator countries, 2008 a. Share of employment in agriculture, 2008 b. Share of GDP in agriculture, 2008 100 50 80 40 IN - 1983 NP - 2008 NP - 2008 IN - 1983 60 BGD - 1984 30 BGD - 1984 PK - 1980 IN - 2008 percent percent PK - 1980 SL - 1981 BGD - 2008 PK - 2008 Thailand 40 SL - 1981 20 PK - 2008 Philippines BGD - 2008 SL - 2008 IN - 2008 Philippines Thailand 20 Brazil 10 SL - 2008 Malaysia Malaysia Brazil Korea, Rep. Korea, Rep. United States United States 0 Singapore United Kingdom 0 United Kingdom Singapore .5 5 10 15 20 25 30 40 50 .5 5 10 15 20 25 30 40 50 2008 GDP per capita in 2005 purchasing power parity dollars (thousands) 2008 GDP per capita in 2005 purchasing power parity dollars (thousands) c. Share of employment in services, 2008 d. Share of GDP in services, 2008 100 100 United States United States United Kingdom 80 80 United Kingdom Singapore Singapore Korea, Rep. Brazil 60 Philippines Korea, Rep. BGD - 2008 IN - 2008 SL - 2008 percent percent 60 Brazil Malaysia NP - 2008 PK - 2008 Philippines BGD - 1984 SL - 1981 Thailand Malaysia 40 PK - 1980 IN - 1983 40 BGD - 2008 SL - 2008 Thailand PK - 2008 BGD - 1984 SL - 1981 20 PK - 1980 IN - 2008 20 NP - 2008 IN - 1983 0 .5 5 10 15 20 25 30 40 50 .5 5 10 15 20 25 30 40 50 2008 GDP per capita in 2005 purchasing power parity dollars (thousands) 2008 GDP per capita in 2005 purchasing power parity dollars (thousands) Sources: Authors, based on data from Aggarwal 2010; ILO 2010; World Bank 2011c; and India National Sample Survey. Notes: The cross-country regression lines shown are shares of employment and GDP by sector regressed on the log of 2008 GDP per capita and the log of 2008 GDP per capita squared (in 2005 purchasing power parity dollars). Figure excludes the transition economies of Europe and Central Asia. The horizontal axis is on a log scale. declines and the share of employment in ser- share of services in employment was below vices grows and that the share of employment the benchmark by nearly 10 percentage in industry rises and subsequently declines (as points in Sri Lanka, but, unlike in India, the workers move into services). The straight lines share of GDP in services was significantly in �gure 2.7 direct attention to the change in above the benchmark, indicating much the share of employment and value added in higher output per worker in services. The South Asian countries from 1980 (or the �rst share of employment in industry was signi�- available year of employment data by sector cantly above the benchmark in Sri Lanka in subsequent to it) through 2008. 2008, but the share regressions for industry The regressions show that the share of �t the data much less well than those for agri- agriculture in employment in India was 14 culture and services. The share of agriculture percentage points above the benchmark in in GDP in Bangladesh—a predominantly 2008, reflecting among other things the lim- agrarian economy at the time of its creation ited absorption of labor in the nonagriculture in 1971—was already below the benchmark economy compared with countries at India’s by 14 percentage points in 2008 (see annex level of per capita income (see column 1 in 2C for full regression results). annex 2C, table 2C.1). The share of services Reallocation across sectors has played a in employment in India was 15 percentage more limited role in boosting TFP in the two points below the benchmark in 2008. The largest countries of South Asia than it did GROWTH AND JOB QUALITY IN SOUTH ASIA 57 in some East Asian countries. 8 Figure 2.8 FIGURE 2.8 Sources of annual growth in total factor productivity in presents the results of a decomposition of India and Pakistan, by sector and reallocation effects, 1980–2008 TFP growth into within- and between-sector contributions over 1980–2008 for India and Pakistan, the two countries in South Asia for a. India 4 which data on capital stocks by broad sector allow growth accounting to be conducted 0.5 for agriculture, industry, and services. On 3 average, reallocation contributed 20 per- cent to aggregate TFP growth in India and 0.5 percent 15 percent in Pakistan during the period. The 2 0.4 0.3 2.2 share of employment in agriculture fell from 1.3 67 percent in 1983 to 54 percent in 2008 in 1.1 1.2 India and from 53 percent in 1980 to 43 per- 1 cent in 2008 in Pakistan. 0.1 0.3 0.6 0.3 Reallocation was considerably more 0.6 0.4 0.5 0.3 important in East Asian countries such as 0 China and Thailand (�gure 2.9).9 In China, 1980–90 1990–2000 2000–08 1980–2008 the contribution of reallocation to the growth reallocation services industry agriculture of total factor productivity was nearly 30 per- cent between 1978, when reforms started, and b. Pakistan 1993. The share of agriculture in employment 4 fell by more than a �fth, from 71 percent to 56 percent, over this period. (Reallocation contributed a mere 5 percent to TFP growth 3 in China between 1993 and 2004, because of the extraordinarily high rate of within-sector percent TFP growth—averaging more than 6 percent 2 a year—in industry.) The share of agriculture 0.1 in employment fell by nearly a �fth, from 56 0.5 0.2 0.2 0.4 percent in 1993 to 47 percent in 2004. Real- 1 0.2 0.3 0.9 0.3 0.4 location amounted to two-thirds of aggre- 0.6 gate TFP growth in Thailand between 1977 0.6 0.4 0.2 0.3 0 0.1 and 1996, a period during which the share 1980–90 1990–2000 2000–08 1980–2008 of agriculture in employment fell by nearly a third, from 65 percent to 45 percent. reallocation services industry agriculture A comparison between China, India, and Pakistan reveals several patterns. Whereas Source: Authors, based on data from Bosworth 2010. the declining share of reallocation in China Note: The contribution of reallocation during a decade is calculated as aggregate TFP growth minus the across the two subperiods studied reflects sum over the three sectors of TFP growth weighted by the share of the sector in GDP at the beginning of the decade. a steep rise in within-sector TFP growth in industry, the increasing share of reallocation in Pakistan since 1980 reflects an across-the- in GDP in South Asia, output per worker board decline in TFP growth. In India, the in industry and services was three to eight share of reallocation falls, but, in contrast times as large as in agriculture in 2008. to Pakistan, it does so as a result of increas- This is evident from table 2.1, which shows ing, rather than declining, within-sector TFP large differences in output per worker across growth and, in contrast to China, in services the three sectors, with agriculture the low- rather than industry. est and, except in Bangladesh, services the As a result of the slower evolution in highest. The differences in sectoral output the shares of employment than in changes per worker are particularly marked in India. 58 MORE AND BETTER JOBS IN SOUTH ASIA FIGURE 2.9 Sources of annual growth in total factor productivity in China, India, Pakistan, and Thailand, by sector and reallocation effects 4 0.2 0.3 3 1.0 0.5 percent 0.6 2 3.1 1.3 0.2 1.4 1 0.4 1.1 0.6 0.3 0.2 0.3 0.1 0.5 0.5 0.2 0.3 0 Pakistan Thailand India China China 1980–2008 1977–96 1980–2008 1978–93 1993–2004 reallocation services industry agriculture Sources: Authors, based on data from Bosworth 2005, 2010; Bosworth and Collins 2008. Note: The contribution of reallocation during a decade is calculated as aggregate TFP growth minus the sum over the three sectors of TFP growth weighted by the share of the sector in GDP at the beginning of the decade. TABLE 2.1 Labor productivity in South Asia and East Asia, by sector, 2008 (per worker, in 2005 purchasing power parity dollars) Total Agriculture Industry Services Ratio Ratio Region/country (1) (2) (3) (4) (3)/(2) (4)/(2) South Asia Bangladesh 4,116 1,754 6,876 5,635 3.9 3.2 India 7,049 2,202 10,368 14,939 4.7 6.8 Nepal 2,577 1,125 3,861 8,691 3.4 7.7 Pakistan 8,287 3,778 11,097 12,430 2.9 3.3 Sri Lanka 12,842 5,257 14,334 17,928 2.7 3.4 East Asia Korea, Rep. of 49,677 17,625 73,013 44,305 4.1 2.5 Malaysiaa 36,156 26,439 60,590 27,335 2.3 1.0 Thailand 14,744 4,324 30,747 15,906 7.1 3.7 Source: Bosworth 2010. a. The utilities industry in Malaysia is included in services rather than industry. Gaps in output per worker between agricul- growth. Reallocation of workers is also ture and the rest of the economy remain necessary to accelerate the movement of in East Asia as well, with larger gaps for resources from low-productivity to high- industry than services. productivity activities within the three Hence an acceleration in the movement broad sectors. However, the limited educa- of resources from agriculture (where TFP tional attainment of the labor force in South growth has been slowest) into industry and Asia, analyzed in chapter 5, implies that services (where growth has been brisker) realizing higher TFP growth through the has the potential to increase aggregate TFP intersectoral and intrasectoral reallocation GROWTH AND JOB QUALITY IN SOUTH ASIA 59 of labor will require substantial investment force participation—the proportion of the in human capital. De�ciencies in infrastruc- working-age population that is in the labor ture, analyzed in chapter 4, imply that real- force—have moved slowly in South Asia, location will require investment in physical implying that the growth of the labor force capital as well. Creating an enabling envi- has tracked that of the working-age popula- ronment for accelerated physical and human tion. Second, in countries where the lack of capital formation and reallocation of labor social safety nets does not allow the luxury to higher-productivity areas must go hand of open unemployment, the proportion of in hand. Doing so represents the most press- the labor force that is unemployed is low ing growth challenge facing South Asia. and does not change very much. At the mar- gin, additional entrants into the labor force are absorbed into low-productivity occu- The track record on pations. Hence the growth of employment employment moves broadly in line with that of the labor This section examines South Asia’s record force. on the quantity of jobs created, the qual- Taken together, these observations imply ity of jobs created, and the degree to which that employment growth can be expected workers move across job categories. to broadly track growth in the working-age population.10 Total employment in South Asia (excluding Afghanistan and Bhutan) Job quantity rose from 473 million in 2000 to 568 million In all South Asian countries, the number in 2010, an average annual rate of growth of of jobs created has grown broadly in line 1.8 percent, ranging from just over 1 percent with the working-age population, for two a year in Sri Lanka to nearly 4 percent a year reasons (figure 2.10). First, rates of labor in Pakistan.11, 12 FIGURE 2.10 Annual growth in working-age population, employment, and labor force in selected South Asian countries 4 3 percent 2 3.6 3.6 3.3 3.3 3.0 2.5 2.6 2.8 2.3 2.2 2.2 2.3 1 1.0 1.2 1.0 0 Sri Lanka India Bangladesh Nepal Pakistan 2000–10 1985–2010 2000–10 1995–2010 2000–10 working-age population employment labor force Sources: Authors, based on data on working-age population from UN 2010 and data on employment and labor force from national labor force surveys. 60 MORE AND BETTER JOBS IN SOUTH ASIA regular wage or salaried workers and fall- Job quality ing poverty rates for the self-employed. Real Two criteria are used to assess job quality. wages in much of South Asia grew 0.1–2.9 The primary criterion is higher average earn- percent a year during various subperiods ings. For wage or salaried workers, it can be between 1983 to 2010 for which comparisons assessed using information on average earn- can be made (�gure 2.11). A higher proportion ings. Survey data do not contain information of the self-employed belong to households on earnings of the self-employed, the larg- that are above the national poverty line in est segment of the labor force in South Asia Bangladesh, India, Nepal, Pakistan, and Sri (except Maldives). Changes in poverty rates Lanka (�gure 2.12). This proportion is used (the percentage of workers living in house- as a proxy for improving job quality for the holds below the poverty line) are used as a self-employed, although falling poverty rates proxy for job quality for this segment of the in households of the self-employed could also labor force. Based on these primary criteria, be a result of an increase in other sources of better jobs are those associated with higher income, such as workers’ remittances (which (average) wage rates for wage workers and are very important in Nepal and somewhat lower poverty rates for the self-employed. important in Bangladesh) or increased hours A secondary criterion of job quality looks worked by household members.14 beyond average income to its variability. Higher proportions of casual work- Variation in income and consumption arising ers and regular wage or salaried workers in from the lack of stable employment exposes Bangladesh, India, and Nepal and all wage workers to the risk of low and uncertain workers in Pakistan and Sri Lanka also belong income. These risks can be major for casual to households above the poverty line (�gure wage workers. Because data limitations in all 2.12).15 This trend is consistent with the evi- countries in the region except India preclude dence of improving job quality provided by a consistent application of this secondary cri- rising real wages and, in Nepal, the poverty- terion, the primary criterion for better jobs reducing impact of workers’ remittances. guides most of this book.13 Poverty rates for all types of workers dur- The creation of better jobs is reflected in ing all time periods also show a decline when rising real wages for both casual workers and the data are disaggregated by location (rural FIGURE 2.11 Average annual increases in mean real wages in selected countries in South Asia 4 2.9 2.8 percent 1.9 2.0 2 0.1 0 Bangladesh India Nepal Pakistan Sri Lanka 2002–05 1983–2010 1999–2008 2000–09 2000–08 Source: Authors, based on data from national labor force and household surveys. GROWTH AND JOB QUALITY IN SOUTH ASIA 61 FIGURE 2.12 Percentage of workers in households below the poverty line in selected South Asian countries, by employment status a. Bangladesh, 2000–10 b. India, 1983–93 and 1999–2004 c. Nepal, 1995–2003 80 80 80 70 67 70 70 61 58 60 60 60 47 47 47 48 50 50 50 44 43 42 percent percent percent 38 38 39 40 40 40 33 31 28 30 39 29 38 28 30 30 30 30 22 21 29 29 28 27 20 24 20 26 20 21 18 18 10 10 15 10 12 7 0 0 0 2000 2005 2010 1983 URP 1993 URP 1999 MRP 2004 MRP 1995/96 2003/04 all regular wage or salaried all self-employed all casual labor all workers d. Pakistan, 2001/02 to 2007/08 e. Sri Lanka, 1995/96–2006/07 80 80 70 70 60 60 50 50 percent 43 percent 40 40 30 30 25 26 20 23 20 24 13 15 10 10 11 0 0 2001/02 2007/08 1995/96 2006/07 all self-employed, all wages, and all wages all self-employed all workers all workers Source: Authors, based on data from national labor force and household surveys. Note: URP = uniform recall period (the period in which respondents were asked to recall all consumption items over the same recall period [for example, 7 days]). MRP = mixed recall period (the period need not be the same for all items, [for example, 7 days for some and 365 days for others]). Figures are for workers age 15–64. or urban) or gender, although details vary by countries, but poverty rates for female work- country. Whereas urban poverty fell some- ers remained higher than for male workers in what faster than rural poverty in Bangladesh urban Bangladesh, India, and Pakistan (�g- and Sri Lanka and considerably faster in ure 2.13 shows data on India). In sum, using Nepal, the opposite was true in India between the primary criteria of higher wages for wage 1983 and 1993 and between 1999 and workers and lower poverty rates for the self- 2004.16 In Pakistan, urban and rural worker employed, South Asia has created better jobs. poverty rates fell equally rapidly between There has also been an improvement in 2001/02 and 2007/08. Poverty rates for both job quality in India based on the secondary male and female workers declined in all �ve criterion, namely, a reduced risk of low and 62 MORE AND BETTER JOBS IN SOUTH ASIA FIGURE 2.13 Percentage of workers in households below the poverty line in India, by employment status and gender a. Male workers b. Female workers 80 80 70 70 60 58 60 59 50 47 50 47 48 43 percent percent 41 41 40 40 40 34 36 38 32 32 31 31 30 28 30 30 27 28 29 22 20 25 20 18 20 22 19 18 17 15 15 12 10 10 0 0 1983 URP 1993 URP 1999 MRP 2004 MRP 1983 URP 1993 URP 1999 MRP 2004 MRP regular wage or salaried self employed casual labor all workers Source: Authors, based on data from national labor force and household surveys. Note: URP = uniform recall period (the period in which respondents were asked to recall all consumption items over the same recall period [for example, 7 days]). MRP = mixed recall period (the period need not be the same for all items, [for example, 7 days for some and 365 days for others]). Figures are for workers age 15–64. uncertain income for casual wage workers Notwithstanding the variation in pov- (India was the only country that had suf�- erty rates across employment types, there is cient data on which to conduct this analysis). a consistent association between poverty and The average number of months for which all type of employment across countries and casual laborers (farm and nonfarm, rural and over time (figure 2.12). Casual labor is the urban) were without work, despite looking for most vulnerable segment of the labor force it, declined between 1999/2000 and 2009/10 and has the highest poverty rates (more than (�gure 2.14).17 Thus, the secondary criterion 40 percent in both r u ral and u rban for better jobs is established for India. Bangladesh, for example). The poverty rates Economic growth in the region has driven of the self-employed are the second highest.18 improvements in the quality of jobs. But not Poverty rates are generally lowest among all countries have enjoyed high or accelerating regular wage or salaried workers; on average, growth. Per capita GDP was virtually stag- they are one-third or less of those for casual nant in Nepal in the 1960s and 1970s and labor. This pattern of association is evident has grown at 2 percent or less a year since the in the consumption distribution by type of 1980s (�gure 2.2). Despite sluggish growth, employment (�gure 2.15). It is also consistent real wages have risen nearly 3 percent a year with observed wage differentials between since the 1980s, and poverty among workers regular wage or salaried workers and casual fell between the mid-1990s and the 2000s. labor (see chapter 3). These patterns have These improvements reflect massive out- endured over time. Hence, better jobs could migration of workers in response to limited be created either through improvement within job opportunities, which has improved labor an employment type or through the realloca- market outcomes for those who stay behind. tion of workers from job types with higher A large inflow of remittances has contributed poverty rates and lower wages to those with to declining poverty (box 2.1). lower poverty rates and higher wages. GROWTH AND JOB QUALITY IN SOUTH ASIA 63 FIGURE 2.14 Average number of months without work in the past year, casual laborers in India, by sector, 1999–2010 2.0 1.8 1.9 1.6 1.5 1.5 1.4 1.4 1.4 1.1 months 1.0 0.9 0.5 0 1999/2000 2004/05 2009/10 agriculture rural nonfarm urban Source: Authors, based on data from Indian labor force and household surveys. Note: Figures are for workers age 15–64 who were available for work during at least part of the month. BOX 2.1 International migration in Nepal and its effects on poverty Despite slow economic growth, Nepal has enjoyed earnings were Nr 16,000. Despite higher living costs higher wages and a signi�cant decline in poverty rates overseas, migrants are able to save, with a typical among workers—thanks in large part to massive out- migrant saving about Nr 8,000 a month—twice the migration and inflow of workers’ remittances. Labor amount earned in Nepal. migration has been a feature of life in Nepal for 200 Nepal has the largest remittances as a share of GDP years. The primary destination for migrants was tra- of any country in the world with more than 10 mil- ditionally India, although since the 1990s migrants lion people. Of�cial remittances totaled $2.7 billion in have increasingly headed to the Middle East and 2009, equivalent to 22 percent of GDP; including infor- Malaysia. The Maoist insurgency from 1996 to 2006 mal flows and remittances from India, total inflows are accelerated the pace of migration, especially after estimated to have exceeded 25 percent of GDP. �ghting intensi�ed in 2001, as many people fled rural Remittances have increased household income sig- communities affected by the hostilities. ni�cantly. An estimated 39 percent of all households The total number of migrants is estimated at and 84 percent of households with recent migration about 4.2 million, equivalent to 13 percent of Nepal’s experience received remittances in 2009. Income population. Migration is widespread, occurring in from remittances accounted for 24 percent of the households of all income groups and from all parts annual income of all households and two-thirds of of the country. Almost half of all households have the income of remittance-receiving households. The had at least one migrant abroad at some time. The additional income is spent largely on consumption, vast majority of migrants (93–94 percent) are men, education, and childcare. More than half the decline most of them 20–40 years old. At least one-third of in Nepal’s poverty rate between 1996 and 2004 (from working-age men in Nepal are migrants. 42 percent to 31 percent) is estimated to have been the Why is migration so prevalent in Nepal? The result of remittances. phenomenon is viewed as a response to limited Migration has also had a signi� cant impact on domestic job opportunities in a stagnant economic the labor force. The male labor supply has fallen, environment with a poor business climate and politi- especially in rural areas. Remittances have caused cal instability. The majority of migrants worked in recipient households to increase their consumption of agriculture in Nepal but moved into manufactur- leisure and reduce labor supply as well. The decline ing, construction, and services (such as hotels and in the male labor supply has reduced domestic unem- catering) after migration. More than 87 percent of ployment and underemployment and led to rising migrants are literate compared with 62 percent of real wages. nonmigrants. Before migration, migrants earned about Nr 4,000 a month; after migration their average Source: World Bank 2011a. 64 MORE AND BETTER JOBS IN SOUTH ASIA FIGURE 2.15 Distribution of per capita household expenditure in occurred over time as a result of transitions India and Nepal, by employment status across the three groups. Workers in South Asia have better jobs than they previously did mainly as a result of increasing quality within a. India, 2005 rather than across employment categories 0.003 (table 2.2). Disaggregating further by sector can reveal more across-type movements in the kernel density 0.002 labor force. Separating India’s rural economy into agriculture and nonfarm (rural-based 0.001 industry and services) sectors reveals that between 1983/84 and 2004/05, the share of casual labor increased (from 25 percent to 0 28 percent) and the share of regular wage or 0 1,000 2,000 3,000 salaried jobs decreased (from 27 percent to monthly per capita expenditure (Indian rupees) 25 percent) (�gure 2.17). The share of casual b. Nepal, 2003 labor in the rural nonfarm sector increased 0.0015 to 38 percent by 2009/10, because most rural nonfarm employment was casual work.20 The World Bank’s poverty assessment of kernel density 0.001 India (World Bank 2011b) notes the increas- ingly bimodal nature of consumption by rural nonfarm regular wage or salaried workers, 0.0005 among whom a minority earns wages that are much higher than average. These trends 0 in the nonfarm economy notwithstanding, 0 1,000 2,000 3,000 median wages of casual and regular wage monthly per capita expenditure (Nepalese rupees) or salaried workers in the rural nonfarm regular wage casual wage economy increased and poverty rates fell dur- self employed ing this period. In addition, the number of months in which rural nonfarm casual labor- ers were available for work but unemployed Source: Authors, based on data from national labor force and household surveys. declined between 2004/05 and 2009/10 (see �gure 2.14). Thus, both the primary and the secondary criteria for improved job quality were met in the rural nonfarm economy. The proportion of workers in the three The effect of a growing labor force, employment types has remained broadly declining poverty rates, and changes in the unchanged over time, in both rural and proportion of workers by different employ- urban areas (figure 2.16). In India, which ment types has generally led to a decline in has the longest time series, the decline in the number of working poor (for details, see self-employment and increase in casual labor annex 2E). In Bangladesh, the number of observed between 1983/84 and 1999/2000 working poor decreased 18 percent between was reversed by 2004/05 but observed again 2000 and 2010, with declines in most types between 2004/05 and 2009/10.19 Given the of employment except rural nonfarm casual hierarchy of poverty rates (which are high- labor, urban casual labor, and urban regu- est among casual workers and lowest among lar wage or salaried workers. In India, the regular wage and salary earners), this implies number of working poor increased 2 per- that no signi�cant change in poverty status cent between 1985 and 1995 for all types of GROWTH AND JOB QUALITY IN SOUTH ASIA 65 FIGURE 2.16 Distribution of rural and urban workers in selected South Asian countries, by employment type a. Bangladesh, 2000–2010 b. India, 1983–2009/10 100 100 18 18 18 15 17 90 22 21 21 90 31 33 31 38 36 39 80 35 38 80 70 70 28 45 41 35 36 60 40 42 41 60 percent percent 50 50 40 49 44 40 61 61 47 60 57 54 30 30 51 20 43 43 20 43 41 41 40 43 10 17 10 14 15 8 7 8 8 8 0 0 00 05 10 00 05 10 83 4 0 5 0 83 4 0 5 0 –9 00 –0 –1 –9 00 –0 –1 20 20 20 20 20 20 19 19 –2 –2 93 04 09 93 04 09 19 99 20 20 19 99 20 20 rural urban 19 19 rural urban casual labor self-employed regular wage or salaried c. Nepal, 1996–2004 d. Pakistan, 2000–2009 e. Sri Lanka, 2000–2008 100 100 100 15 13 12 90 18 17 16 16 19 20 19 31 33 80 80 80 45 44 70 60 60 60 61 46 45 45 percent percent 55 percent 50 70 71 79 83 71 40 40 40 69 67 30 55 56 20 20 20 35 36 36 27 26 10 12 14 13 5 4 0 0 0 96 04 96 04 0 8 9 0 8 9 00 08 00 08 00 –0 –0 00 –0 –0 19 20 19 20 20 20 20 20 –2 –2 07 08 07 08 99 20 20 99 20 20 rural urban rural urban 19 19 rural urban self-employed casual labor self-employed regular wage or salaried wage worker Source: Authors, based on data from national labor force and household surveys. a. Data from the Bangladesh Household Income and Expenditure Surveys (HIES) were used to calculate worker poverty rates. The share of workers by employment type in the HIES differs from the share in the Bangladesh labor force surveys. The difference is likely to be partly driven by how female employment is captured, with female participation rates in the HIES less than half those reported in the labor force survey. Therefore, the changes in the share of workers by type in Bangladesh from the HIES should be interpreted carefully. For example, between 2005 and 2010 the signi�cant increase in the share of regular wage or salaried work in urban areas was driven largely by changes in the female urban workforce reported in the HIES 2005 and HIES 2010. b. Although there is variation in the shares of casual labor and self-employment in rural areas in India, there is no persistent increase or decline in the shares throughout the whole period (for example, the increase in casual labor between 2004/05 and 2009/10 mostly reversed the decline between 1999/2000 and 2004/05); the share of regular wage or salaried workers remained constant throughout the 25-year period. casual labor—agricultural, rural nonfarm, number of working poor decreased 2 percent and urban—and for the urban self-employed. between 1995 and 2005, falling among rural In contrast, the number of working poor fell nonfarm regular wage or salaried workers, 18 percent between 2000 and 2005 and in rural nonfarm self-employed and urban most employment types. 21 In Nepal, the casual labor. 66 MORE AND BETTER JOBS IN SOUTH ASIA TABLE 2.2 Decomposition of decline in worker poverty rates (percent) India Nepal Bangladesh Contribution to decline in worker poverty rates 1983–93 1999–2004 1995–2003 2000–10 Changes in poverty rates of different employment types 101.0 90.7 78.6 93.3 Changes in distribution of employment type –1.9 12.9 20.0 10.0 Interaction/residual 0.9 –3.6 1.5 –3.4 Sources: Authors, based on national labor force and household surveys. Note: Changes in poverty rates holds distribution of employment status constant; changes in distribution of employment status holds poverty rates constant. The interaction term equals 100 percent – (A + B). FIGURE 2.17 Distribution of rural nonfarm workers in India, by employment type, 1983–2009/10 100 90 25 26 28 28 80 38 70 60 percent 50 48 49 47 48 40 41 30 20 27 25 25 24 10 22 0 1983 1993/94 1999/2000 2004/05 2009/10 casual wage self-employed regular wage or salaried Source: Authors, based on data from national labor force and household surveys. into each of the four possible states (the Labor mobility combinations of the two possible employ- The broad constancy in the share of work- ment types in each of the two time periods ers across employment types does not nec- shown in the �gure) are generated by adapt- essarily imply a lack of mobility at the ing a technique developed by Lanjouw, level of individual workers across employ- Luoto, and Mckenzie (2011) to study pov- ment types. As an illustration, the analysis erty transitions. 22 focuses on labor transition for rural workers A sizable share of rural workers in the from agricultural work (less desirable jobs three countries is moving in both directions on average) to rural nonfarm work (more between agriculture and the rural nonfarm desirable jobs on average) and vice versa sector (�gure 2.19). The share of rural work- (�gure 2.18). Lower- and upper-bound esti- ers moving from agriculture to rural non- mates of the shares of rural workers falling farm work was 5–17 percent in Bangladesh GROWTH AND JOB QUALITY IN SOUTH ASIA 67 FIGURE 2.18 Labor transitions in rural areas between 2002 and 2005, 10 –20 percent in India between 1999 and 2004 and 3–13 percent between 2004 and 2007, and 4–11 Second period percent in Nepal between 1996 and 2004. The movement from the rural nonfarm sec- Agriculture Rural nonfarm tor back to agriculture was 5–17 percent in Rural nonfarm Agriculture No transition: More desirable agriculture both transition: Bangladesh between 2002 and 2005, 2–12 periods agriculture to percent in India between 1999 and 2004 First period rural nonfarm and 8–18 percent between 2004 and 2007, Less desirable No transition: and 3–10 percent in Nepal between 1996 and transition: rural nonfarm rural nonfarm to both periods 2004. The data do not allow a conclusion to agriculture be drawn as to which transition was larger, as the bounds for both the more desirable (agriculture to rural nonfarm labor) and less Source: Authors. desirable (rural nonfarm labor to agriculture) FIGURE 2.19 Probability of moving into or out of better jobs in rural Bangladesh, India, and Nepal a. Bangladesh, 2002–05 b. India, 1999–2004 and 2004–07 80 80 70 70 60 60 50 50 percent percent 40 40 30 30 20 20 10 10 0 0 rio re ul to rio m rio re nf to nf to ul to rio m pe ltu pe ltu pe far pe far ric rm no lture no lture ds ric rm re ds ds m m re ds th icu th non th icu tu ar th non ar ag nfa tu ag nfa u u bo agr bo agr no ric ric no ag ag bo bo c. Nepal, 1996–2004 80 70 60 50 percent 40 30 20 10 0 ul to rio m rio re nf to pe ltu pe far ric rm no lture re ds ds m th non th icu tu ar ag nfa u bo agr no ric ag bo Source: Authors, based on national labor force and household surveys. Note: The upper -and lower-bound estimates shown by the bars indicate the share of rural workers in the states of labor transitions described in �gure 2.19. In panels a and c, the red bar represents the given years. In panel b, the red bar represents 1999–2004 and the blue bar represents 2004–07. 68 MORE AND BETTER JOBS IN SOUTH ASIA transitions overlap. In fact, the similarity of 0.1–0.4. The correlation between growth both sets of transition bounds, especially in rates in 1971–80 and growth rates thereafter Bangladesh and Nepal, suggests that the pro- is 0.37 for 1981–90, 0.28 for 1991–2000, portion of workers could be similar in both and negligible thereafter (table 2.3). East- types of transitions. These results are con- erly and others (1993) attribute this � nding sistent with the observed constancy of the to exogenous shocks, such as adverse move- share of the rural workforce engaged in rural ments in the terms of trade and armed con- nonfarm activities noted later in chapter 3. fl ict, both of which are prevalent in South Although the bounds still overlap by a small Asia. The presence of such shocks is an amount in India between 1999 and 2004, the important reason for undertaking reforms size of the more desirable transition is notice- of the business environment (chapter 4) ably larger than that of the less desirable and education systems (chapter 5) to ensure one. This �nding is consistent with the large that the creation of better jobs is not overly increase in the share of the rural workforce in dependent on continued high economic rural nonfarm activities—from 25 percent in growth. 2000 to 30 percent in 2004—noted in chap- Demographic pressures lend further ter 3. The bounds of the more desirable and urgency to the need for reform. Most coun- less desirable transitions overlap much more tries in the region have a demographic between 2004 and 2007, when the share of window of opportunity during which an the rural nonfarm workforce in India was vir- enabling policy framework can help them tually flat. reap a demographic dividend. But the The less desirable transition suggests that opportunity is time bound and will close the nature of the rural nonfarm work itself for most of South Asia around 2040. Pop- might be transient and temporary in nature. ulation projections and the age structure In fact, the variance of the wage distributions of the population are used to develop two for casual nonfarm workers is higher than scenarios (details of the projections are in that for casual agricultural workers in India annex 2D): in all labor force surveys except 2004/05. Chapter 5 takes the labor transition analysis • Scenario 1: South Asia adds 1 million farther by looking at which types of workers entrants to the labor force every month are more likely to make more desirable and between 2010 and 2030. This scenario less desirable transitions. assumes no increase in the rates of female In summary, the labor transition analysis labor force participation, which are suggests that there is labor mobility in South among the lowest in the world. Asia and that labor moves to both more • Scenario 2: Female labor participation desirable and less desirable jobs. These two- rates in Bangladesh, India, and Pakistan way transitions are masked when looking at increase 10 percentage points by 2030, cross-sectional data at the aggregate level. in line with observed behavior in labor force participation rates in Indonesia, the Republic of Korea, Malaysia, and Thai- The urgency of reform land between 1960 and 2000. In this sce- nario, 1.2 million entrants are added to the The continuation of economic growth, labor force between 2010 and 2030, fur- which has been associated with improved ther intensifying labor market pressure. job quality in South Asia during the last three decades, cannot be taken for granted. By way of comparison, an average of Growth rates are famously unstable over just under 800,000 entrants joined the time: across five decades, the correlation labor market in South Asia every month of growth rates of GDP per capita for 94 between 1990 and 2010. The two sce- countries for which data are available in narios thus represent increases of 25–50 the World Development Indicators is just percent above the average for this period. GROWTH AND JOB QUALITY IN SOUTH ASIA 69 TABLE 2.3 Correlations of country growth rates of per capita GDP across decades 1961–70 1971–80 1981–90 1991–2000 2001–10 1961–70 1 0.34*** 0.37*** 0.22** –0.18* 1971–80 1 0.37*** 0.28*** 0.14 1981–90 1 0.43*** 0.07 1991–2000 1 0.18* 2001–10 1 Source: Authors’ calculations, based on data from World Bank 2011c. Note: Findings are based on 94 countries. *** Signi�cant at the 1% level; ** signi�cant at the 5% level; * signi�cant at the 10% level. The critical question is whether increases high economic growth may not continue, of this magnitude will be absorbed at ris- highlight the importance of proceeding ing or low levels of productivity. The two quickly with reform in order to meet the scenarios, together with the possibility that employment challenge. Annex 2A Methodology for decomposing growth A country’s output in any given year depends Given an estimate for and measures of on its factor inputs—labor and (human and Y, L, K, and H, it is straightforward to solve physical) capital—as well as the ef� ciency for A (or a) and construct the decomposition. with which factors are used in production. 23 It is assumed that capital’s share = 0.35. De� ne Y as real GDP, K as the physical capi- An analysis that used actual income shares tal stock, A as the level of technology, and L in each period would allow for the consid- as labor (measured as “bodies of economi- eration of a much wider range of underlying cally active persons�), which is assumed to production functions. However, few coun- be “augmented� by H, an index of the aver- tries are able to allocate the incomes of the age level of labor quality, measured by aver- self-employed between capital and labor, a age years of schooling. Assume that a coun- particular problem in South Asia, where the try’s output can be expressed as a function self-employed make up the bulk of the labor of these inputs, using the speci�c functional force. form shown in equation (2A.1), and that L is employment for industrial countries returns to scale are constant. plus Bangladesh, India, Pakistan, and Sri Y = AK (HL)(1 ) (2A.1) Lanka and labor force for all other countries. The capital stock measure is constructed The results are reported in a form that from investment data using the perpetual decomposes growth in output per worker into inventory method, with annual depreciation the contributions from the growth of physi- rate of 0.05 percent. The construction of H cal capital per worker, education per worker, assumes that human capital is directly related and total factor productivity, as shown in to average years of schooling (S) and that equation (2A.2) (lower-case letters denote a there is a 7 percent return to each additional variable’s average annual growth rate). year of schooling: y/l = (k/l) + (1 )h a (2A.2) H = (1.07)S (2A.3) 70 MORE AND BETTER JOBS IN SOUTH ASIA Annex 2B Sources of average annual growth in output per worker TABLE 2B.1 Sources of average annual growth in output per worker, by region, 1960–2008 (percent) Change in Contribution of Output Physical capital Education per Total factor Region/period Output per worker per worker worker productivity World (83 countries) 1960–80 4.91 3.12 1.29 0.41 1.34 1980–90 3.81 1.96 0.81 0.38 0.60 1990–2000 4.03 2.41 0.97 0.38 0.85 2000–08 4.30 2.73 0.93 0.24 0.99 1980–08 4.03 2.34 0.90 0.34 0.80 Industrial countries (22 countries) 1960–80 4.45 3.21 1.20 0.42 1.45 1980–90 3.08 1.91 0.73 0.31 0.78 1990–2000 2.43 1.61 0.84 0.33 0.57 2000–08 1.83 1.08 0.76 0.14 0.13 1980–2008 2.49 1.56 0.77 0.27 0.52 Sub-Saharan Africa (19 countries) 1960–80 4.53 2.16 0.87 0.18 1.15 1980–90 1.62 (1.47) (0.11) 0.47 (1.78) 1990–2000 1.92 (1.60) (0.76) 0.48 (1.38) 2000–08 4.72 2.24 0.25 0.41 1.58 1980–2008 2.61 (0.47) (0.24) 0.45 (0.69) China 1960–80 4.89 2.65 0.91 0.44 1.27 1980–90 9.29 6.62 2.09 0.39 4.03 1990–2000 10.42 9.11 3.28 0.50 5.12 2000–08 10.18 9.25 3.45 0.39 5.20 1980–2008 9.95 8.25 2.90 0.43 4.75 East Asia less China (7 countries) 1960–80 7.84 4.46 2.58 0.56 1.27 1980–90 7.42 4.26 2.15 0.64 1.18 1990–2000 6.13 3.98 2.35 0.50 0.78 2000–08 4.46 2.53 0.78 0.49 1.30 1980–2008 6.11 3.66 1.83 0.55 1.07 1980–96 7.57 4.68 1.50 0.59 2.31 1999–2008 4.81 2.76 1.50 0.49 0.78 Latin America and the Caribbean (21 countries) 1960–80 6.06 2.84 0.87 0.36 1.58 1980–90 1.49 (1.65) 0.11 0.52 (2.32) 1990–2000 3.15 0.47 0.08 0.48 (0.14) 2000–08 3.52 1.28 0.24 0.39 0.69 1980–2008 2.66 (0.06) 0.14 0.47 (0.69) Middle East (9 countries)           1960–80 5.68 3.40 2.45 0.43 0.95 1980–90 3.78 1.26 0.44 0.60 0.36 1990–2000 3.80 1.11 0.15 0.57 (0.02) 2000–08 4.84 2.74 0.73 0.46 1.39 1980–2008 4.09 1.63 0.42 0.55 0.52 (continues next page) GROWTH AND JOB QUALITY IN SOUTH ASIA 71 TABLE 2B.1 Sources of average annual growth in output per worker, by region, 1960–2008 (continued) Change in Contribution of Output Physical capital Education per Total factor Region/period Output per worker per worker worker productivity South Asia (4 countries) 1960–80 3.93 1.63 1.04 0.27 0.31 1980–90 5.24 3.18 1.11 0.42 1.46 1990–2000 5.37 3.47 1.21 0.37 1.92 2000–08 7.05 4.94 1.67 0.34 2.90 1980–2008 5.80 3.78 1.30 0.38 2.04 Source: Bosworth 2010. Note: Average annual growth in output per worker is the sum of the physical capital, human capital, and TFP contributions to growth in productivity. Totals may not add up to total growth in output per worker because of interaction terms. Table is based on aggregated growth accounts, except for Bangladesh, India, Pakistan, and Sri Lanka data from 1980 onward, which are based on disaggregated growth accounts. For South Asia, growth accounts use employment data from national labor force surveys (latest available in August 2010). Aggregated growth accounts for regions other than South Asia use labor force estimates from World Bank 2011c (taken from the International Labour Organization [ILO]), which include the unemployed population. ILO data include extrapolated data. TABLE 2B.2 Sources of average annual growth in output per worker in South Asia, by country, 1960–2008 (percent) Change in Contribution of Output per Physical capital Education per Total factor Country/period Output worker per worker worker productivity Bangladesh 1960–80 2.39 0.21 0.27 0.21 –0.26 1980–2008 4.66 2.01 0.87 0.44 0.69 India 1960–80 4.67 1.34 1.00 0.25 0.08 1980–2008 6.2 4.39 1.41 0.42 2.60 Pakistan 1960–80 5.94 3.1 2.38 0.19 0.53 1980–2008 5.28 2.67 1.00 0.30 1.40 Sri Lanka 1960–80 4.49 2.18 0.43 0.43 1.31 1980–2008 4.91 3.08 1.62 0.27 1.16 Source: Bosworth 2010. Note: Average annual growth in output per worker is the sum of the physical capital, human capital, and TFP contributions to growth in productivity. Totals may not add up to total growth in output per worker because of interaction terms. The growth accounts use employment data from national labor force surveys (latest available in August 2010). 72 MORE AND BETTER JOBS IN SOUTH ASIA Annex 2C Shares of agriculture, industry, and services in employment and GDP TABLE 2C.1 Regressions of shares of agriculture, industry, and services in employment and GDP, 2008 Employment GDP Agriculture Industry Services Agriculture Industry Services Variable (1) (2) (3) (4) (5) (6) ln (per capita GDP, 2008 –64.77** 28.92 29.87 –44.96*** 69.57 –24.61 in 2005 constant purchasing (25.82) (18.30) (26.93) (14.13) (56.68) (54.82) power parity dollars) ln (per capita GDP, 2008 2.783** –1.505 -0.988 2.040*** –3.780 1.740 in 2005 constant purchasing (1.365) (0.954) (1.425) (0.724) (2.990) (2.899) power parity dollars, squared) ln (land area, square kilometers) 0.316 –0.210 0.0783 –0.226 1.245 –1.019 (0.216) (0.348) (0.470) (0.378) (0.887) (0.826) Bangladesh dummy –12.28 1.169 9.185 –13.90** 11.22 2.679 (7.444) (5.946) (7.745) (5.659) (15.89) (14.92) India dummy 13.96*** 1.102 –15.26*** –3.045 –2.457 5.502 (3.497) (2.895) (3.661) (2.693) (8.026) (7.433) Nepal dummy 4.107 0.396 –6.753 –1.170 1.490 –0.320 (7.754) (6.180) (8.069) (5.842) (16.61) (15.61) Pakistan dummy 2.393 1.812 –5.220 –1.613 –2.101 3.714 (3.380) (2.748) (3.462) (2.776) (7.036) (6.454) Sri Lanka dummy 2.905 4.758*** –9.682*** –2.062 –1.629 3.691** (1.900) (1.462) (2.079) (1.954) (2.232) (1.828) Constant 373.0*** –112.7 –133.0 251.1*** –300.2 149.1   (121.0) (86.40) (125.7) (69.45) (260.1) (250.8) N 55 55 55 53 53 53 Adjusted R-squared 0.860 0.095 0.799 0.827 0.080 0.376 Sources: Authors, based on data from ILO 2010; World Bank 2011c; and India National Sample Survey. Note: Standard errors in parentheses. Sample restricted to countries for which employment data were available from both 1980 (or the earliest year in the subsequent �ve-year period) and 2008 (or the latest year in the preceding �ve-year period). Data exclude transition economies in Europe and Central Asia. *** Signi�cant at the 1% level; ** signi�cant at the 5% level; * signi�cant at the 10% level. Annex 2D Methodology and data sources for labor force projections This annex describes the assumptions used household surveys (mostly 2008) to the 2010 to generate the labor force projections for age-gender populations from the United 2030. The scenarios are developed at the Nations (UN 2008). 24 country level and aggregated for the region. Scenario 1 (demographic case) Historical period (1990–2010) Labor force projections in 2030 under sce- Trends in the region’s labor force cover nario 1 are based purely on demographic the period 1990–2010. The source of the projections, using UN (2008). In this and labor force in 1990 for each country is the the next scenario, the labor force in 2010 for LABORSTA database of the International each country is calculated using the same Labour Organization (ILO). The labor method described in the historical descrip- force in 2010 is obtained by applying the tion. Similarly, the 2030 labor force projec- (�ve-year) age/gender labor force participa- tions are obtained by multiplying the 2030 tion rates obtained from the latest available (five-year) age/gender populations by the GROWTH AND JOB QUALITY IN SOUTH ASIA 73 same current age/gender–speci�c labor force The experience of four Asian countries participation rate, calculated from the latest (Indonesia, the Republic of Korea, Malaysia, household surveys. and Thailand) was used to develop this sce- nario. These countries represent an interest- ing example for South Asia, for a variety of Scenario 2 (behavioral case) reasons. First, as neighbors they share some Scenario 1 assumes no change in labor social and cultural characteristics. Second, force participation rates over the projec- they developed earlier than the large South tion period for speci� c age/gender groups. Asian countries and can therefore provide Scenario 2 assumes that participation by insights as benchmark countries. Third, at women in Bangladesh, India, and Pakistan the beginning of their “take-offs,� Indonesia, will increase between 2010 and 2030. These Korea, and Malaysia (though not Thailand) three countries are selected for this scenario had female participation rates comparable because each has very low female partici- to those of Bangladesh and India and a bit pation rates by international standards. higher than Pakistan. In the past 50 years, to Together with the fact that they dominate different degrees, these countries experienced South Asia’s overall population (accounting strong economic growth and modernization; for 95 percent of the working-age popu- in the process, all of them except Thailand lation), this implies that any increases in experienced increases in female participation female participation in these three large rates (box 2D.1). Tracking these historical countries could have an important effect on development indicators and roughly match- the region’s labor force. Under scenario 2, ing them with projected trends (for example, participation rates are assumed to remain per capita GDP) for the South Asian coun- at current levels for men in all countries tries to 2030 suggest that an increase in and women in the other five countries in the female labor force participation rate of the region (scenarios 1 and 2 are identical 10 percentage points would approximate a for Afghanistan, Bhutan, Maldives, Nepal, convergence with the historical experience of and Sri Lanka). these benchmark countries. 74 MORE AND BETTER JOBS IN SOUTH ASIA BOX 2D.1 Trends in female labor force participation in southeast and East Asian comparator countries Table 2D.1.1 summarizes what happened to female lower end of this range, female participation rates in labor force participation in Indonesia, Korea, Indonesia, Korea, and Malaysia were in the low to Malaysia, and Thailand during the periods when mid-30s—not too different from rates in India and per capita GDP increased through the range pro- Bangladesh today, though about 10 points higher jected for Bangladesh, India, and Pakistan to 2030. than in Pakistan. The case of Thailand demonstrates In Indonesia, Korea, and Malaysia, female participa- that increasing participation is not inevitable as tion rates increased consistently, in some cases, sub- per capita GDP increases. Although female rates in stantially, with per capita GDP (in 2005 purchasing Thailand did rise as per capita GDP grew to about power parity dollars) rising from the mid-$2,000s $5,000, they subsequently stalled and then started to almost $10,000. When per capita GDP was at the to decline. BOX TABLE 2D.1.1 Female labor force participation in Bangladesh, India, and Pakistan and Asian comparator countries Projected per capita GDP 2007–30 Projected changes in female (in 2005 purchasing power Southeast and East Asian labor force participation rate Country parity dollars) comparators (percentage points) Bangladesh Mid-$2,000s to high $3,000s Indonesia 1980–90s Almost +10 Malaysia 1960–70s More than +5 Thailand 1970–80s About +5 Republic of Korea 1960–70s About +10 India High $3,000s to low $9,000s Malaysia 1970s–90s Almost +10 Thailand 1980s onward About –10 Republic of Korea 1970s–90s About +10 Pakistan Mid $3000s to Low $6000s Indonesia 1990 onward About +5 Malaysia 1970–80 About +3 Thailand 1980–90 No change Republic of Korea 1970–80 About +2 Sources: Authors, based on GDP trends from Penn World Tables 6.3 and female labor force participation rates from ILO LABORSTA. Note: Projections of per capita GDP assume that the 1990–2007 annual growth rates will continue through 2030. The per capita GDP data used are 2005 purchasing power parity estimates from the Penn World Tables 6.3. Annex 2E Poverty rates and the number of working poor in South Asia TABLE 2E.1 Percentage of workers in households below the poverty line in Bangladesh, by employment type, 2000–10 2000 2005 2010 Type of Poverty Lower Upper Poverty Lower Upper Poverty Lower Upper employment rate bound bound rate bound bound rate bound bound Rural agricultural regular wage or salaried 39.6 29.5 49.7 58.8 48.8 68.8 54.1 42.5 65.7 Rural agricultural self-employed 39.5 37.4 41.6 29.7 28.0 31.4 22.0 20.3 23.6 Rural agricultural casual labor 72.1 69.9 74.3 66.2 63.9 68.5 51.0 48.7 53.3 Rural nonagricultural regular wage or salaried 33.8 30.7 37.0 25.7 23.2 28.2 26.2 23.9 28.5 Rural nonagricultural self-employed 41.1 38.2 44.0 32.9 30.3 35.5 26.1 24.0 28.2 Rural nonagricultural casual labor 60.9 57.6 64.1 55.7 53.0 58.4 44.8 42.4 47.1 Urban regular wage or salaried 25.7 23.4 28.0 19.2 17.3 21.1 15.6 14.1 17.1 Urban self-employed 29.3 26.8 31.9 18.6 16.9 20.4 13.6 12.1 15.2 Urban casual labor 62.2 58.8 65.6 55.5 52.8 58.2 44.1 41.6 46.6 Regular wage or salaried 30.5 28.6 32.3 23.6 22.1 25.2 20.9 19.6 22.2 All self-employed 38.3 36.9 39.7 28.2 27.0 29.3 21.6 20.5 22.6 All casual labor 67.4 65.8 69.0 60.6 59.1 62.0 47.3 45.9 48.7 All rural 50.0 48.8 51.2 41.7 40.6 42.7 33.8 32.8 34.8 All urban 34.9 33.3 36.5 26.8 25.6 28.0 21.0 20.0 22.1 All workers 46.8 45.8 47.7 37.7 36.9 38.5 30.0 29.3 30.8 Source: Authors, based on data from labor force and household surveys. Note: Lower and upper bounds in the table refer to the 95 percent con�dence intervals of the poverty rate estimate by employment type. 75 76 MORE AND BETTER JOBS IN SOUTH ASIA TABLE 2E.2 Number of working poor in Bangladesh, by employment type, 2000–10 Percentage change in number of working poor Number of working poor (thousands) 2000–05 2005–10 Type of employment 2000 2005 2010 Total Annual Total Annual Rural agricultural regular wage or salaried 186 244 162 31 6 –50 –8 Rural agricultural self-employed 4,025 3,555 2,177 –12 –2 –63 –9 Rural agricultural casual labor 6,036 4,938 4,018 –18 –4 –23 –4 Rural nonagricultural regular wage or salaried 1,478 1,284 1,601 –13 –3 20 5 Rural nonagricultural self-employed 2,293 1,851 1,881 –19 –4 2 0 Rural nonagricultural casual labor 2,687 3,205 3,181 19 4 –1 0 Urban regular wage or salaried 1,017 1,092 1,267 7 1 14 3 Urban self-employed 962 877 617 –9 –2 –42 –7 Urban casual labor 1,249 1,564 1,484 25 5 –5 –1 Regular wage or salaried 2,681 2,620 3,030 –2 0 14 3 Self-employed 7,280 6,282 4,675 –14 –3 –34 –6 Casual labor 9,972 9,707 8,683 –3 –1 –12 –2 All rural 16,704 15,076 13,020 –10 –2 –16 –3 All urban 3,228 3,534 3,369 9 2 –5 –1 All workers 19,932 18,609 16,389 –7 –1 –14 –3 Source: Authors, based on data from labor force and household surveys Note: The number of working poor for each employment type is calculated by multiplying the estimated total employment (estimated in �ve yearly intervals) by the share of employment and poverty rate by employment type from the closest year of the household survey. For example, for the number of working poor in India in 1985, the share of employment and poverty rate by employment type used was estimated from the 1983 labor force survey. TABLE 2E.3 Percentage of workers in households below the poverty line in India, by employment type, 1983–2004/05 1983 1993/94 1999/2000 2004/05 Poverty Lower Upper Poverty Lower Upper Poverty Lower Upper Poverty Lower Upper Type of employment rate bound bound rate bound bound rate bound bound rate bound bound Rural agricultural regular wage or salaried 52.7 51.0 54.5 43.3 40.5 46.1 37.0 34.8 39.2 26.6 24.1 29.2 Rural agricultural self-employed 37.1 36.7 37.4 28.6 28.3 29.0 24.6 24.3 24.9 17.0 16.7 17.3 Rural agricultural casual labor 60.6 60.0 61.1 47.8 47.2 48.4 43.2 42.6 43.8 30.8 30.1 31.4 Rural nonagricultural regular wage or salaried 25.2 24.2 26.1 13.6 12.9 14.2 11.0 10.4 11.6 8.4 8.0 8.8 Rural nonagricultural self-employed 41.2 40.4 42.0 27.2 26.4 27.9 26.3 25.7 27.0 17.6 17.1 18.0 Rural nonagricultural casual labor 47.0 45.9 48.1 38.0 36.9 39.1 30.2 29.2 31.2 24.7 23.9 25.5 Urban regular wage or salaried 26.2 25.6 26.7 17.8 17.3 18.2 15.6 15.1 16.0 13.7 13.3 14.1 Urban self-employed 42.8 42.2 43.4 32.3 31.8 32.9 28.2 27.7 28.7 24.3 23.8 24.8 Urban casual labor 59.9 58.9 60.9 57.1 56.1 58.0 51.5 50.6 52.4 45.5 44.5 46.4 Regular wage or salaried 28.8 28.3 29.2 17.8 17.4 18.2 15.5 15.1 15.8 12.3 12.0 12.6 Self-employed 38.6 38.3 38.9 29.0 28.8 29.3 25.5 25.3 25.8 18.5 18.3 18.7 Casual labor 58.2 57.8 58.7 47.4 46.9 47.9 42.0 41.6 42.4 31.2 30.7 31.6 All rural 43.7 43.5 44.0 33.4 33.2 33.7 29.8 29.5 30.0 20.3 20.1 20.5 All urban 38.8 38.4 39.2 30.7 30.4 31.1 27.2 26.9 27.5 23.1 22.8 23.4 All workers 42.6 42.4 42.9 32.8 32.6 33.1 29.2 29.0 29.4 20.9 20.8 21.1 Source: Authors, based on data from labor force and household surveys. Note: Lower and upper bounds in the table refer to the 95 percent con�dence intervals of the poverty rate estimate by employment type. 77 78 MORE AND BETTER JOBS IN SOUTH ASIA TABLE 2E.4 Number of working poor in India, by employment type, 1985–2005 Percentage change in number of working poor Number of working poor (thousands) 1985–95 2000–05 Type of employment 1985 1995 2000 2005 Total Annual Total Annual Rural agricultural regular wage or salaried 2,290 1,177 1,320 639 –49 –6 –52 –13 Rural agricultural self-employed 36,916 36,007 31,661 25,639 –2 0 –19 –4 Rural agricultural casual labor 30,943 33,898 34,635 22,177 10 1 –36 –9 Rural nonagricultural regular wage or salaried 2,974 2,209 1,965 1,943 –26 –3 –1 0 Rural nonagricultural self-employed 9,156 8,553 8,807 8,302 –7 –1 –6 –1 Rural nonagricultural casual labor 5,641 6,336 6,079 6,701 12 1 10 2 Urban regular wage or salaried 6,210 5,484 5,355 5,438 –12 –1 2 0 Urban self-employed 9,453 10,192 9,772 10,806 8 1 11 2 Urban casual labor 5,988 7,597 7,757 6,514 27 2 –16 –3 Regular wage or salaried 11,474 8,870 8,641 8,021 –23 –3 –7 –1 Self-employed 55,525 54,751 50,240 44,746 –1 0 –11 –2 Casual labor 42,571 47,831 48,470 35,392 12 1 –27 –6 All rural 87,919 88,181 84,467 65,401 0 0 –23 –5 All urban 21,651 23,272 22,884 22,758 7 1 –1 0 All workers 109,570 111,453 107,351 88,159 2 0 –18 –4 Source: Authors, based on data from labor force and household surveys. Note: The number of working poor for each employment type is calculated by multiplying the estimated total employment (estimated in �ve yearly intervals) by the share of employment and poverty rate by employment type from the closest year of the household survey. For example, for the number of working poor in India in 1985, the share of employment and poverty rate by employment type used was estimated from the 1983 labor force survey. TABLE 2E.5 Percentage of workers in households below the poverty line in Nepal, by employment type, 1995/96 and 2003/04 1995/96 2003/04 Poverty Lower Upper Poverty Lower Upper Type of employment rate bound bound rate bound bound Rural agricultural regular wage or salaried 50.9 41.8 59.9 30.1 4.9 55.3 Rural agricultural self-employed 39.7 38.2 41.1 30.7 29.4 32.0 Rural agricultural casual labor 50.2 46.3 54.2 50.0 45.7 54.3 Rural nonagricultural regular wage or salaried 21.2 15.4 27.0 10.0 6.3 13.7 Rural nonagricultural self-employed 30.4 25.9 34.8 16.7 13.3 20.0 Rural nonagricultural casual labor 44.8 39.5 50.0 43.4 37.8 49.1 Urban regular wage or salaried 9.7 6.6 12.8 3.3 1.9 4.8 Urban self-employed 18.4 15.6 21.2 8.9 7.4 10.4 Urban casual labor 45.2 38.2 52.1 16.1 11.7 20.6 Regular wage or salaried 27.6 24.2 31.0 7.4 5.6 9.1 Self-employed 38.1 36.8 39.3 27.3 26.2 28.3 Casual labor 48.1 45.2 51.0 43.7 40.8 46.7 All rural 40.0 38.8 41.2 31.1 29.9 32.2 All urban 21.0 18.7 23.2 8.3 7.2 9.5 All workers 39.0 37.9 40.1 28.0 27.1 29.0 Source: Authors, based on data from labor force and household surveys. Note: Lower and upper bounds in the table refer to the 95 percent con�dence intervals of the poverty rate estimate by employment type. GROWTH AND JOB QUALITY IN SOUTH ASIA 79 TABLE 2E.6 Number of working poor in Nepal, by employment type, 1995–2005 Percentage change in number of working poor Number of working poor (thousands) 1995–2005 Type of employment 1995 2005 Total Annual Rural agricultural regular wage or salaried 92 9 –91 –21 Rural agricultural self-employed 2,456 2,464 0 0 Rural agricultural casual labor 432 463 7 1 Rural nonagricultural regular wage or salaried 60 41 –31 –4 Rural nonagricultural self-employed 181 136 –25 –3 Rural nonagricultural casual labor 207 215 4 0 Urban regular wage or salaried 13 14 11 1 Urban self-employed 50 89 79 6 Urban casual labor 41 33 –20 –2 Regular wage or salaried 165 65 –61 –9 Self-employed 2,687 2,689 0 0 Casual labor 680 711 5 0 All rural 3,427 3,328 –3 0 All urban 104 137 31 3 All workers 3,531 3,465 –2 0 Source: Authors, based on data from labor force and household surveys Note: The number of working poor for each employment type is calculated by multiplying the estimated total employment (estimated in �ve yearly intervals) by the share of employment and poverty rate by employment type from the closest year of the household survey. For example, for the number of working poor in India in 1985, the share of employment and poverty rate by employment type used was estimated from the 1983 labor force survey. Annex 2F Analysis of poverty and unemployment in India A more complete analysis of the links In 1983 and 1993/94, the same house- between poverty and unemployment is pos- holds were sampled for the employment and sible in India, where time series data for a unemployment surveys and the consumption longer period are available. Some data com- expenditure survey. The household consump- parability issues merit attention, however. tion expenditure aggregate in the employment and unemployment surveys is the unabridged uniform recall period measure from the con- Data used sumption expenditure survey. These two Data from the employment and unemploy- datasets are comparable. ment surveys by the India National Sample In 1999/2000 and 2004/05, the employ- Survey—which are equivalent to labor force ment and unemployment surveys and surveys—were used for 1983, 1993/04, consumption expenditure survey sampled dif- 1999/2000, and 2004/05. The old of� cial ferent households. The consumption expendi- state level poverty lines for the same time ture survey used unabridged uniform recall periods were used for the analysis, because period and mixed recall period consumption they are available for all survey years. measures. The consumption module in the Employment and unemployment surveys employment and unemployment surveys used were used instead of consumption expendi- an abridged (fewer questions) mixed recall ture surveys because they contain both con- period to measure the household consump- sumption expenditure and labor force vari- tion expenditure aggregate. ables, which allow the estimation of poverty The estimates of poverty rates using rates by employment types—something the abridged mixed recall period in these two consumption expenditure survey does not. employment and unemployment surveys are 80 MORE AND BETTER JOBS IN SOUTH ASIA comparable with each other but not compa- and 2004/05 show a significant decline in rable with the 1983 and 1993/94 unabridged poverty rates for the whole population, from uniform recall period measures. The abridged 36.0 percent to 27.5 percent. It is highly mixed recall period measures should not be unlikely that poverty increased between interpreted as accurate estimates of poverty 1993/94 and 1999/2000, as this would imply rates, because mixed recall period measures an even higher percentage decline between tend to generate lower poverty rates than 1999/2000 and 2004/05. It is also unlikely uniform recall period measures and abridge- that during a period in which poverty rates ment creates differences in the estimates declined signi�cantly for the whole popula- using unabridged measures. Therefore, the tion, worker poverty rates would not also poverty estimates from the 1999/2000 and have fallen. 2004/05 employment and unemployment surveys should only be used to look at trends Categories of Employment by employment type. Given the changes in the consumption Nine categories of employment are consid- measure, the estimated worker poverty rates ered, based on the main activity of each of 1993/94 and 1999/2000 cannot be com- working member of a household. They pared to determine whether poverty declined. include regular wage or salaried, self– However, official poverty measures using employed, and casual labor in the rural agri- unabridged uniform recall period from the cultural, rural nonagricultural, and urban consumption expenditure survey in 1993/94 sectors. TABLE 2F.1 Official and authors’ estimated poverty rates for urban, rural, and all workers in India, 1983–2004/05 Year/estimate Urban workers Rural workers All workers 1983 Official estimates (based on unabridged uniform recall period from consumption expenditure survey) 40.8 45.7 44.5 Authors’ estimates (based on unabridged uniform recall period from employment and unemployment survey/consumption expenditure survey) 38.8 43.7 42.6 1993/94 Official estimates (based on unabridged uniform recall period from consumption expenditure survey) 32.4 37.3 36.0 Authors’ estimates (based on unabridged uniform recall period from employment and unemployment survey/consumption expenditure survey) 30.7 33.4 32.8 1999/2000 Official estimates (based on unabridged mixed recall period from consumption expenditure survey) 23.6 27.1 26.1 Authors’ estimates (based on abridged mixed recall period from employment and unemployment survey) 27.2 29.8 29.2 2004/05 Official estimates (based on unabridged uniform recall period from consumption expenditure survey) 25.7 28.3 27.5 Official estimates (based on unabridged mixed recall period from consumption expenditure survey) 21.7 21.8 21.8 Authors’ estimates (based on abridged mixed recall period from employment and unemployment survey) 23.1 20.3 20.9 Source: Authors, based on data from national labor force and household surveys. Note: Official estimates include the entire population living in households below the poverty line; authors’ estimates include only people age 15–64. The unabridged mixed recall period measures from the consumption expenditure surveys conducted in 1999/2000 and 2004/05 are roughly but not strictly comparable, because of differences in design. GROWTH AND JOB QUALITY IN SOUTH ASIA 81 Sample growth accounting to be conducted only for Bangladesh, India, Pakistan, and Sri Lanka. The sample includes 209,223 employed indi- 3. The �gures for South Asia are GDP at purchas- viduals in 1993/94, 213,986 in 1999/2000, ing power parity–weighted averages of the �g- and 228,244 in 2004/05, about 30 percent ures for Bangladesh, India, Pakistan, and Sri of whom are women. Rural agricultural reg- Lanka and are therefore weighted most heav- ular wage or salaried workers represent just ily by India. For further details, see annex 2B, 1.7 percent of the sample in 1983 and 0.6 which contains sources of growth by region percent in 2003/04. Given this small sample per decade between 1960 and 2008. 4. Young (1994, 1995) highlights the relatively size, estimated poverty rates for this type of limited role of TFP growth in East Asia less worker should not be treated as reliable. China during its years of rapid investment-led In 1983 and 1993/94, the headcount pov- growth. erty rates from the employment and unem- 5. Bosworth, Collins, and Virmani (2007) ployment surveys were comparable to of�cial develop this point in the context of India. poverty rates, as the consumption expendi- 6. This discussion draws on Bloom, Canning, ture survey sampled the same households and Rosenberg (2011). (table 2F.1). The differences reflect the fact 7. “Potential� is used because without enabling that the estimates are for workers and not the policies, a rising ratio of the working-age whole population. population to the nonworking-age population Between 1999/2000 and 2004/05, offi- will not necessarily boost economic growth. 8. Reallocation during a period is calculated as cial estimates using the unabridged mixed a residual by subtracting the weighted sum recall period measure from the consumption of TFP growth in each sector (agriculture, expenditure survey also show a decline in industry, services) from aggregate TFP growth poverty. The differences between the of�cial during that period, where the weights are the poverty estimates and the authors’ estimates share of each sector in GDP at the beginning reflect the fact that (a) the authors’ estimates of the period. refer to workers and not the whole popula- 9. The choice of countries is dictated by the tion and (b) the mixed recall period measure availability of sectorally disaggregated growth is unabridged in the consumption expendi- accounts. The Thailand numbers are from ture survey and abridged in the employment Bosworth (2005); data from China were and unemployment surveys. The compari- adapted by the authors from Bosworth and Collins (2008). son of the mixed recall period and uniform 10. The ratio of employment to the working- recall period measures from the consumption age population can be written as follows: expenditure survey in 2004/05 also shows (employment/labor force) (labor force/ that the mixed recall period measures esti- working-age population). The �rst term is (1 mate lower poverty rates than the uniform – the unemployment rate); the second is the recall period. Hence, it is not possible to com- labor force participation rate. pare the results from the 1983 and 1993/94 11. Afghanistan and Bhutan are excluded, because employment and unemployment surveys with 2000 data are not available. Including them the 1999/2000 and 2004/05 employment and in 2010 would bring total employment to unemployment surveys. 577 million in 2010. 12. The employment status of workers is de�ned here on the basis of questions on current Notes weekly status (generally the past seven days) in the labor force surveys of Bangladesh, 1. Europe and Central Asia, (not shown in the Bhutan, India, Nepal, Pakistan, and Sri �gure) grew faster than South Asia during Lanka and on the basis of employment status the 2000s, but this reflected, in part, recovery in the last month in the labor force surveys from a transition recession following the exit of Afghanistan and Maldives (where weekly from the command economy. status questions were not included). 2. The decomposition is formally presented 13. Various additional dimensions of job qual- in annex 2A. Data constraints allowed ity are often cited, such as access to nonwage 82 MORE AND BETTER JOBS IN SOUTH ASIA bene�ts and public social protection mecha- 22. The details of the methodology are described nisms, the ability to upgrade skills and receive in appendix B. training on the job, and a safe working envi- 23. This annex is based on Collins (2007). ronment. There is typically a strong correlation 24. The household survey–based participation between better jobs as de�ned by the criteria rates are used to maintain consistency with used in this book and many of these additional the methodology used for the projections. dimensions. ILO estimates of the labor force in 2010 are 14. The use of this proxy for the self-employed slightly different from the estimates developed was proposed in the Indian context by here. The ILO estimate of the total labor force Sundaram (2004). in South Asia in 2010 is about 27 million 15. The household survey data for Pakistan and Sri higher, with an annual growth rate between Lanka used for the poverty-employment anal- 1990 and 2010 of 2.3 percent, compared ysis do not distinguish between casual work- with the 2.1 percent posited in this book. ers and regular wage and salaried workers. 16. This trend prompted an investigation into the cost of living in urban India and a recent revi- References sion of the of�cial poverty lines. Aggarwal, S. 2010. “Labor Input and Its Compo- 17. This question was based not on a reference sition: An Industry–Level Perspective.� Paper week but on a reference year. A worker’s usual presented at the Worldklems conference, Cam- principal activity is determined by the activity bridge, MA, August. the worker spent most of his or her time doing Bloom, D., D. Canning, and L. Rosenberg. 2011. in the year preceding the survey. Any activ- “Demographic Change and Economic Growth.� ity other than the principal status constitutes In Reshaping Tomorrow, ed. E. Ghani. New a worker’s subsidiary status. “Usual’� status Delhi: Oxford University Press. workers include principal status workers (who Bosworth, B. 2005. Economic Growth in Thai- spent most of their time employed or looking land: The Macroeconomic Context. Brookings for jobs) and subsidiary workers (who spent Institution, Washington, DC. part of their time working or looking for jobs ———. 2010. “Update of Bosworth and Collins, in the year preceding the survey). 2003.� Unpublished paper, World Bank, Wash- 18. An exception is Nepal, where the poverty rate ington, D.C. for urban casual workers is signi�cantly lower Bosworth, B., and S. Collins, 2003. “The Empir- than the rate for the rural self-employed, ics of Growth: An Update.� Brookings Papers which is driven by the subsegment of urban on Economic Activity (2): 113–206. casual labor employed in short-term contract ———. 2008. “Accounting for Growth: Compar- work. This group is more educated than daily ing China and India.� Journal of Economic wage workers and has poverty rates that are Perspectives 22 (1): 45–66. closer to those of urban regular wage or sala- ———. 2010. “Update of Bosworth and Collins ried workers. 2003.� Unpublished notes for this book. 19. The number of self-employed workers declined Bosworth, B., S. Collins, and A. Virmani 2007. between 2004/05 and 2009/10. Casual labor “Sources of Growth in the Indian Economy.� accounted for nearly 80 percent of net addi- NBER Working Paper 12901, National Bureau tional employment in those sectors that of Economic Research, Cambridge, MA. expanded employment between 2004/05 and Collins, S. 2007. “Economic Growth in South 2009/10 (the remaining net additional employ- Asia: A Growth Accounting Perspective.� In ment was in regular wage or salaried jobs). Growth and Regional Integration in South 20. This trend toward casualization of the rural Asia, ed. S. Ahmed and E. Ghani, 45– 60. nonfarm labor force was highlighted in the Delhi: Macmillan India Ltd. World Bank’s India poverty assessment for Easterly, W., M. Kremer, L. Pritchett, and L. 1983–2004/05 (World Bank 2011b). Summers. 1993. “Good Policy or Good Luck? 21. The number of working poor in India is likely Country Growth Performance and Temporary to have declined between 1985 and 2005. Shocks.� Journal of Monetary Economics Annex 2F explains why poverty rates for the 32 (3): 459–83. �rst and second subperiods cannot be directly ILO (International Labour Of�ce). 2010. KILM compared. and LABORSTA databases. Geneva. GROWTH AND JOB QUALITY IN SOUTH ASIA 83 Lanjouw, P., J. Luoto, and D. Mckenzie. 2011. ———. 2011a. Large-Scale Migration and Remit- “Using Repeated Cross–Sections to Explore tances in Nepal: Issues, Challenges and Movements in and out of Poverty.� Policy Opportunities. Washington, DC. Research Working Paper 5550, World Bank, ———. 2011b. Perspectives on Poverty in India: Washington, DC. Stylized Facts from Survey Data. Washington, Sundaram, K. 2004. “Growth of Work Oppor- DC. tunities in India: 1983 to 1999–2000.� Paper ———. 2011c. World Development Indicators. presented at a conference in honor of K. N. Raj Washington, DC. on Planning, Institutions, Markets and Devel- Young, A. 1994. “Lessons from the East Asian opment, Thrissur, Kerala, India, October. NICs: A Contrarian View.� European Eco- UN (United Nations). 2008. World Population nomic Review 38: 964–73. Prospects: The 2008 Revision. New York: ———. 1995. “The Tyranny of Numbers: Con- United Nations. fronting the Statistical Realities of the East ———. 2010. World Population Prospects: The Asian Growth Experience.� Quarterly Journal 2010 Revision. New York: United Nations. of Economics 110 (August): 641–80. World Bank. 1993. The East Asian Miracle. Washington, DC. CHAPTER 3 A Pro�le of South Asia at Work Questions and Findings Questions agriculture. Faster intersectoral reallocation of employment into industry and services will • What are they key features of labor markets require the development of not just the urban in South Asia? industrial and services sectors but also the • Where are the better jobs, and who holds rural nonfarm sector. them? • Within industry and services, better jobs are • What are the implications for the region’s with large formal �rms. The majority of work- employment challenges? ers, however, work in informal micro �rms, where value added per worker is lower and Findings which pay lower wages. Creation of better • Although employment in South Asia has been jobs will require faster intrasectoral reallo- expanding, employment rates have remained cation of labor from lower-productivity— steady and are below those in other regions, typically micro and small informal—�rms to as a result of persistently low female employ- higher-productivity—typically medium and ment and participation rates. large formal—�rms within manufacturing and • The majority of workers in the region are services. still engaged in agriculture. Self-employment • The educated are more likely to work outside is the predominant type of employment, and agriculture and be employed in regular wage a high share of wage employment is casual or salaried work. Female workers are less labor. Thus, the vast majority of work in likely to be in better jobs than men, except at South Asia—86–95 percent of total employ- the highest levels of education; they also earn ment and 71–81 percent of nonagricultural less, even after controlling for differences in employment in most countries—is informal educational attainment. Members of ethnic in nature. This picture is unlikely to change minorities are less likely to hold better jobs; signi�cantly in the short to medium term. they also earn less, although much of this dif- • Jobs that pay higher wages and are associated ferential can be explained by differences in with lower poverty rates are found outside of educational attainment. A Profile of South Asia at Work 3 T his chapter profiles employment in conduct labor force surveys (Bangladesh, South Asia. Relying on household India, Nepal, Pakistan, and Sri Lanka), mea- survey data from the region’s eight surement of labor market indicators such as countries, it describes the patterns of par- labor force participation, employment, and ticipation, employment, unemployment, and unemployment is common and generally con- earnings in the region. sistent with international standards. In coun- Describing the labor market in South tries in which other household surveys are Asia is a formidable task. The region’s eight used (Afghanistan, Bhutan, and Maldives), countries vary widely in size, ranging from de�nitions of these (and other) indicators can less than 1 million people each in Bhutan differ from international norms. As a result, and Maldives to 1.2 billion people—about measurement differences explain some of the three-quarters of South Asia’s population—in variation across countries presented in this India. There is diversity in the stages of devel- chapter (Srinivasan 2010 discusses in further opment, economic structures, social and cul- detail how labor market concepts are mea- tural characteristics, and conflict. Even within sured in different surveys). (Annex table 3A.1 countries there is signi�cant diversity. provides more detail on the measurement The profile of South Asia at work pre- of employment and unemployment from sented is based primarily on microlevel data the national surveys as used in this book.) collected by national statistical agencies. Second, as South Asian economies are still The analysis relies on labor force surveys in heavily rural, agricultural, and informal, the some countries and on living standards sur- productive activities of many individuals may veys in others (depending on survey avail- not be fully captured by standard labor mar- ability and data quality). The latest surveys ket indicators. were conducted between 2004 and 2009/10 This chapter is organized as follows. The (see appendix table A.1). first section provides an overview of the Two caveats should be noted regarding main labor market trends, including employ- analysis across countries. First, there are ment, unemployment, and labor force par- limits to the standardization that is pos- ticipation, for the eight countries in South sible, especially between labor force and Asia, with a focus on the employment and living standards surveys. In countries that participation patterns of women. The second 85 86 MORE AND BETTER JOBS IN SOUTH ASIA section takes a closer look at the nature of (2004) to 80 percent in Nepal (2008) (the employment in the region, including loca- other countries in the region have employment tion, sector, employment status, and infor- rates of 50–65 percent). Analysis within coun- mality. The third section examines where tries shows moderate differences in regional the better jobs are. The last section analyses employment rates within countries (see annex how gender, caste/ethnicity, and education 3B). Internationally, the average employment are correlated with access to better jobs. rate is 60–70 percent for low- and lower- middle-income countries (figure 3.2). The employment rate in the three largest countries Overview of employment and in South Asia (India, Bangladesh, and Paki- labor force participation in stan) is signi�cantly below the average rate for South Asia countries at similar levels of development. This section � rst examines employment in These relatively low employment rates in the region. It then addresses labor force par- South Asia refl ect persistently low female ticipation and unemployment. employment rates in all countries except Bhutan and Nepal (figure 3.3). Employ- ment rates among men are not low by Employment international standards. The (unweighted) Total employment in South Asia is estimated national average for male employment in at 574 million in 2010, with India account- South Asia is 77 percent, which is almost ing for 75 percent, Bangladesh 10 percent, identical to the male average for compara- and Pakistan 9 percent of employment in the tor countries ( Bolivia, Cambodia, China, region (�gure 3.1). In the region as a whole, Ghana, Guatemala, Indonesia, Lao Peo- 55 percent of the 1.04 billion working-age ple’s Democratic Republic, Nigeria, and population is employed. the Philippines). In contrast, the aver- Employment rates are low by international age employment rate for women in South standards in all countries except Bhutan and Asia is 21 percentage points lower than in Nepal. Employment growth looks favorable comparator countries. The male-female because of the region’s growing working-age employment rate ratio is 2.2 in the region population, as discussed in chapter 2. The and just 1.3 in comparator countries. picture is less positive in terms of employ- There is no consistent evidence of an ment rates. Employment rates among people upward trend in employment rates in South age 15–64 range from 48 percent in Maldives Asian countries (�gure 3.4). Total employment FIGURE 3.1 Total employment in South Asia, by country, 2010 1,000,000 432,497 100,000 54,103 54,155 14,694 workers (thousands) 7,894 9,666 10,000 1,000 322 90 100 10 1 Maldives Bhutan Sri Lanka Afghanistan Nepal Pakistan Bangladesh India Sources: Authors, based on working-age population �gures from UN 2010 and employment rate data from national labor force surveys. A PROFILE OF SOUTH ASIA AT WORK 87 FIGURE 3.2 Employment rates in lower- and lower-middle-income countries 90 80 Nepal 70 Bhutan 60 Afghanistan Bangladesh India Sri Lanka percent 50 Pakistan Maldives 40 30 20 10 0 0 2,000 4,000 6,000 8,000 10,000 12,000 14,000 2008 gross national income per capita in purchasing power parity dollars Source: Authors, based on data from World Bank 2011b and national labor force and household surveys. Note: Employment rates are for population age 15 years and above. For all countries, gross national income per capita in 2008 is adjusted for purchasing power parity. Employment rates for countries in South Asia are for latest survey year; employment rates for other countries are for 2008. FIGURE 3.3 Male and female employment rates in South Asia, by country 100 85 86 83 81 80 79 78 80 75 67 61 60 percent 44 40 38 35 29 29 22 20 0 Afghanistan Bangladesh Bhutan India Maldives Nepal Pakistan Sri Lanka 2008 2009 2007 2010 2004 2008 2009 2008 male female Source: Authors, based on data from national labor force and household surveys. rates increased in Maldives and Pakistan, Pakistan, declined in Bhutan and India, and declined moderately in India and Nepal and changed little in the other countries. signi�cantly in Bhutan, and remained fairly These employment figures are for the constant in Bangladesh and Sri Lanka.1 These working-age population (15–64). Child labor, trends mirrored those of female employ- which this book does not address, remains an ment rates, which increased in Maldives and important aspect of the overall employment 88 MORE AND BETTER JOBS IN SOUTH ASIA FIGURE 3.4 Trends in employment rates in South Asia, by country 100 90 85 82 80 74 70 67 64 60 56 57 58 57 55 56 percent 52 50 48 48 44 40 30 20 10 0 2008 2002 2009 2003 2007 2000 2010 1998 2004 1999 2008 2000 2009 2000 2008 Afghanistan Bangladesh Bhutan India Maldives Nepal Pakistan Sri Lanka Source: Authors, based on data from national labor force and household surveys. Note: Trend analysis does not take into account cyclical factors. Although the analysis attempted to use standard, consistent de�nitions of employment over time, differences may reflect differences in the questions used to de�ne employment in different survey rounds (see annex table 3A.1 for details). picture in South Asia, as it is in many parts of Underemployment is conventionally de� ned the developing world (box 3.1). as working fewer hours than desired in mature labor markets. This may not be an appropriate definition in developing coun- Labor Force Participation and tries, where people often work for long hours Unemployment even if earnings are very low. In addition, Employment rates in South Asia closely data limitations do not permit a consistent track labor force par ticipation rates, estimate of underemployment. because measured unemployment is very Estimates of the magnitude of underem- low in most countries in the region (table ployment in South Asia vary, based on dif- 3.1). Open unemployment rates in low- ferent definitions. Underemployment was income countries tend to be low, even if estimated at 48 percent of the total workforce labor market conditions are unattractive. in Afghanistan in 2008 (Islamic Republic For the region as a whole, 3.2 percent of the of Afghanistan and World Bank 2010) and labor force—19 million people in a labor 24.5 percent in Bangladesh in 2006 (Rahman force of 593 million—was unemployed in 2008). These �gures are based on a de�nition 2010. The reported unemployment rate that classi�es as underemployed workers who was high only in Maldives (15.3 percent), work 35 or fewer hours a week on average. In where it mainly reflects the methodology India one measure used by the National Sam- used to calculate unemployment. 2 Unem- ple Survey organization (which de�nes under- ployment in other countries ranged from employment as the proportion of the usually 1.1 percent in Pakistan to 5.6 percent in Sri employed according to the usual status crite- Lanka. In Bangladesh, Sri Lanka, and espe- ria not employed the previous week) estimates cially Maldives, women have higher unem- underemployment at 9–17 percent for women ployment rates than men. In the rest of the and 2–4 percent for men in 2004/05 (Gov- region, there is little gender difference. ernment of India 2006). One problem with Although open unemployment is low in these measures is that they may overestimate South Asia, underemployment—the under- underemployment, because they do not take utilization of labor—may be prevalent. into account individuals who did not wish to A PROFILE OF SOUTH ASIA AT WORK 89 BOX 3.1 Child labor in South Asia According to the International Labour Organization 10 children between the ages of 5 and 14 working. (ILO 2010), 215 million children between the ages of Significant numbers of children are also working 5 and 14 were working in South Asia in 2008, with in Afghanistan, Bhutan, and Pakistan. Bangladesh, 115 million engaged in hazardous work. These �g- India, and Sri Lanka report lower incidences. ures indicate that 13 percent of all working children Additional dimensions of the statistical picture of in the world live in South Asia; among these children, child labor in South Asia include the following: 7 percent are engaged in hazardous work. According to these �gures, the incidence of child labor in Asia is • In most countries, boys are somewhat more likely the second highest of all regions, behind Africa. (The than girls to work. However, in Bhutan and Nepal, ILO statistics do not separate South Asia from the two countries with high child labor rates, employ- rest of the continent). ment is higher among girls. Child labor has long played an important role in • Child labor is much more prevalent in rural areas many traditional and agriculturally based societies. than in cities. The vast majority of working chil- It can also be a product of poverty and inequality, dren are engaged in agriculture and �shing. Other poor education, and confl ict. In South Asia, as in sectors with some child labor are commerce (retail some other regions, concerns about child labor are trade) and manufacturing. heightened by the presence of practices such as child • Although some children are employed as wage traf�cking and bonded child labor (ILO 2010). workers (in Bangladesh and, to a lesser extent, in How prevalent is child labor in South Asia? Many India), most are household enterprise workers. In of the surveys used for this book include questions Nepal, for example, 96 percent of working children that provide data on the incidence of child labor. work in household enterprises. The employment rates for children can be computed • The incidence of child labor continues to decline in the same way they have been calculated for the gradually in Bangladesh, India, and Sri Lanka. working-age population. However, surveys differ in In Nepal and Pakistan, where the incidence of their age coverage: in some countries, employment child labor is higher, there is no clear evidence of rates can be computed for the 5- to 14-year age group; decreases in child labor over time. in others, surveys do not cover children under 10. Survey evidence suggests that the incidence of The nefarious effects of child labor can be miti- child labor varies across the region (box table 3.1.1). gated when working children continue their studies. In Nepal has the highest incidence, with about 3 in Nepal, for example, the country with the highest child BOX TABLE 3.1.1 Incidence of child labor in South Asia, by age group and country (percentage of age group) Age group Country Year 5–9 10–14 5–14 Afghanistan 2008 8.5 23.2 16.1 Bangladesh 2005 0.7 7.1 4.1 Bhutan 2003 — 19.7 — India 2008 0.2 3.4 1.8 Nepal 2008 10.9 47.0 29.7 Pakistan 2009 — 11.8 — Sri Lanka 2008 — 1.1 — Source: Authors, based on data from national labor force and household surveys. Note: — = Not available. No data on child labor are available for Maldives. (continues next page) 90 MORE AND BETTER JOBS IN SOUTH ASIA BOX 3.1 Child labor in South Asia (continued) labor rates in the region, almost 90 percent of chil- and other harmful forms of employment. The three dren who are working also attend school. In contrast, most important conventions are the UN Convention although a smaller percentage of children in Afghani- on the Rights of the Child and ILO Conventions 138 stan, Bangladesh, Bhutan, India, and Pakistan work, (minimum age of employment) and 182 (Elimination most are not in school (box �gure 3.1.1). of the Worst Forms of Child Labor). All eight coun- The international community has passed a num- tries in South Asia have rati�ed the UN Convention ber of conventions designed to protect the rights of on the Rights of the Child; only Afghanistan, Nepal, children in the labor market, through both mini- Pakistan, and Sri Lanka have rati�ed the two ILO mum working ages and protection from hazardous conventions (see chapter 6). BOX FIGURE 3.1.1 Percentage of child workers attending school in South Asia, by age group and country 100 14 13 90 21 80 53 70 68 70 60 84 percent 86 94 91 50 99 86 87 40 79 30 47 20 32 30 10 16 14 6 9 0 1 20 tan 20 esh 20 dia 20 pal 20 tan 20 esh 20 tan 20 dia 20 pal 20 tan 20 nka In In Ne Ne 08 05 05 08 08 05 03 05 08 09 08 is is u kis lad lad a Bh iL an an Pa ng ng Sr gh gh Ba Ba Af Af age 5–9 age 10–14 attending school not attending school Source: Authors, based on data from national labor force and household surveys. Note: Data on 5- to 9-year-olds are not available for Bhutan, Pakistan, and Sri Lanka. TABLE 3.1 Male and female labor force participation, employment, and unemployment rates in South Asia, by country Participation rate Employment rate Unemployment rate Country Year All Male Female All Male Female All Male Female Afghanistan 2008 65 85 45 64 83 44 2 3 2 Bangladesh 2009 60 89 31 57 85 29 5 4 8 Bhutan 2007 69 76 62 67 75 61 2 2 2 India 2010 57 82 30 55 80 29 3 3 4 Maldives 2004 57 72 46 48 67 35 15 8 24 Nepal 2008 84 88 80 82 86 79 2 2 1 Pakistan 2009 52 82 22 52 81 22 1 1 1 Sri Lanka 2008 60 81 41 57 78 38 6 4 8 Source: Authors, based on data from national labor force and household surveys. Note: See annex table 3A.1 for de�nition of employment and unemployment used in each country. The term participation refers to the formal de�nition of labor force participation according to international norms. Application of these norms can be problematic in South Asia, because they do not take into account nonmarket activities of women. A PROFILE OF SOUTH ASIA AT WORK 91 work additional hours. In India, de�ning the can be important determinants of whether underemployed as people who worked at least women participate in the labor market. In three days during the week and spent at least a India, for example, where the overall female half day searching for work, results in an esti- participation rate was 30 percent in 2010, mated underemployment rate of 5 percent.3 the rate among women from scheduled tribes Low female employment rate is primar- (46 percent) was 16 points higher and the rate ily a result of low levels of labor force par- among Muslim women (18 percent) almost ticipation among women. The lowest female 12 points lower. In Sri Lanka, where aggre- labor force participation rates are in the gate female participation was 41 percent in three large South Asian countries: Pakistan, 2008, the rate for Indian Tamil women was where almost four out of every �ve women 62 percent and the rate for Sri Lankan Moors do not participate in the labor force, and just 17 percent. Bangladesh and India, where slightly more Female participation is especially low in than two out of every three women do not urban areas. Overall labor force participa- participate.4 tion is generally lower in cities than it is in Because female participation is such an rural areas, where labor-intensive, family- important factor in defining the region’s oriented agricultural production still domi- employment picture and its evolution over nates, but this gap is especially striking for time, this issue merits a more detailed look. women. Female rural participation rates are Before proceeding, a caveat about labor higher than urban participation rates in all force statistics and the concept of partici- countries except Bangladesh (where female pation is needed. All of the surveys collect participation is low everywhere); in Afghani- data that make it possible to measure labor stan and Pakistan, the participation rate force participation according to international norms, but the application of these norms to FIGURE 3.5 Female labor force participation rates in South Asia, low-income, traditional societies can be prob- by age group and country lematic. For this reason, it seems unlikely that the actual participation of women, especially 100 in the region’s large countries, is as low as the rates of participation in the surveys indicate. 90 In what follows, the term participation needs 80 to be understood as referring to the formal de�nition of labor force participation; it does 70 not take into account other activities of South 60 Asian women, including reproduction and percent household labor. 50 Except in Bhutan and Nepal, South Asian 40 countries generally have low female participa- tion rates across age groups (�gure 3.5). This 30 is especially true in Pakistan and, to a lesser 20 degree, Bangladesh, where even in the prime- age groups, the large majority of women are 10 not in the labor force. In all three countries in which data on 0 15–24 25–34 35–44 45–54 55–64 65+ caste/ethnicity are collected (India, Nepal, age group and Sri Lanka), there is considerable variation Nepal, 2008 Bhutan, 2007 Maldives, 2004 in female labor force participation along this Afghanistan, 2008 Sri Lanka, 2008 Bangladesh, 2009 dimension; differences in male labor force India, 2010 Pakistan, 2009 participation are small. The �nding on caste/ ethnicity is not surprising, as cultural factors Source: Authors, based on data from national labor force and household surveys. 92 MORE AND BETTER JOBS IN SOUTH ASIA for rural women is nearly three times that in the labor force. Sociocultural explanations for urban women. The most recent female have also been put forward, based on the urban participation rates are just 10 percent possible stigma attached to educated women in Pakistan, 18 percent in Afghanistan, and who choose to work. 19 percent in India. Moreover, there is little In all countries except India, the surveys evidence of any signi�cant change, with the ask women not participating in the labor force (unweighted) average female urban participa- why they were not employed or searching for tion in the region increasing from 30 percent work. In all countries, household duties were to 33 percent over the periods studied. the number one reason cited for nonpartici- What factors are associated with female pation (table 3.3). This is especially true of labor force participation? The labor force countries with very low rates of female labor participation status of working-age women force participation (Afghanistan, Bangladesh, was regressed on individual and family char- and Pakistan). This finding is consistent acteristics, using separate logit models for with the fact that women in South Asia, like rural and urban women in each country. women across the world, bear a dispropor- Table 3.2 summarizes the key determinants tionate share of household and care respon- of participation in urban areas. sibilities and therefore face high opportunity The negative relationship between edu- costs when they work in the marketplace. cation and the labor force participation of Social norms also affect these tradeoffs. women has been noted by others studying Education was the second-most fre- the region’s labor market (World Bank 2010). quently cited reason, but there are large Various explanations have been put forward cross-country differences. Substantial num- to explain this relationship. One hypothesis bers of young women in Bhutan, Nepal, and is that better-educated women may opt out Maldives report being in school instead of of the labor market because of the scarcity of the labor force. In contrast, in Afghanistan, good jobs that are available to them and that Bangladesh, Pakistan, and Sri Lanka, fewer an income effect may be at play, with rela- than 2 in 10 urban women not participating tively high family incomes reducing the incen- in the labor force cite education as the main tives for well-educated women to participate reason. TABLE 3.2 Factors associated with participation of women in urban areas Factor Effect Age • Age increases probability of female participation in all countries. Effect weakens later in age distribution. Education • More years of schooling decreases the probability of female participation in all countries except Bhutan. Participation rates tend to be lowest for women who complete secondary or lower-secondary school; they rise only at higher- secondary levels and tertiary levels. Household • Living in a larger household reduces the probability of female participation in all countries except Afghanistan, Nepal, characteristics and Maldives (where there is no effect). • The number of children under the age of six reduces the probability of female participation in all countries except Afghanistan and Bhutan. • Ethnic minority or lower caste status increases the probability of female participation in countries for which data are available (India, Nepal, and Sri Lanka). Marital status • Being married reduces the probability of female participation in all countries except Bangladesh (one of two surveys) and Nepal. Characteristics of other • More years of schooling of the best-educated male in the household reduces the probability of female participation adults in the household in all countries, presumably because it signals an income effect. • Having males in the household who are employed increases the probability of female participation in all countries. • Having a migrant away from the household increases the probability of female participation in India and Nepal but not Maldives (no data on other countries). Source: Authors, based on data from national labor force and household surveys. A PROFILE OF SOUTH ASIA AT WORK 93 TABLE 3.3 Reasons why urban women in South Asia do not participate in the labor force, by country (percent) Country/year Old age Illness Household duties Education Discouraged Other Afghanistan, 2008 0.2 0.6 81.0 12.9 2.2 3.2 Bangladesh, 2005 0.3 2.2 81.1 15.0 0.2 1.0 Bhutan, 2007 3.6 2.7 60.4 26.9 2.6 3.8 Maldives, 2004 — 11.1 46.7 22.8 — 19.5 Nepal, 2008 7.0 0.7 51.6 29.6 4.7 6.3 Pakistan, 2009 1.3 1.0 81.1 16.2 0.1 0.4 Sri Lanka, 2008 5.3 1.8 75.5 15.8 0.4 1.3 Source: Authors, based on data from national labor force and household surveys. Note: — = Not available. Improving economic opportunities for and of the region’s working-age population lives educational attainment of women could con- in rural areas and rural employment rates tribute to improved utilization and allocation are higher than urban rates in all countries of South Asia’s female labor force. Women’s except Maldives. decision to participate in market work is not In Afghanistan, Bhutan, India, and Nepal, independent of the occupational and earn- at least half of all employment remains in ings opportunities available to women in the agriculture. Only in Maldives is this sector labor market as these impact incentives to a relatively minor source of employment. participate. Consistent with global evidence Services are important in most countries, on employment segregation by gender (World representing more than 40 percent of total Bank 2012), women in South Asia are less employment in Bangladesh, Maldives, and likely to access the better jobs (see last section Sri Lanka. The industrial sector, including of this chapter). They also earn signi�cantly manufacturing, utilities, and construction, is less for the same type of job, even after con- relatively small, despite the great importance trolling for differences in education. Improv- attached to industrialization since indepen- ing opportunities requires interventions that dence (Srinivasan 2010). In Bangladesh, relax time constraints, increase access to pro- India, Maldives, Pakistan, and Sri Lanka, ductive inputs, and correct institutional and 20–27 percent of the employed workforce market failures that contribute to employment works in industry, with most of them in segregation. (For a comprehensive discussion manufacturing. As expected, these sectoral of options to improve economic opportuni- patterns differ substantially between rural ties for women, see World Bank 2012.) and urban areas. Agriculture is the largest sector of employment in rural areas in all countries except Maldives and Sri Lanka. In The nature of employment urban areas, most workers are in the service sector. Manufacturing accounts for about This section begins by describing employ- a quarter of urban workers in Bangladesh, ment patterns by location and sector in India, Pakistan, and Sri Lanka. South Asia. It then looks at employment sta- Sectoral employment patterns are chang- tus and informality. ing. The share of agriculture employment in total employment has been declining by about 0.5 percentage points a year in recent Employment patterns by location decades in countries where statistics are and sector available over time. In the �ve largest coun- Most South Asians work in rural areas tries in the region, employment growth in (table 3.4). The concentration in rural areas agriculture was slower than other sectors in reflects the fact that more than 70 percent the � rst decade of this century (�gure 3.6). 94 MORE AND BETTER JOBS IN SOUTH ASIA TABLE 3.4 Distribution of employment in South Asian countries, by location and sector (percent) Total Total Rural Urban Country/ year Agriculture Industry Services Rural Urban Agriculture Industry Services Agriculture Industry Services Afghanistan, 2007 59 13 29 85 15 68 12 21 9 19 72 Bangladesh, 2009 39 21 40 76 24 47 18 35 12 30 57 Bhutan, 2007 68 7 24 78 22 85 4 11 9 19 72 India, 2010 50 23 27 74 26 65 19 16 7 34 59 Maldives, 2004 17 27 55 66 34 24 33 43 5 16 79 Nepal, 2008 73 11 16 87 13 80 9 11 31 21 48 Pakistan, 2009 43 21 36 69 31 60 16 24 6 32 62 Sri Lanka, 2008 31 27 42 90 10 34 26 39 2 31 68 Source: Authors, based on national labor force and household surveys. Note: These data pertain to the area in which the worker’s main employment is located. The classi�cation of the area is based on each country’s classi�cation of rural and urban. Sri Lanka’s classi�cation of rural areas includes the tea estate sector, where a large number of workers are employed. Differences in classi�cation may account for some of the variations across countries. FIGURE 3.6 Annual percentage increases in number of employed workers in South Asia, by sector and country 8 7 6.8 6 5.2 5.0 5 4.7 4.3 4.0 4 3.8 percent 3.0 2.8 3 2.7 2.6 2.5 2.4 2.0 2.2 2 1.5 1.1 1 0.0 0 –0.2 –0.3 –1 Bangladesh India Nepal Pakistan Sri Lanka 2000–10 2000–10 2000–10 2000–10 2000–10 agriculture industry services total employment Sources: Authors, based on data from ILO 2011and national labor force and household surveys. A PROFILE OF SOUTH ASIA AT WORK 95 Total agricultural employment increased sig- by only 1 percentage point over nine years. ni�cantly in Nepal and Pakistan during this In contrast, in Bhutan, Maldives, Pakistan, period; it remained constant in Bangladesh and Sri Lanka, it increased 5–11 percentage and declined in Sri Lanka and India. points over six to nine years. The major contributors to job creation In India, employment in the nonfarm sec- everywhere have been industry and services. tor increased steadily for 25 years, rising from Industrial employment has grown very rap- 20 percent of the rural workforce in 1983 to 35 idly in Bangladesh (at almost 7 percent a percent in 2009/10. The pace of diversi�cation year), Pakistan (just over 5 percent a year), away from agriculture increased over time. and India (just over 4 percent a year). Ser- During 1983–1993/94, the average annual vices employment growth has been strongest growth in nonfarm jobs was just over 2 per- in Bangladesh, Nepal, and Pakistan, at 4–5 cent. During 1993/94–1998/99, it increased percent a year. to 3 percent; between 1999 and 2004/05, it The gradual decline in the shares of agri- increased to 4 percent. In the 1980s, of the cultural employment reflects not just rural- nearly 40 million additional rural jobs gener- urban migration but also the growth of the ated in India, 6 out of 10 were in the farm sec- rural nonfarm sector across the region. The tor. In contrast, of the 56 million new rural rural nonfarm sector employs 12–59 percent jobs created between 1993 and 2004, 6 out of the total workforce (15–65 percent of the of 10 were in the nonfarm sector (World rural workforce) in South Asia (�gure 3.7). Bank 2011a). This trend has continued in Countries that are still primarily rural and recent years: between 2004/05 and 2009/10, agricultural (Bhutan, Nepal) have the small- the nonfarm sector increased from 30 percent est rural nonfarm sectors. to 35 percent of the rural workforce. The pace of development of the rural non- According to data from the 2000 and farm sector varies widely across countries 2008 China National Rural Surveys, trans- and time (�gure 3.8). In Nepal, the nonfarm formation of the rural labor market has been sector share of the rural workforce increased one of the most salient trends in China’s FIGURE 3.7 Distribution of employment in South Asia, by sector and country 100 15 13 10 90 24 22 26 34 31 80 31 70 60 16 percent 57 44 50 49 69 41 66 40 30 59 50 20 27 32 28 10 26 17 12 0 Afghanistan Bangladesh Bhutan India Maldives Nepal Pakistan Sri Lanka 2007 2005 2007 2010 2004 2008 2009 2008 rural nonfarm rural agriculture urban Source: Authors, based on data from national labor force and household surveys. 96 MORE AND BETTER JOBS IN SOUTH ASIA FIGURE 3.8 Percentage of rural workers in the nonfarm sector in South Asia, by country 100 90 80 76 70 65 60 percent 50 41 42 40 40 35 35 30 30 24 25 20 19 20 20 15 10 8 0 2002 2005 2003 2007 1983 1994 2000 2005 2010 1998 2004 1999 2008 1999 2008 Bangladesh Bhutan India Maldives Nepal Pakistan Source: Authors, based on data from national labor force and household surveys. development since the 1980s. India’s over- • Local public rural infrastructure provision all pace of transition has been slower than was superior (as a result of higher levels of China’s. The share of rural labor force off decentralization in China). the farm in China was lower than in India in Unlike in East Asia, most nonfarm jobs the early 1980s (a few years after the start of in South Asia are in the service sector, with the major rural reforms adopted in 1978 that commerce the largest subsector, employ- abolished the commune system and intro- ing 12–33 percent of nonfarm rural work- duced the household responsibility system). ers (figure 3.10). The manufacturing sec- However, by 2000 the share had become tor—which in other developing countries, significantly greater than India’s, at 43.5 especially East Asia, was the major source percent; by 2008 the share of China’s rural of employment for workers moving out of labor force that worked off the farm was 62 agriculture—provides less than 30 percent percent—twice that of India (�gure 3.9). of nonfarm jobs. Mukherjee and Zhang (2007) offer a number of explanations for the faster pace of rural nonfarm sector development in China: Employment Status • Local governments in China were incentiv- South Asia is far from a typical, modern ized to support the town and village enter- labor market dominated by wage or salaried prises, as they generated revenue for them. employees. In all countries except Maldives • China’s rural nonfarm sector was less and Sri Lanka, most workers are self- protected than India’s. When the sector employed (box 3.2). The dominance of self- was liberalized, it was more competitive employment is most extreme in Afghanistan, than the protected small-scale sectors in Bhutan, and Nepal, where more than three India, which were not able to compete out of every four workers are self-employed. after liberalization. The scarcity of secure work forms is even • Rural literacy was higher, making it eas- more striking when wage employment is ier for workers to move into the nonfarm broken down into regular wage or salaried sector. workers and casual workers. In Afghanistan, A PROFILE OF SOUTH ASIA AT WORK 97 FIGURE 3.9 Percentage of rural workers in the nonfarm sector in China and India, 1983–2008 100 90 80 70 62 60 percent 50 44 40 30 31 30 24 25 20 20 15 10 0 1983 1994 2000 2004 2008 India China Source: Wang, Huang, and Zhang 2011. Note: Data for China are not available for 1994 and 2004. FIGURE 3.10 Rural nonfarm sector employment in South Asia, by economic activity and country 100 11 15 90 17 16 17 26 3 26 30 2 3 3 4 80 2 9 2 10 1 7 12 4 70 7 18 0 13 9 7 20 60 17 1 5 27 percent 3 26 21 50 24 12 7 33 40 29 8 17 15 10 2 30 17 0 1 23 7 9 1 0 1 20 3 33 28 0 22 27 10 21 21 12 14 0 1 0 2 2 0 1 0 2 Afghanistan Bangladesh Bhutan India Maldives Nepal Pakistan Sri Lanka 2007 2005 2007 2010 2004 2008 2009 2008 other services public administration �nancial, insurance, real estate transportation commerce construction electricity and utility manufacturing mining Source: Authors, based on data from national labor force and household surveys. 98 MORE AND BETTER JOBS IN SOUTH ASIA BOX 3.2 Composition of the labor force by employment status Regular wage or salaried workers are de�ned as regu- self- employed workers are typically farmers work- larly paid wage employees in the public or private sec- ing their own land, though many self-employed tors. These workers are usually on the regular payroll workers work in the rural nonfarm sector. There of the enterprises for which they work and usually are signi�cant gender differences in the type of self- earn leave and supplementary bene�ts. A signi�cant employment, with women much more likely than proportion of regular wage or salaried work is in the men to be classi�ed as family enterprise workers. In public sector, ranging from 27 percent in India (2010) most countries in the region, men are more likely to to 66 percent in Afghanistan (2007) (the proportions work as own-account workers. in other countries were 29 percent in Bangladesh in The category of the self-employed is very hetero- 2005; 42 percent in Nepal in 2008; 40 percent in Pak- geneous. It can be split into two groups:. istan in 2009; and 52 percent in Sri Lanka in 2008). Casual laborers are de� ned as wage workers who • The high-end self-employed subgroup consists of are paid on a casual, daily, irregular, or piece-rate all employers and other self-employed workers basis. These workers typically do not have access who work as of�cials, managers, professionals, to formal instruments of social protection. In rural technicians, and clerks. On average, these work- areas, casual laborers are often landless agricultural ers are more educated than other self-employed help, though a signi�cant number of casual workers workers. Their consumption distribution pro- work in the rural nonfarm sector (in, for example, � le is more similar to regular wage or salaried construction). workers. Self-employed workers consist of employers, • The low-end self-employed subgroup consists own-account workers, and unpaid family enterprise of own account and unpaid family workers who workers. They represent the largest group of work- work as service workers, skilled agricultural work- ers in most South Asian countries, ranging from ers, craftspeople, machine operators, and workers 43 percent in Sri Lanka to 82 percent in Nepal. in elementary occupations. Their consumption The majority of the self-employed are own-account pro�les are similar to those of casual laborers (box workers or family enterprise workers. In rural areas, �gure 3.2.1). BOX FIGURE 3.2.1 Distribution of per capita household expenditure in India and Nepal, by employment status a. India, 2005 b. Nepal, 2003 0.003 0.0015 Kernel density 0.002 Kernel density 0.001 0.001 0.0005 0 0 0 1,000 2,000 3,000 0 1,000 2,000 3,000 monthly per capita expenditure (Indian rupees) monthly per capita expenditure (Nepalese rupees) regular wage casual wage self-employed (high end) self-employed (low end) Source: Authors, based on data from national labor force and household surveys. A PROFILE OF SOUTH ASIA AT WORK 99 Bangladesh, India, and Nepal, more than shift toward regular wage work in Nepal half of wage earners are casual workers. and Pakistan. In contrast, within the rural Only in Bhutan is wage employment domi- nonfarm sectors in Bangladesh (2002–05) nated by regular wage or salaried workers. and India (1983 –2010), there was an The pattern of employment differs by increase in the share of casual labor, both location. A larger share of rural workers than as a share of total rural nonfarm employ- urban workers is self-employed. The major- ment and as a share of rural nonfarm wage ity of rural wage earners are casual laborers, employment. whereas the majority of urban wage earners are regular wage or salaried workers. Informality The distribution of workers by employ- ment status changed very little, if at all, in The issue of informality is a prominent one the past decade (table 3.5). Only in Bhutan in South Asia. Since the term informal sec- and Maldives did employment shift mark- tor was coined, about 40 years ago, consid- edly toward wage work, and it is not known erable efforts have been made to de� ne and if the increase there was in regular wage or measure informality. (For a comprehensive salaried work, casual labor, or both. In the discussion of measurement and statistics, other countries, self-employment continues see ILO 2002. For a de� nition of informal to dominate, with the share of wage employ- employment in India, see the National Com- ment growing very little in the past decade. mission for Enterprises in the Unorganised Within wage employment, there was a slight Sector 2009.) TABLE 3.5 Distribution of employment in South Asian countries, by type of employment (percent) Wage employment Self-employment Regular wage or Country/year salaried Casual labor Employer Own account Family enterprise Latest year Afghanistan, 2008 9 14 0.5 44 33 Bangladesh, 2005 14 22 0.3 41 22 Bhutan, 2007 21 4 0.2 25 50 India, 2010 17 32 1 33 17 Maldives, 2004 55 4 31 10 Nepal, 2008 8 10 1 35 46 Pakistan, 2009 21 17 1 34 28 Sri Lanka, 2008 57 3 29 11 Earlier year Bangladesh, 2002 14 22 0.4 44 19 Bhutan, 2003 14 0.2 18 68 India, 2000 15 31 1 31 21 Maldives, 1998 48 6 37 9 Nepal, 1999 6 10 1 39 44 Pakistan, 2000 19 18 1 41 21 Sri Lanka, 2000 57 2 28 13 Source: Authors, based on data from national labor force and household surveys. Note: Labor force surveys in Bhutan, 2003; Maldives; and Sri Lanka do not allow the separation of wage employment into regular wage workers and casual laborers. Afghanistan has only one survey. 100 MORE AND BETTER JOBS IN SOUTH ASIA Information on employment status and Asia is informal (figure 3.12). Informal- sector, � rm characteristics, and worker edu- ity rates based on pension coverage in all cation from national labor force surveys is South Asian countries except Sri Lanka are used to compare levels of informal employ- higher than in other countries with simi- ment across the region. Informal workers lar levels of gross domestic product (GDP). include all workers in the informal sector as Together with Africa, South Asia has the well as workers in the formal sector perform- highest rate of informal employment in the ing informal jobs: all workers in agriculture; world (�gure 3.12). wage workers in informal enterprises; and Although informal employment has casual laborers, family enterprise workers, traditionally been seen as a labor market and self-employed workers with less than problem—because informal workers tend to senior-secondary education in the nonagri- have low earnings and little access to formal cultural sectors. (Annex table 3A.2 shows social protection systems—recent research, country details.) Based on this de�nition, an especially in Latin America, suggests that in estimated 86–95 percent of employment is some situations, individuals may choose to informal in all countries except Maldives and work informally. Analysis in South Asia has Sri Lanka, and 71–81 percent of nonagricul- emphasized the vulnerability and involuntary tural employment is informal in all countries nature of informality in the region (Chen except Bhutan, Maldives, and Sri Lanka and Doane 2008; National Commission for (�gure 3.11). Enterprises in the Unorganised Sector 2009). The estimated rates are consistent with The next section of this chapter shows that other studies showing that the vast major- informal workers in South Asia are less ity of employment in South Asia is infor- skilled, earn less, and have higher poverty mal. 5 Using lack of pension coverage as a rates than formal workers and that informal proxy, Loayza and Wada (2011) estimate manufacturing and services �rms have lower that 91 percent of the labor force in South labor productivity and pay lower wages than FIGURE 3.11 Percentage of employment in South Asia classi�ed as informal, by country 100 95 92 90 87 88 89 86 79 81 80 80 74 71 71 70 60 58 56 51 percent 50 43 40 30 20 10 0 Afghanistan Bangladesh Bhutan India Maldives Nepal Pakistan Sri Lanka 2008 2005 2007 2010 2004 2008 2009 2008 all sectors nonagricultural Source: Authors, based on data from national labor force surveys. A PROFILE OF SOUTH ASIA AT WORK 101 formal � rms in the same size class. Loayza FIGURE 3.12 Percentage of labor force not covered by pension and Wada (2011) also point out that labor schemes, by region informality is higher than predicted by the level of production informality, suggesting 100 that the productivity of informal workers in South Asia is relatively low compared with that of informal workers in other countries. 80 The pervasiveness of informality in South Asia is likely to remain a core feature for a 60 percent long time. Informality is a complex, multi- faceted phenomenon that is shaped by both 40 the modes of socioeconomic organization and the relationship the state establishes with private agents through regulation and moni- 20 toring. Loayza and Wada (2011) show that the actual rates of labor informality in South 0 Asia are similar to predicted levels based on om ed As d M bbe nd Af and ci� d sia a ric La tral an Pa an the determinants of informality: the legal and on lop i a hA Eu s th Am a d an Ea ica c Af ie i n pe ica rth ast ia r ec eve ut As Ce ro e C er No le E regulatory framework, educational achieve- So st d ar id tin ment, the share of youth or rural popula- tion, and the sectoral production structure (box 3.3). Most of these determinants are Source: Loayza and Wada 2011. BOX 3.3 Determinants of informality Using t wo labor informalit y measures (self- structure tilted toward agriculture favors informality employment and lack of pension coverage), Loayza by making legal protection and contract enforcement and Wada (2011) show that in cross-country regres- less relevant and valuable. Third, larger shares of sions, labor informality is negatively and signi�cantly youth or rural populations are likely to increase infor- related to the strength of law and order, business mality, make monitoring more dif�cult and expensive, freedom from regulations, and average years of sec- place greater demands on resources for training and ondary schooling and positively and significantly the acquisition of abilities, create bottlenecks in the associated with sociodemographic transformation initial school-to-work transition, and make it more factors (the share of agriculture, youth population, dif�cult to expand formal public services. and rural population). All correlation coef�cients are Bangladesh, India, Pakistan, and Sri Lanka have highly statistically signi�cant (p-values of less than larger predicted informality levels than the grow- 1 percent) and of large magnitude (0.68–0.83). The ing East Asian countries (the Republic of Korea, predicted levels of informality for South Asian coun- Malaysia, and Singapore). Sociodemographic factors, tries are similar to actual levels. in particular the region’s high ratio of rural popula- These results have several implications. First, tion, are the largest contributors to the differences informality is more prevalent when the regulatory in predicted informality levels between South Asian framework is burdensome, the quality of government and East Asian comparator countries. Lower busi- services to formal � rms is low, and the state’s moni- ness regulatory freedom (for all countries) and low toring and enforcement power is weak. Second, the levels of education (for all countries except Pakistan) structural characteristics of underdevelopment play an play a moderate but consistent role in explaining dif- important role in explaining informality. Other things ferences in informality. Law and order does not play equal, a higher level of education is likely to reduce a major role in explaining the differences. informality by increasing productivity, potentially increasing returns to formalization. A production Source: Loayza and Wada 2011. 102 MORE AND BETTER JOBS IN SOUTH ASIA structural and take time to change. For this Among South Asian countries in which reason, large informal sectors continue to headcount poverty rates by employment sec- exist even after economies have experienced tor and status of household members can be rapid growth.6 In South Asia, despite grow- analyzed, agricultural casual workers have ing labor productivity and increasing qual- the highest poverty rates in Bangladesh and ity of jobs, there is little evidence from labor Nepal and the second highest in India (after force or industrial surveys that labor infor- urban casual labor) (�gure 3.13).8 Agricul- mality is decreasing.7 tural self-employment provides a relatively As the correlates of informality are largely good source of income (as proxied by pov- structural, the book assumes that formaliza- erty rates) in Bangladesh and India but not tion will be a slow process. Easing restric- in Nepal. The jobs associated with the low- tive labor legislation and other interventions est poverty rates in all countries are regu- to improve the business regulatory envi- lar wage work outside of agriculture in the ronment can contribute to lower informal- urban or rural nonfarm sectors. Urban self- ity, but large increases in formality are not employment in Bangladesh and Nepal is also expected to occur immediately. Therefore, associated with lower poverty rates. the approach taken in this book is to aim to The wage data tell a consistent story: work- improve the quality of all types of jobs by ers in industry and service sectors are better addressing constraints to the productivity of paid than workers in agriculture (�gure 3.14). all workers and � rms, formal or informal. Rural nonfarm employment (regular wage Such an approach is likely to have a positive and casual labor) and urban regular wage effect on reducing informality, as increasing employment offer higher wages than agri- productivity may increase the returns to for- cultural casual labor in Bangladesh, India, malization. As the majority of workers will Nepal, and Pakistan (the countries for which remain informal at least in the near term, wage employment can be split into regular efforts should be made to increase their wage or salaried and casual labor in all time access to programs that help manage labor periods observed): market shocks (see chapter 6). • Rural nonfarm and urban regular wages are 30–100 percent higher than wages for Where are the better jobs? agricultural labor. • Wages for rural nonfarm casual labor This book de� nes “better jobs� as jobs asso- are 10–50 percent higher than wages for ciated with higher wages (for wage workers), agricultural labor. Even with the consis- lower poverty, and a lower risk of low and tent increase in the share of rural nonfarm uncertain income. This section examines jobs, nonfarm casual labor wages in India which sectors, types of employment, and have remained about 30–50 percent higher types of � rms are associated with better job than agricultural wages since 1983. quality. • Wages for urban casual labor are higher than wages for agricultural casual labor By sector in Nepal and Pakistan (20–30 percent) but the same in Bangladesh and India. This Employment in industry and services—in evidence is consistent with the high pov- urban areas or the rural nonfarm sector— erty rates observed among urban casual yields higher wages and is associated with labor in Bangladesh and India. lower poverty rates than agricultural casual labor. Chapter 2 showed that labor produc- Wage differentials between services and tivity (measured as output per worker) is agriculture are especially large (�gure 3.15). much higher in industry and services than in In India, average hourly wages in services agriculture. The higher productivity in these were 135 percent higher than in agriculture in sectors is manifested in higher earnings. 2010. In Nepal, Pakistan, and Sri Lanka, the A PROFILE OF SOUTH ASIA AT WORK 103 FIGURE 3.13 Percentage of workers in households below the poverty line in Bangladesh, India, and Nepal, by employment status a. Bangladesh, 2010 b. India, 2004/05 c. Nepal, 2003/04 agricultural casual 51 urban casual labor 45 agricultural casual labor 50 labor rural nonfarm casual agricultural casual rural nonfarm casual labor 45 31 43 labor labor urban casual labor rural nonfarm casual agricultural 44 25 31 labor self-employed rural nonfarm regular wage rural nonfarm or salaried worker 26 urban self-employed 24 17 self-employed rural nonfarm rural nonfarm self-employed 26 18 urban casual labor 16 self-employed agricultural agricultural rural nonfarm regular self-employed 22 17 10 self-employed wage or salaried worker urban regular wage urban regular wage or salaried worker 16 14 urban self-employed 9 or salaried worker rural nonfarm regular urban regular wage urban self-employed 14 8 3 wage or salaried worker or salaried worker 0 50 0 50 0 50 percent percent percent Source: Authors, based on data from national labor force and household surveys. Note: Figures are for workers age 15–64. Poverty rates for India are based on official poverty lines prevailing until 2010. Using the new official poverty lines for 2004/05 (revised in 2011) would increase poverty rates in rural areas, making the poverty rates of rural workers higher than those of urban workers for the same employment type. The hierarchy in terms of employment type would remain the same. FIGURE 3.14 Ratio of median rural nonfarm and urban wages to agricultural wages in selected South Asian countries 4.0 3.0 3.0 2.8 2.4 2.2 2.2 2.1 2.1 2.0 2.0 ratio 2.0 2.0 1.8 1.8 1.8 1.6 1.7 1.6 1.6 1.6 1.5 1.6 1.5 1.5 1.4 1.5 1.4 1.4 1.3 1.3 1.3 1.3 1.3 1.2 1.1 1.1 1.1 1.1 1.0 1.0 1.0 1.0 0.9 0 2002 2005 1994 2000 2005 2010 1999 2008 2000 2009 Bangladesh India Nepal Pakistan rural nonfarm regular wage or salaried worker rural nonfarm casual labor urban regular wage or salaried worker urban casual labor Source: Authors, based on data from national labor force and household surveys. Note: A ratio of 1 means the median wage was equivalent to the median wage of agricultural casual labor. 104 MORE AND BETTER JOBS IN SOUTH ASIA FIGURE 3.15 Ratio of median industry and service sector wages to agricultural casual labor without controlling agricultural wages in selected South Asian countries for other factors was 40 percent in 2000 and 50 percent in 2008; after controlling for other 2.5 2.35 factors, including educational attainment of 2.1 the workers, it was 24 percent in both years. 2.0 2.0 1.8 There is thus a reallocation gain even without 1.5 1.5 1.4 1.4 additional investment in education. 1.5 ratio 1.0 By employment status and �rm type 0.5 Within industry and services wage employ- ment, better jobs can be defined from two 0 main angles: by worker’s employment status 2010 2008 2009 2008 India Nepal Pakistan Sri Lanka and by type of �rm. Regular wage or salaried service/agriculture workers have considerably higher average industry/agriculture hourly wages than casual laborers (see �gure Source: Authors, based on data from national labor force and household surveys. 3.14). The two types of wage employment Note: A ratio of 1 means the median wage was equivalent to the median wage in agriculture. correspond roughly to formal and informal employment in the wage sector.10 In Bangla- desh, India, Nepal, and Pakistan, regular wage workers earn 20–48 percent more than ratio of wages in services to wages agricul- casual workers in the rural nonfarm sector ture was also very high (112 percent higher and 14–74 percent more than casual workers in Nepal, 84 percent higher in Pakistan, and in the urban sector. These wage differentials 99 percent higher in Sri Lanka). partly reflect differences in education levels Industry and service jobs are better paid between regular wage workers and casual partly because they are higher-skilled jobs. laborers. Casual labor has the highest pov- Nonagricultural workers have higher average erty rates; poverty rates are generally lowest levels of education than agricultural work- for regular wage or salaried workers, among ers. Within the nonagricultural sector, regu- whom poverty rates are on average one third lar wage workers are more educated than or less those for casual labor. casual workers, with a signi�cant proportion In addition to lower wages, casual labor of workers having secondary education or employment offers less stability than regular above. The higher wages observed in industry wage work. In India in 2009/10 (depending and service employment partly reflect higher on the sector of employment), 31–49 percent levels of education. of casual laborers, 8–23 percent of the self- However, the majority of industry and ser- employed (own- account and family helper), vice jobs pay more than agricultural casual and 4–7 percent of regular wage or salaried labor even after accounting for higher levels workers reported spending at least one month of education (and other individual and house- looking for work in the past year. Casual hold characteristics that can affect wages, laborers spent an average 0.9–1.4 months such as age, gender, caste/ethnicity, religion, without work the past year. Self-employed and household wealth).9 The wage premium workers reported 0.2–0.7 months of unem- is highest for public sector regular wage or ployment; regular wage or salaried workers salaried workers, where the premium over reported virtually no time without work.11 the agricultural wage is 41–50 percent in It is more dif� cult to assess the quality Bangladesh, India, and Nepal. Private sector of self-employment because there are no nonfarm sector jobs also offer a wage pre- wage data and self-employed workers are a mium. In India, for example, the wage dif- very heterogeneous group. The small pro- ferential between nonfarm casual labor and portion of workers classi�ed as “high-end� A PROFILE OF SOUTH ASIA AT WORK 105 self-employed have poverty rates that are TABLE 3.6 Distribution of formal and informal similar to those of regular wage or salaried manufacturing �rms in India, by location and size, workers. Most self-employed workers—who 2005 are typically involved in small-scale enter- (percent) prises (enterprises with no more than five Firm Formal Informal workers) as own-account workers or family characteristic sector sector Total enterprise workers—have lower poverty rates Percentage than casual workers but higher rates than reg- of all firms 0.7 99.3 100.0 ular wage workers in India and Nepal. The Location India poverty assessment (World Bank 2011a) Urban 60.2 29.0 29.2 reports that only about half of nonfarm self- Rural 39.8 71.0 70.9 employed workers regard their earnings from Firm size self-employment as remunerative. 1–49 74.4 100.0 99.8 50–99 11.9 0.0 0.1 Analyzing the question from the point of 100+ 13.7 0.0 0.1 view of the �rm is more challenging because of the lack of nationally representative � rm- Sources: Authors, based on data on formal �rms from the Annual Survey of Industries and data on informal �rms from the National Sample Survey level industrial data. Only India conducts manufacturing surveys. national surveys that cover all � rms in the manufacturing and services sectors, infor- levels of firms employing more than 200 mal and formal, rural and urban. Based on workers (�gure 3.16). Informal �rms with one the estimated level of informality and self- to four employees have 25 percent of the value employment (especially own-account and added per worker and less than 50 percent of family enterprise work) indicated in national the wages per worker of formal �rms the same labor force surveys, there is reason to believe size. Although productivity is lower in rural that the statistics presented below for India firms than in urban firms of the same size are typical of other South Asian countries.12 class and type, especially in the informal sec- Among Indian manufacturing and services, tor, real wages (adjusted using consumption- workers in larger formal firms earn higher based household deflators rather than indus- wages, which reflect higher value added per try deflators) in rural and urban �rms of the worker. However, the majority of manufactur- same size and type are similar.14 ing and services employment is in micro/small Although the � rm size productivity/wage informal �rms, which have lower value added differential exists in other regions, it is par- per worker and, hence, pay lower wages. ticularly high in India. In East Asia (China; Indonesia; Korea; Malaysia; the Philippines; India’s manufacturing sector Taiwan, China; and Thailand), the produc- In 2005, 99 percent of the 17.1 million man- tivity of small enterprises (enterprises with ufacturing � rms in India were informal (the 5–49 workers) is about 20–40 percent that of term used in India is unorganized); nearly enterprises with more than 200 workers (ADB all � rms were micro/small enterprises (1–49 2009). In India, small � rms have 12 percent workers) (table 3.6). Even in the formal sec- of the productivity and pay 19 percent of the tor (referred to in India as the organized sec- wages of large �rms. tor), 74 percent of � rms were micro/small.13 A number of factors may accounts for Seventy-one percent of all � rms were rural. the � rm size productivity/wage differentials Average productivity (value added per observed everywhere: 15 worker) and average wages are lower in smaller � rms than in larger � rms and much • Smaller and informal � rms are less capital lower in informal �rms than in formal �rms intensive. In India, for example, informal the same size. Value added and wages per � rms have less than a quarter of the capi- worker in formal firms with one to four tal per worker of formal � rms in the same employees are on average one-quarter the size class. 106 MORE AND BETTER JOBS IN SOUTH ASIA FIGURE 3.16 Average wage, value added, and capital per manufacturing worker in India, by �rm size and type, 2005 (percentage of average wages/value added/capital per worker in formal firms with 200 or more employees) a. Mean wage per worker b. Median wage per worker 100 100 90 90 80 80 70 66 70 70 64 60 60 60 60 percent percent 54 50 50 50 43 40 40 37 32 30 27 30 25 19 23 20 17 20 18 9 11 10 10 0 0 4 9 9 9 9 9 4 9 9 9 9 9 1– 5– –1 –4 –9 19 1– 5– –1 –4 –9 19 0– 0– 10 20 50 10 20 50 10 10 firm size firm size d. Median value added per worker 100 100 c. Mean value added per worker 90 90 80 80 70 70 66 60 56 60 percent 55 percent 53 50 45 50 47 40 39 40 37 30 26 30 26 19 21 19 21 20 20 10 12 10 4 10 6 0 0 4 9 9 9 9 9 4 9 9 9 9 9 1– 5– –1 –4 –9 19 1– 5– –1 –4 –9 19 0– 0– 10 20 50 10 20 50 10 10 firm size firm size e. Mean capital per worker f. Median capital per worker 100 100 90 90 80 80 71 70 70 62 60 60 percent 50 47 50 44 45 40 39 percent 40 32 40 31 31 30 26 30 20 13 20 12 10 5 10 7 7 3 0 0 4 9 19 9 9 4 9 9 9 9 99 1– 5– –4 –9 1– 5– –1 –4 –9 – 1 0– 10 20 50 10 20 50 10 firm size firm size formal firms informal firms Sources: Authors, based on data on formal �rms from the Annual Survey of Industries and data on informal �rms from the National Sample Survey manufacturing surveys. Note: Formal �rms with 200 or more employees = 100 percent. • Workers in larger formal enterprises are perhaps as a result of economies of scale, more skilled. and they may pay “ef�ciency wages� (higher • Large formal enterprises are more produc- than market-clearing wages paid to encour- tive with the labor and capital they have age higher output, raise worker morale, and (their total factor productivity is higher), discourage absenteeism and shirking). A PROFILE OF SOUTH ASIA AT WORK 107 Eighty-one percent of workers in India sector became more dynamic, growing at 3 work in informal firms, and 83 percent of percent a year to 8.7 million. At the same workers are in micro or small firms, with time, employment in the informal sector most working in own-account manufactur- declined by 0.3 percent a year. The opposing ing enterprises with fewer than �ve workers. dynamics of the two periods resulted in very Combined with the large productivity and similar share of employment in the informal wage differentials observed, these figures sector and in micro/small � rms in 1994 and translate into low-productivity and low-wage 2005. jobs for the majority of the 45.1 million man- The share of employment in micro and ufacturing workers in India in 2005. small �rms is much larger in India than it is in The distribution of employment across East Asia (�gure 3.18). A number of factors can � rm size groups has not shifted over time: affect the enterprise size distribution, includ- the share of employment in micro/small �rms ing industrial composition, infrastructure, and the informal sector in 2005 is almost product market segmentation, coordination the same as it was in 1994 (figure 3.17). failures, credit constraints for small firms, Between 1994 and 2000, employment in the regulations that differ according to enterprise informal sector increased 4 percent a year, size, and industrial policy (ADB 2009). Lower- from 28.8 to 37.0 million, whereas employ- income countries tend to produce less complex ment in the formal sector remained stagnant goods, which are better suited to small-scale at 7.7 million (employment in the largest production. � rms actually fell 1 percent a year). Between Part but not all of India’s concentration of 2000 and 2005, employment in the formal employment in small � rms can be explained FIGURE 3.17 Share of manufacturing employment in India, by �rm size and type, 1994–2005 100 90 80 70 60 percent 50 40 30 20 10 0 4 9 9 5 9 0– 9 9 0+ 4 9 9 50 9 10 –99 9 0+ 4 9 20 9 50 9 10 9 9 0+ 1– 5– –1 –4 10 0–9 19 1– 5– –1 –4 19 1– 5– –1 –4 –9 19 20 20 20 0– 0– 10 20 10 20 10 1994 2000 2005 informal directory manufacturing establishments (at least 1 hired worker and more than 6 workers in total) informal nondirectory manufacturing establishments (at least 1 hired worker but fewer than 6 workers in total) informal own-account manufacturing enterprises (no hired workers) formal Sources: Authors, based on data on formal �rms from the Annual Survey of Industries and data on informal �rms from the National Sample Survey manufacturing surveys. Note: The data show a small share (1 percent or less) of informal employment in the larger �rms as well. 108 MORE AND BETTER JOBS IN SOUTH ASIA FIGURE 3.18 Share of manufacturing employment by �rm size in India’s service sector India and selected East Asian economies India had 16.5 million services �rms in 2006, 15 percent of which were establishments and India 83 6 11 85 percent were own-account enterprises.16 Ninety-eight percent of own-account enter- Philippines 70 8 23 prises and 73 percent of establishments have Indonesia 65 6 29 one to four workers, and nearly all � rms are Korea, Rep. 47 24 30 micro or small (table 3.7). In contrast to the manufacturing sector, India’s service sector Thailand 46 13 42 has very few establishments with 50 or more Taiwan, China 39 21 40 workers. Sixty percent of all services � rms are rural. Malaysia 28 20 53 Average productivity (value added per China 25 23 52 worker) and average wages are lower in smaller �rms than in larger �rms (�gure 3.19). 0 20 40 60 80 100 percent For example, mean wages at �rms with fewer than 19 employees are 20–47 percent those micro and small (1–49) medium (50–199) large (200+) of � rms with 20–49 workers. Although the Sources: Authors, based on data from ADB 2009, India Annual Survey of Industries, and National average wage monotonically increases with Sample Survey manufacturing surveys. �rm size, average productivity and capital per Note: Data for India are for 2005. Data for East Asia data are for latest year available between 2004 and 2007. worker do not always increase among � rms with 5–9 and �rms with 10–19 employees. Productivity and wages are also much lower in own-account enterprises �rms than in establishments of the same size class. by its level of income and industrial composi- Own-account enterprises with one to four tion. For example, India’s apparel industry is employees (the vast majority of such enter- dominated by small � rms, whereas China’s prises) have less than half the value added industry is dominated by medium-size and per worker and less than 15 percent of the large � rms (ADB 2009). It is possible that mean wage per worker of establishments the past industrial policies in India, particularly same size. One reason for this differential is the policy of reserving production of certain that own-account enterprises are less capital products for small-scale industries, which intensive. started in the late 1960s, has had signi�cant An estimated 27.5 million workers were effects on � rm size distribution that are still employed in the service sectors covered in being felt today. India’s 2006 National Sample Survey. Sixty Data from the Annual Survey of Indus- percent of service workers worked in own- tries since 2005 show continued dynamism account enterprises and 40 percent in estab- in the growth of employment in the for- lishments. The majority of workers worked mal sector, suggesting that a J-curve effect in lower-productivity and lower-wage micro may be at work: reforms in the 1990s led �rms (own-account enterprises and establish- to an initial decline of employment in the ments with fewer than �ve workers). The con- formal sector in medium-size and large centration of employment in micro and small � rms before increasing from 2000 onward enterprises (enterprises with 1–49 workers) is (box 3.4). (National Sample Survey data even higher in the service sector (96 percent of on the informal sector are not available workers) than in manufacturing (83 percent). after 2005, making it impossible to analyze The employment pattern in services has the growth of the informal manufacturing not changed significantly over time, with employment and determine how the overall the share of employment in micro/small and distribution of manufacturing employment own-account enterprises remaining fairly has evolved since then.) stable between 2001 and 2006 (�gure 3.20). A PROFILE OF SOUTH ASIA AT WORK 109 BOX 3.4 Trends in India’s formal manufacturing sector, 1998–2007 Total employment in India’s formal manufacturing the decade, with employment in nonprivate/state sec- sector follows a U-shaped path, declining from 1998, tor � rms (mostly large � rms) declining steadily and reaching a trough between 2001 and 2003, and growing signi�cantly as the public sector downsized. Initially, rapidly thereafter. This path was driven by employment this pattern led to a decline in overall employment, as in large �rms; employment in small and medium-size private sector employment remained stable through �rms (enterprises employing fewer than 100 workers) 2003. Beginning in 2003, private sector employment was fairly stable through 2003 and grew thereafter. rapidly increased, more than offsetting the continuing The U-shaped pattern in employment can be decline in the state sector. explained by the large compositional change over BOX FIGURE 3.4.1 Employment in India’s formal manufacturing sector, by �rm size, type, and location, 1998–2007 a. By location and �rm size b. By �rm ownership (private versus nonprivate) 12 12 10 10 8 8 millions millions 6 6 4 4 2 2 0 0 98 99 00 01 02 03 04 05 06 07 98 99 00 01 02 03 04 05 06 07 19 19 20 20 20 20 20 20 20 20 19 19 20 20 20 20 20 20 20 20 Source: Authors, based on data from Annual Survey of Industries. Note: Annual data on the formal manufacturing sector are available for 1998–2007. The year refers to the reference period in the survey. For example, the reference period in the 2000/01 survey is the accounting year ending March 31, 2001. As three-quarters of the reference period fell in 2000, the survey is labeled 2000. SME = small to medium enterprise. Employment in establishments increased 1.2 TABLE 3.7 Distribution of service �rms in India, by location and size, percent a year, from 10.7 million in 2001 to 2006 11.4 million in 2006; employment in own- (percent) account enterprises increased 0.4 percent a Own-account year, from 15.9 to 16.2 million. Data after Firm characteristic Establishments enterprises Total 2006 are not available. Percentage of all firms 14.7 85.3 100.0 Although industrial surveys with the Location same coverage of the manufacturing and Urban 63.0 36.1 40.0 service sectors are not available for other Rural 37.0 63.9 60.0 South Asian countries, other enterprise Number of employees and � rm surveys suggest a similar picture 1–49 99.7 100.0 99.96 elsewhere. Rural nonfarm enterprise sur- 50–99 0.17 0.00 0.03 veys conducted in Bangladesh, Pakistan, 100 + 0.04 0.00 0.02 and Sri Lanka show large numbers of rural Source: Authors, based on data from National Sample Survey service sector surveys. 110 MORE AND BETTER JOBS IN SOUTH ASIA FIGURE 3.19 Average wage, value added, and capital per service sector worker in India, by �rm size and type, 2006 (percentage of average wages/value added/ capital per workers in establishments with 20–49 employees) 100 a. Mean wage per worker b. Median wage per worker 100 90 90 80 80 70 70 58 percent 60 percent 60 50 47 50 51 40 33 40 32 30 30 20 20 20 10 3 10 2 0 0 1–4 5–9 10–19 1–4 5–9 10–19 �rm size �rm size c. Mean value added per worker d. Median value added per worker 100 100 90 90 80 80 70 70 66 68 58 percent percent 60 60 50 43 50 40 34 36 40 35 30 30 20 15 20 10 10 0 0 1–4 5–9 10–19 1–4 5–9 10–19 �rm size �rm size e. Mean capital per worker f. Median capital per worker 100 100 90 90 80 80 70 68 63 70 56 percent percent 60 52 60 51 50 50 48 40 40 30 30 20 19 20 17 10 10 0 0 1–4 5–9 10–19 1–4 5–9 10–19 �rm size �rm size establishments own-account enterprises Source: Authors, based on data from the National Sample Survey service sector surveys. Note: Establishments with 20–49 employees = 100 percent. FIGURE 3.20 Share of service sector employment in India, by �rm size and type, 2001 and 2006 100 90 80 70 60 percent 50 40 30 20 10 0 0+ 0+ 9 9 9 9 9 9 4 4 9 9 9 9 5– –1 –4 5– –1 –4 1– 1– 19 19 –9 –9 20 20 10 20 10 20 0– 0– 50 50 10 10 �rm size �rm size 2001 2006 own-account enterprise establishments Source: Authors, based on data from the National Sample Survey service sector survey. Note: Employment in 2006 excludes the �nancial intermediation sector, because the sector was not covered in 2001. A PROFILE OF SOUTH ASIA AT WORK 111 nonfarm enterprises and large numbers of or salaried workers are more educated than workers employed in the sector. The major- self-employed or casual workers. Almost all ity of rural nonfarm enterprises are micro have some education, and a signi�cant pro- family enterprises with few hired workers, portion have secondary education or above. low use of capital, and lower productivity The self-employed are better educated than than urban formal � rms. casual workers. The relative educational lev- els of employment types are found in both the Who holds better jobs? urban and rural nonfarm sectors and across all South Asian countries (�gure 3.21). These This section examines the factors that affect data suggest that education is an important access to better jobs in South Asia. These determinant of access to better jobs and that factors include education, gender, and caste lack of education is likely to be a barrier to and ethnicity. mobility to moving into better jobs. (For an analysis of the effect of education on work- Education ers’ ability to access and move into better jobs, see chapter 5.) Education is an important determinant of access to better jobs. It is strongly correlated with the sector of employment in South Gender Asian labor markets. Industry and services rely on workers who are considerably more Labor force participation rates for women are educated than workers in agriculture (table low in South Asia, particularly in urban areas. 3.8). The average educational attainment is Among women who do participate in the labor particularly high in service sector industries force, the majority live in rural areas, where such as utilities, � nance, and public admin- 70 percent of South Asians of working age istration. Although manufacturing and live. Most rural female workers are employed construction jobs do not usually require as in traditional agricultural activities. The more much education as these service industries, modern and better-paid rural nonfarm sector their workforces have much more education employs mostly men. In all countries except than workers in agriculture. Maldives, a higher percentage of male rural Within industry and services, education workers are in the nonfarm sector. Although is correlated with better jobs. Regular wage more female rural workers are entering the TABLE 3.8 Average years of education in South Asian countries, by sector of employment Public administration Manufacturing Other services Construction Agriculture Commerce Transport Financial Utilities Mining Country/year Afghanistan, 2008 0.9 1.9 1.4 — 2.0 3.5 3.5 — 9.8 4.2 Bhutan, 2007 0.9 3.8 3.5 8.2 4.2 4.2 4.3 9.9 5.7 7.2 India, 2010 4.1 6.1 6.4 10.1 4.4 7.7 6.9 12.9 11.2 9.7 Maldives, 2004 3.5 4.4 3.5 6.8 4.7 7.2 7.8 9.2 7.8 8.2 Nepal, 2008 3.1 2.9 4.5 6.3 3.5 6.2 5.4 9.6 9.8 9.6 Pakistan, 2008 2.7 5.5 5.7 10.4 4.0 6.8 5.6 11.7 9.9 9.9 Sri Lanka, 2008 7.1 7.6 9.4 10.9 8.6 9.3 9.9 12.2 11.6 10.9 South Asia unweighted mean 3.2 4.6 4.9 8.8 4.5 6.4 6.2 10.9 9.4 8.5 Source: Authors, based on data from national labor force and household surveys. Note: Bangladesh is not included, because its surveys do not permit comparable calculations. — = Not available. percent se percent percent 112 se lf- se 0 10 20 30 40 50 60 70 80 90 100 0 10 20 30 40 50 60 70 80 90 100 lf- 0 10 20 30 40 50 60 70 80 90 100 lf- ag em agri em ri c c em agri ag plo ultu pl cul ag ploy ultu ric ym ra oy tu 17 13 m ra 47 ric m ral ul en l 0 1 a en l 2 ul en tu t tu t no ra no ca gric t no ra nf lc nf su ult nf lc as ar as a a u 22 14 ar 23 or m r lab ual or rm r l lab ral 1 or m r lab ual sa eg or 0 sa eg or sa eg or lar ul lar ul lar ul ied ar ied ar ied ar wo wag wo wag 66 59 se se 88 se wo wag lf- rk e lf- rk e 37 16 rk e em n er 28 lf- em n er em n er pl on pl on pl on o fa oy fa m rm 26 o fa 37 no yme rm e 65 5 2 no yme rm nf nt 6 ar ca no nt nf nt ar m ur su n m ur ba ca b al far ur ca 21 or an r lab m 34 ba s or n r la sual sa eg or 54 2 4 or n r lab ual sa eg bo c. Bhutan, 2007 lar ul 1 e. Nepal, 2008 sa eg or lar ul r ied ar a. Afghanistan, 2008 lar ul ied ar ied ar wo wag 87 wo wag 78 se rk e se rk e 90 wo wag lf- lf- er 62 26 se rk e em er em 42 lf- er pl u em pl u oy rb oy rb m a 52 pl u 45 oy rb ur men an 7 ur en n 15 68 m an in South Asia, by employment type and country ba t ba t 12 ur ent nc nc ba as as 37 nc lab ual 48 as lab ual 4 or 9 61 lab ual or Source: Authors, based on data from national labor force and household surveys. 5 percentage with some education or MORE AND BETTER JOBS IN SOUTH ASIA percent se percent se percent se lf- lf- 0 10 20 30 40 50 60 70 80 90 100 0 10 20 30 40 50 60 70 80 90 100 0 10 20 30 40 50 60 70 80 90 100 lf- em agri em agri em agri c ag plo cult ag plo cult ag ploy ultu ric ym ura ric ym ura 63 54 37 ric m ral ul e l ul e l 9 1 3 ul en tu nt tu nt tu t no ra no ra no ra nf lc nf lc nf lc ar as ar as ar a 52 34 or m r lab ual 23 or m r lab ual 0 2 or m r la sual sa eg or 0 sa eg or sa eg bo lar ul lar ul lar ul r ied ar ied ar ied ar wo wag wo wag 92 90 se rk e se rk e 82 wo wag se lf- 40 18 rk e lf- 29 lf- em n er em n er em n er pl on pl on pl on o fa 76 o fa 62 oy fa no ym rm no yme rm 67 2 13 no me rm nf ent nf nt 8 ar ar nf nt ar ur m ur m m ba c ba c percentage with secondary education or above ur 62 51 b ca or n r l asu or n r l asu 0 sa eg ab al sa eg ab al 4 51 d. India, 2010 or an r la sual lar ul or 2 lar ul or sa eg bo f. Pakistan, 2009 ied ar b. Bangladesh, 2005 lar ul r ied ar ied ar wo wag wo wag 93 87 wo wag se rk e se rk e lf- er 85 lf- er 50 28 se rk e em em 42 lf- er pl u pl u em oy rb oy rb m a m a 85 70 pl u oy rb ur en n 7 ur en n 30 m an 75 ba t ba t ur ent nc 18 nc ba as as 69 48 nc lab ual lab ual as 5 1 or or 59 lab ual 6 or FIGURE 3.21 Percentage of workers with some education and percentage of workers with secondary education or above A PROFILE OF SOUTH ASIA AT WORK 113 sector, the gender gap has widened over time, and Nepal, but less than 30 percent of rural increasing from 10 percentage points in 1983 women in India and less than 20 percent in to 16 percentage points in 2010 in India, for Nepal worked in the nonfarm sector at the example (�gure 3.22). end of the period. A cohort analysis for India, Nepal, and A multinomial logit analysis that looks Pakistan provides further evidence that at the determinants of labor participation women are generally stuck in agriculture and occupational choice shows that among across all countries in the region (�gure 3.23). workers with less than upper-secondary edu- This analysis tracks participation in the rural cation, men are more likely than women to nonfarm sector for the same age groups of hold regular wage jobs within the rural non- male and female workers for 8–10 years farm and urban sectors. In contrast, female beginning in 1999/2000. workers with upper-secondary education Several findings emerge. First, for every are more likely than male workers to hold age cohort, women were less likely than men regular wage jobs (see chapter 5). Given the to work in the rural nonfarm sector. Second, small percentage of women who participate younger men were more likely than older in the labor force, the share of working-age men to leave agriculture for the nonfarm sec- women in better jobs is still much lower than tor and stay there. In all countries, the cohort the share of working-age men in better jobs, of men that was 15–24 in 1999/2000 had the even at the highest levels of education. largest share of nonfarm labor by 2009/10 The lower level of educational attainment (40 –55 percent), followed by the 25–34 among women is one reason why they are cohort. Among men in cohorts other than less likely than men to be in better jobs. More 15–24, and in some cases 25–34, the share than half of the female workforce in �ve of of workers in the nonfarm sector by the end the eight countries in the region has no educa- of the period declined or remained the same. tion. Occupational segregation and lower pay Third, younger women were more likely than for the same jobs and quali�cations are other older women to exit agriculture in India likely factors. FIGURE 3.22 Percentage of rural workers in the rural nonfarm sector in South Asia, by gender and country 100 90 90 80 73 70 70 66 64 61 60 58 54 percent 50 49 49 42 40 39 40 35 36 31 30 30 28 28 25 23 22 21 20 16 19 16 15 15 12 11 13 8 10 10 4 0 2003 2007 2002 2005 1998 2004 1999 2008 2000 2009 2000 2008 1983 1994 2000 2004 2010 Bhutan Bangladesh Maldives Nepal Pakistan Sri Lanka India male female Source: Authors, based on data from national labor force and household surveys. 114 MORE AND BETTER JOBS IN SOUTH ASIA FIGURE 3.23 Percentage of rural workers in the rural nonfarm sector in India, Nepal, and Pakistan, by gender and age cohort, 1999–2009/10 a. Male workers in India b. Female workers in India 60 60 50 50 40 40 percent percent 30 30 20 20 10 10 0 0 2000 2005 2010 2000 2005 2010 c. Male workers in Nepal 60 d. Female workers in Nepal 60 50 50 40 40 percent percent 30 30 20 20 10 10 0 0 1999 2008 1999 2008 e. Male workers in Pakistan f. Female workers in Pakistan 60 60 50 50 40 40 percent percent 30 30 20 20 10 10 0 0 2000 2009 2000 2009 age 15–24 age 25–34 age 35–44 age 45–54 Source: Authors, based on data from national labor force and household surveys. Note: The age cohort refers to the age of the cohort in the earliest period, that is, the line of the cohort of age 15–24 in India refers to the same cohort group: the cohort that was 15–24 years old in 2000. A PROFILE OF SOUTH ASIA AT WORK 115 Women earn wages that are 20–40 per- It is important to determine how much of cent lower than those of men, even after con- the observed wage gap is a result of produc- trolling for education levels, type of employ- tive characteristics, such as education. The ment, and other individual and household technique proposed by Blinder and Oaxaca characteristics associated with productivity. (1973) has been used to decompose the wage This result holds true for all countries and gap into the part explained by differences in almost all types of wage employment. The levels of observable characteristics (includ- lower wages could be a result of unobserved ing education) and the part not explained or unmeasured characteristics in the wage by these differences. The unexplained por- regressions; they are also consistent with tion reflects in part differences in returns to some other type of discrimination in the mar- skills.17 Figure 3.24 plots the decomposition ket that is not associated with productivity for rural and urban workers for all survey differentials. If discrimination in the labor years. It indicates that part of the wage gap is market is a major factor behind the observed the result of differences in observable produc- wage gap, a policy focus on increasing the tive characteristics but that most of the wage skill levels of women will not be suf�cient. gap is unexplained. FIGURE 3.24 Decomposition of wage gap between male and female workers in South Asia, by country a. Afghanistan, 2008 b. Bangladesh, 2002 c. Bangladesh, 2005 d. India, 2000 60 40 percent 20 0 –20 rural urban rural urban rural urban rural urban e. India, 2005 f. India, 2008 g. Nepal, 1999 h. Nepal, 2008 60 40 percent 20 0 –20 rural urban rural urban rural urban rural urban i. Pakistan, 2000 j. Pakistan, 2009 k. Sri Lanka, 2000 l. Sri Lanka, 2008 60 40 percent 20 0 –20 rural urban rural urban rural urban rural urban explained unexplained Source: Authors, based on data from national labor force and household surveys. 116 MORE AND BETTER JOBS IN SOUTH ASIA Caste and ethnicity Women are more likely to work, particularly in rural areas, if they belong to an ethnic Caste and ethnicity have long been impor- minority. Among women who work, women tant factors shaping employment in South from ethnic minority groups are more likely Asia. Female labor force participation and to be in casual or self-employment in urban employment rates in India, Nepal, and Sri areas and agriculture in rural areas. Lanka vary by ethnic/caste groups. Ethnic minority groups earn lower wages The multinomial logit analysis shows that than other groups, after controlling for edu- workers from ethnic minority groups, par- cation levels and employment type. The dif- ticularly women, are less likely to be in bet- ference, though smaller in size than gender ter jobs even after controlling for education differentials, appears signi�cant in India but and other factors. Among men in rural areas, not in Nepal.18 members of ethnic minority groups are more The results of the decomposition of the likely to be involved in agriculture. In urban wage differential show that in India and areas, men from ethnic minority groups are Nepal, much of the wage gap is explained more likely to be involved in casual work and by differences in observable characteristics, self-employment. including education (�gure 3.25). In contrast, Similar trends appear for women, and in Sri Lanka most of the wage differential is the differences appear larger in magnitude. unexplained. FIGURE 3.25 Decomposition of wage gaps between nonethnic minority and ethnic minority workers in India, Nepal, and Sri Lanka a. India, 1983 b. India, 2000 c. India, 2005 d. India, 2008 40 30 percent 20 10 0 rural urban rural urban rural urban rural urban e. Nepal, 1999 f. Nepal, 2008 g. Sri Lanka, 2000 h. Sri Lanka, 2008 40 30 percent 20 10 0 rural urban rural urban rural urban rural urban explained unexplained Source: Authors, based on data from national labor force and household surveys. A PROFILE OF SOUTH ASIA AT WORK 117 Annex 3A De�nitions and criteria used in pro�le of South Asia at work De�nitions of employment and Criteria used to de�ne formal and unemployment informal workers Table 3A.1 provides the speci�c de� nitions Table 3A.2 provides information on how the for employment and unemployment used national surveys are used to estimate infor- for each country based on the national mal employment for each country. surveys. TABLE 3A.1 De�nition of employment and unemployment used based on national surveys Definition of unemployment (for people Country Year Definition of employment who are not employed) Afghanistan 2007/08 In past 30 days, did any one of the following: In past 30 days, did any of the following: • Worked for any organization or any individual • Attempted to find a job or start a business • Performed any agricultural work, even without pay, • Found a job that has not yet started on land owned, rented, or used by household • Awaited recall from an employer • Performed any nonagricultural work, on own account, in a business enterprise • Performed occasional paid work, such as helping someone in his or her business • Has a permanent or long-term job from which he or she will be temporarily absent Bangladesh 2002/03 In past seven days, did either of the following: • Was available for work in past seven days and tried to • Worked for at least one hour on any day for pay or find a job during past two months profit, family gain, or final use or consumption • Even if did not work in past seven days, has a job, business, enterprise, or attachment to a job such as a business, farm, or shop 2005/06 Same as Bangladesh 2002 In past seven days, did any of the following: • Was prepared for and tried to find a job • Was waiting for reappointment • Was waiting to start work after appointment 2009 In past seven days, did either of the following: • In past seven days, looked for job • Engaged in any economic activity for at least one hour as paid worker, for family gain or profit, or own use or consumption • Was temporarily absent from work Bhutan 2003 In past seven days, did any of the following: In past seven days, did any of the following: • Any farming, fishing, hunting, or gathering of fruits • Actively looked for a job or tried to start a new • Worked for money or have profitable business business • Performed unpaid work on enterprise or farm of • Was waiting for a job to start friends or relatives • Was waiting for employer’s reply 2007 Same as 2003 In past seven days, did any of the following: • Actively looked for a job or tried to start a new business • Did not actively look for a job because was waiting for employer’s reply • Did not actively look for a job because was waiting for employers recall (continues next page) 118 MORE AND BETTER JOBS IN SOUTH ASIA TABLE 3A.1 De�nition of employment and unemployment used based on national surveys (continued) Definition of unemployment (for people Country Year Definition of employment who are not employed) India 1999/2000, In past seven days, did any of the following: • In past seven days, sought work and/or was available 2004/05, • Worked or helped in household enterprise for work 2007/08, and • Worked as regular or casual wage labor 2009/10 • Was involved in public or other works Maldives 1998 and In past 30 days, did either of the following: • Not currently working because unable to find a 2004 • Was involved in economic activity most of the time suitable job, but if hired, would be available to work • Engaged in any activity that generated income Nepal 1998/99 and In past 7 days, did either of the following: In past 7 days, did either of the following: 2007/08 • Was involved at least one week in any economic • Was available and looked for work in past 30 days activities • Awaited reply on earlier inquiries or start of • Did not work but have a permanent job he or she prearranged job or business will return to Pakistan 1999/2000 In past 7 days, did any of the following: In past 7 days, did any of the following: and 2008/09 • Performed any work for pay, profit, or family gain • Looked for job during for one hour on any day • Did not work but will take a job within a month • Did not work but has a job or enterprise to return to • Was temporarily laid off and awaiting recall • Worked/helped in family business or family farm for family gain Sri Lanka 2000 In past 7 days, did either of the following: In past 7 days, was available or looking for work. • Engaged in economic activity during past 7 days • Did not engage in economic activity but had an economic activity to return to 2008 In past 7 days, did either of the following: • Already obtained a job or made arrangements to • Engaged in economic activity start self-employment activity • Had an economic activity to return to OR • Took some steps to find a job or start self- employment activity within the last four weeks OR • Awaited results of interview for work Source: Authors’ compilation. TABLE 3A.2 De�nition of formal and informal workers used based on national surveys Country Year Informal workers Formal workers Afghanistan 2007/08 • Daily laborers • Workers in public sector • Workers who receive pensions Bangladesh 2002/03 • Casual workers and day laborers • Workers in public sector • Domestic workers • Workers in private or nongovernmental formal • Workers in private informal sector sector 2005/06 • Domestic workers and apprentices • Workers in public sector • Workers working in personal households • Workers in private formal sector • Casual workers and day laborers Bhutan 2003 a • Workers in public sector 2007 • Casual paid employees • Workers in public sector India 1999/2000 • Workers in firms that do not keep written accounts • Workers in public and semipublic sector • Workers in firms with fewer than 10 employees • Workers covered under a provident fund • Workers in firms that keep written accounts • Workers in firms with more than 10 employees (continues next page) A PROFILE OF SOUTH ASIA AT WORK 119 TABLE 3A.2 De�nition of formal and informal workers used based on national surveys (continued) Country Year Informal workers Formal workers 2004/05 and • Casual workers • Workers in public sector 2009/10 • Workers without written contracts • Workers receiving social security benefits • Workers working in firms with fewer than • Workers with written job contracts 10 employees • Workers in firms with more than 10 employees • Workers not receiving any social security benefits Maldives 2004 a • Workers in public sector Nepal 1998/99 • Workers in unregistered organizations • Workers in public sector • Workers in firms with fewer than 10 employees • Workers in firms with 10 or more employees 2007/08 • Workers who are not eligible for paid leave and do • Workers in public sector not receive social security contributions • Workers for whom employers pay social security • Workers in establishments with fewer than contributions 10 employees • Workers who are eligible for paid leave • Workers working in establishments with 10 or more employees Pakistan 1999/2000 • Casual and day laborers • Workers in public sector and 2008/09 • Workers in enterprises that do not keep written • Workers in enterprises that keep written accounts accounts • Workers in establishments with 10 or more • Workers in establishments with fewer than employees 10 employees Sri Lanka 2000 • Daily laborers • Workers in public sector 2008 • Casual workers • Workers in public sector • Workers whose employers do not contribute to • Workers for whom employer contributes to pension pension scheme or provident fund and do not scheme or provident fund or provides paid leave provide paid annual leave • Workers in registered enterprise • Workers in unregistered enterprise • Workers in enterprise that maintains formal written • Workers in enterprise that does not maintain formal accounts written accounts • Workers in firms with 10 or more employees • Workers in firms with fewer than 10 employees Source: Authors’ compilation. Note: All family enterprise workers are in the informal sector. All self-employed workers with less than senior-secondary education are considered informal, and all self-employed individuals with higher education are considered formal. All workers involved in agriculture are informal. Some workers could not be identi�ed as formal or informal. In all countries except Maldives, the unclassi�ed portion was 10 percent or less. In Maldives 32 percent could not be classi�ed. For the subsample of workers that could not be identi�ed as formal or informal in each country, their status was allocated according to the formal/informal split for the employment that could be classi�ed. The assumption that the distribution of formality/informality is the same for classi�ed and unclassi�ed workers may lead to some overestimation of informality, as the classi�ed portion includes some block assignments, such as agriculture to the informal sector. Because the unidenti�ed portion is small in most countries, this should not be a major issue, except perhaps in Maldives. a = No additional criteria speci�c to the survey. Annex 3B Regional employment patterns Figure 3B.1 shows how employment rates difference in employment rates between the vary within each of the South Asian coun- regions with the highest and lowest rates is tries. The overall picture is one of relatively only about 10 percentage points. Although moderate differences in regional rates. the disparities in employment rates may not Unweighted coefficients of variations are be large, there are major differences in the 5–10 percent for all countries except Afghan- nature and quality of employment across istan, where regional differences are much regions (for example, the share of employ- larger. For all countries except Afghanistan, ment that is in the formal wage-earning the range is fairly narrow: in most cases, the sector). 120 MORE AND BETTER JOBS IN SOUTH ASIA FIGURE 3B.1 Regional variations in employment rate in South Asia, by country 90 West Central Mid -Western 85 84 (5.9) 80 75 Central Eastern 70 69 (9.0) percent 67 (16.7) 65 North-Central Sylhet Western 60 58 (4.7) Western 58 (6.7) 55 Chittagong 61 (5.7) Central Punjab Western Central Lagging 50 50 (7.8) 48 (8.8) 45 NWFP South 40 Afghanistan Bangladesh Bhutan India Maldives Nepal Pakistan Sri Lanka 2008 2005 2007 2005 2004 2008 2009 2008 Source: Authors, based on data from national labor force and household surveys. Note: Each diamond represents the employment rate in a region. The regions with the highest and lowest employment rates are named. Each box represents the average employment rate of all regions in the country; �gures in parentheses are coefficients of variation, calculated as the standard deviation of the regional employment rates normalized by the mean of these rates. Notes (identi�ed as activity pursued by the individ- ual for a relatively minor time, but not less 1. The labor force participation rate and the than 30 days during the past year) fell sharply employment rate in India in 2009/10 were compared to previous years. Usual principal 56.6 percent and 54.8 percent, respectively. employment status (de�ned on the basis of the There is a sharp decline from 2004/05, lead- activity pursued for the longest—major—time ing to only a very small increase (in absolute during the reference year) only experienced a terms) in the labor force and total employ- modest decline. The reasons for the decline ment between 2004/05 and 2009/10. The in labor force participation and employment decline has been driven by a sharp fall in rates remain to be analyzed fully, but possible female labor force participation and employ- reasons include (a) younger workers, espe- ment rates (male rates have also declined, cially women, staying longer in education; but far less sharply), which led to a signi�- (b) rising family incomes reducing the need for cant decrease in rural female labor force and women to work; and (c) shocks in the agricul- employment in absolute terms. The decline in ture sector. It is important to note that over female participation was observed in all age the longer period between 1983 and 2010 groups including school age cohorts (15–19 labor force participation and employment years old and 20–24 years old) but also older rates have increased and decreased without cohorts. Rural female employment declined a consistent trend, although the employment particularly in agriculture. Self-employment and participation rates in 2010 are lower declined both in number and as a share of than in any previous period. Using current total employment, while casual wage labor weekly status, the labor force participation accounted for nearly 80 percent of net addi- rate was 59.7 percent, 62.0 percent, 60.4 per- tional employment in those sectors that cent, 62.2 percent, and 56.6 percent in 1983, expanded employment between 2004/05 and 1994, 2000, 2005, and 2010, respectively, 2009/10. Usual subsidiary employment rates and the employment rate was 57.5 percent, A PROFILE OF SOUTH ASIA AT WORK 121 60.1 percent, 58.2 percent, 59.8 percent, and 8. Poverty rates for India are based on the 54.8 percent in 1983, 1994, 2000, 2005, and of�cial poverty lines prevailing until 2010. 2010, respectively. It may thus be premature Using the new of�cial poverty lines for to see the past �ve years as necessarily being 2004/05 (revised in 2011) would increase evidence of a longer-term trend. the poverty rates in rural areas, causing 2. In the absence of better questions in the sur- rural poverty rates to exceed urban work- vey in Maldives, the criterion “available for ers for the same employment type. The hier- work, conditional on getting a job� was archy in terms of employment type would use to de�ne unemployment. This criterion, remain the same, however: agricultural which is much broader than that used in casual labor would have the highest pov- other countries, yielded a much higher unem- erty rate, followed by rural nonfarm casual ployment rate. labor and urban casual labor. Agricultural 3. This measure, based on current weekly sta- and rural nonfarm self-employment would tus questions, is an adjusted version of the have higher poverty rates than urban self- de�nition of underemployment used by the employment. The lowest poverty rates National Commission for Enterprises in the would be among urban regular wage or Unorganised Sector (2009). salaried workers. 4. Using household survey data, Srinivasan 9. Results are based on regressions for each (2010) reviews labor market indicators over country-survey year. Log hourly wages were the longer term for Bangladesh (from regressed on individual and household char- 1983/84), India (from 1972/73), and Pakistan acteristics, including educational attainment (from 1990). He �nds that female participa- and sector and type of employment. tion rates in Bangladesh rose from 8 percent 10. All casual labor is classi�ed as informal in to 30 percent between 1983/84 and 2005/06, this book. Some regular wage work is also female participation rates in India were posi- classi�ed as informal (see annex table 3A.2 tive in rural but not urban areas, and female for details of de�nition used in each country’s participation in Pakistan has been essentially labor force survey). stagnant since 1990. 11. The India poverty assessment (World Bank 5. The informality shares in �gure 3.11 are simi- 2011a, p. 133) also highlights the transient lar to those in Chen and Doane (2008). nature of rural nonfarm casual labor work: 6. For example, more than a third of total “Casual workers tend not to have year-round employment in the Republic of Korea is infor- employment and make ends meet by working mal (OECD 2007). at several jobs, often combining agricultural 7. Based on the de�nition of informality used in and nonfarm activities. In 2004/05, more this book, there is no clear pattern of increas- than half (55 percent) of casual nonfarm ing or decreasing informality. In Bangladesh workers report that they are without work (2002–05), India (2000–05), and Pakistan for one or more months in the year compared (2000–09), there have been no signi�cant to 8 percent of salaried workers or 12 percent changes. In Maldives (1998–2004) and Sri of self-employed. Fourteen percent of casual Lanka (2000–08), informal employment nonfarm workers report that they were seek- declined signi�cantly. In Bhutan (2003–07) ing or available for additional employment and Nepal (1998–2008), it increased sig- even when working.� ni�cantly. However, because the concepts 12. Estimates from labor force surveys of infor- used to measure informality changed from mality in the nonagricultural sector indicate the earlier to the later year in the countries that Afghanistan, Pakistan, and Nepal have in which changes were observed, it is not a higher proportion of informal nonagricul- clear how much of the observed increase tural employment (about 80 percent) than or decrease was real. In Bangladesh and Bangladesh (74 percent), India (72 percent), Pakistan, where there was no change in the and Sri Lanka (68 percent). way of measuring informality, no change 13. “Organized� (formal) manufacturing �rms in the degree of informality was observed. are registered �rms, typically with more than Other studies, using alternative de�nitions, 10 workers. The most commonly used de�ni- have found increasing informality in India tion of the unorganized (informal) sector flows (NCEUS 2009). from the Indian Factories Act (1948), which 122 MORE AND BETTER JOBS IN SOUTH ASIA requires that an enterprise register with the large, are of�cially in the informal sector. The state government if it employs 10 or more service sector survey covers service �rms in workers and uses power or employs 20 or rural and urban locations for a large number more workers and does not use power. Units of service sectors but excludes trade. Dehejia that do not come under the purview of this law and Panagariya (2010) estimate that the sur- constitute the unorganized sector. Businesses vey covers about half of output and employ- that are organized are required to comply ment in the service sector. with health, safety, and welfare requirements. 17. This interpretation becomes problematic if Firms are required to contribute toward insur- the observable characteristics are improp- ance against sickness, disability, and maternity erly measured or omitted or unobservable and to deposit linked provident funds or pen- characteristics, such as motivation, could be sion schemes. The state monitors compliance influencing wages. Such characteristics will through a system of inspections. Unorganized bias the results if the errors and omissions �rms largely fall outside this system. The unor- are systematically different for men and ganized manufacturing sector comprises own- women. account manufacturing enterprises, directory 18. Ethnic minorities are scheduled caste and tribes manufacturing enterprises, and nondirectory in India and Dalits and Janajatis in Nepal. manufacturing enterprises subcategories. Own- account manufacturing enterprises are informal �rms that operate without any hired worker References employed on a regular basis. Nondirectory manufacturing establishments are informal ADB (Asian Development Bank). 2009. Enter- �rms that employ fewer than six workers prises in Asia: Fostering Dynamism in SMEs. (with at least one hired worker); directory Manila. manufacturing establishments employ six or Chen, M., and D. Doane. 2008. “Informality in more workers (with at least one hired worker). South Asia: A Review.� Background working Ninety-nine percent of unorganized manufac- paper for the Swedish International Develop- turing �rms have fewer than 10 workers. ment Cooperation Agency (Sida), Sweden. 14. Among �rms in the formal sector, real wages Dehejia, R., and A. Panagariya. 2010. “Services are slightly higher in rural �rms than urban Growth in India: A Look inside the Black �rms of the same size class. Among �rms with Box.� Working Paper 2010-4, Columbia Uni- one to nine employees in the informal sector versity, Program on Indian Economic Policies, (the vast majority of the unorganized sector), New York. wages in urban and rural unorganized �rms Government of India. 2006. Employment and are very similar. Unemployment Situation in India 2004– 05 15. Assessing the contribution of each factor is National Sample Survey 61st Round. Minis- challenging, given the dif�culty in matching try of Statistics and Programme Implemen- information across workers and �rms. For tation, National Sample Survey Of� ce, New example, the industrial surveys for India, Delhi. which provide �rm productivity and wage ILO (International Labour Of�ce). 2002. Women data, do not survey the education and skills and Men in the Informal Economy: A Statis- levels of workers. See ADB (2009) for a dis- tical Picture. Geneva: International Labour cussion of �rm-size wage differentials. Of�ce. 16. The National Sample Survey service sector ———. 2010. Accelerating Action against Child survey classi�es �rms as either establishments Labour: Global Report under the Follow-Up or own-account enterprises. Own-account to the ILO Declaration on Fundamental Prin- enterprises do not employ hired workers on a ciples and Rights at Work. Geneva: Interna- regular basis; establishments do. Own-account tional Labor Of�ce. enterprises belong to the informal sector; estab- ———. 2011. Key Indicators of the Labour Mar- lishments are a mix of informal and formal ket. Geneva: International Labour Of�ce. sector �rms. As services �rms are not required Islamic Republic of Afghanistan, and World Bank. to register under the Factories Act unless they 2010. Poverty Status in Afghanistan: A Pro�le are engaged in manufacturing, most private Based on the National Risk and Vulnerabil- sector service enterprises, whether small or ity Assessment (NRVA) 2007/08. Ministry A PROFILE OF SOUTH ASIA AT WORK 123 of Economy, Kabul, and Economic Policy and Volume 26. Bangladesh Institute of Develop- Poverty Sector, Washington, DC. ment Studies, Dhaka. Loayza, M., and T. Wada. 2011. “Informality in Srinivasan, T. N. 2010. “The Utilization of Labor South Asia.� Background study conducted for in South Asia.� Background study conducted this book. for this book. Mukherjee, A., and X. Zhang. 2007. “Rural UN (United Nations). 2010. World Population Industrialisation in China and India: Role of Prospects: The 2010 Revision. New York: Policies and Institutions.� World Development United Nations. 35 (10): 1621–34. Wang, X., J., Huang, and L . Zhang, 2011. National Commission for Enterprises in the “Creating the Entrepreneur Farmers Needed Unorganised Sectors. 2009. The Challenge of Yesterday, Today and Tomorrow.� Paper Employment in India: An Informal Economy presented at the International Fund for Agri- Perspective. New Delhi. cultural Development (IFAD) conference on OECD (Organisation for Economic Co-operation new directions for smallholder agriculture, and Development). 2007. “Labour Markets in Rome. Brazil, China, India and Russia and Recent World Bank. 2010. India’s Employment Chal- Labour Market Developments and Prospects lenge: Creating Jobs, Helping Workers. India: in OECD Countries.� OECD Employment Oxford University Press. Outlook. Paris. ———. 2011a. Perspectives on Poverty in India: Rahman, R. I. 2008. “A Review of Open Unem- Stylized Facts from Survey Data. Washington, ployment and Underemployment Concepts: DC: World Bank. Measurement, Methods, and Application� ———. 2011b. World Development Indicators. (in Bengali with title “Unmukto Bekaratta O Washington, DC: World Bank. Angshik Bekaratta Dharanar Porjalochana, ———. 2012. World Development Report: Gen- Parimap Paddhati O Proiog�). Bangladesh der Equality and Development. Washington, Unnayan Shamikkha, Golden Jubilee Number. DC: World Bank. CHAPTER 4 What Is Preventing Firms from Creating More and Better Jobs? Questions and Findings Questions informal �rms. However, without informa- tion on whether inadequate �nance reflects • What are the business environment con- the lack of bankable projects, one cannot straints affecting �rms in South Asia? infer directly that access to �nance is a bind- • Do these constraints vary by sector and �rm ing constraint. Other evidence suggests that characteristics? access to �nance may be an issue for micro • What are the main policy priorities for over- and small �rms in some countries. coming the identi�ed business constraints? Findings Policy focus areas The chapter identi�es constraints facing � rms • The severity of the electricity constraint for all and four areas of policy focus. types of �rms reflects the large gap in the region between demand for and supply of power. Clos- ing the gap requires a substantial increase in Constraints investment, which in turn requires that power • The most binding constraints facing all types sector reforms be sustained and deepened. of urban formal firms (where the highest- • South Asian �rms face high levels of corrup- productivity and highest-paid jobs are) are tion in a range of interactions with public of�- electricity, corruption, and political instability. cials, particularly for utilities and tax inspec- Job-creating �rms are more severely affected tions. Simplifying processes in, for example, than other �rms by virtually the entire range tax administration and reducing unnecessary of constraints. Labor regulations rank high in interaction with local officials could be an India, Nepal, and Sri Lanka. effective way of tackling corruption, as it has • The rural nonfarm sector is an important route the additional bene�t of reducing the cost of out of low-paid agricultural work. For �rms in red tape on �rms. this sector, electricity and political instability • Key policy options for improving access to also rank high on the list of constraints. Trans- finance for micro and small firms include port (poor road quality and inaccessibility) strengthening the institutional environment ranks much higher as a constraint for rural (secured transactions registry, credit informa- �rms, as it affects access to larger markets. tion); creating a conducive environment for • Increasing the productivity of informal �rms downscaling commercial banks and upscal- is fundamental, as they employ the majority ing micro�nance institutions; and providing of workers. Electricity is the top constraint �nancial literacy and training to micro and for informal urban �rms in India. Relative to small �rms. formal urban �rms, they are less concerned • Easing business registration could benefit about corruption, taxes, and labor regulations potential �rm entrants. Signi�cant improve- and more concerned about access to land. ments in the registration process would help • Firms cite inadequate access to finance as entrepreneurs who were previously in wage one of the top �ve constraints in every sec- employment or out of the labor force start tor; this constraint is particularly severe for new activities and create new employment. What Is Preventing Firms from Creating More and Better Jobs? 4 T his chapter looks at the constraints mitigation behaviors by � rms in the enter- faced by South Asian firms in the prise surveys. The business environment is manufacturing and service sectors considered external to the � rm; its compo- where, as shown in chapter 3, jobs are more nents resemble public goods and are consid- productive and better paid. Addressing con- ered common to all � rms in the economy. straints to the operations of � rms in these All enterprise surveys include a standard sectors is therefore important for creating question that asks firms to rate the sever- more and better jobs. ity of different business environment con- The chapter is organized as follows. The straints.2 The question takes the form: “How first section describes the methodologi- much of an obstacle is X to the operation cal framework. The second section looks at and growth of your business?� The firm’s constraints facing � rms in the urban formal response regarding its severity, rated on a sector. The third section extends the analysis �ve-point scale, with 0 being no obstacle and to the rural nonfarm and urban informal 4 being a very severe obstacle, is a measure of sectors. The fourth section suggests policy the marginal cost imposed by the constraint options for overcoming the most binding on the operation and growth of its business. constraints. The last section examines the The cost can be interpreted as the difference regulatory constraints faced by potential �rm between the �rm’s pro�t in the hypothetical entrants. situation in which the business environment poses no obstacle to the firm’s operations and the �rm’s actual pro�t given the existing Methodological framework quality of the business environment. This chapter is based on �ndings from three Differences in firms’ perceptions of the background studies by Carlin and Schaf- severity of constraints and mitigation behav- fer (2011a, 2011b, 2011c), who draw on 30 iors reflect the different costs of the constraint country enterprise surveys, covering more to �rms, which may reflect differences in the than 26,000 � rms, conducted in South Asia quality of the business environment faced between 2000 and 2010.1 The method- by � rms or differences in the � rms’ charac- ological approach taken to identify binding teristics. Differences in � rm characteristics business environment constraints uses the are more likely if the business environment reported severity of constraints and reported is shared. For example, almost all firms in 125 126 MORE AND BETTER JOBS IN SOUTH ASIA Afghanistan, Bangladesh, and Nepal report sufficiently large samples to analyze. The at least one power cut per month, but the sample size of individual surveys generally did valuations of the electricity constraint vary not allow analysis of how the reported sever- across firms. A higher-productivity firm ity of constraints changed over time. should report higher severity of a constraint The ranking of constraints for individual than a lower-productivity � rm, because the countries as well as the region as a whole marginal cost/forgone pro�ts are higher. The guides the prioritization of constraints to data con�rm this prediction. tackle. Some countries, notably India, report In addition to subjective perceptions of the lower absolute levels of severity of constraints severity of constraints, this chapter examines across the board. It is unlikely that there are objective measures, such as the frequency no significant business environment con- of power outages and mitigation behaviors, straints in India. Therefore, it is important including the use of generators and payment to look at the ranking of constraints, not just of bribes. Mitigation behaviors are a � rm’s the reported absolute severity level. reaction to the poor quality of the business The average severity of constraints environment. According to Carlin and Schaf- reported by a benchmark � rm in the urban fer’s methodological framework (2011c), formal sector is compared across countries expenditures on mitigation behavior are at similar levels of per capita gross domestic decreasing in the quality of the external busi- product (GDP).3 These data indicate whether ness environment (the better the environment, the severity reported by South Asian �rms is the less mitigation is needed) and increasing greater than expected given their country’s in the productivity of the �rm (more produc- level of development. tive �rms undertake more mitigation actions, The framework reveals how the reported because their marginal costs/forgone pro�ts level of severity (and ranking) of constraints are higher). The �ndings from these objective and mitigation activities varies for nonbench- measures are largely consistent with those mark � rms (for example, how the reported from the analysis of subjective constraints: severity of constraints changes with �rm size, countries in which �rms report higher sever- sector, location, formality, history of job cre- ity of the electricity constraint also report ation, and so forth). This book is particularly higher frequency of power outages. interested in �rms that expanded employment The average reported severity of con- in the past three years (job-creating firms), straints and the average reported level of because identifying and addressing constraints mitigation activities are measured for a to their operations and growth is likely to lead “benchmark �rm� in each country individu- to the creation of more and better jobs. ally and for the region as a whole (by pooling the country sample data), in order to account Application of methodology to rural for differences in the composition of firms nonfarm and informal �rms in the enterprise surveys and facilitate cross- country comparisons. A benchmark � rm is One of the main drawbacks of standard a medium-size manufacturing � rm with 30 enterprise surveys is that their sampling employees that is domestically owned, does frames are generally limited to the formal not export or import, is located in a large city, urban sector. This bias is problematic in and did not expand employment in the pre- South Asia, given the high levels of informal- ceding three years. The reported severity of ity. Enterprise surveys also omit the rural each element of the business environment was nonfarm sector, an important source of bet- regressed on the �rm’s characteristics. Hence, ter jobs in South Asia. The constraints anal- the average reported severity of a constraint ysis is conducted here for four South Asian for a benchmark � rm is a conditional mean countries that conducted enterprise surveys for the period covered by the surveys avail- of the rural nonfarm and informal sectors. able for each country (2000–10). The country For the rural nonfarm sector, the analysis surveys were pooled across time to generate examines the severity of constraints reported WHAT IS PRE VENTING FIRMS FROM CREATING MORE AND BE T TER JOBS? 127 by a benchmark �rm in Bangladesh, Pakistan, and informal sectors does not lead to substan- and Sri Lanka, where rural enterprise surveys tial creation of new jobs, it is still important have been conducted.4 These results are then for increasing the productivity and job qual- compared with those of an urban benchmark ity of a large proportion of existing workers �rm in these countries. in manufacturing and services. For the urban informal sector, the severity of constraints reported by benchmark � rms Limitations of the methodology is compared with that of the urban formal sector. (The informal survey in India was the The methodology allows the costs of differ- only one with a large enough sample size to ent business environment constraints to be facilitate this comparison. The comparison is assessed and compared. Two constraints that therefore limited to India’s formal and infor- many � rms in South Asia frequently report mal urban manufacturing sector.5) as severe—access to � nance and tax rates— Rural nonfarm and informal sector � rms cannot be analyzed in the same framework are typically microenterprises: in the surveys, as the others, as neither has the character of the median employment in rural nonfarm a public good. Even when the � nance sys- sector �rms in Bangladesh, Pakistan, and Sri tem is working well, one would expect that Lanka is 1.5 (compared with a median of 35 access to � nance would be a constraint to employees in the urban formal sector in these potential borrowers with low-quality proj- countries); in India, the median employment ects. Indeed, less productive firms report in the informal (urban) survey is 4 (compared more severe problems with access to � nance. with 20 for all �rms in the urban formal sec- A high reported severity does not necessarily tor). As the median size of �rms in the rural mean that access to � nance is a problem or and informal sectors is much smaller, the that increasing access to � nance would nec- benchmark firm in the rural nonfarm and essarily boost output and productivity. informal sector analyses is de�ned as having With respect to taxes, managers do not take �ve employees.6 into account the social bene�ts of taxation. In the analysis of the rural nonfarm and In virtually all countries, irrespective of their informal sectors, the focus is on the con- level of development, managers consider high straints to the benchmark firm, which is a tax rates a constraint, but cutting taxation nonexpanding � rm. The average rural and is not a priority everywhere. As the reported informal sector firm reported very little severity cannot be used to assess whether employment growth: access to � nance and tax rates are binding constraints, other indicators of the � nancial • In the Bangladesh rural survey, 81 per- sector and tax system are examined. cent of firms experienced no growth in A second limitation with the methodol- employment between 2005 and 2007 ogy is that it identi�es binding constraints to (World Bank 2008c). existing �rms. It does not identify constraints • In the survey of informal � rms conducted facing potential entrants. Regulatory con- in India in 2005, 40 percent of � rms had straints faced by potential entrants are exam- not grown since they started operations ined later in this chapter. (Ferrari and Dhingra 2009). • In the Sri Lanka rural survey, 90 percent of firms had not increased employment Constraints in the urban the previous year (World Bank 2005). formal sector This section examines constraints fac- For this reason, it was not possible to ana- ing urban formal sector � rms. It compares lyze the constraints facing an expanding �rm benchmark � rms across countries and iden- in these sectors. ti�es the constraints facing both expanding/ Even if addressing the constraints reported job-creating � rms and other types of non- for a benchmark � rm in the rural nonfarm benchmark � rms. 128 MORE AND BETTER JOBS IN SOUTH ASIA Constraints facing benchmark �rms highest-ranked constraint in India and Sri Lanka. Political instability is among the The three most common binding constraints top three constraints in all five countries facing urban formal � rms in South Asia are where it was included in the survey except electricity, corruption, and political instabil- Bhutan. This � nding is not surprising given ity (�gure 4.1 and table 4.1). Although the that South Asia contains some of the most ranking of these constraints varies across confl ict-affected areas in the world. In �ve countries, most countries identify the same of the eight countries, corruption is among top three constraints. In every country the top four constraints. except Bhutan and Maldives, electricity Though not as binding as electricity, is one of the top two constraints. It is the corruption, and political instability, tax administration is a severe concern in some FIGURE 4.1 Severity of constraints reported by South Asian countries, especially India and Pakistan. benchmark �rm in the urban formal sector Labor regulations are important in Bhutan, India, Sri Lanka, and Nepal. (Chapter 6 South Asian benchmark �rm analyzes labor regulations and other labor 2.6 market institutions in more detail). Firms in political instability most countries report tax rates and access electricity 2.2 to � nance as one of the top �ve most severe constraints. However, because of endogeni- corruption 1.8 ety issues described in the methodology, their tax administration 1.7 reported severity cannot be interpreted in the same way as the other constraints or ranked government policy uncertainty 1.4 alongside them. Given the heterogeneous nature of India, macro instability 1.3 separate analyses of firm constraints were competition 1.2 conducted for higher- and lower-income states (see annex 4A).7 The top three constraints crime, theft, and disorder 1.2 (corruption, electricity, and tax administra- tion) are the same for low- and high-income access to land 1.1 states. Firms in low-income states complain customs 1.1 more about inadequate physical infrastruc- inadequately ture (electricity, transport, access to land) 1.1 and crime; � rms in high-income states com- educated labor plain more about policy uncertainty and labor regulations 1.0 labor regulation. These results parallel cross- transport 1.0 country results that show that � rms in poor countries tend to complain more about physi- business licensing 0.9 cal infrastructure and �rms in rich countries 0.9 complain more about labor regulation (Carlin courts and Schaffer 2011b). telecoms 0.7 0 1 2 3 Electricity severity of constraint Most South Asian countries rank electricity as one of the top constraints. The reported Source: Authors, based on Carlin and Schaffer 2011b (from World Bank enterprise surveys). severity of the constraint in Afghanistan, Note: A benchmark �rm is a medium-size manufacturing �rm with 30 employees that is domesti- cally owned, does not export or import, is located in a large city, and did not expand employment Bangladesh, and Nepal is higher than in in the preceding three years. Analysis is based on pooled sample of enterprise surveys conducted other countries at similar GDP levels and between 2000 and 2010. The severity of constraint is rated by �rms on a 5-point scale, with 0 being no obstacle, 1 being a minor obstacle, 2 being a moderate obstacle, 3 being a major obstacle, and among the highest in the world. The down- 4 being a very severe obstacle. Access to �nance and tax rates constraints are excluded. ward slope in �gure 4.2 implies that although WHAT IS PRE VENTING FIRMS FROM CREATING MORE AND BE T TER JOBS? 129 TABLE 4.1 Top �ve constraints reported by South Asian benchmark �rm in the urban formal sector, by country South Asia region Afghanistan Bangladesh Bhutan India Maldives Nepal Pakistan Sri Lanka Electricity 2 2 1 1 2 2 1 Political instability 1 1 2 n.a. n.a. 1 3 n.a. Corruption 3 3 3 2 3 4 Tax administration 4 5 5 3 1 Labor regulations 3 4 5 5 Inadequately educated labor 2 5 2 Access to land 4 4 1 Transport 1 3 Government policy uncertainty 5 4 2 Courts 4 5 Crime, theft, and disorder 5 5 Business licensing 4 Macro instability 3 Competition 4 Source: Authors, based on Carlin and Schaffer 2011b (from World Bank enterprise surveys). Note: A benchmark �rm is a medium-size manufacturing �rm with 30 employees that is domestically owned, does not export or import, is located in a large city, and did not expand employment in the preceding three years. n.a. = Not applicable (question was not asked). Analysis is based on pooled sample of enterprise surveys conducted between 2000 and 2010. Access to �nance and tax rates constraints are excluded. FIGURE 4.2 Cross-country comparisons of reported severity of the electricity constraint 4 Electricity Bangladesh Nepal 3 Afghanistan severity of constraint Pakistan 2 Sri Lanka India 1 Maldives Bhutan 0 6 7 8 9 10 11 log of per capita GDP in purchasing power parity dollars Source: Carlin and Schaffer 2011b (based on World Bank enterprise surveys). Note: The cross-country regression line shows the relationship between the reported severity of the constraint for a benchmark �rm and the log of per capita GDP. The shaded area is the 95 percent con�dence interval band around the regression line. Vertical bars show con�dence intervals of 95 percent around the reported severity of the constraint for countries in South Asia. The lack of overlap between the South Asian country con�dence interval and the regression line con�dence interval is a conservative test of the statistically signi�cant difference between the reported severity of a constraint for the South Asian country and the average reported severity of constraint for countries at the same level of per capita GDP. The reported severity could still be signi�cantly different even when there is an overlap. Analysis is based on pooled sample of enterprise surveys conducted between 2000 and 2010. The severity of constraint is rated by �rms on a 5-point scale, with 0 being no obstacle, 1 being a minor obstacle, 2 being a moderate obstacle, 3 being a major obstacle, and 4 being a very severe obstacle. 130 MORE AND BETTER JOBS IN SOUTH ASIA � rms in richer countries can be expected to Afghanistan, Bangladesh, and Nepal, virtu- make more demands on the electricity grid, ally 100 percent of � rms experience outages which would lead to rising severity of com- (�gure 4.3). In Bangladesh, urban �rms face plaints, those countries are better able to an average of 98.5 power outages a month— meet increased demand, resulting in lower more than three a day (World Bank 2008c). levels of reported severity at higher incomes Predictably, the use of generators to miti- per capita. The severity of the constraint gate the effects of uncertain power supply is increased over time in India (between 2005 higher in South Asia than elsewhere, with 87 and 2010), Nepal (between 2000 and 2009), percent of � rms in Afghanistan, 52 percent and Pakistan (between 2002 and 2007 and in Sri Lanka, and 49 percent in India having between 2007 and 2010). generators. This analysis covers only the urban for- The high reported severity of the electricity mal sector. Electricity is also one of the top constraint and level of power outages reflect binding constraints in the rural nonfarm and the signi�cant demand-supply gap in electric- informal sectors, as discussed later in this ity across the region caused by the failure of chapter. electricity supply to expand rapidly enough Reported frequency of power outages—a to keep up with robust economic and popula- direct � rm-level estimate of the shared busi- tion growth (box 4.1). ness environment—is consistent with the The unreliability of power supply and the reported severity of the constraint. Indeed, in frequency of power outages causes � rms to FIGURE 4.3 Cross-country comparisons of power outages percent of �rms affected by at least one power outage a month Afghanistan Nepal 100 Bangladesh 80 60 India 40 Pakistan Bhutan 20 Maldives Sri Lanka 0 6 7 8 9 10 11 log of per capita GDP in purchasing power parity dollars Source: Carlin and Schaffer 2011b (based on World Bank enterprise surveys). Note: The cross-country regression line shows the relationship between the percentage of �rms experiencing more than one power outage per month and the log of per capita GDP. The percentage of �rms is a conditional mean for benchmark �rms in each country. As the conditional means are calculated by holding constant the effects of the �rm controls (�rm characteristics) using a simple linear procedure, the effect of this can be, for unconditional country values that are at or close to 0 percent or 100 percent, to push the conditional mean under 0 percent or over 100 percent. The shaded area is the 95 percent con�dence interval band around the regression line. Vertical bars show con�dence intervals of 95 percent around the percentage of �rms experiencing power outages for countries in South Asia. The lack of overlap between a South Asian country con�dence interval and the regression line con�dence inter- val is a conservative test of the statistically signi�cant difference between the share of �rms experiencing power outages in the South Asian country and the average share of �rms experiencing power outages in countries at the same level of per capita GDP. The share of �rms could still be signi�cantly different even when there is an overlap. Analysis is based on pooled sample of enterprise surveys conducted between 2000 and 2010. WHAT IS PRE VENTING FIRMS FROM CREATING MORE AND BE T TER JOBS? 131 lose production and incur high costs of self- (table 4.2). In addition to using generators, generation. Firms in South Asia lose a larger there is growing evidence that to cope with share of their output to power losses than unreliable supply, firms adopt second-best �rms in other regions—10.7 percent of sales production technologies, which imply large for the region as a whole, up to 27 percent efficiency gaps (Alby, Dethier, and Straub for Nepal—and spend more on generators 2010). The persistent shortage of electricity BOX 4.1 Electricity challenges facing South Asia Access to and consumption of electricity are low in average level of tariffs, subsidizing agricultural and South Asia, where some 600 million people—more domestic consumers (KPMG 2009). Nepal’s grid- than 40 percent of the worldwide total—lack access to connected installed generation capacity stands electricity. Access rates range from 44 percent of the at 698 MW, but available capacity falls to nearly population in Nepal to 77 percent in Sri Lanka (box one-third of installed capacity during the dry sea- table 4.1.1). Annual per capita electricity consump- son. Peak demand stands at 885 MW, and consum- tion is highest in Bhutan (1,174 kilowatts [KWh]) ers face 16 hours of load shedding in dry winter and lowest in Nepal (KWh 89). India consumes 566 months. In Pakistan, installed capacity is about KWh, Pakistan 436 KWh, and Sri Lanka 409 KWh. 22 gigawatts (GW), but technical problems, includ- The average annual per capita consumption for the ing lack of fuel and inadequate maintenance of region is 500 KWh—lower than in Africa (513 KWh) public sector thermal generation companies over and China (2,500 KWh) and just 3.8 percent of con- prolonged periods, reduced actual available capac- sumption in the United States (13,000 KWh) (World ity to 18 GW in 2010. Electricity demand exceeds Bank 2011e). supply by about 2– 4 GW, and rolling blackouts Robust economic and population growth over the can stretch up to 8–10 hours a day. The Ministry past two decades have led to rapidly rising energy of Finance estimates the cumulative effect of the demand. Although the 2008 global � nancial crisis energy crisis on the economy at more than 2 per- slowed economic activity, regional growth prospects cent of GDP (Government of Pakistan 2010). An remain strong, with annual growth forecast to aver- independent study conducted in 2008 estimated age 6 percent until 2015 (IMF 2010). Electricity the cost of industrial load shedding to the economy consumption is likely to increase at an annual rate at PRs. 210 billion, resulting in the loss of 400,000 of about 6.0 percent in Bangladesh, 7.0 percent in jobs and $1 billion worth of exports (IIP 2009). India, and 8.8 percent in Pakistan, the region’s larg- Industrial consumers typically pay high tariffs, est power markets (by comparison, annual growth subsidizing other consumers. The share of indus- in Organisation for Economic Co-operation and trial consumption is about 26 percent, but indus- Development countries is projected at 0.7 percent). tries contributed 30 percent to revenue, indicating The demand-supply gap is particularly wide in the extent of cross-subsidy. In Sri Lanka, the aver- some countries. In addition, industrial consumers age customer paid $0.12/KWh in 2009, consider- typically pay high tariffs and cross-subsidize other ably more than the average tariff in Bangladesh, consumer categories. In Bangladesh, the gap is esti- India, Pakistan, or Thailand of $0.05–$0.09/KWh mated to be 1,000 megawatts (MW). About 90 (World Bank 2010b). percent of generation is fired by gas, which itself Sector � nancial losses across the region are high, is in short supply. Power outages are estimated to largely as a result of the misalignment of tariffs with cost Bangladesh about $1 billion a year, reducing the cost of supply and high transmission and distri- GDP growth by about 0.5 percent (USAID 2007). bution losses, which arise from theft, faulty meter- In India, the gap between peak demand and supply ing, and poor technology in transmitting power. is about 10 percent (World Bank 2010a). Industrial The total sector de� cit in Bangladesh amounts to consumers in India pay 30 – 60 percent above the almost $300 million a year. In India, the combined (continues next page) 132 MORE AND BETTER JOBS IN SOUTH ASIA BOX 4.1 Electricity challenges facing South Asia (continued) cash loss of state-owned distribution companies distribution companies and below-cost recovery rose by nearly $22 billion a year, $9 billion attrib- tariff is estimated at about $2 billion. These de�cits utable to transmission and distribution losses and are met through subsidy payments by the govern- the rest attributable to tariffs that did not cover the ment, short-term borrowing by companies, and the cost of generation (Mint 2011). The annual sector accumulation of receivables and payables on bal- de� cit in Pakistan resulting from noncollection by ance sheets. BOX TABLE 4.1.1 Selected energy indicators in South Asia, by country Per capita Percentage electricity of population Installed Estimated investment Estimated deficit or surplus consumption with access to capacity required in the medium Country (KWh per capita) electricity (MW) MW Percent term (billions of dollars) Bangladesh 208 49 6,727 –1,000 –13 15.0 (2015) Bhutan 1,174 56 1,498 +1,300 +15 3.1 (2020) India 566 66 159,000 –15,700 –13 280.0 (2015) Nepal 89 44 698 –200 –30 — Pakistan 436 62 22,000 –4,500 –20 32.5 (2020) Sri Lanka 409 77 2,684 +800 +30 10.0 (2020) Sources: Data on access to electricity and per capita electricity consumption data are from International Energy Agency database 2010. Data on estimated supply gaps and investment requirements are from the following sources: Bangladesh, India, and Pakistan: World Bank 2010a; Bhutan: Bhutan Electricity Authority (http://www .bea.gov.bt); Nepal: Banerjee, Singh, and Samad 2011; Pakistan: Trimble, Yoshida, and Saqib 2011; Sri Lanka: Ceylon Electricity Board 2009; World Bank, 2011d. Note: Data are for most recent year available. — = Not available. TABLE 4.2 Electricity constraints faced by �rms, by developing region Percentage of firms citing electricity Average total Value lost due to as major or severe time of power outages power outages Percentage of firms Region/country constraint per month (percent of sales) owning generators Africa  49.7 69.5 6.2 42.2 East Asia and Pacific  24.1 14.4 3.1 29.6 Eastern Europe and Central Asia 35.1 23.3 3.8 16.9 Latin America and the Caribbean  39.0 8.7 4.2 19.5 Middle East and North Africa  42.9 45.9 5.6 36.3 South Asia 53.4 139.4 10.7 42.8 Afghanistan 66.2 280.3 6.4 71.1 Bangladesh 78.4 113.2 10.6 52.3 Bhutan 5.8 8.5 4.3 17.8 India 32.0 — 6.6 41.4 Nepal 75.6 226.2 27.0 15.7 Pakistan 74.5 69.0 9.2 26.3 Sri Lanka 41.3 — — 75.1 Source: World Bank Enterprise Surveys. Note: Regional averages are unweighted. Country averages are unconditional (that is, not for the benchmark �rm). — = Not available. WHAT IS PRE VENTING FIRMS FROM CREATING MORE AND BE T TER JOBS? 133 for industrial and commercial activities is more cautious regarding decisions to invest, likely to be a major factor limiting the com- expand, and hire. petitiveness of �rms and dampening employ- The importance of corruption as a con- ment growth.8 straint for � rms is reflected in both the high reported severity of the corruption constraint Corruption and the high prevalence of bribes. The rela- Corruption is the abuse of power, usually tionship between the reported severity of the for personal gain or the bene�t of a group corruption constraint and per capita GDP is to which allegiance is owed. Corruption in characterized by low levels and low variation government is de� ned as the abuse or mis- in rich countries and higher levels and more use of public of�ce or authority for private variation in low- and middle-income coun- gain and bene�t that occurs at the interface tries (figures 4.4 and 4.5). Bhutan and Sri of the public and private sectors. Lanka have lower levels of reported severity Corruption poses serious challenges to of the corruption constraint than is typical development: it undermines democracy and at their level of per capita GDP. Of the �ve good governance by subverting formal pro- countries in which corruption was among the cesses, reduces accountability and representa- top five constraints (Afghanistan, Bangla- tion in policymaking, and erodes the institu- desh, India, Maldives, and Nepal), only Ban- tional capacity of government as procedures gladesh has higher levels of reported severity are disregarded (Dininio, Kpundeh, and than typical for countries at the same level of Leiken 1988). It may also create distortions per capita GDP. and inef�ciencies by diverting public invest- The data on the prevalence of bribes are ment away from priorities and into sectors consistent with those on corruption. Firms and projects where kickbacks are more likely. in Bangladesh report a very high prevalence Corruption also shields � rms with access to of bribe payments in absolute and relative influence from competition and may prevent terms. Pakistan also has a higher preva- small companies from growing. lence of bribes than is typical at its level of Measuring the impact of corruption on per capita GDP. � rms is challenging because of the dif�culty The high reported severity and prevalence of causal attribution and the fact that corrup- of bribes is fairly consistent with the Cor- tion data are subject to measurement error. ruption Perceptions Index 2010 reported by Some observers claim that corruption reduces Transparency International (2010), which costs by cutting through red tape. In fact, ranks (out of 178 countries) Afghanistan at corruption may increase red tape, in order to 176, Bangladesh at 134, and Pakistan at 143 allow public of�cials to obtain more bribes. A (India is 87). Box 4.2 describes in more detail survey of the literature on econometric studies the nature of corruption in South Asia. that analyze the economic costs of corruption South Asian � rms report signi�cant costs suggests that corruption is bad for economic associated with corruption, including direct growth and bad for a number of factors that costs (for example, the value of bribes paid) tend to be correlated with economic growth: and indirect costs (for example, management domestic investment, the quantity and com- time spent dealing with of�cials). Enterprises position of foreign direct investment, gov- in Afghanistan pay more than 2 percent of ernment expenditures on health and educa- their sales as bribes (World Bank 2008a). tion, the quantity and quality of government Firms in India report spending an average investment in infrastructure, and the returns of 2.2 percent of sales on informal payments to business and trade. Whatever the econo- and an average 8.1 percent of managers’ metric and data problems of measuring the time dealing with government (Ferrari and impact of corruption on �nal outcomes such Dhingra 2009). as growth, corruption affects the perceptions Corruption also distorts the economy of businesses, which in turn leads them to be by creating an unlevel playing field. The 134 MORE AND BETTER JOBS IN SOUTH ASIA FIGURE 4.4 Cross-country comparison of reported severity of corruption constraint 4 3 severity of constraint Bangladesh Afghanistan Maldives 2 Pakistan Nepal India Sri Lanka 1 Bhutan 0 6 7 8 9 10 11 log of per capita GDP in purchasing power parity dollars Source: Carlin and Schaffer 2011b (based on World Bank enterprise surveys). Note: The cross-country regression line shows the relationship between the reported severity of the constraint for a benchmark �rm and the log of per capita GDP. The shaded area is the 95 percent con�dence interval band around the regression line. Vertical bars show con�dence intervals of 95 percent around the reported severity of the constraint for countries in South Asia. The lack of overlap between the South Asian country con�dence interval and the regression line con�dence interval is a conservative test of the statistically signi�cant difference between the reported severity of a constraint for the South Asian country and the average reported severity of constraint for countries at the same level of per capita GDP. The reported severity could still be signi�- cantly different even when there is an overlap. Analysis is based on pooled sample of enterprise surveys conducted between 2000 and 2010. The severity of constraint is rated by �rms on a 5-point scale, with 0 being no obstacle, 1 being a minor obstacle, 2 being a moderate obstacle, 3 being a major obstacle, and 4 being a very severe obstacle. FIGURE 4.5 Cross-country comparison of bribe payments 100 Bangladesh % of �rms reporting having made 80 bribe payments 60 Pakistan 40 Afghanistan India Maldives Sri Lanka 20 Nepal Bhutan 0 6 7 8 9 10 11 log of per capita GDP in purchasing power parity dollars Source: Carlin and Schaffer 2011b (based on World Bank enterprise surveys). Note: The cross-country regression line shows the relationship between the percentage of �rms reporting having made bribe payments and the log of per capita GDP. The percentage of �rms is a conditional mean for benchmark �rms in each country. As the conditional means are calculated by holding constant the effects of the �rm controls (�rm characteristics) using a simple linear procedure, the effect of this can be, for unconditional country values that are at or close to 0 percent or 100 percent, to push the conditional mean under 0 percent or over 100 percent. The shaded area is the 95 percent con�dence interval band around the regression line. Vertical bars show con�dence intervals of 95 percent around the percentage of �rms reporting having made bribe pay- ments for countries in South Asia. The lack of overlap between a South Asian country con�dence interval and the regression line con�dence interval is a conservative test of the statistically signi�cant difference between the share of �rms reporting having made bribe payments in the South Asian country and the average share of �rms reporting having made bribe payments in countries at the same level of per capita GDP. The share of �rms could still be signi�- cantly different even when there is an overlap. Analysis is based on a pooled sample of enterprise surveys conducted between 2000 and 2010. WHAT IS PRE VENTING FIRMS FROM CREATING MORE AND BE T TER JOBS? 135 BOX 4.2 Corruption in South Asia Enterprise surveys, which asks � rms if they have • Pakistan: Electricity (71 percent), water con- encountered bribes by different types of interac- nections (62 percent), and tax meetings (59 tions with public officials, suggest that firms in percent) South Asia face a high level of corruption in a range of interactions with public of�cials, particularly for The high frequency of bribes faced in connecting utilities and tax inspections. to power supply is another dimension of the issue of The government interactions that have the access to electricity. It may be related to businesses highest frequency of bribes vary by country (box having to compete to secure much needed power � gure 4.2.1). The highest frequency occurs in the (World Bank 2008c). following situations: More than half of firms in Bangladesh, India, • Afghanistan: Government contracts (43 percent) and Pakistan are expected to pay bribes during and electrical connection (38 percent) tax inspections. These countries also report very • Bangladesh: Utilities (42–76 percent), tax meetings high severity of tax administration (among their (54 percent), and import licenses (51 percent) top five constraints). The tax systems in these • India: Construction permits (67 percent), tax meet- countries are complex, creating not only high ings (52 percent), operating licenses (52 percent), costs of compliance but also opportunities for electrical connections (40 percent) corruption. BOX FIGURE 4.2.1 Percentage of �rms expected to give gifts to public officials, by type of interaction 100 90 80 70 60 percent 50 40 30 20 10 0 operating import license construction electrical phone water meetings with government license permit connection connection connection tax officials contract Afghanistan Bangladesh Pakistan India South Asia world Source: Authors, based on data from World Bank enterprise surveys. Note: Figures show percent of �rms in South Asian countries citing corruption as one of their top three constraints. 136 MORE AND BETTER JOBS IN SOUTH ASIA constraints analysis finds that expanding/ on average dealing with government offi- job-creating firms, manufacturing firms, cials than do service firms (manufacturing importing and exporting firms, and firms � rms reportedly pay 4.9 percent of sales in in larger cities are more constrained by cor- and devote 12.6 percent of managers’ time ruption than other � rms. The greater sever- to bribes). In Pakistan, manufacturing � rms ity probably reflects the greater use of public are also more likely than service �rms to pay services, such as electricity for manufacturing bribes across a range of interactions with �rms; higher levels of interactions with pub- public of�cials. lic of�cials (to, for example, obtain import licenses); and the fact that larger cities have Political instability higher concentrations of economic resources. In the �ve countries in which surveys asked These firms are also the most productive, about the political instability constraint making the opportunity cost of corruption— (Afghanistan, Bangladesh, Bhutan, India, goods and services not produced because of Nepal, and Pakistan), all except Bhutan resources and time dedicated to overcoming reported it as one of the top three constraints. corruption activities—even higher. The India The reported costs of political instability investment climate assessment (Ferrari and are higher in Afghanistan, Bangladesh, and Dhingra 2009) indicates that manufactur- Nepal than in other countries at similar lev- ing firms on average pay a higher percent- els of per capita GDP (�gure 4.6). These three age of sales in bribes and spend more time countries have some of the highest reported FIGURE 4.6 Cross-country comparison of reported severity of political instability constraint 4 Nepal Afghanistan 3 Bangladesh severity of constraint Pakistan 2 1 Bhutan 0 6 7 8 9 10 log of per capita GDP in purchasing power parity dollars Source: Carlin and Schaffer 2011b (based on World Bank enterprise surveys). Note: The cross-country regression line shows the relationship between the reported severity of the constraint for a benchmark �rm and the log of per capita GDP. The shaded area is the 95 percent con�dence interval band around the regression line. Vertical bars show con�dence intervals of 95 percent around the reported severity of the constraint for countries in South Asia. The lack of overlap between the South Asian country con�dence interval and the regression line con�dence interval is a conservative test of the statistically signi�cant difference between the reported severity of a constraint for the South Asian country and the average reported severity of constraint for countries at the same level of per capita GDP. The reported severity could still be signi�- cantly different even when there is an overlap. Analysis is based on a pooled sample of enterprise surveys conducted between 2000 and 2010. The severity of constraint is rated by �rms on a 5-point scale, with 0 being no obstacle, 1 being a minor obstacle, 2 being a moderate obstacle, 3 being a major obstacle, and 4 being a very severe obstacle. Enterprise surveys in India, Maldives, and Sri Lanka did not ask �rms about political instability. WHAT IS PRE VENTING FIRMS FROM CREATING MORE AND BE T TER JOBS? 137 costs of political instability in the world. organize their activities and produce goods (Chapter 7 examines the effect of armed and services, which in turn affects the struc- conflict, which is one key aspect of political ture of taxation (Bird and Zolt 2007). instability, on �rms and workers.) Firms in India and Pakistan report tax rates as their top constraint in terms of sever- Tax administration ity; Sri Lankan � rms report tax rates to be Tax administration is a major issue in some the third-most severe constraint. However, countries in South Asia (table 4.1). It is the as explained in the methodology, one cannot top-ranked constraint in Pakistan, where directly infer the real cost of the constraint the reported severity is above the average for from the reported severity. Other measures countries at similar levels of per capita GDP have been used to assess whether tax rates are (�gure 4.7). As annex 4B shows, many tax indeed a binding constraint (see annex 4B). systems in South Asia are complex, making It seems likely that it is the complexity and tax compliance very costly for �rms. lack of uniformity of tax systems rather than the tax rates themselves that impose signi�- Tax rates cant costs on � rms. These tax systems create Taxes matter to business and individuals. distortions in the economy and make it dif- Businesses and people talk about them, �cult to raise adequate revenues. Simplify- complain about them, and often try to ing the tax regimes (by reducing the number avoid paying them when they can. Taxation of taxes and exemptions) and widening the also affects the ways in which businesses tax base is likely to bene�t � rms and remove FIGURE 4.7 Cross-country comparison of reported severity of tax administration constraint 4 3 severity of constraint Pakistan 2 Bangladesh India Afghanistan Sri Lanka Bhutan 1 Nepal Maldives 0 6 7 8 9 10 11 log of per capita GDP in purchasing power parity dollars Source: Carlin and Schaffer 2011b (based on World Bank enterprise surveys). Note: The cross-country regression line shows the relationship between the reported severity of the constraint for a benchmark �rm and the log of per capita GDP. The shaded area is the 95 percent con�dence interval band around the regression line. Vertical bars show con�dence intervals of 95 percent around the reported severity of the constraint for countries in South Asia. The lack of overlap between the South Asian country con�dence interval and the regression line con�dence interval is a conservative test of the statistically signi�cant difference between the reported severity of a constraint for the South Asian country and the average reported severity of constraint for countries at the same level of per capita GDP. The reported severity could still be signi�- cantly different even when there is an overlap. Analysis is based on a pooled sample of enterprise surveys conducted between 2000 and 2010. The severity of constraint is rated by �rms on a 5-point scale, with 0 being no obstacle, 1 being a minor obstacle, 2 being a moderate obstacle, 3 being a major obstacle, and 4 being a very severe obstacle. 138 MORE AND BETTER JOBS IN SOUTH ASIA distortions in the economy. By increasing � rms is likely to produce large increases in compliance, governments can maintain tax output as well. revenues while reducing statutory rates. Job-creating � rms—which are similar in all respects to the benchmark firm except that they expanded employment during the Constraints facing expanding/ preceding three years—report higher levels job-creating �rms of severity than the benchmark � rm for 14 The benchmark � rm is useful for comparing of the 16 constraints (�gure 4.8). They also the costs of constraints to � rms across South report higher levels of mitigation activities Asia and the rest of the world. However, the (paying bribes and using generators). This main interest here is in understanding how pattern holds strongly in India and Paki- these constraints affect employment creation. stan, where the difference between the cost This section therefore examines the reported of constraints facing expanding �rms and the cost of constraints facing expanding/job- benchmark �rm is largest.10 creating � rms, because relaxing constraints Although the level of severity changes, on these � rms is likely to produce the largest job-creating and benchmark � rms in most increases in employment. Job-creating � rms countries rank constraints in a similar also perform well: employment growth is order: electricity, corruption, and politi- positively and signi�cantly correlated with cal stability are the top three constraints research and development, introduction of among both types of � rms, and the rank- new processes and products, sales to mul- ings of other constraints are similar (table tinational companies, the education level of 4.3). Only in Pakistan do the rankings by the manager, and in-house training.9 There- the benchmark and job-creating � rms differ: fore, relaxing constraints on job-creating corruption becomes the joint top constraint for expanding � rms. Using alternative per- FIGURE 4.8 Severity of constraints reported by South Asian formance measures, such as productivity benchmark (nonexpanding) and expanding �rm in the urban formal and innovation, reveals a similar pattern: sector well-performing � rms report higher costs but similar ranking of constraints as poorly telecoms performing � rms. political instability 3.0 courts 2.5 electricity business licensing 2.0 Constraints facing nonbenchmark �rms 1.5 corruption 1.0 transport In general, the ranking of constraints facing 0.5 different types of � rms in the urban formal tax administration 0.0 labor regulations manufacturing sector is very similar to the ranking by the benchmark �rm, although the government policy level of severity differs. The patterns described inadequately uncertainty educated labor below are consistent across countries in South macro instability Asia (�gure 4.9). customs Service �rms generally report lower sever- competition access to land ity of constraints than manufacturing �rms. crime, theft, and disorder Firms in the service sector are typically less benchmark (nonexpanding) expanding capital intensive, more dependent on commu- nications, and less engaged in trade than man- Source: Authors, based on Carlin and Schaffer 2011b (based on World Bank enterprise surveys). ufacturing �rms. Indeed, service �rms in South Note: Analysis is based on a pooled sample of enterprise surveys conducted between 2000 and 2010. A point farther away from the origin indicates that the business constraint is considered more Asia report lower constraints for electricity, severe. The severity of constraint is rated by �rms on a 5-point scale, with 0 being no obstacle, 1 labor regulations, and custom administration being a minor obstacle, 2 being a moderate obstacle, 3 being a major obstacle, and 4 being a very severe obstacle. Only statistically signi�cant differences in reported severity between benchmark (but higher cost for telecommunications). The and nonbenchmark �rms are shown. Access to �nance and tax rates constraints are excluded. difference between manufacturing and service WHAT IS PRE VENTING FIRMS FROM CREATING MORE AND BE T TER JOBS? 139 TABLE 4.3 Top �ve constraints reported by South Asian benchmark (nonexpanding) and expanding �rm in the urban formal sector, by country South Asian region Benchmark firm Afghanistan Bangladesh Bhutan India Maldives Nepal Pakistan Sri Lanka Benchmark firm Benchmark firm Benchmark firm Benchmark firm Benchmark firm Benchmark firm Benchmark firm Benchmark firm Expanding firm Expanding firm Expanding firm Expanding firm Expanding firm Expanding firm Expanding firm Expanding firm Expanding firm Constraint Electricity 2 2 2 2 1 1 1 2 2 2 2 1 1 Political instability 1 1 1 1 2 2 n.a. n.a. 1 1 3 1 n.a. n.a. Corruption 3 3 3 4 3 3 2 1 3 4 4 4 1 Tax administration 4 4 5 5 5 5 3 3 1 4 Labor regulations 3 3 4 4 5 5 5 5 Inadequately educated labor 2 2 5 5 2 2 Access to land 4 5 4 4 1 1 Transport 1 1 3 3 Government policy uncertainty 5 5 4 3 2 2 Courts 4 5 5 Crime, theft, and disorder 5 3 5 3 Business licensing 4 4 Macro instability 5 3 3 Competition 4 4 Source: Authors, based on Carlin and Schaffer 2011b (based on World Bank enterprise surveys). Note: Analysis is based on pooled sample of enterprise surveys conducted between 2000 and 2010. n.a. = Not applicable (question was not asked). Access to �nance and tax rates constraints are excluded. � rms is particularly marked in India, where in India and larger in Nepal, Pakistan, and service �rms report far less severe constraints Sri Lanka. In Nepal, and to a lesser extent Sri than manufacturing �rms in electricity, labor Lanka, the cost of labor regulations increases regulations, customs, tax rates and adminis- signi�cantly with �rm size (see annex 4C). tration, and access to �nance.11 Exporters and importers in South Asia Relative to smaller firms, larger formal report more severe constraints, except in �rms which, on average produce higher out- Bangladesh. Importers are more constrained put per worker (see chapter 3), report more by customs administration and anticom- severe constraints and a larger number of mit- petitive behavior. Macroeconomic instabil- igation activities, including management time ity is more problematic for both types of spent on regulations and use of generators.12 � rms, which is likely to reflect sensitivity to They are also subject to more inspections. exchange rate movements and uncertainty. The difference between the benchmark � rm Exporters in Bangladesh report lower costs with 30 employees and a �rm with the same of constraints relative to nonexporters for characteristics but with 60 employees is small electricity and most aspects of institutional 140 MORE AND BETTER JOBS IN SOUTH ASIA FIGURE 4.9 Severity of constraints reported by South Asian corruption would bene�t all types of � rms. benchmark (manufacturing) and service �rm in the urban But larger job-creating �rms in the manufac- formal sector turing sector would bene�t most, as the costs of the constraints are higher for them. a. South Asia telecoms political instability 3.0 courts Constraints in the rural nonfarm 2.5 and informal sectors electricity business licensing 2.0 This section looks at the severity of con- corruption 1.5 transport straints on �rm operations reported by micro 1.0 benchmark � rms (� rms with no more than 0.5 tax administration 0.0 labor regulations �ve employees) in the rural nonfarm sector and informal (urban) sector and compares them with those reported by micro bench- government policy inadequately uncertainty educated labor mark � rms in the urban formal sector. The rural nonfarm sector has been shown to macro instability customs offer higher-paid jobs to rural workers than competition access to land the agricultural sector, with wages on par crime, theft, and disorder with or even higher than similar categories b. India of employment in the urban sector. The courts growth of rural nonfarm enterprises will electricity 3.0 telecoms play an important role in the creation of bet- 2.5 ter jobs. Although informal �rms are smaller corruption 2.0 access to land and less productive than formal � rms, they 1.5 employ the most workers. Addressing their tax administration 1.0 transport constraints is therefore important in creating 0.5 better jobs. 0 labor regulations competition Rural nonfarm sector inadequately macro instability educated labor In general, rural firms face fewer and less government policy severe constraints than urban firms (fig- business licensing uncertainty ure 4.10 and table 4.4). This pattern is not customs crime, theft, and disorder unusual in developing countries, where urban benchmark (manufacturing) services �rms are typically more productive and place more demands on public services than their Source: Authors, based on Carlin and Schaffer 2011b (based on World Bank enterprise surveys). rural counterparts. Electricity, macroeco- Note: Analysis is based on a pooled sample of enterprise surveys conducted between 2000 and 2010. A point farther away from the origin indicates that the business constraint is considered more nomic instability, and political instability severe. The severity of constraint is rated by �rms on a 5-point scale, with 0 being no obstacle, 1 remain important constraints for rural �rms. being a minor obstacle, 2 being a moderate obstacle, 3 being a major obstacle, and 4 being a very severe obstacle. Only statistically signi�cant differences in reported severity between benchmark Transport is a greater concern for rural �rms and nonbenchmark �rms are shown. Access to �nance and tax rates constraints are excluded. than it is for urban �rms. Rural �rms are less concerned about corruption, taxes, and labor regulations than their urban counterparts. infrastructure pattern (see annex 4C). This Like urban �rms, rural �rms rank electric- pattern may be explained by the large num- ity as one of their top constraints and report ber of export processing zones in Bangladesh, similar levels of power outages. In Bangla- which offer better infrastructure and regula- desh, 73 percent of nonmetropolitan nonfarm tory environments. enterprises have an electricity connection. Of In conclusion, addressing the top con- �rms without electricity: 32 percent complain straints of electricity, political stability, and about high connection costs, 32 percent report WHAT IS PRE VENTING FIRMS FROM CREATING MORE AND BE T TER JOBS? 141 not needing electricity, and 25 percent report FIGURE 4.10 Severity of constraints reported by micro no supply in their location. Among �rms with benchmark �rm in urban and rural sectors of Bangladesh, connections, 99 percent report outages. In Pakistan, and Sri Lanka Pakistan, the median number of outages per month is 20 days in villages and 15 days in telecoms electricity 3.0 labor regulations small towns. In Sri Lanka, the shortage and 2.5 unreliability of power is exacerbated in rural political instability 2.0 customs areas where less than 70 percent of enter- 1.5 prises use electricity from the national grid. 1.0 corruption transport The propensity for rural households to start 0.5 up a nonfarm enterprise is 14 percent lower in 0.0 areas where electricity is found to be a major macro instability inadequately constraint.13 educated labor Rural �rms also rank macroeconomic and access to land business licensing political instability as important constraints. One should be cautious in interpreting these government tax administration results, however, as the macroeconomic insta- policy uncertainty courts bility question was asked only in Bangladesh micro benchmark (urban) micro benchmark (rural) and refers to inflationary concerns and the political instability question was asked only Source: Authors, based on Carlin and Schaffer 2011a (based on World Bank enterprise surveys). in Bangladesh and Pakistan. Note: Analysis is based on a pooled sample of enterprise surveys conducted in Bangladesh, Pakistan, Rural � rms (especially microenterprises) and Sri Lanka between 2000 and 2010. A point farther away from the origin indicates that the business constraint is considered more severe. The severity of constraint is rated by �rms on a report access to �nance as one of the top con- 5-point scale, with 0 being no obstacle, 1 being a minor obstacle, 2 being a moderate obstacle, 3 straints. In Pakistan, more than 30 percent of being a major obstacle, and 4 being a very severe obstacle. Only statistically signi�cant differences in reported severity between the micro benchmark �rms in urban and rural sectors are shown. small towns and rural �rms report access to Access to �nance and tax rates constraints are excluded. The low demand constraint is excluded �nance, cost of �nance, and loan procedures because it was asked only in the rural surveys and is susceptible to endogeneity issues. Crime, theft, and disorder as well as competition constraints were not asked in the rural surveys. Macro instability (separately) as a major or severe constraint. was asked only in Bangladesh, and political instability was asked only in Bangladesh and Pakistan. In contrast, none of the other constraints was ranked as major or severe by more than TABLE 4.4 Top �ve constraints reported by micro benchmark �rm in 16 percent of � rms. These results cannot be the urban and rural sectors of Bangladesh, Pakistan, and Sri Lanka interpreted directly as evidence that access to Bangladesh, Pakistan, Sri Lanka �nance is a constraint to operations, though Constraint Urban, formal Rural the � nding is consistent with other evidence that suggests that access to � nance may be Electricity 1 2 an issue for micro and small �rms (including Political instability 2 3 Corruption 3 rural nonfarm enterprises) (see annex 4D). Macro instability 4 1 In Bangladesh, India, and Pakistan, rural Access to land 5 � rms perceive low demand for their products Transport 4 as one of the top �ve constraints, reflecting Source: Authors, based on Carlin and Schaffer 2011a (based on World Bank enterprise surveys). low access to larger markets.14 Rural � rms Note: Analysis is based on a pooled sample of enterprise surveys conducted in Bangladesh, Pakistan, sell predominantly to local customers, limit- and Sri Lanka between 2000 and 2010. Access to �nance and tax rates constraints are excluded. The low demand constraint is excluded because it was asked only in the rural surveys and is susceptible ing the size of the market for their products to endogeneity issues. Crime, theft, and disorder as well as competition constraints were not asked and services. In Bangladesh, the majority in the rural surveys. Macro instability was asked only in Bangladesh and political instability was asked only in Bangladesh and Pakistan. Only the top four constraints are shown for rural �rms, as the of nonmetropolitan enterprises sell their remaining constraints were not reported on average as obstacles. goods and services directly to customers located within the same union or ward; less networks, poor contract enforcement, and than 1 percent of nonmetropolitan enter- low quality of production are the main fac- prises have subcontracting arrangements tors behind limited access to larger markets with metropolitan firms.15 Infrastructure and subcontracting. Concerns over quality bottlenecks, limited contacts and business lead to vertical integration of metropolitan 142 MORE AND BETTER JOBS IN SOUTH ASIA � rms. A quarter of the � rms that did not analysis using Nepal census data also subcontract cited difficulties in ensuring shows that rural nonfarm employment has quality control; another quarter cited lack a statistical and practical positive associa- of predictability and reliability of supply tion with income growth in nearby urban (World Bank 2008c). centers.16A one percentage point decline in Low market demand does not necessar- urban poverty is associated with a 0.1–0.2 ily reflect market failure. It is affected by percentage point increase in the rural non- the �rms’ characteristics (a poor-performing farm employment rate. Urban growth can � rm that does not make goods suitable for increase the demand for goods and ser- the market will face low demand). Endogeni- vices produced by the rural nonfarm sec- ety issues make it dif�cult to interpret high tor, which otherwise might be limited by reported severity as a business environment the small size of their local markets. The constraint to �rms. rural nonfarm sector can produce goods Low market demand is consistent with to be directly consumed by urban consum- � rms citing transport (road conditions and ers or by urban firms through backward accessibility of roads) as one of their top four contracting arrangements. Linkages to constraints. Nonmetropolitan enterprises in urban markets are hindered by weak trans- Bangladesh identify inaccessibility of roads port infrastructure connecting rural areas during certain seasons as a major or severe to urban areas (as well as lack of market constraint. Regression analysis shows that the information and the inability of rural non- greater the distance to the nearest medium- farm firms to produce goods demanded size city (a city with a population of 100,000 in the urban areas). Assessments of rural or more), the greater the likelihood that a transport in South Asia would help policy nonmetropolitan manufacturer will identify makers develop speci�c options. low demand as a major or severe constraint. Rural �rms are much less concerned than In Pakistan, 16 percent of rural enterprises urban �rms with corruption, taxes, and labor surveyed (far more than small town enter- regulations. This �nding is consistent with the prises) cite road quality and 14 percent cite fact that rural � rms report less engagement availability of transport as a major or severe with of�cials than urban � rms (inspections constraint. In Sri Lanka, nearly a third of and management time) and that about half rural enterprises (versus 20 percent of urban of rural �rms are informal (just 40 percent of enterprises) cite transport as a major or severe rural � rms in Bangladesh and 53 percent in obstacle to operation or growth of business. Sri Lanka were registered). Among � rms that describe transport as an obstacle, 35 percent cite road quality, 33 percent cite lack of access to roads, and 32 Informal urban sector percent cite lack of available transport as the Figure 4.11 and table 4.5 contrast the rank- underlying issue. These results are not sur- ing and severity of constraints reported by prising, as poor transport limits access to micro benchmark � rms in the formal and larger, urban markets. informal sectors of urban areas of India. Good-quality transport infrastructure is Like formal firms, informal firms rank critical for facilitating urban-rural nonfarm electricity as one of their top constraints and sector linkages. The India poverty assess- report similar levels of power outages. For- ment (World Bank 2011b) uses the varia- mal � rms in India cite power as their great- tion in the nonfarm sector across India est constraint; informal � rms cite it as their to explore the determinants of rural non- second-greatest constraint (after access to farm growth. It � nds that the expansion of � nance). Power outages and loss of sales as employment in the nonfarm sector is more a result of outages are very high for informal closely linked to urban than agricultural firms (11.3 percent of annual sales lost to growth. A simple multivariate regression power outages). WHAT IS PRE VENTING FIRMS FROM CREATING MORE AND BE T TER JOBS? 143 Informal �rms cite lack of access to �nance FIGURE 4.11 Severity of constraints reported by micro benchmark as the top constraint they face: in 2005, more �rm in India’s urban formal and informal sectors than a third of informal �rms viewed it as the courts single greatest obstacle. In contrast, formal 3.0 electricity telecoms �rms cite access to �nance as the �fth-greatest 2.5 corruption macro instability obstacle. This finding does not necessarily 2.0 mean that access to �nance is a binding con- 1.5 tax administration 1.0 business licensing straint for informal �rms (see the earlier dis- 0.5 cussion of the endogeneity problem), nor does 0.0 it necessarily mean that access to �nance is a crime, theft, and disorder competition greater constraint to informal �rms than for- mal �rms. However, given the overwhelming access to land customs response by informal �rms, this issue merits labor regulations government policy the analysis reported in the next section. inadequately uncertainty transport Informal � rms are more concerned about educated labor access to land and transport than formal micro benchmark (formal) micro benchmark (informal) � rms and (like rural � rms) less concerned about corruption, taxes, and labor regula- Source: Authors, based on Carlin and Schaffer 2011a (based on World Bank enterprise surveys). Note: Analysis is based on a pooled sample of enterprise surveys conducted in India between 2000 tions. This finding is consistent with the and 2010. A point farther away from the origin indicates that the business constraint is considered fact that informal � rms report less manage- more severe. The severity of constraint is rated by �rms on a 5-point scale, with 0 being no obstacle, 1 being a minor obstacle, 2 being a moderate obstacle, 3 being a major obstacle, and 4 being a ment time spent with of� cials than formal very severe obstacle. Only statistically signi�cant differences in reported severity between the � rms. micro benchmark �rms in formal and informal sectors are shown. Access to �nance and tax rates constraints are excluded. The low demand constraint is excluded because it was asked only in the The concern regarding access to land informal surveys and is susceptible to endogeneity issues. Political instability was not asked in the expressed by urban informal � rms in India India surveys. may reflect the impact of regulations that shape the operation of land markets. Den- TABLE 4.5 Top �ve constraints reported by micro benchmark �rm in sity regulations, which limit the ratio of floor India’s urban formal and informal sectors space to plot area, lead cities to expand out- ward instead of upward. Together with lim- India ited accessibility of public transport, such Formal Informal expansion can make it more difficult for Electricity 1 1 informal manufacturing units to locate where Corruption 2 they should—close to buyers and suppli- Tax administration 3 ers. Relaxing density regulations, improving Crime, theft, and disorder 4 urban transport, and increasing the supply Access to land 5 3 of property might help reduce the severity of Competition 2 the land constraint for informal urban �rms Transport 4 (World Bank 2011a). Government policy uncertainty 5 Source: Authors, based on Carlin and Schaffer 2011a (based on World Bank enterprise surveys). Note: Analysis is based on a pooled sample of enterprise surveys conducted in India between 2000 Access to �nance and 2010. Access to �nance and tax rates constraints are excluded. The low demand constraint is excluded because it was asked only in the informal surveys and is susceptible to endogeneity A growing body of empirical research has issues. Political instability was not asked in the India surveys. established a positive association between � nancial sector and economic development Beck, Demirgüç-Kunt, and Maksimovic (countries with higher levels of credit to the 2005). Though in general it is very dif�cult private sector relative to GDP experienced to control for endogeniety in these types of higher real per capita GDP growth rates analyses, the studies suggest that if � nance between 1980 and 2007) and between access is indeed a binding constraint, it could have to external finance and firm productivity a signi�cant adverse effect on productivity (see, for example, Banerjee and Duflo 2008; and growth. 144 MORE AND BETTER JOBS IN SOUTH ASIA Firms in all South Asian countries report relevance of each option for an individual access to �nance as one of their top �ve con- country will depend on country-specific straints. The reported severity is higher for issues and contexts. informal, rural, and small � rms. To circum- vent endogeniety issues, this book looks at Electricity policy options alternative indicators, including financial depth and outreach/access indicators, to see Alleviating the demand gap across all seg- whether there is other evidence suggesting ments in the value chain requires at least that access to �nance is a severe constraint to $300 billion by 2020 ($280 billion for India � rms (see annex 4D). It does not � nd strong only) (see box 4.1). The private sector is pro- evidence that access to � nance is a binding jected to continue to be the leading � nan- constraint for all types of firms in South cier. Since 1990, it has invested $121 billion Asia. Firm-level data from enterprise sur- in the sector, 85 percent since 2000 (table veys show that smaller � rms in South Asia 4.6). India has attracted about 90 percent use less external finance (this evidence is of the regional investment in the power sec- consistent with evidence from other regions). tor with private participation (World Bank Small firms in Afghanistan and Pakistan 2011c). The private sector’s contribution to use less external � nance than small � rms in total power sector investments in India rose other regions. Microenterprises in the infor- from 21 percent under the 10th Plan to 36 mal and rural sector report even lower use percent under the 11th Plan; it is expected of external � nance than small � rms in the to reach about 50 percent in the 12th Plan, formal sector, reflecting less access to bank- beginning in 2012 (Economic Times 2011; ing services in general and the large number Government of India 2009). of � rms that are unable to apply for credit If these improvements in closing the elec- because they cannot meet the loan require- tricity gap are to be sustained or increased, ments, especially with regard to collat- countries will need to continue to create eral. The evidence suggests that addressing enabling environments to improve sector access to � nance issues for micro, small, and performance and foster additional invest- medium-size enterprises may be important ment. Key actions include deepening market in some countries.17 and institutional reforms and improving the After studying the evidence on whether commercial and financial viability of the small and medium-size enterprises in Asia sector. are credit constrained, the Asian Develop- ment Bank (2009) concluded that studies Deepening market and institutional reforms are subject to numerous endogeniety issues. Beginning in the mid- to late 1990s, coun- It noted that although the evidence may be tries put in place overarching policy and inconclusive, policy makers believe lack of legal frameworks and restructured power credit is an important issue and have already utilities to make investments more attrac- introduced many active interventions. The tive and commercialize utilities to enhance next section assesses which of these inter- service delivery. Most countries established ventions is technically appropriate and eco- regulatory bodies; separated segments nomically justi�ed. (generation, transmission, and distribu- tion); and implemented a host of measures to attract the private sector, particularly in Demand-side policy options generation. (See table 4.7 for a summary This section suggests policy options to of market reforms in the power sector in overcome some of the binding constraints South Asia.) Bangladesh and Pakistan also faced by � rms in South Asia. Policy options unbundled their power markets to some are based on experiences from coun- extent, with the prevailing market model tries inside and outside the region. The the single buyer with independent power WHAT IS PRE VENTING FIRMS FROM CREATING MORE AND BE T TER JOBS? 145 producers or rental power plants selling TABLE 4.6 Private investment in electricity in South Asia, by their power at regulated prices. The unbun- country, 1999–2000 and 2001–10 dling of the Water and Power Develop- (millions of dollars) ment Authority in Pakistan into a number Country 1999–2000 2001–10 Total of generation companies was formalized in Afghanistan 0 2 2 2001 and 2002. Bhutan unbundled only the Bangladesh 573 823 1,397 power generation segment by establishing Bhutan 0 201 201 a power company responsible for domes- India 11,039 97,683 108,722 tic power (Bhutan Power Corporation) and Nepal 239 34 273 another company responsible for operating Pakistan 5,937 4,496 10,433 the export-oriented hydropower projects Sri Lanka 177 271 488 (Druk Green Power Company). In India, Total 17,966 103,510 121,475 the generation, transmission, distribution Source: World Bank PPI database (http://www.ppi.worldbank.org). segments include a number of players (both Note: Data on Maldives are not available. TABLE 4.7 Market reforms in the power sector in selected countries in South Asia Market Role of Country structure Regulator public sector Role of private sector Role of competition Bangladesh Unbundled Bangladesh Companies created Government launched program Competition limited and confined Electricity after unbundling in 2010 to attract independent to generation sector Regulatory continue to be power producers and boost Commission state owned. private investment in generation. Because of gas shortages, most independent power producers and rental plants are small and oil fired. India Unbundled Electricity Public sector owns Private sector assets and financing Long-term agreements by regulatory large shares of are concentrated in generation; independent power producers, commissions in the transmission limited private presence in public-private partnerships most states and distribution transmission and negligible in transmission, distribution segments. presence in distribution. franchisees. Policy framework shifting from guaranteed return on assets to tariffs based on competitive bidding Nepal Horizontally None Public sector owns Significant participation of Competition introduced in integrated large shares of independent power producers in generation. the transmission generation. and distribution segments. Pakistan Unbundled National electric Companies created Independent power producers Further strengthening of the power regulatory after unbundling own one-third of generation regulatory framework envisaged authority continue to be capacity. to enable competition and market- state owned. oriented systems for power pricing, trading, and exchange. Sri Lanka Horizontally Multisector Public sector owns Limitation of private investment Limited integrated public utilities large shares of above 25 MW unless there is regulatory the transmission shareholding by the government. and distribution Independent power producers segments. supply oil-dependent power to the grid. Source: Authors’ compilation, based on Besant-Jones 2006. 146 MORE AND BETTER JOBS IN SOUTH ASIA public and private) as well as electricity development of which has been limited by traders, who participate in a competitive political instability). electricity wholesale market. Nepal and There is more scope for private players in Sri Lanka retain their vertically integrated the transmission and distribution sectors, state-owned company. where investments have been considerably Countries also have to strengthen insti- lower than in generation, reflecting the tutions in order to ef�ciently expand capac- natural monopoly nature of transmission ity. The region has pursued ambitious plans and difficulties in tackling longstanding to meet power shortages. In India alone, underpricing and theft in distribution. In power sector investments have to increase India, several public-private partnership from about $7 billion a year to about $30– contracts have been competitively awarded $40 billion a year by 2012, under the 11th on a Build, Own, Operate, and Maintain Plan, which calls for total investment of or a Design, Build, Finance, Operate, and $150 billion. Total investment is expected Transfer basis. India and Pakistan have to double to $300 billion under the 12th also pursued private participation in power Plan (2012–17) (Banerji and Mishra 2010). distribution as a means of improving the India added 50 GW between 2006 and 2011 commercial orientation and service deliv- and implemented a series of “ultra mega� ery. In India, power distribution has been (4,000 MW) generation projects based on privatized in Delhi and the state of Orissa, competitive bidding by independent power and some states, such as Maha rashtra, are producers. It also set up a functioning trad- developing “franchise� models akin to lease ing market for contracts of less than a year. contracts to bring in the private sector. Pak- In 2010/11, the short-term power market istan privatized distribution in the Karachi constituted 10 percent of the total power Electricity Supply Corporation. The results procured in India (Central Electricity Regu- of these initiatives have been mixed, reflect- latory Commission 2011). ing the difficulties in moving to and sus- Bangladesh plans to develop 9,426 MW taining cost-covering tariffs and reducing of new generation capacity by 2015, at an nonpayment and theft. estimated cost of about $15 billion, includ- Skill mapping and skill development need ing associated transmission and distribution to appear on top of the business agenda investments and fuel development and sup- among sector players, as the lack of skilled ply facilities. Bhutan has established public- manpower is emerging as a constraint hin- private partnerships and secured financing dering massive investment. The active par- on commercial terms for a large export- ticipation of private players in the sector has oriented hydropower project (Dagachhu resulted in higher attrition rates from the HPP). Between 2008 and 2010, Pakistan public sector and steep competition for talent. added more than 2,000 MW of capacity. It In India, demand for manpower is expected expects to commission 300 MW of hydro- to increase through 2013/14, when the sector power by 2012. will require an estimated 85,000 additional Further improvements in the operating workers (CRISIL 2011). environment are needed to increase invest- ments in generation. Generation shortages are Exploiting the signi�cant potential of caused primarily by procedural bottlenecks intraregional energy trade (land acquisition, environmental and forest Countries need to build on the unexploited clearances, and provision of water for ther- potential of interregional trade and improve mal plants); limited technical and � nancial the ef�ciency of existing investments, two of capacity to implement large projects; the lack the most cost-effective options for alleviating of fuel and the inability to import it; and dif- shortages in the region. Interregional trade, �culty in accessing domestic resources (such particularly hydro exports from Bhutan to as coal in India, which lies under forests or India, has increased in recent years. Ini- in tribal areas, or hydropower in Nepal, the tiatives continue, with the government WHAT IS PRE VENTING FIRMS FROM CREATING MORE AND BE T TER JOBS? 147 of Bangladesh signing a memorandum of Many states in India have undertaken understanding with India in 2010 for cross- rural feeder segregation programs to reduce border power trade that will facilitate the losses in rural distribution lines and provide import of 500 MW of electricity and a new uninterrupted power to rural nonagricultural Nepal-India electricity and trade project that consumers. These mechanisms have had a will facilitate the trading of up to 1,000 MW transformative effect on rural livelihoods in of electricity. As a result of transmission Gujarat, the state where the scheme has func- bottlenecks, electricity exchange between tioned the longest and been subject to many Nepal and India was limited to about 70 evaluations (box 4.3). MW in 2010. The potential for more ef� - Tariff rationalization—that aligns tar- ciently deploying existing generation capac- iffs with costs and better targets implicit ity through rehabilitation, maintenance, and subsidies—is critical. In Bangladesh, policy technical optimization is also important. makers envisage addressing the sector’s financial deficit through a combination of Improving the commercial and �nancial tariff increases, cost reduction, and demand- viability of the sector side management. Retail tariffs were raised Governments in the region have been pro- by 5 percent in February 2011, but additional active in curbing losses in the power sector. tariff increases are required. In Bhutan, the India’s power development reform program government adopted a more cost-reflective (R-APDRP), launched under the 11th Plan, tariff structure after the corporatization of the which aims to limit losses to 15 percent, electricity utility in 2002. During 2004–09, focuses on demonstrable and measurable the average electricity tariff increased by performance in reducing distribution losses. 66 percent. Even after the increase, however, Pakistan has conducted technical and opera- tariffs need to rise 73 percent for residential tional audits of distribution and generation customers and 28 percent for industrial con- companies, which provide a menu of actions sumers to cover the cost of production (ADB for the government and utilities to imple- 2010). In Pakistan, consumer tariffs were ment (for instance, investments to reduce not increased at all between 2003 and 2007, losses and improve ef�ciency, administrative despite substantial increases in the interna- and commercial steps to enhance account- tional prices of fuel oil and gas. Changes in ability, and upgrading of information sys- the tariff level and structure carried out in tems and management practices). March 2008 were implemented to reduce the BOX 4.3 Bringing light to rural consumers in Gujarat, India Although most villages in the Indian state of Gujarat a cost of $285 million in a record 1,000 days from ini- were electri�ed by 2003, the power supply remained tiation to implementation. The scheme was expected erratic and unreliable, because the feeders to rural to provide better quality power supply to nonagricul- areas were joined and utilities had no incentive to sup- ture consumers, charge farmers for power use, and ply 24-hour power to a group of mostly unmetered ration the power supply to agriculture consumers. consumers paying below cost-recovery tariff. The The impact of the scheme has been signi� cant: Jyotigram Yojana (JGY) aimed to segregate agricul- both the quantity and quality of power and enter- ture from rural domestic and industrial power supply. prise operations improved. As a result, families can Feeders were laid to provide 24/7 uninterrupted sup- now work at night. The Gujurat Electricity Board ply, and farmers were rationed an uninterrupted, pre- also experienced increases in revenue collection and announced 8-hour supply. The scheme was launched made pro�ts in 2010. in 2003 on a pilot basis and then scaled up to include the whole state. The scale-up created 1,888 feeders at Sources: Devaiah 2010; Mishra 2010. 148 MORE AND BETTER JOBS IN SOUTH ASIA �scal burden of tariff subsidies and improve interference in day-to-day utility matters. At the bene�t incidence of such subsidies. Tariffs least eight states in India have passed multi- have since increased by 75 percent, which has year tariff orders. Regulators have also made helped cut the gap between the cost of supply attempts to improve the consumer interface and noti�ed tariffs in half (from 35 percent to by enabling consumers to have input into reg- 17 percent). The Public Utilities Commission ulatory decision making. of Sri Lanka announced a new set of electric- Governance and the institutional capac- ity tariffs for 2011–15 in late 2010. The tar- ity of utilities need to be strengthened. iffs are expected to be gradually restructured Most countries have unbundled their power to reflect cost-effectiveness by 2015. sectors, corporatized previously vertically More remains to be done, particularly in integrated power utilities, or both. How- rationalizing industrial tariffs by reducing ever, many newly formed utilities are still not the cross-subsidy and improving targeting. empowered to act autonomously and suffer Doing so would help industries go back to from severe institutional capacity constraints. the grid (instead of depending on captive Their accountability, ef�ciency, and customer generators) and use open access options service are well below industry benchmarks. (IDFC 2010). The freedom to make important decisions— Regulators with adequate capacity can such as setting up performance-based remu- ensure transparency and accountability in neration systems and providing market-linked tariff setting, a process that continues to be remuneration to new employees—is limited. driven by factors such as political exigen- Robust and effective corporate governance is cies. In Pakistan, the regulator (NEPRA) has widely accepted as the essential foundation exclusive responsibility for determining tar- for performance improvement. Without the iffs after public hearings, but tariffs become support of a strong board and high-quality legally binding only once they have been noti- senior management, the complete transfor- �ed by the government in the of�cial gazette. mation in organizational culture needed to In India, there are instances of tariffs not commercialize in the face of a long tradi- being � led annually and subsidies not paid tion of risk-averse top-down bureaucratic to utilities in time. New initiatives have been control will not be achieved. Some states in implemented, however, particularly the mul- India (such as Andhra Pradesh and West Ben- tiyear tariff, which provides certainty on the gal) have adopted technology initiatives and costs for which utilities can be held account- accountability frameworks to improve sector able, and the stepping back from regulatory performance (box 4.4). BOX 4.4 Improving performance of state-owned power suppliers in Andhra Pradesh In 1999, the government of Andhra Pradesh under- lions of digital meters; they automated meter read- took a comprehensive reform program to control ing and reduced billing cycles. They also consulted theft of electricity from state-owned power compa- extensively with customers and communicated with nies and improve its revenue collection. The cam- the general public about their reforms—a major paign focused on four measures: enacting a new law change from previous practice. Under this cam- to address electricity theft, strengthening enforce- paign, the distribution companies regularized large ment mechanisms, reorganizing the anticorruption numbers of consumers and prosecuted extreme cases function in the utilities, and reengineering busi- of theft of electricity. As a result, they reduced losses nesses processes to improve management control from 38 percent in 1999 to 26 percent in 2003 and and customer service. increased the collection rate to 98 percent. The distribution companies installed modern bulk metering, data-logging equipment, and mil- Source: Bhatia and Gulati 2004. WHAT IS PRE VENTING FIRMS FROM CREATING MORE AND BE T TER JOBS? 149 provides examples of red tape reforms that Corruption policy options have had an impact on corruption. Two broad strategies are discussed for The second broad strategy involves legis- addressing corruption based on experiences lative and administrative interventions, such from countries in South Asia and other as establishing anticorruption laws and set- regions. The first strategy focuses on cut- ting up anticorruption institutions, such as ting red tape and simplifying administrative an independent anticorruption commission. procedures to reduce unnecessary interac- Direct interventions have had a mixed record, tion with public of�cials and reduce oppor- with success depending on the governance tunities for discretionary abuse by of�cials. structure of these institutions, their inde- In addition to reducing corruption, these pendence, and their resources and authority. policies should benefit firms by reducing Box 4.6 provides successful examples of direct the administrative burden on � rms. Box 4.5 legislative and administrative intervention. BOX 4.5 Cutting red tape to reduce corruption Investment climate assessments conducted by the ized. Manual transcriptions are more prone to errors World Bank identify several reforms that have and misplacement. reduced corruption by cutting red tape. Although recent reforms in some states have improved the registration process, it takes on aver- age 87 days to register a property in India. Because Bangladesh port and customs of the lengthy and complex land transfer process, The port of Chittagong, which handles 85 percent of informal payments are considerable. Bangladesh’s trade, was highly inef�cient, with, for Recent reforms include computerization of land example, ship turnaround times of �ve to six days records in Gujurat, Karnataka, Madhya Pradesh, compared with one day at more ef�cient ports. Main Maharashtra, and Tamil Nadu and computeriza- problems included long and complex procedures, tion of the registration of deeds in Andhra Pradesh, labor shortages, and lack of accountability of of� - Karnataka, Maharashtra, Rajasthan, and Tamil cers and employees of customs and port, leading to Nadu. rent-seeking behavior and irregularities. These reforms have led to substantial improve- Following the handover of the Chittagong Con- ments. In cities where reforms have taken place, it tainer Terminal to a private operator, there have takes a little more than one month to register land, been noticeable improvements in handling opera- rather than the more than �ve months in cities with tions. A one-stop service incorporating shipping no reforms. The reform has also had an impact agents, freight forwarders, customs, clearing and on corruption: computerization of land records in forwarding agents, and banks has ensured docu- Karnataka reduced informal payments by Rs 800 mil- ment clearance within 20 minutes. The turnaround lion and saved Rs 66 million in waiting time (Ferrari time of ships has decreased to three to four days. and Dhingra 2009). Port ef� ciency has increased 30 percent, and the cost of doing business in the port has decreased Pakistan tax reform 40 percent. The recent literature on tax reforms stresses the importance of tax administration and the effect it has India land registration on compliance costs. Recent reforms have focused Uncertainty surrounding property titles drives up on improving the management and organization of the cost of land acquisition by limiting the supply revenue administration; updating information tech- of reliable investments. In many states, land records nology and systems to efficiently process massive and registration of deeds have not been computer- information flows from taxpayers; strengthening (continues next page) 150 MORE AND BETTER JOBS IN SOUTH ASIA BOX 4.5 Cutting red tape to reduce corruption (continued) investigation, so that authorities can develop risk problems. However, this solution came at the cost of analysis to closely follow cases that violate tax laws; decreased tax revenue. and credibly conducting audits to improve enforce- A less costly, and arguably more effective, effort ment. Such reforms signal credibility in the tax was made by the Federal Bureau of Revenue to regime, a critical part of improving the system that improve the integrity and accountability of its inspec- can in turn increase voluntary compliance. tors. It implemented a task force’s recommendations Reforms that simplify tax procedures and reduce to allow only minimal contact between taxpayers interactions between tax of� cials and taxpayers, and tax of�cials; establish end-to-end automation of such as the reforms adopted in Chile, Colombia, tax � ling and refunds; improve recruitment policies, and Latvia (Gill 2003), also reduce opportunities to weed out potentially corrupt of�cials; simplify tax for corruption. Other anticorruption measures structures; and reduce the discretion of tax of�cials enabled by automation include restricting access of in determining tax liabilities. employees to scanned copies of original records These attempts to decrease corruption should to prevent tampering and creating audit trails of have an important side benefit in reducing tax administrative decisions and changes to taxpayer administration costs, which firms in Pakistan cite current accounts. as their second-highest constraint (after tax rates). In an attempt to decrease corruption by reduc- Cross-country data on time to prepare and pay taxes ing the interaction between tax officials and pri- show � rms in Pakistan spending 560 hours versus vate agents, the government of Pakistan raised the an average of 314 hours for South Asia (World Bank threshold for general sales tax registration from 2011e). Rs 0.5 million to Rs 5 million, freeing 42,000 pre- viously registered taxpayers from corruption-related Sources: Ferrari and Dhingra 2009; World Bank 2008c, 2009. BOX 4.6 Legislative and administrative interventions to reduce corruption in East Asia The annual Transparency International Corrup- related to corruption; and public information cam- tion Perceptions Index ranks countries in terms of paigns that promote whistleblowing and the � ght the degree to which corruption is perceived to exist against corruption. among public of� cials and politicians. a Singapore Several countries in East Asia have tried to follow has consistently ranked among the least corrupt Singapore’s recipe, with varying degrees of success. countries in the index, ranking third in 2010 (after The Republic of Korea introduced the Anticorruption Denmark and New Zealand). Act (2001), the Money Laundering Prevention Act Singapore has a long tradition of fighting cor- (2001), and the Code of Conduct for Maintaining ruption. It implemented a comprehensive anticor- the Integrity of Public Of�cials (2003). It also created ruption system aimed at reducing the incentives and the Korea Independent Commission against Corrup- opportunities for corruption. This system includes tion (later called the National Integrity Commission). legislation against corruption, called the Prevention Korea’s ranking in the Corruption Perceptions Index Corruption Act; an independent authority called the declined from 48 in 2001 to 39 in 2010. However, Corrupt Practices Investigation Bureau, with strong other East Asian countries that introduced similar powers to implement the law, including functions reforms saw their rankings deteriorate (Malaysia of investigation, prevention, and training in cases ranked 44th in 2006 and 56th in 2010). (continues next page) WHAT IS PRE VENTING FIRMS FROM CREATING MORE AND BE T TER JOBS? 151 BOX 4.6 (continued) Analysis of the factors that may have led to suc- any corruption within the agency must be subject to cess in Singapore suggests that there must be strong harsh penalties. Major causes of corruption (includ- political will to �ght corruption. This political will ing low civil service salaries and excessive red tape) is partially reflected in the issuance of legislation must be tackled and corrupt offenders punished and the creation of an independent implementa- (Quah 2007). tion authority, but the authority must also have full powers, there must be strong political support for a. The Transparency International Corruption Perceptions Index is a composite its actions at the highest level, and it must have the index drawing on corruption related data from expert and business surveys budget and personnel to carry out its functions. The carried out by a variety of independent institutions. The index ranks more than 175 countries in terms of perceived levels of corruption as determined by anticorruption agency must be independent from the expert assessment and opinion surveys (see http://www.transparency.org/ police and political control and incorruptible itself; policy_research/surveys_indices/cpi/2010/results). Access to �nance policy options and medium-size enterprises’ lack of con- ventional collateral Access to � nance is an issue for many micro, • Strengthening credit bureaus, in order to small, and medium enterprises in some South help overcome adverse selection and moral Asian countries. Policies to increase access hazard related to asymmetric information to � nance for such � rms address the various in credit markets demand- and supply-side constraints and • Supporting financial intermediation by the underlying institutional and regulatory facilitating downscaling of commercial environment that affects credit. Commercial banks (for instance, through simpli�cation banks face higher costs in lending to small, of documentation and collateral require- often geographically dispersed businesses, ments, reducing the number of days to because identifying potential borrowers, approve a loan, and so forth) and upscal- conducting due diligence, and maintaining ing micro� nance institutions contact after a loan is made are all more dif- • Providing technical assistance and � nan- �cult. For their part, small businesses often cial literacy training to micro, small, and do not know what is needed to qualify for medium-size enterprises. bank financing and lack market informa- tion. Governments should focus on devel- Annex 4E expands on these policy options. oping an enabling environment that makes small business lending safer, cheaper, and faster rather than on prescriptive measures such as directed lending programs, which The “how� of tackling constraints: have created distortions and inef� ciencies public-private collaboration in � nancial intermediation in many South This chapter so far has identified binding Asian countries. constraints to firms and suggests reforms Good practices from countries that have that may be needed to facilitate the expan- been able to expand access to finance for sion of employment. The next question is micro, small, and medium-size enterprises how to design and implement successful include the following: reforms to address these constraints. Reforms can be designed and imple- • Establishing efficient secured transac- mented through coordinated public and tions registries for available movable private sector efforts. International experi- assets, in order to address micro, small, ences of successful industrial development 152 MORE AND BETTER JOBS IN SOUTH ASIA indicate that such collaboration can play • Identify sector opportunities. The �rst step an important role in identifying binding is to use agreed upon selection criteria and constraints and designing, championing demand-side/market signals (such as factor and implementing reform/actions in a coor- endowments, which identify where a coun- dinated and integrated manner, leading try’s latent comparative advantage lies) to to increased competitiveness and growth. identify existing sectors that have the poten- Efforts often focus on a particular sec- tial to compete in regional and global mar- tor, because many binding constraints and kets, attract investment, and create jobs. market failures are sector speci� c and pri- • Conduct sector analyses. The second step is vate sector players � nd it easier to coalesce to conduct evidence-based technical analy- around a particular sector. Box 4.7 looks at ses of the opportunities and constraints of the role of public-private collaboration to the selected sector—by, for example, per- implement reforms in the export garment forming value-chain diagnostics to identify industry in Cambodia. constraints at each stage of the chain. The World Bank’s Financial and Private • Develop action and implementation plans. Development Network developed the Com- The last step is to create country-level task petitiveness Partnership Initiative (CPI), forces and joint public-private working which is being piloted in several countries, groups that develop action plans to remove including India, Jordan, and Kenya. CPI is constraints and lead their implementation. a collaborative process between the public Types of public, private, and public-private and private sectors around a sector or sectors collaborative actions that could come out with high growth potential. In Kenya, for of this process include both hard and soft instance, the selected sectors are tourism and infrastructure delivery. business process outsourcing. The CPI is developed around three broad This targeted sector approach has several steps: potential drawbacks. First, sectors could be BOX 4.7 Public-private collaboration to implement reforms in the Cambodian garment industry Cambodia’s garment industry— the country’s mechanisms in the world. By reducing business costs, top export earner and top employer—consists of the 11 reforms the forum has implemented have about 270 export-orientated factories. Total gar- saved the private sector an estimated $100 million. ment export revenue increased from $1.2 billion in A wide range of reforms has been implemented 2001 to $2.9 billion in 2010, representing 78 per- to improve the competitiveness of the garment and cent of exports and 15 percent of GDP. The indus- other export sectors. These reforms include the try employs 290,000 workers, 92 percent of them following: women, representing 70 percent of formal manufac- • Simplifying import/export procedures, such as turing employment. customs clearance at point of import In 1999, the Government-Private Sector Forum • Reducing export management fees (G-PSF) was established. Chaired by the prime min- • Reducing contributions to the National Social ister, the G-PSF comprises eight working groups, Security Fund each made up of government of� cials and private • Developing the Trade Union Law sector representatives. The export processing work- • Establishing joint trade missions between the pri- ing group is co-chaired by the Ministry of Com- vate sector and the government merce and the Garment Manufacturers Association. • Conducting government negotiations to obtain The International Finance Corporation (which trade preferences and free trade agreements. funded G-PSF from 2000 to 2010) ranked it as one of the top most effective public-private dialogue Source: Van 2011. WHAT IS PRE VENTING FIRMS FROM CREATING MORE AND BE T TER JOBS? 153 chosen arbitrarily or on the basis of noneco- sector members of the competitiveness com- nomic reasons rather than on latent com- mittee, created in 1998, were members of a parative advantage or potential economic single business lobbying association. In 2007, benefits. To prevent scarce resources from the World Bank and the International Finance being unproductively invested, it is impor- Corporation helped restructure a new collab- tant to agree on objective selection criteria, oration platform, which was more inclusive use demand-side/market signals to identify and included a process for selecting proposals sectors with comparative advantage, gauge that followed good practice. The focus of the trade potential and interest by private inves- proposals was also narrowed to enable better tors, and ensure that a wide set of private scrutiny of the details (large reform proposals sector representatives from different sectors tend to provide opportunities to hide details). is consulted during the process of identifying The results have been broader participation priority sectors. and wider acceptance of the reforms. Second, the approach could create an uneven playing � eld between � rms in dif- ferent sectors and between � rms within a Constraints facing potential sector if the reforms bene�t disproportion- �rm entrants: Business entry ately certain � rms. To reduce the risks of regulations being captured by vested interests of rent- One of the limitations of the analysis of seeking firms with privileged access to � rm constraints is that it does not necessar- political and governance structures, it is ily identify the constraints facing potential important to ensure broad representation entrants. Doing so is important because of private sector players—and not just the new � rms create jobs. The ease of business largest firms—in the public-private dia- entry regulations has been identi� ed as an logue process; use objective, evidence-based important determinant of new � rm entry, analyses to identify constraints and assess with studies showing that large reductions costs and bene�ts of proposed actions; and in business entry regulations—by, for exam- establish processes that elicit structured and ple, introducing a one-stop shop of business transparent dialogue, analysis, and decision registration centers or reducing registration making. fees—can spur the establishment of new Cameroon provides an example of a com- formal � rms by individuals who were pre- petitiveness partnership initially diverted viously wage workers or out of the labor into rent-seeking activity. Most of the private force. Box 4.8 summarizes the evidence. BOX 4.8 Effects of easing business entry regulations on �rm entry, employment, and formalization Large reductions in the ease of opening a business are try’s initial conditions matter: countries that start out positively and strongly associated with higher levels with high registration costs need larger reforms to of business registration and new � rm entry. A cross- induce a signi�cant number of new registrations. country study by Klapper and Love (2010) �nds that In Mexico, a reform that reduced the average large reforms—reforms that achieve more than a 40 number of days needed to start a business from percent reduction in costs, days, or procedures of busi- 30 to 1.4 led to a 5.0 percent increase in the number ness registration—lead to an increase in rates of new of new businesses and a 2.2 percent increase in wage firm entrance. Multiple and simultaneous reforms employment (Bruhn 2008). The increase in registra- have a larger impact on business registration. A coun- tion and employment came from new activities from (continues next page) 154 MORE AND BETTER JOBS IN SOUTH ASIA BOX 4.8 Effects of easing business entry regulations on �rm entry, employment, and formalization (continued) former wage earners. Former unregistered business inspections. In terms of bene�ts, most mentioned owners were not more likely to register their busi- being able to obtain a bank account, apply for bank ness after the reform. loans, operate on a larger and more visible scale, A field experiment in Sri Lanka by De Mel, and get contracts. McKenzie, and Woodruff (2011) aimed to estimate Does formalization of existing informal firms the demand for formality among informal � rms with result in increased productivity and employment? 1–14 workers. The experiment took 500 � rms and The Sri Lanka study reinterviewed the same � rms divided them into 5 groups, each with 100 � rms. The 12–18 months after formalizing, to see whether there � rst group was the control group. The second group was any evidence that formality had bene�tted them. was given information on how to register with the Over this time frame it found little evidence of any of district secretariat and obtain a business registration the main channels through which formality is argued certi�cate and offered reimbursement of the cost of to bene�t � rms: the newly formalized � rms had not doing so. The remaining groups were also given this received more credit or obtained more government information. In addition, they were offered payments contracts, they had not become more visible, and of Rs 10,000 (about $88, just under half a month’s they had not hired more workers. There was also no pro�ts for the median �rm); Rs 20,000; or Rs 40,000 signi�cant impact on sales or pro�ts (although the if they registered within a set timeframe. study could not rule out large negative or positive The study found that providing information impacts given how noisy the data were). These � nd- and reducing the cost of registering to zero was not ings suggest that informal enterprise owners acted enough to induce � rms to register. In contrast, the rationally by formalizing only when provided with Rs 10,000 and Rs 20,000 payments were enough to monetary incentives to do so. induce about one-fifth of eligible informal firms Simplification of entry regulations alone is to register, and the Rs 40,000 payment was enough unlikely to be enough to get informal � rms to for- to induce almost half of eligible informal � rms to malize, although the results show that it does not do so. The � nding that additional monetary incen- take too much of a change in net bene�ts of formal- tives were required supports the “exit� hypoth- izing to induce formalization (half the � rms were esis, which posits that the net bene� ts to formal- ready to formalize for an amount equal to 1.5–2.0 ization may be very low (if not negative) and that months’ pro�ts). Moreover, formalization is not suf- � rms are therefore rational in choosing to remain �cient for increasing productivity. Therefore, rather informal, rather than the “exclusion� hypothesis, than focusing on inducing existing informal � rms which posits that there are high net positive bene�ts to register their businesses, it may be more effec- to formalization and that it is the upfront costs or tive to focus on directly easing constraints to their complexity of registration that keeps � rms infor- productivity. Increasing their productivity is likely mal. Firm owners asked about the possible costs of to have a second-round effect on reducing informal- registering most often cited paying taxes (corporate ity, as increasing productivity can increase returns to and employee) and being subject to tax and labor formalization. The 2010 World Bank Doing Business India has the largest number of procedures indicator for starting a business indicates that (12) and costs in the region and is ranked the average number of procedures, time, and lowest (165th in the world) for this indica- cost of starting a business in South Asia are tor.18 Nepal ranks 96th, Pakistan 85th, Bhu- on a par with or lower than in Africa, East tan 84th, Bangladesh 79th, Maldives 50th, Asia, or Latin America and the Caribbean. Afghanistan 25th, and Sri Lanka 24th. WHAT IS PRE VENTING FIRMS FROM CREATING MORE AND BE T TER JOBS? 155 Annex 4A Business environment constraints in high- and low-income states in India The division into high-income and low- The low income group comprises Arunachal income states in India is based on their net Pradesh, Assam, Bihar, Chattisgarh, Jammu state domestic product per capita over the and Kashmir, Jharkand, Madhya Pradesh, period 1999/2000 to 2009/10. The high Manipur, Meghalaya, Mizoram, Nagaland, income group comprises Andhra Pradesh, Orissa, Rajasthan, Sikkim, Tripura, Uttar Goa, Gujarat, Haryana, Himachal Pradesh, Pradesh, Uttarakhand, and West Bengal. Karnataka, Kerala, Maharashtra, Pun- jab, Tamil Nadu, Andaman and Nicobar Islands, Chandigarh, Delhi, and Puducherry. FIGURE 4A.1 Severity of constraints reported by benchmark �rm in urban formal sector in high- and low-income states in India courts electricity 3.0 telecoms TABLE 4A.1 Top five constraints reported by 2.5 benchmark �rm in the urban formal sector in high- corruption 2.0 macroinstability and low-income states in India 1.5 tax administration 1.0 business licensing Low-income High-income 0.5 Constraint states states 0.0 Electricity 1 3 transport government policy uncertainty Corruption 2 1 Tax administration 3 2 crime, theft, and disorder competition Transport 4 labor regulations customs Crime, theft, and disorder 5 inadequately access to land educated labor Labor regulations 4 India low-income states benchmark Government policy India high-income states benchmark uncertainty 5 Source: Authors, based on Carlin and Schaffer 2011b (based on World Bank Source: Authors, based on Carlin and Schaffer 2011b (based on World Bank enterprise surveys). enterprise surveys). Note: A benchmark �rm is a medium-size manufacturing �rm with 30 employees that is domesti- Note: A benchmark �rm is a medium-size manufacturing �rm with 30 cally owned, does not export or import, is located in a large city, and did not expand employment employees that is domestically owned, does not export or import, is in the preceding three years. Analysis is based on a pooled sample of enterprise surveys conducted located in a large city, and did not expand employment in the preceding in India between 2000 and 2010. A point farther away from the origin indicates that the business three years. Analysis is based on a pooled sample of enterprise surveys constraint is considered more severe. The severity of constraint is rated by �rms on a 5-point conducted in India between 2000 and 2010. Access to �nance and tax scale, with 0 being no obstacle, 1 being a minor obstacle, 2 being a moderate obstacle, 3 being a rates constraints are excluded. Political instability was not asked in the major obstacle, and 4 being a very severe obstacle. Access to �nance and tax rates constraints are India surveys. excluded. Political instability was not asked in the India surveys. Annex 4B Tax rates as a constraint to �rms The reported severity of the tax rate con- other elements of the business environment straint cannot be interpreted directly as evi- is problematic, cross-country comparisons dence that tax rates are binding constraints can be used to show whether the severity in South Asia. This annex looks at alterna- reported in South Asia is atypical. Figure tive indicators. 4B.1 shows the reported severity of the tax rate constraint for a benchmark � rm in dif- ferent countries. It shows that independent Cross-country comparison of reported of the level of income of the country, � rms severity complain signi�cantly about tax rates. Only Although comparing the level of severity Pakistan reports higher severity than is typi- reported for the tax rate constraint with cal of countries at its level of development; 156 MORE AND BETTER JOBS IN SOUTH ASIA FIGURE 4B.1 Cross-country comparison of reported severity of tax rate constraint 4 3 severity of constraint Pakistan Sri Lanka 2 Bangladesh Afghanistan India Bhutan Maldives 1 Nepal 0 6 7 8 9 10 11 log of per capita GDP at purchasing power parity dollars Source: Carlin and Schaffer 2011b (based on World Bank enterprise surveys). Note: The cross-country regression line shows the relationship between the reported severity of the constraint for a benchmark �rm and the log of per capita GDP. The shaded area is the 95 percent con�dence interval band around the regression line. Vertical bars show con�dence intervals of 95 percent around the reported severity of the constraint for countries in South Asia. The lack of overlap between the South Asian country con�dence interval and the regression line con�dence interval is a conservative test of the statistically signi�cant difference between the reported severity of a constraint for the South Asian country and the average reported severity of constraint for countries at the same level of per capita GDP. The reported severity could still be signi�- cantly different even when there is an overlap. Analysis is based on a pooled sample of enterprise surveys conducted between 2000 and 2010. The severity of constraint is rated by �rms on a 5-point scale, with 0 being no obstacle, 1 being a minor obstacle, 2 being a moderate obstacle, 3 being a major obstacle, and 4 being a very severe obstacle. elsewhere in the region, � rms consider tax only 0.85 million out of a registered tax base rates less of an obstacle than do firms in of 1.25 million are actually taxpayers. other countries at the same income levels. Statutory and effective tax rates Total tax revenues to GDP ratio Although tax rates in India, Pakistan, and The ratio of total tax revenues to GDP is Sri Lanka are lower than they were in the low in many countries in the region—just 9 1990s, these countries still have the highest percent in Bangladesh, for example, and 10 marginal corporate tax rates in the region. percent in Pakistan (�gure 4B.2). These ratios Rates are higher than comparator countries do not necessarily reflect low taxes, how- such as Brazil, China, Ghana, Indonesia, the ever. Rather, they represent a lower level of Philippines, and Thailand (�gure 4B.3). development and very narrow tax bases. In Because of poor compliance, however, and Pakistan, the tax base is very small, and �rms system of exemptions and loopholes, there that are in the tax base bear a heavy tax bur- is a signi�cant divergence between statutory den. The same is true for Bangladesh, where and actual average effective tax rates in South only 1.4 million of the country’s 145 million Asia. There is also a wide variance in effective people are registered in the tax base. In 2004, tax rates across activities and sectors, which WHAT IS PRE VENTING FIRMS FROM CREATING MORE AND BE T TER JOBS? 157 FIGURE 4B.2 Tax revenue as a percentage of GDP in South Asia 25 21 20 20 16 15 percent 13 12 12 10 10 10 9 9 6 5 0 an an h an l a ka s sia CD lA d pa ive di es ra an an ist ut ist hA OE In Ne sia lad ald nt e Bh an iL k Ce rop Pa ut ng M Sr gh So Eu Ba Af Source: World Bank 2011e. Note: OECD = Organisation for Economic Co-operation and Development. FIGURE 4B.3 Highest marginal corporate tax rate in South Asian countries and selected comparator countries 40 34 34 35 35 35 30 30 30 27.5 27 25 25 25 percent 20 20 15 10 5 0 an a a h ia s nd a il an ka ne az an in di es s an ist ist ila ne Ch In Br ad pi Gh an iL k a do illi Pa l Th ng Sr gh Ph In Ba Af Source: World Bank 2011e. creates distortions, because decisions about Taxation in Bangladesh is characterized by resource allocation, location, and invest- low collections, highly complex rules, a heavy ment are driven by tax considerations rather administrative burden, and poor compliance than underlying economic reasons. The cases (World Bank 2008c). It has become the norm of Bangladesh and Pakistan are discussed in Bangladesh to pay minimum taxes (by, for below. example, underreporting pro�ts) or to avoid 158 MORE AND BETTER JOBS IN SOUTH ASIA them altogether, because of both lax enforce- income tax. Among listed companies, 1.2 per- ment of laws (tax defaulters are rarely inves- cent paid 53.3 percent of taxes, and the top tigated) and the hassle and cost involved in 40 percent paid almost 100 percent. The top dealing with tax authorities. Only 66 percent 10 percent of corporate taxpayers contrib- of metropolitan � rms were required to pay uted 80 percent of total corporate income tax corporate tax, and only 3 percent reported collection; the top 5 percent contributed 64 actually doing so. Furthermore, the variance percent. in actual tax payments shows signi�cant use In Pakistan, average and marginal corpo- of tax holidays and exemptions. The average rate tax rates vary widely by sector, with the �rm paid taxes of about 16 percent of pro�ts; bulk of the burden falling on large, formal the median �rm paid only 7 percent. Among manufacturing �rms. Although formal man- private unlisted �rms, 25 percent paid 84 per- ufacturing � rms account for just 23 percent cent of the taxes collected (the bulk paid by of output, they account for half of all federal foreign companies), and 13 percent paid no taxes (World Bank 2009). Annex 4C Constraints facing nonbenchmark �rms FIGURE 4C.1 Severity of constraints reported by benchmark �rm and �rm with 60 employees in the urban formal sector in Nepal and Sri Lanka a. Nepal b. Sri Lanka courts access to land 3.0 3.0 political instability business licensing electricity customs 2.5 2.5 2.0 government policy 2.0 business licensing electricity access to land uncertainty 1.5 1.5 1.0 1.0 transport 0.5 telecoms macro instability transport 0.5 0.0 0.0 crime, theft, and competition telecoms corruption disorder inadequately labor regulations labor regulations corruption educated labor competition tax administration crime, theft, and disorder tax administration inadequately customs educated labor benchmark (�rms with 30 employees) firms with 60 employees Source: Authors, based on Carlin and Schaffer 2011b (based on World Bank enterprise surveys). Note: A benchmark �rm is a medium-size manufacturing �rm with 30 employees that is domestically owned, does not export or import, is located in a large city, and did not expand employment in the preceding three years. The �rm with 60 employees is the same as the benchmark �rm except for the �rm size. Analysis is based on a pooled sample of enterprise surveys conducted in Nepal and Sri Lanka between 2000 and 2010. A point farther away from the origin indicates that the business constraint is considered more severe. The severity of constraint is rated by �rms on a 5-point scale, with 0 being no obstacle, 1 being a minor obstacle, 2 being a moderate obstacle, 3 being a major obstacle, and 4 being a very severe obstacle. Access to �nance and tax rates constraints are excluded. WHAT IS PRE VENTING FIRMS FROM CREATING MORE AND BE T TER JOBS? 159 FIGURE 4C.2 Severity of constraints reported by benchmark (nonexporting) and exporting �rm in the urban formal sector in South Asia and Bangladesh a. South Asia b. Bangladesh telecoms labor regulation political instability 3.0 electricity 3.0 courts courts 2.5 2.5 electricity 2.0 business licensing political instability transport 2.0 1.5 1.0 1.5 corruption transport corruption business licensing 0.5 1.0 0.0 0.5 tax administration crime, theft, and labor regulations access to land 0.0 disorder government policy tax administration inadequately inadequately uncertainty educated labor educated labor macroinstability government policy customs telecoms uncertainty competition access to land customs macroinstability crime, theft, and disorder competition benchmark (nonexporting) exporters Source: Authors, based on Carlin and Schaffer 2011b (based on World Bank enterprise surveys). Note: A benchmark �rm is a medium-size manufacturing �rm with 30 employees that is domestically owned, does not export or import, is located in a large city, and did not expand employment in the preceding three years. The exporter �rm is the same as the benchmark �rm except that it exports. Analysis is based on pooled sample of enterprise surveys con- ducted in South Asia and Bangladesh between 2000 and 2010. A point farther away from the origin indicates that the business constraint is considered more severe. The severity of constraint is rated by �rms on a 5-point scale, with 0 being no obstacle, 1 being a minor obstacle, 2 being a moderate obstacle, 3 being a major obstacle, and 4 being a very severe obstacle. Access to �nance and tax rates constraints are excluded. Annex 4D Access to �nance as a constraint to �rms The reported severity of the access to �nance percentage of GDP, and they are generally constraint cannot be interpreted as evidence experiencing rapid growth. Indeed, � nancial that access to � nance is a binding constraint depth has been increasing in all countries in in South Asia. This annex looks at alterna- South Asia except Sri Lanka. tive indicators. In 2008, domestic credit to the private sector as a share of GDP was 29 percent in Cross-country comparison Sri Lanka and Pakistan, 30 percent in Bhu- of reported severity tan, 39 percent in Bangladesh, 51 percent in India, and 56 percent in Nepal (Hoxha 2011). Figure 4D.1 shows the reported severity of Compared with countries with similar levels the access to finance constraint for bench- of per capita GDP (which is positively corre- mark � rms in different countries. It shows lated correlation with � nancial depth), only that in no country in South Asia do �rms cite Bhutan is signi�cantly below average. it as a greater obstacle than in �rms in other countries with the same level of per capita Cost of �nance GDP. Using lending interest rates to assess the cost Financial depth of � nance is dif�cult, because there is a high level of state intermediation in many South Financial depth measures the availability Asian countries: interest rates in Bangladesh, of � nancial resources for the economy and India, and Pakistan are not freely set by the for firms. The financial markets in South market. Nevertheless, lending rates in South Asia are not shallow by measures such as Asia are not high compared with other coun- domestic credit to the private sector as a tries at the same level of per capita GDP. 160 MORE AND BETTER JOBS IN SOUTH ASIA FIGURE 4D.1 Cross-country comparison of reported severity of access to �nance constraint 4 Access to Finance 3 severity of constraint Maldives Bangladesh 2 Afghanistan Pakistan Bhutan Sri Lanka Nepal India 1 0 6 7 8 9 10 11 log of per capita GDP at purchasing power parity dollars Source: Carlin and Schaffer 2011b (based on World Bank enterprise surveys). Note: The cross-country regression line shows the relationship between the reported severity of the constraint for a benchmark �rm and the log of per capita GDP. The shaded area is the 95 percent con�dence interval band around the regression line. Vertical bars show con�dence intervals of 95 percent around the reported severity of the constraint for countries in South Asia. The lack of overlap between the South Asian country con�dence interval and the regression line con�dence interval is a conservative test of the statistically signi�cant difference between the reported severity of a constraint for the South Asian country and the average reported severity of constraint for countries at the same level of per capita GDP. The reported severity could still be signi�- cantly different even when there is an overlap. Analysis is based on pooled sample of enterprise surveys conducted between 2000 and 2010. The severity of constraint is rated by �rms on a 5-point scale, with 0 being no obstacle, 1 being a minor obstacle, 2 being a moderate obstacle, 3 being a major obstacle, and 4 being a very severe obstacle. 5 percent of � rms in Pakistan use banks to Financial outreach and access finance working capital compared with 36 Outreach is a measure of financial inclu- percent in India, 43 percent in Bangladesh, siveness; it is captured here by variables and 52 percent in Sri Lanka. Small � rms in such as the percent of � rms with access to a Afghanistan and Pakistan finance a larger loan or other � nancial products. Enterprise share of working capital (88–91 percent) surveys show large variation in access to from internal financial sources than large external finance for firms both across �rms (63–88 percent). and within countries, with smaller firms Moreover, comparable survey data in having less access, leading to claims of a Afghanistan, Bangladesh, and Pakistan show small and medium-size enterprise � nance no clear trend in terms of the share of invest- gap, particularly in emerging economies, ment financed by internal funds (retained where larger enterprises are frequently the earnings); external � nance (banks and non- main beneficiaries of tax incentives and bank financial institutions); and informal corporate subsidies and aid is targeted at finance (family, friends, moneylenders) in microenterprises.19 the past �ve years for a benchmark � rm in Small firms in the urban formal sector the urban formal sector. Internal funds con- in Afghanistan, Pakistan, and, to a lesser tinue to � nance most investment. extent, India use external � nance less than Use of external �nance is even lower among small � rms in other countries (�gure 4D.2). rural and informal �rms, which are typically Only 2 percent of � rms in Afghanistan and microenterprises, and often reflects lower FIGURE 4D.2 Percentage of �rms with credit line or loan from �nancial institution, by �rm size and region 90 80 78 70 67 68 63 64 60 58 57 55 55 51 50 50 50 50 48 48 percent 45 44 43 40 40 36 38 34 35 32 30 26 20 16 15 11 10 6 2 0 20 tan lA d be d ci� d 20 pal 20 nka 20 tan a 20 tan 20 dia ra an ib an Pa an ric 10 is sia an c Ne 09 10 10 09 kis u In a 10 Af nt pe ar ca an ia iL Bh As Pa e C eri Ce uro gh Sr th Am st Af E Ea tin La small (5–19 employees) medium (20–99 employees) large (100 or more employees) Source: Authors, based on World Bank enterprise surveys. 161 162 MORE AND BETTER JOBS IN SOUTH ASIA access to banking services. In Bangladesh, value of the collateral required was 132 per- only 9 percent of � rms in nonmetropolitan cent of the loan amount. The most common areas have bank accounts, compared with forms of collateral were property/buildings 95 percent in metropolitan areas, and only and land. In summary, there is evidence of 6 percent of nonmetropolitan enterprises use unmet demand for the rural nonfarm sector formal �nance. in Pakistan. In the informal sector in India, 89 percent of enterprises had bank accounts, but less Institutional indicators than 20 percent had a loan (from informal or formal sources). Only 6–9 percent of infor- Creditors care about the broader institu- mal �rms had a loan from a �nancial institu- tional environment (Djankov, McLiesh, and tion (versus 26 percent of small � rms in the Shleifer 2007). A World Bank (2008b) study, formal sector). based on an April 2007 survey of commer- In Pakistan, very few rural firms deal cial banks, addresses the issue of access to with the formal banking sector: only 10–12 credit from the perspective of supply. It percent of enterprises keep savings, and less reviews the factors that limit the willingness than 10 percent maintain checking accounts. of � nancial institutions to lend or increase About 43 percent enterprises in villages and their incentives to charge higher interest 58 percent in small towns made investments rates. Key concerns were problems with col- the year preceding the survey, but nearly lateral registration and enforcement; lim- 90 percent were made with savings. Among ited information (the result of the lack of a surveyed business owners in rural areas, well-functioning credit registry); and poor 24–28 percent wanted to apply for a formal accounting and auditing practices, which loan, but only 4–7 percent actually did so. limit veri�able � nancial information. Some Among � rms in Pakistan that wanted a loan countries in South Asia have weak legal and but did not apply for one, 40 percent cited property rights, but the region as a whole loan procedures that were too complicated, performs better in this regard than Africa 27 percent cited interest rates that were too and the Middle East and North Africa and high, 16 percent cited insuf�cient collateral, is on par with Latin America and the Carib- and 8 percent cited the shortness of the loan bean and East Asia and Pacific. In terms duration. Two-thirds of loan applications of credit, public and private credit bureau were approved. Among rejected applica- coverage in South Asia is lower than in any tions, 36 percent were refused because of other region, although the depth of credit insuf�cient collateral. Eighty percent of for- information is better than in Africa and East mal loans require collateral, and the average Asia and Paci�c. Annex 4E Policy options for increasing access to �nance This annex describes four policy areas that that banks are generally not willing to lend can increase access to � nance, particularly without secure collateral. Micro, small, and for micro, small, and medium-size enter- medium-size enterprises find it harder to prises (Hoxha 2011). meet collateral requirements: in Bangladesh 30.8 percent of small firms (compared to 4.2 percent of large � rms) found collateral Operationalize secure transactions requirements too high (World Bank 2008c). registeries Banks usually require immovable property, Lack of credit information on custom- such as land, buildings, or an owner’s assets, ers and weak contract enforcement means to secure a loan. Most micro, small, and WHAT IS PRE VENTING FIRMS FROM CREATING MORE AND BE T TER JOBS? 163 medium-size enterprise � rms’ property and the existence of private credit bureaus is assets are vested in movable property, such associated with lower � nancing constraints as machinery and inventory. and a larger share of bank financing. In For example, in India an immovable asset addition, small and medium-size enterprises of the enterprise (60 percent of loans) or a tend to have a larger share of bank � nancing personal asset of the owner (25 percent of in countries where private registries exist. loans) is usually required as collateral. Such Currently, credit bureau coverage in South assets account for a small percentage of the Asia is the lowest of any region. It is par- total asset value of unorganized �rms. More- ticularly low in Afghanistan, Bangladesh, over, in about half of cases, the collateral Bhutan, Nepal, and Pakistan. required was more than twice the amount of the loan (Ferrari and Dhingra 2009). In Sri Downscale commercial banks and Lanka, collateral in the form of land is espe- upscale micro�nance institutions cially important for rural enterprises, but high levels of public landownership, unclear Access depends not only on the terms and ownership records, and widespread restric- cost of � nance but also on other barriers, tions on the use and transfer of land make such as eligibility criteria; procedural exclu- it difficult to use land as collateral (World sions (high minimum loan size requirements, Bank 2005). documentation requirements, days to pro- In the absence of a secured transactions cess loans); and lack of appropriate � nancial mechanism, lenders find it risky to accept products for micro, small, and medium-size movable property as collateral. A secured enterprises. transactions mechanism facilitates lending by A crucial function of �nancial intermedi- eliminating information asymmetries about aries is to screen borrowers. The number of collateral and reducing the risk of lending days it takes to process a loan application secured by movables. can be a de facto eligibility barrier, because Successful models of secured transac- some borrowers may be discouraged from tions registry include the creation of a reg- applying for bank loans and seek financ- istration database in which a public record ing elsewhere to avoid long waiting periods. of obligations secured by movable property Beck, Demirgüç-Kunt, and Peria (2007) �nd can be made; the setting of priorities in case that on average loan applications by small of multiple obligations of the same security and medium-size enterprise take more than a giver as secured by the same collateral; and month to process in Bangladesh (43 days) and the establishment of a means to uniformly Pakistan (34 days). In contrast, the median enforce secure interests in case the security number of days for the 60 countries in their giver defaults. sample was nine days. Lack of documentation can create other eligibility barriers. Bangladesh and Nepal Deepen the reach of credit bureaus require at least four documents, including Credit bureaus promote deeper financial an identity card or passport, a recommenda- systems by helping overcome adverse selec- tion letter, a wage slip, and proof of domicile. tion and moral hazard related to asymmet- For the 60 countries Beck, Demirgüç-Kunt, ric information in credit markets (Djankov, and Peria examine, the average number of McLiesh, and Shleifer 2007). As a result of documents was two. Given the high degree of the cost savings achieved through more ef�- informality in South Asia, only a small pro- cient and accurate credit analysis and lower portion of potential borrowers can produce expected losses, lenders can increase their these documents. credit extension. Using cross-country � rm- Standard financial products offered by level survey data and survey data of credit banks, such as lines of credit, are designed bureaus, Love and Mylenko (2003) � nd that for large corporations (the traditional client 164 MORE AND BETTER JOBS IN SOUTH ASIA for South Asian banks). They are inappropri- to reduce transaction costs and generate deal ate for many small businesses, which do not volume. Some banks have been successful in deposit their revenues in banks. downscaling to the small and medium-size To serve small businesses well and prof- enterprise market, achieving high levels of itably, banks need to simplify procedures outreach, scale, and sustainability (box 4E.1). for small business loans; design appropriate Microfinance institutions need to adapt �nancial products for micro, small, and medi- their business model and product offerings um-size enterprises; reduce reliance on col- to the small and medium-size enterprise mar- lateral and modernize lending based on risk ket. One institution that has done so is BRAC and credit information; and use technology Bank (box 4E.2). BOX 4E.1 Successful approaches to small and medium-size business banking ICICI Bank is the largest private sector bank in India. In 2002, Standard Chartered Bank implemented Its small and medium-size enterprise banking model a client-centric model on a global scale, creating a is based on segmentation of the market by industry separate small and medium-size enterprise bank- and business linkages; a proprietary risk evaluation ing operation, with its own resources and credit covering � nancial and non� nancial parameters to processes, within its consumer bank unit. Today, compensate for enterprises’ lack of �nancial informa- this business unit operates in more than 30 coun- tion; and a “beyond lending approach,� which relies tries. The bank follows three customer acquisition primarily on deposit products and other banking and servicing approaches: relationship managers services (used by 95 percent of small and medium- handle larger medium-size enterprises; a portfolio size enterprise clients) rather than lending products team made up of salespeople and virtual relation- (used by just 5 percent of such clients). Client ser- ship managers handle midsize small and medi- vicing is done though multiple channels, including um-size enterprises; and branches handle smaller relationship managers, doorstep banking, branches, businesses, whose needs are similar to those of the Internet, and automatic teller machines. individuals. BOX 4E.2 Scaling up micro�nance institutions: The case of BRAC Bank BRAC Bank Ltd. was established in Bangladesh in sizes relationship banking and collateral-free lend- 2001, with a focus on the small business sector. Its ing of up to $14,000. small and medium-size enterprise banking unit goes Upscaling relies heavily on support from inter- beyond traditional banking, working with entrepre- national and domestic � nancial institutions in the neurs as a business partner by building awareness, start-up phase. Such support is particularly impor- providing training, and arranging road shows to tant in adapting microfinance institutions’ lend- support and develop clients’ businesses. BRAC Bank ing technologies to serve the new clientele and in is the number one small business bank in Bangla- building the capacity of staff. BRAC Bank’s sec- desh in terms of loans outstanding, with nationwide ond- and third-largest shareholders are Shorecap coverage (44 percent of coverage is in rural areas). International Limited and the International Finance The bank’s small business banking model empha- Corporation. WHAT IS PRE VENTING FIRMS FROM CREATING MORE AND BE T TER JOBS? 165 Commercial downscaling and microfi- Notes nance institution upscaling should be led by 1. The �rst study sets out the methodological banks themselves. Governments can facilitate framework and tests predictions of how a the process by creating an enabling regula- �rm’s evaluation of the cost of constraints tory and institutional environment—by, for imposed by its external environment varies example, revising rules on provisioning and with the �rm’s characteristics. These tests credit bureau reporting, which make it expen- con�rm the predictions of the model and sive to serve smaller �rms. Governments can provide the empirical basis for using, in the also look into using credit guarantee schemes second and third studies, the reported sever- and other �nancial products aimed at reduc- ity of constraints to assess which elements of ing the risk associated with lending to small the external environment are most problem- and medium-size enterprises. Such schemes atic for �rms across South Asia and in each would address information asymmetries, the country. 2. Eighteen constraints appear in the enter- high costs of processing small credit transac- prise surveys: electricity; telecommunica- tions, and constraints in the enforcement of tions; transport; access to land; inadequately contracts. educated labor; macroeconomic instability; government policy uncertainty; political instability; access to �nance; competition; Provide technical assistance and tax administration; tax rates; labor regula- �nancial literacy training tions; customs; business licensing; courts; Small and medium-size enterprises face sev- corruption; and crime, theft, and disorder. eral non� nancial barriers that are related to Surveys in Afghanistan, Bhutan, and Nepal did not ask about macroinstability or gov- their own capacity. Managers or owners of ernment policy uncertainty. Surveys in India, such � rms usually perform a wider range of Maldives, and Sri Lanka did not ask about tasks than their managers or owners of larger political instability. firms. Moreover, small and medium-size 3. The World Bank enterprise surveys portal enterprises cannot take advantage of econo- contains data on almost 120,000 urban for- mies of scale related to accounting, business mal �rms (16,000 in South Asia) from more planning, and market research to the same than 230 surveys in 126 countries conducted extent as large � rms can. Enterprise survey between 2000 and 2010 (see http://www data indicate that only 37 percent of small .enterprisesurveys.org/). � rms have their annual � nancial statement 4. The Bangladesh rural survey covered peri- reviewed by an external auditor, compared urban areas, small towns, and rural villages. The Pakistan rural survey covered small with 58 percent of medium-size � rms and towns and rural villages. The Sri Lanka rural 79 percent of large � rms (Hoxha 2011). In survey covered rural areas as de�ned by the Bangladesh, a quarter of smaller urban � rms Department of Census and Statistics. � nd the procedures for applying for a loan 5. Only the 2001 survey was used in the formal too complex (compared with only 1.2 percent analysis of the severity of constraints, because of large urban � rms) (World Bank 2008c). the constraint question in 2005 did not use Only about 5 percent of village enterprises the standard 0–4 scale. The ranking of con- and 11 percent of small town enterprises in straints from the 2005 informal survey was Pakistan prepare � nancial statements. used to check the results. Technical assistance and � nancial liter- 6. In addition, to facilitate better comparison, acy training can be provided to small and �rms with more than 20 employees were dropped from the sample of urban formal medium-size enterprises to increase their �rms. capacity to keep better accounts and develop 7. The states were grouped based on net state and present bankable proposals. Doing so domestic product per capita over the previous would ensure that such enterprises have the 10 years. capacity to successfully apply for loans and 8. Empirical evidence on other developing effectively use loans for the growth of their regions indicates that because they lack businesses. the economies of scale to operate a power 166 MORE AND BETTER JOBS IN SOUTH ASIA generator ef�ciently, small and medium 15. About 305 urban manufacturing �rms had enterprises, which create a large share of new subcontracting arrangements for input pur- jobs, are the most affected in areas without chases, but only 1 percent purchased inputs access to electricity or an unreliable pub- from �rms/distributors located in villages. lic grid (Aterido, Hallward-Driemeier, and 16. Other controls in the ordinary least squares Pagés 2007). regression include the number of urban cen- 9. Results reported are from partial correla- ters in the district, the number of urban cen- tions with other indicators of �rm produc- ters in a region, distance to district headquar- tivity in the pooled sample of South Asian ters, the average height and slope of land, and �rms with country �xed effects and controls the population density in the village. for �rm size, sector, location, exporting and 17. The second Bangladesh investment climate importing status, and ownership (Carlin and assessment (World Bank 2008c) concludes Schaffer 2011c). that overall lending to the private sector had 10. Expanding �rms in Afghanistan and improved and was adequate in Bangladesh Bangladesh �rms report lower costs than the but that lending to small �rms and nonurban benchmark �rm. A hypothesis on why this �rms remained inadequate because of lack of may be the case in Bangladesh is that expand- appropriate lending techniques and products ing �rms are located in export processing for micro, small, and medium-size enterprises. zones, which face especial conditions that 18. A procedure is de�ned as any interaction lower the constraints. between the company founders and exter- 11. This �nding is consistent with a 2007 nal parties (for example, government agen- Organisation for Economic Co-operation cies, lawyers, auditors, or notaries). Cost is and Development (OECD) economic survey measured as a percentage of the economy’s of India that describes the service sector as income per capita. It is measured in two ways: being less dependent on electricity and less (a) all of�cial fees and fees for legal or profes- subject to labor regulations; the sector also sional services if such services are required faces lower indirect taxation and has access by law and (b) the amount the entrepreneur to better infrastructure and a more condu- needs to deposit in a bank or with a notary cive business environment. The OECD sur- before registration and up to three months vey makes the case that growth in the service following incorporation, as a percentage of sector has been more dynamic because these the economy’s income per capita. constraints are less binding. 19. A 2008 World Bank enterprise survey of more 12. The regression model assumes a linear rela- than 50,000 �rms in more than 70 countries tionship between �rm characteristics and indicates that small and medium-size �rms reported severity, so that if, on average, larger rank access to and cost of �nance as one of �rms (more than 30 employees) report more their top constraints to growth. severe constraints than the benchmark �rm (30 employees), on average, smaller �rms (fewer than 30 employees) will report less References severe constraints. 13. Using the rural enterprise and household sur- Alby, P., J. J. Dethier, and S. Straub. 2010. veys, Deininger, Jin, and Sur (2007) �nd that “Firms Operating under Infrastructure and the top �ve reported constraints among rural Credit Constraints in Developing Countries. �rms in Sri Lanka were also the most signi�- The Case of Power Generators.� Policy cant determinants of new nonfarm enterprise Research Working Paper 5497, World Bank, start-ups. In areas where electricity is a major Washington, DC. constraint, the propensity to start up a �rm A DB (A si a n D evelopment B a n k). 2 0 09. is 14 percent lower. Distance to banks, poor Enterprises in Asia: Fostering Dynamism in road access, credit constraints, and lack of SMEs. Manila. market demand are all negatively correlated ———. 2010. Bhutan Energy Sector: Sector with the probability of starting up a business. Assistance Program Evaluation. Manila. 14. Enterprises that sell to larger �rms are less Aterido, R., M. Hallward-Driemeier, and C. Pagés. likely than enterprises that sell directly to 2007. “Business Climate and Employment local consumers to complain about low Growth: The Impact of Access to Finance, demand (World Bank 2008c). Corruption and Regulation Across firms.� WHAT IS PRE VENTING FIRMS FROM CREATING MORE AND BE T TER JOBS? 167 IZA Discussion Paper, Institute for the Study Do They Vary Across Countries?� Background of Labor, Bonn. study conducted for this book. Banerjee, A., and E. Duflo. 2008. “Do Firms Want ———. 2011c. “How Do Business Environment to Borrow More? Testing Credit Constraints Constraints Vary with Firm Efficiency and Using a Directed Lending Program.� CEPR Dynamism?� Background study conducted for Discussion Paper 4681, Center for Economic this book. and Policy Research, Washington, DC. Central Electricity Regulatory Commission. 2011. Banerjee, S. G., A. Singh, and H. Samad. 2011. Report on the State of Short-Term Power Power and People: The Bene�ts of Renewable Market in India. New Delhi. Energy in Nepal. World Bank, South Asia Ceylon Electricity Board. 2009. “Statistical Digest Sustainable Development, Washington, DC. 2009.� Colombo, Sri Lanka. Banerji, D., and M. Mishra. 2010. “Power Sector CR ISIL Risk and Infrastructure Solutions to Lead Infra Investments in 12th Plan.� Limited. 2011. The Study on Skill Gaps and Business Standard, December 29, New Delhi. Employment Opportunities in Power Sector http://www.business-standard.com /india/ in India. Report submitted to World Bank, news/power-sector-to-lead-infra-investments- Washington, DC. in-12th-plan/419972/. Deininger, K., S. Jin, and M. Sur. 2007. “Sri Beck, T., A. Demirgüç-Kunt, and V. Maksimovic. Lanka’s Rural Non-Farm Economy: Removing 2005. “Financial and Legal Constraints to Firm Constraints to Pro-Poor Growth.� World Growth: Does Firm Size Matter?� Journal of Development 35 (12): 2056–78. Finance 60 (1): 137–77. De Mel, S., D. McKenzie, and C. Woodruff. Beck, T., A. Demirgüç-Kunt, and M. Peria. 2007. 2011. “What Is the Cost of Informality? “Banking Banking Services for Everyone? Experimentally Estimating the Demand for Barriers to Bank Access and Use around the Formalization.� Unpublished paper. World.� Policy Research Working Paper 4079, Devaiah, D. 2010. Government of Gujarat: World Bank, Washington, DC. Jyotigram Yojana. http://www.scribd.com/ Besant-Jones, J. 20 06. “Reforming Power doc /33811921 /G over n ment- of- Guja rat- Markets in Developing Countries. What Jyotigram-Yojana. Have We Learned?� Energy and Mining Dininio, P., S. J. Kpundeh, and R. Leiken. Sector Board Discussion Paper 19, World 1988. A Handbook on Fighting Corruption. Bank, Washington, DC. http://siteresources. U.S. Agency for International Development worldbank.org / I N TENERGY/ Resources/ ( U SA I D), C ent er for D emo c rac y a nd Energy19.pdf. Governance, Washington, DC. Bhatia, B., and M. Gulati. 2004. “Reforming Djankov,S., C. McLiesh, and A. Shleifer. 2007. the Power Sector: Controlling Electricity “Private Credit in 129 Countries.� Journal of Theft and Improving Revenue.� Note 272 Financial Economics 8 (4): 299–329. Private Sector Development, World Bank, Economic Times. 2011. “Infra Sector Needs Rs Washington, DC. http://rru.worldbank.org/ 41 Lakh Cr Investment in 12th Plan Period: documents/publicpolicyjournal/272bhatia_ Economic Survey.� February 25. http://articles. Gulati.pdf. economictimes.indiatimes.com/2011-02-25/ Bird, R., and M. Zolt. 2007. “Tax Policy in news/28634119_1_power-sector-national- Emerging Countries.� International Tax forest-land-bank-land-acquisition. Program Paper 0707. University of Toronto. Ferrari, A. and I.S. Dhingra. 2009. India’s Bruhn, M. 2008. “License to Sell: The Effect Investment Climate Assessment: Voices of of B u s i n e s s R e g i s t r at io n R e f o r m o n Indian Business. World Bank, South Asia Entrepreneurial Activity in Mexico.� Policy Region, Washington, DC. Research Working Paper 4538, World Bank, Gill, J. B. S. 2003. The Nuts and Bolts of Revenue Washington, DC. Generation. World Bank, Washington, DC. Ca rlin, W., and M . S chaf fer. 2011a. “A Government of India. 2009. Position Paper C ompa r ison of Busi ne ss E nv i ron ment on Power Sector in India. Department of Constraints between Formal Sector Firms and Economic Affairs, New Delhi. Rural and Informal Sector Firms.� Background Government of Pakistan. 2010. Economic Survey study conducted for this book. 2009–10. Ministry of Finance, Islamabad. ———. 2011b. “Which Elements of the Business http://www.finance.gov.pk /survey_0910. Environment Matter Most for Firms and How html. 168 MORE AND BETTER JOBS IN SOUTH ASIA Hoxha, I. 2011. “Access to Finance in South Trimble, C., N. Yoshida, and M. Saqib. 2011. Asia.� Background study conducted for this Rethinking electricity tariffs and subsidy book. policy in Pakistan. Report 62971-PK, World IDFC (Infrastructure Development Finance Bank, Washington, DC. Company Limited). 2010. “India Infrastructure U S A I D ( U . S . A g e n c y for I nt e r n at ion a l Report: Infrastructure Development in a Low- Development). 2007. Impact and Benefits Carbon Economy.� New Delhi. of Power Trading in the South Asia Growth IIP (Institute of Public Policy). 2009. State of Quadrangle. http://www.sari-energy.org / the Economy: Emerging from the Crises. successdocs/ImpactBene�tsSAGQ.pdf. Second Annual Report. Beaconhouse National Va n , S ou L eng. 2011. “C omp et it ivene ss University, Lahore. Partnerships: How the Cambodia Government IMF (International Monetary Fund). 2010. Private Sector Forum (G -PSF) Helps the Wo rl d E c o n o m i c O u t l o o k D a t a b a s e , Garment Industry Grow.� Presentation to the October. Washington, DC. http:www.imf World Bank, April, Washington, DC. .org/external/pubs/ft/weo/2010/02/weodata/ World Bank. 2005. Sri Lanka: Improving index.aspx the Rural and Urban Investment Climate. Klapper, L., and I. Love. 2010. “The Impact of Washington, DC. Business Environment Reforms on New Firm ———. 2008a. The Afghanistan Investment Registration.� Policy Research Working Paper Climate Assessment. Washington, DC. 5493, World Bank, Washington, DC. ———. 2008b. Finance for All? Policies and K PMG . 20 09. T hi n k BR IC : In d i a . K e y Pitfalls in Expanding Access. Washington, DC. Considerations for Investors Targeting the ———. 2008c. Harnessing Competitiveness Power Sectors of the World’s Largest Emerging for Stronger Inclusive Growth: Bangladesh Economies. Mumbai. Second Investment Climate Assessment. Love, I., and N. Mylenko. 20 03. “Credit Washington, DC. Reporting and Financing Constraints.� Policy ———. 2009. Pakistan Investment Climate Research Working Paper 3142, World Bank, Assessment. Washington, DC. Washington, DC. ———. 2010a. Impact of the Global Financial Mint. 2011. “Can’t Erase State Power Firms’ Crisis on Investments in South Asia’s Electric Losses at One Go.� January 4. Power Infrastructure. India, Bangladesh, and Mishra, P. 2010. Alleviating Energy Poverty Pakistan. Report 56849-SAS. Washington, through Innovation. The Case of Jytotigram DC. Yojana of Gujarat. http://www.worldenergy ———. 2010b. Sri Lanka: Environmental Issues .org/documents/congresspapers/228.pdf. in the Power Sector. South Asia Sustainable OECD (Organisation for Economic Co-operation Development, Washington, DC. and Development). 2007. “OECD Economic ———. 2011a. India Urbanization Review. Surveys: India.� Paris. Washington, DC. Quah, John. 2007. Combating Corruption ———. 2011b. Perspectives on Poverty in India: Singapore-Style: Lessons for Other Asian Stylized Facts from Survey Data. Washington, Countries. Maryland Series in Contemporary DC. Asian Studies 2-2007 (189), University of ———. 2011c. PPI database. Washington, DC. Maryland School of Law, Baltimore, MD. http://www.ppi.worldbank.org. Transparency International. 2010. Corruption ———. 2011d. S r i L ank a Inf ra st r uc ture Pe rc e p t ion s I nd ex 2 010 . h t t p : // w w w Assessment. Washington, DC. .transparency.org/policy_research/surveys_ ———. 2011e. World Development Indicators. indices/cpi/2010/results. Washington, DC. CHAPTER 5 Opening the Door to Better Jobs by Improving Education and Skills Questions and Findings Questions investment in human capital among disad- vantaged groups. • What are the dimensions of the education • The biggest payoffs may come from address- and skills challenge in South Asia? ing poor nutrition and other factors in early • What are the policy priorities for improving childhood—before children enter formal the skills of graduates of education and train- schooling—in order to prevent development ing systems? problems that can hinder learning and con- tribute to early dropout, especially among children from poor families. Findings • Improving the quality of learning in primary • Increasing wage premiums at the upper- and secondary schools calls for measures secondary and tertiary levels of education that strengthen incentives and capacity in the signals increasing demand for more skilled school system. Priorities include strengthen- workers. In addition to knowledge and spe- ing national assessment systems, improving ci�c technical skills for the job, analytical and capacity and accountability at the school level, behavioral skills are being demanded. Upper- and improving the quality and performance secondary and tertiary education also sharply of teachers. Improving parents’ understand- increase the ability of workers to access ing of the returns to education, reducing their better jobs. Despite considerable recent gains, budget constraints through conditional cash three main dimensions to the education chal- transfers, and providing subsidies to private lenge remain: increasing attainment beyond providers to serve the poorest groups are some primary to secondary and tertiary education; strategies that have succeeded in increasing including all groups in this effort; and going education investments among disadvantaged beyond attainment to ensuring that individu- groups in primary and secondary education. als are equipped with the skills demanded in • In tertiary education and preemployment the labor market. training systems, priorities include providing • The targets and strategies of individual coun- information on outcomes and strengthening tries will depend on country-specific out- quality assurance; increasing the role of the comes, constraints, resources, capacity, and private sector in providing training and edu- political imperatives. Although speci�c tar- cation and in managing public institutions; gets will differ, there are two common pri- providing public institutions with greater orities. All countries will need to improve the autonomy and incentives for improved per- learning outcomes and skills of graduates of formance; and increasing contributions from education and training systems at all levels students able to pay while protecting those and address market failures that lead to lower who cannot. Opening the Door to Better Jobs by Improving 5 Education and Skills E ducation and training systems have This chapter takes a close look at the an important role to play in the labor challenges confronting education and train- market. They can influence the deci- ing systems in South Asia. These systems sions of � rms to invest, adopt new technol- need to equip more than 10 million poten- ogies or develop new products, and enter tial entrants a year into the labor force— new markets—decisions that determine including many from groups that have not how the structure of the economy evolves been suf� ciently included—with skills for as well as the number and productivity of productive employment. the jobs that are created. There is a strong The chapter is organized into eight sec- association between education and train- tions. The first looks at skills in demand ing, job creation, and � rm innovation (see relative to those available in South Asian chapter 4). Job growth takes place in larger labor markets. The second looks at how firms that are internationally engaged; education is connected to the ability of these firms are also more likely to intro- workers to access better jobs. The third duce new products and processes, engage discusses the dimensions of the education in research and development, have more challenge in South Asia. The fourth looks educated managers and workforces, and at the potential evolution of educational invest in the training of their workforce. attainment in South Asia’s rapidly expand- Education and training systems also affect ing labor force. The � fth looks at the role of individuals’ decisions to participate in the early childhood development in preparing labor force, their ability to access produc- children for success in school and the labor tive and more secure employment, and market. The sixth looks at primary and their earnings and welfare.1 Education is secondary education. The seventh exam- an important “sorter� in South Asian labor ines tertiary education and preemployment markets, with the less educated overwhelm- training systems. The last section exam- ingly work in agriculture, self-employment ines the extent to which workers in South (mostly low-end own-account or unpaid Asia’s formal sector � rms have opportuni- family work), or precarious and irregular ties to upgrade their skills through on-the- wage employment (see chapter 3). job-training. 171 172 MORE AND BETTER JOBS IN SOUTH ASIA Education and skills in very important but that they were only some- South Asian labor markets what satis�ed with the graduates’ skills. A survey in Sri Lanka of the workforce in This section reviews employers’ perceptions the information technology sector reveals a of skills constraints and wage premiums for similar picture (table 5.1). Together the two different levels of education. It finds evi- surveys provide evidence on the importance dence of considerable gaps in the supply of of behavioral, creative thinking, problem- skills in the labor force. Improving the skills solving, communication, and knowledge and of graduates requires building strong educa- speci�c skills. Some of the largest perceived tion and training systems and strengthening skills gaps are in these areas. The foundation institutional capacity. It does not happen for these skills is developed even before higher overnight. education, in primary and secondary educa- tion, as well as outside the education system. Employers’ perceptions of skills constraints Changes in demand: Trends in wage premiums and the skills Inadequate skills of the labor force rank content of labor among the top five constraints enterprise managers face in Bhutan and Maldives, Some evidence from India suggests that the where it is the second-most important con- skills content of jobs is slowly changing, espe- straint, and India, where job-creating � rms cially in urban areas, as demand for analytical rank lack of skills �fth (comparable in sever- and interactive skills increases. As countries ity to government policy uncertainty). 2 develop and technologies evolve, new occupa- Firms in Afghanistan and Pakistan rate skills tions appear and the skills demanded change. constraints among the least problematic. Autor, Levy, and Murnane (2003) exam- Although formal sector � rms in �ve of the ine the evolution of the skills content of the eight countries in South Asia do not report U.S. labor force over a 40-year period begin- skills as a top constraint, the evidence on ning in 1960. They document the rise of non- wage premiums for India, Nepal, Pakistan, routine cognitive analytical and interactive and Sri Lanka suggests that demand for skills (“new economy� skills) and the decline workers with secondary or tertiary education in routine cognitive and routine and nonrou- is increasing. Moreover, in focused surveys tine manual skills (see table 5.2).3 on skills, employers report dissatisfaction A similar analysis conducted for India with skills. between 1994 and 2010 for this book � nds Focused employer surveys in particular that the skills content remained fairly sta- sectors in India and Sri Lanka suggest con- ble, although there was a slight decrease cerns with the skills of tertiary education in the importance of routine and nonrou- graduates relative to skills in demand. A tine manual skills. The lack of change is survey of employers hiring recent engineer- not surprising, given the large share of the ing graduates in India shows that 64 percent workforce still employed in elementary, of employers are only somewhat satis�ed or agricultural, and trade-related occupations. worse with the quality of their new hires A closer look at the proportion of work- (Blom and Saeki 2011). Employers evalu- ers in occupations in which the content of ated the level of importance of a broad range “new economy skills� is high (4 or 5 on a of skills, including behavioral skills (such 1–5 scale), however, suggests that the share as teamwork, reliability, leadership, and of such occupations, although still modest, willingness to learn); creative thinking and is increasing steadily (see annex table 5A.1). problem-solving skills; communication skills; For urban wage earners, there has been a and speci�c knowledge and technical skills gradual increase in new economy and rou- needed for the job (�gure 5.1). They reported tine cognitive skills and a decline in manual that they considered most of these skills a skills (�gure 5.2). OPENING THE DOOR TO BE T TER JOBS BY IMPROVING EDUCATION AND SKILLS 173 FIGURE 5.1 Employers’ perceptions of skills of recently graduated engineers in India integrity reliability teamwork willingness to learn entrepreneurship self-discipline communication in english self-motivated flexibility leadership responsibility modern tools knowledge math/science/English creativity written communication reading technical skills experiments/data analysis verbal communication basic computer problem solving empathy system design contemporary issues advanced computer customer service 0 not at all not very somewhat very extremely satisfaction importance Source: Blom and Saeki 2011. TABLE 5.1 Perceived demand for and de�ciency in skills in programming and software engineering in Sri Lanka Skills in demand Rank Essential skills Complementary skills Skills deficient in employees 1 Programming Team working Communication skills 2 Systems Design Communication skills Creative thinking skills 3 .Net Creative thinking skills Systems design 4 Systems Analyst Interpersonal skills Proficiency in English 5 Visual Basic C++ Programming 6 Java SQL server Team working 7 System testing Professional ethics Java 8 Business analysis System testing Interpersonal skills 9 Team working XML Professional ethics 10 Interpersonal skills Systems design .Net Source: Sri Lanka Information Communication Technology Association 2007. 174 MORE AND BETTER JOBS IN SOUTH ASIA TABLE 5.2 Routine and nonroutine skills categories Type of skill Tasks Examples of occupations requiring skills Nonroutine cognitive: Analytical Analyzing data/information Thinking creatively Interpreting information for others Lawyers, professors, doctors, training and Nonroutine cognitive: Interpersonal Establishing and maintaining personal relationships development managers Guiding, directing, and motivating subordinates Coaching/developing others Routine cognitive Repeating same tasks Telephone operators, bookkeepers, Being accurate accounting and auditing clerks, meter Following tasks, priorities, and goals established for readers, cashiers the job Routine manual Importance to this job that the pace is determined by the speed of controlling machines and processes Making repetitive motions   Industrial truck operators; machine setters, Nonroutine manual Operating vehicles, mechanized devices, operators, and tenders; construction or equipment carpenters Using hands to handle or control objects or tools Using manual dexterity and spatial orientation Source: Acemoglu and Autor 2010. Note: Acemoglu and Autor use the skills structure in the occupational information network database published by the U.S. Department of Labor, Employment and Training Administration to update the results in Autor, Levy, and Murnane (2003). FIGURE 5.2 Evolution of skills content of urban wage workers in Consistent with these patterns, the wage India, 1994–2010 premium associated with the more skilled occupations has increased relative to elemen- 70 percentile (1994 skills distribution) tary occupations (�gure 5.3). Sharp increases 65 are evident in changes in the premium for 60 skilled occupations over elementary occupa- 55 tions between 1994 and 2010. This pattern is 50 consistent with the picture above, which sug- 45 gests gradually increasing demand for more 40 skilled occupations. It is also consistent with 35 the � nding below on increasing returns for 30 1994 2000 2005 2008 2010 higher levels of education, especially tertiary new economy skills routine cognitive skills manual skills education. The wage premium increased for higher Source: World Bank staff, based on national labor force surveys. levels of education in several countries, even Note: Data cover 21- to 60-year-old urban casual workers and regular wage or salaried workers. Changes in skills contents are assessed using the method proposed by Autor, Levy, and Murnane as the supply of educated workers increased. (2003) and the skills categories identi�ed in the O*NET database, as suggested by Acemoglu and The premiums shown in figure 5.4 reflect Autor (2010). The skills content level of every worker’s job is assessed in each of the �ve skills cat- egories, and a population distribution is created for each of the �ve skills categories. The process is the differential between a particular level repeated in each period surveyed. The skills content level for the median worker (50th percentile) in of educational attainment and the level of the distribution is identi�ed, and the percentile corresponding to the same skills content level in the base year of 1994 is tracked. Thus, for example, the average intensity of new economy skills for the attainment just below. The pattern is one in urban wage workers was around 10 centiles more in 2010 versus 1994. “New economy skills� is an which the premium to lower levels of educa- aggregate of nonroutine cognitive analytical and nonroutine cognitive interpersonal skills. “Manual skills� is an aggregate of routine manual and nonroutine manual skills. tion has been falling while the premium for OPENING THE DOOR TO BE T TER JOBS BY IMPROVING EDUCATION AND SKILLS 175 FIGURE 5.3 Wage premiums over elementary occupations in India, 1994–2010 40 34 33 30 29 25 25 21 22 23 22 23 23 19 22 21 20 17 16 18 14 13 13 14 12 9 percent 10 1 1 0 –4 –5 –8 –10 –9–10 –20 –16 –20 –30 managers professionals technicians clerks service skilled people in crafts machine workers agricultural or trades operators workers 1994 2000 2005 2010 Source: Authors, based on data from national labor force surveys. Note: The bars represent the wage premiums of each occupation relative to elementary occupations. The wage premiums are obtained as the coefficients in a regression of log hourly wage on a number of individual characteristics such as age, education, occupation, caste, and so on. The omitted occupation category is elementary occupations. FIGURE 5.4 Wage premiums in selected South Asian countries, by level of education a. India and Sri Lanka 60 50 43 42 42 43 43 42 40 37 36 36 36 percent 31 32 33 33 30 30 26 26 20 19 21 18 20 17 14 16 15 10 13 10 0 1994 2000 2005 2010 2000 2008 India Sri Lanka 60 b. Nepal and Pakistan 40 37 35 32 34 34 percent 32 28 26 25 24 22 23 24 22 20 14 8 10 11 13 9 0 1999 2008 2000 2008 Nepal Pakistan incomplete primary complete primary complete lower-secondary complete higher-secondary complete tertiary Source: Authors, based on data from national labor force and household surveys. Note: The �rst bar for each country-year pair reflects the wage premium for even some primary education relative to no education; the last bar reflects the wage premium for completing tertiary relative to completing upper-secondary education. The wage premiums represent the differences in the coefficients of a regression of log hourly wage on basic controls (the highest level of education completed, experience, and experience squared). The qualitative results are robust to other speci�cations with additional controls. 176 MORE AND BETTER JOBS IN SOUTH ASIA upper-secondary and tertiary education has long been high, most countries in the region been increasing. This trend is taking place in a have seen rapid declines in the share of the context in which the educational attainment of labor force that does not have any education the labor force has been generally rising, with (�gure 5.5). Declines were particularly sharp an increase in the share of workers with pri- in Nepal, Bangladesh and Pakistan. mary, secondary, and tertiary education (see Despite this progress, the share of work- annex �gure 5A.1). This pattern is thus con- ers in the labor force without any education sistent with a situation in which the supply of at all remains high relative to the region’s workers at lower levels of education is increas- level of development reflecting relatively ing faster than demand while the demand for lower attainment at all levels of education workers with secondary or tertiary education (see annex �gure 5A.2). (The main exception is outpacing the increased supply. is primary attainment in Sri Lanka.) Educa- There are important differences across tional attainment of men in the labor force is countries. India and Nepal have seen increases higher than that of women, but in Afghani- in the premium to both upper-secondary and stan, Bhutan, and Pakistan, the share of the tertiary education. Indeed, the wage premium male labor force with no education is high for tertiary education more than doubled in compared with other countries at similar India between 2000 and 2010, despite an income levels. In the case of women, more increase in the share of the labor force with than half of the labor force in all countries tertiary education from 1 percent to 8 per- other than Sri Lanka and Maldives lacks cent over this period. This increase is seen in any education at all. Much of this difference both the public and private sectors, in urban reflects gaps in educational attainment of and rural areas.4 In Nepal, the largest rela- earlier generations, but gender gaps remain tive increases in the wage premium were at even in young cohorts of the population in the upper-secondary level, suggesting that many countries (see annex �gure 5A.6). demand increases at this level were greater than the small increases in supply (from 2 per- cent of the labor force to 3 percent) between Education and access to 1999 and 2008. In Sri Lanka, wage premi- better jobs ums for upper-secondary education increased There is a strong link between education substantially, signaling signi�cant increases and type of employment. A multinomial logit in demand relative to supply at this level. The model was used to explore the link between share of the labor force with lower-secondary household and individual characteristics— education rose 10 percentage points between including education—and whether and where 2000 and 2008; the increase in the supply of individuals work. It reveals that in all coun- workers with upper-secondary and tertiary tries, an individual’s education level is typi- education was much smaller. cally the most important determinant. In Pakistan, trends in wage premiums in For both working men and women in the private sector are different from the trends rural and urban areas, the likelihood of being shown in �gure 5.4. In the private sector, the in a better job increases with education, par- premium to tertiary education over upper- ticularly with secondary and higher levels secondary increased, whereas the premium to (�gure 5.6). The increases are sharpest begin- all other levels fell or—in the case of primary ning with the completion of upper-secondary education—stayed more or less stable. education, except in Bangladesh, where the sharpest increases occur at the tertiary level (other exceptions include male workers in Educational attainment urban India, where the increases are not as of the labor force sharp at upper-secondary, and female work- With the exception of Afghanistan, for ers in urban Pakistan, where sharp increases which data are available only for 2008, and occur even with completion of lower- Sri Lanka, where primary attainment has secondary school). Disaggregation within OPENING THE DOOR TO BE T TER JOBS BY IMPROVING EDUCATION AND SKILLS 177 FIGURE 5.5 Share of South Asian labor force with no education, with international comparisons a. Female labor force 100 Afghanistan 2008 Bhutan 2003 Pakistan 2000 80 Nepal 1999 Bhutan 2007 Pakistan 2008 India 1994 India 2000 Pakistan 2009 Nepal 2008 India 2005 60 India 2008 Bangladesh 2002 India 2010 percent Bangladesh 2005 Maldives 1998 40 Maldives 2004 20 Sri Lanka 2000 Sri Lanka 2004 Sri Lanka 2008 0 Sri Lanka 2006 0 1,000 2,000 3,000 4,000 5,000 per capita GDP in purchasing power parity dollars b. Male labor force 100 80 Bhutan 2003 Afghanistan 2008 60 Bhutan 2007 percent Nepal 1999 Bangladesh 2002 Maldives 1998 Pakistan 2000 40 Bangladesh 2005 Pakistan 2008 Pakistan 2009 Nepal 2008 India 1994 India 2000 Maldives 2004 India 2005 20 India 2008 India2010 Sri Lanka 2006 Sri Lanka 2000 Sri Lanka 2008 0 Sri Lanka 2004 0 1,000 2,000 3,000 4,000 5,000 per capita GDP in purchasing power parity dollars Sources: Authors, based on data from World Bank 2011b and national labor force and household surveys. Note: The dark line shows the predicted values of the share of the labor force with no education by per capita GDP, based on a cross-country regression (excluding high-income countries). The shaded area is the 95 percent con�dence interval band around the regression line. GDP = gross domestic product. rural areas (not shown in figure) indicates or salaried jobs. Among male workers in that this pattern primarily reflects the fact Bangladesh, Nepal, and Pakistan, the likeli- that workers with secondary or higher levels hood of being in casual nonfarm employment of education are much more likely to work increases with some primary education and in nonfarm self-employment or regular wage falls at higher levels of education. 178 FIGURE 5.6 Predicted probability of working in rural nonfarm and urban regular wage jobs in selected South Asian countries, by level of education and gender a. Likelihood of rural workers in Bangladesh b. Likelihood of urban workers in Bangladesh being in nonfarm jobs, 2005 being in regular wage or salaried jobs, 2005 100 100 80 80 60 60 percent percent 40 40 20 20 0 0 n y nd er nd er y n y nd er nd er y ar ar ar ar io io co low co upp co low co upp im rti y y im rti y y at at ar ar ar ar uc uc te te pr pr ed ed no no se se se se c. Likelihood of rural workers in India d. Likelihood of urban workers in India being in nonfarm jobs, 2008 being in regular wage or salaried jobs, 2008 100 100 80 80 60 percent 60 percent 40 40 20 20 0 0 n y nd er nd er y n y nd er nd er y ar ar ar ar io io co low co upp co low co upp im rti im rti y y y y at at ar ar ar ar uc uc te te pr pr ed ed no no se se se se female male e. Likelihood of rural workers in Nepal f. Likelihood of urban workers in Nepal being in rural nonfarm jobs, 2008 being in regular wage or salaried jobs, 2008 100 100 80 80 60 60 percent percent 40 40 20 20 0 0 n im e y nd er nd er y n im e y nd er nd er y ar ar ar ar pr som pr som io io co low co upp co low co upp im rti im rti y y y y y y at at ar ar ar ar ar ar uc uc te te pr pr ed ed no no se se se se g. Likelihood of rural workers in Pakistan h. Likelihood of urban workers in Pakistan being in rural nonfarm jobs, 2009 being in regular wage or salaried jobs, 2009 100 100 80 80 60 60 percent percent 40 40 20 20 0 0 n im e y nd er nd er y n im e y nd er nd er y ar ar ar ar pr som pr som io io co low co upp co low co upp im rti im rti y y y y y y at at ar ar ar ar ar ar uc uc te te pr pr ed ed no no se se se se female male Source: Authors, based on data from national labor force and household surveys. Note: The likelihood of workers being in a particular sector or type of employment is the predicted share of workers in a particular sector or type of employment from a multinomial logit regression of occupational choice on household and individual characteristics. 179 180 MORE AND BETTER JOBS IN SOUTH ASIA Working women with less than upper- type with varying educational levels among a secondary education (lower-secondary in cross-section of workers, but panel data are Pakistan) are generally less likely than men needed to provide a picture of labor mobility. to hold better jobs. The main exception is In the absence of panel data for South Asian in urban Bangladesh, where women who countries, analysis of employment transitions work are likely to do as well or better than was carried out using pseudo-panel data (see men at all levels of education. Women see a appendix B). Education emerges as one of the sharper increase in the likelihood of being in strongest correlates of labor mobility in all a better job with secondary or tertiary educa- countries and all years considered. tion than do men: at these levels of education, Secondary and higher levels of education women are at least as likely to be in a better increase the ability of workers to move out of job. In rural Bangladesh and Nepal, unlike in agriculture, casual wage jobs, and low-end Pakistan and India, the likelihood that work- self-employment to better jobs. Although ing women with upper-secondary or tertiary the analysis was carried out for Bangladesh, education are in better nonfarm jobs is not Nepal, and two time periods for India, in the higher than that of men. interest of space, figures 5.7 and 5.8 pres- Education also improves labor mobility. ent the results only for worker transitions Multinomial logit regressions provide insights between 2004/05 and 2007/08 in India. The into the distribution of workers’ employment figures show the probability of transition FIGURE 5.7 Conditional probability of moving into and out of better jobs in rural India, by education and gender, 2005–08 a. Transition by men from b. Transition by men from agriculture to nonfarm jobs nonfarm jobs to agriculture 80 80 60 60 percent percent 40 40 20 20 0 0 io o im e y da er nd er y io o im e y da er da er y ar ar ar ar pr som pr som n n on low co pp on low on pp im rti im rti n y ry y n y ry ry ar ar ar se u u te te pr pr at at uc uc c c c ed ed se se se c. Transition by women from d. Transition by women from 80 agriculture to nonfarm jobs nonfarm jobs to agriculture 80 60 60 percent percent 40 40 20 20 0 0 at no im e y nd er nd er y at no im e y nd er nd er y ar ar ar ar pr som pr som co low co pp co low co pp im rti im rti n y y y n y y y ar ar ar ar ar ar io io se u se u te te pr pr uc uc ed ed se se Source: Authors, based on data from national labor force and household surveys. Note: The probability of transition is conditional on being in a speci�c type of employment in the �rst period (e.g., panel a shows the estimated probability that a rural male worker who was in agriculture in the �rst period is working in a nonfarm job in the second period). The probability differs by the level of education of the worker. Upper and lower bounds of the estimated probabilities are shown. The blue lines are drawn through the midpoints of the bounds. OPENING THE DOOR TO BE T TER JOBS BY IMPROVING EDUCATION AND SKILLS 181 FIGURE 5.8 Conditional probability of moving into and out of better jobs in urban India, by education and gender, 2005–08 a. Upward transition by men b. Downward transition by men 80 80 60 60 percent percent 40 40 20 20 0 0 at n o im e y nd er nd er y at no im e y nd er nd er y ar ar pr som ar ar pr som co low co pp co low co upp im rti n y y y im rti n y y y ar ar ar ar ar ar io io se u te te pr pr uc uc ed ed se se se c. Upward transition by women d. Downward transition by women 80 80 60 60 percent percent 40 40 20 20 0 0 no im e y nd er nd er y no im e y nd er nd er y ar ar ar ar pr som pr som co low co upp co low co upp im rti im rti n y y y n y y y ar ar ar ar ar ar io io te te pr pr at at uc uc ed ed se se se se Source: World Bank staff, based on data from national labor force and household surveys. Note: See note for �gure 5.7. An upward transition in the urban context is a move from casual wage work or low-end self-employment to regular wage or salaried work or high-end self-employment. A downward transition is a move from regular wage or salaried work or high-end self-employment to casual wage work or low-end self-employment. between the first and second period con- • Workers with higher levels of education ditional on type of employment in the � rst were much more likely to move to better period, by education and gender. 5 Three jobs than they were to lose them, as shown interesting �ndings emerge: in the lower levels of the transition bars for higher levels of education in panels b • Rural workers who were in agriculture and d of �gures 5.7 and 5.8 than in panels and urban workers who were in casual or a and c. Workers with lower levels of edu- low-end self-employment in the �rst period cation were more likely to lose better jobs were more likely to be able to make the than they were to gain them, as shown in transition to a better job (nonfarm employ- the higher levels of the transition bars for ment in rural areas and high-end self- lower levels of education in panels b and d employment or regular wage or salaried than in panels a and c. jobs in urban areas) if they had secondary • These patterns appear to be sharper for or higher levels of education (panels a and c women in both urban and rural areas, sug- of �gures 5.7 and 5.8). This mobility is typ- gesting that education may play a stronger ically greater for workers with completed role in worker mobility for women. upper-secondary education (in Bangla- desh for workers with tertiary education).6 Workers with lower levels of education The education challenge were more likely to transition out of better The analysis described in the previous jobs (for example, from nonfarm employ- sections suggests three dimensions to the edu- ment to agriculture in rural areas) than cation challenge in South Asia. First, there is workers with higher levels of education. a need to increase the educational attainment 182 MORE AND BETTER JOBS IN SOUTH ASIA of the labor force beyond primary to second- Including all groups ary and higher levels of education, where Figure 5.10 shows mean years of education the payoffs in worker mobility and earnings of young cohorts of the population by gen- are greatest. Second, it will be important to der, location, and socioeconomic group. 8 ensure that all groups—including women, Three facts stand out. First, with the excep- ethnic minorities, and the poor—have equal tion of Sri Lanka, gender, rural, and socio- opportunities to access education. Third, economic disadvantages persist even in although increased opportunities to attend young cohorts. Gender disparities in mean educational and training institutions are nec- years of education in Bangladesh and Mal- essary, they are not suf�cient. It is essential dives are close to being eliminated, but rural that individuals graduate from these institu- and socioeconomic disparities remain.9 tions equipped with skills that will enable Second, rural and low-income women them to access better employment opportu- are the most disadvantaged, and urban and nities and succeed in the world of work. This high-income men are the least disadvantaged. section examines how South Asia is doing on There are important country differences, but these three dimensions. the gap is close to or more than four years on both dimensions in �ve of eight countries. Increasing educational attainment Third, urban and higher-income women have higher educational attainment than Despite signi�cant progress by young cohorts, rural and low- income men in most countries, educational attainment in young cohorts of suggesting that the rural and income disad- the population in South Asia remains low, vantage dominates the gender disadvantage. reflecting low attainment of upper-secondary Ethnic minorities and disadvantaged castes and tertiary education in all countries and of also have lower educational attainment, even primary education in Afghanistan. The aver- among younger cohorts. In India, scheduled age years of education of young cohorts of castes and tribes and Muslim minorities have the population (age 15–34) has increased signi�cantly lower educational attainment. in all countries since the late 1990s or early In Nepal, the Dalits and Muslims are among 2000s (annex table 5A.2). It is still low the most disadvantaged. Ethnic differences in in most countries, however, ranging from education outcomes are also observed in Sri 2.5 years in Afghanistan to 10.2 years in Lanka (see annex �gure 5A.8). These disad- Sri Lanka. About 80 percent of youth com- vantages by gender, location, and socioeco- plete primary education in all South Asian nomic status reflect both a still low likelihood countries except Afghanistan, Nepal, and of starting school and a high likelihood of Pakistan (�gure 5.9; see annex �gures 5A.6 dropping out, with variations across coun- and 5A.7 for a break-out by gender). Primary tries (see annex �gure 5A.9.) attainment is especially low in Afghanistan, Education disadvantages by gender, socio- where only a third of youth complete pri- economic status, and ethnicity play out in mary school. Considerable progress has been the labor market and are an important rea- made in most countries in lower-secondary son for the small number of workers from attainment, although less than 60 percent of disadvantaged groups holding better jobs recent cohorts completed lower-secondary (see chapter 3). in all countries except India, Maldives, and Sri Lanka. Completion of upper-secondary and tertiary by young cohorts remains low in Equipping children with skills all countries in the region, with only 21–23 percent completing upper-secondary even in Children in school in South Asia have very India, Maldives, and Sri Lanka.7 In all coun- low levels of learning, with large numbers of tries, less than 10 percent of youth have com- children acquiring no more than basic literacy. pleted tertiary education. No systematic studies enable comparison of FIGURE 5.9 Share of young cohorts with completed primary, secondary, and tertiary education in South Asia, by country a. Primary attainment by 15–19 year olds and lower-secondary attainment by 20–24 year olds 99 100 98 95 87 88 80 78 75 70 70 73 65 64 63 58 60 60 percent 47 49 48 48 46 42 40 37 33 23 22 20 16 0 2008 2000 2010 2000 2010 1998 2004 1999 2008 2000 2009 2000 2008 Afghanistan Bangladesh India Maldives Nepal Pakistan Sri Lanka completed primary (15–19 year olds) completed lower secondary (20–24 year olds) b. Upper-secondary attainment by 25–29 year olds and tertiary attainment by 30–34 year olds 30 22 23 21 20 18 percent 15 16 10 10 11 10 9 9 8 7 8 6 5 3 4 4 4 4 3 4 2 2 1 0 2008 2000 2010 2000 2010 1998 2004 1999 2008 2000 2009 2000 2008 Afghanistan Bangladesh India Maldives Nepal Pakistan Sri Lanka completed primary (25–29 year olds) completed lower secondary (30–34 year olds) Source: Authors, based on data from national labor force and household surveys. 183 184 MORE AND BETTER JOBS IN SOUTH ASIA learning outcomes in primary, secondary, or 5 competency in math (93 percent in oral tertiary education across South Asian coun- math, 80 percent in written math) (Asadul- tries or over time. There is also limited reliable lah and others 2009). In India and Pakistan, information on learning from national asses- learning levels are signi�cantly below cur- sment systems, particularly on trends over riculum standards (Das, Pandey, and Zajonc time within countries. However, as discussed 2006). There is also evidence that children below, evidence from several in-country stu- have dif� culties with conceptual thinking dies and surveys suggests that the low level and problem-solving skills. In Pakistan, less of learning is perhaps the greatest challenge than 2 percent of students in a representa- facing the education and training systems in tive national sample (in the 2006 national all countries. For example, 62 percent of chil- assessment) were able to reason in settings dren in class 4 in India and Pakistan could involving the application of concept de� ni- read only at the expected level for class 1: tions, relationships, or representations of after four years of schooling, they could at either, and less than 1 percent were able to most read a short paragraph. Results in math demonstrate the choice and application of in both countries are also poor (�gure 5.11). mathematical processes in different situ- In Bangladesh, the vast majority of chil- ations. Similar evidence is available from dren who complete grade 9 attain only grade India from a sample of elite private schools, FIGURE 5.10 Mean years of education of 15–34 year olds in South Asia, by gender and country a. Rural versus urban 12 10.0 10.0 10 9.7 9.6 9.5 9.1 9.2 9.0 8.9 8.1 8 7.7 7.8 7.5 7.3 7.2 7.3 7.3 6.9 6.7 6.5 mean years 6.3 6.0 6 5.5 5.2 5.2 4.2 4.0 3.9 4 3.5 2.9 2.8 2 0.6 0 8 0 07 10 4 8 8 7 00 01 00 00 00 00 20 20 ,2 ,2 l, 2 2 ,2 ,2 n, a, s, an h an ka pa ive di es ta an ist ist Ne In u lad ald Bh an iL k Pa ng M Sr gh Ba Af rural female rural male urban female urban male (continues next page) OPENING THE DOOR TO BE T TER JOBS BY IMPROVING EDUCATION AND SKILLS 185 FIGURE 5.10 (continued) 12 b. Poorest versus richest quintile 10.2 10 9.8 9.7 9.8 9.9 9.6 9.4 9.2 9.0 9.0 8.9 8.8 8.7 8 7.7 7.5 7.6 6.9 mean years 6 5.9 5.8 5.4 4.3 4 4.0 3.5 3.3 3.0 2.6 2.5 2 1.9 1.9 1.2 1.0 0.5 0 8 10 7 10 4 4 8 7 00 00 00 00 00 00 20 20 ,2 ,2 2 l, 2 ,2 ,2 h, a, s, an an an ka pa ive di es an ist ut ist Ne In lad ald Bh an iL k Pa ng M Sr gh Ba Af poorest female poorest male richest female richest male Source: Authors, based on data from national labor force and household surveys. Note: Socioeconomic group is based on per capita consumption quintile of the household. suggesting that concerns of quality are not Except in Bangladesh and Pakistan, average limited to public or low-cost private schools public education expenditures in South Asia (EI 2006). are comparable to or higher than in East Asia (�gure 5.12).10 Although countries may chose to increase budgetary allocations to meet the Two common priorities education challenge, there is also considerable The strategies and targets of countries will room to increase the efficiency with which depend on country-speci�c outcomes, con- existing resources are used. The solid progress straints, resources, capacity, and political in improving educational attainment in recent imperatives. In the face of multiple dimen- years in South Asia has been achieved by the sions to the education challenge, countries combined efforts of governments, households, will determine their own priorities and tar- and the nongovernment and private sector—in gets. For example, achieving completion of the face of resource and capacity constraints. primary remains a priority in Afghanistan These efforts will continue to be important and in some other countries in the region, and, indeed, may be leveraged further. whereas Bangladesh is concerned about Although country’s strategies and targets providing equal opportunities for boys and will differ, all countries share two priorities. girls from low-income households in pri- First, what is ultimately critical for labor mar- mary and secondary school. ket success is what individuals learn. Research 186 FIGURE 5.11 Reading and arithmetic achievement in rural India and Pakistan, by class, 2010 a. Distribution of children by reading level, India b. Distribution of children by arithmetic level, India 100 3 100 2 4 4 9 3 9 20 13 22 17 14 18 29 80 38 41 80 36 27 49 53 58 26 68 36 67 60 32 76 60 36 24 83 percent percent 41 42 29 18 34 37 40 30 40 25 31 22 17 32 35 28 27 20 19 20 20 20 21 34 19 13 15 16 34 21 17 11 14 12 12 10 8 5 12 12 8 9 6 7 4 3 3 8 6 5 3 8 0 3 2 1 1 2 1 0 3 2 1 1 2 1 1 2 3 4 5 6 7 8 all 1 2 3 4 5 6 7 8 all class class nothing letter word standard I text standard II text nothing recognizes numbers 1–9 recognizes numbers 1–99 subtraction division c. Distribution of children by reading level, Pakistan d. Distribution of children by arithmetic level, Pakistan 100 4 100 2 9 4 5 5 11 21 14 21 15 38 80 26 34 80 24 25 50 52 57 24 67 69 72 36 75 60 80 60 41 37 86 84 percent percent 90 38 26 36 37 41 40 33 40 25 30 25 30 28 27 24 18 20 16 20 21 21 29 16 16 32 16 16 12 14 9 8 9 11 11 12 10 8 5 6 15 6 7 6 7 5 4 3 3 2 8 3 3 2 1 4 0 4 2 2 1 2 2 1 2 11 0 4 3 3 2 1 2 1 1 1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10 class class nothing letter word standard I text standard II text nothing recognizes numbers 1–9 recognizes numbers 1–99 subtraction division Source: Pratham 2010. OPENING THE DOOR TO BE T TER JOBS BY IMPROVING EDUCATION AND SKILLS 187 FIGURE 5.12 Public expenditure on education as a share of GDP, in South Asia and other regions 6 5.1 5.2 5 4.8 4.9 4.6 4.4 4.0 4 3.8 percent 3.1 3.1 3.1 3 2.9 2.7 2.4 2 1 0 08 7 t, 2 tio r 9 06 1 9 8 20 ific, e C in 200 a, 20 d 08 , 2 nd No Mid 200 ia, en era fo 00 00 00 00 00 a, an c s 20 20 20 an a a fri 08 Am 8 7 7 06 00 n lA pm p on c ,2 ,2 ,2 l, 2 ,2 ric st 00 Pa nA h, a, , be ric ra lo -o ati rld an ka an sia Af e Ea 7 pa di es nd nt ib e ra an ve Co nis ist ut wo hA Ne In lad Ce a ha rth dl Bh iL k De ic a ia Pa ut om Org Sa ng d Sr As So an b- th Lat Ba ar st Su pe Ea ro Eu an con E d Source: World Bank 2011b. Note: Public expenditure on education is reported to be about 11 percent of GDP in the Maldives (not shown in �gure). GDP = gross domestic product. suggests that the level of skills—as measured dramatically. Figure 5.13 shows the composi- by performance on international student tion of the labor force in 2010 and what it assessments such as the Programme for Inter- could look like under an optimistic but feasi- national Student Assessment (PISA) and the ble scenario of progress in educational attain- Trends in International Mathematics and ment in 203012 (see annex 5B for assumptions Science Study (TIMSS)—predicts economic and more details). This scenario corresponds growth far better than do average schooling to one in which primary attainment of young levels (Hanushek and Woessman 2008).11 cohorts is based on the six best-performing Thus an important focus for governments is South Asian countries over the previous to establish and strengthen the systems that 10 years and secondary attainment follows will lead to improved learning outcomes and the trends of a smaller subset of countries in skills of graduates at all levels. Second, gov- South Asia with the highest progression.13 ernments have an important role to play in With effort, South Asian countries could addressing market failures that lead to lower dramatically improve educational attainment investment in human capital among disadvan- of their workforces over the next 20 years. taged groups. Many countries should be able to more than halve the share of the labor force with no education; in all countries, the percent of the The next 20 years: Can South labor force with only primary education or Asian countries improve the less will fall. The share of individuals with educational attainment of their at least some secondary education could rise labor forces? dramatically. Under this scenario, the per- Should all South Asian countries increase centage of the labor force with some second- education attainment at the highest pace ary education could reach almost 30 percent observed among countries in the region, the in Afghanistan and over 50 percent in India, composition of the labor force could change Nepal, and Sri Lanka. Tertiary attainment 188 MORE AND BETTER JOBS IN SOUTH ASIA FIGURE 5.13 Educational attainment of the South Asian labor force in 2010 (estimated) and 2030 (projected), by country a. 2010 100 2 5 2 6 8 7 10 18 15 22 27 80 15 26 36 32 27 29 60 27 percent 36 19 52 24 40 73 33 53 20 43 42 33 33 23 20 0 3 an h an a s l an ka pa ive di es an ist ut ist In Ne lad ald Bh an iL k Pa ng M Sr gh Ba Af b. 2030 100 4 5 10 9 12 11 29 31 80 29 39 47 54 50 60 54 percent 25 42 40 29 59 29 22 16 18 20 43 25 23 19 23 15 16 9 0 4 1 an h an a s l n a pa ive di k es ta an ist ut In Ne kis lad ald Bh an iL Pa ng M Sr gh Ba Af no education primary secondary tertiary Sources: Authors, based on data from national labor force and household surveys; UN 2009, 2010; KC and others 2010. remains under 12 percent in all countries have lower levels of education. Except in Sri except Maldives and Sri Lanka, although Lanka, the majority of prime-age (35–49) under a more ambitious scenario, most coun- and oldest (50–64) adults in the labor force tries could raise the percentage of the labor will continue to have no more than primary force with tertiary education to at least 20 attainment in 2030 (�gure 5.14). The share percent (see annex table 5B.2). of prime-age adults with primary attainment Progress will be made by younger cohorts or below ranges from 30 percent of the labor and new labor market entrants; a large stock force in Sri Lanka to 88 percent in Afghani- of prime-age and older workers will still stan; the picture is worse in rural areas. These OPENING THE DOOR TO BE T TER JOBS BY IMPROVING EDUCATION AND SKILLS 189 FIGURE 5.14 Projected shares of prime-age (35–49) and oldest (50–64) labor force participants in South Asia with no education or only primary attainment in 2030, by country a. Labor force, 35–49 years old b. Labor force, 50–64 years old 100 100 81 82 80 77 80 76 65 65 60 60 60 58 percent percent 50 47 47 44 39 40 40 40 37 32 33 30 26 22 23 22 20 18 20 16 20 16 15 12 10 8 4 5 0 0 an h an a s l an ka an h an a s l an ka pa pa ive ive di di es es an an ist ut ist ist ut ist In Ne In Ne lad lad ald ald Bh Bh an iL k an iL k Pa Pa ng ng M M Sr Sr gh gh Ba Ba Af Af no education primary Sources: Authors, based on data from national labor force and household surveys; UN 2009, 2010; KC and others 2010. poorly educated will be likely hold informal of health and nutrition early on can have jobs, with lower earnings and higher labor effects that persist throughout life; damage market risks. (Chapter 6 discusses programs from childhood disease and malnutrition that can help informal workers adjust to can be dif�cult to undo. labor market shocks and improve their pro- There are critical periods in childhood ductivity and future earnings potential.) Pol- for acquiring different skills.15 Three types icy options and promising interventions for of outcomes in early years are critical for improving the skills of potential new entrants future development: physical growth and into the labor force (from education and well-being, cognitive development, and training systems), and including individuals socioemotional development. The context in from disadvantaged groups in these efforts, which a child develops and is either stimu- are discussed in the next few sections of this lated or shocked can have a signi�cant influ- chapter. Some of the constraints education ence on how much of his or her potential is and training systems face in improving the eventually realized. skills of potential new entrants into the labor Developmental delays often begin early in force and policy options and promising inter- life and accumulate quickly over time, espe- ventions for addressing them are discussed cially for the poorest children. Low levels of later in this chapter. cognitive development in early childhood are often strongly correlated with low socioeco- nomic status, as measured by wealth, parental Addressing disadvantages education, and malnutrition. Evidence from before school: The role of early Cambodia, Ecuador, Nicaragua, Madagascar, childhood development and Mozambique shows that children from The first years of life—long before for- low-income households begin to fall behind mal schooling begins—are a key period on various measures of cognitive develop- for building human capital.14 The bene�ts ment starting from about 36 months (Nadeau 190 MORE AND BETTER JOBS IN SOUTH ASIA and others 2011). Falling behind is evident employment in higher-paying and white col- not only in language development but also lar jobs and were less likely to live in poor in behavioral (sometimes termed “noncog- households (Hoddinott and others 2011). nitive�) abilities, such as sustained attention These and a number of other studies indicate (Fernald, Galasso, and Ratsifandrihamanana high returns to preventing malnutrition. 2011). The factors behind these shortfalls are Micronutrient deficiencies, which often still being delineated (for a recent review of begin during the prenatal period, can also risk factors, see Walker and others 2011). impair cognitive and motor development in Beyond nutrition and infectious diseases, children. Evidence from India and Nepal is early learning opportunities and caregiver- consistent with global evidence. In India, child interactions, which vary with parental for example, provision of iron supplementa- education and wealth, also play a role. tion and deworming medicine to preschool There is considerable evidence, including children had positive impacts on school from South Asia, that improved nutrition attendance, with much stronger responses enhances lifetime learning and labor mar- among children who were anemic at baseline ket productivity (Behrman, Alderman, and (Bobonis, Edward, and Puri-Sharma 2006). Hoddinott 2004). In a study that tracked Children of Nepali mothers provided with birth cohorts of Indians since 1969, research- iron and zinc during pregnancy performed ers conclude that both birthweight and weight better on cognitive measures as well as on tests gain in the � rst two years of life are strong of working memory, inhibitory control, and predictors of schooling outcomes (Martorell �ne motor functioning (Christian and others and Horta 2010).16A study of Pakistan �nds 2010). Addressing micronutrient de�ciencies that malnourished girls—but not boys—were long before school begins, often with prenatal less likely to enter school (Alderman and oth- micronutrient fortification or supplementa- ers 2001). Consistent with these � ndings, a tion, is often the most ef�cient response. study from Guatemala � nds that adult men South Asia has some of the highest rates of and women who were better nourished in malnutrition in the world as well as high lev- the � rst three years of life (having received els of anemia and iodine de�ciency. It has the nutritional supplements) had significantly highest prevalence of malnutrition in children higher schooling attainment, higher scores under �ve across all indicators (�gure 5.15).17 on cognitive tests, and greater likelihood of Indeed, malnutrition rates are higher than in FIGURE 5.15 Percentage of children under �ve with malnutrition, by region and country 70 60 59 50 47 48 49 47 42 44 42 42 40 39 40 38 percent 32 31 33 32 30 26 23 25 21 20 19 21 19 20 17 15 13 14 12 13 14 12 10 11 10 9 9 78 5 4 2 6 0 As bea d c Af and ld ca ka s an an h a l an sia pa ive i� di b n es or fri an ut ist ist Ea Cari ica a hA ac n a In Ne ad ald W ric nA rth st Bh an iL k dP Pa l ut No a ng M Sr r eE gh ra th Ame So an Ba ha Af dl ia Sa id tin e M b- La st Su stunting underweight wasting Source: Authors, based on data from World Bank 2011b. OPENING THE DOOR TO BE T TER JOBS BY IMPROVING EDUCATION AND SKILLS 191 Africa. Rates of malnutrition, anemia, and by programs in health and education that iodine de�cits decrease with wealth, but only take place in communities and preschools. modestly (annex �gure 5A.10). Most countries in South Asia do not Three conclusions can be drawn from this have integrated policy frameworks for early evidence. First, high malnutrition and micro- childhood development. Early childhood nutrient deficiencies are likely important interventions—nutrition, hygiene, early cog- contributors to developmental delays in low- nitive stimulation, preschool programs—are income groups in South Asia. Second, they among the most cost-effective investments to may also be important factors for overall cog- improve the quality and ef�ciency of basic nitive development of the broader population. education, as well as labor market outcomes. Third, income growth alone will not eliminate Table 5.3 provides an overview of the key malnutrition; focused attention is needed. issues for early childhood development policy Insuf�cient cognitive stimulation among and some types of programs that can help low-income children in South Asia may also prepare children for school. be a factor. Some evidence from Bangladesh National nutrition policies and programs in and Sindh (Pakistan) suggests that moth- Bangladesh and some aspects of public health ers’ limited understanding of when children campaigns related to nutrition in Pakistan respond to communication, for example, and have contributed to improvements in nutri- low levels of learning materials and involve- tion. Neither country has a national policy on ment could be contributing factors.18 In early childhood development, however. Bangladesh, parents have been found to be India has the region’s strongest enabling more interactive with their sons than their policy framework, which derives it founda- daughters, even at a very young age. Lack of tion from the constitution. Its centrally spon- time and maternal depression can also limit sored Integrated Child Development Services stimulation. (ICDS) program, which emanated from the Many children in South Asia come to Fifth Five-Year Plan (1969–79), remains the school with signi�cant disadvantages, which cornerstone for implementing its early child- hinder learning, contribute to early drop out, hood development policy. ICDS is one of and can be difficult or costly to reverse in the few programs at scale in the region that later years. Although outcomes and activities seeks to integrate early childhood interven- at home play a key role, they can be enhanced tions. The program—one of the largest TABLE 5.3 Main issues, interventions, and expected outcomes in early childhood development Key nutrition/early childhood Main issue/age development interventions Expected outcomes Lack of nutrition, health, and cognitive • Promotion of good nutritional practices, • Better physical well-being and health stimulation (0–3) including feeding and personal hygiene • Preschool readiness • Protein-energy malnutrition and deficiencies • Deworming and provision of micronutrients • Improved cognitive and noncognitive in key micronutrients for young children (vitamin A, zinc, iodine) development • Infectious diseases and parasites and their mothers (iron folate) • Provision of micronutrients through food fortification Lack of cognitive stimulation and parental • Parental enrichment training • Improved cognitive and noncognitive interaction (0–5) • Home visits in high-risk cases development Insufficient age-appropriate stimulation • Preschool programs with appropriate • School readiness and development of social skills (4–5) curricula • Improved cognitive and noncognitive achievement • Lower dropout Source: Alderman 2011. 192 MORE AND BETTER JOBS IN SOUTH ASIA child development programs in the world— Improving the quality of learning integrates health and nutrition services, early Improving the quality of learning in pri- stimulation and learning, and nonformal pre- mary and secondary school calls for mea- school education. Although well conceived sures that strengthen incentives and capacity as an approach to holistic child development, in the school system. Poor quality, particu- the program has not had the desired impact larly in public schools, can be attributed to on promoting cognitive development or several governance challenges. First, under- improving child nutrition. Inadequate cover- developed assessment systems limit monitor- age of children under three, an overempha- ing by stakeholders on what children learn sis on food supplementation, and substantial and the extent to which they master needed operational challenges (such as inadequate competencies. This lack of information is a program stewardship and delivery capacity, serious constraint to introducing reforms to weak monitoring and evaluation and supervi- improve quality. Second, with some notable sion, and the lack of community ownership) exceptions—such as the community-based have undermined its potential impact. There management of schools in Nepal— education is, however, wide variation in implementa- systems are still very centralized; decision- tion of the program across states, with some making authority, especially at the school states having demonstrated good results. Sev- level, is limited. Third, neither government eral localized innovations have the potential education management nor parents hold to be scaled up. schools accountable for education outcomes. In part, the absence of programs at scale The potential of efforts to increase local reflects still limited understanding of what autonomy and accountability has not been constitute cost-effective designs in South fully realized, as a result of limited capacity Asia. Several pilot projects can serve as a lab- and information of school and village man- oratory for designing cost-effective programs, agement committees. Fourth, � nancing of but many lack careful plans for evaluation. providers is not typically based on objective Potentially promising efforts are under way, measures of performance; it tends to be his- however. Pilots run by the International Cen- torical and based on inputs. Fifth, political ter for Diarrheal Disease Research (ICDDR) interference in teacher hiring and still under- in Dhaka, for example, have stronger evalua- developed career and pay incentives may be tion designs and have proven the feasibility of affecting teacher performance. This human promoting better parenting and mother-child resource management challenge extends interactions through home visits. The Lady beyond teachers to education management. Health Workers Programme in Pakistan, a Information, capacity, and incentives can community-based preventive care government also be important for private provision. The program, holds promise for promoting nutri- private sector is playing a signi�cant role in tion and child care through scaling up of care- providing even primary education in many fully evaluated cost-effective pilot designs. countries; it has already contributed to sig- nificant expansion in enrollments (figure Primary and secondary 5.16).19 Private provision offers the poten- education tial for increasing access at lower cost, often with comparable or better outcomes than This section begins by examining constraints public schools. to—and priorities for—strengthening pri- A study in two states in India shows that mary and secondary education systems in although private schools do not always achieve order to improve the quality of learning of higher learning outcomes than public schools graduates. It then discusses some innovative (especially after controlling for observable interventions that have had increased invest- differences in children) and achievement gaps ment in primary and secondary education of vary between states, private schools are more disadvantaged groups. cost-effective than public schools (Goyal and OPENING THE DOOR TO BE T TER JOBS BY IMPROVING EDUCATION AND SKILLS 193 FIGURE 5.16 Share of primary and secondary enrollments in public and private institutions in selected South Asian countries 100 1 1 2 2 1 2 0 1 0 1 1 5 5 6 6 7 7 7 8 6 10 15 14 20 17 25 25 25 18 30 24 20 34 80 17 15 60 74 73 percent 99 99 98 98 93 93 94 90 84 83 40 75 74 74 71 71 72 68 69 65 65 20 22 20 0 primary secondary primary secondary primary secondary primary secondary primary secondary primary secondary primary secondary primary secondary primary secondary primary secondary primary secondary 2008 2000 2010 2003 2007 2005 2010 1996 2004 2002 2008 Afghanistan Bangladesh Bhutan India Nepal Pakistan public private other Source: Authors, based on national labor force and household surveys. Note: “Other� includes private-aided schools in India; technical schools in Nepal; and madrasas in Afghanistan, Bangladesh, and Pakistan. Public schools in Nepal are, for the most part, under community management. Pandey 2009b). 20 A study in rural Punjab children attend private schools (Andrabi and (Pakistan) �nds that private school students’ others 2007). In Bangladesh, where second- test scores are 1.5–2.5 years ahead of their ary education is almost entirely provided by public school counterparts and that a sig- the private sector with government financ- ni�cant achievement gap remains even after ing, the majority of �nancing is on the basis controlling for the fact that more affluent of inputs, and serious concerns about quality 194 MORE AND BETTER JOBS IN SOUTH ASIA remain. Insuf�cient attention has been paid through their annual status of education in most South Asian countries to developing reports (Pratham 2010). quality assurance systems for the public and Providing information to local stakehold- private sectors and strengthening incentives ers on roles, responsibilities, and learning and capacity in the school system as a whole. outcomes can improve accountability and The global knowledge base on interven- outcomes in schools. Many countries (includ- tions to improve quality is still being built. ing Bangladesh, India, and Pakistan) are More is known on how to improve access and increasing efforts to address the limited infor- to include disadvantaged groups. An ongoing mation and capacity of school management regional study on the quality of education in committees and to increase the involvement South Asia will provide a more comprehen- of parents and communities. Rigorous evalu- sive look at the region’s education systems ations of some pilot interventions suggest that and a more in-depth discussion of interven- such efforts hold promise, although capacity tions to improve quality. Addressing gover- building requires sustained effort and results nance challenges will need to be at the center can take time. of all country strategies, as discussed below. In India, a study in 12 districts in 3 states evaluated the impact of an intervention that Improving governance and provided information to villagers about the accountability through information roles and responsibilities of school or vil- A key missing element in the accountabil- lage management committees. Although ity regimes in the region is data on student the magnitude and channels of impact (for learning outcomes. Only India, Pakistan, and example, increased knowledge, engage- Sri Lanka have national systems of student ment, and monitoring) varied across states, assessment, and even they are at the primary math competencies increased 10–30 percent level only. Only India has participated in an and teacher absenteeism fell 9–12 percent international assessment of student achieve- (Pandey, Goyal, and Sundararaman 2009). ment (two Indian states participated in the In Pakistan, school report cards were Programme for International Student Assess- distributed to parents and teachers, and ment [PISA 2009+], the results of which will discussion groups took place to ensure that be available at the end of 2011). Although all participants understand the information countries have an extensive system of public reported. Average test scores increased in examinations at the end of secondary edu- villages in which report cards were distrib- cation and often also at the end of primary uted. Increased local competition appeared education, the tests are usually not a good to have spurred the poorest-quality private basis for accountability.21 schools to improve (Andrabi, Das, and Even on a sample basis, national assess- Khwaja 2009). ments can yield important information about the strengths and weaknesses of the system, Strengthening school management which provides guidance for policy makers Most South Asian countries have started and other institutional actors (such as teacher devolving greater responsibilities to schools, training institutions) in designing interven- although the extent of the devolution varies tions. Establishing such assessment is a pri- across countries. Nepal has undertaken the ority area for education systems in South most sweeping reforms. Bangladesh, India, Asia. Countries with assessment systems are Pakistan, and Sri Lanka have all begun to working on strengthening them. India, for engage school committees in the school plan- example, is modernizing its system to bring it ning process, increase the availability of funds to international standards. Nongovernmental to schools, and strengthen the role of school organizations (NGOs) such as Pratham have leadership. Although the global evidence on played a key role in contributing to the edu- school-based management reforms is still cation debate in India (and now in Pakistan) limited, studies find that they can change OPENING THE DOOR TO BE T TER JOBS BY IMPROVING EDUCATION AND SKILLS 195 the dynamics of the school, because parents of any strategy to improve education out- become more engaged, teachers’ behaviors comes. South Asian countries have started change, or both. The studies document a pos- taking bold steps to reduce the politicization itive impact on repetition and failure rates, of recruitment and placement, improve the although evidence on learning outcomes is quality of teachers recruited, and increase mixed (Bruns, Filmer, and Patrinos 2011). motivation through various approaches, The evidence from initiatives in South such as incentive pay and the development Asian countries is promising. In Nepal, of career paths. where the devolution of responsibility for Moving away from patronage-based managing schools to communities is one recruitment to merit-based recruitment of of the most extensive in the world, positive teachers is key. Political interference in recruit- impacts include increased participation of ment and school postings reduces the qual- disadvantaged children and improved grade ity and accountability of teachers, affecting progression (see box 7.5 in chapter 7). Sri student attendance and learning. In the past Lanka’s Program for School Improvement few years, the region has made considerable empowers school management committees progress toward transparent and merit-based to develop and implement school improve- recruitment of teachers. Bangladesh developed ment plans. T he program, which has a teacher registration and certi�cation author- improved math and English language test ity, which administers examinations to certify scores of grade 4 students, is expected to be teachers, who are then assigned on the basis expanded nationally. of demand from schools. Pakistan introduced new policies in various provinces requiring Improving the quality and the use of transparent and merit-based criteria performance of teachers and school-speci�c contracts (box 5.1). Recruiting, motivating, and supporting the Some studies from South Asia suggest that right teachers have to be critical components well-designed schemes that provide incentives BOX 5.1 Recruiting teachers based on merit in Sindh, Pakistan Patronage-based hiring of teachers has long been prev- quali�cations, candidates who live close to the pro- alent throughout South Asia. In Sindh, Pakistan’s sec- posed placement, and women. Qualifying candi- ond most populous province, hiring has historically dates are then placed in understaffed schools under been patronage based, potentially compromising the school-speci�c and � xed-term contracts in an effort quality of incoming teachers and their accountability, to enhance accountability, reduce absenteeism, and as evidenced by rates of teacher absenteeism estimated place teachers according to need. as high as 25 percent. Teacher absenteeism further About 13,500 teachers have been hired over two undermines the regularity and quality of teaching, rounds of recruitment. The � rst round emphasized and, in turn, student attendance and learning. merit-based recruitment; the second round also To improve the quality of incoming teachers, the introduced needs-based placement. Both rounds government of Sindh is administering a program were validated by an external third party. Early � nd- for transparent merit-based recruitment. In a major ings show the rate of absenteeism of newly recruited shift from past recruitment policies and practices, teachers to be 16 percent, compared with 22 percent the government now � rst screens candidates through for regular teachers. A rigorous impact evaluation of an externally developed and administered content- the program is underway to determine the effects of based examination. Higher priority in placement is the program on teaching and learning quality. given to candidates who score higher on the entrance exam and have better academic and professional Source: World Bank staff reports. 196 MORE AND BETTER JOBS IN SOUTH ASIA through the nature of the contract or pay can Muralidharan and Sundararaman 2010). lead to learning gains. There are a few good These improved outcomes come at consider- studies from developing countries, but the able savings, as the cost of hiring teachers on knowledge base is still limited. Moreover, civil service terms and conditions is four to evidence on incentive pay varies across coun- seven times greater than the cost of recruiting tries and scheme designs (see annex table contract teachers (Goyal and Pandey 2009a). 5A.3). The evidence is still based on relatively Few countries have �gured out how to cre- short periods of time, with much less known ate career paths for contract teachers; doing on longer-term impacts. so will be important for ensuring the sustain- These caveats notwithstanding, some stud- ability of such policies. ies show promising results. The use of con- Evaluations from the states of Rajasthan tract teachers appears to have led to learning (Duflo, Hanna, and Ryan 2010) and Andhra gains for students in India, in part because Pradesh (Muralidharan and Sundararaman contract teachers exhibit more effective 2011) show positive impacts of teacher incen- work habits, such as better attendance and tives on attendance and student learning. greater effort (Duflo, Hanna, and Ryan 2010; Box 5.2 provides additional information on the BOX 5.2 Teacher incentives schemes in Andhra Pradesh In a statewide representative sample of 500 schools, schools). The main mechanism for the effectiveness four different approaches to improving learning of the teacher incentive program was not higher were tried. The schemes included two incentives teacher attendance in the incentive schools, but schemes (an individual teacher bonus and a group greater teacher effort conditional on attendance teacher bonus) and two input schemes (provision (teachers report that they assigned more homework, of an additional contract teacher and provision of held extra classes, gave practice tests, and helped a block grant to the school) as well as a compari- weaker students). son group of 100 schools. All schools, including A f ter t wo years, the individual incentive the control schools, were monitored, with students schools performed better than the group incentive tested at the beginning and end of the school year. schools. Averaged across subjects, students in indi- Teacher bonuses were awarded on the basis of aver- vidual incentive schools scored 0.28 standard devia- age improvement of student test scores over each tions higher than the control group, while students school year. In the group incentive schools, bonuses in group incentive schools scored 0.16 standard devi- were awarded on the basis of the average gains of ations higher. However, because the bonuses were all students in the school, and were equally distrib- paid as a function of actual improvement in test uted among teachers. In the individual incentive scores, the bonuses paid out under the group incen- schools, the bonuses were awarded to each teacher tive program were lower, and so the cost per unit on the basis of the average gains in scores of students of increase in test scores of the two programs was taught by that particular teacher. similar (though the individual incentive program After two years, all four schemes improved stu- was more cost effective after including � xed admin- dent learning. However, the schools receiving the istrative costs). teacher incentive programs improved their test scores Although there was no incentive for performance in by signi�cantly more than those receiving additional science and social studies, scores in these subjects also inputs. Students in incentive schools scored 0.27 rose. It seems likely that weak literacy and math skills standard deviations higher in math, and 0.16 stan- were preventing students from doing well in other sub- dard deviations higher in language than students jects, and the results suggest positive spillovers from in the control schools (while the students in input gains in math and language to other subjects. schools scored 0.09 and 0.08 standard deviations higher in math and language than those in control Source: Murlidharan and Sundararaman 2011. OPENING THE DOOR TO BE T TER JOBS BY IMPROVING EDUCATION AND SKILLS 197 intervention in Andhra Pradesh and the rela- Increasing demand from households tive cost-effectiveness of various interventions. As the demand for more educated workers Countries in South Asia are taking steps increases, improving parents’ understand- to improve incentives and performance ing of the returns to education can lead to through capacity building linked to career increased investment in the education of their progression of teachers. Although this is still children. In a study of three states in India, an underdeveloped area of reform in South Oster and Millet (2011) show that opening Asia, recent initiatives by several governments an information technology–enabled service suggest that they recognize the importance of center increased knowledge of improved job this issue. For example, to address the serious opportunities and perceived returns from de�ciencies in teacher education and profes- higher schooling, leading to a 5.7 percent sional development, the Sindh government increase in the number of children enrolled is gradually implementing a comprehensive in a primary school. They find that the teacher education development program. The very localized effects were a result of lim- program envisions an integrated and coher- ited information diffusion outside the areas ent system for preservice and systematic in- where the call centers were opened, suggest- service training (continuous professional ing that policies that improve understanding development) linked to entry and career pro- of job opportunities could play an impor- gression. Reforms in Nepal are focusing on tant role in increasing enrollments. Consis- developing separate professional career paths tent with this � nding, an intervention that for basic and secondary-level teachers and provided three years of recruiting services provision of preparatory and in-service train- for business process outsourcing opportuni- ing. The plan is to link career progression to ties for women in randomly selected rural pedagogical effort/time, seniority, quali�ca- villages in India was associated with large tion, training, and student achievement. increases in enrollment as well as subsequent employment (Jensen 2010). Reducing the costs of schooling or other- Increasing human capital investment wise reducing household income constraints in disadvantaged groups can also contribute to increased participa- In addition to strengthening the school sys- tion. In Afghanistan, and Pakistan, reducing tem to improve the quality of learning, gov- indirect costs by reducing distance to school ernments have a role to play in addressing has increased enrollment, especially for girls. market failures that lead to lower investment In Afghanistan, enrollment of girls in com- in human capital in disadvantaged groups. munities randomly selected to receive schools On the demand side, inadequate information increased by 15 percent more than overall and budget constraints are two reasons for enrollment (a 50 percent increase for all stu- low investment in human capital among low- dents versus a 65 percent increase for girls) income households. There is some evidence (Burde and Linden 2010). After six months, that inaccurate perceptions on returns reduces girls in villages with community schools also investment in education. Direct and indirect reported far higher test scores than boys. costs associated with schooling also reduce Around the world and in South Asia, investment (distance to schools is a problem conditional cash transfer programs, which for girls in Afghanistan and Pakistan). On the help households address budget constraints, supply side, in many remote and rural parts have a strong record in increasing enrollment of South Asia, latent demand for schooling (Fishbein and Schady 2009). Enrollment is going unmet in underserved communities. and attendance of girls in secondary schools The following sections discuss some innova- increased substantially in Bangladesh after tive interventions on the demand and supply the government introduced cash stipends to sides that have shown promise in increasing rural girls under the Female Secondary School participation among disadvantaged groups. Stipend (FSSS) program in the 1990s. Female 198 MORE AND BETTER JOBS IN SOUTH ASIA enrollment rates in lower secondary (grades areas, they are less likely to locate in the most 6–10) rose from about 25 percent in 1991 to underserved areas or to serve the poorest about 60 percent by 2008, leading to gender groups. Governments in the region are there- parity at this level. A follow-up program is fore innovating with programs that use subsi- targeting poor children. Early evaluation of dies to address this market failure. this program shows an estimated average net A PPP program in Punjab (Pakistan) pro- increase in retention of 20 percentage points vides per student subsidies to low-cost private (Asadullah and others 2011). schools for enrolling and raising learning lev- In Pakistan, both Sindh and Punjab prov- els among children from low-income house- inces have large-scale conditional cash trans- holds. An evaluation of the program finds fer programs targeting girls in secondary large effects (of 0.3–0.5 standard deviation) grades in government schools. Evaluations of on student learning—far greater than the the program in Punjab � nd enrollment gains 0.1–0.2 standard deviation gains from most (of about 9 percent), delayed marriage and interventions that have been evaluated in the pregnancy, and no negative spillover effects literature—across two different tests, sug- on boys’ enrollment (Adam, Baez, and del gesting that teaching to the test is likely not Carpio 2011; Chaudury and Parajuli 2010). at play (Barrera-Osorio and Raju 2010). In Sindh, Pakistan, a pilot program that offers Providing accountability-based public per student subsidies to support the subsidies to private providers set-up and operation of private coeducational Public-private partnerships (PPPs) are prov- primary schools in underserved rural com- ing to be a cost-effective means of reaching munities increased school participation by underserved children. Although low-cost pri- 51 percentage points in program communi- vate schools are emerging, including in rural ties. The growth in school participation was BOX 5.3 The Reaching Out-of-School Children project in Bangladesh The Reaching Out-of-School Children (ROSC) students, with support from NGOs as implementation project has increased access to learning opportu- partners. Operational program delivery differs from nity for out of school children ages 7–14 in about the norms in primary schools in order to cater to spe- 60 targeted upazilas (subdistricts). It combines ci�c needs of students (through, for example, flexible accountability-based subsidies to providers (learn- school timing, multigrade teaching, and entry at an ing centers) with demand-side � nancing of students older age). (stipends). Key features include public financing of Initial findings from an independent impact stipends and grants to learning centers, community evaluation � nd that ROSC has succeeded in target- management of learning centers, and public-private ing intended bene�ciaries (children from the poorest partnerships (of community, government, and NGOs) households); reducing the proportion of out-of-school in implementation. Grants to learning centers are children; and improving learning levels, as measured provided on a per capita basis, and continuation by student test scores in math and language (Bangla). of financing of learning centers as well as stipends Compared with the regular primary schooling sys- is contingent on achieving minimum levels of stu- tem, the ROSC project is achieving similar outcomes dent attendance and performance. The community- in access and equity at less than half the per student driven approach focuses on establishing learning spending, implying impressive cost-effectiveness. centers through a functional center management committee that is directly accountable to parents and Source: Asadullah and others 2011. OPENING THE DOOR TO BE T TER JOBS BY IMPROVING EDUCATION AND SKILLS 199 larger for girls than for boys, eliminating the two systems are different, as is the focus of gender gap. The cost for this intervention of the knowledge and technical skills, many inducing a 1 percent increase in school par- of the challenges are similar. Three coun- ticipation lies at the bottom of the range of tries in the region—Bangladesh, India, and estimates for interventions in developing Pakistan—also have small vocational edu- countries subject to rigorous evaluations (Bar- cation streams within the secondary school rera-Osorio and others 2011). Early results system (box 5.4). indicate learning gains of two standard devi- Policy makers are keen to expand the small ations. Bangladesh has also demonstrated the preemployment training and tertiary educa- potential of innovative partnerships with the tion systems. Participation in preemployment private sector (box 5.3). training ranges from about 1.0 percent of secondary enrollment in Afghanistan to 8.1 percent in Sri Lanka. India ranks second in Tertiary education and the region, with 4.5 percent (see annex �gure preemployment training 5A.4). Gross enrollment rates for tertiary edu- systems cation range from 4 percent in Afghanistan to Preemployment vocational training provides about 13 percent in India and 12–21 percent an alternative for students who do not pursue in Sri Lanka (depending on whether exter- tertiary education. It is provided in institu- nal degree programs are included) (see annex tions outside the general schooling system, �gure 5A.5). In most South Asian countries, often targeting early school leavers with policy makers see expansion of these sectors the objective of directly preparing partici- as an important element of their strategy pants for the labor market. Preemployment for increasing the skills of the large number training includes formal accredited courses of new entrants into the labor force. Bhutan and can be part of a national quali�cation seeks to increase the number of its preemploy- framework requiring a minimum entry qual- ment training graduates by a factor of 12 by i�cation. It is typically provided in special- 2013. India is targeting training 500 million ized training centers, training colleges, or people over the next 10 years through a com- polytechnics. bination of preemployment training, in-service Minimum entry quali�cations vary across training, and tertiary education (Government countries. In most countries, premploy- of India 2009). Nepal and Pakistan are also ment training requires completion of grade signi�cantly increasing their investments in 10; it starts after grade 8 in Bangladesh and preemployment training. All South Asian Pakistan and after grade 9 in Afghanistan countries are targeting substantial expansion and Sri Lanka. Depending on entry quali�ca- of their tertiary education systems. tions, courses can range from less than a year Expansion should not come at the to five years, with multiple entry and exit expense of strengthening the systems for points. Courses range from those aimed at better outcomes. As with concerns about future tradespeople and technicians to those skills of graduates of tertiary education dis- providing higher levels of skills. cussed earlier, there are concerns with the Tertiary education refers to academic and quality and relevance of preemployment professional degree and diploma courses pro- training. A 2006/07 tracer study of labor vided by universities, colleges, and profes- market outcomes for graduates of preemploy- sional schools (such as medical schools). The ment training in Bangladesh � nds that only minimum entry quali�cation is completion of 9.7 percent of male participants and just 5.2 higher-secondary education or its equivalent. percent of female participants were employed This section discusses the challenges and (World Bank 2008). A tracer study conducted priorities for strengthening preemployment by the International Labour Organization training and tertiary education systems. in three states in India in 2002/03 notes Although the students targeted by these very low employment of graduates of the 200 MORE AND BETTER JOBS IN SOUTH ASIA BOX 5.4 Vocational education provided in the public school system In Bangladesh, India, and Pakistan, students who Expanding or retargeting secondary vocational do not fare well in the general (academic) education programs is not justi�ed unless a model is found that stream are often tracked into vocational education substantially improves outcomes. One such model is streams, typically at the senior-secondary level (and Chile’s Corporación de Desarollo Social del Sector sometimes earlier). Vocational schools are supposed Rural (CODESSER) program. CODESSER is a non- to equip students with skills that will allow them to pro�t organization created by the National Society of join the labor market or preemployment training Farmers in 1976. Initially, it administered four schools programs. with weak reputations that found it dif�cult to attract The experience with secondary vocational pro- students. Today, some of these schools receive more grams in South Asia, as in many other regions, has than 300 applications for 45 � rst-year openings, and been poor. In India, for example, studies show that schools have been added to accommodate growing 60–70 percent of graduates fail to obtain employ- demand. Recent �gures show that more than 75 per- ment even two to three years after graduation cent of graduates from agricultural schools hold mid (World Bank 2007). level management positions in agriculture. Despite these poor outcomes, policy makers CODESSER has the following features, which remain keen to expand vocational secondary educa- embody some of the principles for reforming the tion. Bangladesh has targeted 20 percent and India vocational education system: 25 percent of secondary education to be vocational. • Private sector players participate in management. There is also a desire to make general secondary A directorate of seven farmers or industrial entre- education more vocational. As in other parts of the preneurs oversees each school. Their participation world, this desire is motivated by the concern that ensures better job skill matches, a direct connec- large numbers of students complete primary and tion to the labor market for graduates, and an secondary education with no occupational skills effective medium for bringing about organiza- and the assumption that having occupational skills tional and productive innovations. improves prospects in the labor market. • Teachers are hired as private sector employees. Such strategies may not be appropriate for the Personnel policy (including selection and promo- following reasons: tion criteria and new contracts) conforms to the • Vocationalizing education is difficult. It requi- labor code that regulates private sector employees. res specially trained instructors, preferably with Teachers’ salaries are about 50 percent higher than work experience in the types of skills being taught. in municipal schools. There has been a consistent Teachers with these quali�cations are hard to recruit effort to upgrade teacher training. and retain. • The focus is on general education. Schools provide • Vocationalizing education is costly. Most variants basic general knowledge in humanities and scien- of vocational training cost more per student class ces, prepare students to work in various occupa- period than general education subjects, primarily tions, teach students to be problem solvers, and because classes are smaller and facilities, equip- encourage them to continue learning. They empha- ment, and consumables cost more. size general growth and the development of respon- • Private sector involvement in running vocational sibility, leadership, and personnel management. education systems, setting course content and • The curriculum is regularly revised. CODESSER curricula, or managing vocational schools is very conducts periodic surveys of job requirements in limited. the areas around each school. The surveys are • Time spent on vocational skills training can detract used to adjust vocation-speci�c components in the from the teaching of basic academic skills, which curriculum. provide the foundation for future learning and are • Schools developed independent sources of funding. increasingly essential for labor market success. The real value of public subsidies fell in the early 1980s; it declined by another 15 percent between A better alternative to expanding vocational train- 1987 and 1991. In 1982, the public subsidy repre- ing in the school system would be to make general sented the bulk of schools’ budgets; it is now less secondary schools more relevant and better able to than 50 percent. cater to a wider range of students. Source: Chhoeda, Dar, and Tan 2010. OPENING THE DOOR TO BE T TER JOBS BY IMPROVING EDUCATION AND SKILLS 201 preemployment training system (ILO 2003). FIGURE 5.17 Percentage of graduates from public and private According to the study, formal sector employ- industrial training institutions employed in the organized sector in ment of graduates from both public and pri- three states of India, 2002/03 vate industrial training institutions was very 100 low. In none of the states did more than half 95 90 of graduates �nd wage employment, become 85 self-employed, or work in a family business 80 75 (�gure 5.17). 70 65 Pressure to expand will continue, as more 60 percent and more students complete secondary edu- 55 50 cation. A key priority for government is to 45 40 strengthen systems to improve quality while 35 leveraging partnerships with a variety of 30 25 stakeholders. Also important will be address- 20 15 ing market failures that prevent children from 10 low-income households from progressing to 5 0 and graduating from higher levels of educa- Orissa Andhra Pradesh Maharashtra tion. In what follows, some of the key chal- industrial training institution industrial training center lenges facing these systems and approaches to Source: ILO 2003. addressing them are examined. the stewardship role, setting policy, ensuring coordination between ministries, oversee- Strengthening policy and regulatory ing � nancing, developing curriculum, han- oversight dling certification and accreditation, and Fragmented system management blurs the providing information on the quality and roles of various agencies and reduces policy effectiveness of institutions. Critical to the coherence. Responsibility for policy direc- success of such a body is the role of employ- tion, leadership, and regulation of and ers. Unless they are given a prominent role accountability for implementation is often in the policy and decision-making process, diffused across different levels of govern- provision may not be adjusted suf�ciently to ment or multiple ministries and agencies. In meet their needs. Bangladesh, for example, both the Ministry Australia provides a successful example for of Education and the Ministry of Labor are preemployment training and tertiary educa- responsible for preemployment training, and tion (box 5.5). Some interesting initiatives are about 10 other ministries run training insti- also underway in tertiary education in South tutes. Financing and regulation of tertiary Asia. If implemented well, for example, estab- education institutions in medicine is also the lishment of the National Council of Higher domain of multiple agencies. Education and Research (NCHER) could Two main problems result from this improve policy coordination and regulation administrative overlap. The � rst is a lack of of tertiary education and research in India. clarity and duplication of functions. The sec- The existing tertiary education bodies—the ond is a tendency to emphasize the delivery University Grants Commission, the All India of training by multiple agencies at the cost of Council of Technical Education, the National ensuring quality standards. It is important Council of Teacher Education, the Distance for governments to transition from a preoc- Education Council, and related academic cupation with provision and � nancing to a councils—are expected to come under the stronger stewardship role. umbrella of the NCHER. There is a need for governments to take In the medium term, the establishment on a stronger stewardship role and clarify of a national quali�cations framework can the functions of various agencies. It may be make programs more coherent and clarify appropriate to entrust one body with playing the roles of various players. Sri Lanka has a 202 MORE AND BETTER JOBS IN SOUTH ASIA BOX 5.5 Learning from Australia’s systemic reforms of training and tertiary education Australia’s federal and state governments created the education and preemployment training (while res- Australian National Training Authority (ANTA) in pecting the mission of each sector) the 1990s. The agency was overhauled over time. • Funding the national training system On July 1, 2009, the government formed a • Maintaining strong industry leadership of and Ministerial Council for Tertiary Education and engagement in preemployment training and ter- Employment (MCTEE) to oversee skills develop- tiary education ment. Membership in the council includes federal, • Overseeing the work of the Australian Qualifi- state, and territory ministers with responsibility for cations Framework Council in strengthening the tertiary education, preemployment training, and Australian Quali�cations Framework employment. MCTEE is chaired by the federal gov- • Engaging with all associated bodies, councils, and ernment minister. ministerial committees to ensure the effectiveness MCTEE has overall responsibility for skills devel- of policy and strategies for tertiary education and opment. Its functions include the following: preemployment training • Responding to business, industry, and stakeholder • Setting national priorities and the strategic policy advice on issues affecting tertiary education and directions for the tertiary education sector and preemployment training preemployment training to meet the skills needs of • Setting national research priorities for skills the Australian economy development. • Establishing streamlined arrangements for natio- nal consistency and harmonization of tertiary Source: Skills Australia 2011. quali�cations framework for preemployment A key role for governments is the pro- training (the National Vocational Qualifi- vision of information and the strength- cations Framework), which is being imple- ening of quality assurance mechanisms. mented. The government is also working on Limited information, especially on quality a full national quali�cations framework that and outcomes, is an important deterrent to will cover primary, secondary, and tertiary improving effectiveness—and to students education, as well as preemployment train- and employers making the right choices. ing. India plans to establish a standards- Limited information about privately pro- based national vocational qualifications vided programs can lead to public sector framework that will permit individuals who crowding out private providers by enter- follow different learning pathways to cer- ing areas where there is already a private tify their skills through testing and facilitate supply. To address these issues, most coun- advancement to higher diplomas and degrees. tries in the region have started establish- Taking advantage of a still nascent tertiary ing education management information education and training system, the govern- systems in tertiary education to provide ment of Afghanistan is setting up a national information on participation and comple- qualifications authority and a national tion, transition from college to work, and qualifications framework with which to � nance. Bangladesh, India, and Nepal have better coordinate and articulate between begun to work on establishing informa- the general education, tertiary education, tion systems for preemployment training. and training systems, as well as between the It is important that these systems collect various players responsible for implement- and disseminate information on outcomes ing tertiary education and preemployment (such as the quality of graduates and their training. employability). Toward this end, some OPENING THE DOOR TO BE T TER JOBS BY IMPROVING EDUCATION AND SKILLS 203 countries have begun to conduct employer bodies of tertiary education institutions in surveys and tracer studies of graduates engineering. It has established quali�cation (in India and Sri Lanka), and student and criteria for their selection and mandated faculty surveys (in Bangladesh and India). greater transparency in the selection process South Asian countries have also begun (Government of Maharashtra 2009; Govern- developing quality assurance and accredita- ment of India 2010). Private sector involve- tion systems in tertiary education. Although ment in institutional management of public there is still a long way to go, and a need to tertiary education and training institutions increase transparency and strengthen capac- is also increasing. Industry associations and ity of accreditation agencies, the region has individual employers are increasingly repre- been moving toward developing these sys- sented on governing boards. In India, the two tems. India has a formal accreditation sys- largest industrial associations—the Federa- tem, but only 15 percent of its more than tion of Indian Chambers of Commerce and 20,000 institutions are accredited (Agarwal Industry and the Confederation of Indian 2009). Nepal is in the process of establish- Industries—are represented and actively par- ing an independent quality assurance and ticipate in the management committees of accreditation council/board. Pakistan estab- tertiary education and training institutions. lished the Quality Assurance Agency in 2005 In Sindh (Pakistan), private sector representa- as a monitoring and regulatory body to facili- tives now head the institutional management tate quality assurance in both public and committees of training institutions. Bangla- private institutions. Nine accreditation coun- desh and Nepal are moving in this direction. cils have accredited 60 programs, and qual- Involving employers in management will ity enhancement cells are being established yield positive results only if governments are within institutions to undertake internal willing to provide institutions with greater quality assurance. Sri Lanka is developing a autonomy. This means letting institutions quality assurance review process covering 60 administer themselves and keep the funds percent of all universities. they raise through fees and other sources. Administrative regulations and budgeting rules provide scant incentives for ef�ciency Strengthening institutional governance and innovation. Possibly to counter the lack and accountability of local oversight, many governments in the Politicization and insufficient representa- region have highly speci�c rules on hiring, tion of key stakeholders in governing bodies �ring, and rewarding faculty (administrative undermines institutional accountability and autonomy) and managing revenues, includ- effectiveness. Vice chancellors of public uni- ing fee setting and expenditures (financial versities are often politically appointed. In autonomy). In Afghanistan, Bangladesh, addition, many colleges and universities in and several states of India, teachers are hired South Asia do not have a board of governors centrally, removing the power of institu- with representation of key stakeholders. tional leaders to hire or promote teachers. In This failure undermines the transparency of Afghanistan, all revenue generated by univer- decision-making process and accountability sities, including fees, has to be returned to the to bene�ciaries, particularly students, par- Ministry of Finance. ents, and employers. Governing bodies of Elements of autonomy are also being intro- preemployment training institutions are less duced in the region. In India, public training politicized. A key omission in both systems institutions and engineering colleges, in con- is the lack of representation of the private sultation with employers, decide on their own sector in the governing bodies. training programs, have the freedom to hire Some countries have initiated reforms in part-time teachers, and retain part of the rev- this regard. India has begun to ban politi- enues generated by selling goods and services. cians from serving as chairs of the governing Nepal’s tertiary education system is also 204 MORE AND BETTER JOBS IN SOUTH ASIA moving in this direction, albeit slowly. The Since about 2000, several governments in move toward greater autonomy and account- South Asia have laid a cornerstone by allocat- ability needs to be accompanied by measures ing some tertiary education funding through to increase awareness among and strengthen competitive or innovation funds. These funds the capacity of members of oversight commit- transparently � nance investments in public tees and governing boards. and private tertiary education based on objec- In most cases, funding mechanisms still tive policy goals, such as improved quality and provide limited incentives to institutions to relevance of undergraduate programs. One improve performance. As countries develop, example is the Quality Enhancement Fund for funding mechanisms for tertiary education Higher Education in Sri Lanka, launched in gradually shift from incremental budgeting 2005, which is progressively being strength- (historically negotiated budgets) to perfor- ened based on experience. mance-based approaches (table 5.4). This Financing systems for training are also transition has not yet occurred in South Asia. seeing reforms. Bangladesh and Nepal are Training institutions are typically guaran- introducing performance grants to training teed continued funding irrespective of per- institutions that meet agreed upon perfor- formance, with levels of finance allocated mance criteria (such as the pass rate for stu- to poorly performing institutions with high dents or the employment rates of students six drop-out rates equal to those allocated to months after graduation). In Pakistan, contin- institutions that maintain a high quality of ued funding under the Benazir Bhutto Sha- teaching and performance. heed Youth Development Program (BBSYDP) TABLE 5.4 Mechanisms for funding tertiary education in South and East Asia, by economy Historically Formula Competitive Performance-based Economy negotiated budget funding funding contracts High-income Hong Kong SAR, China • Japan • • Korea, Rep. • • Singapore • • Middle-income China • • Indonesia • a • Malaysia • • Philippines • • Thailand • • Low-income Cambodia • Lao PDR • Vietnam • • South Asia Bangladesh • b India • b Nepal • b Pakistan • b Sri Lanka • • Sources: Authors, based on data on East Asia from World Bank 2011a and data on South Asia from sector staff. Note: a = Formula funding incorporating performance indicators has been piloted in �ve autonomous universities (and some nonautonomous universities). b = Government projects �nanced by the World Bank have started competitive funding for selected institutions and disciplines. OPENING THE DOOR TO BE T TER JOBS BY IMPROVING EDUCATION AND SKILLS 205 now requires training providers to achieve as key constraints to setting up new and minimum levels of job placement. upgrading existing institutions. 22 Although not uniform across countries, other concerns include excessive barriers to entry and gov- Increasing the role of the private sector ernment bureaucracy in registering institu- Private provision of education is signi�cant tions and accrediting and certifying courses. in South Asia, particularly in preemployment International experience suggests that training, and is growing. Enrollment in pri- when private providers are not discouraged by vate preemployment institutions ranges from stringent regulations, rapid industrial growth 6 percent of total enrollment in Afghanistan can be accompanied by strong private sector to 82 percent in Nepal; in tertiary education, response in supply. Streamlining the process the �gures range from less than 5 percent in of registration/licensing based on clear merit- Afghanistan and Bhutan to more than 50 based criteria is critical. Imposing strict con- percent in Bangladesh (figure 5.18). Even sequences for malpractice by bureaucrats can in Afghanistan, where formal institutes are also help improve the environment for private still very much under government domain, providers. In India, where the private sector strong private and NGO–based initiatives is more heavily engaged in training in com- are developing. mercial areas than in other countries, up to Easing barriers to entry and leveling the 15 percent of all providers now provide tech- playing �eld can generate a stronger private nical training, indicative of private sector sector response. Private providers in some response despite the higher initial investment South Asian countries identify lack of access costs for technical fields. Enabling private to credit and �nancing of initial investments institutions to compete for public resources FIGURE 5.18 Share of preemployment training and tertiary education enrollment in public and private institutions in South Asia, by country 100 3 0 1 4 6 12 24 20 25 80 32 54 57 56 60 26 80 77 82 percent 96 100 94 88 40 76 80 67 43 48 20 43 44 20 23 18 0 20 tan 20 esh 20 tan 20 dia 20 pal 20 tan 20 nka 20 tan 20 esh 20 tan 20 dia 20 pal 20 tan 20 nka In In Ne Ne 08 10 08 10 04 08 07 Ba 08 08 09 05 03 08 08 is u kis is u kis lad lad a a Bh Bh iL iL an an Pa Pa ng ng Sr Sr gh gh Ba Af Af tertiary education vocational education and training public private other Sources: Authors, based on data from Chhoeda, Dan, and Tar 2010 for vocational education and training; labor force and household surveys for tertiary education for all countries except Bhutan and Sri Lanka; Thinley 2009 for Bhutan; World Bank 2009 for Sri Lanka. Note: “Other� includes private-aided schools in India and madrasas in Bangladesh and Pakistan. For Sri Lanka, enrollment in external degree programs—which accounts for 58 percent of total tertiary enrollment—is included in public enrollment. If shares in the public and private sectors are computed excluding enrollment in external degree programs, private enrollment represents 28 percent of total tertiary enrollment. 206 MORE AND BETTER JOBS IN SOUTH ASIA on a level playing �eld, particularly in prior- be possible only by diversifying the resource ity education and training areas, could reduce base. barriers. The recent trend toward competitive One important mechanism is tuition, funding is a promising step in this direction. which is being increased in public tertiary PPP arrangements are just being initiated and preemployment training institutions in the region. Potential models include the pri- in Bangladesh, India, Nepal, and Pakistan. vate sector investing in and managing defunct The much higher fees being paid at private public institutions as a concession or setting institutions demonstrates both the demand up new institutions in underserved areas with for services and the willingness of better-off public subsidies. Early experience suggests sev- segments to pay. Governments can offset the eral lessons for making PPPs work, including adverse impact of such fees on equity by pro- the importance of sound project fundamen- viding targeted scholarships to low-income tals; attention to the timing, scope, and dura- groups, as Bangladesh, India, Nepal (box tion of transaction; high-quality advisors; a 5.6), and Pakistan are doing. Student loans strong government implementation team; and can help ease liquidity constraints. Student transparency throughout the process (based loan schemes for tertiary education have been on unpublished presentation by the Interna- implemented in India and are being developed tional Finance Corporation 2010). in Pakistan. Diversifying the resource base On-the-job training Until recently, little attention was paid to Formal on-the-job training can be an impor- exploring ways to enhance the resource pool tant channel through which workers can to finance expansion and improvements. build human capital. Although this chapter Throughout the world, the unit costs of both focuses primarily on education and training preemployment training and tertiary educa- that takes place before entry into the labor tion are significantly higher than those of force, a significant part of human capital general education. In the face of competing accumulation takes place in the work place. demands and limited fiscal space, signifi- As analysis of business enterprise data in cant expansion and enhancing quality will South Asia reveals, � rms that innovate, are BOX 5.6 Providing scholarships for tertiary education in Nepal In 2003/04, only about 1 percent of children in example, types of walls, roof, cook stove, and toi- each of the two poorest quintiles were enrolled in let; number of siblings; parent education; and caste tertiary education compared to 88 percent in the and ethnic group). All household characteristics are richest quintile. To address the problem, in 2009 the observables; information can therefore be veri� ed government initiated a scheme to provide � nancial through household visits to prospective bene� cia- support for meritorious and needy students through ries. The calculation gives preference to disadvan- scholarships. These scholarships � nance education taged castes and ethnic groups. A minimum share of and living expenditures up to Rs 40,000 ($530) for bene�ciaries must be women. As of February 2010, the duration of study. 176 undergraduate students had received benefits The crucial innovation of the scheme is proxy under this scheme. means testing. A poverty score is calculated based on a set of observable household characteristics (for Sources: Nepal Living Standards Survey 2003/04 and World Bank 2010. OPENING THE DOOR TO BE T TER JOBS BY IMPROVING EDUCATION AND SKILLS 207 internationally engaged, and have a more region (�gure 5.19). Across the world, small educated workforce are also more likely to and medium-size �rms are less likely to train, invest in the training of their workforce. but even holding �rm size constant, the inci- This finding is consistent with the strong dence of training in South Asia is relatively links found in the international literature low. In East Asia and Latin America and the between on-the-job training and individuals’ Caribbean, for example, more than 70 per- and � rms’ productivity. Despite econometric cent of the largest �rms and 30 percent of the challenges associated with identifying the smallest �rms provide training to their work- impact of on-the-job training on wages and ers. In South Asia, less than 40 percent of the firm productivity (see Almeida, Behrman, largest firms train their workers, and only and Robalino 2011 for a review of the evi- about 5 percent of the smallest �rms do so. dence), there is convincing evidence, par- Training incidence also varies with ticularly from developed countries, of the worker characteristics. It is lowest among positive effects of training. It may be the case workers with less schooling, confirming that on-the-job training may be more criti- complementarities between investments in cal in producing skilled workers in develop- general schooling and on-the-job training. ing countries than in developed countries, in Evidence from India also shows distinct dif- view of the considerable skills gaps of gradu- ferences across industries in their propen- ates of education and preemployment train- sity to offer formal in-service training, with ing systems in these countries. industries such as textiles, garments, leather The incidence of formal on-the-job train- products, food processing, automobile parts, ing is lower in South Asia than in any other and metal products less likely to train and FIGURE 5.19 Percentage of �rms providing on-the-job training in regions and selected South Asian countries, by �rm size 100 86 80 76 74 74 66 65 60 55 53 53 percent 51 48 43 43 39 40 40 35 34 30 30 32 31 27 27 29 25 26 21 20 16 16 17 16 14 12 13 7 9 6 4 3 0 0 Af ran ci� a lA d be d Af and sia 20 esh 20 dia 20 tan 20 nka Pa si ra an ib an d tA hA In sia a c an a 02 06 02 04 ha kis lad a ric ric nt e ar ca rth st iL an Eas Ce rop Sa Pa ut No e Ea e C ri ng Sr th Ame So b- Eu Ba dl Su id tin M La < 10 11–50 50–250 > 250 Source: World Bank staff, based on establishment survey data compiled by Almeida and Aterido 2010. 208 MORE AND BETTER JOBS IN SOUTH ASIA industries such as chemicals, pharmaceuti- poaching trained employees could also lead cals, machine tools, electrical white goods, to suboptimal investments. Policies should electronic products, and software more be designed to address the primary source of likely to do so. In-house programs (offered market failure. by 23 percent of all � rms) tend to be more International evidence suggests three main common than external programs (offered by reasons why employers do not train workers. 14 percent of � rms). This evidence is consis- Responses of managers in enterprise surveys tent with the notion that workforce training suggest that in the majority of � rms, formal can complement general skills of workers, training may not be needed (informal training with training tailored to the production and is considered suf�cient or the technology does technological requirements of individual not require training) (Almeida, Behrman, enterprises (Riboud, Savchenko, and Tan and Robalino 2011; Batra and Stone 2004). 2007). High worker turnover or fear of poaching (or The low incidence of formal training does the ability to hire workers from elsewhere), not necessarily imply the need for public inter- lack of financing, and information barri- vention. For many �rms in developing coun- ers also appear to be factors. Improving the tries, investment in on-the-job training may investment climate should improve incentives not be necessary, given technological choices, for employers to invest in both physical and the skills components of most of their jobs, human capital. Policies that improve access to or their reliance on informal training. Other new technologies and funding for investments firms, however, may face severe informa- in technology and skills upgrading would tional constraints in, for example, identifying also increase �rms’ demand for training. relevant training programs, recruiting train- PPPs and collective industry action can ing experts, or � nancing programs. Market strengthen the in-house training capabilities failures associated with the risk of other �rms of �rms. Firms, industry associations, buyers BOX 5.7 Industry-government cooperation: The Penang Skills Development Centre The Penang Skills Development Centre (PSDC) is shared training facilities without having to duplicate a joint corporate training center established with their in-house capability. Small and medium-size government support. It has evolved to become a enterprises enjoy technology transfer from and can full private continuing education institution pro- benchmark their standards against multination- viding certificate and diploma-level training. The als. Multinationals, in turn, receive better support Malaysian government invests in the center and uses services. Vendors donate equipment to familiarize it to conduct public training programs. workers with their products and promote sales. With more than 100 member companies, the Although still a joint training center, the PSDC PSDC runs both standardized and customized pro- has taken on institutional functions to provide train- grams. It charges at cost and is largely self-� nancing. ing for school-leavers, thus moving toward becom- Companies can recoup their expenditure from the ing a professional training entity in its own right. Human Resource Development Fund, a training The PSDC is not easily replicated. Eleven of fund � nanced by a 1 percent levy on payrolls. Malaysia’s 16 states and federal territories have The initiative for the PSDC came from the launched similar centers, with varying degrees Penang state government, which provided the land of success. The success of this sort of enterprise and buildings. The founding members included large depends on a number of circumstances, most impor- multinational companies with training traditions of tant among them the people initiating and managing their own. Members donate equipment, laboratories, the process. Centers like PSDC cannot be imposed; training modules, and trainers. They have access to they need to grow in already fertile ground. OPENING THE DOOR TO BE T TER JOBS BY IMPROVING EDUCATION AND SKILLS 209 and equipment suppliers, and private training demand-driven, timely, and calibrated train- institutes are an important means of expand- ing without the overhead investment and ing the resources available for developing delay of building a new center. workforce skills. Numerous training programs, especially Successful examples of such collective in the information and communications tech- action—a potential response to the externali- nology sector in India, are meeting industry ties associated with poaching—are in place needs. A randomized field experiment that in the region. The Chittagong Skills Devel- provided five months of intensive manage- opment Centre is the � rst industry-led, non- ment consultancy to help poorly managed pro�t skills training center in Bangladesh. It textile � rms adopt standard internationally is modeled on the successful Penang Skills used systems (for quality, inventory, and pro- Development Center (box 5.7). The center duction control) found gradual but signi�cant provides skills training to corporate mem- improvements in management practices and bers and other private companies. Its high- � rm performance (Bloom and others 2011). quality, cost-saving advantage stems from its Possible reasons for poor initial management corporate members’ willingness to share their included informational constraints (firms training resources, technologies, costs, train- were not aware of many modern management ers, space, and equipment, which ensures practices). 210 MORE AND BETTER JOBS IN SOUTH ASIA Annex 5A Additional tables and �gures on education and skills TABLE 5A.1 Share of Indian labor force requiring high concentration of nonroutine cognitive analytical and interpersonal skills, 1994–2010 (percent) Type of nonroutine cognitive skill 1994 2000 2005 2008 2010 Analytical Analyzing information 3.9 5.0 4.2 9.5 11.8 Thinking creatively 3.3 4.3 3.9 6.5 9.5 Interpreting information for others 1.3 1.1 1.0 1.4 1.3 Interpersonal Establishing and maintain relationships 6.9 7.4 7.9 11.9 13.2 Guiding, directing, and motivating subordinates 1.7 2.9 2.7 4.6 8.2 Coaching and developing others 1.2 1.0 1.1 1.3 1.2 Source: Authors, based on national labor force surveys. Note: High concentration of a skill in an occupation is when the content of the skill in the occupation is valued at 4 or 5 on a 1–5 scale. TABLE 5A.2 Mean years of education of 15–34 year olds in South Asia, by gender and country All Male Female   Country Earliest Most recent Earliest Most recent Earliest Most recent (period studied) year year year year year year Afghanistan (2008) —  2.5 —  3.7 —  1.4 Bangladesh (2000–10) 4.5 5.9 5.2 6.1 3.9 5.8 Bhutan (2003–07) 3.0 4.5 3.8 5.2 2.4 3.9 India (2000–10) 5.3 7.1 6.4 7.9 4.3 6.3 Maldives (1998–2004) 7.2 7.8 7.5 8.1 7.0 7.6 Nepal (1999–2008) 3.2 5.7 4.5 7.1 2.2 4.8 Pakistan (2000–09) 4.7 5.9 6.0 6.9 3.5 4.9 Sri Lanka (2000–08) 8.7 10.2 8.5 9.9 8.9 10.4 Source: World Bank staff, based on national labor force and household survey data. Note: — = Not available. OPENING THE DOOR TO BE T TER JOBS BY IMPROVING EDUCATION AND SKILLS 211 TABLE 5A.3 Summary of randomized experiments on teacher incentives Level of Strategy of Unintended Study Description incentive estimation Impact effects Duflo, Hanna, and Local program (113 School and teacher Randomized Reduction of teacher None Ryan (2010) schools, 57 treated) in experiment absence by 21 percentage India monitors teachers’ points presence at school and Increase in instructional provides rewards for time of 30 percent reduction in Positive impact on test absenteeism. scores (0.17 standard deviation) Effects come from incentives and not from monitoring Glewwe, Ilias, and Local program (100 School Randomized No change in homework Teaching to the Kremer (2008) schools, 50 treated) in experiment or pedagogy test (for example, Kenya provides cash No impact on test linked strategies for not rewards to teachers for to incentives leaving blank multiple reducing dropouts and Short-term impact on choice questions); increasing test scores. scores linked to test encouragement of No impact on dropout rate students to take and prepare for the test Lavy (2002) Local program School: secondary Regression Positive impact on None (involving more than schools (grade 7–12); discontinuity: average test score and 62 schools) conducts all teachers receive schools in same average credits tournament, offering prizes community Increase in proportion of prizes to top third of students for matriculation schools. certificate (especially among low-income groups) Reduction in dropout rate McEwan and Global program Individual teachers Regression No impact on test scores Permanent increases Santibáñez (2005) (all public teachers) (all levels, all grades) discontinuity: in wages without any provides rewards based schedule of benefit on student test results. points Muralidharan and Program in Andhra Group bonus Randomized Increase in test score of None cited Sundararaman Pradesh provides scheme covered all experiment 0.15 standard deviation in (2011) bonuses to schools and teachers in school; first year, 0.22 in second individual teachers tied individual bonus year in both programs to student achievement scheme covered Impacts fairly robust across (200 schools including teachers who taught grades, test difficulty levels, 100 in group bonus grades tested and most baseline school program, 100 in and student characteristics individual bonus Higher impacts in math program, and than language; impacts 100 controls). also observed on subjects not included for bonus After two years, impact of individual bonuses larger than group bonuses Source: World Bank staff compilation. 212 FIGURE 5A.1 Educational attainment in the labor force in South Asia, by country 100 2 1 1 3 1 1 2 3 4 3 4 1 2 2 4 7 7 2 3 5 6 6 7 5 4 3 8 10 8 4 11 10 9 4 4 4 6 6 13 5 5 5 90 7 14 8 18 7 16 9 21 9 20 12 80 21 22 6 24 25 25 14 25 25 12 28 70 30 16 10 17 38 12 15 13 15 60 19 12 48 5 14 16 16 11 16 percent 12 3 50 10 15 26 7 15 4 3 9 3 40 78 9 75 26 67 9 4 65 30 54 54 51 49 22 47 43 44 45 44 43 20 40 34 33 30 19 10 10 5 3 0 2008 2000 2010 2003 2007 1994 2000 2005 2008 2010 2000 2008 1998 2004 1999 2008 2000 2006 2008 2009 Afghanistan Bangladesh Bhutan India Sri Lanka Maldives Nepal Pakistan no education incomplete primary primary lower secondary higher secondary tertiary Source: Authors, based on data from national labor force and household surveys. FIGURE 5A.2 Share of labor force with at least primary, upper-secondary, and tertiary education, in South Asia and international comparators a. Male labor force with at least primary completed b. Female labor force with at least primary completed Sri Lanka 2008 100 Sri Lanka 2006 80 Sri Lanka 2000 Sri Lanka 2006 Sri Lanka 2008 Sri Lanka 2004 80 Sri Lanka 2000 Maldives 2004 Sri Lanka 2004 60 India 2010 India 2008 Maldives 2004 India 2005 Pakistan 2009 percent percent 60 India 2000 Pakistan 2008 40 India 2010 Nepal 2008 India 2008 Pakistan 2000 Nepal 2008 Maldives 1998 India 1994 India 2005 2 India 2000 200 05 Pakistan 2009 40 Bangladesh 2005 sh 20 ade esh 20 Pakistan 2008 Nepal 1999 Bangladesh 2002 Ba ngl lad India 1994 Bhutan 2007 Maldives 1998 g Pakistan 2000 Bhutan 2007 Ban Nepal 1999 Afghanistan 2008 Afghanistan 2008 Bhutan 2003 20 Bhutan 2003 0 0 1,000 2,000 3,000 4,000 5,000 0 1,000 2,000 3,000 4,000 5,000 per capita GDP in purchasing power parity dollars per capita GDP in purchasing power parity dollars c. Male labor force with at least higher secondary completed d. Female labor force with at least higher secondary completed 60 50 40 40 30 Sri Lanka 2004 Sri Lanka 2008 percent percent 20 02 05 Sri Lanka 2000 Maldives 2004 20 20 Sri Lanka 2006 sh sh 20 India 2008 India 2010 Maldives 2004 de ade 994 la l 1 India 2005 India 2008 India 2010 India 2005 ng ng ia India 2000 Pakistan 2009 Sri Lanka 2004 Ba Ba Ind India 2000 Pakistan 2009 Sri Lanka 2008 Pakistan 2008 Bhutan 2007 Afghanistan 2008 Pakistan 2008 Nepal 2008 Pakistan 2000 10 Sri Lanka 2000 Sri Lanka 2006 0 Afghanistan 2008 Maldives 1998 Bhutan 2003 Nepal 2008 B Ban India 1994 Bhutan 2007 Nepal 1999 an gla Pakistan 2000 gla de Nepal 1999 de sh sh 20 Maldives 1998 Bhutan 2003 0 20 05 02 –20 0 1,000 2,000 3,000 4,000 5,000 0 1,000 2,000 3,000 4,000 5000 per capita GDP in purchasing power parity dollars per capita GDP in purchasing power parity dollars (continues next page) 213 214 FIGURE 5A.2 Share of labor force with at least primary, upper-secondary, and tertiary education, in South Asia and international comparators (continued) e. Male labor force with tertiary completed f. Female labor force with tertiary completed 20 20 15 10 percent percent 10 India 2010 Pakistan 2009 India 2008 India 2010 Sri Lanka 2006 India 2008 Pakistan 2008 Bangladesh 2005 Pakistan 2009 Bangladesh 2002 Pakistan 2000 Lanka 2008 Sri Lanka 2004 SriMaldives 2004 India 2005 Sri Lanka 2000 Pakistan 2008 Bhutan 2007 Afghanistan 2008 Sr Bangladesh 2005 India 1994 Maldives 1998 iL 0 Nepal 2008 an Bhutan 2003 Nepal 1999 5 Bangladesh 2002 Bhutan 2007 India 2000 ka Pakistan 2000 20 Nepal 2008 06 Afghanistan 2008 Maldives 1998 Maldives 2004 Nepal 1999 India 2000 India 2005 000 Sri Lanka 2004 Sri Lanka 2008 ka 2 Bhutan 2003 0 India 1994 Lan Sri –10 0 1,000 2,000 3,000 4,000 5,000 0 1,000 2,000 3,000 4,000 5,000 per capita GDP in purchasing power parity dollars per capita GDP in purchasing power parity dollars Sources: Authors, based on data from World Bank 2011b and national labor force and household surveys. Note: The dark line shows the predicted values of the share of the labor force with completed primary, upper-secondary, and tertiary education by per capita GDP. Predicted values are obtained from cross-country regressions (excluding high-income countries) of the share of the labor force with completed primary, upper-secondary, and tertiary education on the log of per capita GDP. The shaded area is the 95 percent con�dence interval band around the regression line. OPENING THE DOOR TO BE T TER JOBS BY IMPROVING EDUCATION AND SKILLS 215 FIGURE 5A.3 Gross and net enrollment rates in primary and secondary education, by region and country a. Gross enrollment rate 120 100 80 percent 60 40 20 0 8 09 9 08 09 2 09 9 eC nA 0 n om Org 20 i�c, t, 2 tio r As , 20 nd en era fo 00 00 00 00 th Lati ca, 2 ara 20 20 20 20 ar m 09 an 09 09 00 n n a pm p on c ,2 ,2 l, 2 ,2 h Pa Ea bea rica h, a, s, n, Af b-Sa lo o-o ati an an ka 9 pa ive di es ta d an ib e ist ut ve C nis In Ne kis lad ald Su Bh iL an De ic a ri ia Pa ng M Sr gh Ba Af st an con E d b. Net enrollment rate 100 80 60 percent 40 20 0 8 06 09 08 07 4 09 7 eC nA 0 n De ic a 00 , t, 2 tio r As n, 2 and om Org 2 i�c en era fo 00 00 00 th Lati ca, 2 ara 20 20 20 20 20 ar m 08 an 9 ve C nis 7 00 n pm p on c ,2 l, 2 ,2 ia 00 h Pa Ea bea rica h, n, a, s, n, Af b-Sa lo o-o ati an ka 8 pa ive di es a ta d an ib e st ut In Ne kis ad ald ni Su Bh iL ri Pa gl ha M Sr n g Ba Af st an con E d primary secondary Sources: Authors, based on data from UIS Database and EdStats. Net and gross enrollment rates for Afghanistan, Nepal, and Sri Lanka were calculated using national household surveys. Note: For Afghanistan, primary grades are 1–6 (ages 6–12) and secondary grades 7–12 (ages 12–18). For Nepal and Sri Lanka, primary grades are 1–5 (ages 5–10) and secondary grades are 6–12 (age 11–18). 216 MORE AND BETTER JOBS IN SOUTH ASIA FIGURE 5A.4 Enrollment in vocational education and training as a share of secondary enrollment, by region and country 30 25 25 20 17 percent 15 10 10 8 8 5 5 3 3 3 1 2 0 an an tin Afr ran Ea ean d om Org aci� nd lo at for a h ka an l di ib an pa es ist ist an P a ut ica an ic C ani c en n In ha ve per n lad Ne ar ca ia pm io an De o tio k Bh iL Sa Pa As t e C ri ng b gh d o- sa Sr th Ame b- st Ba Af Su La on Ec Source: Chhoeda, Dar, and Tan 2010. Note: Data are for latest year available. They primarily reflect preemployment training for South Asian countries. Although vocational education provided by the public school sector is also included, this sector is absent in many South Asian countries and is very small in others. FIGURE 5A.5 Gross enrollment rate in tertiary education, by region and country 80 72 60 percent 40 37 28 21 20 13 8 7 6 5 6 4 12 0 n h an a l an a tin Afri an b d om Org ci� d lo at for pa di nk sta es ib an Pa an ut t r In Ne Am ca Ea ean an ic C ani c en n ha kis ve per n ad La i ar ca ia pm io Bh De o tio an Sa Pa l As t i ng e C eri Sr gh d o- sa b- st Ba Af Su th La on Ec Sources: Authors, based on data from UNESCO 2011 and World Bank 2011b. Note: Data are for latest year available. Gross enrollment of 21 percent in Sri Lanka includes students enrolled in external degree programs, some of whom may be only nominally enrolled. Gross tertiary enrollment excluding external degree enrollment is about 12 percent. FIGURE 5A.6 Share of young cohorts with completed primary and lower-secondary education in South Asia, by gender and country a. Primary attainment by 15–19 year olds and lower-secondary attainment by 20–24 year olds, women 100 97 95 99 91 82 84 80 77 67 68 63 59 59 62 60 56 percent 52 48 39 42 41 40 37 29 31 21 20 20 14 8 0 2008 2000 2010 2000 2010 1998 2004 1999 2008 2000 2009 2000 2008 Afghanistan Bangladesh India Maldives Nepal Pakistan Sri Lanka completed primary (15–19 year olds) complete lower secondary (20–24 year olds) b. Primary attainment by 15–19 year olds and lower-secondary attainment by 20–24 year olds, men 100 98 95 98 89 83 84 80 73 77 73 77 72 69 64 60 62 60 57 55 57 56 percent 53 43 45 46 40 32 25 27 20 0 2008 2000 2010 2000 2010 1998 2004 1999 2008 2000 2009 2000 2008 Afghanistan Bangladesh India Maldives Nepal Pakistan Sri Lanka completed primary (15–19 year olds) complete lower secondary (20–24 year olds) Source: Authors, based on data from national labor force and household surveys. 217 218 FIGURE 5A.7 Share of young cohorts with completed upper-secondary and tertiary education in South Asia, by gender and country a. Upper-secondary attainment by 25–29 year olds and tertiary attainment by 30–34 year olds, women 30 27 22 20 18 18 percent 13 10 10 7 8 6 7 5 4 5 3 2 3 4 1 1 2 3 3 2 1 0 1 0 2008 2000 2010 2000 2010 1998 2004 1999 2008 2000 2009 2000 2008 Afghanistan Bangladesh India Maldives Nepal Pakistan Sri Lanka completed primary (25–29 year olds) complete lower secondary (30–34 year olds) b. Upper-secondary attainment by 25–29 year olds and tertiary attainment by 30–34 year olds, men 30 27 25 20 20 18 18 percent 16 16 12 14 14 13 12 9 10 10 10 8 6 6 6 7 5 4 2 1 3 3 0 2008 2000 2010 2000 2010 1998 2004 1999 2008 2000 2009 2000 2008 Afghanistan Bangladesh India Maldives Nepal Pakistan Sri Lanka completed primary (25–29 year olds) complete lower secondary (30–34 year olds) Source: Authors, based on data from national labor force and household surveys. OPENING THE DOOR TO BE T TER JOBS BY IMPROVING EDUCATION AND SKILLS 219 FIGURE 5A.8 Mean years of education of 15–34 year olds in selected South Asian countries, by caste/ethnicity 12 10.4 9.8 10 8.9 8.5 8.2 8 7.6 7.7 7.1 years of education 5.8 5.8 6 5.3 5.2 4.4 4 3.3 2.9 2 0 rs Ca du lim ste e r ri i te lit lim la r n il jat wa oo m ib ka et he ha Da as rd in Ca Tr ste na us us Ta hh nM an Ne eC Sin Ot wa r H M M led Ja led an /C iL ka dl ck the an Sr di du du id an In m O iM he he iL ah Sc ra Sr Sc Br Te Ba India 2010 Nepal 2008 Sri Lanka 2008 Source: Authors, based on data from national labor force and household surveys. 220 MORE AND BETTER JOBS IN SOUTH ASIA FIGURE 5A.9 Share of 20–28 year olds in Afghanistan and Nepal with different years of education completed in Afghanistan (2008) and Nepal (2004), by gender and socioeconomic status a. Afghanistan all urban rural 80 60 percent 40 20 0 0 5 10 15 0 5 10 15 0 5 10 15 years of education b. Nepal all urban rural 100 percent 50 0 0 5 10 15 0 5 10 15 0 5 10 15 years of education boys girls poorest middle richest Source: Authors, based on national household surveys. FIGURE 5A.10 Percentage of children under age 5 with stunting in selected South Asian countries, by socioeconomic status 70 62 63 64 63 60 58 56 59 56 53 53 54 50 49 49 47 46 42 43 42 percent 40 36 30 30 20 10 0 a h n l pa di es ta In Ne kis lad Pa ng Ba Q1 Q2 Q3 Q4 Q5 Source: Van De Poel and others 2008. Note: Bars indicate income quintile. OPENING THE DOOR TO BE T TER JOBS BY IMPROVING EDUCATION AND SKILLS 221 Annex 5B Projections of the educational attainment of South Asia’s population and labor force The projections used in this book are based attainment projections by age and gender are on an adaptation of the international pro- adapted from KC and others (2010). Labor jections by KC and others (2010) to South force participation rates for each age/sex/ Asia. Projections of educational attainment educational attainment/rural-urban group were made under three scenarios, a base- were calculated from the latest national labor line scenario and two optimistic scenarios. force and household surveys. They were then The assumptions under the three scenarios applied to the educational attainment projec- are summarized in table 5B.1. tions of the population to obtain projections Projections for the male and female labor on the educational attainment of the labor force for 2030 were obtained by explicitly force. The assumption is that there would be factoring in trends in educational attainment no behavioral change in labor force participa- and urbanization. Various sources of data tion, that current participation rates for each were used: population projections by gender cell would hold in 2030. Changes in labor and �ve-year age groups are from World Pop- force participation are therefore driven by ulations Prospects: The 2008 Revision (UN trends in urbanization and educational attain- 2010); urban and rural projections are from ment of the population. Table 5B.2 shows the UN’s World Urbanization Prospects: The current and projected attainment of the labor 2009 Revision (UN 2011); and educational force under various educational scenarios. TABLE 5B.1 Assumptions underlying scenarios used to project educational attainment of population in South Asia Scenario Primary education (age 10–14) Secondary education (age 20–24) Tertiary education (age 30–34) Baseline SA–GET Primary attainment over past Global Education Trend (GET), GET (SA8) 10 years in all South Asian countries based on trends from 120 countries Optimistic SA-GET Primary attainment over past Rate of increase for secondary age cohorts GET (SA6) 10 years in all South Asian countries whose gains were higher than the GET in except Pakistan (stalled) and past 10 years Maldives (universal) Fast Track (FT1) 99 percent by 2020 60 percent Tertiary progression by 2030, 100 percent by 2040 following Singapore rate Source: Authors. Note: GET scenarios as in KC and others 2010. SA = South Asia. 222 MORE AND BETTER JOBS IN SOUTH ASIA TABLE 5B.2 Educational attainment of the South Asian labor force in 2010 (estimated) and 2030 (projected) under various scenarios, by country (percent) 2010 2030 SA8   SA6   FT No Education No Education No Education No Education Secondary Secondary Secondary Secondary Primary Primary Primary Primary Tertiary Tertiary Tertiary Tertiary Country Total Afghanistan 73.1 15.3 9.6 2.0 43.1 33.4 20.1 3.5 43.1 24.8 28.6 3.5 33.3 30.0 23.0 13.7 Bangladesh 32.9 35.6 25.6 6.0 14.6 37.5 38.1 9.9 14.5 28.9 46.6 9.9 14.3 28.4 32.1 25.2 Bhutan 53.3 26.9 15.2 4.6 23.7 39.0 28.5 8.8 23.4 29.1 38.7 8.7 24.3 28.9 25.4 21.4 India 33.0 24.0 35.5 7.5 15.4 23.3 49.5 11.7 15.5 18.3 54.3 11.9 14.9 18.8 41.7 24.6 Maldives 20.4 32.6 28.8 18.2 4.3 29.9 36.9 28.9 4.3 24.6 42.4 28.7 4.0 24.5 32.9 38.5 Nepal 43.2 27.2 27.4 2.2 18.9 29.9 46.6 4.5 19.0 22.2 54.2 4.6 18.4 23.1 45.3 13.2 Pakistan 41.8 19.5 31.6 7.1 22.8 22.3 43.7 11.2 22.9 16.0 49.6 11.4 19.8 18.3 38.1 23.9 Sri Lanka 2.9 23.2 51.9 22.1 0.8 12.4 55.5 31.4 0.8 8.9 58.9 31.5 0.8 8.6 40.1 50.5 Male Afghanistan 64.3 20.3 13.1 2.4 34.8 36.8 25.1 3.3 35.0 28.0 33.7 3.4 27.3 34.6 26.1 11.9 Bangladesh 28.8 37.3 27.4 6.6 13.9 37.6 38.5 10.0 13.9 29.5 46.5 10.1 14.0 29.5 32.9 23.6 Bhutan 45.7 32.4 16.2 5.7 20.2 42.7 27.7 9.3 20.2 35.0 35.5 9.4 20.8 35.0 23.8 20.5 India 24.7 25.4 41.4 8.5 10.9 22.3 54.7 12.1 10.9 18.1 58.8 12.2 10.5 18.6 45.5 25.4 Maldives 19.5 33.8 29.5 17.2 4.2 30.8 38.1 26.9 4.2 25.8 43.1 26.9 3.9 25.9 33.3 36.9 Nepal 26.8 33.3 36.3 3.6 10.5 30.3 52.6 6.6 10.5 24.7 58.2 6.6 9.6 25.6 46.4 18.4 Pakistan 33.9 21.8 36.8 7.5 17.9 23.4 48.0 10.7 17.9 17.6 53.6 10.8 15.7 19.9 42.0 22.4 Sri Lanka 2.4 25.9 55.3 16.5 0.8 15.2 62.0 22.0 0.8 11.1 66.0 22.1 0.8 11.1 43.8 44.4 Female Afghanistan 91.5 4.9 2.3 1.2 62.8 25.1 8.3 3.9 62.3 17.3 16.5 3.9 47.2 19.3 15.7 17.8 Bangladesh 46.0 30.1 19.7 4.1 16.7 37.1 36.8 9.5 16.4 27.2 47.1 9.4 15.0 25.5 29.7 29.8 Bhutan 64.6 18.8 13.7 2.9 28.8 33.5 29.7 8.0 28.0 20.8 43.4 7.8 29.4 20.3 27.6 22.7 India 56.5 20.0 18.8 4.7 29.7 26.6 33.1 10.6 30.4 18.9 39.8 10.9 28.6 19.1 30.0 22.3 Maldives 21.8 30.5 27.8 19.9 4.5 28.7 35.0 31.8 4.4 22.8 41.4 31.4 4.2 22.5 32.3 41.0 Nepal 60.4 20.8 18.0 0.8 28.5 29.5 39.8 2.2 28.8 19.5 49.5 2.3 28.4 20.2 44.2 7.2 Pakistan 73.7 10.1 10.6 5.5 46.6 16.7 22.8 13.8 47.6 8.0 29.9 14.4 38.4 10.7 20.0 30.9 Sri Lanka 3.7 18.5 46.1 31.7 0.9 7.9 44.9 46.4 0.9 5.3 47.4 46.5 0.8 4.9 34.7 59.7 Sources: Authors, based on data from national labor force and household surveys; UN 2009, 2010; and KC and others 2010. Note: See text for descriptions of scenarios. FT = fast track, SA = South Asia. Notes about the constraints facing existing enter- prises. They are not useful in understanding 1. Access to education is a basic human right. It constraints perceived by potential �rms that also has considerable value beyond its role in did not enter in the �rst place. the labor market: it enables individuals to be 3. Autor, Levy, and Murnane (2003) argue that better parents, more responsible citizens, and as a result of computerization, the shift in more effective stewards of the natural environ- demand for labor has contributed to the secu- ment. This book focuses on the skills imparted lar increase in the premium for tertiary educa- and how they play out in the labor market. tion in the United States. 2. Other constraints are more binding for for- 4. The increase in the public sector likely reflects mal sector �rms in the region as a whole. adjustments in wages to reduce wage compres- Enterprise surveys provide information only sion. In India, the premium for upper-secondary OPENING THE DOOR TO BE T TER JOBS BY IMPROVING EDUCATION AND SKILLS 223 relative to lower-secondary increased in both is a broad category that combines lower-sec- the private and the public sector. ondary and completed upper-secondary. 5. Instead of point estimates, the probabilities of 14. This section bene�ted from the draft chap- worker mobility are estimated as upper and ter on early childhood development and lower bounds. The lines shown in the �gures preschool prepared for the ongoing regional link the midpoints of the bounds; they pri- study on the quality of learning in South Asia marily show how the bounds vary with edu- (Alderman 2011). cation. The midpoints of the bounds should 15. A growing neuroscience literature provides not be interpreted as true point estimates. insights into biologically determined win- 6. These transition probabilities are conditional dows in childhood for these processes. For on being in a particular state in the initial example, the number of neurons devoted to year. The likelihood of a worker with tertiary language peaks before a child turns one and education being in agriculture or in casual then declines, implying that children lose plas- employment is relatively low. ticity even before they engage in rudimentary 7. Annex �gure 5A.3 shows enrollment rates at conversations. primary and secondary. Annex �gure 5A.5 16. Martorell and Horta (2010) �nd that weight shows enrollment at tertiary institutions in gain during the �rst two years of life has the South Asia and other regions. strongest association with schooling, followed 8. For all countries, this section uses the multi- by birthweight. A one standard deviation purpose household surveys that permit analysis increase in birth weight (about 0.5 kilograms) is by consumption quintiles (see appendix A to associated with 0.21 years more schooling and the report). The only exception is India, where an 8 percent decrease in the risk of grade fail- the Labor Force Survey is used. ure. A one standard deviation increase (about 9. Between 2000 and 2008, young female 0.7 kilograms) in conditional weight gain in cohorts in Bangladesh still had lower attain- the �rst two years of life is associated with 0.43 ment at the upper-secondary and tertiary lev- years more schooling and a 12 percent reduc- els, but they made signi�cantly larger gains tion in the risk of grade failure. Gains in educa- at the primary and lower- secondary levels, tion as a result of weight gain were particularly surpassing young men. In Maldives, primary large for babies who were born small. attainment among young female cohorts was 17. Wasting, stunting, and underweight indica- comparable to that of males, but attainment tors refer to the proportion of children under at higher levels was still lower. In Sri Lanka, �ve whose weight for height, height for age, educational attainment among young female and weight for age are more than two stan- cohorts was higher than of males at all lev- dard deviations from the medians of an inter- els of education (see annex �gures 5A.6 and national reference population recognized by 5A.7). the World Health Organization. Low height 10. Private expenditures by households are also for age (also termed stunting) is a measure high in some countries. In India, for example, of chronic malnutrition. Progress toward the they are equal to 1.4 percent of gross domes- Millennium Development Goal is tracked tic product (Tilak 2009). based on underweight. Trends in underweight 11. Hanushek and Woessman �nd that an increase and stunting tend to move together. Although of one standard deviation in student reading underweight has a somewhat higher risk of and math scores (roughly equivalent to improv- mortality, stunting is more strongly associated ing a country’s performance ranking from the with long-term development (human capital median to the top 15 percent) is associated and economic) outcomes. with a 2 percentage point increase in annual 18. These �ndings are based on indicators devel- gross domestic product per capita growth. oped by the United Nations Children’s Fund, 12. The scenario discussed here is the SA6 including “family care indicators,� an index scenario in annex 5B. of inputs and resources, and more compre- 13. Projections are based on a methodology used hensive “home observations for measurement for global projections; as a result, there are of the environment� (HOME), which can be differences in the de�nitions of educational adapted to household surveys. Family care attainments by level used in this section. In the indicators have been validated as a predictor projections in this section, “primary� combines of cognitive development in diverse settings. incomplete and complete primary; “secondary� See Alderman (2011) for a discussion. 224 MORE AND BETTER JOBS IN SOUTH ASIA 19. The role of the private sector varies greatly Almeida, R., and R. Aterido. 2010. “Investment across countries. Whereas almost all second- in Job Training: Why Are SMEs Lagging So ary schools in Afghanistan are government Much Behind?� Policy Research Working schools, almost all secondary schools in Paper 5358, World Bank, Washington, DC. Bangladesh are private. Almeida, R., J. Behrman, D. Robalino. 2011. The 20. Some caution should be exercised in inter- Right Skills for the Job? Rethinking Effective preting these results, as these studies may Training Policies for Workers. World Bank, not take into account all the characteristics Social Protection. Human Development, that influence the choice of public or private Washington, DC. school. Desai and others (2008) control for Andrabi, T., J. Das, and A. Khwaja. 2009. the endogeneity of school choice in India. “Report Cards: The Impact of Providing They �nd modest gains in math and reading School and Child Test Scores on Educational scores in private schools compared with gov- Markets.� Working Paper, Kennedy School ernment schools, with higher gains for chil- of Government, Harvard University. http:// dren from lower economic strata. www.hks.harvard.edu/fs/akhwaja/papers/ 21. These tests do not typically measure impor- RC_08Oct09Full.pdf. tant skills and knowledge against established Andrabi, T., J. Das, A. Khwaja, T. Vishwanath, learning standards, tending instead to rely T. Zajonc, and the LEA PS Team. 2007. heavily on rote learning and multiple choice “Learning and Educational Achievement in answers. In addition, the results are generally Punjab Schools (LEAPS): Insights to Inform not comparable across years. the Education Policy Debate.� The LEAPS 22. Technical training is costly to set up, caus- Project. Punjab. ing private entrepreneurs to shy away from Asadullah, M. N., N. Chaudhury, D. Parajuli, L. this kind of training. Commercial training (in R. Sarr, and Y. Savchenko. 2011. Reaching languages and secretarial skills, for example), Out-of-School Children (ROSC) Project: which costs less to provide, attracts more pri- Evaluation Report. World Bank, South vate providers. Asia Region, Human Development Unit, Washington, DC. Asadullah, M. N., N. Chaudhury, D. Parajuli, References L. R. Sarr, Y. Savchenko, and S. R. Al-Zayed. Acemoglu, D., and D. Autor. 2010. “Skills, 2009. “Bangladesh Secondary Education Tasks and Technologies: Implications for Quality and Access Enhancement Project: Employment and Earnings.� In Handbook of B asel i ne Repor t.� World B a n k , S out h Labor Economics, vol. 4B, ed. O. Ashenfelter Asia Region, Human Development Unit, a n d D. C a rd . 10 43 –7 2 . A m s t e rd a m : Washington, DC. Elsevier. Autor, D., F. Levy, and R. Murnane. 2003. “The Agarwal, P. 2009. Indian Higher Education: Skill Content of Recent Technological Change: Envisioning the Future. New Delhi: Sage An Empirical Exploration.� Quarterly Journal Publications. of Economics 118 (4): 1279–33. Alam, A., J. Baez, and X. del Carpio. 2011. “Does Barrera-Osorio, F., D. S. Blakeslee, M. Hoover, Cash for School Influence Young Women’s L. L. Linden, and D. Raju. 2011. “Expanding Behavior in the Longer Term? Evidence from Educational Opportunities in Remote Parts Pakistan.� Policy Research Working Paper of the World: Evidence from a RCT of a 5669, World Bank, Washington, DC. Public Private Partnership in Pakistan.� Paper A ld e r m a n , H . 2 011. “E a rly C h i ld ho o d presented at the Third Institute for the Study Development and the Role of Pre-school.� of Labor (IZA) workshop, “Child Labor in Background paper prepared for Regional Developing Countries,� Mexico City. http:// Quality of Education Study, South Asia www.iza.org/conference_files/childl2011/ Region, Human Development Unit, World blakeslee_d6783.pdf. Bank, Washington, DC. Barrera-Osorio, F., and D. Raju. 2010. “Short- Alderman, H., J. Behrman, V. Lavy, and R. Run L earning Dynamics under a Test- Menon. 2001. “Child Health and School Based Accountability System: Evidence from Enrollment: A Longitudinal Analysis.� Journal Pakistan.� Policy Research Working Paper of Human Resources 36 (1): 185–205. 5465, World Bank, Washington, DC. OPENING THE DOOR TO BE T TER JOBS BY IMPROVING EDUCATION AND SKILLS 225 Batra, G., and A. Stone. 2004. Investment Climate, An Experiment in Cross-Cultural Learning.� Capabilities and Firm Performance: Evidence Working Paper, World Bank, South Asia from the World Business Environment Region, Human Development Unit, Surveys. World Bank, Investment Climate Washington, DC. Department, Washington, DC. Das, J., P. Pandey, and T. Zajonc. 2006. “Learning Behrman, J., H. Alderman, and J. Hoddinott. Levels and Gaps in Pakistan.� Policy Research 2004. “Hunger and Malnutrition.� In Global Working Paper, World Bank, Washington, DC. Crises, Global Solutions, ed. B. Lomborg, 363- Desai, S., A. Dubey, R. Vanneman, and R. Banerji. 420. Cambridge: Cambridge University Press. 2008. “Private Schooling in India: A New Blom, A., and H. Saeki. 2011. “Employability Educational Landscape.� Brookings–National and Skill Set of Newly Graduated Engineers in Council of Applied E conomic Research India.� World Bank Policy Research Working (NCAER) India Policy Forum, New Delhi. Paper, Washington, DC. Duf lo, E ., R. Hanna, and S. Ryan. 2010. Bloom, N., B. Eifert, A. Mahajan, D. McKenzie, “Incentives Work: Getting Teachers to Come and J. Roberts. 2011. “Does Management to School.� Working Paper, Department of Matter?� World Bank Policy Research Working Economics, Harvard University, Cambridge, Paper 5573, Washington, DC. MA. http://www.hks.harvard.edu/fs/rhanna/ Bobonis, G., M. Edward, and C. Puri-Sharma. Incentives_Work_30may2010.pdf. 2006. “Anemia and School Participation.� EI (Educational Initiatives). 2006. “Student Journal of Human Resources 41 (4): 692–721. Learning in the Metros.� Working Paper 2, Bruns, B., D. Filmer, and H. A. Patrinos. 2011. Ahmedabad, India. Making Schools Work: New Evidence on Fernald, L ., E . Galasso, and L . Ratsifan- Accountability Reforms. Washington, DC: drihamanana. 2011. “Socioeconomic Gradients World Bank. and Child Development in a Very Low-Income Burde, D., and L. Linden. 2010. “The Effect Population: Evidence from Madagascar.� of Village-Based Schools: Evidence from a Developmental Science 14 (4) 1–16. Randomized Controlled Trial in Afghanistan.� Fishbein, A., and N. Schady. 2009. Conditional Working Paper, J-PAL, Cambridge. Cash Transfers: Reducing Present and Future Carlin, W., and M. Schaffer. 2011. “Which Povery. Washington, DC: World Bank. Elements of the Business Environment Matter Gill, I., F. Fluitman, and A. Dar, eds. 2000. Most for Firms and How Do They Vary across Vocational Education and Training Reforms: Countries?� Background study conducted for Matching Skills to Markets and Budgets. this book. Oxford: Oxford University Press. Chaudury, N., and D. Parajuli. 2010. “Conditional Glewwe, P., N. Ilias, and M. Kremer. 2008. Cash Transfers and Female S chooling: “Teacher Incentives in the Developing World.� The Impact of the Female School Stipend In Performance Incentives: Their Growing Programme on Public School Enrolments in Impact on K–12 Education, ed. Matthew Punjab, Pakistan.� Applied Economics 42 (28): Spr i nger, 2 95 – 32 6 . Wa sh i ng ton , D C : 3565–83. Brookings Institution Press. Chhoeda, T., A. Dar, and H. Tan. 2010. Vocational Government of India. 2009. National Policy on Education and Technical Training in South Skills Development. Ministry of Labour and Asian Countries: Summary of Issues and Employment, New Delhi. Recommended Policy Reforms. World Bank, ———. 2010. Project Implementation Plan for South Asia Region, Human Development Unit, the Second Phase of the Technical Education Washington, DC. Quality Improvement Project. National Christian, P., L. E. Murray-Kolb, S. K. Khatry, Project Implementation Unit, Noida. J. Katz, B. A. Schaefer, P. M. Cole, S. C. Government of Maharashtra. 2009. “May 2009 LeClerq, and J. M. Tielsch. 2010. “Prenatal Amendment to the Maharashtra University Micronutrient Supplementation and Intellectual Act 1994.� Mumbai and Motor Function in Early School-Aged Goyal, S., and P. Pandey. 2009a. Contract Children in Nepal.� Journal of the American Teachers. Washington, DC: World Bank. Medical Association 304 (24): 2716–23. ———. 2009b. “How do Government and Private Clarke, P. 2005. “Technical and Vocational Schools Differ? Findings from Two Large Education Systems in Australian and India: Indian States.� Working Paper 30, World 226 MORE AND BETTER JOBS IN SOUTH ASIA Bank, South Asia Region, Education Sector, Low- and Middle-Income Countries.� Journal Washington, DC. of Nutrition 40 (2): 348–54. Gustabo, B., C. Puri-Sharma, and E. Miguel. McEwan, P. J., and L. Santibáñez. 2005. “Teacher 2006. “Anaemia and School Participation.� and Principal Incentives in Mexico.� In Work i ng Paper 337, e S oci al Sci e n c es . Incentives to Improve Teaching: Lessons http://www.esocialsciences.com /articles/ from Latin America, ed. E. Vegas, 213–53. displayArticles.asp?Article_ID=337. Washington, DC: World Bank. Hanushek, E ., and L . Woessmann. 20 08. Muralidharan, K., and V. Sundararaman. “The Role of Cognitive Skills in Economic 2010. “Contract Teachers: Experimental Development.� Journal of Economic Literature Evidence from India.� Paper presented at the 46 (3): 607–68. Massachusetts Avenue Development Seminar, Hoddinott, J., J. Maluccio, J. Behrman, R. Flores, Washington, DC, May 24. and R. Martorell. 2008. “Effect of a Nutrition ———. 2011. “Teacher Performance Pay: Intervention during Early Childhood on Experimental Evidence from India.� Journal Economic Productivity in Guatemalan Adults.� of Political Economy 119 (1): 39–77. Lancet 371 (February 2): 411–16. Naudeau, S., S. Martinez, P. Premand, and D. Hoddinott, J., J. Maluccio, J. R. Behrman, Filmer. 2011. “Cognitive Development among R. Martorell, P. Melgar, A. R. Quisumbing, Young Children in Low-Income Countries.� In M. Ramirez-Zea, A. D. Stein, and K. M. Yount. No Small Matter: The Interaction of Poverty, 2011. “The Consequences of Early Childhood Shocks, and Human Capital Investment, ed. Grow th Failure over the Life Course.� H. Alderman, 9–50. Washington, DC: World IFPRI Discussion Paper 01073, International Bank. Food Policy Research Institute, Washington, Oster, E., and B. Millet. 2011. “Do Call Centers DC. Promote School Enrollment? Evidence from I LO (International Labour Office). 20 03. I ndia.� Working Paper, Department of “Industrial Training Institutes in India: The Economics, University of Chicago. Ef�ciency Study Report.� Subregional Of�ce Pandey, P., S. Goyal, and V. Sundararaman. 2009. for South Asia, ILO, New Delhi. “Community Participation in Public Schools: International Finance Corporation. 2010. Impact of Information Campaigns in Three “Public Private Partnerships in Education: Indian States. “ Education Economics 17 (3): Sharing IFC’s Experience.� Presentation to the 355–75. National Summit on Higher Education, New Pratham. 2010. Annual Status of Education Delhi, India. Report. Rural 2010. New Delhi. Jensen, R. 2010. “Economic Opportunities Riboud, M., Y. Savchenko, and H. Tan. 2007. and Gender Differences in Human Capital: “The Knowledge Economy and Education Experimental Evidence for India.� NBER and Training in South Asia.� World Bank, Working Paper 16021, National Bureau of South Asia Region, Human Development Unit, Economic Research, Cambridge, MA. Washington, DC. KC, S., B. Barakat, A. Goujon, V. Skirbekk, W. SAFED (South Asia Forum for Education and Sanderson, and W. Lutz. 2010. “Projection Development). 2010. Annu al Status of of Populations by Level of Educational Education Report 2010. Lahore, Pakistan. Attainment, Age, and Sex for 120 countries for Skills Australia. 2011. “Ministerial Council for 2005–2050.� Demographic Research 22 (15): Tertiary Education and Employment (MCTEE) 383–472. Industry Forum.� Canberra. http://www Lavy, V. 2002. “Paying for Performance: The .skillsaustralia.gov.au/MCTEEIndustryForum Effect of Teachers’ Financial Incentives on .shtml. Students’ Scholastic Outcomes.� CEPR S r i L a n k a I n for m at ion C om mu n ic at ion Discussion Paper 3862, Center for Economic Tech nolog y A ssociation. 20 07. R i sing and Policy Research, Washington, DC. Demand: The Increasing Demand for IT Martorell, R., and B. Horta. 2010. “Weight Gain Workers Spells a Challenging Opportunity for in the First Two Years of Life Is an Important the IT Industry. Colombo. Predictor of Schooling Outcomes in Pooled Thinley, K. 2009. “Higher Education in the Analyses from Five Birth Cohorts from Kingdom of Bhutan: Cherishing Dreams and OPENING THE DOOR TO BE T TER JOBS BY IMPROVING EDUCATION AND SKILLS 227 Confronting Challenges.� Report prepared for Hen n i ng ha m , S . Cha ng, J. H a mada n i, UNESCO, Paris. B. Lozoff, J. Gardner, C. Powell, A. Rahman, Tilak, J. 2009. Household Expenditure on and L. Richter. 2011. “Inequality Begins Education and Implications for Rede�ning the in Early Childhood: Risk and Protective Poverty Line in India. Government of India, Factors for Early Child Development.� Planning Commission, New Delhi. Lancet. U N ( Un ited Nat ions). 20 09. T he Wo rl d World Bank. 2007. Skills Development in India: Populations Prospects: The 2008 Revision. The Vocational Education and Training Department of Economic and Social Affairs, System. Washington, DC. Population Division, New York. ———. 20 08. The Bangladesh Voc ation al ———. 2010. World Urbanization Prospects: E d u c a t i o n a n d Tra i n i n g S y s t e m : A n The 2009 Revision: Highlights. Department Assessment. South Asia Region, Human of Economic and Social Affairs, Population Development Unit, Washington, DC. Division, New York. ——— . 20 09. T he To we rs of L e a r n i ng : UNESCO (United Nations Education, Scienti�c Performance, Peril, and Promise of Higher and Cultural Organization). 2011. Data Education in Sri Lanka. Washington, DC. from U N E SCO I nstit ute for Statistics, ———. 2010. Nepal: Second Higher Education Montreal. Project: Implementation Status Results Van de Poel, E., A. Hosseinpoor, N. Speybroeck, T. Report. Washington, DC. Van Ourti, and J. Vega. 2008. “Socioeconomic ———. 2011a. Skills and Research for Productivity Inequalities in Malnutrition in Developing and Growth: Higher Education in East Asia. Countries.� Bulletin of the World Health Washington, DC: World Bank. Organization. 86 (4) 282–91. ———. 2011b. World Development Indicators. Walker, S., T. Wachs, S. Grantham-McGregor, Washington, DC: World Bank. M. Black, C. Nelson, S. Huffman, H. Baker- CHAPTER 6 The Role of Labor Market Regulations, Institutions, and Programs Questions and Findings Questions income support and active labor market pro- grams available to workers. The distributional • What is the role of labor market policies and effects imply political economy challenges to institutions in encouraging job creation and this approach, but the aggregate welfare gains protecting workers in the formal and infor- could be substantial. One policy, especially mal economy? relevant in Sri Lanka, is to transform the • How can South Asian countries increase current severance system into an unemploy- access of informal sector workers to programs ment benefit scheme, perhaps through the that help them manage labor market shocks? creation of funded worker accounts. Simul- • What are the directions for labor market poli- taneously, cost-effective employment services cies moving forward? and retraining programs need to be ramped up to help workers adjust to changes in the labor market. Findings • In the short run, shifting strategy in this way • Employment protection rules create disin- would benefit only those informal sector centives for expanding employment in the workers who �nd jobs in the formal sector. formal sector, especially in India, Nepal, Policy makers need to pay attention to the and Sri Lanka. Job security laws are often vast majority of workers in South Asia— too restrictive, compliance too complicated, disproportionately poor and vulnerable— and enforcement too weak and discretion- who will not. Universal social insurance ary. Labor regulations also tend to fall short schemes that cover all informal sector work- of their intended goal of protecting workers, ers may not be feasible, but coverage could because most workers fall outside the cover- grow organically by encouraging the most age of these rules and employers search for effective among the plethora of plans spon- ways to circumvent regulations. sored by government and civil society. • South Asian countries need to reorient their • Public works, self-employment assistance, and labor market policies from protecting jobs to training programs are also important. All of protecting workers. Doing so calls for coor- these instruments are already in place through- dinated strategies that relax statutory pro- out the region. The priority is to improve their tections (especially the procedures and costs administration and delivery, thereby expand- associated with dismissals) and build up the ing access and increasing cost-effectiveness. The Role of Labor Market Regulations, Institutions, and 6 Programs L abor market regulations, programs, (for example, activation programs), preven- and institutions can have important tion (for example, social insurance), and pro- effects on creating jobs, increas- tection (for example, social assistance). ing productivity and earnings, and helping The focus of this chapter is on policies, workers manage risks of unemployment and regulations, and programs that cover the low and uncertain earnings. However, any � rst two of these functions.1 The discussion assessment of the labor market framework recognizes the distinctions between the for- in South Asia must be based on the reality mal and informal sectors. How are policies that the region is still dominated by infor- affecting the size and quality of employment mal and casual employment. The prevalence in each sector and movement between them? of low-income, precarious, and informal How much access do workers in the two sec- employment reflects the stage of development tors have to programs that could help them characterizing much of South Asia. Strong improve their position in the labor market growth could gradually change the quality and manage labor market-related risks? This of employment in the region, but this process chapter considers the formal and informal is slow, and the experience of other regions sectors separately, beginning with the formal suggests that the relationship between devel- sector. opment and the formalization of the labor market is not as linear as it was once thought to be. Labor market institutions, Given this reality, the policy challenge is policies, and programs in the to reform regulations to encourage the cre- formal sector ation of better jobs while broadening the Labor laws and social security programs access of all workers to programs that help apply largely to regular wage or salaried them manage labor market shocks. Doing so workers in nonagricultural employment. will require choices on regulation of the labor Although the structure of South Asia’s market and design of social protection pro- economies is changing, the labor force is grams for workers that are well adapted to the still predominantly rural and agricultural. A South Asian context. These interventions can majority of workers are either self-employed be seen as a system that includes promotion or work in family enterprises.2 Formal sector 229 230 MORE AND BETTER JOBS IN SOUTH ASIA workers make up only a small minority of in the private sector. In contrast, coverage is the workforce. high among regular wage or salaried work- Nonetheless, the institutional and policy ers in the public sector. In addition to non- framework still has a signi�cant impact on contributory pension schemes, workers in the region’s labor markets. Where they are this sector enjoy considerable job security enforced, labor laws can protect the rights of and, in some countries, signi�cantly higher workers. Job security rules, wage regulations, earnings (see Palacios and Whitehouse 2006 and the financing of social insurance can for pension schemes; see chapter 3 for wage provide important bene�ts to workers, but differences). Coverage of labor laws that set they also affect the cost of labor and, thus, out basic rights of workers, de�ne minimum employment levels in the formal sector. working conditions, and specify contractual Reform of labor law and the strengthening of requirements can be expected to be even employment programs could improve incen- lower than coverage of formal social insur- tives for formal sector employment while ance programs, because most labor regula- increasing the real protection available to tions apply only to � rms above a minimum workers. size threshold.3, 4 Wages, working conditions, and social protection for workers can also be deter- Who is covered by worker protection? mined through collective bargaining. Wage employment accounts for less than half Collective agreements negotiated at the of employment in all South Asian countries � rm, industry, or national level can exist in other than Maldives and Sri Lanka. Even in parallel with legislation, potentially estab- the wage sector, nonregular or casual work lishing more generous provisions than stipu- remains very important. None of these work- lated by law. ers is typically covered by social protection Unions and collective bargaining play programs. In fact, formal social security and a limited role outside the public sector in labor laws cover less than 10 percent of the most South Asian countries. 5 Figures 6A.1– workforce in most South Asian countries 6A.3 in the annex provide estimates of and only a fraction of wage employees. Cov- trade union membership in India, Pakistan, erage of formal social insurance (based on and Sri Lanka, based on administrative pension coverage) is low not only in absolute data. Combined with information on total terms but also relative to per capita gross employment these estimates suggest that domestic product (GDP) in all countries less than 1 percent of workers in Pakistan except Sri Lanka (figure 6.1). Even in Sri (2008), about 2 percent in India (2005), and Lanka, which has the most extensive social just below 10 percent in Sri Lanka (2008) protection system in the region, only about are unionized. Although there are problems a third of all workers are covered by formal with these data, alternate sources provide social security, refl ecting the high share of a consistent message. 6 For countries on self-employed, temporary, and casual work- which data are available, coverage of collec- ers (social security covers about 9 percent of tive bargaining is even lower. For example, workers in India and 4 percent of workers based on a questionnaire administered to in Nepal). the Ministry of Labor in Sri Lanka, an esti- Figure 6.2, based on household survey mated 4 percent of the workforce is covered data, takes a closer look at social security by a collective agreement. Union member- coverage within the wage sector in Sri Lanka ship is higher in the public sector than the and India. In Sri Lanka, coverage is high, private sector (see annex figure 6A.4 for except among casual wage employees. India’s India). Although their overall numbers may profile is more typical of the region as a be relatively low, where unions are present, whole, with only a small fraction of workers they can play important roles in negotiat- bene�ting from formal protection, especially ing benefits and protection of members, THE ROLE OF LABOR MARKE T REGULATIONS, INSTITUTIONS, AND PROGRAMS 231 FIGURE 6.1 Percentage of workforce not covered by formal pension scheme 100 Nepal Bangladesh Pakistan India Bhutan 80 Maldives Sri Lanka 60 percent 40 20 0 0 10,000 20,000 30,000 40,000 50,000 per capita GDP in purchasing power parity dollars Sources: Authors, based on data from Loayza, Servén, and Sugawara 2010; World Bank 2011c. Note: Data are for latest year available. FIGURE 6.2 Percentage of wage employees in India and Sri Lanka covered by social security, by type of worker 100 93 91 80 77 60 percent 51 40 26 23 20 9 1 0 rk ge iva r bl r rk ge pr ge o pu e o wo l wa wo l wa s te ic s ied ag er er ied wa lar r w al a su lar r sa gula sa ula ca g re re India, 2010 Sri Lanka, 2008 Source: Authors, based on data from national labor force and household surveys. Note: Social security coverage in India includes eligibility for pension or provident fund only or in combination with other bene�ts, such as gratuity (a lump sum payable to an employee on termination of employment), health care, and maternity bene�ts. Coverage in Sri Lanka includes employers’ contributions to the pension scheme or provident fund on workers’ behalf. The category of casual workers includes temporary workers in Sri Lanka. In India and Sri Lanka, questions about social security coverage were asked only of wage workers. 232 MORE AND BETTER JOBS IN SOUTH ASIA increasing compliance, and influencing the also reflect the greater likelihood of high- more general policy debate. productivity workers � nding jobs that offer Workers covered under one form of formal protection. In India and Sri Lanka, work- protection tend to be covered under others, ers with higher levels of education are more but most workers have no coverage at all. likely to hold jobs that provide formal protec- This duality is illustrated in table 6.1, which tion; young, old, and less educated workers presents simple correlation coefficients for are less likely to hold such jobs (table 6.2). social security coverage, trade union mem- The higher coverage of women in Sri Lanka bership, and employment in workplaces of 10 likely reflects the high degree of selection of or more employees (as a proxy for coverage more educated women into better jobs. through labor laws) in India. As the matrix shows, all of these indicators of formal social Protection of workers’ protection are positively correlated, with sta- fundamental rights tistically signi�cant coef�cients. There are strong associations between Labor laws play an important role in protect- access to formal instruments on the one hand ing the fundamental rights of workers. The and poverty and overall welfare on the other. core labor standards, reflected in the Inter- The poverty rate for wage workers with national Labour Organization (ILO) 1998 social security is much lower than the rate Declaration on Fundamental Principles and for workers who are not covered (4 percent Rights at Work, represent an internationally versus 24 percent in India in 2005; 3 percent agreed upon group of basic rights relevant versus 29 percent in Nepal in 2003/04). The regardless of the level of development.8 The poverty rate for nonunion workers (22 per- declaration covers four principles and rights, cent) in India was more than twice the rate each covered by two ILO conventions: for union members (10 percent) in 2010. In • Freedom of association and the effec- Sri Lanka, nearly one-third of the workers tive recognition of the right to collective covered by the employment protection law bargaining come from the richest quintile of the popula- • Elimination of all forms of forced or com- tion (Heltberg and Vodopivec 2004). pulsory labor Earnings are higher for workers with • Effective abolition of child labor access to social security or employment pro- • Elimination of discrimination in respect tection.7 Although these correlations may in of employment and occupation. part reflect the effectiveness of formal instru- ments in helping workers manage risk, they Ratification of the eight core conven- tions does not by itself imply implementa- tion, but it does signal political will. As such, it is a useful metric by which to assess TABLE 6.1 Correlation coefficients among union membership, countries’ commitment to protecting basic social security coverage, and employment in firms with 10 or more worker rights. employees, India, 2010 Ratification of the core conventions in Access to Firm size of South Asia has been only partial. Pakistan Union social 10 or more and Sri Lanka rati�ed all of the conventions membership security workers associated with the core labor standards Union membership 1.00     (table 6.3). The other countries in the region Access to social security 0.51a 1.00   (excluding Bhutan, which is not a member Firm size of 10 or more workers 0.19a 0.42a 1.00 of the ILO, and Maldives, which became Source: Authors, based on data from the 2009/10 National Sample Survey. an ILO member only in 2009) have rati�ed Note: Social security coverage includes pensions and provident funds only. Union membership some but not all standards. The conventions includes registered trade unions, memberships in associations of owners, and self-employed workers. It is not possible to isolate trade union membership. on freedom of association and collective bar- a. Pearson correlation coefficient signi�cant at the 1 percent level. gaining and on abolition of child labor have THE ROLE OF LABOR MARKE T REGULATIONS, INSTITUTIONS, AND PROGRAMS 233 TABLE 6.2 Percentage of workers with access to formal protection instruments in India and Sri Lanka, by worker characteristic (percentage of workers) India Sri Lanka 2005 2010 2008 Social In workplace Social In workplace In workplace Worker security of 10+ Union security of 10+ Social security of 10+ characteristic coverage employees membership coverage employees coverage employees All 29 22 11 26 24 51 28 Male 29 22 13 25 24 44 26 Female 28 19 7 27 24 63 31 15–24 years 6 16 5 8 20 42 42 25–44 years 30 22 12 25 25 54 30 45–54 years 54 27 16 42 27 54 20 55–64 years 47 18 12 42 20 39 14 Primary or below 9 14 6 6 15 30 20 Secondary 41 27 14 26 24 58 31 Tertiary 76 57 36 67 50 93 60 Regular wage 45 50 31 44 53 95 87 Casual wage 2 26 5 1 23 10 39 Public 83 70 57 77 67 93 — Private 12 16 8 12 19 34 60 Source: Authors, based on national labor force and household surveys. Note: — = Not available. TABLE 6.3 Year of rati�cation of International Labour Organization core conventions, Declaration on Fundamental Principles, and Rights at Work, by country Elimination of discrimination in Freedom of association Elimination of forced and respect of employment and collective bargaining compulsory labor and occupation Abolition of child labor Convention Convention Convention Convention Convention Convention Convention Convention Country 87 98 29 105 100 111 138 182 Afghanistan a a a 1963 1969 1969 2010 2010 Bangladesh 1972 1972 1972 1972 1998 1972 a 2001 India a a 1954 2000 1958 1960 a a Nepal a 1996 2002 2007 1976 1974 1997 2002 Pakistan 1951 1952 1957 1960 2001 1961 2006 2001 Sri Lanka 1995 1972 1950 2003 1993 1998 2000 2001 Source: ILOLEX database, http://www.ilo.org/ilolex/. Note: Bhutan is not a member of the ILO. Maldives became a member in 2009 and has not yet rati�ed any core conventions. a = Not rati�ed. the poorest rati�cation records. Most coun- or national legislation may broadly recognize tries have indicated their intention to ratify the principles and rights reflected in them.10 all outstanding core conventions, but some In practice, however, limited enforcement dif�culties remain.9 capacity and large informal sectors, among Even in countries that have not rati�ed all other factors, make protecting the basic rights conventions, the constitution, national policy, of workers a challenge in South Asia.11 234 MORE AND BETTER JOBS IN SOUTH ASIA Employment protection legislation and Table 6.4 summarizes the rules for fixed- its effect on labor market outcomes term contracts, based on the World Bank’s Doing Business database.14 Most countries in Employment protection legislation refers to the region take a flexible approach, allowing the rules governing the initiation and termi- the use of �xed-term contracts for permanent nation of employment. These rules cover the tasks and imposing no limits on the duration kinds of contracts permitted, the conditions of such contracts. under which workers can be terminated, and Although the Doing Business database the procedures for termination.12 Employ- indicates no restrictions in India, the situa- ment protection arrangements are often char- tion with respect to �xed-term contracting is acterized along a continuum, ranging from actually more uncertain. The Contract Labor protective to unregulated (or rigid to flexible). (Regulation and Abolition) Act of 1970 is A stylized characterization of this continuum intended to prevent denial of job security is presented in �gure 6.3. where it is feasible (for example, situations Employment protection rules provide job that warrant longer-term employment). In security—but not without controversy. Leg- this regard, it empowers the appropriate gov- islation can provide protection by restricting ernment to prohibit contract labor in some the ability of employers to hire workers on situations. In practice, states in India adopt an explicitly nonpermanent basis or by mak- different stances toward contract labor.15 ing dismissal costly. Controversies surround Pakistan is the most restrictive country in the employment protection legislation because region, prohibiting � xed-term contracts for of sharp differences in views about the costs permanent activities and limiting their dura- and benefits of these policies.13 One per- tion to nine months.16 spective sees restrictions on nonpermanent In contrast, dismissal procedures in much hiring and employer dismissal rights as pro- of the region are among the most onerous viding important social protection for work- in the world. A number of countries rely on ers. Increased employment security can also high administrative and financial costs of encourage longer working relationships and dismissal as the principal means of protect- the accumulation of firm-specific human ing permanent workers. India, Nepal, and Sri capital. An opposing perspective emphasizes Lanka stand out in that they require not only that these regulations raise the cost of labor, noti�cation but prior approval by the state discouraging job creation and favoring privi- to lay off workers (see table 6.4).17 The three leged “insiders.� The impacts of employment countries are among only 33 of 183 countries protection legislation are discussed later in worldwide that require prior approval for this section. individual dismissals and among only 42 that Most countries in South Asia have flexible require prior approval for collective dismissals hiring rules, although there are exceptions. FIGURE 6.3 The continuum of employment protection legislation RIGID OR The Employment Protection Continuum FLEXIBLE OR PROTECTIVE UNREGULATED • restricted fixed-term contracting • unrestricted fixed term contracting • restricted temporary agency work • unrestricted temporary agency work • restricted employer dismissal rights • unrestricted dismissal rights • substantial severance and • no severance or notice required advance notice/approval required • simple administrative procedures • substantial administrative for layoffs requirements for layoffs Source: Authors. THE ROLE OF LABOR MARKE T REGULATIONS, INSTITUTIONS, AND PROGRAMS 235 TABLE 6.4 Selected hiring and redundancy rules in South Asia, by country Hiring rules Redundancy and severance rules Fixed-term Average contracts notice Maximum length of Coverage of Notification Approval prohibited period for fixed-term contracts severance pay required required for redundancy permanent Single 1 9 1 9 dismissal Country  tasks contract Renewals Excluded categories worker workers worker workers (months) Afghanistan No No limit No limit — Yes Yes No Yes 4.3 Bangladesh Yes No limit No limit Firms with fewer Yes Yes No No 4.3 than five workers, managerial positions Bhutan No No limit No limit — Yes Yes No No 5.3 India No No limit No limit Firms with fewer than Yes Yes Yes Yes 4.3 50 workers Maldives No 24 months 24 months — No No No No 4.3 Nepal Yes No limit No limit Firms with fewer Yes Yes Yes Yes 4.3 than 10 workers Pakistan Yes 9 months 9 months Firms with fewer No No No No 4.3 than 20 workers Sri Lanka No No limit No limit Firms with fewer than Yes Yes Yes Yes 4.3 15 workers, members of cooperatives Sources: Authors, based on data in World Bank 2011a and Holzmann and others 2011. Note: The notice period for redundancy dismissal is 4.3–10.3 months in Bhutan and 2.0–8.7 months in Maldives for workers with tenures of 9 months–20 years. For Bhutan and Maldives, �gures show notice period for workers with one year of tenure. For other countries, the notice period is the same irrespective of tenure length. — = Not available. (World Bank 2011a). In addition, severance the requirements for maximum compensa- pay requirements, which typically require tion. Although statutory severance rights are minimum tenure and increase with seniority, generous, nonpayment or partial payment is are high in Bangladesh, Nepal, Pakistan, and common (as it is in some other South Asian particularly Sri Lanka (�gure 6.4). countries).20 Indeed, Sri Lanka has one of the most Benchmarking by the Organisation for costly severance pay systems in the world, Economic Co- operation and Development though workers do not always bene�t from (OECD) suggests that India’s level of statu- it. Sri Lanka’s Termination of Employment of tory employment protection is considerably Workmen Act (TEWA), which governs non- higher than that of most OECD member disciplinary termination in private �rms with countries as well as most other major emerg- more than 14 workers, has heavy reporting ing economies.21 In terms of dismissal protec- demands and requires of�cial authorization tion for permanent workers, India’s labor laws for intended layoffs.18 Retrenchment costs are stricter than all OECD countries except are extremely high by international stan- Portugal and all non-OECD countries except dards (see �gure 6.4), as well as complicat- Indonesia (�gure 6.5). Indian employers in ed.19 TEWA requires that workers be paid establishments with 100 or more workers 0.5–2.5 months’ salary for each year of ser- cannot lay off any worker (or close down vice, depending on the number of years of operations) without permission from the service. The maximum compensation pay- appropriate government. In addition, unfair able is 48 months (capped at Rs 1.25 million), dismissals can lead to reinstatement by the although only very rarely do workers ful�ll court following lengthy court procedures 236 MORE AND BETTER JOBS IN SOUTH ASIA FIGURE 6.4 Weeks of wages required to be paid in severance in regions, country income groups, and selected South Asian countries, by length of service a. International comparators by region 98 100 80 60 54 weeks 47 50 43 43 40 29 33 29 27 25 26 25 16 16 17 21 21 21 20 14 15 12 14 4 4 3 3 6 10 4 11 6 9 2 11 4 4 5 11 0 lo at for h Af ran tin acifi ia Af and on an a l an a be d sia pa di nk es ib an P s ist ist iti d tA hA In Ne a Am c an a en n ha ve per n lad La ric ric rth st ar ca pm io ns De o tio an k an s Sa Pa ut No e Ea t Ea i ng e C eri tra Sr d o - sa gh So b- Ba an ic C i Af an dl Su id om Org M th La on Ec b. International comparators by country income group 98 100 80 60 50 54 weeks 47 43 43 40 32 29 25 26 21 21 21 25 20 16 20 17 20 15 12 11 10 6 9 11 5 3 5 4 4 2 4 4 0 a h tri e tri e rie - nt le- an a l an sia pa nt le nk di un m un m es ist ist ou idd ou idd hA In Ne es es s co inco co inco lad s La rie an k ec m ec m Pa ut ng Sri gh - w- m er- m r- So gh Ba co pe Af lo co w hi in up in lo 1 year 5 years 10 years Source: Holzmann and others 2011. (see annex table 6A.1).22 Although there are (see Freeman 2009). This research �nds mod- no additional costs or noti�cation require- est impacts on employment and unemploy- ments/regulations for collective dismissals, ment, often statistically insigni�cant in the the effective costs of dismissals (the sum of case of unemployment. The empirical �ndings costs for individual dismissals and any addi- are stronger in the case of dynamic effects: in tional costs for collective dismissals) puts general, more restrictive employment protec- India, together with Indonesia, among the tion legislation is associated with lower labor top third of countries included in the OECD turnover, job creation, and destruction; longer benchmarking. India’s employment protec- job tenure; and longer unemployment dura- tion legislation for temporary and �xed-term tion. The research also suggests that there are contracts, though more relaxed, is just above distributional impacts of strong employment the OECD mean. protection rules, with prime-age males and The effects of employment protection leg- skilled workers bene�ting and women, youth, islation on labor market outcomes have been and unskilled workers losing. extensively researched, especially in OECD Studies on the impact of labor laws in countries (for overviews of this literature, South Asia have concentrated on India. 23 A see OECD 2006, 2007b). Studies conducted generally robust set of conclusions emerges: in developing countries, particularly in Latin states with more protective legislation have America, reach broadly similar conclusions significantly lower output, employment, FIGURE 6.5 Employment protection indicators in selected countries OECD employment protection benchmark score 4 3.5 3 0.3 2.5 2.19 1.6 0.6 1.6 0.9 1.2 1.1 1.6 2 1.3 1.1 0.0 0.9 0.8 0.3 0.5 1.5 1.35 0.7 0.9 0.6 1.6 0.0 0.4 0.5 0.3 0.0 0.2 1 0.88 0.1 0.1 0.5 0.5 0.3 0.6 0.3 1.8 1.7 1.5 0.1 0.4 0.5 0.1 0.0 1.4 0.5 1.1 1.2 1.0 1.0 1.1 1.2 0.5 0.6 0.6 0.9 0.7 0.7 0.4 0.5 0.6 0 0.1 a es a m d ile . y ico l ey ain g ca n il a sia k ep ga di an ar az ad in ur tio lan do at fri rk Ch ne ex In ,R Sp Ch tu nm Br bo rm n ra St Tu hA a ng Ca do M r a Ze de Po m Ge De re d Ki ut In ite xe Ko Fe w So d Ne Lu Un ite ian Un ss Ru selected OECD countries selected non-OECD countries protection of permanent workers against individual dismissal (1) speci�c requirement for collective dismissals (2) regulation of temporary employment (3) OECD average individual (1) OECD average individual + collective (1+2) OECD average total (1+2+3) Source: OECD 2009. Note: The OECD employment protection indicators cover three aspects of employment protection: individual dismissal of permanent workers, regulation of temporary employment, and speci�c requirements for collective dismissal. Each subindicator ranges from 0 to 6, with 0 the least restrictive and 6 the most restrictive. The overall indicator—the sum of the three subindicators, weighted at 5/12, 5/12, and 2/12—also ranges from 0 to 6. 237 238 MORE AND BETTER JOBS IN SOUTH ASIA number of establishments, investment, and legislation. Using urban enterprise survey labor productivity in the formal manufac- data for the early 2000s, Gunatilaka and turing sector and higher output in informal Vodopivec (2010) �nd that Sri Lanka has very manufacturing (see Ahsan and Pagés 2009; low job flows relative to other countries (see Gupta, Hasan, and Kumar 2009; OECD annex �gure 6A.5) and suggest that its strin- 2007a). Gupta, Hasan, and Kumar (2009) gent job security legislation may be a signi�- �nd that output growth following the signi�- cant contributing factor. Analysis carried out cant deregulation of the 1980s and 1990s was for this book using a newly available panel of higher in labor-intensive industries in states large (more than 100 employees) manufactur- with more flexible labor regulations. Gains in ing �rms in India �nds higher rates of job cre- employment generation were higher in states ation and destruction (and therefore job real- with more flexible regulations, irrespective of location/turnover) for contract labor relative the labor intensity of the industry. to permanent workers (annex table 6A.2).24 It Ahsan and Pagés (2009) � nd evidence of also �nds higher job turnover for states with strong complementarities across labor laws in more flexible labor regulations (�gure 6.6). India. They �nd that the output and employ- Although labor regulations are not binding ment costs of employment protection legisla- on all �rms in the region, enterprise managers tion are higher in states and time periods in in India, Nepal, and Sri Lanka cite them as which it is more dif�cult to resolve disputes an important constraint to the operation and and not very important when such costs are growth of their businesses. Job-creating �rms low. Workers do not bene�t from these laws, report that labor regulations are the third most because any gains in employment protec- important constraint in Bhutan, the fourth tion are more than offset by the decline in most important constraint in India, and the employment, resulting in a decline in the total �fth most important constraint in Nepal and wage bill (Ahsan and Pagés 2009; Besley and Sri Lanka. In contrast, formal sector �rms in Burgess 2004). Afghanistan and Bangladesh rate labor regu- Evidence from India and Sri Lanka on job lations as among the least problematic. Labor turnover suggests efficiency costs of labor regulations generally become more costly to FIGURE 6.6 Job turnover rates and labor regulations in Indian states 30 25 gross job turnover (percent) Haryana Karnataka Goa Kerala Himachal Pradesh Madhya Pradesh Gujarat 20 Tamil Nadu Uttaranchal Rajasthan Maharashtra Delhi Punjab Chattisgarh Orissa Uttar Pradesh 15 Jharkhand Andhra Pradesh West Bengal Bihar 10 Assam 5 20 30 40 50 60 labor rule reform index Sources: Authors, based on various sources. Gross job turnover data are calculated from the Annual Survey of Industries panel data for 2000–08 for �rms with more than 100 employees. The Labor Rule Reform Index is the OECD index of state-level reform in India, developed on the basis of data collected through a customized survey instrument (Dougherty 2008; OECD 2007a). The proportional index, which captures the state’s share of total reforms, is used here. Note: Higher values of the reform index reflect a greater degree of change in procedures to reduce transactions costs (such as limiting the scope of the regulation, providing greater clarity with regard to application, and simplifying compliance procedures in eight speci�c labor law areas). THE ROLE OF LABOR MARKE T REGULATIONS, INSTITUTIONS, AND PROGRAMS 239 FIGURE 6.7 Cross-country comparison of reported severity of the labor regulation constraint 4 3 severity of constraint 2 Sri Lanka Nepal Maldives Bhutan India Pakistan 1 Bangladesh Afghanistan 0 0 6 7 8 9 10 11 log of per capita GDP in purchasing power parity dollars Source: Carlin and Schaffer 2011 (based on World Bank enterprise surveys). Note: The cross-country regression line shows the relationship between the reported severity of the constraint for a benchmark �rm and the log of per capita GDP. The shaded area is the 95 percent con�dence interval band around the regression line. Vertical bars show con�dence intervals of 95 percent around the reported severity of the constraint for countries in South Asia. The lack of overlap between the South Asian country con�dence interval and the regression line con�dence interval is a conservative test of the statistically signi�cant difference between the reported severity of a constraint for the South Asian country and the average reported severity of constraint for countries at the same level of per capita GDP. The reported severity could still be signi�- cantly different even with overlap. Analysis is based on pooled sample of enterprise surveys conducted between 2000 and 2010. The severity of constraint is rated by �rms on a 5-point scale, with 0 being no obstacle, 1 being a minor obstacle, 2 being a moderate obstacle, 3 being a major obstacle, and 4 being a very severe obstacle. firms as GDP per capita rises (figure 6.7). various channels, including the rules deter- India, Nepal, and Sri Lanka report higher lev- mining collective bargaining, pay equity els of this constraint than other countries at policies intended to eliminate wage discrimi- their level of development.25 nation, and minimum wages. Another way In India, labor regulations were mark- in which public policy can affect compen- edly more problematic in 2010 than in 2005. sation is through rules mandating contribu- Firms in high-income states report greater tions by employers, employees, or both to severity and ranking of the labor regulations social insurance (such as pension plans and constraint than low-income states. Nearly health insurance) and other benefits. The 30 percent of Indian �rms that perceive labor quality and ef�ciency of the insurance and regulations as a major or severe constraint bene�ts � nanced by such contributions are report that restrictions on dismissal are a in themselves major policy issues. However, constraint to hiring, about 25 percent report from the perspective of this book, the rele- that restrictions on employing casual work vance of these nonwage labor costs pertains are constraints, and about 20 percent com- to their labor market implications. Depend- plain about restrictions on temporary work. ing on the details of the contributions and who ultimately pays for them, these contri- butions can affect total labor costs incurred Compensation and minimum wages by the employer (and thus labor demand) or Government policy can affect compen- net earnings of the employee (and thus labor sation—and thus employment—through supply). 240 MORE AND BETTER JOBS IN SOUTH ASIA Where social contributions are large, All countries in South Asia have minimum employment impacts can be substantial. wage rules; some have multiple minimum Labor economists typically measure the wages (see annex table 6A.3). It is important magnitude of social contributions through to consider local labor market conditions in the “tax wedge�—the difference between assessing whether the level set for the mini- total labor costs and net earnings. 26 Most mum wage strikes the right balance between South Asian countries have tax wedges protection and efficiency. In India, where that are below the international average states set the minimum wage, there is wide (see annex � gure 6A.6). However, the tax variation in how it relates to market wages, wedges in India, Nepal, and Sri Lanka although overall the level is low. In a number are high enough that any policy reforms of Indian states, for example, the minimum must take into consideration the poten- wage is less than 20 percent of the median tial employment disincentives of further formal wage. It is highest in Dadra and Nagar increases. Haveli and Delhi, but even in those states it Social insurance systems can also affect was less than 40 percent of the median wage worker mobility. The lack of portabil- in the formal sector in 2008. In most states, ity of bene�ts in some countries (including the minimum wage is below the median wage India, Nepal, and Sri Lanka) with multiple for casual workers (see annex �gure 6A.7). In schemes could inhibit the movement of work- Nepal, Pakistan, and the manufacturing sec- ers between the public and private sector. In tor of Sri Lanka, it is higher as a proportion 2010 (for the public sector) and 2011 (for the of the median formal sector wage (46–61 private sector), Maldives introduced a new percent) (�gure 6.8). In Nepal and Pakistan, contributory pension scheme that is designed it is at or above the median wage for casual to address this issue. The design of de� ned labor, suggesting that the majority of these bene�t schemes can also penalize movement wage workers are unlikely to �nd formal sec- out of covered employment (for example, for tor employment. Although minimum wages in workers covered by India’s employee pension Nepal, Pakistan, and Sri Lanka are not high scheme). by international standards, they approach Minimum wage rules are another inter- the level at which lower-productivity workers vention with potentially important employ- will be effectively excluded from formal wage ment effects. Like employment protection employment, suggesting caution with respect rules, minimum wages can provide security to further increases. to workers adversely affected by market fail- ures. When they are set too high, however, FIGURE 6.8 Minimum wages as a proportion of they can have the unintended consequence median formal sector wages in Nepal, Pakistan, of reducing employment and encouraging and Sri Lanka informality. 100 For the most part, the international litera- ture �nds the employment effects of minimum 80 wages to be modest or negligible (Card and 61 Krueger 1997; Neumark and Wascher 2006). 60 percent 50 46 In developing countries, it suggests that mini- 38 40 mum wages often raise the pay of low-paid workers in the formal sector and can even 20 have a spillover effect on the pay of informal sector wage workers.27 However, the impacts 0 all sectors all sectors manufacturing construction of minimum wages vary considerably, with Nepal 2008 Pakistan 2009 Sri Lanka 2008 examples of negative employment effects in some countries and groups of workers within Sources: Authors, based on national sources for minimum wages and labor countries. force surveys for median wages. THE ROLE OF LABOR MARKE T REGULATIONS, INSTITUTIONS, AND PROGRAMS 241 The main concern regarding minimum returns under each law; and the lack of a sin- wages in South Asia involves the institu- gle system of inspection for labor legislation tional arrangements used to implement (Debroy 2011). them. Labor regulations typically indicate Enforcement is also weakened by collu- that there should be a minimum wage but sion between and corruption of inspectors do not de� ne the mechanisms to set its level and employers. According to World Bank or enforce it. Regulations are also often enterprise surveys, 50–60 percent of firms silent about the structure of the body in in India (2005) and Pakistan (2007) report charge of setting minimum wages, the role that inspectors expect gifts. About 20 percent of employers and employees, and the mecha- of managers in India in 2002 reported that nism to adjust the minimum wage over time, inspectors responded to unof�cial payments creating uncertainty for both employees and by reducing the number of visits to their employers. establishments.31 Implementation is further hampered by inef�cient procedures for resolving disputes. Compliance and dispute resolution In India, Pakistan, and Sri Lanka, options Lack of compliance weakens the impact of for resolving disputes and grievances labor market rules. 28 Firms employ vari- through negotiation appear to be limited, ous strategies to circumvent statutory rules, and the legal framework favors adjudication reducing de facto the regulation of labor over negotiation or collective bargaining. In markets. Incentives for noncompliance can India, for example, the Industrial Disputes be particularly strong where compliance can Act sets the conciliation, arbitration/media- be costly, as in the case of dismissals. Strat- tion, and adjudication procedures to be fol- egies firms use include increasing the use lowed in industrial disputes. If conciliation of contract and temporary workers (India, fails, the government may refer the case Pakistan, Sri Lanka), using voluntary retire- to the industrial tribunal or a labor court, ment schemes for collective dismissals which normally leads to a long litigation pro- (India, Sri Lanka), fragmenting production cess. 32 The backlog means that an average and “leasing� production to smaller � rms adjudication takes three to �ve years (some through subcontracting (India, Sri Lanka), cases have been pending for more than a using disciplinary grounds to justify � ring a decade). In Pakistan, unfair dismissal cases worker or harassing workers to make them can involve protracted processes at labor quit (Sri Lanka), adopting capital-intensive courts and appeals to high courts. In Sri technologies (India), locking out workers Lanka, the process of inquiry in unfair dis- (India); employing foreign workers instead missal cases can take a substantial amount of local workers (Nepal), and terminat- of time and is often arbitrary and nontrans- ing workers during the period of probation parent. Although a 2003 amendment to the (Pakistan). 29, 30 TEWA mandated a 4:3:2 month formula for Weak enforcement is another factor. One labor tribunal cases, arbitration, and deci- problem is that the administrative machinery sions by the commissioner, the process still for enforcement is inadequate. Bangladesh takes a long time, and most cases drag on and Nepal, for example, suffer from short- for two to three years. Poorly resourced ages of inspectors. Nepal faces challenges of institutions and weak capacity are evident in terrain and transport. most countries. Another problem is the sheer complex- Strengthening enforcement and dispute ity of compliance. This problem is espe- resolution capabilities are necessary for cially severe in India, with its multitude of improving compliance, but they require com- labor laws, which are often poorly speci�ed plementary measures. There is consensus in and interpreted as well as contradictory; its most countries on improving the ef�ciency of requirements to maintain registers and file inspections and strengthening enforcement 242 MORE AND BETTER JOBS IN SOUTH ASIA capabilities. Inspection must play a key role employment objectives than the current focus in enforcement, but the sheer number of �rms on protecting jobs. The strategy of protecting (especially small � rms) limits the impact it workers is based on allowing the labor market can have. Improving inspection must there- to function more freely than it does in regimes fore be complemented by other strategies, with strong job protection but giving workers including strengthening complaint and griev- better access to active and passive programs ance redressal mechanisms and streamlining that help them adjust to market fluctuations and addressing the lack of clarity and con- and increase their future employability. sistency in laws and regulations. Options to Table 6.5 contrasts the different approaches reduce �rms’ incentives to circumvent legisla- to worker protection. The challenge for coun- tion include reducing the stringency of such tries in South Asia—like other low-income legislation. These options, on which there is countries with small formal sectors and typically less consensus, are discussed below. indeed many middle-income countries—is to move from the lower-right-hand cell in table 6.5 upward and to the left. Shifting protection from jobs to workers Reducing job protection for formal sec- Easing protective regulations would improve tor workers is dif�cult, because it increases labor market outcomes but weaken job secu- the prospects of layoffs and dismissals from rity for formal sector workers. Less restric- previously protected jobs. Many countries— tive regulations, especially pertaining to including several transition economies of dismissal, could create incentives for formal Central and Eastern Europe—have taken this sector job creation; bene�t certain groups, dif�cult step by combining it with the intro- including women; and encourage compli- duction of better programs to protect workers ance with the law. The minority of workers (Holzmann and others 2011; see Vodopivec, that currently bene�t from these regulations Wörgötter, and Raju 2005 for a discussion would lose from such a reform, especially of the reform paths and challenges in these in India, Nepal, and Sri Lanka, where job countries). security rules are strongest. One aspect of this shift is to extend the Better tools to protect workers, such as legality of nonpermanent contracts. Most income support and active labor market countries have reasonably flexible contract- policies, would provide a more broadly effec- ing rules. Bangladesh, Nepal, and Pakistan tive mechanism for achieving social and set limits on the use of � xed-term contracts, TABLE 6.5 Stylized characteristics of protecting workers versus protecting jobs Protecting jobs (employment protection legislation) Low High Protecting High • Broad coverage of social protection • Some duplication in social protection workers (income • Positive impact on job creation • High cost (regulation plus taxes) support, active • High incidence of short-term and temporary • Incentives for informal activity and low compliance labor market employment • Encourages tenure and human capital investment policies) • High adjustment flows handled through active and • Strict regulation has negative effects on labor market passive programs dynamism and protects insiders Low • Little protection offered • Little social protection support for outsiders • Positive impact on job creation • Incentives for informal activity and low compliance • Precarious employment and weak safety net • Encourages tenure and human capital investment • Unfavorable for productivity growth • Strict regulation has negative effects on labor market dynamism and protects insiders Source: Authors. THE ROLE OF LABOR MARKE T REGULATIONS, INSTITUTIONS, AND PROGRAMS 243 however, and in India, lack of clarity has led who currently bene�t from job security rules. to widely different interpretations, regula- One option would be to grandfather existing tions, and practices across states. Opening bene�ciaries while adjusting provisions for up the use of fixed-term contracting runs new entrants. Another would be to combine the risk of encouraging employers to use this this measure with some form of income sup- form of employment to avoid obligations of port in the event of unemployment and active permanent contracts, but it can also have the programs that help unemployed workers �nd positive effect of shifting workers from casual new jobs. employment relationships (without contracts) The design of such income support pro- to more formal contracted positions. If coun- grams needs to draw on international expe- tries also reduce some of the procedural and riences while making adjustments for income � nancial costs associated with dismissal (as and administrative conditions in South Asia. A discussed below), incentives for avoiding per- number of countries, especially in the OECD manent contracts in legitimately long-term and Latin America, have introduced funded employment relationships could be mini- unemployment accounts (box 6.1). Some mized and formalization of casual employ- countries in Eastern Europe have implemented ment could be the more dominant effect. flat unemployment benefits, which are Combined with strengthened enforce- relatively simple to implement and monitor. ment, minimum income security provi- Studies by the World Bank have looked at sions for contract workers can be improved. design principles, particularly those suited Although pools of temporary contract and for Sri Lanka, a higher-income country with casual workers are generally similar—with some administrative capacity (box 6.2). A key women and young, old, and poorly educated design question involves how income support workers overrepresented in both groups— programs are �nanced. If �nancing is through data from India’s 2010 National Sample social insurance contributions, countries Survey show that social security covers need to be careful that these contributions do more than twice as many contract workers not increase the tax wedge to the point that (28 percent) as casual workers (12 percent). employment is negatively affected. Indeed, increasing protection of contract Income protection could be complemented labor by extending minimum wage and by strengthened active labor market pro- social security requirements, as well as regu- grams, including cost-effective training and lating the conditions of work, is an impor- employment services. Active labor market tant objective of India’s Contract Labor programs can enhance labor supply, create Act. Noncompliance with requirements for labor demand, and improve the function- minimum wage and social security bene�ts ing of the labor market, potentially increas- is widespread, however (NCEUS 2009). ing employment, productivity, and earnings. Opening up the use of contract labor and They can be tools for helping unemployed improving enforcement of basic provisions workers. However, the evaluation literature of income security could increase the share underscores the fact that programs often do of workers with effective basic protection. not meet their intended objectives and are Reducing the costs of dismissal is an often not cost-effective (see, for example, important aspect of reform, but it would need Betcherman and others 2004, 2007). Careful to be accompanied by improvements in sup- design and implementation matter. port for the unemployed. Many stakeholders If it is responsive to the real needs of the support reducing severance pay (in Sri Lanka) labor market, training could potentially and termination-related procedural costs be an important support for unemployed (in India, Nepal, and Sri Lanka) in order to formal sector workers. Training can help bring them more in line with international unemployed workers upgrade their skills arrangements. More universal support would or learn new ones. However, programs in require steps to ensure protection of workers many countries are not responsive to the 244 MORE AND BETTER JOBS IN SOUTH ASIA BOX 6.1 Severance reforms in Austria and Chile In 2003, Austria converted its severance pay to This scheme has reduced employers’ severance a fully funded, contributory system. The reform payment obligations, which are being partly repla- extended entitlement to workers with short tenures ced by the new unemployment insurance system. and removed obstacles to worker mobility, granting The system is effectively funded with individual full portability and allowing the accumulation of accounts managed by a freestanding administra- bene�ts from the beginning of the employment spell. tor selected through a competitive tender. To sti- Employers contribute 1.5 percent of the payroll to mulate reemployment, the scheme requires bene�t each worker. Resources are held in a central account recipients to � rst draw resources from their own and invested in the capital market. Laid-off workers accounts; when they deplete these funds, they with job tenure of three or more years can withdraw draw from the solidarity account. Withdrawals accumulations from their accounts or keep them from individual accounts are triggered by separa- and claim them upon retirement. (Workers who tion from the employer, regardless of the reason. separate voluntarily and workers with tenures of less Withdrawals from the common fund are triggered than three years are denied the right of immediate by insufficient resources in individual accounts. withdrawal.) Claimants must satisfy the usual eligibility condi- In 2002, Chile introduced a new unemployment tions under unemployment insurance (not working insurance system that combines social insurance and being available and searching for a job). They with self-insurance. Both workers and employers are limited to two withdrawals over a five-year pay contributions. Unemployment contributions are period. Bene�ts are linked to past earnings, with a split between individual accounts and a common declining schedule. solidarity account, which is partly � nanced by the government. Source: Holzmann and others 2011. occupational demands of the labor market, livelihoods of the poor (these programs are and curricula are often not in tune with the discussed in the next section). No serious requirements of � rms. International experi- evaluation evidence exists on the impact of ence has shown that training and retraining training programs in the region, but there programs need to have a strong connection appear to be many obstacles that likely to the employer community for design and affect their success. Many programs are implementation. Establishing such a connec- more about life skills and basic job readiness tion can be done in various ways, including than vocational skills training (Rahman promoting competition among trainers, cre- and Shamsunnahar 2011). Another concern ating performance incentives through out- is the lack of correspondence between the come-based fees for suppliers, and involving training provided and the needs of employ- employers in the planning of training, the ers. To strengthen the relevance of training, supervision of trainers, and the direct provi- some countries have initiated public-private sion of training. partnerships. The most prominent example South Asian countries have a variety of is India’s National Skill Development Cor- training programs for unemployed workers. poration, which provides gap � nancing to Training is delivered by government train- trainers in 21 key manufacturing and ser- ing institutes, enterprises, nongovernmental vice sectors. organizations (NGOs), and commercial Employment services have generally been training companies. Many programs, par- found to be a cost-effective active labor ticularly those funded by governments and market policy. These services include job international donors, seek to improve the brokerage, labor market counseling, testing THE ROLE OF LABOR MARKE T REGULATIONS, INSTITUTIONS, AND PROGRAMS 245 BOX 6.2 Unemployment bene�t proposals for Sri Lanka Analysis by Vodopivec (2004) and Abidoye, Orazem, bene�ts (for example, a maximum potential dura- and Vodopivec (2009) identi�es design principles for tion of 6–10 months, with a replacement rate of moving to a more effective protection system for the about 50 percent), to take into account the fact unemployed in Sri Lanka. Their proposals take into that potential bene�ciaries would be from better- account short-term political economy needs as well off segments of the population. as the need for a longer-term transition to an unem- • Workers would be able to access their accounts on ployment bene�t program. Key elements include the becoming unemployed. following: • Existing administrative systems—such as the Employees Provident Fund network—would be • Starting eligibility conditions and coverage would used to pay out bene�ts into bene�ciaries’ accounts be essentially the same as under the Termination (to address the administrative costs of setting up of Employment of Workmen Act, with retrenched new systems). workers in �rms with more than 14 workers recei- • Unused funds remaining in individual accounts ving both severance and unemployment bene�ts. on retirement would be transferred to pension (Workers whose employers went bankrupt would accounts. at least get unemployment benefits, even if they • Employment services would be provided (through were unable to get TEWA bene�ts.) JobsNet, for example). • The continuing eligibility requirements (other than employment in the formal sector) found in OECD In the long run, employers’ obligations under programs would be eliminated. Recipients would TEWA could be fully transformed into their regular be automatically entitled to a maximum amount monthly contributions to the individual accounts of of bene�ts, paid as a lump sum. This feature takes workers and coverage extended to all workers cove- into account Sri Lanka’s large informal sector, red by the Employees Provident Fund scheme, which in which workers self-protect through informal does not have a size-related threshold. The insurance employment, and the associated challenges and function could be improved through the addition of costs of monitoring. a social insurance component to supplement indivi- • Employee or a combination of employer and dual savings for workers who exhaust their accounts employee contributions would be made to indi- (see the example from Chile in box 6.1). vidual worker accounts, which are transferrable across enterprises, based on a modest level of Sources: Abidoye, Orazem, and Vodopivec 2009; Vodopivec 2004. and assessment, and job search assistance. There is considerable room for improv- Evaluations from developed and developing ing the effectiveness of national employment countries suggest that employment services services. India’s national employment service are among the most cost-effective mecha- runs almost 1,000 employment exchanges, nisms, especially under favorable economic which register job seekers, notify them about conditions (Betcherman and others 2004). vacancies, collect and disseminate labor When labor demand is sluggish, employ- market information, and provide vocational ment services need to be combined with guidance to students and youth. The limited other interventions. In developing coun- information available suggests that most (but tries, where employment agencies are weak not all) of these exchanges—like exchanges and most hiring is done through informal elsewhere—are of limited effectiveness channels, considerable capacity needs to (World Bank 2010). In Nepal, the Depart- be built for public employment agencies to ment of Labor and Employment Promotion be effective. The market for private employ- operates skills development of�ces for regis- ment agencies also needs to be further tering job applicants and provides labor mar- developed. ket information. 246 MORE AND BETTER JOBS IN SOUTH ASIA There are promising models and emerg- delivered by government, donors, NGOs, ing good practices in the region, all of which or the private sector. Indeed, South Asia include a central role for information tech- has a rich mix of informal worker associa- nology in reaching jobseekers and employers tions and trade unions, self-help groups, and and in providing detailed information for other membership- and community-based employers to review. There has also been an organizations, all of which play important increase in job fairs, which enable employers roles in the labor market (box 6.3). The role and workers to engage directly. Employment of labor market institutions and programs exchanges in Gujarat, India, and in Sri Lanka in the informal sector is of great relevance have strong linkages with private employers. in South Asia, given that this sector employs In Gujarat, a group of NGOs provides local the vast majority of workers. labor market information. The social protection challenge in low-in- The state of Karnataka has set up India’s come countries is how to cover the gap in pro- � rst employment exchange based on a pub- tection for informal sector workers while not lic-private partnership. Under this arrange- constraining incentives for creating income ment, the state government and TeamLease from informal activities. This gap covers Services, a staffing solutions company, all aspects of the social protection system, have created the Karnataka Employment including activation programs, which pro- Center. mote opportunities for workers; social and In Sri Lanka, JobsNet—which intends to self-insurance plans, which prevent declines set up a wide network of job service centers in income from shocks; and safety net pro- and is one of the largest active labor market grams, which protect against destitution. services in terms of registrations—provides Countries in South Asia rely on three types intermediation, career guidance and counsel- of programs to support workers in the infor- ing, and job search assistance. It is � nanced mal sector: public works, training, and sup- by a commission of employers. It receives port for self-employment/small business. 33 special funds from the government treasury These interventions play both an activation for programs that target youth and inter- (promotion) and a safety net (protection) nally displaced people. JobsNet’s board of role. There are also efforts to extend social directors—which has representation from insurance (prevention) options to informal government, the private sector, and trade workers—the thrust of proposals made by unions—is responsible for framing policy pri- the National Commission for Enterprises in orities and overseeing the implementation of the Unorganised Sector for minimum univer- programs. Several private �rms are involved, sal social security in India (NCEUS 2009), as both job providers and providers of pri- for example. The different approaches need vate sector perspectives on industry needs. A to be considered in terms of their effects on transparent database facilitates monitoring of adequacy of protection, ef�ciency, function- the program. ing of the labor market, and productivity. Labor market institutions, Public works policies, and programs in the Public works have a long history in South informal sector Asia and continue to be an important instru- Labor market regulations, institutions, and ment for activation, insurance, and safety programs are important even in sectors that nets in the region. A global review of public are informal: public works, self-employment works programs � nds that South Asia has assistance including micro�nance, and many more participants than any other region (del training programs are oriented to workers Ninno, Subbarao, and Milazzo 2009). Most who are outside the formal sector. These and countries in the region have public works other types of programs may be � nanced or programs, which provide cash or food in THE ROLE OF LABOR MARKE T REGULATIONS, INSTITUTIONS, AND PROGRAMS 247 BOX 6.3 Nongovernment players in South Asia’s informal labor market An impressive range of NGOs provides various 1970s. BRAC, the largest NGO in the world, has forms of protection to workers in informal labor an extensive micro� nance operation as part of its markets in South Asia, carrying out functions wide-ranging development activities. India’s Self- that are typically the responsibility of government Employed Women’s Association (SEWA) is a leading agencies and trade unions in the formal sector. example of an organization devoted to improving the Informal worker associations, micro� nance institu- security and self-reliance of workers in the informal tions, self-help groups, and other membership- and sector. The organization—actually a registered trade community-based organizations are involved in the union—provides a broad range of services, including full range of worker protection activities. They pro- micro� nance and banking services, training, insur- mote income-generating activities, largely through ance, housing, and lobbying and negotiating for bet- microfinance services and training, and mitigate ter wages and working conditions. labor market– related shocks, through social and These (and some other) organizations operate on self-insurance plans and by advocating for the basic a very large scale and have major impacts. Many rights of informal workers. Programs target the poor smaller organizations also provide services to the hun- and vulnerable, often with a focus on women. dreds of millions of South Asian workers who work Some of South Asia’s organizations are well beyond the reach of labor laws, statutory social insur- known internationally. Grameen Bank has been a ance programs, and formal collective agreements. pioneer in micro� nance, offering credit and related banking services to the poor in Bangladesh since the Source: Authors. return for work on projects typically devel- works since the 1950s. Bangladesh has �ve oped by governments (at the national, sub- programs that directly create employment national, or local level); donors; or NGOs. through public works. The emphasis is on Various public works designs exist across the rural poor. The program with the largest the region (table 6.6). Public works are an number of participants is Employment Gen- important tool in the reconstruction of eration for the Hard Core Poor, which served Afghanistan, for example, where programs 1.8 million beneficiaries between October include food-for-work programs, supported 2009 and May 2010. This program oper- by the World Food Programme, and the ates in 16 poor districts, with the objectives National Rural Access Program, which of creating temporary employment, reducing reconstructs and maintains rural infrastruc- poverty, and contributing to rural develop- ture. Nepal has implemented a range of ment. The program provides up to 100 days public works activities, some reaching out of wage employment to one person per poor to the remote communities. In some cases family. This model is similar in some ways (for example, the Poverty Alleviation Fund), to India’s Mahatma Gandhi National Rural public works have been part of larger pro- Employment Guarantee Act (MGNREGA). grams with various components. The Rural Bangladesh also has a food-for-work pro- Community Infrastructure Works Program gram, which has the largest budget of all is a food-for-work program operated jointly public works programs. with the World Food Programme. Pakistan India has had a succession of public works and Sri Lanka have provided some pub- programs over the past half century. The lic works programs, but public works have national and state governments, NGOs, and been used less there than elsewhere in the the private sector have all been, and continue region. to be, involved. The most prominent program Bangladesh and India, where public works is the Mahatma Gandhi National Rural are most prominent, have been using public Employment Guarantee Program, created 248 MORE AND BETTER JOBS IN SOUTH ASIA TABLE 6.6 Selected public works programs in South Asia, by country Number of participants Country Program Target population Benefits (latest year available) Afghanistan Protracted Relief and Food insecure in rural areas, Food, basic education, and 700,000 Recovery Operations women, internally displaced training for girls and women (with World Food Programme) people, excombatants National Rural Access Program Rural unemployed Short-term employment, 1,435,000 labor days training for excombatants Bangladesh National Service Program Unemployed rural youth Wages and training 13,000 and women with secondary education and above Rural Employment Poor rural female household Wages (portion put into savings 5,000 Opportunities for Public heads account for future Assets self-employment) Employment Generation for Rural poor Wages 1,800,000 the Hard Core Poor Rural Employment and Road Poor rural women Wages (portion put into savings 10,000 Maintenance Program account for self-employment in future), training, other supports for self-employment Food for Work Food insecure in rural areas; Food 56,000 people affected by disasters India National Rural Employment Rural poor Wages 41 million rural households Guarantee Program Swarna Jayanti Urban poor and unemployed Wages 71 million person-days under Shahari Rozgar wage program Yojana (Urban Wage Employment Program) Nepal Rural Community Food insecure in rural areas Food 30,000 households from Infrastructure Works food-insecure regions Support Activities for Poor Households affected by Food 30,000 Producers of Nepal (SAPPROS) drought Social Safety Net Project Households in chronically Food 168,000 food-insecure districts Pakistan People Works Program Poor and unemployed youth Wages 228,000 Source: Authors’ compilation. in 2006 under MGNREGA. With its rights- work has been higher than other programs, based design—which guarantees 100 days of qualitative evidence and �eld studies indi- work annually for any rural household will- cate that there is substantial unmet demand ing to perform unskilled manual labor, at a for MGNREGA work in some states. 34 rate no more than the wage speci�ed by the About one-fifth of rural households who scheme—MGNREGA is being watched with sought work in the program did not get interest (box 6.4). it (table 6.7). This evidence, together with MGNREGA has created much more field monitoring reports from the Profes- employment than other Indian public works sional Institutional Network, suggests that programs have, but the employment guar- the employment guarantee does not always antee has not been fully implemented. In operate in the way set out in the act.35, 36 2009/10, the program provided employment Evidence suggests that MGNREGA ben- to about one-fourth of rural households e�ciaries tend to be poor and that the pro- (table 6.7). Although the number of days of gram is largely successful in targeting the THE ROLE OF LABOR MARKE T REGULATIONS, INSTITUTIONS, AND PROGRAMS 249 BOX 6.4 Key features of India’s Mahatma Gandhi National Rural Employment Guarantee Act Under India’s Mahatma Gandhi National Rural act also provides for childcare facilities for children Employment Guarantee Act, the government guar- under six if more than �ve such children are pres- antees 100 days of work per rural household per ent at a worksite. Together with the provision that year. Each rural household is entitled to a free job the job be within �ve kilometers of the worker’s resi- card with photographs of all adult members living dence, these features make MGNREGA work a fea- in it. A job card holder may then apply for employ- sible option for women. ment. The government must provide employment The act focuses on labor-intensive rural develop- within 15 days. If it fails to so, a daily unemploy- ment works, with a focus on water and irrigation ment allowance has to be paid to the applicant. activities as well as connectivity. In addition to the Work must be provided within �ve kilometers of the list of speci�c types of works allowed under the act, applicant’s residence. If it is not, the worker receives there is a provision for other types of works based a 10 percent premium over the scheme wage. How on consultations between the state and the central the household distributes the 100 days among its government. Overall, the scheme maintains a 60:40 members is entirely the household’s decision. The labor to capital ratio. Contractors and machines are scheme is therefore completely self-targeted and fol- explicitly banned. lows a demand-driven, rights-based approach. Panchayati Raj institutions (PRIs) and communi- The central government in the union budget of ties play a central role in implementing MGNREG. 2009/10 promised to provide a real wage of Rs 100 Unlike previous public works and social protection a day (indexed to state-speci� c infl ation) for work programs, PRI leaders and communities are sup- performed under MGNREGA. Payment is based on posed to identify the list of works through discus- a rural schedule of rates that depends on the amount sions in the gram sabha (village council). PRIs also of work performed by a person. This schedule of participate in implementation (at least half of works rates should be set such that able-bodied workers by value are to be implemented through the gram should be able to produce the output that could earn panchayat [local self-governments in villages and them the program wage. small towns]), supervision, and monitoring of works Wages are paid in cash—a departure from previ- (including through social audits). This program fea- ous public works programs, which typically included ture represents a signi�cant shift from earlier public a food component. In a move to counter corruption work schemes which were based on central rules, in forging job cards and tampering with muster with PRIs involved only in supportive implementa- rolls (registries of eligible households) and improve tion or distribution functions. transparency in wage payment, the government To implement MGNREG, the act calls for dedi- announced that as of April 1, 2008, all MGNREGA cated administrative structure. Unlike previous pub- wages would be paid directly into worker bank or lic works programs, there is a speci� c provision of post of�ce accounts, opened free of charge. The idea administrative costs, which are borne by the center was that separating the implementing agency from and supplemented by states. There is also provi- the payment mechanism would reduce room for sion for a dedicated cadre of MGNREGA of� cials fraud and harassment. and functionaries at the district, block, and gram Adequate worksite facilities are to be provided. panchayat level responsible for implementing the To facilitate manual labor, the act mandates provi- program. sion of certain basic facilities at the worksite, includ- The act emphasizes accountability, through the ing provision of shade, drinking water, crèches for use of information and communication technology children, and � rst-aid facilities. tools and reliance on communities and third-party Several provisions of the act favor the participa- monitoring. Management information systems tion of women. The act mandates that one-third have been established to handle administrative data of workers be women. It ensures equal wages for on employment generated and assets created at the men and women, with women typically having to lowest level. Field-based monitoring is supposed perform less work to earn the program wage. The to be performed by national field-level monitors (continues next page) 250 MORE AND BETTER JOBS IN SOUTH ASIA BOX 6.4 Key features of India’s Mahatma Gandhi National Rural Employment Guarantee Act (continued) reporting to the central ministry, as well as local to a maximum limit), working on an assumed 60:40 vigilance and monitoring communities. Some labor to capital ratio. States are responsible for pro- states are also using community-based monitoring viding unemployment allowances should they fail to through social audits. provide work within 15 days of it being demanded. The central government bears 90 percent of the This penalty is designed to create a strong incentive total cost of the program. It covers all wage costs for the state government to provide work. and three-quarters of the nonwage component (materials and most administrative expenses, subject Source: World Bank 2011b. TABLE 6.7 Participation in the Mahatma Gandhi National Rural Employment Guarantee Program, by consumption quintile, 2010 (percentage of households, except where otherwise indicated) Mean person Holds Sought but did days for households Mean person days MGNREGA Reports working in not obtain that worked in for all rural Quintile job card MGNREGA program MGNREGA work MGNREGA program households 1 (poorest) 47 34 25 34 11 2 41 30 22 36 11 3 39 27 21 38 10 4 33 23 18 40 9 5 22 14 14 40 6 All 35 24 19 37 9 Source: Authors’ estimates, based on the 2009/10 National Survey Sample. disadvantaged. 37 Nationally, households in and localities to design projects that are poorer quintiles are considerably more likely accessible to women. Scheduled castes and than households in richer quintiles to be reg- tribes are also well represented in the pro- istered with and report employment through gram, although there is some evidence that the program (table 6.7).38 Rural households their share of total work declined in recent in the poorest quintile report about twice as years.40 many days of MGNREGA work as house- The design of MGNREGA partially holds in the richest quintile. However, a addresses some of the traditional problems small but nontrivial share of MNREGA with public works, including the lack of com- work opportunities go to better-off house- munity participation, the leakage of funds holds, and the number of days of work is and related forms of corruption, and the lack actually higher for participants from richer of monitoring and evaluation (see box 6.4). quintiles.39 However, an implementation review reveals Administrative data show that women problems with the issuance of job cards; account for almost half of work days funded manipulation of program data; discrimina- by MGNREGA (figure 6.9). Although tion against women, scheduled castes and households determine who will participate, tribes, and people with disabilities; fraudu- the program sets targets for the share of lent payments and anomalies; and slippage jobs held by women (box 6.4). To achieve in execution of work undertaken (NCAER– this target, MGNREGA encourages states PIF 2009). Moreover, concerns remain about THE ROLE OF LABOR MARKE T REGULATIONS, INSTITUTIONS, AND PROGRAMS 251 the poor quality of the assets created, which FIGURE 6.9 Participation of women and members of scheduled critics cite as indicative of a primary focus on castes and tribes in the Mahatma Gandhi National Rural employment generation at the expense of the Employment Guarantee Program, 2006/07–2010/11 creation of productive assets and improved (share of person days of employment provided) future livelihood opportunities.41 The effec- tiveness with which MGNREGA has been 100 implemented has varied considerably across 90 states. 80 MGNREGA has potentially important effects on the functioning of rural labor mar- 70 kets. A concern is that it may be affecting 60 labor supply by offering wages above what percent 50 49 would normally be offered for unskilled labor. 45 47 Given the scale of the program, this issue is 41 43 40 36 important. Although there is anecdotal evi- 31 29 30 dence that in some areas MGNREGA wages 25 27 25 21 22 may be discouraging labor supply outside the 20 17 18 program (in, for example, labor-importing 10 states like Punjab and Haryana [Wall Street 0 Journal 2011]), there is no systematic evidence women scheduled scheduled with which to rigorously assess this claim. castes tribes This issue should be carefully analyzed. 2006/07 2007/08 2008/09 2009/10 2010/11 Public works can provide an effective safety net for poor and vulnerable house- Sources: World Bank 2011b and MGNREGA website (http://nrega.nic.in). holds. The evidence both internationally and in the region suggests that public works can have a positive impact on bene�ciaries’ chances for about half of participants but (short-term) income and ability to smooth does not improve earnings (Betcherman and consumption (del Ninno, Subbarao, and others 2004).42 Milazzo 2009). Could these programs play a greater role in The cost-effectiveness of public works as enhancing the labor market prospects of par- a safety net is enhanced when a number of ticipants? Some programs attempt to activate design and implementation-related features participants in various ways. One way is to are incorporated. These features include combine work with training. Another, used encouraging self- targeting (by setting the in a few programs in Bangladesh, is to allo- wage or in-kind rate appropriately); estab- cate a portion of wages to a savings account lishing the optimal labor intensity; respond- used to support future self-employment. ing to seasonality variations; exhibiting The lack of evaluation evidence makes gender sensitivity; maintaining transparency it dif�cult to empirically establish whether and accountability; and ensuring commu- these, or other, design features are asso- nity involvement in project selection. (See del ciated with success. In any event, policy Ninno, Subbarao, and Milazzo 2009 for an makers need to be cautious in choosing the elaboration of these points.) employability-related objectives they estab- Less clear is whether and how public lish for public works programs. Experiment- works can be an effective “activation� strat- ing with different activation designs would egy that helps workers improve their employ- be worthwhile, especially in urban labor ability and income-earning potential. The markets as the region continues to urban- limited evaluations on the effects of public ize. Better monitoring and evaluation will works on employability suggest that partici- be essential to identify and then disseminate pation improves postprogram employment good practices. 252 MORE AND BETTER JOBS IN SOUTH ASIA of small enterprises and income-generating Training activities (such as textile manufacturing, tai- Even under optimistic scenarios of educa- loring, knitting, carpentry, batik painting, tional attainment, the majority of prime-age plumbing, house wiring, and radio and TV and older workers in South Asian countries maintenance). Pakistan’s Training for Rural other than Sri Lanka will continue to have Economic Empowerment program offers only primary education or less over the next vocational, entrepreneurial, managerial, and 20 years (see chapter 5). Workers with little literacy/numeracy skills to 2,400 trainees. education will likely continue to hold infor- Training institutions play a limited role mal jobs, suggesting both the importance of in improving skills in the informal sector. developing initiatives to help workers in the In most of South Asia, the mandate of for- informal sector and the challenge of doing mal training institutions includes training so with workers with a low skills base. workers for the informal sector, but evidence Most public training agencies and private shows that they are rarely successful in doing sector trainers focus on the formal sector. so. In India, for example, only 12 percent Although some initiatives are in place that of graduates of formal training institutions seek to improve the skills of informal work- enter the informal sector (World Bank 2007). ers, until recently they have been ad hoc in Although similar numbers are not available nature, making it difficult to estimate the from other countries, anecdotal evidence number of workers they cover. from Bangladesh and Nepal suggests that Many training initiatives do cover infor- they face a similar situation. mal sector workers, but the numbers involved Overall, current efforts through formal are small. In Afghanistan, vibrant private programs and training institutions are inad- sector and NGO training activities exist for equate to meet the training needs of the large the informal sector, although it is extremely workforce in the informal sector, many of dif� cult to assess the size of these efforts. whom have never been to school. Because In India, several initiatives provide skills to of its entry requirements and concentration the informal sector. Community polytech- in urban areas, the formal training system is nics provide skills training within commu- not designed to offer skills to low-educated nities, with no entry prerequisites. The Jan people, particularly people in the rural non- Shikshan Sansthan (Institute for People’s farm sector. Education) focuses on unskilled workers in What distinguishes some informal sector both rural and urban areas, offering courses employment from formal wage employment ranging from candle and incense stick mak- is the breadth of tasks that need to be per- ing to sewing and embroidery to computers. formed, especially by self-employed workers, India’s National Institute of Open School- who typically need to complete speci�c jobs by ing provides skills training to disadvantaged themselves, from beginning to end (table 6.8). youth. The Ministry of Labor and Employ- These self-employed workers must perform ment’s Modular Employable Skills program a variety of functions, including conduct- is also geared toward the informal sector. It ing initial market surveys, maintaining cost provides vocational training to school leav- and quality control, and handling � nancing ers, existing workers, graduates of industrial and marketing. Technical and business skills training institutes, and others to improve are important in enhancing productivity and their employability by optimally using the the quality of goods and services produced. training infrastructure available in public Improving skills will strengthen the ability to and private institutions and within industry. compete. Complementary services (access to Nepal’s Department of Cottage and Small credit, technology, markets, and information) Industries and the Small Industry Develop- appear to lead to better outcomes.43 ment Board provide skill-oriented training Reorienting public training institutions is programs and assistance in the operation probably not the best solution to meeting the THE ROLE OF LABOR MARKE T REGULATIONS, INSTITUTIONS, AND PROGRAMS 253 TABLE 6.8 Skill requirements in the informal sector Skill area Training needs Technical • General upgrading of technical skills used in trade • Improved knowledge of materials used in trade • Practical ways to reduce waste of materials • Basic reading of designs and drawings • Repair of own equipment • Skills required for new product designs • Understanding of more advanced equipment and improved technologies • Basic knowledge of industrial production techniques Management • Costing, pricing, and related aspects of financial administration • Various aspects of marketing, including rudimentary market research • Customer relations, including creation of a customer data base • Division of labor in the workshop and personnel management • Input stock planning • Quality control • Workshop layout • Legal and fiscal regulations Literacy and numeracy • Functional language skills and higher educational attainment (to enhance trainability) Other • Knowledge of recent technological developments in the trades • Improvement in the teaching skills of master craftspeople (to increase the effectiveness of the training) • Ability to work cooperatively (why and how to work together, informally, or as a trade association) Source: Johanson and Van Adams 2004. training needs of the informal sector. Public with NGOs to provide skills training in institutions would �nd it dif�cult to make the diverse areas (such as business development, changes and serve both the formal and infor- horticulture, food processing, and livestock) mal sectors. Major investments would be to farmers, underemployed youth, and poor needed to upgrade facilities and equipment, women. In almost all provinces, people who develop and retain new staff, and design cur- received training reported signi�cantly higher ricula and materials to provide the package wages than those who did not. These positive of skills described above. Such improvements impacts likely reflect the fact that the NGOs would require far greater resources than are providing tailor-made training catering to are currently provided as public subsidies. the needs of the local labor market. Increased training fees would not generate A significant—and undocumented— much income in view of the lack of clear ben- amount of training in the informal sector e�ts of past training and constraints on fam- occurs through informal apprenticeships and ily incomes of the target groups. learning by doing. The model for this train- Trainers from local NGOs are often more ing is a simple one: an experienced tradesper- effective in providing services that meet the son or master craftsperson teaches a trade to needs of the informal economy. Anecdotal younger apprentices and workers. evidence on the impact of these programs is More young people acquire competence positive, although only a few programs have through traditional (informal) apprentice- been evaluated. ships than would be possible through the One such program that has been carefully formal education system. Although data for evaluated is the Afghanistan Skills Develop- South Asian countries are not available, �g- ment Program. In view of its own inability to ures for other countries with informal sectors provide services in hard to reach and insecure of similar size suggest that 50–70 percent of rural provinces, the government contracted employees in microenterprises in India may 254 MORE AND BETTER JOBS IN SOUTH ASIA have obtained their skills through apprentice- involvement in skills development.44 NSDC ships (World Bank 2007). will perform three key functions: Effective informal apprenticeships have • Stimulate private sector training provid- several advantages. They provide trainees ers, by providing seed capital and viability with flexible and dynamic skills, are self- gap funding. regulating, cost the government virtually • Facilitate the direct involvement of private nothing, and require little initial skill or sector � rms in the development of train- experience, just a willingness to be trained ing programs and curricula, in part by by a master, often a relative. But they also supporting sector skills councils in prior- have important limitations. They are based ity sectors on traditional technologies and ideas from • Focus on the informal sector, underprivi- previous generations, and the quality of leged segments of society, and “backward� training is only as good as the skills of the areas, in which residents have not received master and his or her willingness and ability adequate opportunities to acquire skills. to pass them on. The theoretical aspect of learning is weak or absent; only the simplest Although evidence is sparse on effective skills are learned, resulting in low-quality training programs for informal workers, products or services. Apprentices also often some international experiences show some lack exposure to modern training systems promise. One successful approach has been and technologies as well as the innovative Mexico’s Integral Quality and Moderniza- aspect of learning. The range of skills taught tion Program (CIMO, later renamed PAC tends to be narrow, limited to a particular and then PAP). Like India, Mexico is domi- product or phase in production. These skills nated by small and medium-size � rms. The may quickly become outdated in rapidly program shows that partnerships between changing labor markets. the public and private sector to provide Governments may not be the best deliver- training and a range of support services ers of training for informal workers, but they can enhance the productivity of workers in have an important role to play as facilitators. the informal sector (box 6.5). The program Instead of delivering training themselves, was successful not only because the govern- governments could focus on creating an envi- ment facilitated the provision of training to ronment to support nonpublic providers by small and medium-size enterprises but also performing the following roles: because enterprises were provided with an integrated package of services, including • Establishing a policy framework (regula- information on technology, new production tions and incentives) processes, quality control techniques, mar- • Supporting curriculum development, the keting, and how to run a business. training of trainers, and competency- Some public-private partnerships have based skills testing improved the quality of informal apprentice- • Stimulating investment through tax ships. Their strategy focuses on upgrading incentives or � nancial support, in order both the technical and management skills to increase the capacity and the quality of and the pedagogical skills of masters. Tra- training ditional apprenticeships can be linked with • Revising apprenticeship acts that are out- specialized training providers or master dated and contain regulations that ham- craftspeople, with the government acting as per enterprise-based training. facilitator. There are examples of innovative An interesting concept in this regard is � nancing as well, as exempli�ed by Kenya’s India’s National Skills Development Cor- Jua Kali project, in which vouchers for train- poration (NSDC), set up in 2009 as a non- ing are distributed to informal sector entre- profit company to promote private sector preneurs (box 6.6). THE ROLE OF LABOR MARKE T REGULATIONS, INSTITUTIONS, AND PROGRAMS 255 BOX 6.5 Mexico’s proactive approach to supporting small and medium-size enterprises Mexico’s Integral Quality and Modernization enterprises face multiple constraints, training alone program, known as CIMO, is a good example of may not enhance their productivity. In recognition a comprehensive program that assists small and of those constraints, CIMO provided an integrated medium-size enterprises and has a sound embedded package of training and consulting services through monitoring and evaluation effort. Launched in 1987, independent private instructors (while promoting CIMO was one of several Mexican programs tar- � rms’ participation in de� ning their own training) geting small and medium-size enterprises. It focused as well as a range of industrial extension services. on training and consulting by specifying its goal to Another noteworthy feature of the CIMO/PAC improve productivity by increasing training levels, program is its monitoring, evaluation, and data col- helping design training plans, and expanding the lection. The collection of panel data allows for rig- availability of training. orous evaluation of these interventions. Evaluations The program, administered by the Ministry of of CIMO before its decentralization � nd that it was Labor, was decentralized to the state level in 2002 effective in improving the performance of targeted under the name Training Support Program (Programa companies. Compared with � rms in a control group, de Apoyo a la Capacitación [PAC]) and in 2009 under CIMO � rms increased investments in worker train- the name Productivity Support Program (Programa de ing, had higher rates of capacity use, and were more Apoyo para la Productividad [PAP]). There was also likely to adopt quality control practices. These a change in the design of the program: before 2000, improved outcomes were associated with an increase all training was provided by the private sector; after in net productivity of 6–11 percent for CIMO par- 2000, the program used only public providers. With ticipants. Evaluations of the program’s performance an estimated budget of about $75 million, CIMO/ after 2000 do not � nd signi� cant effects on � rms’ PAC trained about 1.6 million workers between 2001 productivity, although they do find an increase and 2006 and bene�ted more than 226,000 � rms. in worker training and the use of quality control Before its decentralization, one of the unique fea- practices. tures of CIMO was its comprehensive approach to addressing training. Because small and medium-size Sources: López-Acevedo and Tinajero 2010; Tan and López-Acevedo 2005. Support for self-employment and where microfinance focuses on linking microenterprises microentrepreneurs to the formal credit Interventions that support self-employ- market—micro� nance in South Asia focuses ment or microenterprise development can on reducing poverty. help informal sector workers improve their Microfinance providers in South Asia potential for generating income. Financial reach more than 50 million clients.46 Cover- assistance has been the primary means of age of micro�nance services ranges from 0.1 encouraging income-generating activities in percent of the population in Afghanistan to most South Asian countries. In fact, since more than 20 percent in Bangladesh (table the 1970s and the founding of the Grameen 6.9). Demand for microcredit services far Bank, South Asia has been at the forefront outstrips supply. Micro� nance in the region of the micro� nancing movement, which has has concentrated on women in rural areas, become prominent in international develop- although in Bangladesh, and increasingly ment.45 Almost half of the people worldwide India, the share of urban clients is growing. who use micro� nance services live in South The principle underlying microcredit is Asia (World Bank 2009). In contrast to activation: providing credit to individuals other regions—particularly Latin America, without access to formal banking services 256 MORE AND BETTER JOBS IN SOUTH ASIA BOX 6.6 Training informal workers: Kenya’s Jua Kali experience The Jua Kali (informal sector) project, funded by 50 percent. According to anecdotal evidence, some the International Development Association, aimed participants who received a voucher for basic train- to upgrade skills and technology for about 25,000 ing paid the full cost of more advanced training. informal sector manufacturing workers, increase the Lessons from this project include the following: access of informal sector entrepreneurs to services, • The use of a voucher mechanism enabled the proj- and improve the policy and institutional environ- ect to stimulate demand for training, technology, ment by removing restrictive laws and policies. A management, and marketing consultation by micro key feature of the project was a voucher program and small enterprises. A supply response was gen- intended to introduce consumer choice, enabling erated and a training market established to address informal sector operators to purchase the training these �rms’ needs. they want wherever they want. Intermediaries (allo- • An unexpected impact of the voucher training pro- cation agencies) were selected by competitive tender gram was the emergence of skilled craftspeople as to market, allocate, and redeem vouchers in a decen- the leading providers of training. In the �rst phase tralized way throughout Kenya. Allocation agen- of the project, 85 percent of all vouchers went to cies received a fee equal to 3 percent of the value of pay for their services; only 15 percent went to train- vouchers issued. Vouchers could be used for any kind ing institutions. Entrepreneurs preferred receiving of training from any registered training provider. training from master craftspeople to training in Over the course of the project in the 1990s and formal institutions. This training was usually well the earlier part of the last decade, about 700 train- adapted to entrepreneurs’ need for short, practical ing providers prequali�ed for providing training. By training. early 2001, some 18,000 training vouchers had been • Implementation experience underscores the impor- issued. tance of appropriate management arrangements. The impact of the project on bene�ciaries, eval- A project for the private sector is best managed uated through two tracer studies, has been highly by the private sector, with government playing a positive. Employment among graduates was 50 facilitating role. percent higher than employment before training, and the income of surviving enterprises increased Source: Johanson and Van Adams 2004. TABLE 6.9 Coverage of micro�nance in South Asia, by country, in order to enable them to initiate or expand 2009/10 productive activities, usually in the form of Population Number of microfinance Percentage self-employment or a household enterprise. 2010 borrowers 2009/2010 of An important limiting factor is the level of Country (millions) (millions) population human capital among entrepreneurs. For this Afghanistan 31 0.3 0.1 reason, adding training, counseling, and vari- Bangladesh 149 30.0 20.2 ous forms of technical assistance to micro�- India 1,225 22.6 1.8 nance (“micro�nance plus�) is key to promot- Nepal 30 1.8 6.0 ing business development among small-scale Pakistan 174 1.9 1.1 entrepreneurs. A key challenge from an acti- Sri Lanka 21 2.5 12.0 vation perspective is to develop instruments Sources: Authors, based on data on number of micro�nance clients from South Asian that allow providers of micro�nance to iden- Micro�nance Network (http://www.samn.eu/?q=micro�nance-asia) and population �gures tify the entrepreneurs most likely to bene�t from UN 2010. Note: Data on number of clients are for 2010 for Afghanistan and Pakistan and 2009 for other from credit or associated subsidies. countries (http://esa.un.org/unpd/wpp/unpp/panel_population.htm). With the exception of a few The institutional structure of micro�nance limited efforts to provide microcredit services as part of integrated development projects, Bhutan and Maldives do not have speci�c micro�nance sectors. programs varies widely. NGOs, nonprofit THE ROLE OF LABOR MARKE T REGULATIONS, INSTITUTIONS, AND PROGRAMS 257 micro� nance institutions, and community- more efficient institutions account for a based organizations, including cooperatives, large share of the market; the majority of account for most of the outreach of �nancial institutions are relatively weak, sustained services available to the poor in the region. mainly by the high level of subsidies provi- The retail operations of development and ded by donors or governments (World Bank commercial banks; post office banks; and, 2009). In Bangladesh, three of the top four less often but increasingly, insurance com- microfinance institutions—ASA, BRAC, panies provide a small amount of additional and Grameen Bank—show positive returns. lending services. Commercial banks are less These organizations account for some 11 mil- important to micro� nance than they are in lion of an estimated 16 million clients. some other regions, such as Latin America, India has a large number of loss-making although in all countries except Afghanistan micro� nance institutions and relatively few and Bangladesh, micro�nance does have sig- viable ones. Political and social restrictions ni�cant commercial bank linkages (EDA and have inhibited many microfinance institu- APMAS 2006).47 tions from setting actuarially neutral interest This diverse set of institutional arrange- rates, which affects their performance. ments of microfinance programs is mir- In Pakistan, the provision of micro�nance rored by a diversity of delivery systems. The on a sustainable basis is particularly challeng- Grameen model refers to a methodology �rst ing, partly as a result of substantial pressure applied by the Grameen Bank of Bangladesh. from government of�cials, elected represen- Borrowers have to be a part of a small group tatives, and the press to charge low interest that gets together with 7–10 other groups rates. To date, the only micro�nance institu- from the same village or neighborhood, tion that has become fully sustainable is the forming a “center.� By aggregating liabilities Kashf Foundation. within each center, this model reduces the In Sri Lanka, the schemes under the likelihood of default. This delivery system is Samurdhi program offer subsidized inter- the dominant methodology in Bangladesh. est rates. They face default rates of 30–50 All retail development banks in Nepal, some percent. institutions in Afghanistan and Pakistan, South Asia has numerous programs invol- and some leading micro�nance institutions in ving not only micro� nance but other forms India also use this system. of assistance to support self-employment and The self-help group is the main approach microenterprise development. Table 6.10 lists to micro� nance in India. It is based on the some of these programs. principle of revolving members’ own savings. Both internationally and within the region, Savings thus precedes borrowing by mem- there is little evidence on the impact on bers. Groups, made up of 15–25 members, employability and earnings of programs that operate a revolving fund that is supplemented support self-employment.48 The few studies with external �nancial assistance. that exist suggest that the provision of ser- Another mode of delivery is cooperative vices to support entrepreneurs often results programs, in which micro�nance institutions in higher rates of business success and higher provide � nancial services to individual cli- earnings (Betcherman and others 2004, ents organized into credit and savings coop- 2007). In general, a combination of access to eratives. All borrowers need to be members credit and support services, such as counsel- of the cooperative. Sri Lanka, with its large ing, seems to have the best outcomes. number of cooperative banks, relies heavily Few evaluations use rigorous methods or on this approach. India and Nepal have simi- try to calculate cost-bene�t ratios. One excep- lar initiatives. tion is a recent evaluation of a Sri Lanka pro- One of the main issues facing microfi- gram that uses training and �nancial support nance institutions relates to financial sus- to encourage women to create and expand tainability. A small number of gradually businesses (box 6.7). 258 MORE AND BETTER JOBS IN SOUTH ASIA TABLE 6.10 Selected programs supporting self-employment and microenterprises in South Asia, by country Country/program Institutional set-up Target population Services Bangladesh Vulnerable Group Development Directorate of Women’s Affairs; funded Poor, landless women Subsistence allowance, training for Ultra Poor by the European Union Employment Guarantee Scheme Bangladesh Rural Development Board Poor Training assistance in for Hard-Core Poor in Northern (BRDB) organizing income-generating region activities Urban-Based Marginal Women’s Jatio Mahila Sangstha Poor, assetless women Training, financial assistance Development Agency India Swarnjayanti Gram Swarozgar Ministry of Rural Development Poor families in rural Financial assistance, training, Yojana areas support for self-help groups (now the National Rural Livelihoods Mission) Swarna Jayanti Shahari Rojgar Ministry of Housing and Urban Poverty Urban unemployed or Financial assistance, training, Yojana Alleviation underemployed support for self-help groups Prime Minister’s Employment Khadi and Village Industries Commission Unemployed youth Credit and subsidies Generation Programme Rajiv Gandhi Udyami Mitra Ministry of Micro, Small, and Medium Potential first- generation Technical assistance, counseling Yojana Enterprises entrepreneurs Nepal Great Himalayan Trail Ministry of Tourism and Civil Aviation, in Local communities Skills training, support for Development Program partnership with Stichting Nederlandse entrepreneurs Vrijwilligers (a Dutch development organization), local authorities, industry associations, and communities Production Credit for Rural Ministry of Local Development, various Poor women Microcredit Women banks; funded by the International Fund for Agriculture Development Rural Self-Reliance Fund Nepal Rastra Bank Poor and specified Wholesale provision of vulnerable groups microcredit Microenterprise Development Ministry of Industry, multilateral donors Poor and specified Training, microcredit Programme vulnerable groups Poverty Alleviation Fund Board reports to prime minister, Rural poor and Training, technical assistance community-based organizations, and vulnerable groups NGOs; funded by International Development Association Development Project Service Executive director reports to seven- Poor women Microcredit, training, Center (DEPROSC) Nepal member board; implementation follows entrepreneurial development guidelines of Nepal Rastra Bank Pakistan Rural Support Programs National and provincial Rural Support Poor people, especially Microcredit, microinsurance, Programs women training Khushali Bank Created under Rural Support Programs; Poor people, particularly Microcredit four banks are shareholders women Khasf Foundation Pakistan Poverty Alleviation Fund, U.K. Women from low- Microcredit Department of International income households Development, and Agha Khan Foundation provided core funding (continues next page) THE ROLE OF LABOR MARKE T REGULATIONS, INSTITUTIONS, AND PROGRAMS 259 TABLE 6.10 (continued) Country/program Institutional set-up Target population Services Sri Lanka Samurdhi Authority Ministry of Economic Development Low-income groups Microcredit Small Enterprise Development Ministry of Youth Affairs Youth, especially Microcredit Division dropouts Ministry of Labour Relations Unemployed Microcredit Sarvodaya Economic Enterprise SEEDS (NGO), Poor people, women Development Services donors and girls, migrants Microcredit Source: Authors’ compilation, based on country research on active labor market programs. BOX 6.7 Business training and �nancial support for self-employed women in Sri Lanka Self-employment is a potentially important form of for the sample of women who had been out of the employment for women in South Asia. However, labor force. Firms created by women who received most women-owned enterprises are small and barely training earned higher pro�ts 16 months later than generate a subsistence-level income for their owners. � rms created by women in the control group. These Policy efforts have sought to enhance the likelihood results suggest some success for business training of women owning small businesses and increase the among women interested in opening a business who earnings of women business owners. currently do not work. Business training programs are a popular policy For women already operating businesses, training used for this purpose. Preliminary results from a alone led to some improvement in business practices randomized experiment in Sri Lanka offer insight (they were more likely to implement some of the ideas into whether such training works and whether it can taught in the training course), but it did not lead to complement � nancial assistance. any improvement in pro�ts, sales, or capital stock. In The study worked with two random samples, contrast, the combination of business training and of about 600 women each. One sample included the cash grant led to large and signi�cant improve- women operating subsistence enterprises. The second ments in business performance for these enterprises, included women who were out of the labor force but with pro�ts rising 36–50 percent and capital stock interested in starting a business. Both samples were increasing by about the amount of the grant. As pre- randomly divided into three groups: a control group; vious work with female microenterprises � nds little a group invited to attend a seven- to nine-day business return to capital alone, the authors of this study training course based on the International Labour argue that the results suggest important complemen- Organization’s “Generate, Start, and Improve Your tarities between training and access to capital. Business� programs; and a group invited to receive Together with other recent experiments with this training and receive a cash grant of Rs 15,000 women-owned microenterprises (Bruhn and Zia ($130) upon completing the training. 2011; Gine and Mansuri 2011), these results suggest Business training sped up the creation of new that it is much easier to get women to start enter- businesses, but it did not ultimately lead to more prises than it is to get these enterprises to grow. Only new businesses. Women who were out of the labor 1–2 percent of the enterprises started by women force were more likely to have a business 4 months who underwent business training employed any and 8 months after the training program, but by 16 paid workers who were not family members. Female months the control group had the same rate of busi- self-employment is thus largely a means to increase ness ownership. Receiving cash in addition to train- income and provide a living for enterprise owners ing led to even larger short-run effects but no differ- themselves rather than a means of generating addi- ence in business ownership after 16 months. tional jobs for others. Although training did not seem to lead to more businesses, it did result in more successful businesses Source: de Mel, McKenzie, and Woodruff 2011. 260 MORE AND BETTER JOBS IN SOUTH ASIA There is more evaluation evidence on also � ll a promotion function, by provid- micro� nance programs. The focus of these ing workers with more room to maneu- studies has typically been on the antipov- ver in making human capital investments erty effects and effects on microenterprise and transitioning to better employment creation and survival, but studies have also opportunities. considered other outcome variables (for Bismarkian social insurance systems linked example, health, education, empowerment to formal employment relationships have lim- of women). Longer-run labor market out- ited scope in South Asia. In the absence of comes (for example, employability and earn- extensive formalization, the key question is ings) have been included less often. whether the formal social insurance model There is a some controversy over the impact can be modi�ed and adapted to the realities of micro�nance, but, on balance, evaluations of informal sector workers in the region. point to positive effects.49 Findings are het- Plans to extend social insurance to infor- erogeneous, with welfare effects ranging mal workers need to take into account pos- from nonexistent to signi�cant (Armendáriz sible effects on the incidence of informal- de Aghion and Morduch 2004). Many evalu- ity itself. Employers may be more likely to ations use fl awed methodology, with selec- seek ways to opt out of formal contributory tion bias and other issues confounding results systems if they know employees can access (EDA and APMAS 2006). social protection in other ways. Depend- Many concerns can be addressed with ing on how programs are � nanced, workers random experimental designs, which tend themselves may prefer to remain informal. to be rare in the study of micro�nance. One Although the extension of social insurance exception is a randomized evaluation in (for example, for pensions, health, and dis- Hyderabad that found that the opening of ability) would address the vulnerability branches of a microfinance institution had problem associated with informality, other signi�cant and positive impacts on business problems, including the � scal and produc- start-ups and the pro�tability of existing busi- tivity disadvantages of large informal sec- nesses, but no impact on average household tors, would remain. Thus, although there consumption. The results led the authors to would be bene�ts to extending social protec- conclude that “microcredit therefore may not tion coverage beyond the formal sector, the be the ‘miracle’ that is sometimes claimed on tradeoffs should not be ignored. its behalf, but it does allow households to Various social insurance schemes for borrow, invest, and create and expand busi- informal workers have been introduced in the nesses� (Banerjee and others 2009, p. 21). region. To encourage informal sector partici- pation, the new pension law in Maldives pro- vides for matching contributions for infor- Extension of social insurance mal sector workers such as �shers. Sri Lanka to informal workers has voluntary pension schemes for infor- A key question for labor market policy in mal farmers, �shers, and the self-employed. South Asia is whether social insurance NGOs also provide microinsurance schemes models can be extended to protect infor- that protect against speci�c risks. mal workers from risks related to income India has been active in trying to extend and employment and to provide a vehicle social insurance to informal sector work- for pension, disability, and health coverage. ers. Numerous welfare funds and insurance Ef� cient programs with adequate bene�ts schemes exist nationally, at the state level (in, and wide coverage play an important pre- for example, Kerala and Tamil Nadu), and vention function, by reducing the vulner- by occupation (covering, for example, work- ability of workers and their families to the ers making hand-rolled cigarettes). Informal negative impact of various shocks. They worker associations, self-help groups, and THE ROLE OF LABOR MARKE T REGULATIONS, INSTITUTIONS, AND PROGRAMS 261 other organizations have also been active in alongside existing contributory programs for this area (IHD 2011). the formal sector.51 Proposals have been made —by, for The potential for extending social insur- example, the National Commission for ance to the informal sector depends heavily Enterprises in the Unorganised Sector on � nancing and the degree to which indi- (NCEUS)—to institute a national minimum viduals are able to contribute. Workers in the social security. The NCEUS proposal sought informal sector fall into three broad catego- to guarantee minimum coverage for health, ries, with the �rst accounting for only a very death and disability, and pension bene�ts small share of the population: for all workers, organized and unorga- • High-income individuals (for example, nized, earning less than Rs. 6,500 in 2005. professionals and large landowners), who Bene�ts were to have been � nanced through are able to protect themselves through equal contributions, at least notionally, from savings and insurance schemes already government, employers, and employees. In available in the market fact, the burden would have been greatest • Middle-income individuals, who may be on government, as it would have had to able to bene�t from insurance and pen- cover the employer share when an employer sion coverage and could afford to pay could not be identified and the employee for a signi� cant portion of the required share in the case of the poor. Although the contributions NCEUS proposal was the spark behind the • Low-income individuals, who are unable Unorganised Workers Social Security Act of to make any contributions beyond a very 2008, that bill was much more modest in marginal level. scope (IHD 2011). 50 The limited scope of the Unorganised Given the high prevalence of individuals Workers Social Security Act underlines the with no or limited capacity to contribute, cov- dif�culties of implementing an overarching erage can be expanded only if governments policy framework for social insurance in the are willing to provide signi�cant subsidies. informal sector. Given the heterogeneity of The level of subsidies needs to be determined informal workers, it seems more likely that in light of the mandate of the programs, broader coverage could be achieved by creat- and the subsidies need to be allocated and ing and expanding different programs with �nanced in ways that minimize labor market different design features—programs that run distortions. 262 MORE AND BETTER JOBS IN SOUTH ASIA Annex 6A Additional tables and �gures on labor market regulations and institutions FIGURE 6A.1 Trade union membership in India, 1987–2006 10,000 8,000 6,000 thousands 4,000 2,000 0 87 88 89 90 91 92 93 94 95 96 97 98 99 00 01 02 03 04 05 06 19 19 19 19 19 19 19 19 19 19 19 19 19 20 20 20 20 20 20 20 Source: Indiastat.com. Note: Trend line indicates three-year moving average of membership in workers’ unions submitting returns in India. FIGURE 6A.2 Trade union membership in Pakistan, 1999–2008 600 500 400 thousands 300 200 100 0 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 Source: Federal Bureau of Statistics Pakistan 2011. Note: Trend line indicates three-year moving average of membership in reporting unions. THE ROLE OF LABOR MARKE T REGULATIONS, INSTITUTIONS, AND PROGRAMS 263 FIGURE 6A.3 Trade union membership in Sri Lanka, 1987–2006 1,400 1,200 1,000 800 thousands 600 400 200 0 96 97 98 99 00 01 02 03 04 05 06 07 08 09 19 19 19 19 20 20 20 20 20 20 20 20 20 20 Source: Department of Labour, Sri Lanka. Note: Fluctuation in total membership is a result of poor reporting by unions. Trend line indicates three-year moving average. 264 MORE AND BETTER JOBS IN SOUTH ASIA FIGURE 6A.4 Percentage of unionized workers in India, by employment status and sector, 1994 and 2000 a. By employment status 100 80 72 70 60 58 percent 40 32 29 20 16 18 15 18 12 10 11 5 4 6 4 5 2 0 s iva d pu ied oy d oy d or er pr rie pl -en pl en lab rk er lar te ic ed ed m w- er la em igh bl wo rk r sa rk sa al lo su h wo e or all wo e o ca e g g lf- lf- wa wa se se lar lar gu gu re re b. By sector 100 80 65 60 58 57 percent 40 26 26 29 29 22 19 19 20 15 15 2 4 7 4 7 2 0 tu nd ica ag , ilit d n es hi e un or ce io ut an fis ur vic ac a m st an ng g tio e, ies ct d lt uf ing d n rin an ricu er tru m rt, ur an atio n rs co o s an min ns d sp , in ag he ity str co an tran nce ot ric ni ct mi a m te n ele c ad ta , fi es rce i bl , al e pu re mm co 1994 2000 2010 Source: Authors, based on data from various rounds of National Sample Survey. Note: Union membership, based on household data, includes not only registered trade unions but also membership in associations of owners, self-employed workers, and any association whose objective is to look into the interests of its members. It is not possible to break membership down into trade unions and other associations. THE ROLE OF LABOR MARKE T REGULATIONS, INSTITUTIONS, AND PROGRAMS 265 FIGURE 6A.5 Job creation and destruction flows in Sri Lanka and selected groups of countries 30 25 20 percent 15 10 5 0 job job job excess job creation destruction reallocation reallocation Sri Lanka developing countries transition economies developed countries Source: Gunatilaka and Vodopivec 2010 (based on 2003 Sri Lanka urban enterprise survey using employment data for three years in the early 2000s). Note: Transition economies include Estonia (1997–2000), Hungary (1994–2000), Latvia (1994–99), Romania (1994–2000), and Slovenia (1992–2000). Developed countries include Canada (1985–97), Denmark (1982–94), France (1991–96), Finland (1989–98), Germany (1978–99), Italy (1988–93), Portu- gal (1984–94), the Netherlands (1994–97), the United Kingdom (1987–98), and the United States (1989–97). Developing countries include Argentina (1997–2001) and Mexico (1987–2000). Job reallocation is the sum of job creation and job destruction, where job creation is the sum of new jobs in �rms that expanded during the previous year, and job destruction is the sum of employment losses in �rms that contracted. Another measure, net employment growth, is the difference between the job reallocation and net employment growth. Using these measures, excess job reallocation computes the difference between job reallocation and net employment growth. This indicator measures the extent of job flows over and above what is needed to accommodate a needed change in net employment; hence, it reflects the intensity of enterprise restructuring. All variables are expressed as a percentage of the average of the stocks at the beginning and end of the period. 266 MORE AND BETTER JOBS IN SOUTH ASIA FIGURE 6A.6 Tax wedges in South Asian and international comparator countries Morocco 35.7 India 35.6 Belarus 35.5 Bosnia 34.9 Albania 33.4 Nepal 33.3 Iran, Islamic Rep. 32.5 Moldova 32.4 Sri Lanka 32.3 Uruguay 31.9 Kyrgyz Republic 31.6 Brazil 31.6 Russian Federation 31.0 Azerbaijan 29.8 Tajikistan 29.6 Costa Rica 28.8 Kazakhstan 28.2 Yemen 28.2 world Georgia 26.7 average Maldives 24.6 Jordan 23.9 Ecuador 23.8 Paraguay 23.3 Libya 21.9 Lebanon 21.0 Niger 19.5 Chile 15.5 Pakistan 15.4 Iraq 15.2 Bhutan 15.2 Bahrain 11.2 Afghanistan 5.8 0 5 10 15 20 25 30 35 40 tax wedge (percent) Sources: Authors, based on latest data on social security contributions available from the International Social Security Administration. Data on average wages and income taxes were collected by staff at the country level. THE ROLE OF LABOR MARKE T REGULATIONS, INSTITUTIONS, AND PROGRAMS 267 FIGURE 6A.7 Ratio of minimum wage to median casual and formal sector wage, by states in India, 2008 a. Ratio of minimum wage to median casual wage Delhi 1.45 Chattisgarh 1.20 Bihar 1.15 Madhya Pradesh 1.13 Haryana 1.09 Sikkim 1.09 Dadra and Nagar Haveli 1.08 West Bengal 1.06 Karnataka 1.06 Punjab 1.04 Orissa 1.01 Rajasthan 0.96 Maharashtra 0.92 Daman and Diu 0.88 Uttar Pradesh 0.85 INDIA AVERAGE 0.85 Gujarat 0.83 Uttaranchal 0.83 Goa 0.77 Arunachal Pradesh 0.76 Meghalaya 0.76 A and N Islands 0.76 Assam 0.76 Manipur 0.76 Chandigarh 0.75 Jharkhand 0.74 Himachal Pradesh 0.73 Tamil Nadu 0.69 Jammu and Kashmir 0.66 Mizoram 0.66 Tripura 0.66 Andhra Pradesh 0.62 Kerala 0.58 Lakshadweep 0.56 Puducherry 0.46 Nagaland 0.31 0 0.50 1.00 1.50 2.00 ratio (continues next page) 268 MORE AND BETTER JOBS IN SOUTH ASIA FIGURE 6A.7 Ratio of minimum wage to median casual and formal sector wage, by states in India, 2008 (continued) b. Ratio of minimum wage to median formal sector wage Dadra and Nagar Haveli 0.37 Delhi 0.35 Haryana 0.32 Kerala 0.28 A and N Islands 0.28 Daman and Diu 0.27 Madhya Pradesh 0.27 Jammu and Kashmir 0.27 Punjab 0.27 Rajasthan 0.26 Sikkim 0.25 Uttaranchal 0.25 Lakshadweep 0.24 Himachal Pradesh 0.24 Meghalaya 0.23 Chandigarh 0.23 INDIA AVERAGE 0.23 Chattisgarh 0.22 Goa 0.22 Karnataka 0.22 Mizoram 0.22 Orissa 0.21 Manipur 0.21 Bihar 0.21 Uttar Pradesh 0.21 Tripura 0.20 West Bengal 0.20 Assam 0.18 Andhra Pradesh 0.18 Nagaland 0.18 Gujarat 0.17 Jharkhand 0.17 Tamil Nadu 0.16 0. Maharashtra 0.15 Arunachal Pradesh 0.14 Puducherry 0.12 0 0.10 0.20 0.30 0.40 ratio Sources: Authors, based on data on minimum wages and state GDP per capita from IndiaStat.com and data on wages from labor force survey. THE ROLE OF LABOR MARKE T REGULATIONS, INSTITUTIONS, AND PROGRAMS 269 TABLE 6A.1 Selected aspects of employment protection in India Type of employment protection Legal provisions Procedures for individual notification and dismissal • Written notice to employee • Employee given sufficient warning and opportunity to respond • Employers with at least 100 workers must give government 60 days’ notice before dismissing worker and await its approval before doing so Notice period for individual dismissals • Workers small firms with 4 or more years of tenure: 1 month • Workers at large firms with at least 20 years of tenure: 3 months Severance pay for individual dismissals • Workers with 4 or more years of tenure: 2 months • Workers with at least 20 years of tenure: 10 months Economic reason as valid reason for dismissal • Permitted, but in absence of agreement between worker and employer, last person to be employed in category should be one dismissed Maximum time to make complaint of unfair dismissal • No time limit Compensation for unfair dismissal • Court orders reinstatement in most cases (with or without back pay) • In extreme cases, court may order compensation instead of reinstatement • Typical compensation for employee with 20 years of tenure is 42 months’ wages • Typical three to four years for labor courts to settle disputes and make award Definition and notification requirements or additional • No special regulations or additional costs procedures for collective dismissals Valid cases for fixed-term contracts • Limited for work of temporary nature likely to be finished within a limited time • Some industries (information technology, business process outsourcing); export processing zones, and special economic zones in some states are exempted Maximum number of fixed-term contracts • No limits on number of renewals or maximum duration Temporary agency work • Allowed in noncore activities; prohibited in some industries or firms • No limits on durations or renewals • Contractors and firms with more than 20 employees required to obtain license (renewable every 12 months) and pay fees before engaging contract workers • Contractors required to report changes in number of workers or conditions of work to licensing authority • Wages and conditions of work of contract worker must be same as direct employees performing same type of work Source: Authors’ compilation, based on information in OECD 2009. 270 MORE AND BETTER JOBS IN SOUTH ASIA TABLE 6A.2 Job creation and destruction rates in large manufacturing �rms in India, by employment type, 2001–08 (percent) Type of Job creation Job destruction Net job Job reallocation Job excess worker/year rate rate growth rate rate reallocation rate Contract workers 2001 15.3 18.7 –3.3 34.0 30.7 2002 19.1 19.1 0.0 38.2 38.2 2003 18.2 16.0 2.2 34.1 32.0 2004 21.3 16.2 5.2 37.5 32.3 2005 21.4 14.5 6.9 35.9 29.0 2006 24.8 14.5 10.3 39.3 28.9 2007 22.9 14.0 8.9 36.8 28.0 2008 23.6 13.8 9.9 37.4 27.5 Total 24.5 15.6 8.9 40.1 31.2 Noncontract workers 2001 6.5 10.1 –3.6 16.6 13.0 2002 7.6 10.1 –2.4 17.7 15.3 2003 7.5 9.0 –1.5 16.5 15.0 2004 8.2 10.0 –1.8 18.2 16.4 2005 9.5 7.9 1.6 17.4 15.8 2006 9.7 7.9 1.8 17.6 15.8 2007 11.5 7.4 4.1 18.9 14.8 2008 11.9 6.7 5.2 18.6 13.4 Total 9.1 9.8 –0.7 18.9 18.1 All employees 2001 6.6 10.2 –3.6 16.7 13.2 2002 7.8 9.8 –2.0 17.7 15.6 2003 7.9 8.8 –0.9 16.7 15.8 2004 8.9 9.3 –0.4 18.2 17.8 2005 10.1 7.4 2.7 17.6 14.8 2006 10.9 7.1 3.8 18.0 14.3 2007 12.2 7.0 5.2 19.2 14.0 2008 13.0 6.6 6.4 19.6 13.3 Total 9.8 8.6 1.2 18.4 17.1 Source: Authors, based on data from various rounds of the Annual Survey of Industries. Note: See �gure 6A.5 for de�nitions. THE ROLE OF LABOR MARKE T REGULATIONS, INSTITUTIONS, AND PROGRAMS 271 TABLE 6A.3 Minimum wage policies in South Asia, by country Minimum wage Latest available Year of last update of Country setting Coverage minimum wage minimum wage Afghanistan One minimum wage set for Public sector workers Af 4,000 a month 2009 public sector workers. Bangladesh Three-person national minimum Minimum wage does not Sectors not covered by Garment industry: 2010 wage board convenes every five apply to categories of workers industry-specific minimum Shrimp industry: 2009 years to set minimum wages by excluded from general coverage wages: Tk 1,800 a month industry. of the Labor Act of 2006 under Garment industry: Tk 3,000 §1(4) (for example government a month offices, small establishments run Shrimp industry: Tk 2,510 by family members). a month Bhutan Ministry of Labor and Human National coverage Nu 3,000 a month 2010 Resources, in consultation with the government, employers, and employees, fixes minimum wage or wages. India Minimum wage set at state Workers in sectors/occupations National minimum wage 2011 level by scheduled sectors and categorized as scheduled sector/ (nonbinding): Rs 115 a day occupations. occupation in a particular state Maldives One minimum wage for public Public sector is covered. Rf 2,600 a month 2008 sector. Employment Act of 2008 provides a mechanism to establish a minimum wage in the private sector. Nepal Government sets monthly Private sector workers and Minimum monthly wage for 2011 minimum wage set for four skill employees in enterprises of unskilled workers: NPR 6200 categories (unskilled, semiskilled, more than 10 workers and Minimum daily wage: NPR 231 skilled, and high skilled). Daily enterprises operating in minimum wage set for daily industrial districts established wage workers. Separate by the government. provisions made on industry basis. Pakistan Federal government sets Workers in agricultural sector PRs 7,000 a month 2010 minimum wages for unskilled and people employed by federal workers at the national level, or provincial governments are through an ordinance. Provincial not covered by minimum wage governments then adjust federal legislation. minimum wage to account for regional differences and skill levels. Sri Lanka Minister of Labour may set Bases on scope of wage boards. Agriculture (average for 2009 up wages boards by trade plantation sector): (occupation), sector, or region to SL Rs 235.15 a day recommend minimum wages by Manufacturing: grade levels of jobs within each SL Rs 247.04 a day category. Construction: SL Rs 292.50 a day Sources: Authors, based on data from ILO database; U.S. Department of State; national labor force and household surveys. 272 MORE AND BETTER JOBS IN SOUTH ASIA timely manner or not at all. They can there- Notes fore provide an unreliable picture especially 1. Labor market and social protection inter- on trends. Statistics cited in the 2010 U.S. ventions often straddle these functions. For Department of State Country Reports on example, public works can play both an acti- Human Rights Practices at http://www.state vation and a safety net role. This chapter does .gov/g/drl/rls/hrrpt/ give a similar picture— not address wider aspects of social protection, with estimated union membership of 13–15 such as the provision of pension bene�ts and million in India, and estimated union densi- health care services, except in terms of how ties of approximately 10 percent in Sri Lanka, they may affect job creation and the function- under 1 percent in Nepal, 3 percent in Pakistan, ing of the labor market. and 4 percent in Bangladesh. In India, where 2. As discussed in this chapter, voluntary social household data for 2010 provide information security schemes have begun to emerge for on union membership, the estimated density informal sector workers, but they cover only is higher (11 percent) (see annex �gure 6A.4), a small share of them. The various schemes in but the �gure includes self-employed workers India are estimated to cover less than 10 per- and members of associations of owners and cent of the unorganized labor force (World other associations whose objective is to look Bank 2011b). In Sri Lanka, voluntary social after the interests of their members. security schemes cover an estimated 0.5 mil- 7. Median monthly earnings in for wage work- lion members of the workforce of roughly ers with social security were Rs. 5,332 in India 7.5 million (Gunatilaka 2011). (2010), NPR 5,128 in Nepal (2008), and SL 3. For example, India’s 1948 Factories’ Act, one Rs7,881 in Sri Lanka (2008). These earnings of the main laws governing working conditions, exceeded those of workers without coverage by applies to manufacturing establishments with at a factor of 3.1 in India, 1.9 in Nepal, and 1.7 in least 10 workers if the establishment uses power Sri Lanka. Wage workers who were members and at least 20 workers if it does not. See table of unions had mean/median monthly earnings 6.4 for coverage of laws that protect workers that were about 2.5 times those of nonunion against redundancy in South Asian countries. members in India (National Sample Survey 4. IHD (2011) and NCEUS (2009) provide 1999/2000). Heltberg and Vodopivec (2004) estimates of the intended coverage of vari- estimate that the earnings of wage workers in ous labor laws in India. They estimate that Termination of Employment of Workmen Act 12 million workers were covered by the (TEWA) jobs were 11–12 percent higher than Factories Act in 1999/2000. The Ministry of those of workers in informal jobs in 1999/2000 Labor estimates coverage (generally based on after taking into account differences in educa- enterprises that have submitted returns under tion and other characteristics. the act) of 8.8 million workers the same year. 8. See http://www.ilo.org/declaration/lang--en/ The �rst estimate translates into 3 percent of index.htm. the total workforce; the second translates into 9. Afghanistan has signaled its intention to ratify 2.2 percent of the total workforce. Together, its three conventions that are outstanding, but they suggest coverage of about three-quarters the government has indicated the need for ILO of the workers intended to be covered. assistance to ensure and support this process. 5. The possible exception is Nepal, where esti- Nepal has delayed rati�cation of C.87, citing mates of union density in the private sector political instability and the period of political range from about 29 percent of the work- transition as reasons for delay. Bangladesh iden- force (for �rms in the formal sector surveyed ti�es dif�culties in addressing the large infor- in the 2009 enterprise surveys) to 75 percent mal sector where child labor is engaged as an of the formal work force in 2010 (http:// obstacle in ratifying C. 138 in 2001. India has www.state.gov/g/drl/rls/hrrpt/). In Sri Lanka, cited various issues complicating rati�cation of the corresponding estimates are 11 percent of the collective bargaining and freedom of asso- the nonagricultural private labor force and ciation and child labor conventions. For details 26 percent of the workforce of �rms included on rati�cation status by country, see http://www in the 2004 enterprise survey sample. .ilo.org/declaration/ 6. Data are typically based on returns �led 10. In Afghanistan, India, and Nepal, the basic by unions. Unions may not �le returns in a legal provisions on freedom of association THE ROLE OF LABOR MARKE T REGULATIONS, INSTITUTIONS, AND PROGRAMS 273 and collective bargaining are provided in the 17. In Sri Lanka, prior approval is necessary constitution as well as in national legislation; only when the worker does not consent in in India, judicial decisions have also upheld writing to the lay-off. This is different from these rights. In Bangladesh, the Labor Act India, where Chapter VB of the Industrial of 2006 not only establishes minimum age Dispute Acts requires establishments with requirements, it also strengthens the labor 100 or more workers to gain prior approval administration for enforcement. In addition from government even if workers give their to the constitution, about a dozen pieces of consent. Under Chapter VA of the act, which legislation provide the basic legal provisions applies to enterprises employing 50 or more on child labor in India. In Nepal, the prin- workers, employers are obligated to inform ciple of and rights to freedom of association the government of dismissals rather than ask and collective bargaining are enshrined in the for their consent. The clause does not speci�- 2007 interim constitution as well as national cally mention worker consent. However, if legislation. workers do not consent, they resort to dis- 11. See http://www.ilo.org/declaration/ and http:// pute resolution. www.state.gov/g/drl/rls/hrrpt/. 18. Employers need the authorization of the com- 12. Particularly relevant are the rules governing missioner of labor for each intended layoff if whether and how workers can be dismissed the employee worked for at least 180 days for “economic� reasons (shrinking markets, the year before termination and does not give increasing competitiveness, and so forth). written consent to the termination. The com- This class of terminations stands in contrast to missioner has the power to grant or refuse dismissals for “noneconomic� reasons, such permission to terminate services. as job performance. 19. The level of pay is determined by two vari- 13. Freeman (1993) characterizes these two per- ables: years of service with the employer spectives as the “institutionalist� and “distor- (past service) and the age of the worker tionist� views. For an updated interpretation, (denied future service). It is relatively high for see Freeman (2009). young workers and those with short service, 14. Cross-country comparisons of employment exceeding the average for all the regions of protection rules need to be viewed with some the world. For older workers and those with caution, especially because of the impor- long service, the severance pay determined by tance of enforcement, which is dif�cult to the formula is more in line with international take into account. National settings can practice (Vodopevic 2004). matter as well. The Doing Business database 20. “Nonperformance� is a generic problem includes almost all employment protection with unfunded severance systems: employ- legislation. Its indicators are based primar- ers’ liabilities rise when they are likely to be ily on laws and regulations, with country in �nancial dif�culty. This problem becomes scores calculated by local experts. See www most acute during recessions. During the .doingbusiness.org for scores and method- recent economic crisis, labor authorities ological details. took unprecedented steps to enable �rms to 15. The main sources of uncertainty related to weather the crisis. Employers were allowed the legislation arise from interpretation and to temporarily lay off workers for a period court judgments on what constitutes “core� of three months on half pay on application to or “perennial� activities, for which contracts and approval by the commissioner of labor. are prohibited, and pressure to absorb con- If after three months the employer was still tract labor into permanent employment. unable to resume operations as normal, he or Andhra Pradesh has relaxed the use of con- she was required to apply for a second tem- tract labor in “core� activities; Goa has porary lay-off (Gunatilaka 2011). tightened the prohibition of contract labor in 21. The OECD measure calculates strictness of “noncore� activities. employment protection legislation based on 21 16. The relevant legislation is the West Pakistan indicators covering three areas of employment Industrial and Commercial Employment protection: dismissal protection for individual (Standing Orders) Ordinance of 1968. Fixed- workers with regular contracts, additional term contracts can be used only in “objec- costs for collective dismissals, and regulations tively de�ned cases.� governing temporary contracts. Summary 274 MORE AND BETTER JOBS IN SOUTH ASIA scores for each area are calculated as weighted 28. The discussion of compliance and dis- averages of the relevant indicators. An overall pute resolution draws on Kyloh (2008) for employment protection legislation measure is Nepal, World Bank (2006) for Pakistan, and calculated as a weighted average of the three Gunatilaka (2011) for Sri Lanka. summary scores. For details, see http://www 29. Information for India suggests that pay- .oecd.org/employment/protection. offs under voluntary retirement schemes are 22. Since the Industrial Disputes Act was amended far above the severance pay provided in the in 1965, any nonmutual layoff, retrenchment, Industrial Disputes Act and that average or closure becomes an industrial dispute, going observed costs place India among the most �rst to the Ministry of Labor for conciliation costly in the world for dismissals (Debroy and then, if conciliation fails, to tribunals of 2011; OECD 2007a). Use of golden hand- courts, which often order reinstatement. shakes does require approval by trade unions. 23. The more recent research builds on the work 30. Labor legislation may be only part of the story. of Besley and Burgess (2004), who exploit In Sri Lanka, for example, it is dif�cult to dis- variations across Indian states and over time entangle the effects of TEWA, the Payment of in amendments to central legislation. The Gratuity Act, and the Inland Revenue Act, as Besley-Burgess index of state employment all three kick in at a �rm size threshold of 14 protection legislation has been controversial workers (Gunatilaka 2011). Gunatilaka and (see Ahsan and Pagés 2009; Gupta, Hasan Vodopevic (2010) �nd suggestive evidence that and Kumar 2009; and OECD 2007 for criti- all three are likely responsible for the irregular cisms and approaches to addressing them). size distribution of �rms and limited growth 24. Analysis of the full sample of manufactur- of �rms with 14 workers. Abidoye, Orazem, ing �rms in the Annual Survey of Industries and Vodopivec (2009) test this hypothesis by data for 2000–08 shows increases in formal applying a difference-in-differences method to sector manufacturing beginning in 2005 in a panel data set of formal sector �rms. They both large (100 or more employees) �rms and distinguish between �rms in export processing smaller �rms. Large �rms account for 70 per- zones (EPZs), which enjoy tax exemptions cent of formal manufacturing employment. and other incentives offered by the Board of The analysis also shows rapid growth in the Investment, and others. They �nd that non- use of contract labor: the share of contract EPZ �rms at or below the threshold are less labor increased from 12 percent of total for- likely to increase employment than non-EPZ mal manufacturing employment in 1999 to 24 �rms above the threshold. Above the threshold, percent in 2008. The biggest increases were non-EPZ �rms are more likely to shed workers, in large �rms (100 or more employees), but and EPZ �rms are more likely to add workers. the share also increased (from 12 percent to At all sizes, EPZ �rms are more likely to add 19 percent) in smaller �rms. The job creation/ workers than non-EPZ �rms. job destruction analysis, which relies on panel 31. Firms reporting incidents reported that on data, uses all �rms for which two consecu- average, making informal payments cut the tive years of data are available. This analy- number of visits and time spent in dealing sis is possible only for �rms with more than with inspections by about half. 100 employees. Results for contract workers 32. Conciliation normally succeeds in cases in should be treated with some caution, as a zero which an agreement could have been arrived for contract workers in one year could simply at without conciliation (NCEUS 2009). reflect nonreporting and hence lead to arti�- 33. Information on these programs comes from cially large job creation or destruction �gures. country studies of active labor market pro- 25. Comparisons of enterprise survey results over grams commissioned for this book. time are not shown in �gure 6.7, which pools 34. Drèze and Khera (2009) �nd considerable all surveys for individual countries. latent demand for public works in their study 26. Many countries in the Europe and Central Asia of six northern states. In other states, such as region have large tax wedges, with unfavorable Kerala, �eld studies report a lack of demand employment consequences (Rutkowski 2007). (Jacob and Varghese 2006). 27. See Freeman (2009) for a more detailed 35. The Professional Institutional Network of the review. Minimum wages are not an effective Mahatma Gandhi National Rural Employment antipoverty tool. Guarantee Act (http://www.nrega.net/pin) is a THE ROLE OF LABOR MARKE T REGULATIONS, INSTITUTIONS, AND PROGRAMS 275 network of institutions, including the Indian likely required by young women in their Institutes of Management, the Indian Institutes households, communities, and the labor of Technology, the National Institute of Rural market, such as negotiation skills, dealing Development, agriculture universities, the with discrimination, workers’ rights edu- Administrative Staff College of India, think cation, reproductive health, basic �nancial tanks, civil society organizations, and other management, control over the use of their professional institutes. It was created by the own income, and mentoring). About 60 per- Ministry of Rural Development as an inte- cent of the �rst round of trainees has been grated structure for monitoring, appraising, gainfully employed after completing training diagnosing implementation constraints, and (Employment Fund 2010). recommending remedial action and sustain- 44. NSDC is jointly owned by the private (51 per- able interventions to enhance the quality of cent) and public (49 percent) sectors. It was the program. established as a nonpro�t company under 36. Moreover, there is little evidence that state- Section 25 of the Indian Companies Act. funded unemployment bene�ts are always 45. Micro�nance services include microcredit, made. deposit services, money transfer services, and 37. In their survey of 100 MNREGA worksites microleasing. spread across six northern states, Drèze and 46. See http://www.samn.eu/?q=micro�nance-asia. Khera (2009) �nd that a majority of people 47. These linkages are miniscule from the point of employed in the program come from the most view of the banks (EDA and APMAS 2006). disadvantaged segments of society. 48. For reviews of the limited evaluation evidence 38. An important question to be examined is the on self-employment programs in developing extent to which this national picture applies countries, see Betcherman and others (2004, at the state level, where there are important 2007). differences in implementation. 49. Some of the controversies relate to issues 39. A more detailed picture of participation in that are tangential to this book, including MGNREGA and a fuller evaluation of the the role of commercial banking institutions program should be obtained from special pur- in micro�nance, the interest rates charged pose surveys designed to assess the program. on microloans, the debt accumulated by 40. These shares of participation by social group poor people through microloans, and the from administrative data for 2010 are con- politicization of some programs. There sistent with those obtained from the 2009/10 is also a claim that the bene�ts of micro- National Sample Survey. �nance have been oversold. Although these 41. See Ambasta, Shankar, and Shah (2008) and are important concerns, they do not speak CSE (2008) for a discussion of the quality of directly to the question of whether micro- assets. �nance can be an effective instrument for 42. At the time of this review, nine rigorous helping informal workers manage labor evaluations of public works programs in market–related risks. developing countries were identi�ed. Of these, 50. In effect, the act provides for the formulation �ve found that participants had higher post- by the central and state governments of social program employment rates than nonpartici- insurance schemes for unorganized workers. pants in the control group, and four found no It does not establish a universal minimum effect. None of the three studies that looked social insurance program. at earnings found any positive effect. 51. The design of the bene�t side of social insur- 43. Some training programs in South Asia ance programs is beyond the scope of this are also beginning to reorient themselves book. Promising initiatives exist throughout to the nature of informal employment. In the region. One prominent example is India’s Nepal, for example, the recently introduced recently introduced national health insurance Adolescent Girls Employment Initiative tar- scheme, Rashtriya Swasthya Bima Yojana gets unemployed girls and girls working in (RSBY), designed to provide hospitalization the informal sector. In addition to providing coverage to households below the poverty technical training, the initiative also pro- line. For more details on the program and its vides job placement services and life-skills early implementation experience, see World training (covering a range of life skills most Bank (2011b). 276 MORE AND BETTER JOBS IN SOUTH ASIA References Countries?� Background study conducted for this book. Abidoye, B., P. Orazem, and M. Vodopivec. 2009. CSE (Centre for Science and Environment). 2008. “Firing Cost and Firm Size: A Study of Sri An Assessment of the Performance of The Lanka’s Severance Pay System.� Social Pro- National Rural Employment Guarantee Pro- tection Discussion Paper 0916, World Bank, gramme in Terms of its Potential for Creation Washington, DC. of Natural Wealth in India’s Villages. New Ahsan, A., and C. Pagés. 2009. “Are All Labor Delhi. Regulations Equal? Evidence from Indian de Mel, S., D. McKenzie and C. Woodruff. 2011. Manufacturing.� Journal of Comparative Eco- “Business Training and Female Enterprise nomics 37 (1): 62–75. Start-up and Growth in Sri Lanka.� Develop- Ambasta, P., P.S. Vijay Shankar, and M. Shah. ment Economics/Finance and Private Sector 2008. “Two Years of NREGA: The Road Development, World Bank, Washington, DC. Ahead.� Economic and Political Weekly 43 Debroy, B. 2011. “India’s Labour Market: Laws, (8): 41–50. Regulations and Reforms.� Background study Armendáriz de Aghion, B., and J. Morduch. conducted for this book. 2004. “Micro� nance: Where Do We Stand?� del Ninno, C., K. Subbarao, and A. Milazzo. In Financial Development and Economic 2009. “How to Make Public Works Work: A Growth: Explaining the Links, ed. C. Good- Review of the Experiences.� Social Protection hart 135–48. London: Palgrave. Discussion Paper 48567, World Bank, Wash- Banerjee, A., E. Duflo, R. Glennerster, and C. ington, DC. Kinnan. 2009. “The Miracle of Microfi- Dougherty, S. 2008. “Labour Regulation and nance? Evidence from a Randomized Evalu- Employment Dynamics at the State Level in ation.� Department of Economics and Abdul India.� OECD Economics Department Work- Latif Jameel Poverty Action Lab Working ing Paper 624, Organisation for Economic Paper, Massachusetts Institute of Technology, Co-operation and Development, Paris. Cambridge, MA. Drèze, J., and R. Khera. 2009. “The Battle for Besley, T., and R. Burgess. 2004. “Can Labor Employment Guarantee.� Frontline 26: 1. Regulation Hinder Economic Performance? Dutta, K. 2011. “Active Labor Market Programs Evidence from India.� Quarterly Journal of in India.� Background study conducted for this Economics 119 (1): 91–134. book. Betcherman, G., A. Dar, A. Luinstra, and EDA (EDA Rural Systems Pvt Ltd), and APMAS M. Ogawa. 2004. “Active Labor Market (Andhra Pradesh Mahila Abhivruddhi Society). Programs: Policy Issues for East Asia.� Social 2006. Self-Help Groups in India: A Study of Protection Discussion Paper 0005, World the Lights and Shades. Gurgaon and Hydera- Bank, Washington, DC. bad, India. Betcherman, G., M. Godfrey, S. Puerto, F. R. Employment Fund. 2010. Employment Fund Rother, and A. Stavreska. 2007. “Global Inven- Half Annual Report (January–June 2010). tory of Interventions to Support Young Work- Kathmandu. http://www.employmentfund ers: Synthesis Report.� World Bank, Social .org.np/annual-reports.html/. Protection Unit, Washington, DC. Federal Bureau of Statistics Pakistan. 2011. Bruhn, M., and B. Zia. 2011. “Stimulating Man- Paki s t a n S t a ti s ti c al Ye a r Book 2 011. agerial Capital in Emerging Markets: The Islamabad. Impact of Business and Financial Literacy for Freeman, R. 1993. “Labor Market Institutions Young Entrepreneurs.� Policy Research Work- and Policies: Help or Hindrance to Economic ing Paper 5642, Development Research Group, Adjustment?� Proceedings of the World Bank Finance and Private Sector Development Team, Annual Conference on Development Econom- World Bank, Washington, DC. ics, ed. M. Bruno and B. Pleskovic, 117–44. Card, D., and A. Krueger. 1997. Myth and Mea- Washington, DC: World Bank. surement: The New Economics of the Mini- ———. 2009. “Labor Regulations, Unions, and mum Wage. Princeton, NJ: Princeton Univer- Social Protection in Developing Countries: sity Press. Market Distortions or Ef�cient Institutions?� Carlin, W., and M. Schaffer. 2011. “Which Ele- N BER Working Paper 14789, National ments of the Business Environment Matter Bureau of Economic Research, Cambridge, Most for Firms and How Do They Vary across MA. THE ROLE OF LABOR MARKE T REGULATIONS, INSTITUTIONS, AND PROGRAMS 277 Gine, X., and G. Mansuri. 2011. “Constraints to Paper 5186, World Bank, Latin America and Female Entrepreneurship: Ideas or Capital.� the Caribbean Region, Poverty and Gender World Bank, Development Research Group, Unit, Washington, DC. Washington, DC. Munshi, F. 2011. “L abor Regulations in Gunatilaka, R. 2011. “Institutional Constraints to Bangladesh: Constraints to More and Better Decent Job Growth in Sri Lanka’s Segmented Jobs?� Background study conducted for this Labour Market.� Background study conducted book. for this book. NCAER–PIF (National Council of Applied Eco- Gunatilaka, R., and M. Vodopivec. 2010. “Labor nomic Research–Public Interest Foundation). Market Institutions and Labor Market Seg- 2009. Evaluating Performance of National mentation in Sri Lanka.� In The Challenge Rural Employment Guarantee Act. New of Youth Unemployment in Sri Lanka, ed. Delhi. R. Gunatilaka, M. Mayer, and M. Vodopivec, NCEUS (National Commission for Enterprises in 49–68. Washington, DC: World Bank. the Unorganized Sector). 2009. The Challenge Gupta, P., R. Hasan, and U. Kumar. 2009. “Big of Employment in India: An Informal Econ- Reforms but Small Payoffs: Explaining the omy Perspective. Final Report of the National Weak Record of Growth and Employment in Commission for Enterprises in the Unorganised Indian Manufacturing.� MPRA Paper 13496, Sector. Government of India, New Delhi. Munich Personal RePEc Archive, University Neumark, D., and W. Wascher. 2006. “Minimum Library of Munich. Wages and Employment: A Review of Evidence Heltberg, R., and M. Vodopivec. 2004. Sri from the New Minimum Wage Research.� Lanka: Unemployment, Job Security, and NBER Working Paper 12663, National Bureau Labor Market Reform. http://ssrn.com / of Economic Research, Cambridge, MA. abstract=1208662. OECD (Organisation for Economic Co-operation Holzmann, R., Y. Pouget, M. Vodopivec, and M. and Development). 2006. Employment Out- Weber. 2011. “Severance Pay Programs around look. Paris. the World: History, Rationale, Status, and ———. 2007a. Economic Survey of India. Paris. Reforms.� Social Protection Discussion Paper ———. 2007b. Employment Outlook. Paris. 62726, World Bank, Washington, DC. ———. 2009. Indicators of Employment Protec- IHD (Institute for Human Development). 2011. tion. Paris. Human Development in India: Emerging Palacios, R., and E. Whitehouse. 2006. “Civil Issues and Policy Perspectives. New Delhi. Service Pension Schemes around the World.� IPS (Institute of Policy Studies of Sri Lanka). 2011. Social Protection Discussion Paper 0602, Review of Active Labor Market Programs World Bank, Washington, DC. (ALMPS) in Sri Lanka. Colombo. Rahman, R., and N. Shamsunnahar. 2011. “A Jacob, A., and R. Varghese. 2006. “NREGA Review of Active Labour Market Programs implementation 1: Reasonable Beginning in in South Asian Countries: Bangladesh.� Back- Palakkad, Kerala.� Economic and Political ground study conducted for this book. Weekly 2 (December): 4943–45. Rifaqat, S. 2011. “A Review of Active Labor Mar- Johanson, R., and A. Van Adams. 2004. Skills ket Programs in Pakistan.� Background study Development in Sub-Saharan Africa. World conducted for this book. Bank, Regional and Sectoral Studies, Wash- Rutkowski, J. 2007. “Taxation of Labor.� In Fis- ington, DC. cal Policy and Economic Growth in ECA, Kyloh, R. 2008. From Confl ict to Cooperation: ed. C. Gray, T. Lane, and A. Varoudakis, 281– Labour Market Reforms That Can Work in 313. Washington, DC: World Bank. Nepal. International Labour Of�ce, Geneva, SAMN (South Asian Microfinance Network). and Academic Foundation, New Delhi. 2010. “Micro�nance in the South Asian Coun- Loayza, N., L. Servén, and N. Sugawara. 2010. tries: Quick Facts.� New Delhi. http://www “Informality in Latin America and the Carib- .samn.eu/?q=micro�nance-asia. bean.� In Business Regulation and Economic Tamiru, K. 2011. “Active Labor Market Policies Performance, ed. N. Loayza and L. Servén, Inventory Review for Afghanistan.� Back- 157–96. Washington, DC: World Bank. ground study conducted for this book. López-Acevedo, G. and M. Tinajero. 2010. Tan, H., and G. López-Acevedo. 2005. “Evaluat- “Impact Evaluation of SME Programs Using ing Training Programs for Small and Medium Panel Firm Data.� Policy Research Working Enterprises: Lessons from Mexico.� Policy 278 MORE AND BETTER JOBS IN SOUTH ASIA Research Working Paper 3760, Latin America Formation in the Manufacturing Sector. South and the Caribbean Region, Poverty Sector Asia Region, Finance and Private Sector Devel- Unit, World Bank, Washington, DC. opment, Washington, DC. UN (United Nations). 2010. World Population ———. 2007. “Skill Development in India: The Prospects: The 2010 Revision. New York. Vocational Education and Training System.� U.S. Department of State. 1999–2010. Human South Asia Region, Human Development Unit, Rights Report. Washington, DC. Washington, DC. Vodopivec, M. 2004. Introducing Unemploy- ———. 2009. Financing the Bottom of the Pyra- ment Bene�ts to Sri Lanka. Washington, DC: mid: Microfinance Strategy for South Asia. World Bank. South Asia Region, Washington, DC. Vodopivec, M., A. Wörgötter, and D. Raju. 2005. ———. 2010. India’s Employment Challenge: “Unemployment Bene�t Systems in Central and Creating Jobs, Helping Workers. India: Oxford Eastern Europe: A Review of the 1990s.� Com- University Press. parative Economic Studies 47 (4): 615–51. ———. 2011a. Doing Business Database. Wash- Wall Street Journal Asia. 2011. “India’s Guar- ington, DC: World Bank. anteed Joblessness.� July 27. http://online.wsj. ———. 2011b. Social Protection for a Changing com/article/SB1000142405311190359110457 India. South Asia Region, Human Develop- 6467562817474174.html. ment Unit, Washington, DC. World Bank. 2006. Pakistan Labor Market ———. 2011c. World Development Indicators. Study: Regulation, Job Creation and Skills Washington, DC: World Bank. CHAPTER 7 Creating Jobs in Conflict-Affected Areas Questions and Findings Questions low-skilled workers, and more likely to have vulnerable jobs. • How does armed conflict affect labor markets? • In the initial postconflict phase, labor market • What are the key constraints to job creation programs and policies can play important in conflict-affected areas of South Asia? roles in providing and promoting employ- • What are the policy priorities for creat- ment, through such efforts as disarmament, ing more and better jobs in conflict-affected demobilization, and reintegration programs areas? and broader-based public works. Govern- ments also need to take initial steps to attract the private sector (such as reforming business Findings regulations, issuing temporary tax breaks, • Even when it is over, armed conflict remains a and forging public-private partnerships) so serious obstacle to job creation in South Asia. that the private sector can eventually become Conflict affects both the demand for and the a more signi�cant provider of jobs. Interna- supply of labor. It affects demand by reduc- tional organizations and foreign governments ing both the incentives and the ability of �rms have important roles to play in providing to invest and create jobs. Lack of security, funding and building capacity in these early inadequate infrastructure and services, and a stages, particularly in cases of nationwide weak governance and regulatory environment conflict. are the most signi�cant deterrents for private • Policy makers should plan three important sector job creation in conflict-affected areas. transitions for policies and programs over Loss of skills, migration, fluctuations in fam- the medium term: increasing the role of the ily income, and demographic changes also private sector; refocusing attention from the affect the nature of the workforce in conflict agricultural sector toward facilitating higher- zones. Conflict-affected areas in South Asia productivity employment in other sectors; and remain more rural, more dependent on agri- moving from targeted programs for vulnerable culture, and more isolated than low-intensity groups (such as excombatants, at-risk youth, conflict areas. They also remain less likely to war victims, and displaced people) to broad- have private employers, more likely to have based employment generation. Creating Jobs in Conflict-Affected Areas 7 A rmed conflict represents a major that led to conflict in the � rst place. House- challenge to job creation. Conflict hold surveys and focus group discussions in destroys communities, institutions, conflict-affected areas in Afghanistan and Sri and physical and human capital and leaves Lanka show that economic recovery and the the affected population living in insecurity restoration of livelihoods are people’s main and often without sources of livelihood. priorities after the cessation of hostilities More than 58,000 people worldwide died (World Bank 2010a). The fact that millions directly as a result of armed conflict in of new entrants will enter the South Asian 2009—more than a third of them in South labor force in the next two decades further Asia (IISS 2010). Most countries in South increases pressure to create more and better Asia are immersed in—or only recently jobs, especially in postconflict environments emerged from—armed conflict, ranging and areas at risk of new or renewed violence. from the insurgencies in Afghanistan and This chapter describes the key challenges to Pakistan and low-level localized insurgency job creation in conflict-affected environments in India to the recently ended civil wars in and outlines what governments in the region Nepal and Sri Lanka.1 Some of these con- can do to overcome them.2 The next section flicts have of�cially ended, but even in these discusses the characteristics and intensity of cases sustainable peace is not guaranteed. armed conflict in South Asia. The second The creation of more and better jobs section discusses the constraints to labor should be at the center of any recovery plan demand and labor supply in conflict-affected for confl ict-affl icted areas, as the literature areas. The third section takes a closer look suggests that poor economic conditions and at the profile of labor markets in conflict- labor market prospects increase the likelihood affected areas in South Asia. The fourth of conflict. At the same time, a minimal level section focuses on the role the private sector of security is essential to economic recovery can play in creating jobs in confl ict-affected and to the success of policies and programs areas, the main constraints it faces, and the that encourage job creation. Jobs and sustain- lessons international experience yields regard- able peace can therefore positively reinforce ing effective interventions. The fifth sec- each other, as long as programs and policies tion discusses education delivery in confl ict are inclusive and sensitive to the grievances situations. The sixth section focuses on the 281 282 MORE AND BETTER JOBS IN SOUTH ASIA role of labor market policies and programs. of nonstate actors and directed against the The last section discusses some of the key apparatus of the state or with the intent of transitions needed in conflict-affected areas if destabilizing the state. This definition of job creation is to be sustainable. It identi�es internal conflict is used in the Uppsala Con- the main actors, the key sectors to focus on, flict Data Program and Centre for the Study and the groups to target. of Civil War at the International Peace Research Institute (UCDP/PRIO) conflict database. It is similar to the definition of Characteristics and intensity terrorism adopted in most incident-level data of armed conflict in South Asia sets dealing with internal conflict. The de�ni- South Asia has experienced one of the high- tion excludes interstate conflicts and violence est levels of armed confl ict in the world. On between individuals. average, countries in the region experienced The focus in this chapter is on internal armed conflict for half of the years since confl ict against the state for two main rea- 2000, as measured by the incidence of at sons. First, internal armed confl ict directed least 25 battle-related deaths a year (�gure against the state arguably poses the greatest 7.1). This is the highest incidence of confl ict challenge to the ability of the state to imple- in absolute terms and the second-highest ment policies and programs conducive to proportion of deaths from armed conflict employment generation. Second, the majority in the world. Among the top 10 countries in of confl icts in the developing world, includ- terms of direct deaths from armed conflict in ing South Asia, are internal confl icts. Since 2008, 4 (Afghanistan, Pakistan, India, and World War II, internal conflicts have resulted Sri Lanka) were in South Asia (�gure 7.2).3 in three times as many deaths as interstate This chapter focuses on internal armed wars (Fearon and Laitin 2003). violence against the state—violent incidents The nature, duration, and intensity of perpetrated by a relatively organized group internal armed conflicts vary across South FIGURE 7.1 Proportion of country-years in armed conflict, by region, 2000–08 0.6 0.50 0.5 0.4 proportion 0.3 0.20 0.2 0.16 0.15 0.1 0.09 0.05 0.0 sia ca Af and ia ia be d ib an As As fri hA a an nA ric rth st ar ca l st ra ut No le Ea Ea e C ri nt ra th Ame So Ce ha d Sa id d tin an M b- La Su pe ro Eu Source: UCDP/PRIO Armed Conflict Dataset version 4-2009 (http://www.prio.no/CSCW/Datasets/Armed-Conflict/UCDP-PRIO/Armed-Conflicts-Version-X-2009/). Note: Armed conflict refers to internal armed conflicts between the government of a state and one or more internal opposition groups that result in at least 25 battle-related deaths a year. CREATING JOBS IN CONFLIC TAFFEC TED AREAS 283 FIGURE 7.2 Top 15 countries in number of deaths Asian countries. Ongoing conflicts in the from armed conflict, 2008 region include the confl icts in Afghanistan and Pakistan, the long-running insurgency Lebanon movements in India’s northeastern regions, Algeria Chad and the escalating violent activities of left- Russian Federation leaning groups in the eastern and central Philippines parts of India. Table 7.1 provides informa- Colombia tion on the major internal armed conflicts Sri Lanka in South Asia over the past decade (see Iyer Sudan 2009 for a discussion of these conflicts). Thailand Congo, Dem Rep. The challenges posed by armed conflict India vary signi�cantly with the type of confl ict. Somalia The focus here is on two dimensions of con- Afghanistan flict of particular importance for job creation: Pakistan the geographic scope and the stage of the Iraq conflict. Geographic scope refers to whether 0 1,000 2,000 3,000 4,000 5,000 number of deaths the confl ict is nationwide or localized. The stage of the conflict refers to whether the Source: National Counterterrorism Center 2009, based on the Worldwide confl ict is ongoing or has ended. A confl ict Incidents Tracking System (WITS), which contains details about incidents of violence against civilians and noncombatants from publicly available is classi�ed as having ended if there has been information. a clear military victory or a peace agreement TABLE 7.1 Major internal armed conflicts in South Asia since 1978 Outstanding Estimated number number of Approximate start of fatalities internally Country/region and end dates (2000–09) displaced people Afghanistan Nationwide 1978– 51,580 More than 235,000 Bangladesh Nationwide 2005–07 93 — India Assam 1990– 3,838 257,000 Manipur 1982–2008 2,712 180,000 Nagaland 1992–2007 710 62,000 Tripura 1978–2006 1,708 — Eastern and central states 1980– 5,290 60,000 Nepal Nationwide 1996–2006 13,592 60,000 Pakistan Baluchistan 2004– 1,294 80,000 North West Frontier Province (renamed Khyber Pakhtunkhwa in 2010) and the Federally Administered Tribal Areas 2002– 20,261 900,000 Sri Lanka Northern and Eastern Provinces 1980–2009 51,815 320,000 Source: Authors, based on start and end dates from the UCDP/PRIO Armed Conflict Dataset version 4-2009 (http://www.prio.no/CSCW/Datasets/Armed- Conflict/UCDP-PRIO/Armed-Conflicts-Version-X-2009/) and number of fatalities and displaced people from the Armed Conflict Database compiled by the International Institute for Strategic Studies, based on data from the Internal Displacement Monitoring Centre (http://www.iiss.org/publications/armed- conflict-database/). Note: — = Not available. 284 MORE AND BETTER JOBS IN SOUTH ASIA TABLE 7.2 Armed conflicts in South Asia, by conflict stage, FIGURE 7.3 Effects of conflict on demand for and geographic scope, and number of casualties per thousand people supply of labor Conflict stage Conflict Ongoing Postconflict Nationwide Afghanistan (1.80) Nepal (0.475) Factors affecting labor Factors affecting Geographic demand labor supply Localized Pakistan (0.130) Sri Lanka (2.570) scope India (0.013) • Lack of security • Lack of security • Poor regulatory • Losses of education Sources: Authors, based on data on casualties from the armed conflict database compiled by the framework (regulatory International Institute for Strategic Studies (http://www.iiss.org/publications/armed-conflict- and skills uncertainty, weak database); population data from World Development Indicators; and Afghanistan population governance) • Displacement of estimate from the CIA Factbook. • Infrastructure losses population Note: Figures in parentheses are estimates of number of casualties per 1,000 people between 2000 • Changes in household and 2009. The classi�cation of conflict stage is based on the status as of December 2010. • High input prices income Source: Authors. between the warring parties and the number of conflict-related deaths has declined signi�- cantly as a result. quality and composition of the workforce. The conflicts in South Asia between 2000 Security concerns in traveling to work can be and 2010 show considerable variation along a major constraint to labor supply. Disrup- both of these dimensions (table 7.2). Armed tions to education and the loss of job-related conflicts in Afghanistan and Nepal were wide- skills and training reduce the quality of the spread in almost all parts of the country. In workforce in conflict areas, with potentially Afghanistan, the conflict continues. In Nepal, long-term effects. Population displacement the conflict was formally brought to an end by and migration, as well as the challenges posed a comprehensive peace agreement in 2006. by the reintegration of returnees after the end of confl ict, place further restrictions on the availability of skilled labor. Finally, labor Constraints to job creation supply will also be affected by changes in in conflict-affected areas household income resulting from the conflict Armed conflict affects the demand for labor, as, for example, women start working to sup- because firms and businesses are reluctant plement the family’s income in the absence of to invest in the face of signi�cant threats to male family members who join the �ghting. the security of life and property (�gure 7.3). Although all of these constraints apply in Inadequate or confusing regulations and the some degree to all confl ict situations, poor poor quality of governance in affected areas governance and inadequate infrastructure are also discourage firms from investing. The more likely to be severe constraints in areas of presence of warring parties and fragile states nationwide conflict, because regional conflict adds to uncertainty regarding both what the typically coexists with a functioning state in regulations are and who is enforcing them. the rest of the country. This means that the Infrastructure losses and unreliable access government has a tax base on which to draw, to key services, such as electricity and trans- the ability to provide public services and port, restrict the scale of �rm activity, and the infrastructure, and functioning institutions absence of markets increases input prices. At (army, police forces, bureaucracy, judicial sys- the height of the armed conflict in northern Sri tem), which can be extended to the conflict- Lanka in 2007, for example, the price of a bag affected region when the conflict ends. of cement was more than four times higher in In contrast, nationwide conflicts often Jaffna (Northern Province) than in other parts involve the building up or complete reform of the country (World Bank 2010c). of state institutions when the confl ict ends. Conflict also affects the capacity of the As discussed in the World Development population to supply labor, as well as the Report 2011 (World Bank 2011c), building CREATING JOBS IN CONFLIC TAFFEC TED AREAS 285 institutions is a long-term affair that gener- of confl ict (Collier, Hoeffler, and Sambanis ally takes a generation. Policy makers in 2005; Hoeffler 2011; Smith 2004; World countries emerging from nationwide conflict Bank 2011). But how does confl ict change face the additional challenge of reestablish- labor markets in confl ict zones? ing connectivity with the rest of the world or This section begins by briefly discussing neighboring countries, in addition to rebuild- the impact of armed conflict on aggregate ing internal connectivity. The lack of such economic growth and unemployment. It then external connectivity can signi�cantly affect analyzes the impact of confl ict on employ- access to foreign goods and capital, access to ment, unemployment, sectoral composition, foreign markets, and the ability of interna- and the types of jobs, using national house- tional agencies to deliver humanitarian and hold and labor force surveys. Multiple rounds other assistance. of household data were used to compare Internal migration to safer parts of the trends in high-conflict and low-conflict areas country is likely to be signi�cant in cases of of India, Nepal, and Sri Lanka (see annex localized conflict. In Pakistan, for example, 7A for de� nitions of high- and low-confl ict an estimated 60 percent of workers in the areas). Within-country variation in the inten- Federally Administered Tribal Areas (FATA) sity of armed conflict was exploited to better leave their tribal area to find employment understand the relationship between conflict (World Bank and ADB 2010).4 Migration and labor markets. can help households access more and better Two caveats should be heeded in interpret- jobs and send home money that can be used ing the results. The �rst refers to interpreting for human capital investments or business results as causal. Labor market characteris- start-ups. The presence of large numbers of tics may be different in conflict areas because internal migrants and displaced people raises of the conflict itself or because of preexisting an important policy question related to job conditions. Where information for more than creation, however: should jobs for internal one point in time is available, the data can be migrants be created in their current areas interpreted as caused by the confl ict, to the of residence, or should the focus be primar- extent that the two types of areas would have ily on creating jobs in conflict-affected areas, followed similar trends in the indicators of thereby providing incentives for the displaced interest in the absence of the armed conflict.5 to return? For Afghanistan, where household data for When the confl ict is ongoing, it may not only one year are available, results should be pay to rebuild infrastructure, which could be interpreted with more care. damaged during �ghting. In postconflict situ- The second caveat refers to the potential ations, the intensity of violence has decreased, bias in the sample composition, as areas allowing better planning and reducing uncer- affected by conflict are often underrepre- tainty in investment and resource allocation sented in household surveys. If these areas decisions. But violence also characterizes were included in the surveys, it is possible many postconflict situations. Peace is very that labor market conditions in conflict- fragile: historically, 40 percent of postconflict affected areas would look even worse, as the situations have reverted to conflict within worst-affected areas are probably omitted. a decade (Collier, Hoeffler, and Söderbom Annex 7B describes the difference in trends 2008). for South Asian countries for employment and unemployment, disaggregated by gender.6 Armed conflict and labor markets Conflict, growth, and the peace dividend Labor market indicators (such as lack of jobs for young people and low wage growth) Confl ict can have aggregate effects on eco- are important predictors of the likelihood nomic growth and therefore on the quantity 286 MORE AND BETTER JOBS IN SOUTH ASIA and quality of jobs the economy is able to FIGURE 7.4 Annual growth in GDP and number of generate.7 battle deaths in India and Nepal, 2000–08 Assessing the economic costs of armed a. India conflict is empirically complicated, because it 12 is dif�cult to specify an appropriate counter- 10 factual, data are not available from conflict- affected countries or regions, and different 8 percent methodologies often yield widely different 6 results (Gardeazabal forthcoming). For exam- 4 ple, in 1998 the macroeconomic impact of the civil war in Sri Lanka was of�cially 2 estimated at 2–3 percent of gross domestic 0 1,000 2,000 3,000 4,000 5,000 6,000 product (GDP) growth a year since its start number of battle deaths (Central Bank of Sri Lanka 1998). More b. Nepal recent estimates for the period 1984–96 put 6 the annual cost at 10.8–15.8 percent of 1996 5 GDP (Arunatilake, Jayasuriya, and Kelegama 4 2001). percent Cross-country estimates find that coun- 3 tries that experienced civil war between 1960 2 and 1999 saw their annual GDP growth rate 1 reduced by 2.4 percentage points on average (Hoeffler and Reynal-Querol 2003). Even for 0 0 1,000 2,000 3,000 4,000 countries with some confl ict during 1980– number of battle deaths 2008, growth was 1.2 percentage points Source: Authors, based on data from World Bank 2011b. higher, on average, in years of peace than Note: Each data point represents a single year. in years of conflict. In South Asia as well, periods with more conflict have usually been FIGURE 7.5 Unemployment rates in the Northern periods of slow growth (�gure 7.4). and Eastern provinces of Sri Lanka, 1997–2001 and This evidence suggests that the peace divi- 2002–04 dend for growth and job creation is poten- tially large. Chen, Loayza, and Reynal-Querol 16 14 (2007) analyze 41 countries that experienced 12 civil war between 1960 and 2003. They �nd 10 that once the war ended, recovery in eco- percent 8 nomic performance, health, education, and 6 political development was signi�cant. 4 Once violence ceased, conflict-affected 2 communities also experienced a faster reduc- 0 ce ce de vin in i nw tion in poverty than communities less directly ov ro tio Pr nP na rn er exposed to the conflict within the same ste rth Ea No country. Poverty reduction was particularly conflict years (1997–2001) ceasefire years (2002–04) high in Sri Lanka, where it fell 40 percent in confl ict-affected areas (the average decline Source: Annual Reports of the Central Bank of Sri Lanka. elsewhere in the country was 34 percent) (World Bank 2010a).8 During the cease� re to 10.5 percent in the Eastern Province, at a that lasted from 2002 to 2004, unemploy- time when the national unemployment rate ment fell from 13.0 percent to 9.2 percent in decreased only slightly, from 8.8 percent to the Northern Province and from 15.9 percent 8.3 percent (�gure 7.5). CREATING JOBS IN CONFLIC TAFFEC TED AREAS 287 Employment rate FIGURE 7.6 Percentage of working-age population employed in high-conflict and low-conflict areas of India Armed confl ict seems to be associated with and Nepal an increase (or a smaller reduction) in high- conflict areas in the share of the working-age a. India 70 population that is economically active and employed.9 In Afghanistan, the share of the 65 working-age population that was employed was 55.6 percent in low-confl ict provinces percent and 68.3 percent in high-confl ict areas. In 60 India, employment rates in confl ict-affected areas fell less than in peaceful areas. In 55 Nepal, the proportion of the working-age population that was employed increased 50 1998 2000 2002 2004 2006 2008 about 2 percentage points between 1996 (preconflict) and 2004 in low-conflict 90 b. Nepal areas and more than 4 percentage points in high-confl ict areas (�gure 7.6). This evi- 85 dence suggests that confl ict could increase labor supply even in the face of decreased percent 80 demand.10 Together these forces are likely to reduce equilibrium wages in conflict areas over time, leading to a decline in liv- 75 ing standards. Increased labor force participation and 70 1994 1996 1998 2000 2002 2004 2006 employment in confl ict zones seems to arise low-conflict areas high-conflict areas primarily from the increase (or smaller reduction) in employment of women. In Source: Authors, based on national household and labor force surveys. Afghanistan, for example, the proportion of men employed is 4 percentage points FIGURE 7.7 Percentage of working-age population higher in high-confl ict areas, but the pro- employed in high-conflict areas of Nepal, by gender, portion of working-age women employed is 1996–2004 19 percentage points higher. The difference in employment patterns over time is statis- 95 tically insignificant for men in India and Nepal; among women, the likelihood of 90 men being employed increases with the intensity of the confl ict in both countries. In high- 85 percent women conflict areas of Nepal, for example, the proportion of working-age men increased 80 about 3 percentage points between 1996 (preconfl ict) and 2004, whereas the propor- 75 tion of women increased almost 6 percent- 1996 1998 2000 2002 2004 age points (�gure 7.7). Source: Authors, based on data from Nepal Living Standard Measurement Study Surveys. There are two potential reasons for the increased participation of women. One is the “added-worker� effect, whereby additional these conditions are likely to apply in conflict- members of the household enter the labor affected areas, given that men often leave to force in response to economic necessity or �ght and economic conditions in the house- the absence of key earning members. Both of hold usually worsen as a result of the conflict. 288 MORE AND BETTER JOBS IN SOUTH ASIA A rise in employment among women dur- likelihood of armed conflict on the one hand ing dif�cult economic times has been docu- and youth unemployment and other indica- mented in cross-country studies (Bhalotra tors of economic opportunities on the other and Umana-Aponte 2010); during the Asian (Collier, Hoeffler, and Sambanis 2005; Urdal crisis (Jones and others 2009); and in Nepal 2006). A similar result is obtained across (Menon and van der Meulen 2010). Several states in India, especially when combined researchers have also documented increased with slower wage growth in rural areas female labor force participation where men (Urdal 2008). There is also evidence that are absent because of war (see Finegan and young adolescents are the optimal target of Margo 1994, who document this phenom- rebel forces (Beber and Blattman 2010). enon for advanced economies during World Other work has failed to � nd economet- War II, and Schweitzer 1980). Another pos- ric evidence of a relationship between youth sible reason for the increased labor force unemployment and armed confl ict. The evi- participation of women in conflict-affected dence overall remains mixed, as discussed in areas could be the higher prevalence of the World Development Report 2011 (World agricultural employment in these areas, Bank 2011c). an activity where women traditionally In South Asia, some limited evidence sup- participate.11 ports the hypothesis that male youth unem- ployment may be a risk factor for confl ict. In Nepal, where labor market information Unemployment before the violence started is available, pre- Unemployment, particularly of youth, conflict unemployment rates among men is often associated with a higher risk of 15–35 years old were 2.4 percentage points conflict. higher (almost twice as high) in districts Two complementary theories try to where the conflict started (in the midwestern explain the relationship between (male) youth hills) than in the rest of the country, even after unemployment and armed conflict. The �rst, accounting for differences in poverty, loca- the youth bulge theory, states that when a tion (urban or rural), and education levels.12 large proportion of the population consists In South Asia as a whole, the more important of young men who are marginalized from question is likely to concern the relationship political and economic opportunities, coun- between the quality of jobs available to youth tries are more vulnerable to instability, and and conflict, as open unemployment is gener- violence becomes a rational option for engag- ally low. ing the system and making demands. The evidence from South Asia on the The second theory is that youth unemploy- impact of confl ict on overall unemployment ment and poverty correlate with the onset of is mixed. In Nepal, the conflict seems to have armed conflict because the opportunity cost been associated with a reduction in the unem- of joining rebel organizations is lower. In ployment rate in high-conflict areas. In India Peshawar, Pakistan, for example, a militant and Sri Lanka, the difference is not statisti- recruit can earn PRs 15,000–20,000 ($176– cally signi�cant (see annex 7B). There is also $235) a month—nearly twice the remunera- no clear pattern across countries in relative tion for unskilled work in the area (World unemployment rates before the start of the Bank and ADB 2010). The high unemploy- conflict (see annex 7B).13 ment among people 15–24 (close to 10 per- cent in Bangladesh, India, and Pakistan; Employment patterns by location 16 percent in Maldives; and almost 20 per- and sector cent in Sri Lanka) bodes ill for the possibil- ity of conflict. Confl ict-affected countries tend to be less Some cross-country empirical evidence urbanized, in part because the mountain- finds a positive correlation between the ous and forested terrain is usually favorable CREATING JOBS IN CONFLIC TAFFEC TED AREAS 289 for rebellion (see Collier and Sambanis FIGURE 7.8 Urban workers as share of all 2005; Do and Iyer 2010). At the global level, workers in India and Nepal 58 percent of the population in conflict- affected countries lives in rural areas—a 30 a. India proportion that is 11 percentage points higher than the average in nonconfl ict coun- 25 tries (World Bank 2011b). Jobs in confl ict areas are therefore more likely to be in rural 20 areas and the agricultural sector. percent In South Asia, too, the workforce in 15 high-conflict areas is also signi�cantly more 10 likely to be in rural areas. In Afghanistan, for instance, 22 percent of the working- 5 age population lives in urban areas in low- conflict areas; the corresponding figure 0 for high-conflict areas is 5 percent. Work 00 02 04 06 08 20 20 20 20 20 is primarily in agriculture: 71 percent of workers in high-conflict areas work in this 30 b. Nepal sector, compared with 52 percent in the rest of Afghanistan. In India, the share of 25 employment in agriculture was 61 in high- conflict and 58 percent in low-conflict 20 areas. In Nepal, 87 percent of workers in high-conflict and 78 percent in low-conflict 15 countries work in agriculture. percent Conflict delays urbanization and the 10 structural transformation of the economy. 5 The proportion of the workforce located in urban areas increased by 10 percent- 0 age points in low-confl ict areas of Nepal 96 98 00 02 04 19 19 20 20 20 between 1996 (preconfl ict) and 2004 but by only 4 percentage points in the high- low-conflict areas high-conflict areas confl ict areas (�gure 7.8). To the extent that Source: Authors, based on national household and labor force surveys. urban areas are able to provide better jobs and better access to public services, this pattern reflects a slowdown in the improve- ment of household welfare. The same pat- Mozambique, Rwanda, and Uganda (Brueck tern is evident in India. 2004; Deininger 2003; McKay and Lover- In India and Nepal, armed conflict seems idge 2005). to have also delayed the transition of the Economic activities outside of agriculture workforce out of agriculture. In India, for are very limited in confl ict-affected areas in example, the share of workers employed South Asia, as evidenced by the small share in agriculture fell throughout the country of the workforce employed by a third party between 2000 and 2007, but the decline was and the limited presence of manufactur- 3 percentage points lower in conflict-affected ing, construction, and commercial activities areas. In Nepal, agriculture fell in all districts (see annex 7B). This finding is important, between 1996 and 2004, but the average because the international evidence shows that decline was 2 percentage points lower in high- diversi�ed economies experience less conflict conflict areas (see annex 7B). These results than those that rely on a few sectors (Collier, echo findings in other settings, including Hoeffler, and Sambanis 2005). It is especially 290 MORE AND BETTER JOBS IN SOUTH ASIA relevant for countries like Afghanistan and in the likelihood of being an unpaid family Nepal, which remain overwhelmingly agrar- worker. This pattern partly reflects the higher ian. Fostering small and medium-size non- concentration of employment in agriculture, farm enterprises in conflict-affected areas can but it is also evident in urban areas. contribute to the economic development of The patterns on casual wage work are those areas and may also contribute to peace. mixed, with confl ict associated with lower shares of wage work in casual activities in Afghanistan and Nepal, higher shares in Sri Types of jobs Lanka (possibly because of the sizable expan- Jobs in high-conflict are more likely to sion of the construction sector), and no appar- remain vulnerable as a result of the con- ent association in India. A mixed picture also flict, for two main reasons. First, the dearth emerges for informal work. It is perhaps not of employers in high-confl ict areas means surprising that the evidence on vulnerability that people are significantly less likely to of jobs is stronger in the context of unpaid become regular wage or salaried employees. family workers than for casual wage work, In Nepal, the share of regular wage or sala- as casual wage work typically requires third- ried workers increased 1.2 percentage points party employers. between 1996 and 2004 in low-conflict dis- tricts but dropped in high-confl ict districts. Education In India, the share of workers in regular wage or salaried jobs increased 3 percentage Levels of education are signi�cantly lower in points in low-conflict areas, but the increase the areas of South Asia most affected by con- was only 1 percentage point in high-conflict fl ict. In Afghanistan in 2008, for instance, areas (see annex 7B).14 the share of the working-age population with Second, a larger proportion of workers in no education was 68 percent in low-confl ict confl ict areas remains dependent on unpaid areas and 82 percent in high-conflict areas. In family work. The share of workers engaged every South Asian country, a smaller propor- in unpaid family labor declined sharply in tion of the workforce had completed second- low-conflict areas of India; high-conflict ary education in conflict-affected areas than areas saw a much smaller decline (�gure 7.9). in low-conflict or peaceful areas, with the In Sri Lanka, the armed conflict was associ- differences both pre-dating the confl ict and ated with an increase of 7 percentage points resulting from the conflict (�gure 7.10). (For a breakdown of the working-age population by education level, see annex 7B.) The differ- FIGURE 7.9 Percentage of workforce employed ence in education levels is not related only to in unpaid family labor in India, 2000–08 lower urbanization: differences are smaller 30 but remain when urban and rural areas are analyzed separately. 25 These effects take place through the impacts 20 of conflict on supply and demand factors that affect schooling decisions. On the supply side, percent 15 armed conflict may lead to the destruction of 10 schools or complementary infrastructure and the absence or death of teachers. In Nepal, 5 for example, almost 5 percent of all victims 0 of the civil war have been identi�ed as teach- ers or students (INSEC 2010). In Pakistan’s 00 02 04 06 08 20 20 20 20 20 North West Frontier Province, 8 percent of low-conflict areas high-conflict areas schools were destroyed between 2002 and Source: Authors, based on labor force surveys. 2009 (World Bank 2009b). The availability CREATING JOBS IN CONFLIC TAFFEC TED AREAS 291 and quality of school facilities has been asso- FIGURE 7.10 Percentage of working-age population with ciated with student attendance and achieve- completed lower-secondary education in low- and high-conflict ment (Glewwe 2002). areas of selected South Asian countries Several factors—including the decline in 100 household income and job opportunities, 90 increased security risks, and lowered expected 80 returns to education—can reduce demand for 70 60 percent schooling in conflict areas: 50 40 • Increased pressure to supplement house- 30 hold income (or join the conflict) can lead 20 10 to children dropping out of school.15 0 • Lack of security makes it more dif�cult for Afghanistan, 2008 India, 2005 Nepal, 2004 Sri Lanka, 2008 students to reach school, and schools may low-conflict areas high-conflict areas be attacked. In Afghanistan, for instance, Source: Authors, based on data from national household and labor force surveys. threats of sexual violence reportedly keep many girls at home (Amnesty Interna- tional 2003). � nd that in Rwanda, where the conflict was • Conflict potentially reduces the expected short but intense, school enrollment rates returns to education by reducing job returned to their preconflict levels within �ve opportunities for educated people. years of the end of the violence. In Nepal, Between 2000 and 2008, the rate of emi- as discussed later in this chapter, important gration among South Asians with tertiary progress in education was made in confl ict- education was 4.7 percent in nonconflict- affected areas during the civil war. affected countries and 14.8 percent in con- In summary, labor markets in conflict-af- flict-affected ones (World Bank 2011b). fected areas present unique features. Overall, • Population displacement makes it dif�cult conflict-affected areas tend to be more rural to continue schooling. and more dependent on agriculture, rely less The literature reveals mixed results on the on regular wage or salaried employment and direction in which conflict has a causal effect more on unpaid family work, and have work- on schooling and skill acquisition. Using forces that are signi�cantly less educated. In detailed household and individual surveys, most cases, these characteristics reflect both several researchers � nd a signi�cant loss of the conflict and conditions predating it. education and experience for children and These characteristics of the labor markets in youth exposed to confl ict (see Verwimp and conflict-affected areas mean that job creation Van Bavel 2010 on Burundi; Akresh and de efforts need to initially focus on agriculture Walque 2008 on Rwanda; Shemyakina 2011 or other low-skilled sectors, such as construc- on Tajikistan; Blattman and Annan 2010 on tion. Additionally, providing inputs such as Uganda). Research by Stewart, Huang, and �nance, infrastructure, and training needs to Wang (2001), however, suggests that coun- be tailored to the informal and casual sector, tries with favorable institutions are able to in addition to the formal sector. maintain enrollment rates during armed con- fl ict or recover quickly to their preconfl ict enrollment levels once the conflict is over. Facilitating private sector In particular, they � nd that primary school job creation enrollment fell in only 3 of 18 African coun- Armed conflict, even conflict that has ended, tries affected by internal conflict and actually is a serious obstacle to job creation in South improved in 5 (it is important to note that the Asia. Chapter 4 showed that political insta- authors lacked appropriate counterfactuals bility is one of the three most common bind- in their exercise). Lopez and Wodon (2005) ing constraints for formal urban and rural 292 MORE AND BETTER JOBS IN SOUTH ASIA nonfarm �rms in South Asia. Almost 60 per- conflict was most violent reported that they cent of formal �rms in South Asia included were more severely constrained than �rms in in the World Bank’s enterprise surveys rank more peaceful areas in most aspects of the political instability as a major or severe con- business environment. The biggest differ- straint to doing business. In Afghanistan and ences between high- and low-conflict areas Nepal, respondents identi�ed political insta- were reported in infrastructure (electricity bility as the leading constraint; in Bangla- and transport); the governance and regu- desh and Pakistan, they ranked it among the latory environment (corruption, business top three constraints. Only in Bhutan was it licensing, courts, land); security (crime and a negligible concern.16 political instability); and skills (�gure 7.11). The second most important constraint The data on other South Asian countries do to � rm activities is the provision of electric- not allow for a similar analysis, because too ity. Key governance (e.g., corruption) and few �rms operating in conflict-affected areas regulatory issues are also cited as major con- were included in the enterprise surveys. The straints. These issues—insecurity, inadequate rest of this section discusses each of these infrastructure, and a poor regulatory frame- constraints to job creation, focusing on ways work—are precisely the factors that are likely in which different actors—the state, inter- to be exacerbated in conflict situations. national donors, nongovernmental organi- In Afghanistan’s 2008 enterprise survey, zations (NGOs), local communities—can for example, firms located in areas where address them. FIGURE 7.11 Severity of business environment constraints (average) reported by �rms in low-conflict and high-conflict areas of Afghanistan, 2008 political instability electricity crime, theft, and disorder corruption telecoms access to land business licensing transport tax administration courts competition inadequately educated labor customs labor regulations 0 1.0 2.0 3.0 4.0 severity of constraint high-conflict areas low-conflict areas Source: Authors, based on data from the 2008 Afghanistan enterprise survey. Note: High-conflict provinces are de�ned as provinces in which the number of deaths per 1,000 population caused by terrorist incidents in 2007 was greater than 0.1. By this de�nition, Baghlan, Ghazni, Farah, Helmand, Kandahar, Khost, Kunarha, Nimroz, Paktika, Paktya, Panjsher, Takhar, Urozgan, Wardak, and Zabul are high-conflict areas. Only differences that are statistically signi�cant at at least the 5 percent in an ordinary least squares regression including �rm size and industry �xed effects are shown. CREATING JOBS IN CONFLIC TAFFEC TED AREAS 293 Improving security Special economic safe zones Special economic safe zones are modeled Security risks, which affect the economic after special economic zones (SEZs).17 SEZs returns to investment, persist even after typically provide duty-free imports and good armed confl ict of�cially ends, putting both infrastructure, with a view to encouraging facilities and workers at risk. Firms need exports. In addition to easing the regula- additional funds to ensure security by hir- tory system or providing direct incentives, ing private security or paying off the warring such as tax breaks and loan guarantees, safe sides in a conflict. zones provide targeted security and other Firms also experience dif�culties in obtain- public services. ing supplies for production. Farmers do not SEZs have been established in several plant crops or go fishing when they live in postconfl ict countries, including Cambodia fear. The lack of security may also manifest and Mozambique, where they have helped itself in the hesitation of the local community attract foreign investors and spur job cre- to take part in economic activities led by the ation. In Cambodia, for example, export private sector (or the government for that processing zones have been associated with matter). the country’s impressive job creation per- Even after a confl ict has of�cially ended, formance after the war, especially among considerable risks to security remain. Nepal, women (female employment in industry grew for instance, witnessed 514 deaths attributed 20 percent a year compared with 7 percent to groups other than the Maoists and the for men) (McLeod and Davalos 2008). government in 2007, after the peace agree- In Sri Lanka, the government has been ment had been signed, an increase over the planning to establish economic zones in 327 such deaths in 2006 (INSEC 2008). conflict-affected areas (Indian Express Continued violence is attributable largely 2009). Policy makers should think of these to the inability of a fragile state to provide zones as an interim step in providing a good effective security to all parts of the country; business environment, gradually extending it highlights the risk that armed conflict can bene�ts such as better regulation and infra- transform itself into other forms of violence structure to �rms throughout the country. or instability, as discussed in the World SEZs could be further complemented with Development Report 2011 (World Bank “resource corridors,� which would more 2011c). Once � rms leave, they often do not directly link local communities, farmers, and come back. In Assam, India, and Sri Lanka, workers to areas with high concentration of for example, communities affected by armed natural resources (by, for example, strength- conflict report that factories and planta- ening logistics and infrastructure). Afghani- tions that shut down during confl ict did not stan’s mining sector, for example, has great reopen when security improved (World Bank potential. Although not labor intensive itself, 2010a). it could be better linked to existing economic Governments can manage some of the activities and livelihoods. This approach has security risks in conflict situations, by, for been tested predominantly in nonconflict instance, implementing disarmament, demo- environments but could also be a promising bilization, and reintegration (DDR) programs path in countries recovering from conflict. (discussed later in this chapter) and creating “safe� economic zones. Together with the pri- Private sector and local community vate sector, the government can also involve initiatives local communities in security arrangements. The private sector has found innovative The � rst two options are more likely to be ways of alleviating some of the challenges successful when violence has been signifi- stemming from lack of security in confl ict- cantly reduced and an end to the confl ict is affected areas by partnering with local in sight. entrepreneurs. Firms in confl ict areas often 294 MORE AND BETTER JOBS IN SOUTH ASIA hire private security firms or strike deals government in the host country and joint with warring parties. Foreign � rms also take ventures/alliances with local companies as political risk insurance to cover some risks, the two main tools they use to mitigate polit- through, for example, the World Bank’s ical risks in these countries (MIGA 2011). Multilateral Investment Guarantee Agency Another option is to involve local commu- (MIGA) or the U.S. government’s Overseas nities in security arrangements, as Roshan, a Private Investment Corporation (OPIC). telecommunications provider in Afghanistan, Respondents to a recent survey of � rms in has done (box 7.1). The community-based confl ict-affected cited engagement with the strategy resulted in better security for BOX 7.1 Private sector solutions to the security constraint: Lessons from Afghanistan Roshan is the largest mobile phone company in Taking this bold experiment a step farther, Afghanistan. Since beginning operations in 2003 in Roshan implemented it in Kandahar, one of Afghan- six major cities, the company has expanded opera- istan’s most confl ict-affected regions. Within a few tions to all 34 provinces and more than 230 cities months, the company saw dramatic results and soon and towns. By 2009, it had more than 3.5 million expanded the program to all areas of the country. active subscribers. The program is very popular in local communi- Since 2003, the company has invested more than ties, where it has generated jobs for local youth and $390 million all over the country. Employing more provided much needed infrastructure in a war-torn than 1,000 people directly (and more than 25,000 country. The initiative also reduced the company’s indirectly, as dealers, distributors, contractors, and security bill for its cell towers from $14 million to suppliers), Roshan is the largest private employer in $7.5 million. Afghanistan. One-� fth of its labor force is women, Roshan’s experience highlights three major issues in contrast to a women’s workforce participation of in conflict zones. First, demand for jobs is high. 1 percent in Afghanistan. Roshan is also the larg- Roshan reported receiving 10,000 applications for est taxpayer in Afghanistan, accounting for about every job it � lled in the early stages of the program. 6 percent of government revenues. Second, private sector companies can overcome the In 2008, Roshan faced a surge in attacks on disadvantages of operating in a high-risk environ- its cell network towers. These attacks strained ment, including security-related ones. Third, the Roshan’s operating budget, given that the company private sector can be a useful source of funding for was already paying $14 million a year to a private local infrastructure, without compromising pro�t- company to provide security. ability (Roshan has been pro�table since its second In response, Roshan decided to implement a new quarter of operation). community-based security model. The � rst step was Is Roshan’s experience replicable? One key to to consult with more than 1,000 local shuras (local success appears to be the company’s orientation councils) regarding their views of Roshan. The pri- toward social development in addition to pro�t mak- mary complaint was that Roshan was not providing ing. This orientation probably reflects the fact that jobs for locals; communities therefore had no inter- Roshan’s majority shareholder is the Aga Khan Fund est in supporting the company. for Economic Development, which has a longer-term In response, Roshan signed explicit contracts with outlook and a commitment to building sound busi- each community. The company paid a leasing fee to the nesses in the developing world. Ninety percent of community for the use of the tower site, and the local Roshan’s employees are Afghans, and the expatriate community was required to ensure that the instal- component is being further reduced. As CEO Karim lations were not attacked. Roshan also introduced Khoja says, “My proudest moment will be to trans- other incentives, such as bonus payments if the tower fer the company to an Afghan CEO.� remained safe for a given a number of months, and it made signi�cant investments in local infrastructure, based on the community’s wishes (in some communi- Sources: Authors, based on Herman and Dhanani 2010; remarks by CEO Karim Khoja at HBS Social Enterprise Conference, March 2010 and personal communi- ties, Roshan provided funds for a local well; in others cation with Karim Khoja, July 2010; and Roshan web site (http://www it funded solar electricity for the local mosque). .roshan.af). CREATING JOBS IN CONFLIC TAFFEC TED AREAS 295 Roshan’s facility, at a significantly lower FIGURE 7.12 Sri Lanka infrastructure accessibility index cost than the alternative of hiring an inter- national private security agency. In addition, it provided important bene�ts for the local community. Rebuilding infrastructure Armed conflict destroys or inflicts heavy damage to physical infrastructure instru- mental to providing basic services and con- necting communities and � rms to markets. Electricity and water supplies, roads, rail networks, bridges, communication systems, and sanitation systems are often destroyed or rendered unusable. Based on the limited evidence from investment climate surveys, lack of access to electricity and transport seem to be a particularly important con- straint to job creation in conflict-affected areas. In Afghanistan, for example, more than three decades of war left the country largely in ruins: in 2004, only 16 percent of households in rural Afghanistan had access to electricity, and three-quarters of the popu- lation had no access to public transportation in their communities (World Bank 2005). After two decades of violence in Sri Lanka, in 2003 only 64 percent of households in the Northern and Eastern provinces had access to electricity, compared with a national aver- Source: World Bank 2007. age of 73 percent.18 Road networks were Note: The accessibility index is calculated for every point as the sum of the population totals of surrounding cities and towns inversely weighted by the road network travel time to each town. also considerably inferior in conflict-affected provinces (those located in the north and the east of the island) (�gure 7.12). 10 selected items was 48 percent in May–July Infrastructure limitations make it more 2008; one year later, the price premium on expensive to do business and create jobs in the same products had declined to 24 percent confl ict-affected areas. In Afghanistan, for (World Bank 2010c). example, 87 percent of �rms have to operate Rebuilding much needed infrastructure with a private generator (chapter 4). in the aftermath of violent conflict faces sig- In Sri Lanka, the end of the conflict in ni�cant challenges. Financing requirements the Northern Province in 2009 provided an are great, renewed violence could destroy the opportunity to assess some of the economic rebuilt infrastructure, and the vast scale of benefits associated with improved infra- needs makes it dif�cult to prioritize. structure and access to markets, as the A9 Private, public, NGO, and international highway that connects the north with the actors have found ways to address some rest of the country was reopened. Improved of these challenges. Three complementary connectivity allowed for more fluid mobility approaches are adopted: of people and goods, resulting in the rapid convergence of prices. The average price pre- • Providing infrastructure on a limited basis, mium in Jaffna compared with Colombo on possibly in SEZs, which are geographically 296 MORE AND BETTER JOBS IN SOUTH ASIA limited and managed separately from the likely to be available only in the medium rest of the economy. SEZs could also help term, however, as they can be very com- address some of the regulatory and secu- plex and often require a stronger � nancial rity concerns associated with job creation and regulatory framework than exists in conflict-affected areas, as discussed immediately after confl ict. above. • Supporting the community-based/decen- • In the m edium te r m , t apping into tralized building of infrastructure and innovative forms of � nancing. Confl ict- service provision. An innovative experi- affected countries can use some inno- ment in progress in Afghanistan is the vative financing mechanisms—such as National Solidarity Program, which uses diaspora bonds, future-flow securitiza- a decentralized community-driven mech- tion, and performance-indexed bonds— anism coupled with international funds to alleviate � nancing constraints (Ketkar to execute local infrastructure projects and Ratha 2009). These mechanisms are (box 7.2). A similar program in Nepal BOX 7.2 Community-led infrastructure provision: Afghanistan’s National Solidarity Program Afghanistan’s National Solidarity Program (NSP), and bookkeeping. Average economic rates of return launched in 2003, is a community-led rural devel- across four sampled sectors (irrigation, power, water opment initiative focused mainly on reconstruction supply, and roads) have been estimated at 18.9 per- and rural infrastructure. The program provides cent. An early randomized evaluation shows that the block grants (averaging $32,800 per community) to program has led to the creation of functional local local governance bodies known as community devel- councils, thereby decreasing the power of traditional opment councils (CDCs) for infrastructure proj- elites. The program has led to greater engagement ects selected by bene�ciaries. Grants are disbursed of women in community life and improved the per- directly to the councils and managed through a bank ception of government figures. The impact of the account with a public bank. Participation of women program on access to services, infrastructure and in the CDCs is mandatory, although they sometimes utilities, economic activities, communities’ trust, or participate in women-only councils. Communities the likelihood of violence, however, has been mixed are also required to contribute 10 percent of the (Beath and others 2010). cost of each project, with funds in the form of cash The successful roll-out of the NSP shows that donations, work hours, or construction materials. community-driven development initiatives can be The program is implemented by Afghanistan’s Min- an important instrument amid confl ict, when gov- istry of Rural Rehabilitation and Development and ernments may be in too feeble a position to imple- � nanced by a consortium of international donors. ment programs. In addition to direct economic The NSP is the only program to have reached all bene�ts, these schemes can serve to foster cohesion provinces in Afghanistan, affecting the lives of more and develop local capacities. However, signi� cant than 17 million people. As of 2009, some 22,000 challenges remain. Can coverage be extended to the communities had successfully elected CDCs, and 9,000 communities most affected by confl ict, which more than 50,000 community projects had been par- are currently not part of the scheme? Will the NSP tially or fully � nanced (about 31,000 had been com- be sustainable without international donor funding? pleted). Most projects focus on building or restoring Can these small projects be scaled up? Will the CDCs irrigation facilities, power supply, roads, and bridges continue to represent their communities, or will they and bringing drinking water to communities. be captured by the traditional elites over time? The program is estimated to have employed some 4,000 Afghan nationals. It has also devel- Sources: Authors, based on Beath and others 2010; World Bank 2009a; World Bank Afghanistan National Solidarity Program website (http://web.worldbank. oped the skills of 600,000 local council members org/WBSITE/ EXTERNAL/COUNTRIES/SOUTHASIAEXT/0,,contentMDK:21166174 in areas such as conflict resolution, accounting, ~pagePK:146736~piPK:146830~theSitePK:223547,00.html). CREATING JOBS IN CONFLIC TAFFEC TED AREAS 297 (the Poverty Alleviation Fund) is a rural where land rights are at the center of the dis- development program focused on devel- pute. The Indian government has taken steps oping rural community infrastructure to address this constraint and source of griev- linked to communities’ priorities. In both ance, enacting, for example, the Scheduled programs, local communities identify, Tribes and Other Traditional Forest Dwellers plan, build, operate, and maintain infra- Act in 2006 (box 7.3). Although it is prob- structure. Community-based infrastruc- ably too early to assess the effects of this law, ture development and service provision the act is a very positive sign. Land issues could help alleviate some security risks also appear to be an important constraint in and help policy makers prioritize in an Afghanistan, where 68 percent of companies environment of tight fiscal constraints that tried to obtain additional land between and vast needs. This type of intervention 2005 and 2008 reported being unable to do is less relevant for large infrastructure so, according to the country’s 2008 enterprise projects with cross-regional spillovers survey. (ports, airports, highways), which are bet- Improving the regulatory environment ter addressed by higher-level government is key to encouraging private sector job administrations.19 creation in conflict situations. To attract investment, governments can focus on “soft measures,� such as reducing barriers to � rm registration and operations and clean- Strengthening the regulatory ing up at least the most blatant pockets of framework and governance corruption. In enterprise surveys, private businesses There is significant room for improve- in South Asia cite lack of effective regu- ment. Afghanistan, for example, ranks lation and governance as a major con- 160th in the world on the overall ease of straint, especially in conf lict-affected doing business and last in the world on pro- areas. Almost 40 percent of firms in the tecting investors, trading across borders, region cite corruption as a major or severe and closing a business. But some of the top constraint, the third-most-cited obstacle performers in improving their Doing Busi- to business after political instability and ness rankings in recent years are postcon- electricity. Tax administration and policy fl ict countries, such as Liberia (box 7.4) and uncertainty are also among the top con- Rwanda. Given the limited upfront costs, straints. There is some variation across simplification of business regulations is a countries in the importance of different key � rst step to attracting the private sector. elements of regulation and governance, as Other components of the regulatory frame- discussed in chapter 4. Average severity work, such as tackling pervasive corrup- appears higher on all dimensions of regu- tion, are likely to take more time to show lation and governance in confl ict-affected improvement. areas (� gure 7.11). This � nding is consis- Three broad lessons for the develop- tent with findings from conflict-affected ment of the private sector in conflict-af- countries outside the region, where politi- fected areas emerge from the success stories cal risk—mainly in the form of potential described in this section. First, the state adverse regulatory changes, which are sig- plays an important role as an enabler of pri- ni� cantly more likely than in other devel- vate sector activity by rebuilding infrastruc- oping countries—are a main concern of ture, strengthening the regulatory environ- investors (MIGA 2011). ment and the legitimacy of institutions like Regulatory issues—related to land, for the judiciary, and providing security for the instance—can be at the heart of armed con- labor force. Second, communities can play flict. In India, for example, the Maoist insur- a vital role in the success of private compa- gency is heavily concentrated in forest areas, nies, sometimes by becoming stakeholders 298 MORE AND BETTER JOBS IN SOUTH ASIA BOX 7.3 Improving the land rights framework in India The Scheduled Tribes and Other Traditional Forest collection of NTFPs. Livelihoods in forest areas are Dwellers (Recognition of Forest Rights) Act of 2006 severely constrained by a lack of infrastructure and represents a departure from past forestry policy government services. By providing for the adminis- in India. The act seeks to address forest dwellers’ trative transformation of forest villages into revenue demands for land rights by granting both individual villages, the Forest Rights Act should increase the and communal rights to people living and cultivat- supply of roads, electricity, schools, and other gov- ing land in forested areas. The Ministry of Tribal ernment services to forest-dwelling communities. Affairs, in conjunction with state governments and Because of slow and partial implementation, village councils, is required to verify occupancy and many of the bene�ts of the Forest Rights Act have yet grant titles to forest dwellers that meet the require- to be realized. According to the Ministry of Tribal ments of the law. Affairs, which periodically publishes data on the sta- Production in forest areas in India is a combina- tus of the implementation of the act, implementation tion of timber, agriculture, and collection of non- varies widely from state to state. As of September timber forest produce (NTFP). Forestry and logging 2010, Chhattisgarh had received more than 16 times accounted for about 1 percent of GDP in 2009. Col- the number of claims as Jharkhand and distrib- lection of fuelwood employed an estimated 11 mil- uted more than 11 times as many titles. Orissa had lion people, generating revenue of about $17 billion issued titles for more than half of all claims submit- in 2001. Although data on forest production are ted, whereas Gujarat had issued titles for less than limited, the Ministry of Environment and Forests 10 percent of claims. No state had accepted claims estimates that 200 million people rely on timber and or issued titles for a signi�cant number of communal NTFPs for part or all of their livelihoods. claims. Instead, the focus has been on implementing The Forest Rights Act is therefore expected to the individual titles that represent part of the rights have a signi�cant impact on the livelihoods of for- granted to forest dwellers under the act. The coming est dwelling communities. Increased tenure security years should prove critical to the success of this act should provide incentives for investments in improve- and the livelihoods of the populations dependent on ments to the land. Granting communities rights and forestry. responsibility over nonagricultural land should lead Sources: Authors, based on World Bank 2006; Government of India 2007, 2010. to more ef�cient forest management and sustainable The full text of the act is available at http://forestrights.nic.in. in private � rms. Third, in their own inter- ways around some of them. Developing a est, successful � rms take a holistic approach vibrant private sector takes a long time, to business, investing in their core economic however. In Afghanistan, for example, 10 activities (by creating jobs and develop- years after the start of the latest armed ing human resources), contributing to their confl ict, domestic � rms still produce pri- communities (by supporting education and marily for the small domestic market, few health programs), and conducting policy foreign companies (less than 2 percent dialogue with and supporting the reforms of of all firms) operate, the quality of jobs the government. Community development remains low (World Bank 2008a). The initiatives are a way for � rms to ensure local public sector is therefore called on to play ownership, enhance their reputations, and a more active role through labor market improve security and the pro�tability and programs in confl ict-affected areas during sustainability of their activities. the initial phase of the recovery, before the Although constraints to job creation in private sector is able to absorb workers and confl ict-affected areas are severe, there are generate jobs. CREATING JOBS IN CONFLIC TAFFEC TED AREAS 299 BOX 7.4 Improving the regulatory framework in a postconflict situation: Lessons from Liberia Liberia is still recovering from the ravages of 14 years private dialogue for further proposals to improve the of conflict (1989–2003), during which it experienced business environment. one of the largest economic collapses in the world After debuting in the Doing Business ranking since World War II, with GDP decreasing by more at number 170 in 2007, Liberia moved to 155th than 90 percent. More than 6 percent of Liberia’s pop- place by 2010. The number of days needed to open ulation was killed, an estimated 500,000–1 million a business was reduced from 68 in 2008 to 20 in people were internally displaced (IDMC 2010), and 2010. Obtaining construction permits was made an estimated 70 percent of schools and government easier by abolishing the requirement to obtain a tax buildings were destroyed. Apart from privately owned waiver certi�cate before submitting documents. As a generators, there was no electricity in most of Liberia result, the number of days needed to obtain a permit from 1991 to 2006 (Radelet 2007). dropped from 398 in 2008 to 77 in 2010. The trade After the restoration of peace, in 2003, and the process was expedited by creating a one-stop shop election of Ellen Johnson Sirleaf, in 2005, Liberia bringing together various ministries and agencies pursued a private sector–driven growth strategy and streamlining the inspection regime. In 2009, (World Bank 2008b). A cabinet-level working group Liberia also committed to the Extractive Industry was created to think about the nature and sequenc- Transparency Initiative, under which it must comply ing of the reforms needed to attract the private sec- with the global standard for transparent manage- tor. As a � rst step, the committee identi� ed areas ment of revenues from its extractive industries (iron, where “quick wins� could be obtained, in the form ore, diamonds, gold, timber, and rubber). of simpli�cations to the business registration process These reforms appear to be paying off: net inflows and the establishment of a one-stop-shop to advise of foreign direct investment increased from $21 mil- potential entrepreneurs of opportunities and pro- lion in 2002 to $83 million in 2005 and $144 mil- vide them with technical support. As a second step, lion in 2008. GDP per capita growth was 9.4 percent the committee focused on longer-term reforms that in 2007 and 7.1 percent in 2008, despite the global required legislative changes, such as the investment � nancial crisis. and revenue legal codes. In 2007, the Liberia Better Business Forum was established to facilitate public- Sources: Radelet 2007; Doing Business Indicators; World Development Indicators. Education service delivery Skills provision in conflict situations in conflict situations requires innovative delivery mechanisms. Security concerns and the lack of physical Confl ict usually forces people to miss out (schools) and human (teachers) infrastructure on years of education and training. In are three important challenges. The dif�culty South Asia, conflict-affected areas start of providing education and training in con- from a very low skill base, with illiteracy flict-affected areas is compounded by exten- rampant among the workforce. As dis- sive migration and displacement. cussed in chapter 5, improving educa- Although the public sector is likely to take tional attainment and skills of the future the lead, NGOs, international aid organiza- workforce is fundamental if workers are tions, private companies, and community ini- to access more and better jobs. In confl ict tiatives can also be active in providing trained areas, the education system can also play personnel or funds to rebuild, building infra- a critical role in addressing some of the structure and improving security and access. societal divisions associated with violence The diaspora could play a vital role, as its (World Bank 2011). members have the knowledge, ideas, and 300 MORE AND BETTER JOBS IN SOUTH ASIA BOX 7.5 Improving schooling despite armed conflict through community schools in Nepal The government of Nepal had managed schools in community management. A rigorous random- the country since 1971—with poor educational ized control trial compared a set of 40 treat- results on average and wide inequalities across wealth ment schools that had been encouraged to apply categories. In 2001, the nationwide literacy rate was for a transfer of management with 40 control 54 percent, and gross enrollment in primary schools schools, which were not given any such encour- ranged from 22 percent among the poorest � fth of agement. Communities whose schools came under households to 90 percent among the top �fth. community management as a result of this promo- In 2001, the government amended the Educa- tion scheme were found to have a smaller share of tion Act, renaming all government-funded schools out-of-school children, particularly among disad- as community schools. Communities were empow- vantaged groups. These schools also had better ered to take over the management of public schools grade progression rates and enhanced community by applying formally to the government and putting participation and parental involvement. There together a school management committee consisting were no signi� cant differences in teacher absen- of parents and influential local citizens. The school teeism or student learning levels within the time- management transfer option was available to all com- frame of the evaluation. munities, and the transfer process was voluntary. The experience in Nepal provides an example of The process of transferring the management of an alternative service delivery mechanism that could schools to communities began in 2002. It accelerated be helpful in times of armed confl ict. Community after 2003, when the World Bank provided � nan- ownership (and the corresponding autonomy from cial and technical support through the Community the government) means that schools, students, teach- School Support Program (CSSP). The CSSP had four ers, and school management are in a better position main components: incentive grants of $1,500 in the to avoid altercations with the groups �ghting in the year of transfer and performance grants for increas- war. In fact, progress in school enrollment in con- ing enrollment and promotion rates; scholarships for fl ict-affected areas, where community schools where children who were disadvantaged or had never been most prevalent, was more rapid than in the rest of enrolled in school; training programs for teachers, the country. As a result, high-conflict areas had community members, and supporting organizations; higher primary school enrollment rates than low- and monitoring and evaluation of strategies and confl ict areas by 2004. The increase in enrollment interventions in recently transferred schools. rates was particularly pronounced for girls, among As of July 2010, more than 10,000 of Nepal’s whom enrollment rates increased 27 percentage 25,000 public schools had been transferred to points in high-confl ict areas (box �gure 7.5.1). BOX FIGURE 7.5.1 Primary school enrollment rate in Nepal among children 6–10 years old, 1996–2004 a. All students b. Girls 100 100 90 90 80 80 70 70 60 60 percent percent 50 50 40 40 30 30 20 20 10 10 0 0 96 98 00 02 04 96 98 00 02 04 19 19 20 20 20 19 19 20 20 20 low-conflict areas high-conflict areas (continues next page) CREATING JOBS IN CONFLIC TAFFEC TED AREAS 301 BOX 7.5 (continued) Can Nepal’s experiment be generalized to other committees; it was not clear whether their function- areas? Nepal’s government decided to transfer ing changed as a result of this initiative. This his- schools to community management not on the basis tory of community management may be absent in of objective data but because such management other settings, making Nepal’s experience dif� cult was already prevalent in other places, such as for- to replicate. estry conservation and small irrigation areas. Many schools already had functioning school management Sources: Chaudhury and Parajuli 2009; World Bank 2001, 2010b. market experience obtained while working and cut off from population centers. The abroad as well as local knowledge—and pos- experience of Nepal’s community schools sibly still relevant socioeconomic ties—that can provide some lessons for the delivery of may allow them to be successful. educational services amidst armed confl ict Educational services such as “school in a (box 7.5). box� kits and mobile educational program- Distance education through television, ming could be useful in camps and settle- video, and particularly radio is another ments for internally displaced people. In the way of working around security problems. wake of the 2010 earthquake in Haiti, for According to the British Broadcasting Cor- example, schooling was provided inside buses poration (BBC), almost half of the Afghan that moved from community to community population listens to its radio dramas every or camp to camp.20 In areas with signi�cant week, suggesting great potential to use this numbers of internally displaced people, tem- medium as an educational tool, raise aware- porary suspension of any documentation ness, and promote social cohesion. requirements could ease some of the con- straints they face when enrolling in school and allow them to register for school without Labor market policies having to return to their areas of origin.21 and programs Other alternatives, such as community Because reforms to improve the business delivery of educational services, could be environment take time to bear fruit, labor helpful in addressing security issues. Com- market institutions, policies, and programs munity- and home-based education initia- such as those discussed in chapter 6 have tives are two ways to get around security a particularly important role to play at the constraints, especially during the transitional early postconflict stage. DDR programs; stage. Community-level schemes could be public works; training programs; and poli- useful by having women in a community, for cies to support self-employment and the example, organize themselves to accompany informal sector play both an activation (pro- children and women to educational centers. motion) and a safety net (protection) role. In Afghanistan, many of these programs have As discussed below, however, the design of been supported by NGOs, both before and these programs is more complex in confl ict after the ousting of the Taliban in 2001. The situations. Ministry of Education has also participated as an of�cial partner (Sigsgaard 2009). Disarmament, demobilization, Community- and home-based education and reintegration programs initiatives can also facilitate the expansion of educational services into places affected The DDR of excombatants has become a key by conflict, which are often dif�cult to reach element of peace building in conflict-affected 302 MORE AND BETTER JOBS IN SOUTH ASIA countries. DDR programs usually target expensive, with costs per demobilized per- excombatants in three broad phases: disar- son of $2,278 in Afghanistan and $2,500 in mament (collecting and disposing of weap- Nepal (Carames and Sanz 2009). The disar- ons); demobilization (disbanding military mament and demobilization stages usually structures); and reintegration (the process represent a small share of the budget (less of facilitating the return of former combat- than 10 percent); most of the spending goes ants to civilian life, the armed forces, or the toward reinsertion and reintegration. In this police). The last phase is essentially an acti- phase, excombatants who participate in the vation program. Reintegration packages— reintegration program often receive eco- including vocational training, formal edu- nomic compensation to substitute for some cation, physical and mental health services, of the income they would have earned by incentives for entrepreneurship, employment participating in violent activities. Afghani- services, public works, and cash transfers— stan paid excombatants a total of $990, on are at the heart of this phase (UN 2010). average, to participate ($180–$480 a month DDR programs are underway in Afghani- over a period of two to four months)— stan and Nepal. Afghanistan’s $141 million the second-highest payment in the world demobilization program, which ran from after Colombia ($2,750). Nepal pays $46 2003 to 2005, demobilized 63,000 profes- a month (the duration of payments is not sional soldiers and 7,500 child soldiers. clear). DDRs may be unaffordable for many In the reintegration phase, 46 percent countries, especially when the conflict is of excombatants attended vocational or nationwide. small business training, and 43 percent Reintegration is the most dif�cult part of opted to go into agriculture. The rein- the DDR process. Providing relevant training tegration phase was to have been com- and helping people �nd economic opportuni- pleted in 2006 but was extended an addi- ties during or after a confl ict is always dif- tional two years after it was revealed that �cult. The task is even harder in the case of 56 percent of all demobilized excombatants excombatants. Reintegration is often about were earning less than $1 a day (Carames and more than just � nding new income sources, Sanz 2009). A new reintegration program as excombatants have also lost physical secu- directed at the entire affected community rity, political influence, social networks, and (not only excombatants), the Afghanistan prestige with the end of the conflict (Willibald Peace and Reintegration Program, is now in 2006). Failure to address these losses may place. No formal evaluation exists, but by jeopardize the DDR process and peace more June 2011, 1,700 former �ghters had publi- generally (box 7.6). cally joined the program (ISAF 2011). The extensive use of cash transfers in the In Nepal, nearly 20,000 members of the reintegration process illustrates some of the People’s Liberation Army have been demo- risks of such interventions. The effectiveness bilized, at a cost of about $50 million. 22 of the cash transfers is unclear a priori. On the These excombatants were to have been one hand, to the extent that � nancing is the gradually reintegrated into the national binding constraint, transfers could encourage armed forces, but political disagreements compliance with disarmament commitments have prevented this phase from being and help restore livelihoods. On the other completed.23 hand, transfers, especially in cash, could DDR programs can be costly. I n a exacerbate the illicit arms trade and generate global survey of programs in 19 countries, resentment in the community. The effective- researchers found that the average cost per ness of transfer programs is therefore likely demobilized person was $1,434 —about to depend on economic opportunities and the 3.5 times per capita income in these coun- details of the implementation of the program, tries (Carames and Sanz 2008). In Afghani- such as payment location, eligibility criteria, stan and Nepal, programs have been more and targeting mechanisms. CREATING JOBS IN CONFLIC TAFFEC TED AREAS 303 BOX 7.6 Lessons from efforts to reintegrate Ugandan youth In Uganda, a country affected by more than two Some of the lessons drawn by the team could be decades of civil confl ict, a group of researchers put helpful in the design of other DDR programs in other together a program in the northern part of the coun- parts of the world. Three � ndings deserve attention: try aimed at reintegrating youth. The program—the • Catching up on missed education is a pressing Survey of War Affected Youth (SWAY)—included need, requiring a strong focus on secondary school an analysis of the needs and dif�culties experienced and adult literacy. There is a need to go beyond the by youth returning from the Lord Resistance Army usual focus on primary education. (LRA), one of the main rebel groups in the country. • In part as a consequence of lost education and It surveyed more than 1,000 households and 741 work experience, excombatants were less than half young men in 2005 and 2006 and 619 young women as likely as noncombatant peers to be engaged in in 2007. a skilled trade or a business. Their earnings were The program focused on youth who lived in the about a third those of youth who had not served areas of interest before the escalation of the violence in the LRA. Cash was one of the most crucial in 1996 and who had recently returned after hav- inputs needed for new income-generating activi- ing been part of the LRA. As the LRA selected and ties; even the modest amounts of cash associated abducted members randomly, outcomes for these with the reinsertion package helped excombatants youth could be compared with those of youth who regenerate livelihoods. Interventions such as were not selected or abducted to serve in the LRA. facilitation or business skills training could have The idea was to use the results of the survey to important impacts on the effectiveness of these inform policy makers who were about to design a funds. full DDR program for excombatants. • Reinsertion packages for excombatants should Assistance to youth in Northern Uganda took two be introduced in tandem with larger programs of main forms: reinsertion assistance and longer-term support for all youth. SWAY researchers identi�ed reintegration and development services. Reinsertion increased resentment and stigmatization on the assistance included interim counseling; vocational part of people not receiving these bene�ts, making training; basic health care; and a reinsertion package reintegration more dif�cult. Where cash disburse- that comprised household items, agricultural tools, ments are paid, it is crucial that they be designed seeds, and an unconditional cash payment. Cash pay- to minimize the risks of stigmatization, resentment, ments were provided not only to returnees but also to misuse, and theft. other families living in affected areas. Interventions included vocational training, cash, and assistance in starting small enterprises and psychosocial care. Source: Annan and others 2007. Despite the prevalence of DDR programs in evaluations of DDR programs, allowing for postconflict environments, the individual and heterogeneous effects across individuals. country-speci�c factors that lead to successful International experience with DDR reintegration are still not well understood. In does suggest some lessons for improving their study of the highly regarded program the chances of successful reintegration or in Sierra Leone, Humphreys and Weinstein activation: (2007) � nd that it was of limited effective- ness in reintegrating more educated people, • Because DDR is a long process, a strong higher-ranking members of the rebel forces, commitment to peace from all parties and and people who had been part of military suf�cient � nancing—often from interna- units that had been most abusive of civil- tional donors—is crucial.24 ians. The fact that the effects of the pro- • A strong training component seems to gram varied depending on the characteris- be critical. Such training is especially tics of the individuals calls for more detailed relevant in a country like Nepal, where 304 MORE AND BETTER JOBS IN SOUTH ASIA BOX 7.7 Training and employing displaced people: The case of Asocolflores in Colombia Colombia has suffered from violence for more than and packing fl owers. The course was followed by two decades. The primary sources of violence are a four-month work period on flower farms, after confl icts among the government; right-wing para- which participants found permanent employment in military groups, such as the Revolutionary Armed the fast-growing flower industry. Forces of Colombia (FARC); and left-leaning gue- The program was funded by Asocolflores and rillas. By 2008, 2.65–4.36 million people had been flower farmers; it also received funds and techni- displaced by the confl ict, the second-largest popula- cal support from the U.S. Agency for International tion of displaced people in the world (after Sudan). A Development (USAID) and the Pan-American Devel- disproportionate number of the displaced are Afro- opment Fund (PADF). Participants were provided Colombians or members of indigenous tribal groups, with a stipend, lodging, and psychological support who have low levels of literacy and numeracy. to help them recover from displacement. By 2008, Colombia is also the second-largest exporter of with an investment of $1.8 million, the school had cut flowers in the world. In 2006, 70 percent of all benefited more than 1,631 families. A key lesson roses and 97 percent of the carnations sold in the from the Colombian experience is that the private United States came from Colombia. sector, supported by public funds, can be a useful In 2003, the Colombian Association of Flower source of training and jobs for displaced popula- Exporters (Asocolflores) established a school of flo- tions, especially in fast-growing industries that riculture to enable people displaced by the confl ict employ unskilled labor. to find work. It developed a nine-month training course focused on cultivating, harvesting, grading, Sources: Caycedo and Mendoza 2008; IDMC 2009; PADF 2008; USAID 2006. the military is already large and there are programs have many features that make few prospects for absorbing all demobi- them well-suited for confl ict situations:27 lized excombatants into it or the police forces. • The workforce in conflict and postconflict • Public employment is likely to play an areas is more likely to be in rural areas, important role for excombatants, who where agriculture is the primary source often lack employable skills and are cut of income. Most workfare programs tar- off from social networks. get rural areas and involve the building • In certain cases, the private sector can be a or rehabilitating of community infrastruc- useful partner, as it has been in Colombia, ture, which is likely to increase the produc- where public-private partnerships have tivity of the agricultural sector and rural created more than 1,800 jobs for internally nonfarm activities. displaced people (box 7.7). This experi- • People in conflict areas have lower lev- ence can be extended to excombatants.25 els of education than people in noncon- fl ict areas. Workfare programs are ideal for such people, because they provide work for unskilled workers (some pro- Public work programs grams also provide additional vocational Public works programs are often imple- training). mented in nonconflict situations (chapter 6).26 • People in confl ict and postconfl ict areas In an analysis of 102 communities in coun- are more likely to work in informal tries affected by confl ict in 2005, research- jobs, which provide little or no social ers found that 44 percent of conflict-af- protection. By instituting a workfare pro- fected communities had at least one public gram, governments can provide informal works project (World Bank 2010a). These unemployment insurance. CREATING JOBS IN CONFLIC TAFFEC TED AREAS 305 • Confl ict and postconfl ict areas are char- priate wage can be difficult if private acterized by mobile populations. Work- labor markets have not functioned for fare programs, along with a traditional some time. In such situations, practitio- resettlement campaign, can help ease the ners must be creative. In Sierra Leone, transition of internally displaced people to for example, preconflict household survey their places of origin by providing them data, along with monthly price data on with temporary employment and guaran- staple foods and � sh, were used to come teed income for a period of time. up with a wage estimate. • Some communities may be so destroyed Implementing a public work program in that the scope of public works may be conflict and postconflict situations poses broader than usual. In such communities, certain challenges, however: it could be desirable to include the rebuild- • Security is a major concern. In most cases, ing of private homes as part of the work- workfare programs can proceed only fare program, as was done in Sri Lanka. when there is suf�cient order to provide a • Excombatants may not be able to work secure work environment and transporta- with other members of society, given the tion to work sites, especially for women. animosities generated during the confl ict. Land mines and other conflict-induced Fearing reprisals, many excombatants in hazards must be removed before a work- Sierra Leone did not wish to return to fare program can begin. their villages, where a workfare program • Most workfare programs use the wage had been established. The solution was to rate as a targeting device, setting it at a allow such people to work in the workfare level that is high enough to induce peo- program set up in the capital. ple to participate in the program and low enough to attract only the very poor. In Sri Lanka’s public works program over- confl ict-affected areas, setting an appro- came some of these problems (box 7.8). BOX 7.8 Implementing public works in a postconflict environment: Sri Lanka’s Northern Province Emergency Recovery Project The decades-long conflict between the Liberation Projects included debris removal and land clearance, Tigers of Tamil Eelam (LTTE) and the Sri Lanka well restoration and construction, small-scale irriga- government ended in May 2009. The conflict had led tion canal rehabilitation, small road rehabilitation to the destruction of social and economic infrastruc- and repairs, classroom rehabilitation, health clinic ture and the displacement of hundreds of thousands of repairs and rehabilitation, and repair and construc- people from the �ve northern districts of Jaffna, Kili- tion of community drinking water facilities. nochchi, Mannar, Mullaitivu, and Vavuniya. As part A maximum of three adults (including the head of its resettlement plan, the government established a of household) from each returning household were cash-for-work program to repair community infras- eligible to participate in the program, with only one tructure and provide returning internally displaced member allowed to participate on a given day. This populations with short-term income support. feature of the program was included to encourage Community leaders and project officers of the participants to seek other work. �ve districts identi� ed projects. Community mem- In order to encourage only the very poor to self- bers mobilized returning households to form com- select into the program, the wage rate was set at 15 munity cash-for-work committees, which identi�ed percent below the market rate for unskilled labor for projects of immediate need for the communities. the district. The market wage was established by a (continues next page) 306 MORE AND BETTER JOBS IN SOUTH ASIA BOX 7.8 Implementing public works in a postconflict environment: Sri Lanka’s Northern Province Emergency Recovery Project (continued) (district) wage-� xing committee that included repre- with any other job if their homes remained in disre- sentatives of the government (who consulted the pri- pair. Repairing one’s own home also increased the vate sector), NGOs, and donors. sense of revival and con�dence in the future. Participants could work for a total of 50 days. By 2010, about 275,000 internally displaced peo- Given that the vast majority of internally displaced ple had returned to their villages. Participants in the people returned to homesteads that had been seve- program worked more than 380,000 person-days, rally damaged during the confl ict, participants were repairing 1,500 kilometers of roads and 145 wells and allowed to work on their own homes during the � rst cleaning more than 900 hectares of public spaces. 10 days. The rationale was that they would not be able to work well with the workfare program or Source: Government of Sri Lanka 2010. Training by the common reallocation packages offered in postconfl ict situations. Sometimes, how- The considerably lower skill base of the ever, people do not have the right opportuni- workforce in confl ict-affected areas means ties at home or fear returning. Following the that reskilling—with an emphasis on pro- ceasefire in Sri Lanka, for example, unem- moting literacy, noncognitive skills, and ployment rates increased signi�cantly for the practical skills—is key for postconflict most educated (especially people who had recovery. Given the large informal sectors in recently moved back) in the Northern and these areas, the discussion of training pro- Eastern provinces. With unemployment at grams in chapter 6 is relevant here. 29 percent, this group became the most likely Public work and DDR programs often to be unemployed in the conflict-affected include training components, which attempt areas—a change from confl ict times (World to cater to the diverse training needs of tar- Bank 2007). get groups. In the Afghanistan Peace and Reintegration Program, for instance, training includes not only literacy skills but also prac- Support for self-employment tical training in farming, farm equipment and microenterprise development maintenance, nursery establishment and Other interventions seek to support self- management, watershed management, road employment and microenterprise devel- building and maintenance, and basic electri- opment to help informal sector work- cal work and plumbing. ers improve their potential for generating Initially, the government takes the lead, income. Particularly relevant are microfi- often in partnership with international aid nance, the strengthening of business sup- organizations and NGOs. Over time, the ply chains, and business training in foster- private sector can play an important role, ing entrepreneurship in the informal sector particularly as it is more likely to provide (World Bank 2011). Much of the discussion job-related training and an employment path- of interventions that promote self-employ- way for trained workers (see box 7.7). ment in chapter 6 is also relevant for conflict- The dif�culty of providing education and affected areas. It is important to recognize, training in conflict-affected areas is com- however, that the presence of a large mobile pounded by extensive migration and displace- population, security risks, malfunctioning ment. The usual approach is to wait for peo- product and input markets, and the erosion ple to come back to their homes, as evidenced of trust in the community make microcredit CREATING JOBS IN CONFLIC TAFFEC TED AREAS 307 and related operations particularly challeng- especially when the conflict is nationwide. In ing in conflict-affected areas. Afghanistan, for example, of�cial develop- One example of a program that tries to ment assistance represented 45.8 percent of tackle these issues in a postconflict environ- gross national income in 2008 (World Bank ment is the Nepal Poverty Alleviation Fund 2011b). This dependency on foreign aid may (PAF), which seeks to promote entrepreneur- pose risks to postconflict recovery by crowd- ship through its income-generating compo- ing out domestic resource mobilization, nent. PAF provides loans, complemented with making unsustainable expenditure commit- funds for infrastructure needed to better link ments, and raising the exchange rate (Boyce products to markets.28 2007). Moreover, international aid declines Initiatives directed at bringing producers significantly after the initial postconflict and markets together—such as initiatives in period (Chen, Loayza, and Reynal-Querol Kosovo and Rwanda in the tourism, coffee, 2007), before local revenue sources have and dairy industries—could also bring early recovered sufficiently. Some postconflict results (World Bank 2011). As with other areas also depend heavily on international interventions, programs directed at sup- organizations and NGOs for well-paid jobs, porting self-employment need to be further which could further distort labor markets.30 evaluated before more speci�c lessons can be Despite these caveats, international orga- drawn on what works, when, and where.29 nizations have an important role to play, especially in the short term, by ensuring that local � rms have the opportunity to partici- Fiscal and capacity constraints pate in aid-� nanced tenders (in the logistics to public employment programs and construction sectors, for example). The Most conflict-affected areas lack govern- Afghan First initiative by the North Atlantic ments with sufficient fiscal resources or Treaty Organization is a promising example capacity to administer large bureaucracies of how to build capacity and increase the par- or public employment programs. Both of ticipation of local � rms in confl ict-affected these constraints are likely to be particularly areas.31 severe in cases of nationwide confl ict. Conflict-affected countries tend to have FIGURE 7.13 Central government debt as a percentage of GDP high levels of indebtedness (�gure 7.13), low in conflict and nonconflict areas of South Asia and the world, tax revenues, and high dependency on for- 1990s and 2000s eign aid. Tax revenues are low across South Asia, especially in Afghanistan (6 percent 100 of GDP) and Pakistan (10 percent of GDP) (World Bank 2011b). Moreover, the military 80 takes up a large share of central government expenditures in South Asia, estimated at 10 60 percent percent in Afghanistan and Bangladesh, 15 percent in India and Sri Lanka, and 18 per- 40 cent in Pakistan (the average for conflict- 20 affected countries worldwide is 3.2 percent) (World Bank 2011b). 0 These � scal constraints and the consid- 1990s 2000s erable capacity constraints in confl ict envi- no conflict South Asia conflict South Asia ronments suggest an important role for no conflict worldwide conflict worldwide international organizations and foreign gov- Sources: Authors, based on data from World Bank 2011b and UCDP/PRIO conflict database. ernments in funding and capacity building. Note: Differences between conflict and nonconflict are signi�cant at the 5 percent level. A country is considered to be in conflict in a given year if it experienced 25 or more battle deaths from confron- A risk, however, is that postconflict countries tations between the military forces of two parties, at least one of which is the government of a state. become highly dependent on foreign aid, Results do not change if India is excluded from the analysis. 308 MORE AND BETTER JOBS IN SOUTH ASIA A jobs transition path key provider of jobs to a greater role for the in conflict zones private sector, from low-skilled agricultural jobs to higher-productivity nonfarm jobs, Job-related interventions need to be and from targeted programs to broad-based sequenced depending on the stage and job creation. Table 7.3 summarizes these the type of the conflict. The constraints to transitions. job creation vary depending on whether the conflict is ongoing or over, nation- wide or localized. These constraints will Who creates jobs? determine the design of all labor market programs discussed in this chapter, from Given the constraints on private business, public works and microcredit programs the public sector is often the main creator to changes in regulation and the provision of jobs initially after confl ict. Direct pub- of basic services, including education and lic employment should be only a transi- infrastructure. tion strategy, however. Fiscal conditions, Security is a key determinant of the sus- efficiency considerations, and unintended tainability and desirability of investment adverse effects of public employment (such in conflict-affected areas. Insecurity is a as corruption and politicization of jobs) all continuum: even postconflict situations expe- caution against maintaining the state as a rience a certain degree of violence. A mini- long-term provider of jobs. Given these con- mum level of security is needed before certain cerns, policy makers need to develop a strat- investments can be made in conflict-affected egy for transitioning to a greater role for the areas; achieving security is likely to drive the private sector. sequencing of interventions.32 There is little private investment in conflict-affected areas, What jobs to create? and opportunities for wage employment are restricted. Large infrastructure projects or In the initial stages, it is natural that jobs be other �xed investments may not be feasible in created in the agricultural sector, where the areas where violence and insecurity are still bulk of the workforce is already involved, rampant; innovative interventions that try as well as in other unskilled sectors. Sectors to maintain human capital may represent a such as construction, which have the poten- better investment. tial to employ local unskilled labor and help Once a minimum level of security is in postconflict rebuilding, should be encour- achieved, job creation efforts in conflict aged, through, for example, public works zones need to undergo three major types and training in needed skills. The Pakistan of transitions: from the public sector as the postcrisis needs assessment identified TABLE 7.3 Labor market transitions in postconflict zones Job transition path Key question Constraints/concerns From To Who creates jobs? Security, infrastructure, Public sector, international Private sector, public-private poor regulations, fiscal organizations partnerships constraints What jobs to create? Security, Agricultural, unskilled, Nonfarm, skilled, regular rural, unskilled work force casual wage or salaried For whom to create jobs? Vulnerable populations Targeted programs Broad-based job creation (excombatants, youth at risk, war victims) Source: Authors. CREATING JOBS IN CONFLIC TAFFEC TED AREAS 309 construction as the sector with the greatest with special needs. The �rst is excombatants, immediate potential for job generation in the who need to be integrated into the civilian or confl ict-affected areas. Yet key sectors are military workforce and be given incentives to often not prioritized, especially in view of refrain from violence. The second is at-risk the need for quick gains. In the worldwide youth and war victims. This group includes community surveys conducted as part of the families who have lost some of their members; Moving out of Poverty series (World Bank people with physical or mental disabilities; 2010a), for example, only 16 percent of con- and households that are extremely vulnerable flict-affected communities reported receiving because they have no steady streams of assistance with agriculture and only 7 per- income. The third is people displaced by the cent bene�ted from a �nance/credit project. conflict, who may wish to return to their The transition plan should seek to move homes but need to be able to �nd jobs and workers toward higher-productivity jobs, feel secure there. which will pay more and result in greater wel- Targeting is dif�cult in confl ict-affected fare improvements. Initial steps toward doing areas. Identifying and locating bene�ciaries so can include making the agricultural sector often proves dif�cult. One way to overcome more productive, by, for example, providing these problems can be to target programs better inputs, such as improved seeds, fertil- for speci�c areas rather than speci�c people, izer, and storage facilities; creating function- thereby allowing self-selection to determine ing local markets; and rebuilding transport who receives bene�ts. This is the approach and communication networks to access more followed in Afghanistan, for example, under distant markets. Longer-term transition plans the Peace and Reintegration Program (Islamic could include retraining the workforce to par- Republic of Afghanistan 2010). ticipate in manufacturing or services, as well Over the longer term, broad-based job as facilitating the growth of �rms in the non- creation is needed in these areas. To work farm sector. These measures are likely to create toward this goal, government can operate more jobs that pay regular wages or salaries. targeted programs for a limited time while The private sector can be involved, through encouraging the growth of other activities public-private partnerships, for example. and firms to provide a broader set of jobs. Table 7.4 summarizes some policies that can be implemented at each step of this transition For whom to create jobs? process. In the early stages of the conflict, policy Because of the violent nature of confl ict, interventions should be directed at managing the impacts of policy interventions in con- the risks of renewed confl ict and addressing flict or postconflict environments have rarely the lack of economic opportunities for the been rigorously analyzed. As a result, policy vulnerable. Without credible economic alter- makers cannot draw on a large body of evi- natives to people involved in the fighting, dence.34 Some evidence on World Bank proj- an end to the violence is unlikely. Job cre- ects suggests that urban development projects ation can play an important role in helping are most likely to succeed and that private address the concerns and grievances that led sector development projects need a longer to the conflict, which is fundamental in but- time horizon for success (Chauvet, Collier, tressing con�dence in the peace process. 33 and Duponchel 2010). The quality of super- Nepal—where persisting disagreements on vision is also an important determinant of how to integrate excombatants have left the project success, but evidence remains limited. country politically unstable almost �ve years Piloting some interventions and progressively after the peace agreement—illustrates the expanding programs could help shed light importance of addressing such concerns. on program effectiveness. In many conflicts, Job creation efforts in conflict-affected areas governments do not regain control of all have to deal with three types of populations confl ict areas at the same time. Starting the 310 MORE AND BETTER JOBS IN SOUTH ASIA TABLE 7.4 Policy interventions in the initial postconflict stages Goal Possible interventions Create some jobs quickly. • Set up labor-intensive public works, especially in rural areas. • Create employment agencies where needed. • Respond to the needs of the displaced and internally mobile by creating job opportunities outside areas directly affected by conflict. • Distribute tools and inputs for farming and construction (starting with rehabilitation of people’s own houses). Promote agriculture and other informal • Rehabilitate complementary infrastructure needed for production and getting products to sector activities, such as jobs in market (this intervention should be combined with appropriate business services). construction and self-employment. • Provide credit, funds, and training to foster entrepreneurship (microcredit, cash transfers to start businesses, business development training). Create the base for better jobs by • Focus on progressively getting children back to school and providing practical training and improving workers’ skills. literacy courses for youth and adults using existing networks, community initiatives, and NGOs when available. • Address security concerns related to accessing schools or training, especially for girls (through mobile schools, community involvement in security). • Focus on the needs of displaced people and internal migrants, allowing for reskilling outside areas directly affected by conflict. Improve the investment climate. • Create “safe� zones that provide better regulation, services, and security. • Tap into innovative source of financing, such as diaspora bonds, as well as innovative forms of guarantees for foreign direct investment. • Establish investment information centers that match interested foreign investors with potential local partners. • Create links between firms and local communities to ensure the sustainability and profitability of enterprises. • Improve the overall business environment by simplifying regulations and facilitating processes for entry and exit. Source: Authors. piloting of labor market programs and poli- In short, the employment challenge in con- cies in areas where control has been restored flict-affected areas is larger than just creating and then scaling up to the rest of the conflict- jobs. It concerns the fundamental problem of affected areas can improve policy results. The establishing functioning economies, creating National Solidarity Program in Afghanistan occupations different from those that existed is a case in point: the program was piloted in in the war economy, promoting growth, gen- locations where violence was less pronounced erating government revenue, and addressing before being progressively extended to the some of the underlying economic and social rest of the country. In Sri Lanka, many of grievances that may have led to the armed the labor market and development programs conflict in the �rst place and could reinitiate the government is planning for the North- violence if not addressed. There are examples ern Province were �rst piloted in the Eastern of promising interventions, but achieving Province. these goals is likely to take a long time. CREATING JOBS IN CONFLIC TAFFEC TED AREAS 311 Annex 7A De�nitions of high-conflict and low-conflict regions in selected South Asian countries This annex describes the classification of Nepal high- and low-confl ict regions in Afghani- Classi�cation is at the district level. High- stan, India, Nepal, and Sri Lanka. conflict districts are those in which the number of confl ict deaths per 1,000 people Afghanistan in 2004 was greater than the median; low- conflict districts are those in which the num- Classi�cation is at the province level. High- ber of confl ict deaths per 1,000 people was conflict provinces are defined as provinces less than the median level of 0.67. The time in which the number of deaths per 1,000 period considered was 1996–2004 (two people caused by terrorist incidents in 2007 years before the end of the confl ict). By this was greater than 0.1. By this de� nition, 15 de� nition, half of Nepal’s 36 districts are provinces are in high-conflict areas (Baghlan, classified as high-conflict areas (Achham, Ghazni, Farah, Helmand, Kandahar, Khost, Arghakhanchi, Bajura, Banke, Bardiya, Kunarha, Nimroz, Paktika, Paktya, Panjsher, Bhojpur, Dadeldhura, Dailekh, Dang, Takhar, Urozgan, Wardak, and Zabul) areas. Dhading, Dolakha, Dolpa, Doti, Gorkha, Data on terrorism deaths were obtained Humla, Jajarkot, Jumla, Kailali, Kalikot, from the National Counterterrorism Center Kanchanpur, Kavrepalanchok, Lamjung, (2008). Mugu, Myagdi, Nuwakot, Okhaldhunga, R a mech hap, Rolpa , Ru ku m , S a lya n , India Sankhuwasabha, Sindhuli, Sindhupalchok, Solukhumbu, Surkhet, and Taplejung). Classification is at the state level. Low- Data on confl ict deaths were obtained from conflict states are states that recorded no the annual Human Rights Yearbooks pub- conflict-related fatalities in 2000 or 2007. lished by the Informal Sector Service Centre High-conflict states are states that transitioned (INSEC), an NGO based in Kathmandu. to a state of conflict during this period—that is, states that recorded no casualties in 2000 but did record confl ict-related casualties in Sri Lanka 2007. This means that states in conflict in 2000 are excluded from this analysis. By this Classi� cation is at the province level. The de�nition, 10 states are in high-conflict areas Northern Province was not included in the (Bihar, Chhattisgarh, Jharkhand, Karnataka, analysis because of lack of data. The East- Maharashtra, Meghalaya, Nagaland, Orissa, ern Province is classi�ed as high conflict; all Uttar Pradesh, and West Bengal). Confl ict- other provinces are classi�ed as low conflict. related fatalities are based on data from the The analysis for Sri Lanka should be inter- RAND-MIPT Terrorism Database, collected preted with caution, however, as the data by the National Memorial Institute for the reflect almost two decades of conflict. In Prevention of Terrorism (MIPT), in collabo- addition, the analysis for Sri Lanka differs ration with the National Counterterrorism from that of other countries in that conflict- Center (NCTC). Data are available for the affected regions are moving from conflict to period January 1998–April 2007. peace. 312 MORE AND BETTER JOBS IN SOUTH ASIA Annex 7B Labor market characteristics and educational attainment in high-conflict and low-conflict areas of selected South Asian countries TABLE 7B.1 Labor market characteristics in high-conflict and low-conflict areas of Afghanistan, 2008 2008 Low- conflict High-conflict Afghanistan areas areas Difference Difference (rural) Percentage of working age population employed 55.67 68.37 12.705*** 9.192*** Male 76.09 81.08 4.984*** Female 35.49 54.76 19.273*** Unemployment rate 1.96 0.66 –1.305*** –1.008*** Male 2.82 0.83 –1.989*** Female 1.12 0.47 –0.644*** Labor force participation rate 57.63 69.03 11.400*** 8.184*** Male 78.91 81.90 2.995*** Female 36.60 55.23 18.629*** Percentage of workers in urban areas 21.77 5.29 –16.481*** Percentage of workers by sector of employment Agriculture 51.46 70.50 19.038*** 9.698*** Manufacturing 6.62 1.43 –5.191*** –4.742*** Construction 9.43 4.90 –4.532*** –4.469*** Commerce 12.65 8.45 –4.195*** –0.502* Transportation 3.13 2.94 –0.183 0.902*** Public administration 4.87 2.21 –2.664*** –0.440*** Percentage of workers by type of employment Regular wage or salaried 11.92 4.24 –7.673*** –2.966*** Casual 16.05 10.33 –5.719*** –7.371*** Employer 0.56 0.39 –0.176** –0.230*** Self-employment 42.98 44.97 1.986*** 3.346*** Unpaid family worker 28.46 40.05 11.590*** 7.232*** Percentage of workers who are informal 89.09 96.26 7.170*** 1.872*** Source: Authors’ calculations, based on household and labor force surveys of Afghanistan. Note: The “difference� column indicates the difference in labor market characteristics between households residing in high-conflict areas and those in low-conflict areas, as de�ned in annex 7.A. The “difference (rural)� columns computes the same difference for rural areas only. *** Signi�cant at the 1% level; ** signi�cant at the 5% level; * signi�cant at the 10% level. TABLE 7B.2 Labor market characteristics in high-conflict and low-conflict areas of India, 2000 and 2008 2000 2008 Low- High- Low- High- conflict conflict conflict conflict Difference-in- Difference-in- India areas areas areas areas difference difference (rural) Percentage of working age population employed 60.75 55.98 58.25 54.37 0.896** 1.118** Male 81.00 81.17 80.74 80.88 –0.033 Female 40.01 29.81 35.36 27.17 2.011*** Unemployment rate 2.23 2.24 2.11 2.09 –0.024*** 0.034*** Male 3.10 3.59 3.09 3.31 –0.260*** Female 1.34 0.85 1.11 0.83 0.228*** Labor force participation rate 62.99 58.22 60.35 56.46 0.871*** 1.153*** Male 84.11 84.76 83.83 84.19 –0.293*** Female 41.35 30.65 36.46 28.00 2.238*** Percentage of workers in urban areas 24.05 21.45 27.87 22.36 –2.902*** (continues next page) CREATING JOBS IN CONFLIC TAFFEC TED AREAS 313 TABLE 7B.2 Labor market characteristics in high-conflict and low-conflict areas of India, 2000 and 2008 (continued) 2000 2008 Low- High- Low- High- conflict conflict conflict conflict Difference-in- Difference-in- India areas areas areas areas difference difference (rural) Percentage of workers by sector of employment Agriculture 58.18 61.35 50.18 56.38 3.027*** 2.327*** Manufacturing 11.20 10.65 13.79 11.27 –1.960*** –0.525 Construction 5.71 4.08 8.81 6.46 –0.717*** –1.133*** Commerce 10.77 10.62 11.19 10.66 –0.381 0.061 Transportation 3.84 3.69 4.95 4.40 –0.397* –0.516** Public administration 2.79 2.38 1.86 1.97 0.515*** 0.010 Percentage of workers by type of employment Regular wage or salaried 15.94 13.70 18.94 14.83 –1.865*** –1.312*** Casual 30.73 31.45 29.34 30.55 0.491*** –0.259 Employer 1.03 0.60 1.44 0.96 –0.053 –0.265** Self-employment 29.61 32.71 30.90 32.92 –1.086** –1.010* Unpaid family worker 22.69 21.53 19.38 20.73 2.514*** 2.846*** Percentage of workers who are informal 86.37 87.94 91.27 92.07 –0.762** –1.033*** Source: Authors’ calculations, based on household and labor force surveys of India. Note: The “difference-in-difference� column indicates the difference in labor market trends between households residing in high-conflict areas and households residing low-conflict areas between 2000 (when there was no conflict in any of the areas considered here) and 2007 (when only the conflict group was in conflict). The “difference-in-difference (rural)� columns computes the same difference for rural areas only. *** Signi�cant at the 1% level; ** signi�cant at the 5% level; * signi�cant at the 10% level. TABLE 7B.3 Labor market characteristics in high-conflict and low-conflict areas of Nepal, 1996 and 2004 1996 2004 Low- High- Low- High- conflict conflict conflict conflict Difference-in- Difference-in- Nepal areas areas areas areas difference difference (rural) Percentage of working age population employed 73.84 81.92 75.91 86.34 2.352* 0.588 Male 79.16 86.26 83.41 89.50 –1.010 Female 69.07 78.09 69.59 83.83 5.222*** Unemployment rate 1.68 1.71 2.32 1.17 –1.181*** –0.788** Male 2.68 2.23 3.16 1.89 –0.819 Female 0.79 1.25 1.62 0.59 –1.484*** Labor force participation rate 75.52 83.62 78.24 87.51 1.172 –0.200 Male 81.84 88.49 86.56 91.39 –1.829 Female 69.87 79.34 71.21 84.42 3.738** Percentage of workers in urban areas 7.34 2.31 17.40 6.16 –6.213*** Percentage of workers by sector of employment Agriculture 77.79 87.16 72.67 84.76 2.726** –1.946 Manufacturing 4.52 2.65 7.56 3.08 –2.619*** –1.151* Construction 2.47 2.57 2.99 2.84 –0.247 0.265 Commerce 7.45 3.00 7.30 3.02 0.161 1.267* Transportation 1.21 0.58 1.64 0.49 –0.521* –0.530* Public administration 0.62 0.21 0.76 0.23 –0.121 0.183 (continues next page) 314 MORE AND BETTER JOBS IN SOUTH ASIA TABLE 7B.3 Labor market characteristics in high-conflict and low-conflict areas of Nepal, 1996 and 2004 (continued) 1996 2004 Low- High- Low- High- conflict conflict conflict conflict Difference-in- Difference-in- Nepal areas areas areas areas difference difference (rural) Percentage of workers by type of employment Regular wage or salaried 7.66 4.80 8.91 3.89 –2.167*** 0.485 Casual 17.86 11.66 16.30 7.02 –3.084*** –3.835*** Unpaid family worker 49.35 57.22 47.78 60.08 4.420** 3.048* Percentage of workers who are informal 92.25 95.75 90.79 95.98 1.685** –0.831 Source: Authors’ calculations, based on household and labor force surveys of Nepal. Note: The “difference-in-difference� column indicates the difference in labor market trends between households residing in high-conflict areas and households residing low-conflict areas between1996 (before the conflict) and 2004 (after the conflict had begun). The “difference-in-difference (rural)� columns computes the same difference for rural areas only. *** Signi�cant at the 1% level; ** signi�cant at the 5% level; * signi�cant at the 10% level. TABLE 7B.4 Labor market characteristics in high-conflict and low-conflict areas of Sri Lanka, 2004 and 2008 2004 2008 Low- High- Low- High- conflict conflict conflict conflict Difference-in- Difference-in- Sri Lanka areas areas areas areas difference difference (rural) Percentage of working age population employed 54.70 45.92 56.51 45.95 1.787 1.655 Male 75.55 75.52 77.80 76.27 1.504 Female 34.96 19.74 37.60 18.99 3.384** Unemployment rate 4.73 5.35 3.32 4.03 –0.083 –0.084 Male 4.79 5.68 3.15 4.01 0.032 Female 4.67 5.06 3.48 4.05 –0.180 Labor force participation rate 59.43 51.27 59.84 49.98 1.704 1.571 Male 80.33 81.20 80.95 80.28 1.536 Female 39.64 24.80 41.08 23.03 3.204** Percentage of workers in urban areas 12.13 22.75 10.24 21.42 –0.558 Percentage of workers by sector of employment Agriculture 31.66 27.39 30.85 24.03 2.545 3.362* Manufacturing 18.58 11.75 19.21 12.66 –0.273 –1.876 Construction 5.36 4.42 6.32 10.52 –5.148*** –5.303*** Commerce 14.66 24.55 15.38 19.91 5.362*** 5.024*** Transportation 5.82 5.87 6.11 4.63 1.537* 1.750* Public administration 7.38 9.34 6.78 10.80 –2.059* –2.340* Percentage of workers by type of employment Regular wage or salaried 37.64 26.59 27.18 20.27 –4.138** –5.741*** Casual 23.00 28.54 29.84 43.15 –7.758*** –9.203*** Employer 2.79 2.91 2.95 2.11 0.964 1.025 Self-employment 27.27 31.50 29.25 29.42 4.065** 6.078*** Unpaid family worker 9.31 10.45 10.78 5.05 6.868*** 7.842*** Percentage of workers who are informal 81.24 79.10 71.13 75.99 –6.992*** –6.556*** Source: Authors’ calculations, based on household and labor force surveys of Sri Lanka. Note: The “difference-in-difference� column indicates the difference in labor market trends between households residing in high-conflict areas and households residing in low- conflict areas between 2008 (after the cease �re) and 2004 (during the cease �re). The “difference-in-difference (rural)� columns computes the same difference for rural areas only. *** Signi�cant at the 1% level; ** signi�cant at the 5% level; * signi�cant at the 10% level. TABLE 7B.5 Educational attainment in high-conflict and low-conflict areas of South Asia, by country (percentage of working-age population) Low- High- Low- High- conflict conflict conflict conflict Difference Difference-in- Difference-in- Counry/level of education areas areas areas areas Difference (rural) difference difference (rural) Afghanistan 2008 No education 67.87 82.19 n.a n.a 14.320*** 4.500*** n.a n.a Some primary school 8.74 4.32 n.a n.a –4.418*** –3.450*** n.a n.a Completed primary school 10.04 6.73 n.a n.a –3.315*** –0.906*** n.a n.a Lower-secondary schooling 5.45 3.67 n.a n.a –1.786*** 0.433*** n.a n.a Upper-secondary schooling 5.94 2.59 n.a n.a –3.340*** –0.346** n.a n.a Tertiary education 1.96 0.50 n.a n.a –1.460*** –0.231*** n.a n.a India 2000 2008 No education 37.48 45.51 28.79 35.80 n.a n.a –1.017*** –1.563*** Some primary school 9.23 8.37 8.36 7.66 n.a n.a 0.162 0.405 Completed primary school 13.92 10.06 15.30 13.53 n.a n.a 2.094*** 2.144*** Lower-secondary schooling 28.08 25.67 31.70 29.49 n.a n.a 0.202 0.500 Upper-secondary schooling 10.85 9.90 8.96 7.54 n.a n.a –0.464* –1.072*** Tertiary education 0.43 0.49 6.90 5.98 n.a n.a –0.975*** –0.414*** Nepal 1996 2004   No education 61.27 69.90 48.22 57.94 n.a n.a 1.092 -2.343 Some primary school 7.90 10.49 9.46 11.71 n.a n.a –0.340 0.237 Completed primary school 12.69 10.25 13.98 13.93 n.a n.a 2.387** 1.923* Lower-secondary schooling 15.58 8.43 22.89 14.69 n.a n.a –1.053 0.029 Upper-secondary schooling 1.63 0.58 3.46 1.21 n.a n.a –1.214*** 0.104 Tertiary education 0.93 0.34 1.99 0.53 n.a n.a –0.873*** 0.051 Sri Lanka 2004 2008   No education 4.79 7.56 3.13 5.28 n.a n.a 0.626 0.145 Some primary school 14.99 22.73 8.36 13.24 n.a n.a 2.853*** 2.543** Completed primary school 23.02 26.10 19.61 25.95 n.a n.a –3.260*** –4.847*** Lower-secondary schooling 43.39 35.58 53.89 43.77 n.a n.a 2.307* 2.833* Upper-secondary schooling 11.64 6.04 12.34 9.30 n.a n.a –2.549*** –0.381 Tertiary education 2.17 1.97 2.69 2.46 n.a n.a 0.024 –0.293 Sources: Authors’ calculations, based on household surveys and conflict data described in annex 7A. Note: The “difference� column indicates the difference in labor market characteristics between households residing in high-conflict areas and households residing in low-conflict areas, as de�ned in annex 7A. The “difference (rural)� column computes the same difference in rural areas only. The “difference-in-difference� column is calculated as in tables 7B.2–7B.4 for the respective countries. The “difference-in-difference (rural)� column computes the same difference in trends in rural areas only. n.a. = Not applicable. *** Signi�cant at the 1% level; ** signi�cant at the 5% level; * signi�cant at the 10% level. 315 316 MORE AND BETTER JOBS IN SOUTH ASIA Notes provide enough coverage of conflict-affected areas. 1. This chapter focuses on internal armed con- 7. Although establishing causality is dif�cult, flict against the state. Although the civil wars there is some evidence that lower economic in Nepal and Sri Lanka recently ended, both growth and poor economic performance countries are included in the analysis, because increase the likelihood of armed conflict. A many of the challenges addressed—and the very robust result in the cross-country litera- policy options discussed—refer to postcon- ture on conflict is that poorer countries and flict areas. Moreover, a better understanding countries with slower growth are at higher of the impact of conflict in these countries can risk of experiencing civil war (Collier and provide insights for the rest of South Asia. Hoeffler 2004; Fearon and Laitin 2003). Using With this in mind, throughout this chapter rainfall as an exogenous determinant of eco- the term conflict affected is used to character- nomic growth in Africa yields a similar result, ize areas that are either experiencing violence suggesting that there is a causal relationship from internal armed conflict or recovering between a lack of economic growth and the from armed conflict. incidence of conflict (Miguel, Satyanath, and 2. Many of the issues discussed in this chapter Sergenti 2004). Within-country analyses also could also be relevant for other types of dis- �nd a strong link between poverty and eco- ruptions, such as natural disasters. Asia as nomic shocks on the one hand and the subse- a whole is the continent most often hit by quent emergence of conflict on the other (see natural disasters. Since 1950, it has experi- Dube and Vargas 2008 on Colombia; Do and enced more than 3,900 disasters, 42 percent Iyer 2010 on Nepal; and Iyer 2009 on South of the global disasters in this period. These Asia). disasters killed more than 5.6 million people 8. In this particular study, de�nitions of poverty (83 percent of the world total) and affected are based on subjective measures and a locally more than 5.4 billion others (90 percent of de�ned poverty line for each community. A the world total) (CRED 2011). “ladder of life� was used to determine who in 3. Measuring the intensity of armed conflict is a community quali�ed as poor. During a lad- very dif�cult. Due to data limitations and for der of life exercise, focus group participants comparability reasons, the indicator of inten- describe the steps (or levels) of overall well- sity used here is the number of direct deaths being of households in their villages. They from armed conflict. Data sources and meth- then rank up to 150 households in their vil- odology are discussed in annex 7A. lage/neighborhood according to their step on 4. Migration can also play an important role the ladder of life, currently and in the past. as a source of jobs for skilled and unskilled The results of this ranking exercise produce workers from conflict-affected countries, a community mobility matrix for each village both during and after conflicts (World Bank that can be used to estimate the evolution 2011). of prosperity, including the share of house- 5. This assumption is probably stronger in some holds that moved out of poverty, the share of cases than in others, given that important dif- households that experienced upward mobil- ferences in labor market indicators in conflict ity of any kind, and the degree of mobility and nonconflict areas often predate the con- (either up or down). flict. Lack of data prevents analysis of pre- 9. Smuggling and other illegal activities (such conflict trends. as poppy cultivation in Afghanistan) are 6. In India and Nepal, the earlier year is a year also a major source of employment in many of peace; during the later year, there was con- conflict-affected areas in the region. The scope flict in areas of both countries. In Sri Lanka, of these activities is dif�cult to quantify. In the year of conflict is 2008; 2004 was a year 2005, illegal opium revenues were estimated of relative peace (the middle of a cease�re). to reach nearly $3 billion, equivalent to half of The effect of a few additional years of vio- Afghanistan’s legal economy (UNODC 2005). lence is measured. In Afghanistan, where only Households in Afghanistan that cultivate one round of household survey data is avail- poppy are signi�cantly less likely to be poor able, the simple differences in outcomes for (World Bank 2005). high-conflict and low-conflict areas are com- 10. The effect of conflict on the employment pared. In Pakistan, household surveys do not rate—as on other labor market indicators—is CREATING JOBS IN CONFLIC TAFFEC TED AREAS 317 likely to vary depending on the context and 2008). Bangladesh, India, Nepal, Pakistan, the subgroups of the population considered. and Sri Lanka all have SEZs. Afghanistan has The effect may also change in the immedi- no duty-free import zones. It is considering ate aftermath of the conflict. In Pakistan, for establishing trade facilitation zones, export example, after the �rst wave of the militancy processing zones, or both to enhance export crisis in the North West Frontier Province potential. Legislation before the U.S. Congress (Khyber Pakhtunkhwa), women’s movements would establish special import tax status for were severely limited after internally displaced certain categories of goods made in recon- people arrived. As a result, women became struction opportunity zones (http://www.state less likely to work (World Food Programme .gov/e/eeb/rls/othr/ics/2010/138776.htm). 2011). 18. Even this �gure for the north is likely to be an 11. As discussed in chapter 3, the female employ- overestimate, as the survey could not cover ment rate in Nepal is more than twice the the remote districts of Kilinochchi, Mannar, unweighted average of employment rates of and Mullaitivu, where up to 90 percent of the the other South Asian countries (40 percent- households may not have power (World Bank age points higher). Conflict seems to explain 2007). only a small share of this, as the evidence indi- 19. For more information on community-driven cates that the conflict led to a 5 percentage development projects in postconflict environ- point increase in employment rates among ments, see Barron (2010). women in areas most affected by conflict. 20. See http://www.clintonfoundation.org/haiti_ 12. In India, the difference in preconflict youth longstanding/haiti_fact_sheet.php. unemployment rates is not statistically sig- 21. For a discussion of interventions to provide ni�cant. educational services to displaced populations, 13. Except possibly in Sri Lanka, unemployment see Mooney and French (2005). is not a very strong indicator of labor market 22. The program is not of�cially a DDR program conditions in South Asia, where few people but rather a “campment, UN monitoring, can afford to be unemployed, making these and reintegration� (CMR) program. In the trends dif�cult to interpret. long term, the goals of both programs are 14. What look like worse labor market outcomes essentially the same (Gurung 2006). in conflict environments do not necessarily 23. The Nepal Peace Project also includes activa- translate into lower average labor incomes. tion programs to assist families. In Nepal, the median household labor income 24. The review of DDR programs by Carames per capita in low-conflict areas was already and Sanz (2009) �nds that the disarmament 1.5 times that of conflict-affected areas before and demobilization stages take about 16 the conflict; the conflict increased this gap to months on average, ranging from 4 months 1.7 by 2004. In contrast, in Afghanistan and in Angola to 27 months in Sierra Leone. The Sri Lanka, labor incomes were higher in con- reintegration phase usually takes consider- flict-affected areas. ably longer, lasting two years on average. 15. For example, school enrollment often falls 25. For a more detailed discussion on the lessons signi�cantly after natural disasters, as chil- learned from DDR programs, see Meek and dren join the labor force (Jacoby and Skou�as Malan (2004). 1997; Jensen 2000; Santos 2007). 26. For a review of experiences with public work 16. In India, Maldives, and Sri Lanka, �rms were programs around the world, see Del Ninno, not asked about political instability. Surveys Subbarao, and Milazzo (2009). there referred to “government policy insta- 27. The authors are grateful to Ernest Sergenti bility,� which includes both macroeconomic (World Bank) for his contributions to this and political instability. In Sri Lanka, gov- section. ernment policy instability was cited as the 28. See the Poverty Alleviation Fund website second-most severe constraint; in India and (www.pafnepal.org.np/index.php). Maldives, it was not among the top �ve con- 29. For a more detailed discussion of micro�- straints. nance, cooperatives, and self-help groups in 17. SEZs have become increasingly popular in the postconflict situations, see ILO (2003). developing world: as of 2008, an estimated 30. In conflict-affected communities in Afghani- 3,000 SEZs in more than 130 countries had stan, for example, employment by NGOs created more than 68 million direct jobs (FIAS was rated as highly desirable, and households 318 MORE AND BETTER JOBS IN SOUTH ASIA with an NGO employee were rated as best off Barron, P. 2010. “CDD in Post-Conflict and (World Bank 2010a). Conflict-Affected Areas: Experiences from 31. Afghan First has not been rigorously evalu- East Asia.� Background paper for the World ated, but initial data from a capacity build- Bank World Development Report 2011. ing project for local enterprises suggest http://wdr2011.worldbank.org/sites/default/ that Afghan companies supported by the files/pdfs/WDR%20Background%20Paper_ project over the years are increasingly win- Barron_0.pdf. ning contracts. The increase has become Beath, A., F. Christia, R. Enikolopov, and S.A. steeper since the introduction of Afghan Kabuli. 2010. “Randomized Impact Evaluation First in 2009. For more information, see of Phase-II of Afghanistan’s National Solidarity http://www.nato.int/cps/en/SID-BF6A3963- Programme (NSP): Estimates of Interim 402643FF/natolive/official_texts_62851. Program Impact from First Follow-up Survey.� htm?mode=pressrelease). http://www.nsp-ie.org/reports/BCEK-Interim_ 32. Of course, the security situation itself can be Estimates_of_Program_Impact_2010_07_25 influenced by policy—not only by direct secu- .pdf. rity measures like policing but also by other Beber, B., and C. Blattman. 2010. “The Industrial interventions discussed in this chapter. Organization of Rebellion: The Logic of 33. For instance, in countries in which ethnic Forced Labor and Child Soldiering.� Working differences were at the heart of the conflict, paper, Department of Political Science, postconflict public work programs need to be Yale University, New Haven, CT. http:// careful not to favor one group over others. In chrisblattman.com/documents/research/2010 cases where the conflict arose as a result of .IOofRebellion.pdf. issues in the management and distribution of Bhalotra, S., and M. Umana-Aponte. 2010. natural resources, such as forests, any post- “The Dynamics of Women’s Labor Supply conflict job creation strategy needs to ensure in Developing Countries.� IZA Discussion that property rights are strengthened and that Paper 4879, Institute for the Study of Labor, the rules of the game for the exploitation of Bonn. the resources are clear and fair. Blat tman, C ., and J. A nnan. 2010. “T he 34. Some notable exceptions are the ongoing Consequences of Child Soldiering.� Review of impact evaluation of the National Solidarity Economics and Statistics 92 (4): 882–98. Program in Afghanistan (Beath and others Blattman, C., and N. Fiala. 2009. “Uganda: 2010) and ongoing work evaluating DDR Post-War Youth Vocational Training.� http:// programs and employment interventions in chrisblattman.com/projects/nusaf_yo/. Uganda (Blattman and Fiala 2009). Blattman, C., and E. Miguel. 2010. “Civil War.� Journal of Economic Literature 48 (1): 3–57. Boyce, J. 2007. “Public Finance, Aid and References Post-Conflict Recovery.� Working paper, Akresh, R., and D. de Walque. 2008. “Armed Department of Economics, University of Conflict and Schooling: Evidence from the Massachusetts, Amherst, MA. 1994 Rwandan Genocide.� IZA Discussion Brueck, T. 2004. “Coping Strategies Post-War Paper 3516, Institute for the Study of Labor, Rural Mozambique.� Working paper, Queen Bonn. Elizabeth House, University of Oxford, Amnesty International. 2003. “No One Listens to Oxford. Us and No One Treats Us as Human Beings: Carames, A., and E. Sanz. 2008. “Analysis Justice Denied to Women.� Report ASA of Disa r m a ment , D emobi l i z at ion a nd 11/023/2003, London. Reintegration. DDR. Programmes in the Annan, J., C. Blattman, K. Carlson, and D. World during 2007.� School for a Culture of Mazurana. 2007. “Making Reintegration Peace, Bellaterra, Spain. http://escolapau.uab Work for Youth in Northern Uganda.� .cat/img/programas/desarme/ddr005i.pdf. Research Brief, Survey of War Affected Youth ———. 20 09. “A nalysis of Disarmament, (SWAY), 11. Demobilization and Reintegration. DDR. Arunatilake, N., S. Jayasuriya, and S. Kelegama. Programmes in the World during 2008.� 2001. “The Economic Cost of the War in School for a Culture of Peace, Bellaterra, Spain. Sri Lanka.� World Development 29 (9): ht tp: //escolapau.uab.es /img /prog ramas / 1483–1500. desarme/ddr/ddr2009i.pdf. CREATING JOBS IN CONFLIC TAFFEC TED AREAS 319 Card, D. 1999. “The Causal Effect of Education Dube, O., and J. Vargas. 2008. “Commodity on E arnings.� I n Han dbook of L abor Price Shocks and Civil Conflict: Evidence Economics, vol. 3, ed. O. Ashenfelter and D. f rom C olombi a .� Work i ng Pap er 14 , Card, 1801–63. Amsterdam: Elsevier. Harvard University, Center for International C ayc e d o , A . , a n d E . M e n d o z a . 2 0 0 8 . Development, Cambridge, MA. “Colombian Flowers and Floriculture.� Fearon, J., and D. Laitin. 2003. “Ethnicity, http://issuu.com/acaycedo/docs/colombian Insurgency, and Civil War.� American Political _grown_208_pages_single. Science Review 97 (1): 75-90. Central Bank of Sri Lanka. Various years. Annual FIAS (Foreign Investment Advisory Service). Reports. Colombo. 2008. “Special Economic Zones: Performance, Chaudhury, N., and D. Parajuli. 2009. “Pilot Lessons Learned and Implications for Zone Evaluation: Nepal Community School Support Development.� World Bank, Washington, DC. Project.� http://siteresources.worldbank.org/ Finegan, T., and R. Margo. 1994. “Work Relief SOUTHASIAEXT/Resources/223546-11924 and the Labor Force Participation of Married 13140459/4281804-1215548823865/Nepal Women in 1940.� Journal of Economic CSSP.pdf. History 54 (1): 64-84. Chauvet, L., P. Collier, and M. Duponchel. 2010. Gardeazabal, J. Forthcoming. “Methods for “What Explains Aid Project Success in Post- Measuring Aggregate Costs of Conflict.� In Conflict Situations?� Policy Research Working Handbook of the Economics of Peace and Paper 5418, World Bank, Washington, DC. Confl ict, ed. M. Gar� nkel and S. Skaperdas. Chen, S., N. Loayza, and M. Reynal-Querol. Oxford: Oxford University Press. 2007. “The Aftermath of Civil War.� Policy Glew we, P. 20 02 . “Schools and Skills in Research Working Paper 4190, World Bank, Developing Countries: Education Policies Washington, DC. and Socioeconomic Outcomes.� Journal of Collier, P., and A. Hoeffler. 2004. “Greed and Economic Literature 40 (2): 436–82. Grievance in Civil Wars.� Oxford Economic Government of India. 2007. “India’s Forests.� Papers 56 (4): 563–95. Ministry of Environment and Forests, New Collier, P., A. Hoeffler, and N. Sambanis. 2005. Delhi. http://tribal.nic.in /writereaddata/ “The Collier-Hoeffler Model of Civil War mainlinkFile/File1256.pdf. Onset and the Case Study Project Research ———. 2010. “Status Report on Implementation Design.� In Understanding Civil War: Evidence of the Scheduled Tribes and Other Traditional and Analysis, ed. P. Collier and N. Sambanis, Forest Dwellers. Recognition of Forest Rights 1–33. Washington, DC: World Bank. Act, 2006.� Ministry of Tribal Affairs New Collier, P., A. Hoeffler, and M. Söderbom. 2008. Delhi. http://tribal.nic.in /writereaddata/ “Post-Conflict Risks.� Journal of Peace mainlinkFile/File1256.pdf. Research 45 (4): 461–78. Government of Sri Lanka. 2010. “Operational Collier, P., and N. Sambanis. 2005. Understanding Guidelines for E N ReP Cash for Work Civil War: Evidence and Analysis. Washington, Program.� Ministry of Nation Building and DC: World Bank. Estate Infrastructure Development (MNBEID), CRED (Centre for Research on the Epidemiology Colombo. of Disasters). 2011. EM-DAT: The OFDA/ Gurung, C.B. 2006. “So Far, So Good.� Nepali C R E D E mergenc y Disaster Dat abas e. Times, December 7, Kathmandu. Université Catholique de Louvain, Louvain, Herman, L., and Q. Dhanani. 2010. “Roshan: Belgium. http://www.emdat.be. Light at the End of the Tunnel in Afghanistan.� Deininger, K. 2003. “Causes and Consequences H a r va rd Busi ness S chool Case St udy, of Civil Strife: Micro-Level Evidence from Cambridge, MA. Uganda.� Oxford Economic Papers 55: Hoeffler, A. 2011. “On the Causes of Civil War.� 579–606. Forthcoming in Oxford Handbook of the Del Ninno, C., K. Subbarao, and A. Milazzo. Economics of Peace and Conflict, ed. M. 2009. How to Make Public Works Work: A Gar� nkel and S. Skaperdas. Oxford: Oxford Review of the Experiences. SP Discussion University Press. Paper 0905, World Bank, Washington, DC. Hoeffler, A., and M. Reynal-Querol. 2003. Do, Q., and L. Iyer. 2010. “Geography, Poverty “Measuring the Costs of Confl ict.� Working and Conflict in Nepal.� Journal of Peace paper, Centre for the Study of African Economies Research 47 (6): 735–48. and St. Antony’s College, University of Oxford, 320 MORE AND BETTER JOBS IN SOUTH ASIA Oxford. http://www.conflictrecovery.org/ Review, Papers and Proceedings 90 (2): bin/2003_Hoeffler_Reynal-Measuring_the_ 399–404. Costs_of_Conflict.pdf. Jones, N., R. Holmes, H. Marsden, S. Mitra, and Humphreys, M., and J. Weinstein. 20 07. D. Walker. 2009. “Gender and Social Protection “Demobilization and Reintegration.� Journal in Asia: What Does the Crisis Change?� of Conflict Resolution 51 (4): 531–67. Background paper for the Conference on the Indian Express. 2009. “Lanka Plans to Set Up Impact of the Global Economic Slowdown Free Trade Zones in LTTE Bastions.� October on Poverty and Sustainable Development in 23. http://www.indianexpress.com /news/ Asia and the Pacific, Overseas Development lanka-plans-to-set-up-free-trade-zones-in-ltte- Institute, London. bastions/532356/. Ketkar, S., and D. Ratha. 2009. “New Paths I DMC (Internal Displacement Monitoring to Funding.� Finance and Development 46 Center). 2009. Internal Displacement: Global (2). http://www.imf.org /external/pubs/ft/ Overview of Trends and Developments in fandd/2009/06/ketkar.htm. 20 08. http://www.internal-displacement Lopez, H., and Q. Wodon. 2005. “The Economic .org. Impact of Armed Conflict in Rwanda.� Journal ———. 2011. IDP statistics. http://www.internal- of African Economies 14 (4): 586–602. displacement.org /8025708F0 0 4C E90B / McKay, A., and S. Loveridge. 2005. “Exploring the (httpRegionPages. /2DE3ACEE54F9A63B802 Paradox of Rwandan Agricultural Household 570A6005588C1?OpenDocument. Income and Nutritional Outcomes in 1990 and IISS (International Institute for Strategic Studies). 2000.� Staff Paper 2005–2006, Department 2010. Armed confl ict database. http://www. of Agricultural Economics, Michigan State iiss.org/publications/armed-conflict-database/. University, East Lansing, MI. ILO (International Labour Of� ce). 2003. Jobs McLeod, D., and M. Davalos. 2008. “Sustainable after War: A Critical Challenge in the Peace Po s t- C o n f l i c t E mploy m e nt C r e at io n : and Reconstruction Puzzle, ed. E. Date- From Stabilization to Poverty Reduction.� Bah. Geneva. Geneva: International Labour Working paper, Department of Economics, Of�ce. Fordham University, New York. http:// INSEC (Informal Sector Service Center). 2008. w w w. ford h a m . e du /e c onom ic s /m cle o d / Human Rights Yearbook 2008. Kathmandu. PostConflictEmployment10.pdf. ———. 2010. “Number of Victims Killed by Meek, S., and M. Malan, eds. 2004. Identifying State and Maoists in Connection with the Lessons from DDR Experiences in Africa. People’s War.� http://www.insec.org.np/ Institute for Security Studies Monograph 106, pics/1247467500.pdf. Pretoria, South Africa. ISAF (International Security Assistance Force). Menon, N., and Y. van der Meulen. 2010. “Gender 2011. Afghanistan Peace and Reintegration and Conflict in Nepal: Testing for ‘Added Prog ram website. ht tp: //w w w.isaf.nato Worker’ Effects.� Working paper, Rutgers .int /article/focus/afghanistan-peace-and- University, Department of Economics, New reconciliation-program.htm. Brunswick, NJ. Islamic Republic of Afghanistan. 2010. Afghanistan M IGA (Multilateral Investment Guarantee Peace and Reintegration Program. National Agency). 2011. World Investment and Political Security Council, Kabul. http://www.sipri Risk. Washington, DC. .org/blogs/Afghanistan/Afghan%20Peace%20 Miguel, E., S. Satyanath, and E. Sergenti. 2004. and%20Reconciliation%20Programme-%20 “Economic Shocks and Civil Conflict: An draft-%20Apr%2010%20.pdf. Instrumental Variables Approach.� Journal of Iyer, L. 2009. “The Bloody Millennium: Internal Political Economy 112 (4): 725–53. Conflict in South Asia.� Working Paper 09-086, Mooney, E., and C. French. 2005. “Barriers and Harvard Business School, Cambridge, MA. Bridges: Access to Education for Internally Jacoby, H., and E. Skou�as. 1997. “Risk, Financial Displaced Children.� Working paper, Brookings Markets, and Human Capital in a Developing Institution, Washington, DC. Country.� Review of Economic Studies 64 (3): National Counterterrorism Center. 2008. 2007 311–35. NCTC Report on Terrorism. www.nctc.gov. Jensen, R. 2000. “Agricultural Volatility and ———. 2009. 2008 NCTC Report on Terrorism. Investments in Children.� American Economic www.nctc.gov. CREATING JOBS IN CONFLIC TAFFEC TED AREAS 321 PADF (Pan-American Development Foundation). U S A I D ( U . S . A g e n c y for I nt e r n at ion a l 2008. “Socially-Conscious Flowers: Flowers Development). 2006. Success Story: Flower Help to Suppor t Displaced Colombian Industry Gives Jobs to Displaced. http://www Families through the School of Floriculture.� .usaid.gov/stories/colombia/ss_co_flowers http://www.padf.org/DOCUMENTS/News .html. Stories/080204_asocolflores_prnewswire Verwimp, P., and J. Van Bavel. 2010. “Violent .pdf. Conflict, Gender, and Schooling: Micro-Level Radelet, S. 2007. “Reviving Economic Growth Evidence from Burundi.� Paper presented at a in Liberia.� Working Paper 133, Center for workshop on gender and conflict, World Bank, Global Development, Washington, DC. Washington, DC, June 9. Santos, I. 2007. “Disentangling the Effects Willibald, S. 2006. “Does Money Work? Cash of Natural Disasters on Children: 2001 Transfers to Ex-Combatants in Disarmament, Earthquakes in El Salvador.� Ph.D. dissertation, Demobilization and Reintegration Processes.� Kennedy School of Government, Harvard Disasters 30 (3): 316–39. University, Cambridge, MA. World Bank. 2001. “Nepal: Priorities and Schweitzer, M. 1980. “World War II and Female Strategies for Education Reform.� Report Labor Force Participation Rates.� Journal of 26509-NEP, Washington, DC. Economic History 40 (1) 89–95. ———. 2005. Afghanistan Poverty, Vulnerability Shemyakina, O. 2011. “The Effect of Armed and Social Protection: An Initial Assessment. Conflict on Accumulation of Schooling: Results Washington, DC. from Tajikistan.� Journal of Development ———. 2006. India: Unlocking Opportunities Economics 95 (2): 186–200. for Forest-Dependent People in India. Main Sigsgaard, M. 2009. Education and Fragility Report, vol. 1, Washington, DC. in Afghanistan: A Situational Analysis. ———. 2007. “Sri Lanka Poverty Assessment. International Institute for Educational Planning, Engendering Growth and Equity: Opportunities United Nations Educational, Scientific and and Challenges.� South Asia Region, Poverty Cultural Organization (UNESCO), Paris. Reduction and Economic Management Sector Smith, D. 2004. “Trends and Causes of Armed Unit, Washington, DC. Conflict.� In The Berghof Handbook for ———.2008a. “The Afghanistan Investment Conflict Transformation ed. D. Bloomfeld, Climate in 2008.� http://siteresources.world M. Fischer, and B. Schmelzle. Berlin: Berghof bank.org/AFGHANISTANEXTN/Resources/ Research Centre for Constructive Conflict 305984-1237085035526/5919769-1258729 Management. http://www.berghofhandbook 848597/1AFInvestmentClimatesurveyReport .net/articles/smith_handbook.pdf. Final.pdf. Stewart, F., C. Huang, and M. Wang. 2001. ———. 2008b. “Liberia Poverty Reduction “Internal Wars in Developing Countries: An Strategy Paper.� June, Washington, DC. Empirical Overview of Economic and Social ———. 2009a. “Emergency Project Paper on a Consequences.� In War and Underdevelopment, Proposed Additional Grant in the Amount of vol. 1, ed. F. Stewart and V. Fitzgerald, 67–103. SDR 50.9 Million.� April. Washington, DC. Oxford: Oxford University Press. ———. 2009b. “Pakistan: Preliminary Damage U N ( Un ite d Nat ion s). 2 010. I n t eg ra t e d and Needs Assessment: Immediate Restoration Disarmament, Demobilization, and and Medium Term Reconstruction in Crisis Reintegration Standards. http://www.unddr Affected Areas. Washington, DC. .org/iddrs/framework.php. ———. 2010a. Moving out of Poverty: Rising UNODC (United Nations Of�ce on Drugs and from the Ashes of Conflict, ed. D. Narayan Crime). 2005. Afghanistan Opium Survey. and P. Petesch. Washington, DC. Kabul. ———. 2010b. “Nepal Community School Support Urdal, H. 2006. “A Clash of Generations? Youth Project: Project Performance Assessment Bulges and Political Violence.� International Report.� Report 55407, Washington, DC. Studies Quarterly 50 (3): 607–30. ———. 2010c. “Sri Lanka: Reshaping Economic ———. 2008. “Population, Resources and Violent Geography.� South Asia Region, Poverty Conflict: A Sub-National Study of India 1956– Reduction and E conomic Management 2002.� Journal of Conflict Resolution 52 (4): D e p a r t m e n t a n d S r i L a n k a C ou nt r y 590–617. Management Unit, Washington, DC. 322 MORE AND BETTER JOBS IN SOUTH ASIA ———. 2011a. Doing Business Dat aba se. World Bank and ADB (Asian Development Washington, DC: World Bank. Bank). 2010. Post Crisis Needs Assessment. ———. 2011b. World Development Indicators. Islamabad. Washington, DC: World Bank. World Food Programme. 2011. “Pakistan-Food ———. 2011c. World Development Report Security and Market Assessment in Crisis- 2011: Confl ict, Security, and Development. Affected Areas of NWFP and FATA.� Working Washington, DC. paper, Rome. Appendix A were better suited, such as education pro- Household surveys used vision or educational attainment by socio- Table A.1 summarizes the labor force and economic group. The most recent labor living standards surveys used throughout survey in Bangladesh is a welfare-monitoring this book. In countries where both types survey, which differs from a typical labor of data are available (Bangladesh, Nepal, force survey. It was used in order provide the and Pakistan), labor force surveys were the latest estimates. This survey does not allow primary source of data, as they tend to be for the calculation of all indicators used in better suited for the analysis of the labor this book. Finally, either due to coverage or market. For these countries, living standards year of surveys, only particular national sur- surveys were used to complement labor force veys were suitable for the analysis of labor survey data and to analyze poverty and market outcomes and conflict. Table A.1 employment (household consumption is not provides a summary of the different national available in most labor force surveys). These surveys used, and identi�es when a partic- living standard surveys were also used for ular survey was used to examine selected examining specific issues for which they issues only. 323 324 APPENDIX A TABLE A.1 Labor force and living standards surveys used Country Survey Year Comments Afghanistan National Risk and Vulnerability Assessment 2007/08   Bangladesh Labor Force Survey 2002/03   Labor Force Survey 2005/06   Monitoring of Employment Survey (MES) 2009 Core indicators only Household Income and Expenditure Survey 2000 Poverty analysis and education analysis only Household Income and Expenditure Survey 2005 Poverty analysis only Household Income and Expenditure Survey 2010 Poverty analysis and education analysis only Bhutan Living Standards Survey 2003   Living Standards Survey 2007   India National Sample Survey (Employment Schedule) 1983   National Sample Survey (Employment Schedule) 1993/94   National Sample Survey (Employment Schedule) 1999/2000   National Sample Survey (Employment Schedule) 2004/05   National Sample Survey (Employment Schedule) (Thin Round) 2007/08   National Sample Survey (Employment Schedule) 2009/10   Maldives Vulnerability and Poverty Assessment 1998   Vulnerability and Poverty Assessment 2004   Nepal Labor Force Survey 1998/99   Labor Force Survey 2007/08   Living Standards Survey 1995/96 Poverty analysis, education analysis, and conflict analysis only Living Standards Survey 2003/04 Poverty analysis, education analysis, and conflict analysis only Pakistan Labor Force Survey 1999/2000   Labor Force Survey 2007/08   Labor Force Survey 2008/09   Social and Living Standards Measurement Survey 2001/02 Poverty analysis and education analysis only Social and Living Standards Measurement Survey 2007/08 Poverty analysis and education analysis only Sri Lanka Labor Force Survey 2000   Labor Force Survey 2004 Education analysis and conflict analysis only Labor Force Survey 2006 Education analysis only Labor Force Survey 2008   Household Income and Expenditure Survey 1995/96 Poverty analysis only Household Income and Expenditure Survey 2006/07 Poverty analysis and education analysis only Source: Authors’ compilation. Appendix B Methodology used to analyze The following (minimal) econometric labor transitions expression illustrates the method. Let yij be binary variables that represent the various The book is interested in labor transitions— labor outcomes of interest for individual i, speci�cally, the ability of workers to move i = 1, . . . , N, in survey round (or period) j, across sectors and types of employment over where j = 1 or 2. The main labor outcomes time and the extent to which they move from of interest are the employment statuses of less desirable to more desirable jobs (and individuals: for workers in rural areas, being vice versa). Such transitions are typically employed in the rural nonfarm sector (versus measured with panel data on individuals or agriculture); for workers in urban areas, being households over time. Such panel data are employed as casual or low-end self-employed only rarely available in South Asia, however. workers (versus regular wage or salaried To compensate for the absence of true panel workers or high-end self-employed workers). data, the authors constructed synthetic panel Although the same subscript i is used to index data from multiple rounds of cross-sectional individuals, the only data available are cross- surveys for Bangladesh, India, and Nepal, sectional. adapting a technique developed by Lanjouw, The labor transitions of interest can be Luoto, and Mckenzie (2011) for studying represented by poverty transitions. The key idea behind the method is to P(yi1 < z1 | yi2 < z2) (B.1) exploit the relationship between an individ- P(yi1 < z1 | yi2 > z2) (B.2) ual’s employment status, which varies from period to period, and time-invariant indi- P(yi1 > z1 | yi2 < z2) (B.3) vidual and household characteristics in order P(yi1 > z1 | yi2 > z2) (B.4) to examine labor transitions over time. Indi- vidual time-invariant characteristics can be The first quantity represents the prob- obtained from cross-sectional surveys (with ability that individual i stays in the same certain assumptions). They thus form the employment status (for example, working connectors that help convert cross-sectional in agriculture) in both survey rounds (that data into synthetic panel data. is, no labor transition takes place over the 325 326 APPENDIX B time spanned by the two survey rounds). using both nonparametric and parametric The second quantity represents the prob- methods. ability that individual i improves his or The method for analyzing poverty tran- her employment status in the second time sitions was validated with true panel data period (for example, by moving from agri- (including data on Bosnia-Herzegovina, Lao culture in the first survey round to the People’s Democratic Republic, Peru, and rural nonfarm sector in the second survey Vietnam). The method for labor transitions round). The cut-off point is z j, set at 0.5. was validated with true panel data from Below this point (indicated by <), workers Nepal. The estimates of the lower and upper are in the less desirable employment status; bounds for poverty and labor transition above this point indicated by >), they are in rates largely encompass the true rates, pro- the more desirable employment status. An viding empirical support for the validity of analogy can be made to poverty analysis, this method. (For more technical details on where a similar cut-off point is known as the methodology used to construct pseudo- the poverty line: households are considered panel data, see Dang and Lanjouw 2011 and poor if their consumption falls below this Lanjouw, Luoto, and McKenzie 2011.) cut-off point. The synthetic panel data are constructed References from repeated rounds of cross-sectional data by applying estimated coef�cients and Dang, H.-A., and P. Lanjouw. 2011. “Measuring error terms obtained from the regressions Poverty Dynamics with Pseudo-Panels Based on Cross-Sections: An Application to Vietnam.� of individual labor outcomes on individual Working paper, Development Research Group, time-invariant characteristics in one sur- World Bank, Washington, DC. vey round to a second round. These data Lanjouw, P., J. Luoto, and D. McKenzie. 2011. can then be used to predict lower-bound “Using Repeated Cross-Sections to Explore and upper-bound estimates of labor tran- Movements in and Out of Poverty.� Policy sitions (B.1–B.4) overall and for various Research Working Paper 5550, World Bank, population groups. Estimates are made Washington, DC. ECO-AUDIT Environmental Benefits Statement The World Bank is committed to preserving Saved: endangered forests and natural resources. The • 32 trees Office of the Publisher has chosen to print • 13 million British thermal More and Better Jobs in South Asia on recy- units of total energy cled paper with 50 percent postconsumer fiber • 3,258 pounds of net in accordance with the recommended stand- greenhouse gases ards for paper usage set by the Green Press • 14,691 gallons of waste Initiative, a nonprofit program supporting water publishers in using fiber that is not sourced • 931 pounds of solid from endangered forests. For more informa- waste tion, visit www.greenpressinitiative.org. South Asia Development Matters South Asia, which is home to more than 40 percent of the world’s absolute poor, will contribute nearly 40 percent of the growth in the world’s working age population over the next several decades. More and Better Jobs in South Asia attempts to answer three questions: Has South Asia been creating an increasing number of jobs and better jobs? What has determined the quality of job creation and what is the employment challenge going forward? What demand- and supply-side bottlenecks need to be eased to meet South Asia’s employment challenge in the face of intensifying demographic pressure? The region has created just under 800,000 jobs per month during the past two decades, a rate that broadly tracks growth in its working age population. The quality of jobs— measured in terms of higher wages for wage workers and lower poverty for the self- employed—has been improving, thanks to strong economic growth in some countries and massive out-migration and workers’ remittances elsewhere. But there is absolutely no room for complacency. South Asia will add between 1 million and 1.2 million new entrants to the labor force every month for the next two decades. Absorbing them into the labor force at rising levels of labor productivity is the crux of the employment challenge. Meeting the employment challenges calls for a reform agenda that cuts across sectors. It includes enhancing access to reliable electricity supply for �rms in urban and rural settings, reducing corruption in dealings between �rms and the state, facilitating access to land, strengthening transport links between town and country, focusing on better nutrition in early childhood, improving the quality of education to equip workers with relevant skills for the world of work, and protecting workers in both informal and formal sectors rather than protecting jobs for the few in the formal sector of the economy. Recognizing that South Asia is the most conflict-affected of the major regions in the world, the book also includes a chapter on how the challenges of job creation are magni�ed in such environments as well as some priorities going forward for labor market policies and programs. More and Better Jobs will be of interest to policy makers and their advisers, not only in South Asia but also in other parts of the world where creating better jobs is a signi�cant challenge, as well as to the international development community and students of economics. ISBN 978-0-8213-8912-6 90000 9 780821 389126 SKU 18912