42114 The Knowledge Economy and Education and Training in South Asia The Knowledge Economy and Education and Training in South Asia Michelle Riboud Yevgeniya Savchenko Hong Tan Human Development Unit South Asia Region Copyright © 2007 The International Bank for Reconstruction and Development/ The World Bank 1818 H Street, N.W. Washington, D.C. 20433, USA All rights reserved Manufactured in the United States of America First printing September 2007 The findings, interpretations, and conclusions expressed in this book are entirely those of the authors and should not be attributed in any manner to the World Bank, to its affiliated organizations, or to members of its Board of Executive Directors or the countries they represent. The World Bank does not guarantee the accuracy of the data included in this publication and accepts no responsibil- ity for any consequence of their use. The boundaries, colors, denominations, and other information shown on any map in this volume do not imply on the part of the World Bank Group any judgment on the legal status of any territory or the endorsement or acceptance of such boundaries. The material in this publication is copyrighted. The World Bank encour- ages dissemination of its work and will normally grant permission to reproduce portions of the work promptly. Design and layout: James E. Quigley, World Bank Institute. Cover photo: Ray Witlin, World Bank. Contents Acknowledgments ...............................................................xi Executive Summary...........................................................xiii 1. Introduction......................................................................1 Objectives of the Regional Study......................................................................3 Data Sources........................................................................................................4 2. Trends in Education Attainment: Stocks and Flows.........7 Data Sources........................................................................................................8 Stock of Skills in the Population.......................................................................8 Flows of Human Capital: Investments in Educating New Generations....13 3. Returns to Investment in Education...............................23 Data Sources and Methodology......................................................................23 Wage Regressions for South Asia....................................................................27 Rates of Return to Education..........................................................................30 Gender Gap .......................................................................................................32 Changes over Time in Returns to Education................................................34 Differences in Returns to Education by Sector.............................................39 v vi The Knowledge Economy and Education and Training in South Asia 4. School to Work Transitions ............................................43 Definitions of Labor Force States....................................................................44 Unemployment Rates by Education...............................................................45 Youth Unemployment and School to Work Transitions..............................47 The Case of Sri Lanka.......................................................................................54 5. Postschool Training in the Labor Market .......................61 Surveys of Postschool training........................................................................61 Incidence of Postschool Training ...................................................................63 Trends in Postschool Training ........................................................................68 Postschool Training and Wages......................................................................79 6. In-Service Training by Employers ...................................83 Constraints to Investing in Training..............................................................90 Correlates of In-Service Training ...................................................................93 Productivity and Wage Outcomes of Training .............................................98 7. Concluding Remarks.....................................................107 Demand for and Supply of Formal Education............................................108 Unemployment and the School to Work Transition..................................110 Postschool Training .......................................................................................111 Implications ....................................................................................................113 Appendix 1: Hourly Wage Regressions, India...................115 Appendix 2: Hourly Wage Regressions, Pakistan.............119 Appendix 3: Hourly Wage Regressions, Sri Lanka............121 Appendix 4: Unemployment Rates by Level of Education and Age Cohort, Economically Active Population Aged 15­64, Selected South Asian Countries and Years ......123 Appendix 5: Unemployment Rates by Education and Years of Potential Work Experience, Economically Active Population Aged 15­64, Selected South Asian Countries and Years .....................................................127 Appendix 6: Percentage of the Population Trained by Field of Training and Average Duration of Training, India, 2004....................................................................131 Contents vii Appendix 7: Percentage Trained by Field of Training, Bangladesh 1995..........................................................133 Appendix 8: Percentage Trained and Duration of Training by Training Institution, India 2004..................135 Appendix 9: Number of People with Vocational Education by Training Field and Year, Pakistan............137 References.......................................................................139 Boxes 6.1 Investment Climate Surveys.....................................................................84 Figures 2.1 Educational Attainment, Selected South Asian Countries and Years...9 2.2 Educational Attainment in China and Malaysia, Selected Years.........10 2.3 Distribution of the Workforce by Level of Education in Manufacturing, by Region, 2000­4............................................................12 2.4 Proportion of Population Who Completed at Least Grade 5, Selected Asia Countries and Years.............................................................14 2.5 Proportion of Population Completing Grade 5 by Gender and Age, Selected Asian Countries and Years...........................................................16 2.6 Proportion of Population Who Completed at Least Grade 10, Selected Asian Countries and Year ............................................................18 2.7 Proportion of Population Who Completed at Least Grade 12, Selected Asian Countries and Year ............................................................19 2.8 Gross Enrollment Rates, Secondary Education, Selected Asian Countries, 2004.............................................................................................21 2.9 Gross Enrollment Rates, Tertiary Education, Selected Asian Countries, 2004.............................................................................................22 3.1 Predicted Log Hourly Wage by Gender and Level of Education, Selected South Asian Countries and Years ...............................................35 3.2 Returns to Education over Time by Level of Education and Gender, Selected South Asian Countries and Years ...............................................38 3.3 Rate of Return to Higher Secondary and Tertiary Education for Males by Sector, Selected South Asian Countries and Years..................40 4.1 Unemployment Rates by Education and Potential Labor Market Experience, Males Aged 15­64, Selected South Asian Countries and Year ........................................................................................................51 4.2 Unemployment Rates by Education and Potential Labor Market Experience, Males, Selected South Asian Countries and Years..............53 4.3 Time to First Job by Level of Schooling Completed, Sri Lanka, Pooled Sample 1996­2002 ..........................................................................57 5.1 Proportion of the Population That Received Vocational or Technical Traning by Age and Gender, Pakistan, 1993­4 and 2003­4...................70 5.2 Annual Gross Domestic Product Growth, Pakistan, 1993­2004........71 6.1 Incidence of In-Service Training in Selected South Asian Countries and Years.....................................................................................85 6.2 Incidence of Formal In-Service Training in Manufacturing, Regional Averages, Selected Years..............................................................86 6.3 Incidence of Formal In-Service Training in Manufacturing by Selected Countries, Selected Years.............................................................91 6.4 Rankings of Investment Climate Constraints, Rated Severe or Very Severe, Selected South Asian Countries and Years.........................92 6.5 Ranking of Reasons for Not Providing In-Service Training, Selected Developing Countries and Years.................................................93 6.6 Incidence of Formal Training by Exports and Research & Development, Selected South Asian Countries and Years......................95 Tables 2.1 Country Rankings by Educational Attainment and Net Enrollment Rates, Selected Asian Countries and Years ...............................................17 2.2 Educational Attainment, Selected Levels of Education, Selected Asian Countries............................................................................................20 3.1 Wage Regressions for South Asia ............................................................28 3.2 Rate of Return to Schooling by Education Level, Selected South Asian Countries and Years..........................................................................31 3.3 Rate of Return to Schooling by Education Level and Gender, Selected South Asian Countries and Years ...............................................33 4.1 Unemployment Rates by Level of Education, Economically Active Population Aged 15­64, Selected South Asian Countries and Years ....46 4.2 Unemployment Rates by Age and Gender, Economically Active Population Aged 15­64, Selected South Asian Countries and Years ....49 4.3 Unemployment Rates by Years of Potential Labor Market Experience and Gender, Economically Active Population Aged 15­64, Selected South Asian Countries and Years...............................................................49 4.3 Time to First Job with and without Postschool Training, Sri Lanka ..59 5.1 Percentage of the Population Aged 15­64 Obtaining Any Vocational Training by Level of Education of Gender, Selected South Asian Countries and Year.......................................................................................64 5.2 Percentage of the Workforce Obtaining Vocational Training by Occupational Category, Selected South Asian Countries and Years.....66 5.3 Percentage of the Workforce Obtaining Vocational Training by Sector of Employment, Selected South Asian Countries and Years ......67 Contents ix 5.4 Percentage of the Workforce Obtaining Training by Source of Vocational Training and Gender, Bangladesh 1995 and India 2004 .....68 5.5 Percentage of the Population Aged 15­64 That Received Vocational Training by Level of Education and Gender, Pakistan, Selected Years..72 5.6 Composition of Vocational Training Received by the Population Aged 15­64, Pakistan, Selected Years........................................................74 5.7 Training Trends by Education and Gender, Sri Lanka, Selected Years...............................................................................................................77 5.8 Percentage of the Population Trained by Age Group and Education, Sri Lanka, Selected Years.............................................................................78 5.9 Postschool Training and Wages, Selected South Asian Countries and Years...............................................................................................................81 6.1 Share of Workers Trained by Skill Group, Selected South Asian Countries and Years.....................................................................................87 6.2 Percentage of Firms Providing Training by Source, Selected South Asian Countries and Year............................................................................89 6.3 Probits of Any Formal In-Service Training, Selected South Asian Countries.......................................................................................................97 6.4 Training and Productivity Results, Simple Indicator of Any Formal In-Service Training, Selected South Asian Countries .............100 6.5 Training and Productivity Results, Share of Workers Training and In-House Versus External Sources of Training, Selected South Asian Countries.....................................................................................................101 6.6 Training and Wages Results, Selected South Asian Countries..........104 6.7 Training and Wages Results, Training by Source and Share of Workers Trained, Selected South Asian Countries................................105 Acknowledgments Michelle Riboud, Yevgeniya Savchenko, and Hong Tan pre- pared this regional report. We gratefully acknowledge fund- ing support from the Human Development Department of the World Bank's South Asia Region, and supplemental fund- ing from the South Asia Regional Studies Program and the Knowledge and Skills Trust Fund of the U.K. Department for International Development to organize the regional confer- ence. The report benefited greatly from feedback provided by participants at the Regional Conference on the Knowledge Economy and Education and Training Policy in South Asia hosted by the government of India in September 2006, and by Gordon Betcherman, Ejaz Ghani, Harry Patrinos, and other World Bank staff who took part in a June 2007 report review meeting. The report reflects their detailed comments and suggestions. Finally, this report would not have been possible without Alice Faintich's careful editing and James Quigley's layout and design expertise. xi xii The Knowledge Economy and Education and Training in South Asia The views expressed are those of the World Bank team and do not necessarily reflect the views of governments in the South Asia region or of the World Bank, its executive directors, or the countries they represent. Please send any comments to Michelle Riboud at mriboud@worldbank.org or Hong Tan at htan@worldbank.org. Executive Summary How education and training systems respond to the sweeping changes brought about by globalization and the knowledge economy can have far-reaching implications for developing countries in terms of sustainability of growth, competitive- ness, job creation, and poverty reduction. This issue is espe- cially pertinent to the countries of South Asia, which are cur- rently growing at a rapid pace and are gradually becoming more integrated into the world economy. Despite this, little systematic research has been undertaken on the progress the region has made in relation to skills development (broadly defined to include both education and training) and how skills affect labor market outcomes. Even less is known about how the pace of progress differs across countries in South Asia; whether the supply response for skills in the region is adequate given global trends in trade, knowledge generation, and technology diffusion; and what their competitor coun- tries in other regions are doing. Thisregionalstudyisafirstattempttoaddresstheseques- tions. Its main objective is to document and compare trends in education and training in the countries of South Asia, as xiii xiv The Knowledge Economy and Education and Training in South Asia well as the associated changes in earnings and employment. It draws upon household, labor force, and firm-level surveys from 1990 to the most recent year available. The analysis fo- cuses on Bangladesh, India, Pakistan, and Sri Lanka (coun- tries with well-developed surveys), with some references to Bhutan, the Maldives, and Nepal, along with comparisons with countries in East Asia and with other regions. The analysis in each chapter provides many policy-rele- vant insights, but the following conclusions stand out: · Despite ongoing progress and clear commitment to education in all the countries of South Asia, other than the Maldives, none of the countries is currently up- grading the skills of its population at a speed that will allow it to catch up with East Asia and the rest of the world over the medium term. Some indications even suggest that the gaps relative to some East Asian com- petitor countries may be widening rather than closing. · Progressacrosscountrieshasbeenuneven.Asidefrom Sri Lanka, which is an outlier in the region given its early achievement of universal primary education, in the near future, those countries that started with the lowest levels of education in 1990 seem likely to catch up with the front-runners with respect to the achieve- ment of universal primary education. · Progressintermsofgenderequalityhasalsobeenun- equal. In recent decades, the gender gap has dimin- ished substantially in all the countries and has disap- peared in some of them at the level of primary educa- tion. The challenge now is to repeat this achievement at levels beyond primary education. Executive Summary xv · Thesupplyofskillsisclearlylaggingbehinddemand. Returns to higher secondary and tertiary-level educa- tion have remained high, and even increased relative to returns to lower levels of education, despite sizable investments by governments in the region, indicating that education policies and programs have not yet ful- ly responded to the high and rising demand for skills. This phenomenon is particularly striking in India. · Inrecentdecades,SouthAsiancountrieshavefocused their efforts on promoting elementary education. Even though pockets of excellence can be observed at oth- er levels of education and differences are noticeable across countries, secondary and higher education, vo- cational education and training, and in-service train- ing have not yet received the same attention from the public sector, and most of the expansion that has taken place in these areas is due to the private sector. Post- school training is a particularly neglected area despite evidence of large, positive impacts of such training on wages. For example, the incidence of training in manu- facturing is among the lowest in the world. · Thehighunemploymentrateamonguniversitygradu- ates should not detract policy makers from investing in education. While the more educated initially have higher rates of open unemployment because of their more intensive search for a good job match, their un- employment rates are subsequently lower than those of other groups as those with more education gain labor market experience. · Availabledataoneducationandtrainingarerelative- ly robust and amenable to rigorous analysis. They are, however, limited, and substantial improvements in xvi The Knowledge Economy and Education and Training in South Asia survey design, systematic data collection, and analy- sis would be required to allow governments to better monitor skills requirements and labor market out- comes and to design appropriate education and train- ing policies. Some indications suggest that South Asian countries are becoming aware of the pressures on skills resulting from glo- balization. Both employers and the media increasingly flag the shortage of skills as a critical issue, especially in the re- gion's rapidly growing countries. Countries are taking vari- ous initiatives to accelerate the achievement of universal primary education, develop strategies for reforming higher education, and make the vocational training sector more re- sponsive to the skill needs of the private sector. What is not yet clear, however, is whether governments are as yet fully aware of the crucial importance of education and training policies for sustaining the current high rates of growth in the region, and whether such policies feature prominently in re- lation to other national priorities on governments' agendas. The foregoing findings and the analyses detailed in this re- gional study suggest that this is not yet the case. 1 Introduction Globalization and the knowledge economy pose numerous challenges as well as opportunities for developing countries, not least in the area of skills development. Expanding trade and the globalization of production and capital create pres- sures for economies to restructure, making it imperative to retrain those made redundant in declining industries and to upgrade the skills of those employed in new industries. In addition, the increased global flow of information made pos- sible by new information technologies creates demand for higher-level cognitive skills and for continuous learning over the work life, as the skills people acquire in school and in the workplace become obsolete more quickly and they need new and more complex skills to respond to accelerating techno- logical change. How education and training systems respond to these sweeping changes and the challenges they pose will have far-reaching implications for the economic growth and competitiveness of South Asian countries and for income 1 2 The Knowledge Economy and Education and Training in South Asia growth, employment, job creation, and poverty reduction. Some effects of globalization and the knowledge econ- omy on the growing relative demand for skills are well known. Economists have documented diverging trend changes in earnings distributions by level of education for many developing countries and regions in the late 1980s and 1990s, paralleling similar trends in countries of the Organi- sation for Economic Co-operation and Development that started in the 1970s (Berman, Bound, and Machin 1998). Some have attributed this global phenomenon to skill-bi- ased technological change whereby the diffusion of skill- intensive, advanced technologies developed in countries of the Organisation for Economic Co-operation and Devel- opment generates a corresponding, but lagged, pattern of change in relative skills demand in developing countries. How important an influence skill-biased technology has on relative pay by skill level will also depend on supply-side changes in skills and on the speed of globalization. Edu- cation and training policies, as well as policies regarding trade liberalization and market orientation, can offset de- mand shifts, and thereby mitigate the effects of skill-biased technology on relative pay by skill level. Policy makers in the South Asia region are already grap- pling with the challenges of reforming national education and training systems. For example, the release of a report on the knowledge economy in India (Dahlman and Utz 2005) has sparked policy interest in how best to reposition edu- cation and workforce skills to take advantage of the oppor- tunities afforded by the knowledge economy. Pakistan, rec- ognizing the imperative of expanding access to postschool vocational education and training (VET), has established the National Vocation and Technical Education Commission, an Introduction 3 apex training body, to develop and implement a scaled up national training strategy for the workforce. The World Bank is also helping the governments of Bangladesh, India, and Sri Lanka with education, vocational training, and labor market sector studies and/or projects. Objectives of the Regional Study This volume seeks to complement and inform these ongoing, but still nascent, initiatives through a cross-country study of education and training in the South Asia region. The focus is on Bangladesh, India, Pakistan, and Sri Lanka, for which data on education and training are available for large samples of households and firms from several different surveys. The objectives of this regional study are to · identifyandassembleavailablehouseholdandfirm- level survey data for the four countries from the 1990 to the most recent year for which data are available; · documentandcomparetrendsintheeducationand training of the workforce in these four countries and associated changes in the earnings of groups that differ in terms of level of education and demographics; · ascertainwhatkindsofeconomicanalysescanbedone with existing data on the life cycle choices individuals, families, and employers make about education, pre- employment VET, and in-service training and the out- comes of such human capital investments on school to work transitions, employment, earnings, and produc- tivity growth. 4 The Knowledge Economy and Education and Training in South Asia The findings reported here suggest that the available data on education and training, while limited, are relative- ly robust and amenable to more technically rigorous analy- sis. Improvements over time in survey design and sustained collection of better data on education and training should allow governments in the region to better monitor the skill requirements necessitated by globalization and the knowl- edge economy and to design and implement education and training policies that better address those skill needs. Data Sources The regional study relies principally on two main data sourc- es, namely: · Household surveys and labor force surveys (LFSs). Each of the four South Asian countries has household sur- veys and LFSs for several points in time. All contain information on educational attainment, demographic attributes, employment, wages and salaries or incomes, industry of employment, and region of residence. The surveys do not cover postschool technical and voca- tional training as well. LFSs are available annually in Sri Lanka and periodically in Pakistan, while house- hold surveys with information on education, employ- ment, and earnings are available for selected years in India (selected rounds of the national sample survey or NSS) and Pakistan (integrated household survey or PIHS). For Bangladesh, the household income and ex- penditure survey (BHIES) is available, but for only two rounds in 2000 and 2004. Introduction 5 · Investment climate surveys (ICSs). For each of the four South Asian countries, cross-sectional information on enterprise-based training (by in-company programs and by external public and private sector training pro- viders) is available from firm-level surveys of manu- facturing establishments conducted between 2002 and 2005. For these countries, the cross-sectional relation- ships between education, training, and outcomes on firm productivity and wages can be investigated, as can simple hypotheses about the demand-side roles of trade, investment, foreign ownership, and skill-biased technological change. 2 Trends in Education Attainment: Stocks and Flows This chapter begins by looking at the evolution of educational attainment in South Asia over a period of two to four decades depending on the country. Following a review of data sources, it looks at the distribution of educational attainment, or the stock of human capital, in the population at different points in time. The stock of human capital at a given time may be characterized by the percentage of the total population aged 15 years and older that has attained the following four levels of education: is illiterate (no education), has completed pri- mary schooling, has completed secondary schooling, or has achieved a level of education above secondary. In all cases, this grouping refers to the highest level of education attained.1 1. For India, the NSS defines education levels as follows: illiterate--not literate, literate through attending nonformal education centers or alternative education centers or by means of the total literacy campaign, literate but below primary; primary--primary or middle school completed; secondary--secondary or higher secondary completed; above secondary--university graduate and above. For Pakistan, the PIHS education categories are as follows: illiterate --not literate, completed grades 1­4 (less than primary); primary--completed grades 5­9 (primary or middle school completed); 7 8 The Knowledge Economy and Education and Training in South Asia The chapter then turns to the speed at which each country is upgrading the skills of its population. To study changes, or flows, in the stock of human capital over time, we compare changes in the distribution of educational achievement across cohorts of individuals born at different times. Data Sources The data for this exercise are based on household surveys. For India and Pakistan, where relatively long time series data are available, we use several rounds of India's NSS and Pakistan's PIHSsforseveralroughlycomparableyears.Weusesecondary data from Barro and Lee (2000) for Bangladesh and Sri Lanka and for two East Asian comparator countries (China and Malaysia), either because we do not have access to household survey data or because such data are not available for compa- rable periods. The Barro and Lee classification of education levelsisbasedoncriteriaadoptedbytheInternationalStandard Classification of Education (United Nations Educational, Scientific, and Cultural Organization 1976). Stock of Skills in the Population The stock of skills at a given point in time reflects past in- vestments in education. When the mean number of years of schooling in a country is low, the distribution of educational attainment resembles a pyramid. The base, which corre- sponds to the fraction of the population with no education secondary--completed grades 10­13 (secondary or higher secondary); and above secondary--bachelor's degree and above. Trends in Education Attainment: Stocks and Flows 9 Figure 2.1 Educational Attainment, Selected South Asian Countries and Years (percentage of the population aged 15 and older) Figure 2.1 Educational Attainment, Selected South AsianBangladeshand Years Countries Sri Lanka 3 3 2000 14 33 2000 50 34 50 14 2.1 2.5 1995 14 1995 48 34 29 16 54 1.2 43 1.7 1985 43 1985 14.4 12 21 63 0.3 1960 26.7 45.8 27 India Pakistan 5.5 3.3 2004 16 32 2001 12 23 47 61 4.7 2000 16 27 2.6 53 1994 12 20 4.7 65 1994 12 25 58 1.8 2.9 1985 8.8 1984 8.5 17.2 24.5 72 64 Illiterate Primary Secondary Above secondary Source: Household surveys. Barro and Lee 2000. or with less than a primary education, is relatively wide, and the middle and top sections taper off to reflect the smaller shares of the population with higher levels of education. This pattern characterizes Bangladesh, India, and Pakistan since the mid-1980s (figure 2.1). Over time, as these coun- 10 The Knowledge Economy and Education and Training in South Asia Figure 2.2 Educational Attainment in China and Malaysia, Selected Years (percentage of the population aged 15 and older) Malaysia China Figure 2.2 Educational Attainment in 2000 China and5Malaysia, Selected 3 36 Years 42 2000 45 34 16 18 2 1 1980 25 34 46 1980 31 27 34 1.5 0.9 1960 10.1 38.6 1975 31.4 27.5 60 40 Illiterate Primary Secondary Above secondary Source: Barro and Lee 2000. tries have upgraded the skills of the population by focusing on the lower levels of education, the base has narrowed and the middle sections have become wider. Nevertheless, in both Bangladesh and India, about half the population aged 15 and older is still illiterate, and in Pakistan the figure is even higher. When countries pursue their investments in education to the point where more adults have primary education than are illiterate, the education distribution takes on a diamond shape. This has been the case in Sri Lanka since the early 1960s. The middle sections of the distribution have continued to grow since that time, and by 2000, more than 80 percent of the population had either completed primary education (34 percent) or secondary education (50 percent). Educational Trends in Education Attainment: Stocks and Flows 11 progress has not, however, been such that the distribution of education resembles an inverted pyramid shape. When looking at how South Asia compares with East Asian countries such as China or Malaysia, which have en- joyed longer periods of economic and total factor produc- tivity growth, figure 2.2 shows that the region is far behind East Asia. The proportion of the population that was illit- erate in India in 2004 was similar to that observed around 1970 in China or around 1960 in Malaysia. The fraction of the population that had completed secondary education in India in 2004 (16 percent) is half of the figure that had pre- vailed in China in 1975. Bangladesh and Pakistan lag even further behind. It is only at the level of tertiary education that the South Asian countries resemble their East Asian coun- terparts, with India actually having a slight advantage over China and being roughly on a par with Malaysia. However, when taking the population as a whole into account, South Asia lags behind East Asia by about 30 years. A comparison with other parts of the world also shows that the distribution of educational attainment in South Asia today is similar to that observed in Latin American countries in the 1960s (de Ferranti and others 2003). Only Sri Lanka, a clear outlier, did much better, but its comparative advantage has been gradu- ally eroded over time. Investment climate surveys (ICSs), which were conduct- ed worldwide between 2000 and 2005, provide another use- ful source of information for comparing the stocks of human capital across regions. These are broadly comparable firm- level surveys that the World Bank has carried out in the man- ufacturing sectors of more than 40 developing countries to obtain employers' assessments of the business environment in the country. They include indicators of governance, of the 12 The Knowledge Economy and Education and Training in South Asia predictability of economic policy, of the judicial system, of access to finance, and of general constraints to business op- erations.2 In addition to these indicators, the ICSs elicited in- formation on the educational distribution of the workforce. Figure 2.3 shows the distributions separately for six regions for which country ICS samples are weighted using the firm Figure 2.3 Distribution of the Workforce by Level of Education in Figure 2.3 Distribution of the Workforce Manufacturing, by Region, 2000­4 by Level of Education in 100 Manufacturing, by Region, ce 90 2000­4 80 workfor 70 the 60 of 50 40 centage 30 Per 20 10 0 Europe and East Asia Latin America Sub-Saharan Middle East South Asia Central Asia and the Pacific and the Africa and North Caribbean Africa Some primary education Some middle and/or secondary school education Some higher secondary education Some university education Source: ICSs. 2. To ensure the comparability of ICSs across countries, a sampling frame is used that is based on the distribution of private firms in each country by sector, size, number of employees, and location. Each ICS includes information on firm size (number of em- ployees, extent of sales and assets); years in operation; debt and growth performance; sources of finance; and a mix of qualitative and quantitative indicators of the business environment. Trends in Education Attainment: Stocks and Flows 13 size distribution of India as the norm.3 The figure suggests that South Asia's stock of human capital differs little from that of the Middle East and North Africa region and lags be- hind that of most other regions. Flows of Human Capital: Investments in Educating New Generations South Asia's stock of human capital is clearly still low com- pared with that in other parts of the world. However, the evi- dence indicates continuous skill upgrading in the region over time. How rapid has this progress been? Has it been differ- ent across countries and is South Asia likely to catch up with other regions? Trends in enrollment rates over time could answer these questions, but the limited availability of household surveys at different points in time for all countries in the region makes the use of enrollment rates to compare trends over time dif- ficult. To overcome this difficulty, we use data from the most recently available survey and look at the educational attain- ment of age cohorts of individuals born at different times. For example, individuals aged 50­59 years in 2000 were born 3. The countries in South Asia are India (2002), 1,824 firms; Pakistan (2002), 914 firms; and Sri Lanka (2004), 451 firms. The countries that make up the other comparator regions are as follows: Sub-Saharan Africa, 2,387 firms, 11 countries--Eritrea (2002), Ethiopia (2002), Kenya (2003), Mali (2003), Mozambique (2001), Nigeria (2001), Senegal (2003), South Africa (2003), Tanzania (2003), Uganda (2003), and Zambia (2002); East Asia and the Pacific, 3,083 firms, 4 countries--Cambodia (2003), China (2002), Indonesia (2003), and the Philippines (2003); Europe and Central Asia, 280 firms from Kosovo (2003), Montenegro (2003), and Serbia (2003); Latin America and the Caribbean, 5,112 firms, 8 countries--Bolivia (2000), Brazil (2003), Ecuador (2003), El Salvador (2003), Guatemala (2003), Honduras (2003), Nicaragua (2003), and Peru (2002); and the Middle East and North Africa, 2,889 firms, 5 countries-- Algeria (2002), Egypt (2004), Morocco (2004), Oman (2003), and the Syrian Arab Republic (2003). 14 The Knowledge Economy and Education and Training in South Asia in the 1940s, those aged 40­49 were born in the 1950s, and so on. With this perspective, we can identify changes in educa- tional investments across different generations and compare the speed at which the human capital stock was upgraded over time. As this only requires using the most recent survey, we were able to add information on additional countries in South Asia, namely, Bhutan, the Maldives, and Nepal. For purposes of comparison across regions, we also add similar data on Malaysia, a rapidly growing East Asian country. Figure 2.4 shows the share of the population completing at least grade 5 in the countries under consideration. It de- picts changes in primary school achievement across differ- Figure 2.4 Proportion of the Population Who Completed at Least Grade Figure 2.4 Proportion of Population Who 5, Selected Asian Countries and Year 100 Completed at Least Grade 5, Selected Asia Countries and 96.3 Bangladesh, 2004 95.4 90 Years Bhutan, 2003 India, 2004 80 Malaysia, 2004 77.2 Maldives, 2004 73.8 Nepal, 2002­3 70 65.1 Pakistan, 2001­2 Sri Lanka, 1999­2000 60 cent 56.9 Per 50 48.0 40 30 20 10 0 50­59 40­49 30­39 20­29 15­19 Age group Source: Leopold R. Sarr and authors' calculations for Bangladesh, India, and Malaysia. Trends in Education Attainment: Stocks and Flows 15 ent generations ranging from those now in their 50s to those aged 15­19 at the time of the surveys. Once again, Sri Lanka is the outlier in the South Asia region: more than 70 percent of Sri Lankans born in the late 1940s had completed at least five years of education, and continuous progress during the next 40 years led to practically universal primary education. For all the other South Asian countries, the starting point was much lower, ranging from 5 percent for Bhutan to 35 percent for In- dia. Countries that started with the lowest educational level improved at a more rapid pace. The most spectacular changes took place in Bhutan and in the Maldives. Over a 20-year peri- od, Bhutan moved from a situation where only a tiny propor- tion of children went to school to a situation where almost half of children spend at least five years in school, and the Maldives was able to increase access to primary education to practically 90 percent of children and catch up with Sri Lanka. Nepal also stands out, with a 4.5-fold increase in the proportion of chil- dren completing at least five years of schooling. Bangladesh, India, and Pakistan made slower progress. During four decades, those three countries increased the proportion of children who completed at least a primary education about 2.5-fold. India has continued to fare better than Bangladesh, which in turn has fared better than Paki- stan; however, these differences are not extremely large, and may be overstated, as the data for Bangladesh and India refer to 2004, while the data for Pakistan refer to 2001, and the country's enrollment rates have increased dramatically. At this level of education, only Sri Lanka can be compared with Malaysia: both countries have the same starting and ending points, although Malaysia's progress toward universal prima- ry education has been faster. 16 The Knowledge Economy and Education and Training in South Asia In relation to the pace of progress for boys and girls, clearly this has not been similar. In all the countries, includ- ing Malaysia (figure 2.5), the proportion of boys completing primary education 40 years ago was significantly higher than that of girls. In some of them, such as Bhutan, Nepal, and Pakistan, only a tiny fraction of women had access to educa- tion. Forty years later, the gender gap has narrowed every- where and has disappeared in Bangladesh, the Maldives, and Sri Lanka. Girls have clearly benefited the most from progress Figure 2.5 Proportion of the Population Completing Grade 5 by Gender Figure 2.5 Proportion of Population and Age, Selected Asian Countries and Years Completing Grade 5 by Gen- 100 der and Age, Selected Asian Countries and Years 80 60 cent Per 40 20 0 Bangladesh Bhutan India Malaysia Maldives Nepal Pakistan Sri Lanka Males, aged 15­19 Females, aged 15­19 Males, aged 50­59 Females, aged 50­59 Source: Household surveys. during this time. Perhaps the most spectacular changes have Trends in Education Attainment: Stocks and Flows 17 Table 2.1 Country Rankings by Educational Attainment and Net Enrollment Rates, Selected Asian Countries and Years Share of the population aged 15­19 with at least Net enrollment rates Country five years of schooling Country in primary education Maldives 95.4 Maldives 87.7 Sri Lanka 95.4 Sri Lanka 77.0 India 77.2 Nepal 72.4 Bangladesh 73.8 Bhutan 69.4 Nepal 65.1 Bangladesh 66.5 Pakistan 56.9 India 62.5 Bhutan 48.0 Pakistan 52.0 Source: Authors' calculations based on household surveys. taken place in Bangladesh, where the proportion of women with a primary education is now larger than that of men, and in the Maldives, which achieved universal primary education for both boys and girls over a relatively short period. What will be the situation in these South Asian countries by 2015, when countries are supposed to have met the Edu- cation for All goal of universal primary education? If current trends persist, most likely Bangladesh and India will have moved closer to, although not yet have reached, that goal, and the region's other countries would still have a long way to go. However, the most recent available data on net enroll- ments suggest that all the countries in the region have ac- celerated their investments in primary education, and that those that made the least progress in past decades are now trying hard to catch up with the front-runners. As table 2.1 shows, the ranking of countries by educational attainment differs from the ranking of countries by net enrollment rates during similar periods. 18 The Knowledge Economy and Education and Training in South Asia A number of points stand out when we turn to figures 2.6 and 2.7 and focus on secondary education. First, efforts to upgrade skills beyond primary education have been steady in the region. Trend lines in relation to the attainment of at least 10 years of school are broadly parallel, with the excep- tion of the Maldives and Sri Lanka, which have experienced faster progress for the youngest generation. The Maldives in particular is now approaching the level of attainment in Sri Lanka, which has the highest proportion of children attained grade 10. Second, with respect to the achievement of 12 years Figure 2.6 Proportion of the Population Who Completed at Least Grade 10, Selected Asian Countries and Year 80 Figure 2.6 Proportion of Population Who Completed at Least Grade Bangladesh, 2004 Bhutan, 2003 70 10, Selected Asian Countries India, 2004 and Year 67.0 Malaysia, 2004 60 Maldives, 2004 Nepal, 2002­3 Pakistan, 2001­2 50 Sri Lanka, 1999­2000 48.4 cent 40 Per 35.5 30 29.7 26.3 23.9 20 23.2 13.7 10 0 50­59 40­49 30­39 20­29 Age group Source: Leopold R. Sarr and authors' calculations for Bangladesh, India, and Malaysia. Trends in Education Attainment: Stocks and Flows 19 Figure 2.7 Proportion of the Population Who Completed at Least Grade 12, Selected Asian Countries and Year Figure 2.7 Proportion of Population Who 35 Completed at Least Grade Figure 2.7 Proportion of Population Who Bangladesh, 2004 12, Selected Asian Countries Completed at Least Grade Bhutan, 2003 30 and Year 12, Selected Asian Countries India, 2004 and Year 28.6 Malaysia, 2004 Maldives, 2004 25 Nepal, 2002­3 23.2 Pakistan, 2001­2 Sri Lanka, 1999­2000 20 19.9 cent 18.0 Per15 12.9 12.2 10 6.3 5 4.1 0 50­59 40­49 30­39 20­29 Age group Source: Leopold R. Sarr and authors' calculations for Bangladesh, India, and Malaysia. of schooling, Sri Lanka no longer appears as an outlier.4 It has concentrated its efforts on basic education and focused much less on levels of schooling beyond that level. While 48 percent of children from the youngest generation achieved at least 10 years of schooling,5 the proportion of those with 12 years of schooling drops to less than 20 percent, and India now performs almost as well. Recent data on secondary-lev- 4. Estimates for Sri Lanka are lower than and not fully consistent with those provided by Barro and Lee (2000). This discrepancy may be because Barro and Lee do not measure completion of a full cycle of education, but only "some" primary or second- ary education. LFSs also give lower estimates. 5. The numbers for secondary school attainment are significantly lower than those Bar- ro and Lee (2000) report. 20 The Knowledge Economy and Education and Training in South Asia Table 2.2 Educational Attainment, Selected Levels of Education, Selected Asian Countries Category Bangladesh Bhutan India Maldives Nepal Pakistan Sri Lanka Malaysia Share of population aged 15­19 who attained at least grade 5 Male 70.0 55.1 81.9 95.1 74.5 67.5 95.1 96.3 Female 78.4 41.4 71.6 95.7 56.3 46.7 95.8 96.4 All 73.8 48.0 77.2 95.4 65.1 56.9 95.4 96.3 Share of population aged 20­29 who attained at least grade 10 Male 28.9 19.3 35.8 34.4 33.5 33.7 45.7 62.9 Female 18.5 9.5 23.6 37.1 17.2 19.5 51.1 71.3 All 23.2 13.7 29.7 35.5 23.9 26.3 48.4 67.0 Share of population aged 20­29 who attained at least grade 12 Male 16.4 5.7 21.3 22.2 9.2 16.0 16.8 25.2 Female 8.7 2.9 14.6 24.6 4.2 10.0 22.9 32.2 All 12.2 4.1 18.0 23.2 6.3 12.9 19.9 28.6 Source: Authors' calculations based on household surveys. el net enrollment confirm that secondary school attendance is increasing faster in India than in Sri Lanka.6 Bangladesh, while still behind, is also rapidly catching up with Sri Lanka. The final point that emerges from figures 2.6 and 2.7 is that South Asia is unlikely to catch up with East Asia in terms of education, at least in the medium term. With the excep- tion of the Maldives, none of the countries in the region has adopted a path that will enable it to reach the education lev- els Malaysia has attained in the near future if those coun- tries continue to invest in their human capital at the same rate. Indeed, differences between South Asian countries and Malaysia are larger for younger than for older generations, suggesting that the gap is widening over time. Note that In- 6. The net enrollment rate at the secondary level is equal to 42.2. in Bangladesh (2004), 49.4 in India (2004), and 45.6 in Sri Lanka. Trends in Education Attainment: Stocks and Flows 21 dia, Malaysia, and Sri Lanka shared almost the same starting point for grade 12 completion (those aged 50­59 in figure 2.7), but their trends diverged over time. Table 2.2 provides similar information disaggregated by gender for the youngest cohorts completing at least grades 5, 10, and 12. In countries that have made the most progress in education, such as the Maldives, Sri Lanka, and the compara- tor country Malaysia, the proportion of girls achieving a giv- en number of years of schooling is larger than that of boys. The reverse -- higher levels of attainment by boys than by girls -- is generally true for the other South Asian countries at all three grade levels, except for Bangladesh at the level of primary education. Figure 2.8 Gross Enrollment Rates, Secondary Education, Selected Asian Countries, 2004 90 80 Figure 2.8 Gross Enrollment Rates, Sec- 70 ondary Education, Selected Asian Countries, 2004 60 population 50 the of 40 centage 30 Per 20 10 0 Afghanistan Pakistan Nepal Bangladesh India China Maldives Malaysia Sri Lanka Source: World Bank 2006b. 22 The Knowledge Economy and Education and Training in South Asia Figure 2.9 Gross Enrollment Rates, Tertiary Education, Selected Asian Countries, 2004 35 Figure 2.9 Gross Enrollment Rates, 30 Tertiary Education, Selected Asian Countries, 2004 25 population 20 the of 15 centage Per10 5 0 Maldives Afghanistan Pakistan Nepal Bangladesh Sri Lanka India China Malaysia Source: World Bank 2006b. Despite ongoing progress, the speed at which the South Asian countries are currently upgrading their populations' skills will clearly not allow them to catch up quickly with other parts of the world, especially East Asia. Comparisons of enrollment rates at the secondary and tertiary levels across countries confirm this conclusion (figures 2.8 and 2.9). Even though the proportion of the population with higher educa- tion was similar in China, India, and Malaysia, differences in enrollment rates suggest that the two regions are not making similar efforts in terms of the flow of human capital. South Asia is clearly lagging behind East Asia, with the implication that levels of attainment of tertiary education are likely to di- verge further over time. 3 Returns to Investment in Education The previous chapter documented the current status of hu- man capital accumulation in South Asia both in terms of the distribution of skills in the population at a given time and in terms of changes in educational investments over time. This chapter turns to the use of these skills by the labor market and their profitability or rate of return. Data Sources and Methodology To calculate rates of returns to education, we use house- hold surveys for Bangladesh (BHIESs), India (NSSs), and Pakistan (PIHSs) and LFSs for Sri Lanka. Surveys at differ- ent points in time are available that cover about one decade for Pakistan and Sri Lanka and two decades for India. For Bangladesh, BHIESs were only available for 2000 and 2004, so no comparisons of long-term trends in schooling returns 23 24 The Knowledge Economy and Education and Training in South Asia were possible. We focus on the sample of males and females aged 15­64 who work for salaries or wages. We exclude the self-employed and those for whom compensation for work is not reported.1 We use information on the sample's wages, salaries, and cash and in-kind payments for their primary occupation or employment to calculate hourly wages, adjust- ing for the number of hours worked last week. Following the standard methodology popularized by Mincer (1974), we estimate the rate of return to education by regressing the logarithm of wages on years of schooling, a measure of years of potential work experience, and a set of other control variables. In this semi-log wage model specifi- cation, the coefficient of schooling is interpreted as the rate of return to an additional year of schooling and the coeffi- cient of potential experience measures returns to postschool investments in on-the-job training. Given our interest in in- vestigating the potentially different rates of return by level of schooling, we estimate the following expanded specification of the Mincer wage model: log(hourlywagei) = fn(EDUCi, EXPi, OTHERi, LOCATION) The dependent variable, the logarithm of hourly wage, is related to the following sets of explanatory variables: · EDUC consists of five 0,1 indicator variables (six in the case of India) for levels of schooling completed: liter- ate, below primary = 1 if the person is literate but has 1. As this section focuses on the returns to human capital, we exclude those earners for whom no compensation is reported, as well as the self-employed, whose incomes include a profit component that reflects returns to their (unmeasured) capital equip- ment. Returns to Investment in Education 25 not completed primary education; primary = 1 if pri- mary education is the highest level of education com- pleted; middle = 1 if middle school is the highest level of education completed; secondary and higher second- ary = 1 if secondary or higher secondary education is the highest level of education completed; tertiary = 1 if any level of tertiary education has been completed; technical education dummy (India only) = 1 if any tech- nical education has been completed. In the regression analysis, the illiterate group is omitted. · EXPmeasuresyearsofpotentialexperience,measuredas age ­ education ­ 5 (in the case of Pakistan) or 6 (in the case of Bangladesh, India, and Sri Lanka), and its qua- dratic EXP2 or years of potential experience squared. · OTHER is a vector of individual attributes, including male = 1 if the respondent is male; SCST = 1 if the per- son belongs to a scheduled caste or scheduled tribe (India only); regular worker = 1 for those who receive monthly or annual salaries; and regular worker = 0 for casual workers, that is, those who are paid on a daily basis. · LOCATION controls for place, where urban = 1 if the household lives in an urban area and = 0 if it lives in a rural area. The Sri Lankan LFS distinguishes between urban, rural, and estate location, and to reflect this pos- sibility, both urban and rural dummies are used, with estate as the omitted category. The underlying human capital model establishes a link between investments in different levels of education, as prox- ied by foregone earnings while in school, and the value that the labor market attributes to skills thus acquired. The es- 26 The Knowledge Economy and Education and Training in South Asia timated coefficients on the different educational categories allow us to calculate what the corresponding annualized pri- vate rates of return are to completing that level of education. Several caveats are important. First, the analysis does not capture the full social value of human capital for a country, as it does not measure nonmarket benefits and possible ex- ternalities. Second, it does not take into account either gov- ernment spending on education or direct outlays by families. What the analysis provides are estimates of private rates of return. The omission of families' outlays is unlikely to have an important effect, as international experience indicates that foregone earnings represent the bulk of private costs. In ad- dition, this effect may be offset by the omission, on the ben- efit side, of earnings from secondary jobs. Third, investments in education are crudely measured by the number of years that reaching a given level of education normally takes, for example, five years for completing primary education. The data do not allow us to take class repetition or the quality of education into account. The estimated schooling coefficients may also be biased by the endogenous choice of education. Selectivity bias aris- es because some of the individual and household attributes that shape schooling choices, such as ability, motivation and social background are unobserved, and these are correlated with wage outcomes. Even though we are familiar with selec- tivity biases and the literature on correcting for such biases using multi-equation and panel data models (for example, Barnow, Cain, and Goldberger 1981; Heckman 1979; Patri- nos, Ridao-Cano, and Sakellariou 2006), we do not explic- itly address these issues for several reasons. First, identifying valid instrumental variables (correlated with the choice vari- able but not with the outcomes of interest) that are common Returns to Investment in Education 27 across all the different surveys is difficult; using different in- strumental variables in each country would limit the compa- rability of results across countries. Second, the literature sug- gests that while selectivity corrections often reduce estimated rates of return to education, corrected estimates do not sub- stantially change the key policy findings based on simpler models. Finally, comparisons of estimated returns over time across countries are still valid if, as is plausible, selectivity (and ability) biases do not change systematically over time. Despite those caveats and the potential for selection bias, our estimates of private rates of return can provide useful first insights into the interaction between the demand for and supply of skills and changes over time in the balance of supply and demand across the different countries. Wage Regressions for South Asia Table 3.1 reports the wage regressions estimated for each of the four South Asia countries using the most recent data available. From empirical evidence based on numerous stud- ies in many countries that covers many different periods of time, we would expect to find that earnings increased with level of educational attainment and, for any given level of ed- ucation, that earnings increased with the number of years of labor market experience, although at a decreasing rate. Our results are fully consistent with these expectations. Investing in formal education is profitable in all the coun- tries, and additional investment increases earnings substan- tially. Despite some well-founded concerns about the low quality of primary education, some schooling, even with- out completion of primary education, results in a significant 28 The Knowledge Economy and Education and Training in South Asia Table 3.1 Wage Regressions for South Asia Bangladesh, India, Pakistan, Sri Lanka, Category 2004 2004 2000­1 2001­2 Literate, below primary 0.087 0.195 0.108 0.057 (1.94) (12.53) (4.69) (2.36) Primary 0.236 0.249 0.225 0.185 (9.29) (18.38) (12.63) (7.44) Middle 0.461 0.421 0.341 (34.31) (18.72) (13.52) Secondary and higher secondary 0.717 0.788 0.606 (52.25) (44.38) (24.08) Secondary 0.443 (10.26) High 0.585 (12.14) Tertiary 0.943 1.329 1.397 0.875 (22.90) (79.64) (61.34) (26.31) Technical education dummy 0.18 (10.87) Potential experience (years) 0.024 0.056 0.06 0.026 (6.82) (53.99) (36.90) (18.63) Potential experience squared ­0.001 ­0.001 ­0.001 ­0.001 (­5.27) (­39.40) (­27.69) (­16.76) Male 0.601 0.446 1.089 0.403 (17.05) (47.68) (63.42) (40.72) Urban 0.110 0.221 0.189 0.271 (6.28) (26.06) (15.73) (12.69) Rural 0.059 (3.03) SCST indicator 0.005 (0.67) Regular worker indicator 0.272 0.798 0.362 (10.71) (81.86) (33.31) Constant 1.103 ­0.219 0.581 2.163 (16.91) (­13.29) (21.32) (68.73) Number of observations 4,729 39,190 16,200 20,838 R2 0.319 0.546 0.396 0.292 Source: Authors' calculations. Note: t-statistics in parentheses. Returns to Investment in Education 29 wage gain. The wage gains from completing secondary and higher levels of education are significantly greater than for primary education. The estimated wage-experience profiles are also consistent with wages increasing with labor force ex- perience, although at a decreasing rate. Table 3.1 also indicates several other noteworthy points. First, as might be expected, regular workers command high- er wages than casual workers for any given level of education and experience. A similar observation applies to workers lo- cated in cities compared with rural areas or estates (in the case of Sri Lanka). Second, belonging to a scheduled tribe or caste in India does not have a significant impact on earnings after controlling for the level of education and other personal attributes. For these disadvantaged groups, the difficulty is access to education, but for those who succeed in accessing education, the returns are no different than for the popula- tion at large. Finally, the earnings received by men and wom- en differ strikingly, with men, on average, earning 40 to 100 percent higher wages than women for a given level of educa- tion and controlling for other attributes. The wage regression results are broadly similar across all four countries, but some differences are apparent. For in- stance, in Sri Lanka, the returns to incomplete education are low compared with those in the other countries. Sri Lanka is also noteworthy for the relatively lower return to invest- ment in higher education, as well as its much flatter wage- experience profile, which may reflect the increased supply of those with a tertiary-level education relative to the demand for such workers. Another point that stands out is the large wage premium that men in Bangladesh and Pakistan receive relative to that earned by observationally comparable wom- en workers. 30 The Knowledge Economy and Education and Training in South Asia Rates of Return to Education Comparing the profitability of investments in different levels of education and how they vary over time and across coun- tries is greatly facilitated by calculating standardized rates of return to education. Taking into account the number of years normally required to complete any particular level of educa- tion, one can use the coefficients of the regression to calcu- late standardized rates of return for that level of education. The normal time taken to complete each level of education is as follows: · primary--primary coefficient/five years for all the countries under consideration, · middle--middle coefficient minus primary coefficient/ three years for India and Pakistan and four years for Sri Lanka, · secondary and higher secondary--secondary coef- ficient minus middle coefficient/three years for India and Pakistan and four years for Sri Lanka, · tertiary--tertiary coefficient minus secondary coeffi- cient/four years for India and Pakistan and three years for Sri Lanka.2 The results are interpreted as the rate of return for one additional year of schooling at a given level of education.3 2. Because of specifics of the education system in Bangladesh, the levels of education, and therefore the methodology for calculating returns, differ from those used for other countries: primary = primary coefficient/five years, secondary = secondary coefficient minus primary coefficient/five years, high secondary = high secondary co- efficient minus secondary coefficient/two years, tertiary = tertiary coefficient minus high secondary coefficient/four years. 3. We compared these estimates by level of schooling to the coefficient of a continuous measure of years of schooling and can reject the null hypothesis that the rates of re- Returns to Investment in Education 31 Table 3.2 Rate of Return to Schooling by Education Level, Selected South Asian Countries and Years (percent) Country and level of education Survey and year Bangladesh BHIES 2000 BHIES 2004 Primary -- 7.0 4.7 Secondary -- 6.4 4.1 Higher secondary -- 10.8 7.1 Tertiary -- 10.7 9.0 India NSS 1993 NSS 1999 NSS 2004 Primary 8.3 8.5 8.5 Middle 9.5 8.4 10.7 Middle 23.3 22.7 16.8 Higher secondary 11.7 15.0 16.3 Tertiary 12.6 15.2 18.9 Pakistan PIHS 1993­4 PIHS 1996­7 PIHS 2000­1 Primary 4.4 4.5 4.8 Middle 5.7 6.4 6.6 Secondary 9.5 9.3 14.2 Higher secondary 10.1 11.4 13.9 Tertiary 13.5 11.5 13.9 Sri Lanka LFS 1992­3 LFS 1997­8 LFS 2001­2 Primary 5.6 5.0 5.8 Middle 13.2 12.1 11.6 Secondary 10.6 7.8 8.8 Higher secondary 14.4 16.0 18.4 Tertiary 7.1 9.9 9.6 Source: Authors' calculations. Note: -- = not available. Table 3.2 shows the rates of return to different levels of education. Except for Bangladesh, estimates are calculated turn to schooling are the same for completion of all levels of schooling. 32 The Knowledge Economy and Education and Training in South Asia from wage regressions estimated at three points in time: the early 1990s, the late 1990s, and the early 2000s (appendixes 1­3). In India, the profitability (rate of return) of each year of primary education averages 8.5 percent and the return for each of the following three years of middle education is between 8.4 and 10.7 percent. Table 3.2 also shows that the profitability of such investments tends to rise with the level of educational attainment, most dramatically in India and Paki- stan, and to a lesser extent in Sri Lanka. Gender Gap One of the striking findings in the previous chapter was the size of wage differentials between men and women by level of education. While gender differences in gross wages of the or- der of 30 to 40 percent are not uncommon in other countries, those differences usually narrow when wages are standard- ized by education, age, hours of work, and other individual characteristics. In South Asia, by contrast, even after stan- dardization, gender-related wage differentials ranging from 50 percent in India and Sri Lanka to almost 300 percent in Bangladesh and Pakistan are still observable. Many possible explanations may account for this, including type of employ- ment, sector, and discrimination. Table 3.3 shows estimates of the rates of return to invest- ments in different levels of education by gender. They are cal- culated from wage regressions estimated separately for men and women that control for work experience, location, and type of employment. One finding common to all the coun- tries is the sharp change observed after primary education. While returns to primary education are significantly higher Returns to Investment in Education 33 Table 3.3 Rate of Return to Schooling by Education Level and Gender, Selected South Asian Countries and Years (percent) Country and level of education Survey and year BHIES 2000 BHIES 2004 Bangladesh Female Male Female Male Female Male Primary -- -- 14.1 5.8 13.4 4.2 Secondary -- -- 10.7 4.8 11.6 3.3 Higher secondary -- -- 15.3 10.0 2.2 7.5 Tertiary -- -- 5.0 10.9 105 8.9 NSS 1993 NSS 1999 NSS 2004 India Female Male Female Male Female Male Primary 5.5 8.0 6.9 8.2 6.7 8.2 Middle 14.3 8.7 9.3 8.2 10.3 8.2 Middle 45.0 20.1 42.0 20.0 31.5 20.0 Higher secondary 13.7 10.8 14.6 14.2 20.7 14.2 Tertiary 9.4 12.8 11.5 15.6 16.8 15.6 PIHS 1993­4 PIHS 1996­7 PIHS 2000­1 Pakistan Female Male Female Male Female Male Primary 4.3 4.3 12.9 4.0 5.4 4.1 Middle 13.1 5.6 7.2 6.5 17.1 6.2 Secondary 12.1 9.0 17.2 8.1 30.2 12.3 Higher secondary 7.6 9.8 12.8 11.2 18.5 11.9 Tertiary 15.4 13.3 11.2 11.0 18.9 11.9 LFS 1992­3 LFS 1997­8 LFS 2001­2 Sri Lanka Female Male Female Male Female Male Primary 2.6 5.7 1.5 7.1 1.9 7.6 Middle 18.2 12.0 13.7 11.7 17.8 10.0 Secondary 11.5 10.2 9.6 7.1 10.0 8.1 Higher secondary 8.5 17.0 13.6 16.5 14.7 19.6 Tertiary 9.3 5.6 14.2 6.2 11.7 7.5 Source: Authors' calculations. Note: -- = not available. 34 The Knowledge Economy and Education and Training in South Asia for men than for women in India and Sri Lanka (in Bangla- desh and Pakistan, returns to primary education are higher for women than for men), returns to higher levels of educa- tion are usually much higher for women, especially at the secondary level, and to a lesser extent at the tertiary level. Thus estimates of average wage ratios, even when stan- dardized, hide an important phenomenon, namely, that ac- cess to higher levels of education allows women to reduce the gender gap. For example, when comparing the wages of men and women in India who are otherwise similar, say regular workers living in urban areas with some 20 years of experi- ence, figure 3.1 indicates that the relative wage differential drops by half when the level of education is secondary or higher. This pattern is particularly strong in India and Paki- stan. The results suggest that in countries where access to higher levels of education is more difficult for women than for men and where labor force participation by women is still low, women who succeed in overcoming these obstacles do relatively well in the labor market. This may imply that part of the returns to education estimated for women may actu- ally reflect the greater motivation and ability of the educated women entering the labor market (please refer to the earlier discussion on unmeasured ability in selection bias). Changes over Time in Returns to Education The evidence also indicates that rates of return to higher secondary and tertiary education increased over time in the three countries for which we have time series data. These in- Returns to Investment in Education 35 Figure 3.1 Predicted Log Hourly Wage by Gender and Level of Education, Selected South Asian Countries and Years Figure 3.1 Predicted Log Hourly Wage by Gender Bangladesh 2004 and Level of Education, Selected South 4.5 Asian Countries and Years 4.0 3.5 3.0 2.5 2.0 men women 1.5 1.0 0.5 0.0 illiterate illiterate, primary middle secondary tertiary below and higher primary secondary India 2004 4.5 4.0 men 3.5 women 3.0 2.5 2.0 1.5 1.0 0.5 0.0 illiterate illiterate, primary middle secondary tertiary below and higher primary secondary continued... 36 The Knowledge Economy and Education and Training in South Asia Figure 3.1 (continued) Pakistan 2000­1 4.5 4.0 3.5 3.0 2.5 2.0 men women 1.5 1.0 0.5 0.0 illiterate illiterate, primary middle secondary tertiary below and higher primary secondary Sri Lanka 2001­2 4.5 4.0 3.5 3.0 2.5 2.0 men women 1.5 1.0 0.5 0.0 illiterate illiterate, primary middle secondary tertiary below and higher primary secondary Source: Authors' calculations. Note: These calculations are for regular workers who live in urban areas and have 20 years of experience. In addition, in India they do not belong to a scheduled tribe or caste and do not have a technical education. Returns to Investment in Education 37 creased returns were most pronounced for India:4 between 1993 and 2004, as reported in Table 3.2, the returns to higher secondary education for both males and females rose from 12 to 16 percent and the returns to tertiary education for males rose from 13 to 19 percent. More modest increases in returns were registered for Sri Lanka and Pakistan during the same decade. The increases in returns for Sri Lanka were 14 to 18 percent for higher secondary and 7 to 10 percent for tertiary education; the corresponding increases for Pakistan were 10 to 14 percent and 13 to 14 percent. These time trends resem- ble similar increases in the relative returns to higher educa- tion reported in other regions, including Latin America,5 and may reflect the effects of globalization and/or of skill-biased technological change. These time trends are more readily apparent when rates of return are presented graphically. Figure 3.2 shows the es- timated returns to different levels of schooling for all avail- able years. For each country, the data are shown separately for males and females. The figure confirms the following re- sults. First, returns to education have grown over time, espe- cially for higher secondary and tertiary education. Second, as noted earlier, returns to education are especially high for females, and they too have grown over time. Finally, the re- turns tend to be higher for the high-growth countries (India and Pakistan) and lower for slower growing Sri Lanka. These results suggest that the demand for highly edu- cated and skilled workers is increasing in South Asia and is 4. Patrinos, Ridao-Cano, and Sakellariou (2006) analyze 16 East Asian and Latin Amer- ican countries and obtain similar results. In almost all the countries they look at, re- turns to university qualifications exceeded returns to all other levels. 5. For evidence from Brazil and Mexico, two countries with long time series data on re- turns to education, see Blom, Holm-Nielsen, and Verner (2001) and Lachler (1998). Also see Giovagnoli, Fiszbein, and Patrinos (2005) for evidence of increasing returns to higher levels of education in Argentina during 1992­2002. 38 The Knowledge Economy and Education and Training in South Asia Figure 3.2 Returns to Education over Time by Level of Education and Gender, Selected South Asian Countries and Years (percent) India, Males India, Females 40 45 35 40 30 35 30 25 Figure 3.2 Returns to Education over Time by Level 25 20 of Education and20Gender, Selected South 15 Asian Countries and Years 15 10 10 5 5 0 0 primary middle secondary higher tertiary primary middle secondary higher tertiary secondary secondary NSS 1983 NSS 1988 NSS 1993 NSS 1999 NSS 2004 Pakistan, Males Pakistan, Females 40 40 35 35 30 30 25 25 20 20 15 15 10 10 5 5 0 0 primary middle secondary higher tertiary primary middle secondary higher tertiary secondary secondary PIHS 1003­4 PIHS 1996­7 PIHS 2000­1 Sri Lanka, Males Sri Lanka, Females 40 40 35 35 30 30 25 25 20 20 15 15 10 10 5 5 0 0 primary middle secondary higher tertiary primary middle secondary higher tertiary secondary secondary LFS 1992­3 LFS 1997­8 LFS 2001­2 Source: Authors' calculations. Returns to Investment in Education 39 doing so more rapidly than the supply of graduates, and also that this phenomenon coincides with periods of fast growth. This is consistent with the evidence observed in other devel- oping and developed countries and with the hypothesis that openness to trade, rapid growth, and technological innova- tions fuel increasing demand for skilled relative to unskilled labor. Education and training policies in South Asia have not yet responded to the needs of and signals provided by the labor market. Differences in Returns to Education by Sector Globalization and economic growth may also create dif- ferential demand for a more educated workforce across dif- ferent sectors of the economy. To explore this possibility, we estimated the returns to education for five sectors: (a) manu- facturing; (b) utilities and construction; (c) wholesale, retail, hotels, and restaurants; (d) business services; and (e) pub- lic administration, education, and social services. Figure 3.3 presents estimates of the time trends in returns to higher secondary and tertiary education for India, Pakistan, and Sri Lanka by sector. In India, where economic growth has been most dra- matic since the mid-1990s, a clear trend of rising returns to tertiary education is apparent after 1993, with the most rapid increase being in the business services sector. This is consis- tent with the well-publicized growth in demand for highly educated workers in call centers and finance, where the use of information technology is intensive. For higher secondary graduates, the returns to schooling have been growing since 1983, though growth rates had slowed by 2004. 40 The Knowledge Economy and Education and Training in South Asia Figure 3.3 Rate of Return to Higher Secondary and Tertiary Education for Males by Sector, Selected South Asian Countries and Years (percent) Figure 3.3 Rate of Return to Higher India, secondary education Secondary and Tertiary India, tertiary education 25 Education for Males by Sec- 25 tor, Selected South Asian 20 Countries and Years 20 15 15 10 10 5 5 0 0 NSS 1983 NSS 1988 NSS 1999 NSS 2004 NSS 1983 NSS 1988 NSS 1993 NSS 1999 NSS 2004 Pakistan, secondary education Pakistan, tertiary education 25 25 20 20 15 15 10 10 5 5 0 0 PIHS 1993­4 PIHS 1997­8 PIHS 2001­2 PIHS 1993­4 PIHS 1997­8 PIHS 2001­2 Sri Lanka, secondary education Sri Lanka, tertiary education 24 25 19 20 14 15 9 10 4 5 ­1 0 LFS 1992­3 LFS 1997­8 LFS 2000­1 LFS 1992­3 LFS 1997­8 LFS 2000­1 manufacturing business services utilities, construction public administration, education, social services wholesale, retail, hotel, restaurants Source: Authors' calculations. Returns to Investment in Education 41 In Pakistan, the picture looks different possibly because of the fits and starts in economic growth during the peri- od under review. The returns to tertiary education declined during 1993­2004 in manufacturing; business services; and public administration, education, and social services, al- though during the same period they increased in utilities and construction and in wholesale, retail, hotel, and restaurants. In contrast, for higher secondary education, the returns rose in business services and manufacturing, with increasing re- turns being especially pronounced for the former. However, the returns to higher secondary education declined over this period for utilities and construction and for wholesale, retail, hotels, and restaurants. In Sri Lanka, the returns to both tertiary and higher sec- ondary were unchanged over time for construction and for public administration, education, and social services. Great- er variability over time is apparent in the returns for the other sectors. For tertiary education, the returns generally rose during 1992--98, followed by a decline to 1992 levels by 2001, although in manufacturing, the returns declined until 1998, after which they rose to 1992 levels. For higher second- ary education, the returns to all sectors stayed relatively con- stant during the period. 4 School to Work Transitions This chapter turns to how individuals completing different levels of education fare as they enter the labor market. We ask several questions about youth, defined as those aged 15­29: What are unemployment rates like for youth in the four countries under review? Does more education facili- tate school to work transitions? Are job search and school to work transitions improved through additional postschool training? These issues are of considerable interest to policy makers concerned about high rates of open unemployment among youth in South Asia, especially the most educated. They also raise thorny questions about whether high rates of youth unemployment reflect the low quality and workplace relevance of education or whether the region's economic growth rates are inadequate to generate sufficient new jobs to meet the rising inflow of new labor market entrants. 43 44 The Knowledge Economy and Education and Training in South Asia Definitions of Labor Force States In comparing the school to work transitions of youth across the four South Asian countries, we first need to define broad- ly comparable measures of the different labor force states: employed, unemployed, and out of the labor force. Broadly similar definitions of these three labor force states are pos- sible with the available household surveys in Bangladesh (BHIESs) and India (NSSs) and with the LFSs in Pakistan and Sri Lanka. In all four countries, the past week is the refer- ence period,1 and this is used to define · employed--eitherengagedinsomeformofeconomic activity,2 or employed but not at work because of sick- ness or other reasons; · unemployed--notengagedineconomicactivityand either making tangible efforts to seek work or being available for employment if work is available;3 · notinthelaborforce--notengagedinanyeconomic activity and also not available for work. 1. In Pakistan, the PIHSs use the past month as the reference period for defining em- ployment status, which lowers estimates of open unemployment in Pakistan relative to the other countries, as the likelihood of working for at least one hour in past four weeks is likely to be much higher. Fortunately, the LFS uses the past week as the ref- erence period for defining unemployment status. 2. The NSS (2004) defines economic activities as being self-employed, an employer, a helper in a household enterprise, a regular salary or wage employee, or a casual wage laborer or being employed but not at work because of sickness or other reasons. 3. In Bangladesh, an individual is unemployed if not working but is available for work, which includes seeking employment and not actively seeking employment. School to Work Transitions 45 Unemployment Rates by Education We estimated unemployment rates by level of educational at- tainment for all years for which household or labor force sur- veys were available in each of the four South Asian countries. The surveys available in each country were as follows: · Bangladesh--BHIESs2000and2004; · India--NSSs1988,1993,1999and2004; · Pakistan--LFSs 1993­94, 1996­97, 1999­2000, and 2003­4; · SriLanka--LFSs1992,1995,1998,2000,and2002. Table 4.1 reports the open unemployment rates estimat- ed for the economically active population aged 15­64 in each of the four South Asian countries by survey year and level of educational attainment. Several points stand out. First, open unemployment rates are quite low. In Bangladesh, India, and Pakistan, open unemployment rates in the most recent year for which data were available ranged from 1.5 percent in Bangladesh to 5.1 percent in India. Sri Lanka is the outlier in this group, recording an open unemployment rate of 9.0 percent, or almost double that of the other countries. Second, open unemployment rates for the economically active population tend to rise with level of educational at- tainment in all four South Asian countries. This is most pro- nounced in India, Pakistan, and Sri Lanka where unemploy- ment rates for university graduates are double or almost tri- ple those of people with only a primary school education. Differentiation by education level is much less pronounced in Bangladesh, with open unemployment rates for primary 46 The Knowledge Economy and Education and Training in South Asia Table 4.1 Unemployment Rates by Level of Education, Economically Active Population Aged 15­64, Selected South Asian Countries and Years (percent) Country and level of education Year Bangladesh 2000 2004 Illiterate 3.81 0.65 Literate, less than primary 5.90 1.00 Primary 7.85 1.97 Secondary 8.24 3.11 Higher secondary 8.27 1.48 Tertiary 7.15 3.79 Total 5.57 1.51 India 1987­8 1993­4 1999­2000 2004 Illiterate 2.98 2.05 2.97 2.74 Literate, less than primary 3.34 2.02 2.92 3.15 Primary 4.92 2.79 3.62 4.29 Middle 7.98 5.02 5.62 6.03 Secondary 11.69 7.98 7.44 7.81 Higher secondary -- 11.09 10.17 9.20 Tertiary 13.06 11.99 11.14 11.86 Total 5.11 3.88 4.73 5.07 Pakistan 1993­4 1997­8 1999­2000 2001­2 2003­4 Illiterate 0.71 1.08 2.17 2.05 1.83 Literate, less than primary 1.28 1.57 4.13 4.27 3.37 Primary 1.65 2.21 3.50 3.57 3.57 Middle 2.69 4.04 7.06 5.51 5.43 Secondary 6.14 6.63 6.95 7.37 8.80 Higher secondary 5.30 6.87 6.65 8.96 9.86 Tertiary 5.05 6.08 5.93 7.80 8.21 Total 1.88 2.65 3.87 4.03 4.29 Sri Lanka 1993­4 1997­8 1999­2000 2001­2 2003­4 Illiterate 2.98 1.97 1.14 1.32 1.16 Literate, less than primary 1.83 3.32 2.44 1.02 2.07 Primary 9.65 7.56 4.93 4.17 3.85 Middle 21.63 17.10 11.92 9.47 10.67 Secondary 22.37 18.53 13.436 11.06 13.40 Higher secondary 26.09 23.71 19.33 16.45 18.47 Tertiary 6.31 6.63 6.93 5.61 8.82 Total 14.88 12.73 9.16 7.52 8.96 Source: Household and labor force surveys. Note: -- = not available. School to Work Transitions 47 school leavers being much more similar to those of univer- sity graduates. Third, the three countries with long time series labor force data -- India, Pakistan, and Sri Lanka -- exhibit quite different time trends in relation to open unemployment. Sri Lanka's unemployment rate shows a downward secular time trend, from 15 percent in 1992 to 9 percent in 2002, while Pakistan's unemployment rate rises secularly over time, from 2 percent in 1993­94 to more than 4 percent in 2003­4. In the case of India, open unemployment rates vary within a narrow band of 4 to 5 percent to more than 5 percent during 1987­88 to 2004, with a slight rising trend after 1993­94. Finally, the data show different time trends of unemploy- ment by level of educational attainment in the three coun- tries. In Pakistan, the rise in overall unemployment rates from 1993­94 to 2003­4 is mirrored in rising unemployment rates across all educational groups. In Sri Lanka, the opposite trend is apparent, with declines over time in the unemployment rates for all educational groups except university graduates from the high levels of unemployment prevailing in the early 1990s. In India, by contrast, unemployment rates for those with a secondary education or lower show a rising trend from 1993 onward, while unemployment rates for those with an upper secondary education or a university degree or above either fall over time or remain roughly unchanged. Youth Unemployment and School to Work Transitions The unemployment rates, even when disaggregated by level of educational attainment, are not particularly informative 48 The Knowledge Economy and Education and Training in South Asia about youth unemployment issues and the job search dy- namics that underlie school to work transitions by differ- ent educational groups. The higher unemployment rates for more educated workers observed in all four countries are the outcome of factors related to both age and time in the labor market. The unemployment rates shown in table 4.1 mix up workers in different age categories, for example, a group of people of the same age might include both new university graduates and workers with several years of labor market ex- perience, and also combine rates for males and females, who may have quite different career aspirations and job search experiences. To address this, tables 4.2 and 4.3 presents unemploy- ment rates estimated from the most recently available house- hold or labor force survey in each country disaggregated by gender, age cohort, and years of potential work experience (for more details, see appendixes 4 and 5). Potential work ex- perience is defined as age minus age at which primary school was started (five years for Pakistan and six years for Bangla- desh, India, and Sri Lanka) minus number of years of edu- cation.4 Table 4.2 shows that high open unemployment rates are essentially a youth problem. Indeed, in all four countries, open unemployment rates are significantly higher among males aged 20­24 than among males aged 40­49. In Bangla- desh, the unemployment rates among young males are un- der 4 percent, compared with less than 1 percent for prime age males. The corresponding figures are 10 and 2 percent in India, 8 and 1 percent in Pakistan, and 21 and less than 2 percent in Sri Lanka. 4. If the number of years of education were not available, we imputed the average num- ber of years of education that a person would have on completion of a certain level of education without repeating or postponing any grades. School to Work Transitions 49 Table 4.2 Unemployment Rates by Age and Gender, Economically Active Population Aged 15­64, Selected South Asian Countries and Years (percent) Age cohort (years) Country and gender 15­19 20­24 25­29 30­34 35­39 40­49 50­64 Total Bangladesh 2004 Males 3.37 3.14 2.21 1.22 0.48 0.29 0.08 1.35 Females 5.19 3.82 1.93 4.23 2.37 1.07 3.07 2.96 India 2004 Males 11.02 9.79 6.54 3.50 2.62 2.14 2.15 5.00 Females 8.27 12.07 7.84 4.96 2.64 2.02 1.90 5.22 Pakistan 2002­3 Males 8.42 7.59 4.98 2.82 1.10 1.33 1.01 3.94 Females 8.42 12.21 7.86 5.35 4.00 2.30 1.17 6.06 Sri Lanka 2001­2 Males 27.40 21.11 7.33 2.50 1.44 1.30 0.73 6.51 Females 33.53 32.52 18.30 8.84 4.53 1.67 0.77 12.30 Source: Household and labor force surveys. Table 4.3 Unemployment Rates by Years of Potential Labor Market Experience and Gender, Economically Active Population Aged 15­64, Selected South Asian Countries and Years (percent) Potential labor market experience (years) Country and gender 0­4 5­9 10­14 15­19 20­24 25­34 > 34 Total Bangladesh 2004 Males 6.99 4.73 1.64 0.73 0.38 0.36 0.17 1.35 Females 11.36 1.79 1.94 0.24 2.66 3.21 2.40 2.96 India 2004 Males 18.66 9.76 5.51 3.07 2.18 2.37 2.36 5.00 Females 26.29 12.52 6.37 4.57 3.23 2.43 1.91 5.22 Pakistan 2002­3 Males 19.28 11.77 6.77 3.17 1.32 1.31 1.01 3.94 Females 30.81 19.34 6.63 6.43 3.81 2.93 1.92 6.06 Sri Lanka 2001­2 Males 35.67 19.05 7.03 3.22 1.60 1.21 0.61 6.51 Females 45.57 28.27 13.23 9.62 4.22 1.85 0.72 12.30 Source: Household and labor force surveys. 50 The Knowledge Economy and Education and Training in South Asia The higher unemployment rates among youth than among their older counterparts are consistent with the out- come of a time-dependent job search process. Information about both available jobs and the quality of job matches is initially scarce, so job search tends to be concentrated early in the labor market experience. Some school leavers find a job match quickly and enter employment, while others fail to find a job and continue their job search. With new informa- tion, those still seeking work adjust their expectations about wages and career goals, and either enter employment or con- tinue their job search, and so on. To see this, the unemploy- ment data in table 4.2 are recast in terms of years of potential labor market experience (table 4.3). Unemployment profiles are initially higher (one-and-a-half to three times higher in the 0­4 years of potential labor market experience interval than in the 15­19 years age interval), but then fall off more quickly with time in the labor market than profiles related to chronological age. This unemployment distribution with time in the labor market is consistent with the outcome of the job search process described earlier. Finally, tables 4.2 and 4.3 indicate that with the excep- tion of India, females of all ages are more likely to be un- employed than males at any level of education, age, or years of potential labor market experience. In Bangladesh, Paki- stan and Sri Lanka, profiles of unemployment broken down by potential labor market experience are one-and-a-half to two times greater for females that for males. In India, by contrast, unemployment profiles for males and females are roughly similar. With these insights, figure 4.1 revisits the earlier obser- vation that open unemployment rates are higher among the more educated. It graphs unemployment rates by potential School to Work Transitions 51 ,y 15­64, andy fory y secondar y Aged tiar 2002 primar secondar higher secondar Bangladesh ter 2004 Males Experience, Lanka, India, Sri Market earY and 34 > 34 Experience, Labor > 25­34 25­34 Countries 20­24 Market Potential 20­24 15­19 of 15­19 of and Asian years Labor experience years 10­14 experience 10­14 5­9 South 5­9 0­4 0­4 5 0 Education 0 35 30 tn25 ecrep 20 15 10 5 45 40 Potential 20 15 10 by 45 40 35 30 25tnecrep and earY Selected Rates and 15­64, 34 > 34 > Education 25­34 Aged veys. 25­34 by Countries 20­24 sur Unemployment Males 20­24 15­19 ce of for 15­19 of Rates Asian years 10­14 experience 4.1 e years 10­14 experience 5­9 5­9 labor 0­4 South Figur 0­4 5 0 45 40 35 and 30 tnecre20p 25 15 10 5 0 tn25 ecrep20 15 10 45 40 35 30 Unemployment Selected 2004 2004 4.1 Household e ce: Figur Bangladesh, Pakistan, Sour 52 The Knowledge Economy and Education and Training in South Asia labor market experience for three groups of males -- those with primary schooling, those with secondary education and upper secondary education combined, and those with ter- tiary education -- using the most recently available survey in each country. The figure shows that the higher unemploy- ment rates among the more educated are concentrated in the first 5 to 10 years in the labor market. Subsequently, with time in the labor market, the more educated tend to experi- ence open unemployment at lower rates than their less edu- cated counterparts.5 This pattern suggests that the more educated tend to search more intensively for a good job match. One interpre- tation is that they have more specific skills than their less ed- ucated counterparts, and as such need more time to find a job that requires those specific skills. Alternatively, the more educated enter the labor market with higher career goals and wage expectations that are more difficult to match with avail- able employment opportunities. The more educated may also come from higher-income households that are able to sup- port their job search over an extended period of time, where- as less educated youth unable to finance job search may be- gin working more quickly. Have unemployment rates by potential labor market ex- perience profiles changed over time as suggested by the ag- gregate unemployment rate figures reported earlier? Figure 4.2 graphs these profiles for males in the three countries with long time series data by three levels of education (primary, secondary and higher secondary, and tertiary) and for two points in time (1992 or 1993 and 2002 or 2004). For India, the aggregate data reveal a rising trend in unemployment 5 This trend is also common to other countries outside South Asia, including Chile, Malaysia, Thailand, and Turkey (World Bank 2006a, chapter 5). School to Work Transitions 53 Figure 4.2 Unemployment Rates by Education and Potential Labor Market Experience, Males, Selected South Asian Countries and Years India, 1993 and 2004 35 30 primary, 1993 25 secondary and higher secondary, 1993 tertiary, 1993 20 cent primary, 2004 per 15 secondary and higher secondary, 2004 10 tertiary, 2004 5 0 0­4 5­9 10­14 15­19 20­24 25­34 > 34 years Figure 4.2 Unemployment Rates by Education and Potential Pakistan, 1993 and 2004 35 Labor Market Experience, Males, Selected South primary, 1993 20 Asian Countries and Years secondary and higher secondary, 1993 15 tertiary, 1993 cent primary, 2004 per 10 secondary and higher secondary, 2004 tertiary, 2004 5 0 0­4 5­9 10­14 15­19 20­24 25­34 > 34 years Sri Lanka, 1992 and 2002 60 50 primary, 1992 secondary and higher secondary, 1992 40 tertiary, 1992 cent 30 primary, 2002 per secondary and higher secondary, 2002 20 tertiary, 2002 10 0 0­4 5­9 10­14 15­19 20­24 25­34 > 34 years Source: Household and labor force surveys. 54 The Knowledge Economy and Education and Training in South Asia rates after 1993. Figure 4.2 confirms that unemployment pro- files for those with a primary education shifted upwards over time, while those for tertiary graduates shifted downwards. In the case of Pakistan, aggregate increases in unemploy- ment rates over time are mirrored by modest upward shifts for those with a primary education and larger upward shifts for those with a tertiary-level education. Sri Lanka, which experienced a secular decline in aggregate unemployment rates, saw downward shifts in unemployment for those with a primary education and larger upward shifts for those with a tertiary-level education. The Case of Sri Lanka The previous graphical analyses for the four South Asian countries suggested that while more educated youth may experience higher initial rates of open unemployment, their subsequent likelihood of remaining unemployed declines more with time in the labor market compared with their less educated peers. This section examines this stylized fact more closely for Sri Lanka, taking advantage of the existence of a long annual time series of LFSs that include relatively detailed information about early years in the labor force and postschool training. The school to work transition of Sri Lankan youth is of particular concern to the country's policy makers because of the long time many youth appear to spend in job search be- tween the time they leave school and find employment. Ac- cording to the 2002 LFS, almost 85 percent of youth aged 15­29 who are currently unemployed report never having a job. This figure rises from about 75 percent for those with School to Work Transitions 55 a lower secondary education to almost 95 percent for uni- versity graduates. While these figures highlight the serious- ness of this issue, as noted earlier, they can be misleading, as they mix more and less educated youth with different years of potential work experience, and thus different amounts of time spent in job search. Here we look at the same issue from another perspective, that of time to first job after completing schooling.6 Another question examined is whether school to work transitions are aided by postschool training, holding the level of education constant. This issue is also of considerable in- terest to policy makers concerned with high rates of youth unemployment and keenly interested in knowing wheth- er additional training after formal education is an effective strategy for reducing youth unemployment. This issue can be addressed using information from the LFS on whether indi- viduals received postschool formal or informal training, as well as the duration of that training. Estimating Time to First Job Studying school to work transitions requires information on the date of first employment after schooling completion.7 The challenge of using the Sri Lankan LFS is to determine the date of first recorded employment for each individual with a given level of education,8 from which the time from 6. The analysis in this section draws upon Tan and Chandrasiri 2004. 7. None of the household or labor force surveys in South Asian countries elicit this kind of information, although the Sri Lankan LFS comes closest. 8. Note that the first recorded employment is not necessarily the first job. Some indi- viduals may have had several jobs prior to the recorded job, so time to first recorded job may overstate the duration of job search, but no other information is available. 56 The Knowledge Economy and Education and Training in South Asia schooling completion to first employment can be calculated. Beginning in 1996, the LFS asked the employed how long they had held their current job, so the start date of that job can be ascertained.9 For the unemployed, the LFS asked whether they had ever had a job, and if so, the duration of time since the previous job.10 If prior jobs are assumed to be of similar duration as those held by their currently employed peers (about two years), then this information and the inter- vening unemployment spell can be used to determine the start date of the previous job. For those who have never had a job, the duration of search for a first job is still ongoing (or censored in the sense that the end date is unknown). Finally, search time is adjusted for those with technical and vocational training by subtracting time spent in training to reflect individuals' withdrawal from active job search while undergoing training. These time to first job calculations were done for 39,000 individuals from the pooled LFS sample covering 1996­2002 and were restricted to those with some schooling up to uni- versity graduates and with 0­10 years of potential labor mar- ket experience to keep the focus on youth. Figure 4.3 pres- ents the resulting distributions of time to employment for different levels of schooling attainment.11 Figure 4.3 suggests that low levels of schooling attain- ment disadvantage youth in their job search while higher- level school qualifications facilitate the school to work transi- 9. The 1996 LFS also started asking detailed questions on years of schooling, from which more precise schooling completion dates can be calculated than in the past. 10. The intervening unemployment spell is reported in several intervals, ranging from a few months to an open-ended five or more years. Some assumptions are needed to impute duration (in years) of unemployment to these categories. 11. Note that these graphs understate time to first job because they include unemployed youth who had still not found employment at the time of the LFS. School to Work Transitions 57 10 10 Figure 4.3 Time to First Job by Level of 8 8 y 6 Schooling Completed, Sri Lanka,6 ee Pooled Sample 1996­2002 years 4 degr years 4 1996­2003 secondar 2 2 upper 0 university 0 Sample 20 15 10 5 0 20 15 10 5 0 boj tsrif gnidnif boj tsrif gnidnif noitalupop fo egatnecrep noitalupop fo egatnecrep Pooled Lanka, 10 10 Sri 8 level 8 y 6 6 years tificate years 4 4 secondar cer advanced Completed, 2 2 lower 0 0 general 20 15 10 5 0 education, 20 15 10 5 0 boj tsrif gnidnif boj tsrif gnidnif Schooling noitalupop fo egatnecrep of noitalupop fo egatnecrep of Level 10 10 by 8 2004. Job levely 8 6 6 First y tificate dinar 4 years years to cer or 4 primar 2 2 Chandrasiri imeT 0 0 and general 20 15 10 5 0 education, 20 15 10 5 0 4.3 boj tsrif gnidnif anT of boj tsrif gnidnif e noitalupop fo egatnecrep noitalupop fo egatnecrep ce: Figur Sour 58 The Knowledge Economy and Education and Training in South Asia tion. Those with less schooling -- primary and lower second- ary -- are more likely to face a protracted job search before securing their first employment. Their distributions of time to first employment are concentrated around four to seven years after completion of schooling. In comparison, most of those completing upper secondary schooling and with gen- eral certificate of education ordinary level (GCE O-level) or general certificate of education advanced level (GCE A-level) qualifications find their first job fairly soon after schooling completion. Their distributions of time to first job are con- centrated around none to four years, tapering off with time in the labor market. However, the school to work transition of those with uni- versity degrees resembles that of youth with lower secondary schooling more than those with GCE A-level qualifications. The distribution for university graduates is bimodal, that is, some find a job within the first year, while many others ap- pear to take about three to five years after graduation from university. The protracted job search of the latter group may reflect the difficulty of finding an appropriate job match for its members' more specialized tertiary-level training or, as some have speculated (World Bank 1999), may reflect queuing for limited but prestigious employment in the public sector. Survival Models of Time to First Job These figures do not control for other factors that may also shape school to work transitions, such as gender, household characteristics, location, and postschool technical and voca- tional training. The joint effects of schooling attainment and these other factors on time to employment can be studied School to Work Transitions 59 within a regression framework that accounts explicitly for the fact that one part of the sample is still actively searching for the first job.12 Table 4.3 reports the results of estimating this regression model for the sample of youth as a whole and separately by training status to investigate how technical and vocational training affects school to work transitions. Table 4.3 Time to First Job with and without Postschool Training, Sri Lanka (dependent variable = time to employment) All youth Without training With training Category Coefficient z-stat Female Male Female Male Lower secondary ­0.329 ­15.6 ­0.340 ­15.8 ­0.166 ­1.7 Upper secondary ­0.471 ­24.2 ­0.492 ­24.7 ­0.294 ­3.1 GCE O-levels ­0.434 ­21.0 ­0.448 ­20.9 ­0.284 ­3.0 GCE A-levels ­0.454 ­20.8 ­0.445 ­19.4 ­0.350 ­3.6 Degree ­0.340 ­10.8 ­0.276 ­8.1 ­0.459 ­4.2 Formal training ­0.069 ­6.0 Informal training ­0.106 ­5.2 Male ­0.070 ­8.1 ­0.069 ­7.3 ­0.077 ­4.0 Married 0.113 9.6 0.136 10.4 0.028 1.0 Urban 0.030 2.9 0.049 4.1 ­0.040 ­1.8 Provincial dummies Yes Yes Yes Constant 1.964 54.0 1.979 50.2 1.771 14.7 Sample size 33,206 26,274 6,932 Number finding jobs 24,605 19,678 4,927 Source: Sri Lanka LFSs 1996­2002. Note: The regressions are estimated by maximum likelihood using a parametric survival time model fit with a lognormal distribution. About one quarter of the samle were censored. The regression model included control variables for parental education and for LFS years. 12. Survival models are ideally suited for studying the determinants of time to a failure event, in this case, time taken to find a job after schooling completion, and for ac- commodating censored spells of job search. Such models may be fitted using alter- native distributional assumptions about the underlying process, but the model used here is the lognormal distribution. 60 The Knowledge Economy and Education and Training in South Asia The results in table 4.3 make several points. First, com- pared with youth with primary schooling, more educated groups find employment much faster, though as figure 4.2 suggests, those with a university degree are more like those with an upper secondary education than those with, say, GCE O-level or GCE A-level qualifications. Second, gen- der differences are important, and males appear to find em- ployment faster than females. A contributing factor to this gender gap may be marital status, as marriage is often as- sociated with withdrawal from the labor market, and thus with delayed time to employment. Location also matters: job search is longer in urban areas and varies across provinces (not reported here). Finally, trends estimated by year dummy variables (not reported here) indicate that the overall length of job search has declined over time in parallel with falling overall unemployment rates. As for the effects of training, the second column of table 4.3 indicates that formal and informal training are both as- sociated with shorter search time, with informal training ap- pearing to have a larger impact (­0.10) than formal training (­0.07). The columns reporting results estimated separately by training status make the additional point that while hav- ing more education reduces time to employment for both those without training and those with training, the impact of education is more pronounced for the group with training. The relative contributions of different levels of education to shortening time to employment in the group without train- ing peaks with upper secondary education. In contrast, the contribution of schooling of the group with training rises lin- early with level of education, peaking with university gradu- ates. In other words, education and training interact posi- tively to reduce the time spent in job search. 5 Postschool Training in the Labor Market This chapter turns to an exploration of the pre-employment and on-the-job training that individuals may acquire after completing their formal education. The analysis of school to work transitions and postschool training in Sri Lanka reported earlier suggests that training can improve young people's labor market outcomes by complementing their formal education. Here we identify household and labor force surveys in other South Asian countries that include information on postschool training to provide a broad overview of postschool training in South Asia, to ascertain its incidence among individuals with different levels of education, and to document some early find- ings on the impact of training on wages. Surveys of Postschool training Information on postschool training in South Asia is limit- ed. Pakistan and Sri Lanka's LFSs have elicited information 61 62 The Knowledge Economy and Education and Training in South Asia on postschool vocational training since the early 1990s. In the other South Asian countries, such information is rarely asked, and if asked, only periodically. Our review identified the following surveys with training information: · Bangladesh--the BHIES 1995 asked, for just one year, whether respondents had received any vocational training, and if so, the type and length of training, the training institution, and the utility of the training to re- spondents' current work. Information on respondents' occupation and industry is available, but these data cannot be linked to individual employment and wage data to study the labor market outcomes of vocational training. · India--the NSS (2004) asked individuals about voca- tional training for the first time and restricted ques- tions to those with at least a middle school education and aged 15­29. If respondents had received vocation- al training, the survey asked about the field of training; the name of the training institution; the duration of the training; whether respondents had received a degree, a diploma, or a certificate; and whether the training had been useful for respondents' current jobs or for taking up other jobs. The NSS also elicited information on oc- cupation and sector of employment. · Pakistan--the 1993­2004 LFSs and the 1997 PIHS asked individuals about whether they had completed vocational and technical training. In addition, the sur- veys elicited information on occupation and employer characteristics, such as industry and which of four em- ployment size categories the respondents belonged to. Postschool Training in the Labor Market 63 · Sri Lanka--the 1992­2002 LFSs asked all individuals whether they had received vocational training, and if so, whether the training had been formal or informal and how long it took. Information on the types of vo- cational training received was elicited, but rarely cod- ed. In addition to the usual LFS questions, the survey asked about current occupation and sector of employ- ment. Incidence of Postschool Training Table 5.1 shows the proportion of the population aged 15­64 reporting vocational training by educational attainment and gender in Bangladesh, India,1 Pakistan, and Sri Lanka. The table is based on the most recent survey available for each country, typically in the early 2000s for India, Pakistan, and Sri Lanka and 1995 for Bangladesh. The table indicates that the incidence of postschool vocational training is quite low in South Asia. It is lowest in Pakistan, 2.4 percent, and high- est, 12.0 percent, in Sri Lanka. Overall levels aside, the incidence of training shows a strong tendency to rise with the level of educational attain- ment across all the countries. For example, of the Sri Lankan population with a lower secondary education, 9.8 percent had also received vocational training compared with 34.5 percent of graduates. The incidence of postschool vocational training tends to peak at or after high school, after which it declines before peaking again after the first degree. These are the times when individuals end their formal education and 1. As noted earlier, the India NSS (2004) only asked people aged 15­29 who had com- pleted middle school about training. 64 The Knowledge Economy and Education and Training in South Asia 0.4 2.0 4.2 13.7 25.3 39.2 33.8 48.4 9.1 Female Selected, Lanka 2.0 6.0 Male 11.4 18.7 24.8 36.0 35.1 42.1 15.1 Sri Gender All 1.2 4.7 9.8 17.4 25.0 37.4 34.5 45.1 12.0 of y y y O-levels A-levels Education Illiterate Primar Lower secondar Upper secondar GCE GCE Graduate Post- graduate otalT Education of 0.5 1.3 1.7 1.4 3.0 4.7 4.8 5.6 1.2 Female Level by Male 1.7 2.6 3.0 3.1 5.1 7.4 10.7 8.5 3.6 Pakistan rainingT All 0.9 2.1 2.5 2.5 4.3 6.4 8.6 7.6 2.4 y y y mal y ocationalV for ee Education No Below primar Primar Middle Secondar Higher secondar Degr Post- graduate otalT Any 2002. -- -- 1.1 3.4 7.4 Female 48.4 16.3 18.3 3.6 LFS Lanka, Obtaining India Male -- -- 0.7 4.4 8.9 62.7 17.1 18.1 4.4 Sri 2004; 15­64 All -- -- 0.9 4.0 8.3 58.6 16.8 18.2 4.0 LFS Aged earY y y y Pakistan, tificate and Education Illiterate Primar Middle Secondar Higher secondar Diploma cer Graduate Post- graduate otalT 2004; 60 Population NSS the 2.7 0.0 0.0 5.8 Female 49.5 0.0 67.5 18.6 0.0 0.0 of Countries India, Asian Male 1.4 4.4 8.1 11.3 12.5 19.7 11.3 6.7 27.5 4.6 1995; Bangladesh centage BHIES Per South All 1.5 4.3 9.2 11.1 13.3 19.7 11.1 6.7 27.5 4.7 available. 5.1 ,y 6­8 ,y not = y 9 Bangladesh, -- tificate tificate ableT general with Education Illiterate Primar Secondar grades Secondar grade School cer Higher cer BA BA honors dnaAM ce: above otalT Sour Note: Postschool Training in the Labor Market 65 obtain postschool vocational or technical training, either to become skilled workers after high school or to become pro- fessionals after completing their tertiary education. Table 5.1 also shows that women are less likely to receive postschool vocational training than their male counterparts with the same level of education. In India, 4.4 percent of males receive vocational training versus 3.6 percent of fe- males. The corresponding gender differences are 3.6 and 1.2 percent in Pakistan and 15.1 and 9.1 percent in Sri Lanka. Bangladesh appears to be an anomaly in South Asia, with females being more likely to obtain vocational training (5.8 percent) than males (4.6 percent). The reason for this is un- clear and requires further study. As concerns which occupational groups are most like- ly to receive vocational training, even though definitions of occupations vary from one country to another, the figures reported in table 5.2 suggest that professionals, technicians, and clerical personnel in South Asia are more likely to receive vocational training than those in other occupational groups. This makes sense, as these are occupations that tend to in- clude a large number of the highly educated. In Pakistan and Sri Lanka, relatively high shares of plant and machine opera- tors and assemblers and craft workers also receive training. The occupations with the lowest share of individuals receiv- ing vocational training are employees in sales, services, and agriculture, where educational requirements tend to be low. The incidence of postschool training also varies across sectors. Table 5.3 tabulates the percentage of the workforce acquiring postschool training by sector of employment. In Bangladesh, India, and Pakistan, the utilities sector tends to have the highest share of employees with postschool train- ing, followed broadly by real estate and finance, and pub- 66 The Knowledge Economy and Education and Training in South Asia Table 5.2 Percentage of the Workforce Obtaining Vocational Training by Occupational Category, Selected South Asian Countries and Years Pakistan, Sri Lanka, India, Occupation 2004b 2002b Occupation 2004a Professionals 46.3 9.3 Professional, technical, and 24.6 related workers Technicians and associate 27.8 11.5 professionals Production, transport 5.9 operators, and laborers Plant, machine operators and 30.0 10.8 assemblers Clerical and related workers 17.2 Craft and related workers 29.1 11.1 Administrative and managerial 9.7 workers Clerical and related workers 20.7 8.7 Service workers 1.4 Legislators and senior officials 19.2 4.9 Sales workers 4.5 Service and sales workers 9.2 3.3 Farmers, fishermen, hunters, 5.3 Skilled agricultural and fishery 6.1 1.0 loggers, and related workers workers Elementary occupations 3.7 1.2 (manual labor, simple and routine tasks, etc.) Source: India, NSS 2004; Pakistan, LFS 2003­4; Sri Lanka, LFS 2002. Note: Bangladesh is excluded because its occupational classification system differs so dramatically from that used in the other countries. a. Respondents aged 15­29. b. Respondents aged 15­64. lic administration and social services. In the manufactur- ing sector, a relatively smaller percentage of workers receive postschool training. The sectors with the smallest share of workers obtaining vocational training are trade, construc- tion, hotels and restaurants, and agriculture. The difference in the extent of training in the mining sector are striking and unexplained, with 37.7 percent of employees in Pakistan receiving training, compared with 1.7 percent in India and none in Bangladesh. Finally, table 5.4 reports the principal sources of post- school vocational training for Bangladesh and India (for these two countries, information is also available on fields of Postschool Training in the Labor Market 67 Table 5.3 Percentage of the Workforce Obtaining Vocational Training by Sector of Employment, Selected South Asian Countries and Years Bangladesh, 1995a India, 2004b Pakistan, 2003­4a Sector Percentage Sector Percentage Sector Percentage Electricity and gas 39.5 Electricity, gas, and 23.6 Utilities 17.7 water supply Finance, real estate, 17.7 Finance and business 12.8 and financial services Real estate, renting, 19.4 business activities Social services and 8.8 Social and personal 12.9 public administration services Financial 14.6 intermediation Transport 6.9 Transport 8.3 Community, social, 10.6 Manufacturing 10.0 Manufacturing 13.9 and personal service Trade 2.9 Housing and 7.4 activities construction Construction 4.1 Public administration 9.0 Business, hotels, and 2.5 and defense, Mining 37.7 restaurants compulsory social security Agriculture 0.9 Mining and quarrying 0.0 Transport 7.2 Agriculture 1.4 Manufacturing 7.1 Trade 5.5 Construction 4.4 Hotels and restaurants 3.8 Mining and quarrying 1.7 Hunting, forestry 1.4 Source: Bangladesh, BHIES 1995; India, NSS 2004; Pakistan, LFS 2004­5.. Note: Sri Lanka was excluded because the industrial classification system changed. a. Respondents aged 15­64. b. Respondents aged 15­29. training received, see appendixes 6 and 7). In India, indus- trial training institutes and industrial training centers are by far the most important sources of vocational training (27.3 percent of trainees), especially for males (38.9 percent). In contrast, females were more likely to have received vocation- al training from tailoring, embroidery, and stitch craft insti- tutes (22.5 percent). What institutes fall into the other insti- tutes category is unclear (see Appendix 8 for a complete list of institutions). In Bangladesh, 30.8 percent of workers who 68 The Knowledge Economy and Education and Training in South Asia Table 5.4 Percentage of the Workforce Obtaining Training by Source of Vocational Training and Gender, Bangladesh 1995 and India 2004 Bangladesh Indiaa Training institutions All Male Female Training institutions All Male Female Government 30.8 29.4 53.5 Industrial training 27.3 38.9 7.2 institution institutes or centers Private institution 27.0 28.3 4.3 Tailoring, 8.8 0.9 22.5 Family member 13.8 12.5 34.25 embroidery, or Private employer 7.4 7.6 3.9 stitch crafts Nongovernmental 1.5 1.4 4.1 Polytechnics 5.8 7.6 2.8 organization Secondary school 5.2 5.6 4.6 Public sector 0.8 0.8 0.0 offering vocational employer courses Other 18.8 19.9 0.0 Other institutes 52.8 47.0 62.9 Source: Bangladesh, BHIES 1995; India, NSS 2004. a. Appendix 8 provides more detailed breakdowns of training by training institution. receive training do so in government training institutions, a figure that is especially high for females (53.5 percent) com- pared with males (29.4 percent). Private training institutions are an important source of training for males, and family members are an important informal source of training for females. Trends in Postschool Training Pakistan's and Sri Lanka's LFSs have time series data that can be used to examine trends in postschool training over a 10- year period. In Pakistan, the incidence of postschool train- ing declined from 4.1 percent of the workforce in 1993 to 2.4 percent in 2003, but started to rise again in 2004. In Sri Lanka, the overall fraction of the workforce that received postschool training remained unchanged at about 12 percent Postschool Training in the Labor Market 69 during 1992­2002, although the figures conceal considerable compositional changes by education, age, and type of train- ing. We now exploit the availability of long annual time series data to look at training trends in the two countries. The Case of Pakistan Pakistan's LFSs cover the period from 1993 through 2004. While they contain rich information on individual and household attributes and on labor force variables, informa- tion on postschool training is relatively limited. The training variable is limited to only two questions: (a) whether the re- spondent ever completed any technical or vocational train- ing, and (b) if so, the type of training. The LFSs do not ask whether the training was formal (that is, if a diploma or cer- tificate was received) or informal, about the duration or year of the training, or which institution provided it. Figure 5.1 reports trends in training incidence of the working population during 1993­94- through 2003­4. For the workforce as a whole, the share that received vocational training rose slightly from 4.1 percent in 1993­94 to 4.6 per- cent in 1996­97, then fell to l.4 percent by 2001­2 before ris- ing to 2.5 percent in 2003­4, a level roughly half that prevail- ing in 1996­97. These trends in the proportion of the workforce receiv- ing training appear to mirror the overall growth of the econ- omy, with perhaps a two- or three-year lag (figure 5.2). As noted earlier, training incidence, especially of youth, rose between 1993­94 and 1996­97 following an increase in the economy's annual growth rate from 1.8 in 1993 to 5.0 percent in 1995. When economic growth slowed down thereafter, the 70 The Knowledge Economy and Education and Training in South Asia Figure 5.1 Proportion of the Population That Received Vocational or Technical Training by Age and Gender, Pakistan, 1993­4 to 2003­4 By age 7.0 age 15­64 6.0 Figure 5.1 Proportion of the Population That age 15­29 5.0 Received Vocational age 30­64 or Technical Traning by Age and Gender, 4.0 Pakistan, 1993­4 cent Per and 2003­4 3.0 2.0 1.0 0 1993­4 1996­7 1997­8 1999­2000 2001­2 2003­4 By gender 7.0 male, age 15­29 6.0 male, aged 30­65 female, age 15­29 5.0 female, age 30­64 4.0 cent Per3.0 2.0 1.0 0 1993­4 1996­7 1997­8 1999­2000 2001­2 2003­4 Source: LFSs. Note: The data are weighted based on weights provided by the LFSs. Postschool Training in the Labor Market 71 Figure 5.2 Annual Gross Domestic Product Growth, Pakistan, 1993­2004 7 6 Figure 5.2 Annual Gross Domestic Product Growth, Pakistan, 5 1993­2004 4 cent per 3 2 1 0 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 Source: World Bank 2006b. incidence of training also declined, and did not pick up until the economy started growing again. This correlation between growth and training suggests that individuals and employers invest pro-cyclically in training, seeing greater employment opportunities and rising demand for a skilled workforce when the economy is growing. Figure 5.1 shows that Pakistani women are not only less likely than men to receive training, but that the gender gap in training incidence has increased over time: for youth and adults combined, the gender gap in training widened from 5.5 percent of men and 2.6 percent of women in 1993­94 to 3.6 of men and 1.3 percent of women by 2003­4. Also note- worthy is that adult males are usually more likely than young men to report vocational education, while the opposite is 72 The Knowledge Economy and Education and Training in South Asia Table 5.5 Percentage of the Population Aged 15­64 That Received Vocational Training by Level of Education and Gender, Pakistan, Selected Years Males Females Level of education 1993­4 1999­2000 2003­4 1993­4 1999­2000 2003­4 No formal 3.53 1.28 1.69 1.70 0.46 0.52 Below primary 6.32 2.17 2.62 4.43 0.49 1.26 Primary 8.07 3.24 3.02 6.91 1.23 1.67 Middle 5.99 2.37 3.07 4.12 1.17 1.41 Matriculation 7.54 5.02 5.10 6.26 3.20 2.96 Intermediate 7.97 8.96 7.43 5.99 2.06 4.67 (grades 11 and 12) Degree 6.94 9.20 10.70 5.39 5.06 4.85 Postgraduate degree 8.32 10.42 8.55 8.09 4.86 5.60 Source: LFSs. true for women, in that young women are more likely than adult women to obtain vocational training, though the dif- ference between the two groups has narrowed over time. Table 5.5 reports cross-tabulations of training and edu- cation by gender at the start, midpoint, and end of the 1993­ 94 to 2003­4 period. The table brings out three main points. First, training incidence rises with educational attainment, from 0.5 to 1.7 percent for females and from 1.3 to 3.5 per- cent for males for those with no formal education and from 4.9 to 8.1 percent for females and from 8.3 to 10.4 percent for males for those who have a postgraduate degree.2 Second, fe- males are less likely to get training at any given level of educa- tion, and this gender gap by education stays roughly constant 2. One exception is evident in 1993­94, when the incidence of vocational training is extremely high for those who completed primary school. Those numbers are higher than for any other superior level of education except for those with a postgraduate degree. Postschool Training in the Labor Market 73 over time. Third, while the incidence of training declines over time, those with the lowest levels of schooling (no for- mal education through middle school) experience the great- est declines, though for males with degrees and postgraduate education, the incidence of training increased slightly (from 8.3 to 8.6 percent) between 1993­94 and 2003­4. This is con- sistent with the finding for other South Asian countries that the demand for skills rises with globalization and growth of the knowledge economy. Pakistan's LFSs, unlike similar surveys in other South Asian countries, also elicit details about 43 types of voca- tional training received (appendix 9 provides a complete list of trades and the number of people who received vocational education in those trades). Most people report receiving vo- cational training in five trades in 2003­4: 57 percent of those trained reported training in computers, driving, embroidery and knitting, garment making, or electrician courses, while the remaining 43 percent reported training in 38 other trades each accounting for less than 3 percent of the total work- force trained. As might be expected, some trades were more popular among men than women, for example, 70 percent of those trained in embroidery and knitting and garment mak- ing were women, who only accounted for 9 percent of those who received driving or electrician training. Table 5.6 shows the top 10 trades in which males and fe- males received vocational training over 1993­94 to 2003­4 ranked by popularity in 2003­4. Several points stand out. First, the composition of trade training changed significantly during the decade. Among men, the proportion getting train- ing in masonry and garment making declined significantly, while increased training was reported in embroidery and knitting and civil engineering technology. Among women, 74 The Knowledge Economy and Education and Training in South Asia Table 5.6 Composition of Vocational Training Received by the Population Aged 15­64, Pakistan, Selected Years Type of training received 1993­4 1996­7 1997­8 1999­2000 2001­2 2003­4 Males Computers 12.3 18.4 17.1 25.7 16.0 17.4 Driving 17.4 18.7 24.3 14.0 14.9 16.7 Electrician 6.2 5.8 5.4 7.3 6.8 7.9 Automobile mechanics 5.5 3.5 4.6 7.0 7.1 5.1 Embroidery and knitting 1.0 2.1 1.8 2.1 1.7 4.8 Garment making 9.2 7.0 6.4 3.8 1.6 3.8 Carpentry 4.4 4.4 4.1 1.9 5.1 3.4 Masonry 12.0 6.2 3.7 2.6 1.3 2.9 Welding 1.9 2.4 2.1 2.9 2.2 2.4 Civil engineering technology 1.3 3.1 2.9 3.2 3.9 2.4 Females Embroidery and knitting 11.5 15.4 32.8 18.3 12.1 32.2 Garment making 59.1 51.5 37.1 29.5 19.3 25.3 Computers 4.4 13.2 11.9 27.8 13.2 12.6 Weaving 8.4 1.5 3.4 3.3 4.0 2.7 General nursing 0.2 1.3 0.2 1.3 4.4 2.4 Health visitor 0.4 1.7 1.0 4.0 1.7 2.1 Electrician 0.2 0.4 0.7 1.2 0.9 2.0 Driving 0.5 0.7 0.3 1.6 0.2 1.7 Drafting 0.2 2.5 1.6 0.5 3.5 1.6 Civil engineering technology 1.6 3.1 2.1 2.1 5.4 1.6 Source: LFSs. the proportion with training in garment making declined, but the proportion trained in embroidery and knitting in- creased. The proportion of women taking computer, general nursing, health visitor, electrician, and driving courses also increased. Postschool Training in the Labor Market 75 A second point that emerges is the large increase in the proportion of the workforce that reported training in com- puters over this decade. This was true for both men, from 12.3 to 17.4 percent, but especially for women, from 4.4 to 12.6 percent. The increase for both men and women was es- pecially pronounced in 1999­2000, and may be explained by the increasing use of information technology in a growing number of jobs. The emergence of the knowledge economy and the mounting use of information technology in manu- facturing and service sector jobs increase the demand for workers with computer literacy and, if this demand is not met by rising supply, lead to rising wages as well. The Case of Sri Lanka Figure 5.3 shows the weighted proportions of the working- age population that reported having received vocational or technical training separately for any training and for formal or certificated training. Within each category of training, the proportions are shown separately for all ages, youth, and adults. Several trends emerge from the figure. First, training incidence shows a secularly rising trend between 1992 and 1999, a stagnation and marked decline in 2001 in line with negative economic growth, and recovery by 2002. Second, the type of training received is increasingly more formal over time: the proportion of the workforce receiving any training rises from 11 to 13 percent during 1992­2002, but the pro- portion obtaining formal training rises from 7 to 10 percent. Finally, in each year a higher proportion of youth aged 15­29 years reported training than did adults aged 30­65, and over 76 The Knowledge Economy and Education and Training in South Asia Figure 5.3 Percentage of the Workforce with Vocational and Technical Training, Sri Lanka, 1992­2002 1.4 any training, aged 15­29 formal training, aged 15­65 1.2 any training, aged 30­65 formal training, aged 15­29 formal training, cent 1.0 aged 15­65 per formal training, aged 30­65 0.8 0.6 1992 1994 1996 1998 2000 2002 Source: LFSs. time, these age-related differences in training widened. In other words, recent entrants into the labor market are more likely to have received training than their counterparts from years past, which may reflect an increased supply of techni- cal and vocational training, an increased derived demand for skills from employers, or some combination of both factors. Table 5.5 reports cross-tabulations of training and educa- tion by gender at the start, midpoint, and end of the 1993­94 to 2003­4 period. The table highlights two time trends. First, females are less likely than males to get training at any given levelofeducation,andthisgendergapbyeducationhasstayed roughly constant over time. Second, while the incidence of training declines over time for most groups, those with the lowest levels of schooling (no formal education through mid- Postschool Training in the Labor Market 77 dle school) experienced the greatest declines. Males with de- greesandpostgraduateeducationaretheexceptions,andtheir training incidence increased slightly (from 8.3 to 8.6 percent) between 1993­94 and 2003­4, which is consistent with the finding for other South Asian countries that the demand for skills rises with globalization and growth of the knowledge economy. Table 5.7 Training Trends by Education and Gender, Sri Lanka, Selected Years (percent) Males Females Education completed 1992 1997 2002 1992 1997 2002 Percentage receiving any training No schooling 2.6 3.7 2.4 1.2 0.7 0.7 Primary 8.0 6.2 5.5 2.0 1.2 1.5 Lower secondary 11.5 10.5 9.9 3.5 2.7 2.3 Upper secondary 15.9 14.8 15.8 8.7 6.8 6.8 GCE O-levels 21.0 22.7 21.2 15.9 16.3 13.6 GCE A-levels 29.0 34.0 37.3 27.8 29.3 32.6 Graduate 29.9 33.0 39.6 21.9 24.1 31.4 Postgraduate 57.5 53.3 46.9 41.8 48.9 46.7 Percentage of training that is formal No schooling 4.2 29.4 27.5 12.2 23.9 13.4 Primary 23.7 26.9 24.6 43.7 45.4 33.6 Lower secondary 35.7 37.2 46.8 47.1 54.3 38.8 Upper secondary 52.0 62.8 68.7 72.6 67.7 72.2 GCE O-levels 78.6 83.7 84.6 84.4 81.1 87.4 GCE A-levels 88.6 91.5 92.2 94.2 93.4 94.7 Graduate 97.9 94.8 96.3 95.6 100.0 93.1 Postgraduate 97.9 100.0 100.0 100.0 98.1 100.0 Source: LFSs. Note: Figures are for the population aged 15­65 years, weighted using Department of Census and Sta- tistics sampling weights. 78 The Knowledge Economy and Education and Training in South Asia Table 5.8 Percentage of the Population Trained by Age Group and Education, Sri Lanka, Selected Years Aged 15­29 Aged 30­65 Education completed 1992 1997 2002 1992 1997 2002 No schooling 1.2 3.5 1.4 1.7 1.1 1.2 Primary 5.1 4.2 4.0 5.1 3.7 3.4 Lower secondary 6.8 5.9 5.7 8.2 7.3 6.6 Upper secondary 11.7 10.5 10.7 13.6 11.1 11.8 GCE O-levels 16.8 17.1 15.1 20.0 21.3 19.1 GCE A-levels 24.9 30.0 35.8 33.1 32.9 33.4 Graduate 27.4 24.0 28.7 26.0 30.4 37.5 Postgraduate 41.6 40.4 53.9 54.6 52.4 46.2 Source: LFSs. Note: Weighted using Department of Census and Statistics sampling weights. To examine training profiles as individuals complete their formal schooling and acquire work experience in the labor market, table 5.8 reports training data by educational attainment separately for two broad age groups, youth a and adults. Two points stand out. First, among youth aged 15­29, the incidence of training for those with GCE A-levels and above increases dramatically, but not for those with GCE O- levels and below. Among adults aged 30­65, the only group to show a rising trend in training is university graduates. Figure 5.3 showed similar age-related differences in training, but across all education groups. Second, at each level of edu- cation, a roughly equal or higher proportion of adults reports having training than similarly educated youth, which is con- sistent with the cumulative probability of training as individ- uals age, though at a slower pace as they become older. Postschool Training in the Labor Market 79 Postschool Training and Wages The main labor market outcomes of investments in post- school training that are of policy interest are unemployment, job search, and earnings. This subsection asks whether post- school training affects wages, and if so, how the returns to vocational training compare with those from investments in formal education. We use the term returns loosely, as post- school training is measured as an indicator variable -- with a value of 1 if the individual reported getting vocational train- ing and 0 otherwise -- and not in terms of time (fraction of years) spent in training as was the case for schooling. As such, the estimated coefficient should be interpreted as the return to an average spell of postschool training.3 In training as in education, selectivity bias can arise be- cause unmeasured productivity attributes of the individual are correlated with both the training choice and the outcome variables of interest. While econometric techniques to ad- dress selectivity bias in estimating the returns to training are available (Barnow, Cain, and Goldberger 1981), these are not pursued here for the same reasons noted earlier in relation to using simple models to estimate returns to schooling.4 We estimated broadly comparable wage models for India (2004), Pakistan (2004), and Sri Lanka (2002) using the most recent survey available for each country and including all in- 3. The India NSS (2004) was the only survey that reported training duration (appendix 8). Time spent in training ranged from a low of three months in carpet weaving cen- ters to a high of three years in polytechnics, with one to two years being the average duration of postschool training. 4. Another reason for using simple models is that training returns corrected for selec- tivity bias are often imprecisely estimated. For example, see the studies using a simple treatment effects model to estimate the returns to training (such as Tan and Batra 1995) or more sophisticated studies using panel data (such as Dearden, Reed, and van Reenen 2006). 80 The Knowledge Economy and Education and Training in South Asia dividuals aged 15­64 who worked for wages and salaries last week. We calculated the logarithm of hourly wages based on the reported number of hours worked in the relevant interval and regressed it on indicator variables for postschool train- ing, individual characteristics (years of education, gender, a quadratic measure of potential work experience), indicator variables for employment status and caste (India), and geo- graphic location. Table 5.9 reports the results. The table suggests that the average returns to postschool training are positive and statistically significant in all three countries, even after controlling for educational attainment and other worker attributes. · In India, the returns to formal vocational training are about 8 percent, almost equivalent to the 8.4 percent return to an additional year of education. · In Pakistan, the returns to formal vocational training are comparable to those in India and equal 8.1 percent. This number is slightly lower than the returns to one additional year of education in Pakistan of about 9 per- cent. When vocational training is differentiated by type (see model 2), the results indicate that the returns to computer training are substantially higher, 18 percent, than those from all other types of vocational training combined, 6 percent.5 · In Sri Lanka, formal vocational training is associated with relatively high returns of 17 percent, more than 5. The estimated high returns to computer training might plausibly explain both its popularity and its rising incidence among the Pakistani workforce aged 15­65 dur- ing 1994­2004. Using a time series of labor force surveys, Savchenko and Tan (2007) show that the proportion of male Pakistani workers who received computer training rose from 12 percent in 1994 to more than 17 percent by 2004. This trend was even more dramatic for women: the incidence of computer training among women tripled during this period from 4 to 12 percent. Postschool Training in the Labor Market 81 Table 5.9 Postschool Training and Wages, Selected South Asian Countries and Years (dependent variable = log[hourly wage]) Pakistan Sri Lanka Independent variables Indiaa Model 1 Model 2 Model 1 Model 2 Years of education 0.084 0.090 0.089 0.079 0.078 (21.09)** (63.02)** (62.82)** (54.80)** (53.35)** Formal vocational training 0.080 0.081 0.170 0.211 (2.55)* (2.95)** (13.68)** (15.22)** Computer vocational training 0.186 (2.81)** Other vocational training 0.061 (2.02)** Informal vocational or technical training 0.035 (1.46) Male indicator 0.340 0.285 0.285 0.328 0.333 (14.5)** (16.26)** (16.23)** (33.47)** (33.89)** Urban location 0.190 0.294 0.297 (11.0)** (13.47)** (13.63)** Rural location 0.035 0.04 (1.76) (1.99)* SCST indicator ­0.003 (­0.15) Years of potential experience 0.065 0.060 0.060 0.025 0.025 (7.83)** (32.68)** (32.73)** (17.67)** (17.66)** Experience squared ­0.001 ­0.001 ­0.001 ­0.000 ­0.000 (­2.53)* (­23.15)** (­23.21) (­14.66)** (­14.71)** Regular worker dummy 0.626 0.075 0.075 (34.32)** (5.75)** (5.74)** Constant 0.27 ­0.460 ­0.462 2.078 2.084 (4.66)** (­16.89)** (­16.93) (77.87)** (78.13)** Number of observations 8,299 13,515 13,515 21,328 21,328 R2 0.29 0.34 0.34 0.25 0.25 Source: Authors' calculations. Note: * = statistically significant at the 5 percent level, ** = statistically significant at the 1 percent level. Figures in parentheses are t-statistics. a. Only those aged 15­29 who had completed middle school were asked about vocational education. double those of an additional year of formal education. Differentiating between formal and informal vocation- al training results in statistically significant returns to formal vocational training of 21 percent, whereas the 82 The Knowledge Economy and Education and Training in South Asia returns to informal training of 3.5 percent are not sig- nificant. These estimates of the average returns to training should be treated cautiously, given the paucity of information about the reported training event and caveats about selectivity bias. Improved estimates of training returns will require house- hold and labor force surveys to collect more detailed infor- mation about training, namely, when training took place (before employment or as part of in-service training), what the duration of training was, and who provided the training (the employer or public or private training institutes). The availability of panel data created by following individuals and their training and earnings experiences over time would also improve the estimation of training returns correcting for se- lectivity bias and unmeasured ability. 6 In-Service Training by Employers Households and individuals take decisions about education, but once individuals get to the world of work, the more de- cisions about postschool skill development are taken jointly with employers. Household and labor force surveys do not typically elicit information on employers and the skills they require; in the best of cases they may ask about industry or employer size. To obtain insights into the factors that shape employers' demand for skills and their in-service training, we turn to firm-level surveys to study the in-service training practices of manufacturing firms in South Asia, their deter- minants, and their consequences for labor productivity and wages.1 In the four South Asian countries under review, compa- rable information on in-service training was elicited from em- ployers as part of the World Bank's ICSs (box 6.1). The ICSs 1. This section draws heavily on Tan and Savchenko (2005, 2006). 83 84 The Knowledge Economy and Education and Training in South Asia Box 6.1 Investment Climate Surveys The World Bank has carried out ICSs in more than 40 de- veloping countries. In addition to a wealth of information about firm characteristics, production, wages, and the business environment, ICSs also collected data on enter- prise innovation, research and development, use of new technologies, and workers' education and training. The training questions elicited information on formal training provided by employers and number of workers trained by occupation and source of training, and they distinguished between in-house training and training obtained from vari- ous external training providers, both public and private. asked employers detailed questions about their workforce and training practices. These data, together with information about different enterprise attributes and production, allow us to ask not only which firms provide in-service training, who they train, how much training they provide, and the source of the training, but also to examine the productivity and wage outcomes of training. The World Bank has undertaken similar ICSs in many developing countries, therefore the in-service training prac- tices of South Asian firms can be compared with those of similar firms in other countries. Such comparisons across countries can provide insights into whether or not the inci- dence of in-service training in South Asia is low, and if it is, can help policy makers design training policies to remedy identified weaknesses. We ask several questions: How much in-service training goes on in manufacturing enterprises and do firms in South In-Service Training by Employers 85 Asia train more or less than their competitors, both region- ally and globally? If levels of enterprise training are low, what factors constrain employers from providing training to their employees? Who are the main providers of in-service training: employers, public training institutions, private sector provid- ers, or other firms? What are the factors that shape employ- ers' decisions to provide employees with training? Is invest- ing in in-service training worthwhile in terms of improving firms' productivity and is it beneficial to workers in the form of higher wages? Figure6.1compareslevelsofin-servicetraininginBangla- desh, India, Pakistan, and Sri Lanka. Estimates are presented with and without adjustments to reflect differences in the firm size distribution of ICS samples across countries, in particular, Figure 6.1 Incidence of In-Service Training in Selected South Asian Countries and Years (percentage of surveyed firms) Figure 6.1 Incidence of In- 40 Service Train- ing in Selected 35 unweighted South Asian Countries and 30 weighted ms Years fir 25 veyed sur 20 of 15 centage Per10 5 0 Pakistan India Bangladesh Sri Lanka Source: ICSs. 86 The Knowledge Economy and Education and Training in South Asia that the Bangladesh ICS includes a higher proportion of large firms, which tend to train, while the India ICS has a more rep- resentative sample of firms of different sizes. The simple, un- weighted tabulations suggest that at 37 percent, the incidence of in-service training is highest in Sri Lanka, followed by Ban- gladesh (26 percent), India (17 percent), and Pakistan (8 per- cent). The weighted incidence of in-service training using the size distribution of India as the norm yields the same country rankings, but reduces cross-country disparities. As figure 6.2 shows, compared with other regions, the in- cidence of training in South Asia is among the lowest in the world, being almost half the average for Europe and Central Figure 6.2 Incidence of Formal In-Service Training in Manufacturing, Regional Averages, Selected Years (percentage of surveyed firms) Figure 6.2 Incidence of 60 Formal In-Ser- vice Training in Manufacturing, 50 Regional Aver- ages, Selected ms fir Years 40 veyed sur 30 of centage 20 Per 10 0 Middle East South Asia Europe and Africa Latin America East Asia and North Africa Central Asia and the Caribbean and the Pacific Source: ICSs. In-Service Training by Employers 87 Asia and less than half the average for East Asia and the Pa- cific, Latin America and the Caribbean.2 This training deficit is especially pronounced when South Asian countries are com- pared with individual East Asian countries such as China and Malaysia (figure 6.3). If an educated and trained workforce is critical for technological change and for the knowledge econ- omy, then low levels of education and this postschool training deficit put South Asia at a distinct competitive disadvantage relative to its neighbors in East Asia. The ICSs in all four South Asian countries included ques- tions about which groups of workers received in-service train- ing and how many were trained. Table 6.1 tabulates the per- centage of workers receiving in-service training in each of four groups -- managers, professionals, production workers, and nonproduction workers -- separately by country and weight- ed by firm size to make the estimates comparable across the four countries. Table 6.1 Share of Workers Trained by Skill Group, Selected South Asian Countries and Years Production Nonproduction Level of education Managers Professionals workers workers Bangladesh (2002) 1.9 3.0 1.2 0.4 India (2002) 6.0 7.3 7.0 2.9 Pakistan (2002) 2.0 3.5 3.3 0.4 Sri Lanka (2004) 10.4 11.3 22.4 6.0 Source: ICSs. Note: Estimates are weighted using India's firm size distribution. 2. The cross-country and regional averages shown in figures 6.2 and 6.3 are based upon ICS data from 35 countries and a total of 17,941 firm respondents. An earlier foot- note lists countries and sample sizes in each region. 88 The Knowledge Economy and Education and Training in South Asia Thecross-countryrankingsoftheshareofworkerstrained, or training intensity, vary with per capita income and years of schooling of the workforce in the country. Sri Lanka has the highest training intensity, followed by India, Pakistan, and Bangladesh. How do these estimates for South Asia compare with the level of in-service training for different groups of workers in the fast-growing economies of East Asia. A World Bank (1997) study of Malaysian manufacturing estimated that in 1994, the overall proportion of workers receiving formal in-service training was 22 percent, or 24 percent of managers, 32 per- cent of technicians, and between 13 and 16 percent of produc- tion workers. South Asian employers are apparently not only less likely to provide in-service training to their workers than employers in other regions, but those that do provide train- ing extend training opportunities to a smaller fraction of their workforce than their counterparts in other regions, especially those in East Asia. This training deficit in terms of the propor- tion of workers trained is especially significant in Bangladesh and Pakistan. In relation to the main sources of in-service training in South Asia, table 6.2 presents information for Bangladesh, India, Pakistan, and Sri Lanka. Conditional on a positive re- sponse to the in-service training question, employers were askedaboutwhethertrainingwasprovidedoncompanyprem- ises (henceforth referred to as in-house training) or at off-site locations by external training providers such as universities or VET schools and training institutes (henceforth referred to as external training). For convenience, these external sources of training may be clustered into two groups: public training pro- viders (universities, VET schools, and government institutes) In-Service Training by Employers 89 school 31.1 46.3 34.73 n.a. VET institute 19.8 53.1 49.93 41.3 training Private nal earY exter nment and oviding 17.6 34.7 33.91 59.1 pr institute Gover those for Countries ce tner Sour par 25.7 10.2 18.66 15.9 Asian Private South 6.9 10.2 29.20 7.6 University 2002. Selected LFS ce, Lanka, Sour training Sri by nal 13.1 8.0 5.04 18.0 2004; Exter LFS rainingT ovided pr training Pakistan, 17.7 13.8 6.63 oviding 15.7 training 2004; Pr mal In-house 60 ms For NSS Fir of India, Any 24.1 16.9 8.15 25.0 1995; centage BHIES Per applicable. (2002) not 6.2 (2002) (2004) = y Bangladesh, (2002) n.a. ableT Lanka ce: Countr Bangladesh India Pakistan Sri Sour Note: 90 The Knowledge Economy and Education and Training in South Asia and private sector training providers (private training insti- tutes and partner firms). Several points stand out from table 6.2. First, while enter- prises in all four South Asian countries rely on both in-house and external training providers, with the exception of Sri Lan- ka, in-house programs are a more common source of training than external training courses. Second, while firms in all four countries use a mix of public and private sources for their ex- ternal training, the most common providers are government training institutes in Sri Lanka, private training institutes in India and Pakistan, and private sector partner firms in Ban- gladesh. Constraints to Investing in Training What accounts for the relatively low levels of in-service train- ing in South Asia? The literature has suggested two broad sets of hypotheses. First, the business environment may not be conducive to investments of any kind, whether physical or human. Second, specific market or policy-induced failures may inhibit employers from making socially optimal levels of investment in worker training. Figure 6.4 shows how firms in South Asia rank the severity of different investment cli- mate constraints. All four countries rank tax rates, economic and regulatory uncertainty, and access to finance as the top three constraints todoingbusiness.Theskillsandeducationofavailableworkers are not ranked as being as constraining as other factors such as access to land, transportation, and telecommunications, sug- gesting that South Asian employers may not yet recognize the importance of workers' skills for improving productivity. By In-Service Training by Employers 91 Figure 6.3 Incidence of Formal In-Service Training in Manufacturing by Selected Countries, Selected Years Figure 6.3 Incidence of Formal In- China Service Training in Manu- Brazil facturing by Selected Peru Guatemala Countries, Selected Years El Salvador Kenya Malaysia Tanzania Algeria Serbia Montenegro Sri Lanka Zambia Ethiopia Bangladesh India Morocco Philippines Egypt, Arab Rep. of Indonesia Pakistan 0 10 20 30 40 50 60 70 80 90 100 Percentage of surveyed firms Source: ICSs. contrast, Malaysian employers ranked skills availability as the top constraint (World Bank 2005a). The South Asian ICSs did not elicit information on why employers might invest little in training, but this informa- tion is available in the world business environment surveys (WBESs) for a broad range of developing countries (Batra and Stone 2004).3 The WBESs asked respondents to rank a series of statements about what factors influenced their decisions to 3. The WBES was an enterprise survey fielded using a standard core questionnaire to more than 10,000 firms in 80 countries between late 1998 and mid-2000 to investi- gate issues concerning the investment climate and firm performance. The analyses reported in Batra and Stone (2004) are based on a special survey module adminis- tered in 28 of the WBES countries that focused on competition, trade, technology, and worker training. 92 The Knowledge Economy and Education and Training in South Asia Figure 6.4 Rankings of Investment Climate Constraints, Rated Severe or Figure 6.4 Rankings of Investment Climate Very Severe, Selected South Asian Countries and Years Constraints, Rated Severe or Very Severe, Selected South Asian tax rates Countries and Years Bangladesh India economic and regulatory policy uncertainty Pakistan access to finance Sri Lanka (e.g., cost of collateral) labor regulations skills and education available to workers transportation access to land telecommunications 0 10 20 30 40 50 percentage of firms that rated constraints as severe or very severe Source: ICSs. invest in training workers. Figure 6.5 graphs these rankings separately for firms that train (using in-house or external fa- cilities) and for those that do not. Firms that do not train are substantially more likely than firms that do train to agree with the following key reasons for not training. First, a majority of firms that did not provide training identified the technologies they were using as mature, and hence indicated that their staff did not require training or skills upgrading to use new technology. Second, many not- ed that they could not afford training because of limited re- sources, which might suggest a weakness in financial markets. Third, many alluded to the high labor turnover of trained staff, an externality that prevented firms from recouping the costs of In-Service Training by Employers 93 Figure 6.5 Ranking of Reasons for Not Providing In-Service Training, Selected Developing Countries and Years Figure 6.5 Ranking of Reasons for Not Providing In-Service Training, Selected Developing mature technology used Countries and Years and new workers proficient unaffordable because of firm's limited resources cost of high labor turnover adequate in-house informal training skilled workers hired elsewhere lack of knowledge firms that used external about techniques facilities for training adequate skills firms that used their own acquired from school facilities for training firms that did not skepticism about provide training benefits of training 1.00 1.50 2.00 2.50 3.00 3.50 4.00 extent of agreement with reason (5 = maximum agreement) Source: Batra and Stone 2004. trainingemployees.Finally,manyemployersopinedthatinfor- mal on-the-job training was adequate or that skilled workers were readily available, which suggests low skill requirements, possibly because of the use of mature technologies. Separate WBES tabulations by region indicate that the small sample of firms from South Asia that participated in the WBES cited the same key constraints. Correlates of In-Service Training To provide insights into the possible roles that integration into global markets and the knowledge economy play in pro- 94 The Knowledge Economy and Education and Training in South Asia viding incentives for employers to provide in-service train- ing, figure 6.6 compares the incidence of training in the four South Asian countries by crude proxy variables for enterpris- es' export orientation and level of technology. A firm's export orientation is measured by an indicator variable that takes a value of 1 if the firm exports and 0 otherwise, and its tech- nology level is captured by an indicator variable for whether enterprises engage in research and development (R&D).4 Figure 6.6 suggests that firms in South Asia that export or are engaged in R&D activities are more likely to report in-ser- vice training than those that do not. The differential incentive to train by export status is most apparent for India, Pakistan, and Sri Lanka. For Bangladesh, the incidence of training is not strongly correlated with export orientation. Export orienta- tion can motivate firms to provide training so they can pro- duce high-quality products that meet the exacting standards of foreign buyers and also increase their labor productivity to meet competitive pressures (Batra and Stone 2004; Tan and Batra 1995). In addition, the second panel of figure 6.6 strong- ly indicates that the incidence of in-service training is higher in enterprises that engage in R&D activities, a result that holds true equally across all four South Asian countries. This rela- tionship between training and technology is consistent with studies that suggest that effective use of new technology re- quires a more skilled and trained workforce (Bell and Pavitt 1992; Enos 1962). The importance of these and other training correlates can be investigated within a regression framework using a probit 4. Studies have used several proxy measures for technological capabilities, including investments in R&D, the percentage of the workforce dedicated to R&D, the presence of technology licensing agreements, the recent introduction of new products, and the adoption of new technologies within the last three years. In-Service Training by Employers 95 Figure 6.6 Incidence of Formal Training by Exports and Research & Development, Selected South Asian Countries and Years 40 Figure 6.6 Incidence of Formal Training by Exports and export no 35 Research & Development, Selected South export yes Asian Countries and Years ms 30 fir 25 espondingr 20 of 15 centage Per 10 5 0 Pakistan Bangladesh India Sri Lanka 45 40 R&D no 35 R&D yes ms fir 30 25 espondingr of 20 15 centage Per10 5 0 Pakistan India Bangladesh Sri Lanka Source: ICSs. 96 The Knowledge Economy and Education and Training in South Asia model. The advantage of regression analysis over simple com- parisons is that the independent effects of each variable or set of variables can be analyzed holding the effects of other hy- pothesized correlates constant. The probit model estimates the probability of in-service training by regressing the (0,1) in- dicator variable "any formal training" on a set of explanatory variables, including measures of firm size, exports, technology level, public sector or foreign ownership, workforce character- istics such as education, and unionization status. Table 6.3 re- ports the results of these probit regressions, and the estimated coefficients can be interpreted as the partial probabilities of training from a unit change in the explanatory variables. Several points emerge from table 6.3. First, the incidence of training rises with establishment size, a common finding for all countries for which data are available, and reflects size-re- lated differences in access to finance, scale economies in train- ing provision, education levels of workers, managerial capa- bilities, and use of new technologies. Second, some support is found for the hypotheses that the demand for in-service train- ing is shaped by export orientation and technology. For India and Sri Lanka, both variables are positive and statistically sig- nificant; for Bangladesh, exports are positive and marginally significant; and for Pakistan, technology is positive and statis- tically significant. Table 6.3 also indicates that formal education and post- school training are complementary forms of human capi- tal. The probability of training rises with the average years of schooling attainment of the firm's workforce, a result consis- tent with the empirical evidence from many developing coun- tries.5 Educated workers are not only more productive when 5. See Tan and Batra (1995) for estimates of the relationship between education and training from five developing countries in East Asia and Latin America and Tan In-Service Training by Employers 97 Table 6.3 Probits of Any Formal In-Service Training, Selected South Asian Countries Probability of any formal training Dependent variable Bangladesh India Pakistan Sri Lanka Small firms (16­100 workers) 1.24 0.58 0.11 0.11 (2.80)*** (5.02)*** (0.37) (0.29) Medium firms (101­250 workers) 1.29 0.88 0.84 0.23 (2.88)*** (5.10)*** (2.06)* (0.53) Large firms (more than 250 workers) 1.56 1.40 1.55 0.97 (3.42)*** (7.25)*** (3.57)*** (2.09)* Average years of education 0.03 0.02 0.04 0.03 (1.89)* (0.99) (2.07)* (2.25)* Education of general manager ­0.05 ­0.51 0.35 0.11 (­2.84)*** (­3.13)*** (3.19)*** (2.84)*** Share of female workers ­0.16 0.19 1.54 0.18 (­0.65) (0.68) (2.51)** (0.63) Age of the firm 0.00 0.00 0.01 ­0.01 (­1.01) (0.14) (1.93)* (­1.50) Unionization dummy 0.55 0.22 0.36 ­0.34 (4.09)*** (1.71) (1.08) (­1.48) Export dummy 0.24 0.33 0.39 0.53 (1.79)* (3.04)*** (1.44) (2.57)** R&D dummy 0.15 0.27 0.47 0.60 (1.31) (2.61)** (2.23)* (2.31)** Foreign ownership dummy ­0.29 0.29 0.29 0.03 (­1.04) (1.19) (0.55) (0.15) Government ownership dummy 1.04 0.53 0.30 (2.11)* (2.06)* (1.14) Intercept term ­1.61 ­1.60 ­10.61 ­1.95 (­3.08)*** (­4.18)*** (­5.24)*** (­3.71)*** Number of observations 1,426 974 771 411 R2 0.22 0.09 0.53 0.24 Source: Authors' calculations. Note: * = statistically significant at the 10 percent level, ** = statistically significant at the 5 percent level, *** = statistically significant at the 1 percent level. The figures in parentheses are t-statistics. 98 The Knowledge Economy and Education and Training in South Asia performing given tasks, but they benefit more from training than less educated workers. A related hypothesis -- that more educated managers understand the benefits of training and are therefore more likely to implement in-service training -- found mixed support. Firms with more educated general man- agers in Pakistan and Sri Lanka were more likely to train; in Bangladesh and India, the opposite and counterintuitive result was found. Finally, the share of females in the workforce was not significantly related to the likelihood of training, except in Pakistan, where the greater the share of female employees, the greater the likelihood that the firm would provide training. Productivity and Wage Outcomes of Training Providing in-service training only makes sense if employers' investments in training their employees and upgrading their skills yield positive returns in the form of higher productiv- ity and profits.6 In making these investment decisions, em- ployers also need to decide where to obtain this training and who should receive it. Important considerations will be what type of training has the highest impact on the bottom line and which workers will benefit the most from the training. If training results in positive impacts on productivity, em- ployers also need to determine whether, or to what extent, to share these productivity gains with workers in the form of higher wages. This calculus will depend on the transferability of skills gained from training to other potential employers (2000) and World Bank (1997, 2005a) for related training analyses for Malaysia. 6. Cross-sectional studies have found a strong positive association between in-service training and firms' productivity and wage levels (Batra and Stone 2004; Tan and Ba- tra 1995). In-Service Training by Employers 99 (Acemoglu and Pischke 1998; Becker 1975; Tan 1980). We address these questions using the ICS data for the four coun- tries under review. For the productivity analysis, we use a simple production function approach. The dependent variable -- the logarithm of value added -- is regressed on the logarithms of capital (book value of physical plant and equipment assets), employ- ment, measures of in-service training, a set of control variables for worker attributes (mean years of education), location, and industry. In-service training is a choice variable, and employ- ers make decisions about whether or not to provide training to their workforce based not only on an economic calculus of the profitability of such an investment, but also on its own unob- served (to the researcher) productivity attributes. To the extent that more productive firms are also more likely to train, these latter attributes can give rise to biased estimates of the returns to training. We recognize the potential for selectivity bias, but given the complexity of addressing it consistently across coun- tries, decided to proceed with a single equation estimate of the production function that treats training as being exogenously determined.7 Theanalysisexperimentedwithalternativemeasuresofin- service training: simple (0,1) indicator variables for any formal training, in-house company versus external training by public or private sector providers, as well as the same training vari- 7. Addressing the selectivity bias in training is complex, especially with cross-sectional data such as the ICSs. This issue is more tractable with panel firm data, which pro- vide longitudinal information on the same firms over time. Repeated data on the training and productivity of the same firms allow researchers to factor out firms' un- observed ability attributes and estimate the effects of changes in training practices on productivity growth free of selectivity bias. Panel studies of training that report posi- tive effects of in-service training on productivity growth and wages include Dearden, Reed, and Van Reenen (2006) for the United Kingdom; Tan (2000) for Malaysia; and Tan and Lopez-Acevedo (2003) for Mexico. 100 The Knowledge Economy and Education and Training in South Asia ables measured in terms of the proportion of workers trained. These latter training measures were included to investigate the possible productivity ramifications of making training avail- able to only a few workers. Tables 6.4 and 6.5 report the results of this productivity analysis for the four countries. Before turning to the training results, some parameters es- timated by these models are noteworthy. First, the estimated production function coefficients of capital and labor are posi- tive and statistically significant and resemble those estimated for many other countries. Second, consistent with the belief that education raises firm-level productivity, the results for Bangladesh and India indicate that increased educational at- tainment of the firm's workforce by one year is associated with higher levels of firm-level productivity: 3.5 percent for Ban- gladesh and 5.8 percent for India (the results for both Pakistan Table 6.4 Training and Productivity Results, Simple Indicator of Any Formal In-Service Training, Selected South Asian Countries (dependent variable = log([value added]) Explanatory variable Bangladesh India Pakistan Sri Lanka Log(capital) 0.247 0.216 0.290 0.162 (14.05)*** (14.36)*** (8.44)*** (5.31)*** Log(labor) 0.767 0.849 0.700 0.786 (24.09)*** (27.21)*** (12.59)*** (13.71)*** Mean years schooling 0.035 0.058 0.0002 0.017 (3.93)*** (5.83)*** (1.32) (1.52) Formal training indicator 0.066 0.156 0.667 0.364 (1.03) (1.78)* (3.23)*** (2.72)*** Intercept 10.186 11.254 14.026 11.342 (58.52)*** (49.96)** (19.89)*** (32.27)*** Number of observations 969 1,790 892 374 R2 0.108 0.662 0.507 0.743 Source: Authors' calculations. Note: * = statistically significant at the 10 percent level, ** = statistically significant at the 5 percent level, *** = statistically significant at the 1 percent level. The figures in parentheses are t-statistics. In-Service Training by Employers 101 Table 6.5 Training and Productivity Results, Share of Workers Training and In-House Versus External Sources of Training, Selected South Asian Countries (dependent variable = log[value added]) Bangladesh India Pakistan Sri Lanka Training Training Training Training measured by measured by measured by measured by share of the Training share of the Training share of the Training share of the Training workforce measured by workforce measured by workforce measured by workforce measured by receiving source receiving source receiving source receiving source Explanatory variable training of training training of training training of training training of training Log(capital) 0.246 0.248 0.216 0.207 0.286 0.290 0.162 0.162 (14.07) (14.07) (14.41) (13.11) (8.27) (8.40) (5.34) (5.30) Log(labor) 0.768 0.767 0.859 0.829 0.741 0.716 0.808 0.768 (24.30) (23.98) (28.32) (25.06) (13.61) (12.84) (14.31) (13.21) Mean education 0.032 0.034 0.058 0.062 0.003 0.003 0.019 0.017 (3.65) (3.82) (5.76) (5.96) (1.71) (1.65) (1.72) (1.58) Training measures Share trained 0.575 0.285 0.351 0.715 (3.36) (1.66) (0.65) (3.19) In-house training 0.089 0.069 0.397 0.151 (1.19) (0.65) (1.62) (0.97) External training ­0.009 0.397 0.113 0.393 (­0.11) (2.96) (0.37) (2.53) Constant 10.202 10.189 11.217 11.360 13.972 13.97 11.27 11.418 (58.94) (58.21) (50.03) (48.41) (19.71) (19.74) (32.30) (32.25) Controls Missing values Yes Yes Yes Yes Yes Yes Yes Yes City Yes Yes Yes Yes Yes Yes Yes Yes Industry Yes Yes Yes Yes Yes Yes Yes Yes Number of observations 969 969 1,790 1,660 892 892 374 374 R2 0.710 0.708 0.662 0.648 0.501 0.503 0.745 0.747 Source: Authors' calculations. Note: The figures in parentheses are t-statistics. and Sri Lanka were not statistically significant). Third, several characteristics of firms -- those with a smaller share of female workers, those that undertake R&D, those with some foreign ownership, and unionized firms -- tend to be associated with higher productivity levels across countries, with mixed results for export-oriented firms (not reported here). 102 The Knowledge Economy and Education and Training in South Asia In-service training is typically associated with higher productivity across South Asian countries. In table 6.4, where in-service training is measured by a simple (0,1) indicator variable, its productivity effect is always positive, although the magnitude of the estimated impact and its significance level vary: 67 percent for Pakistan and 36 percent for Sri Lan- ka, both significant at the 1 percent level; 16 percent for In- dia, significant at the 10 percent level; and 7 percent for Ban- gladesh, though not significantly different from zero. In table 6.5, when training is measured by the share of the workforce receiving training, its effect on productivity is positive and statistically significant only for Bangladesh and Sri Lanka. In Pakistan, the effect of the share of workers trained is posi- tive, though not significant, which is curious given the strong positive result using a simple indicator measure. When train- ing is distinguished by source), only external training has a positive productivity impact and then only for India and for Sri Lanka. In-house training also has a positive estimated ef- fect, but never attains statistical significance. For the wage analysis, we use a wage model that exploits all the occupation-specific information elicited in the ICSs. For each of five skill groups -- managers, professionals, skilled production workers, unskilled production workers, and non- production employees -- firms reported not only the num- ber of workers trained (though not by source), but also mean monthly wages. This means that the wage model can be esti- mated for the pooled sample of occupations across all firms that had usable occupation-specific information on numbers trained, wages, and number of workers in that occupation. In the wage model, the logarithm of hourly wages per worker is regressed on the training variables and a set of control vari- ables for occupation, worker attributes (years of education, In-Service Training by Employers 103 age, tenure, and proportion of female workers), firm size, ex- port and R&D indicators, unionization, and industry. Tables 6.6 and 6.7 report the regression results for the wage model. Controlling for location and industry, firm char- acteristics have mixed effects on wages, usually higher in larg- er firms (Bangladesh), firms that are unionized (Bangladesh and Sri Lanka), and firms that are export-oriented (Pakistan). However, in all the countries wage premiums are not associ- ated with firms undertaking R&D. In terms of worker char- acteristics, employers pay higher wages for a more educated and experienced workforce (education effects are particularly significant), but tend to pay lower wages when the workforce is predominantly female. Compared with the omitted skill group (skilled production workers8), managers and professionals are paid more, while unskilled and nonproduction workers re- ceive lower pay. Relative wages across these broad occupations appear to be similar across the countries. In-service training has mixed effects on wages in South Asia. When training is measured as a simple indicator variable for the receipt of any formal in-service training, its wage effects never attain statistical significance except in Sri Lanka (table 6.6). Similarly, when the source of training is distinguished, again using indicator variables, only in-house training is sta- tistically significant, and then only for Sri Lanka, where any training is associated with positive wage gains. When training ismeasuredbytheproportionofworkerstrainedineachoccu- pational group, training is associated with positive and statisti- cally significant wage gains in both Bangladesh and Sri Lanka, where previously no significant wage effects were found using 8. In India, the ICS does not distinguish between skilled and unskilled production workers, so the omitted skill group consists of production workers. 104 The Knowledge Economy and Education and Training in South Asia Table 6.6 Training and Wages Results, Selected South Asian Countries (dependent variable = log[hourly wage]) Explanatory variable Bangladesh India Pakistan Sri Lanka Firm characteristics Small firms 0.19 0.03 0.08 0.06 (2.63)** (0.57) (1.33) (0.48) Medium firms 0.29 0.20 0.16 ­0.14 (3.80)*** (0.99) (1.34) (­0.86) Large firms 0.24 ­0.01 ­0.44 ­0.01 (2.92)*** (­0.02) (­2.04)* (­0.05) Exporter indicator 0.05 ­0.02 0.20 0.13 (1.09) (­0.25) (2.96)*** (1.68) R&D indicator 0.02 0.06 0.04 0.00 (0.52) (0.55) (0.83) (0.00) Unionization indicator 0.09 0.11 0.00 0.24 (1.97)* (0.59) (0.01) (2.54)** Worker attributes Managers 1.45 1.32 0.90 1.17 (40.48)*** (34.89)*** (30.44)*** (22.54)*** Professionals 0.98 0.99 0.65 0.80 (32.34)*** (25.08)*** (19.88)*** (12.55)*** Unskilled workers ­0.42 ­0.32 ­0.32 (­16.98)*** (­9.62)*** (­5.82)*** Nonproduction workers ­0.17 ­0.19 ­0.24 ­0.06 (­5.72)*** (­5.62)*** (­7.53)*** (­0.84) Mean years of education 0.02 0.07 0.02 0.01 (3.23)*** (3.69)*** (1.69) (2.21)* Any training indicator 0.06 ­0.12 0.00 0.20 (1.47) (­0.90) (0.00) (2.65)** Share of female workers ­0.22 ­0.59 0.00 ­0.19 (­2.44)** (­2.03)* (0.00) (­1.73) Mean years job tenure 0.01 0.00 0.02 (1.26) (0.08) (2.63)** Constant term 2.00 2.50 17.20 3.68 (19.57) (11.02)*** (142.37)*** (19.93)*** Missing values indicator Yes Yes Yes Yes City indicator Yes Yes Yes Yes Industry indicator Yes Yes Yes Yes Number of observations 3,012 3,076 3,175 1,263 R2 0.55 0.32 0.37 0.39 Source: Authors' calculations. Note: * = statistically significant at the 10 percent level, ** = statistically significant at the 5 percent level, *** = sta- tistically significant at the 1 percent level. The figures in parentheses are t-statistics. In-Service Training by Employers 105 Table 6.7 Training and Wages Results, Training by Source and Share of Workers Trained, Selected South Asian Countries (dependent variable = log[wage]) Bangladesh India Pakistan Sri Lanka Training Training Training Training measured by measured by measured by measured by share of the Training share of the Training share of the Training share of the Training workforce measured by workforce measured by workforce measured by workforce measured by receiving source receiving source receiving source receiving source Explanatory variable training of training training of training training of training training of training Indicator In-house training 0.052 ­0.126 0.00 0.138 (1.23) (­0.88) (0.00) (1.91)* External training ­0.006 0.314 0.00 0.096 (­0.11) (1.22) (0.00) (1.28) Intensity Share trained 0.24 ­0.288 0.00 0.379 (3.30)*** (­1.11) (0.00) (4.41)*** Number of observations 3,012 3,012 3,076 3,076 3,175 3,175 1,263 1,263 R2 0.547 0.548 0.370 0.368 0.370 0.370 0.387 0.392 Source: Authors' calculations. Note: The figures in parentheses are t-statistics. the any training indicator. In India and Pakistan, no significant wage effects were found for training however measured. 7 Concluding Remarks Using available household, labor force, and firm-level sur- veys, this study of South Asia, which focused on Bangladesh, India, Pakistan, and Sri Lanka, sought to (a) document and compare trends in the education and postschool training of the workforce in each of the four countries; (b) identify what kinds of economic analyses can be done on the life cycle choices individuals, families, and employers make about ed- ucation, pre-employment VET, and in-service training and on the outcomes of human capital investments on school to work transitions, employment, wages, and productivity; and (c) draw out the implications of globalization and the knowl- edge economy for education and training policy in South Asia. The findings reported here suggest that the South Asia data pertaining to education and training from household, labor force, and firm surveys can yield empirically robust findings and insights that are consistent with economic the- ory. The main findings and policy implications follow. 107 108 The Knowledge Economy and Education and Training in South Asia Demand for and Supply of Formal Education In relation to formal education, the analyses highlighted sev- eral important trends: on progress toward universal primary education and at higher levels of schooling, on gender equal- ity in education, on the profitability of investments in differ- ent levels of schooling, and on what continued high rates of return to schooling imply about the demand for education. · Despite commitment to education and continuous progress, the stock of human capital in South Asia is still low compared with that elsewhere, in particular, East Asia. About half of the adult population in the largest South Asian countries is still illiterate. Except for the Maldives, none of the countries is currently up- grading the skills of its population at a speed that will allow them to catch up quickly with East Asia and the rest of the world over the medium term. Indeed, these gaps may be widening rather than closing relative to some East Asian countries. · Progress has been unequal over time across countries within the region. Sri Lanka is clearly an outlier with its early achievement of universal primary education, and the Maldives is rapidly catching up with Sri Lan- ka. Among all other South Asian countries, Bhutan and Nepal, which started with the lowest educational levels, showed a faster pace of improvement, yet not rapid enough to catch up with Bangladesh, India, and Pakistan. In the near future, however, some of the slower countries will likely catch up with the front- runners with regard to universal access to primary Concluding Remarks 109 education. Only Pakistan still seems to be making slow progress in this direction. · Progress over time has also been uneven in terms of gender equality. Forty years ago, with the exception of Sri Lanka, only a tiny fraction of girls had access to education. Since then, the gender gap has diminished substantially at the primary education level in most of the countries under review, and even disappeared in some of them. The challenge for the future is to repeat this achievement at levels beyond primary education, where the gap is still sizable. · The evidence suggests that investments in formal edu- cation are profitable in all the countries and at all levels of education. Despite well-founded concerns about the low quality of education, having some schooling, even an incomplete primary education, provides individuals with a significant wage gain. Belonging to a scheduled caste or tribe does not affect earnings negatively. For such individuals, the primary issue is access to educa- tional opportunities. · Despite increased investments in education over time, the returns to higher secondary and tertiary-level edu- cation have remained high, and even increased relative to lower levels of schooling, suggesting a rising relative demand for higher levels of education. Education poli- cies have not yet responded to this increased demand. · A large gender gap is apparent in wages for given levels of education and work experience, especially in Ban- gladesh and Pakistan. As levels of education increase, the gender gap is dramatically reduced by significantly higher returns to education for women than for men, 110 The Knowledge Economy and Education and Training in South Asia but the higher relative returns are still insufficient to completely eliminate the gender gap in wages. Unemployment and the School to Work Transition The analyses addressed policy concerns that high unemploy- ment rates among more educated youth might reflect the low quality and limited workplace relevance of education and of- fered an alternative explanation based on time-intensive job search during the school to work transition. · Even though the countries exhibit quite different time trends in open unemployment, the most recent data show a low open unemployment rate overall in South Asia. Open unemployment rates range from 1.5 to 5.0 percent depending on the country, with the exception of Sri Lanka, which has open unemployment rates of 8.9 percent. · When disaggregated, unemployment rates in all the countries rise with education levels, but these gross fig- ures obscure the fact that while more educated youth have higher initial rates of unemployment during the school to work transition, they face lower unemploy- ment rates than other groups as they acquire more work experience. · Unemployment rates also vary by age, and high un- employment rates are essentially a youth problem. Youth unemployment is essentially the outcome of a job search process that underlies school to work tran- sitions by groups that differ in terms of their level of education, with the more educated tending to search Concluding Remarks 111 more intensively for a good job match that requires their specific skills. Postschool Training The analyses of pre-employment and in-service training pro- vide insights into the incidence of and trends in postschool training in the region and its effects on the school to work transition, wages, and for one sector (manufacturing) on productivity. · The available data on postschool VET suggest that in- vestments in VET facilitate school to work transitions and yield wage returns roughly comparable to or larger than those from education. · The incidence of postschool training is still quite low in South Asia, being lowest in Pakistan and highest in Sri Lanka, with the other countries falling in be- tween. Trend patterns (which could only be analyzed for two countries) differ markedly, reflecting overall trends in macroeconomic growth in the two countries. While the incidence of VET in the workforce remained roughly unchanged in Sri Lanka over the past decade, it declined sharply in Pakistan until 2002, after which it started to rise again. · Overall levels aside, the incidence of postschool train- ing rises with levels of educational attainment in all the countries, reflecting the complementarity between education and training observed in other parts of the world. It is particularly low in some sectors, such as wholesale and retail trades, construction, agriculture, 112 The Knowledge Economy and Education and Training in South Asia and hotel and restaurant businesses. It is higher for men than for women, mirroring the gender gap ob- served in formal educational attainment. · The incidence of in-service training in manufacturing firmsinSouthAsiaisamongthelowestintheworld,and is less than half the average for East Asia, Europe and Central Asia, and Latin America and the Caribbean. To the extent that training is associated with productivity growth and is required for technological change, a low incidence of training has negative implications for the competitiveness of the region's countries. The deficit is particularly pronounced when South Asian countries are compared with their competitor countries, such as Malaysia, where the incidence is double that in South Asia, and China, where it is triple that in South Asia. · South Asian employers are not only less likely to pro- vide their workers with in-service training than em- ployers in other regions, but those who do extend training opportunities to a smaller fraction of their workforce. This training deficit in terms of the propor- tion of the workforce trained is especially significant in Bangladesh and Pakistan. · The low level of training suggests that employers may not yet recognize that a lack of worker skills is one of their primary constraints to doing business. Neverthe- less, one can observe an incipient response to incen- tives shaped by integration into global markets and the knowledge economy, as firms that train tend to be larger, export oriented, and innovators. While we can- not draw causal inferences from cross-sectional data, the results also suggest that firms that train are more productive and tend to pay above average wages. Concluding Remarks 113 Implications Despite data limitations, this study provides substantial evi- dence that the demand for highly educated and skilled work- ers in South Asia is increasing more rapidly than the supply. It also shows that concerns about unemployment among the more educated -- which is essentially a temporary school to work transition phenomenon -- should not distract policy makers from investing more in the education of their popu- lations. All the countries are increasingly aware that an educat- ed and trained workforce is critical for technological change and for creating a knowledge economy, but many of them have yet to make education and training high priorities, and many education and training policies and programs have yet to respond fully to the needs of and signals from the labor market. The relatively low levels of education and postschool training put South Asian countries at a distinct competitive disadvantage relative to some of their East Asian neighbors, and the challenge for South Asia is to shift emphasis to high- er levels of education without neglecting the unfinished edu- cation agenda at the primary level. This study did not focus on how education and training policies and programs in the region could be improved to address the skill needs of globalization and the knowledge economy. That is a subject for future research. Considerable research is already under way on improving the quality of and access to primary and secondary education. As concerns highereducation,researchisstartingonissuesofgovernance, financing of tertiary education, and the role of universities in science and technology development. Research on post- school training is even more nascent, and there is a need for 114 The Knowledge Economy and Education and Training in South Asia careful studies on reforming public VET institutions to make them more responsive to the skills needs of employers and the market, on the role of private sector training providers, and on policies and programs to encourage greater provision of in-service training by employers. South Asian governments have considerable scope for improving the kinds of questions labor force and household surveys ask to help policy makers better monitor chang- es over time in education and training investments and la- bor market outcomes. More precise information about the timing and duration of training, about training providers (whether public, private, or part of a government program), and about financing sources could help governments formu- late appropriate VET policies. For many kinds of labor mar- ket analysis, knowing the age at which individuals complete their formal schooling is important so that the extent of the school to work transition (or of time in the labor market) can be determined with greater accuracy. Knowing if individu- als participated in training programs during spells of unem- ployment or did so as part of in-service training sponsored by employers is also useful. Developing countries in other regions, especially Latin America, have incorporated such improvements into their surveys and are now using them to monitor and evaluate their education and training policies and labor market interventions. Appendix 1 Hourly Wage Regressions, India 115 116 The Knowledge Economy and Education and Training in South Asia Females 0.058 (2.17) 0.07 (2.59) 0.408 (10.68) 0.998 (30.29) 1.314 (33.16) 0.126 (3.23) 0.154 (9.85) 0.063 (4.93) 0.021 (10.13) 0.000 (­10.17) 0.637 (34.36) ­0.128 (­4.50) 19,665 0.322 1993­4 Males 0.103 (10.17) 0.22 (22.26) 0.377 (34.15) 0.671 (57.78) 1.094 (70.68) 0.14 (8.90) 0.294 (40.99) ­0.042 (­5.94) 0.046 (46.78) ­0.001 (­38.60) 0.602 (77.53) 0.018 (1.34) 61,856 0.406 NSS All 0.107 (11.21) 0.212 (22.56) 0.384 (35.78) 0.712 (64.14) 1.123 (76.76) 0.161 (10.90) 0.473 (69.78) 0.266 (40.28) ­0.012 (­1.98) 0.04 (45.50) ­0.001 (­38.92) 0.620 (85.46) ­0.392 (­30.75) 81,521 0.447 Females 0.086 (2.78) 0.161 (5.11) 0.428 (10.65) 1.072 (34.68) 1.56 (45.72) 0.136 (4.14) 0.173 (10.38) 0.175 (11.74) 0.035 (15.80) ­0.001 (­13.86) 0.568 (27.54) ­0.216 (­6.61) 13,756 0.455 1987­8 Males 0.143 (10.02) 0.192 (14.49) 0.377 (26.97) 0.673 (49/95) 1.197 (76.87) 0.182 (12.22) 0.151 (13.71) 0.013 (1.28) 0.058 (48.26) ­0.001 (­35.42) 0.591 (58.99) 0.119 (6.25) 33,812 0.409 NSS All 0.177 (13.57) 0.23 (18.77) 0.42 (31.59) 0.762 (61.08) 1.288 (89.53) 0.195 (14.05) 0.486 (55.86) 0.186 (20.66) 0.07 (8.54) 0.051 (48.76) ­0.001 (­37.69) 0.612 (66.93) ­0.388 (­24.95) 47,568 0.526 Females 0.058 (2.17) 0.07 (2.59) 0.408 (10.68) 0.998 (30.29) 1.314 (33.16) 0.126 (3.23) 0.154 (9.85) 0.063 (4.93) 0.021 (1013) 0.000 (­10.17) 0.637 (34.36) ­0.128 (­4.50) 19,665 0.322 -statistics.te ar 1983­4 Males 0.103 (10.17) 0.22 (22.26) 0.377 (34.15) 0.671 (57.78) 1.094 (70.68) 0.14 (8.90) 0.294 (40.99) ­0.042 (­5.94) 0.046 (46.78) ­0.001 (­38.60) 0.602 (77.53) 0.018 (1.34) 61,586 0.406 entheses NSS par in All es 0.107 (11.21) 0.212 (22.56) 0.384 (35.78) 0.712 (64.14) 1.123 (76.76) 0.161 (10/90) 0.473 (69.78) 0.266 (40.28) ­0.012 (­1.98) 0.04 (45.50) ­0.001 0.62 (­38.92) (85.46) ­0.392 (­30.75) 81,521 0.447 Figur. y vey sur y secondar sample ed primar higher dummy vations calculations. variable national below andy squar education obser = worker of Authors' y y dummy dummy dummy NSS tiar cept ce: Dependent Literate, Primar Middle Secondar erT echnicalT Male Urban SCST Experience Experience Regular Inter Number 2R Sour Note: Hourly Wage Regressions, India 117 Females 0.248 (6.83) 0.198 (6.12) 0.414 (11.69) 0.972 (27.27) 1.640 (41.15) 0.162 (4.07) 0.292 (13.83) 0.079 (4.32) 0.047 (19.96) ­0.001 (­15.55) 0.666 (26.17) ­0.180 (­4.86) 8,580 0.529 2004 Males 0.157 (9.13) 0.233 (15.53) 0.439 (30.03) 0.656 (44.01) 1.229 (66.71) 0.179 (9.90) 0.201 (21.99) ­0.014 (­1.54) 0.060 (51.59) ­0.001 NSS (­37.34) 0.815 (78.22) 0.230 (13.21) 30,682 0.519 All 0.195 (12.53) 0.249 (18.38) 0.461 (34.31) 0.717 (52.25) 1.329 (79.64) 0.180 (10.87) 0.446 (47.68) 0.221 (26.06) 0.005 (0.67) 0.056 (53.99) ­0.001 (­39.40) 0.798 (81.86) ­0.219 (­13.29) 39,190 0.546 Females 0.187 (8.52) 0.240 (9.72) 0.398 (14.91) 1.126 (46.18) 1.619 (58.81) 0.292 (6.61) 0.217 (15.31) 0.088 (7.40) 0.046 (27.17) ­0.001 (­23.08) .0620 (34.52) ­0.035 (­1.35) 18,494 0.519 -statistics.te ar 1999­2000 Males 0.156 (15.66) 0.262 (26.23) 0.420 (43.23) 0.729 (72.98) 1.277 (104.55) 0.272 (16.96) 0.215 (34.39) 0.019 (3.03) 0.055 (65.86) ­0.001 (­48.88) 0.679 (92.80) 0.352 (29.15) 61,614 0.520 entheses NSS par in All es 0.181 (19.84) 0.280 (30.08) 0.438 (47.78) 0.800 (86.03) 1.355 (120.79) 0.268 (17.51) 0.423 (68.19) 0.219 (37.78) 0.037 (6.77) 0.052 (70.26) ­0.001 (­53.78) 0.680 (99.06) ­0.075 (­6.57) 80,108 0.552 Figur. y vey sur y secondar sample ed primar higher dummy vations calculations. variable national below andy squar education obser = worker of Authors' y y dummy dummy dummy NSS tiar cept ce: Dependent Literate, Primar Middle Secondar erT echnicalT Male Urban SCST Experience Experience Regular Inter Number 2R Sour Note: Appendix 2 Hourly Wage Regressions, Pakistan 119 120 The Knowledge Economy and Education and Training in South Asia Females 0.213 (1.90) 0.247 (2.75) 0.752 (5.98) 1.506 (20.53) 2.288 (28.69) 0.250 (5.46) 0.062 (10.10) ­0.001 (­7.41) 0.289 (3.28) 2,045 0.402 2000­1 Males 0.071 (3.20) 0.193 (11.31) 0.376 (17.54) 0.691 (39.90) 1.216 (53.16) 0.169 (14.24) 0.062 (38.96) ­0.001 (­29.85) 1.696 (71.00) 14,155 0.266 PIHS All 0.108 (4.69) 0.225 (12.63) 0.421 (18.72) 0.788 (44.38) 1.397 (61.34) 1.089 (63.42) 0.189 (15.73) 0.060 (36.90) ­0.001 (­27.69) 0.581 (21.32) 16,200 0.396 Females 0.428 (1.55) 0.636 (4.78) 0.847 (5.39) 1.270 (15.02) 1.775 (17.92) 0.414 (6.95) 0.052 (7.02) ­0.001 (­5.45) 0.864 (8.02) 776 0.498 1996­7 Males 0.010 (0.29) 0.198 (13.32) 0.393 (22.44) 0.627 (44.87) 1.106 (57.60) 0.196 (19.68) 0.063 (42.44) ­0.001 (­32.84) 1.794 (85.78) 10,813 0.373 PIHS All 0.037 (1.01) 0.223 (14.46) 0.415 (22.79) 0.675 (47.23) 1.174 (60.72) 0.650 (32.89) 0.215 (21.17) 0.060 (39.86) ­0.001 (­30.63) 1.152 (42.76) 11,589 0.405 entheses. Females 0.868 (3.12) 0.212 (1.85) 0.602 (4.27) 0.913 (12.41) 1.455 (17.97) 0.182 (3.37) 0.047 (6.93) ­0.001 (­6.22) 1.061 666 (10.60) 0.444 par in 1993­4 Males 0.152 (4.97) 0.213 (14.48) 0.379 (20.66) 0.619 (43.17) 1.151 (62.50) 0.220 (21.83) 0.058 (39.90) ­0.001 (­31.52) 1.368 (66.29) 9,887 0.417 -statisticst. PIHS vey sur All 0.173 (5.55) 0.220 (14.79) 0.390 (21.00) 0.643 (45.13) 1.183 (65.56) 0.377 (19.14) 0.214 (21.32) 0.056 (38.95) ­0.001 (­31.12) 1.008 (37.95) 10,553 0.422 household y y secondar integrated ed primar higher vations calculations. Pakistan variable below andy squar obser = of Authors' y y dummy dummy PIHS tiar cept ce: Dependent Literate, Primar Middle Secondar erT Male Urban Experience Experience Inter Number 2R Sour Note: Appendix 3 Hourly Wage Regressions, Sri Lanka 121 122 The Knowledge Economy and Education and Training in South Asia Females 0.024 (0.63) 0.059 (1.41) 0.308 (7.32) 0.621 (15.38) 0.909 (17.82) 0.089 (2.47) ­0.19 (­5.97) 0.014 (5.86) 0.000 (­5.64) 0.447 (19.70) 2.466 (49.05) 6,829 0.302 2001­2 Males 0.092 (2.84) 0.245 (7.51) 0.377 (11.47) 0.615 (18.46) 0.849 (18.75) 0.402 (14.95) 0.219 (8.77) 0.033 (19.86) ­0.001 (­17.50) 0.332 (27.22) 2.316 (55.03) 14,009 0.255 LFS All 0.057 (2.36) 0.185 (7.44) 0.341 (13.52) 0.606 (24.08) 0.875 (26.31) 0.403 (40.72) 0.271 (12.69) 0.059 (3.03) 0.026 (18.63) 0.000 (­16.76) 0.362 (33.31) 2.163 (68.73) 20,838 0.292 Females ­0.042 (­1.44) 0.052 (1.54) 0.243 (7.12) 0.532 (16.47) 0.877 (18.82) 0.033 (1.20) ­0.207 (­8.76) 0.013 (5.75) 0.000 (­6.87) 0.365 (18.93) 2.650 (64.12) 7,808 0.282 1997­8 Males 0.071 (2.52) 0.245 (8.54) 0.406 (13.85) 0.591 (19.89) 0.798 (18.05) 0.448 (21.26) 0.216 (11.31) 0.028 (17.20) 0.000 (­14.95) 0.305 (26.01) 2.373 (66.13) 14,421 0.245 LFS All 0.017 (0.86) 0.172 (8.24) 0.337 (15.68) 0.553 (25.85) 0.828 (26.07) 0.381 (40.98) 0.277 (16.64) 0.041 (2.76) 0.023 (17.21) 0.000 (­15.93) 0.324 (32.19) 2.273 (86.64) 23,229 0.281 Females 0.101 (3.76) 0.318 (11.33) 0.610 (22.11) 0.926 (35.48) 1.143 (29.67) ­0.153 (­7.61) ­0.280 (­13.68) 0.031 (15.84) ­0.001 (­13.75) 2.493 (58.54) 12,563 0.249 1992­3 Males 0.062 (2.38) 0.279 (10.76) 0.522 (20.13) 0.857 (33.54) 1.074 (29.72) ­0.134 (­8.55) ­0.259 (­16.18) 0.031 (19.88) 0.000 (­16.06) 2.661 (73.29) 19,168 0.219 LFS entheses. par in All 0.039 (1.87) 0.224 (10.58) 0.485 (23.00) 0.852 (42.03) 1.088 (36.79) 0.302 (31.99) ­0.130 (­9.35) ­0.244 (­17.15) 0.030 (21.34) 0.000 (­18.13) 2.392 (89.11) 24,535 0.258 y -statisticst. vey sur y secondar ce for primar y ed higher dummy vations calculations. variable below andy squar Labor obser = worker of Authors' y secondar y dummy dummy dummy LFS tiar cept ce: Dependent Literate, Primar Lower Secondar erT Male Urban Rural Experience Experience Regular Inter Number 2R Sour Note: Appendix 4 Unemployment Rates by Level of Education and Age Cohort, Economically Active Population Aged 15­64, Selected South Asian Countries and Years Age cohort Country, gender, and level of education 15­19 20­24 25­29 30­34 35­39 40­49 50­64 Total Bangladesh, 2000 Males Illiterate 4.69 0.93 0.56 0.96 0.85 0.48 1.00 1.17 Literate 10.30 3.33 0.00 0.00 1.04 0.00 3.59 2.76 Primary 13.37 5.35 1.81 0.86 0.42 1.33 1.23 3.56 Secondary (grades 6­10) 38.51 4.99 2.09 3.83 0.00 0.81 1.15 3.93 High (grades 11­12) 30.10 17.79 3.30 1.70 2.76 2.28 0.00 4.60 Tertiary 43.07 27.60 17.31 1.15 0.58 0.00 0.95 5.61 Total 9.92 4.60 2.18 1.08 0.75 0.78 1.14 2.41 Females Illiterate 24.01 26.52 20.76 16.71 21.94 17.86 29.83 21.73 Literate 49.47 39.39 17.16 64.57 26.20 0.00 67.33 42.51 Primary 65.09 44.03 56.80 59.28 47.27 60.13 72.65 57.65 Secondary (grades 6­10) 74.97 53.85 28.45 63.64 23.95 12.63 0.00 43.17 High (grades 11­12) 78.83 40.23 62.27 10.06 23.37 0.00 0.00 42.52 Tertiary 100.00 13.60 35.61 5.08 16.11 0.00 58.19 22.61 Total 48.61 34.34 31.32 26.62 25.28 22.13 34.98 31.44 123 124 The Knowledge Economy and Education and Training in South Asia Age cohort Country, gender, and level of education 15­19 20­24 25­29 30­34 35­39 40­49 50­64 Total India, 2004 Males Illiterate 4.85 3.52 2.93 2.56 2.93 2.38 2.60 2.88 Literate 9.06 2.85 1.47 2.56 3.72 2.56 2.55 3.31 Primary 9.84 5.92 4.07 3.53 2.03 2.43 1.90 4.29 Middle 13.44 8.52 6.26 3.05 2.16 2.27 2.26 5.82 Secondary 20.53 15.41 8.10 3.99 2.57 1.56 1.36 6.94 Tertiary 49.81 32.66 17.01 5.87 2.57 0.89 0.42 9.12 Total 11.02 9.79 6.54 3.50 2.62 2.14 2.15 5.00 Females Illiterate 3.38 3.59 3.39 2.78 2.43 1.92 2.15 2.58 Literate 4.96 2.80 1.15 4.20 2.46 1.58 0.77 2.55 Primary 5.48 5.75 4.34 6.02 2.05 2.88 0.72 4.27 Middle 12.81 9.34 6.18 5.55 2.80 2.49 1.38 7.23 Secondary 24.25 26.02 17.87 16.80 5.18 1.36 0.13 17.04 Tertiary 29.43 49.44 36.17 13.11 4.26 2.54 0.00 24.54 Total 8.27 12.07 7.84 4.96 2.64 2.02 1.90 5.22 Age cohort Country, gender, and level of education 15­19 20­24 25­29 30­34 35­39 40­49 50­64 Total Pakistan, 2003­4 Males Illiterate 4.48 2.32 1.37 1.03 0.98 1.06 0.57 1.60 Literate 9.16 3.74 1.84 1.06 1.00 0.57 1.63 3.22 Primary 6.84 2.91 1.91 2.00 1.27 1.07 1.45 2.87 Middle 11.65 8.37 4.41 2.22 1.11 1.62 0.94 5.30 Secondary 18.46 14.23 9.64 4.19 0.66 2.46 2.08 7.92 Tertiary n.a. 21.94 11.65 7.91 2.33 0.76 1.84 6.75 Total 8.42 7.59 4.98 2.82 1.10 1.33 1.01 3.94 Females Illiterate 2.98 3.22 2.61 2.29 2.82 2.39 1.26 2.46 Literate 6.67 12.08 0.00 0.00 0.00 4.88 0.00 5.55 Primary 10.74 12.28 15.65 13.21 16.20 2.98 0.00 11.10 Middle 15.26 8.40 4.54 0.00 0.00 0.00 0.00 7.81 Secondary 27.89 26.00 18.03 17.34 10.05 3.12 0.00 19.75 Tertiary n.a. 35.88 17.67 10.80 0.90 0.00 0.00 15.95 Total 8.42 12.21 7.86 5.35 4.00 2.30 1.17 6.06 Unemployment Rates by Level of Education and Age Cohort 125 Age cohort Country, gender, and level of education 15­19 20­24 25­29 30­34 35­39 40­49 50­64 Total Sri Lanka, 2001­2 Males Illiterate 4.88 6.10 1.30 0.00 0.77 0.00 0.00 0.88 Literate 12.48 7.46 3.25 0.51 0.48 0.83 0.52 1.64 Primary 17.77 10.61 4.20 1.38 0.44 1.31 0.48 3.36 Middle 30.17 19.34 6.82 2.66 1.53 1.47 0.86 8.79 Secondary 44.65 31.26 9.63 3.95 3.09 1.75 0.98 10.04 Tertiary 77.19 2.62 29.29 4.74 0.00 0.00 2.69 5.52 Total 27.40 21.11 7.33 2.50 1.44 1.30 0.73 6.51 Females Illiterate 8.19 3.85 0.00 1.37 0.00 1.12 0.56 1.02 Literate 12.36 8.08 5.48 1.91 2.72 0.97 0.39 2.21 Primary 22.79 14.52 8.08 2.13 2.88 2.82 0.94 5.29 Middle 31.34 21.59 12.25 10.97 3.74 1.60 0.91 12.56 Secondary 48.95 45.07 24.84 11.99 7.96 1.84 1.06 21.41 Tertiary 100.00 34.02 32.26 15.02 3.22 0.00 1.52 12.72 Total 33.53 32.52 18.30 8.84 4.53 1.67 0.77 12.30 Source: Bangladesh, household income and expenditure survey 2000; India, national sample survey 60; Pakistan, labor force survey 2003­4; Sri Lanka, labor force survey pooled 2001 and 2002. Appendix 5 Unemployment Rates by Education and Years of Potential Work Experience, Economically Active Population Aged 15­64, Selected South Asian Countries and Years Years of potential work experience Country, gender, and level of education 0­4 5­9 10­14 15­19 20­24 25­34 > 34 Total Bangladesh, 2000 Males Illiterate 5.54 3.57 0.81 0.57 0.93 0.64 1.17 Literate 9.32 5.26 0.00 0.00 0.63 2.40 2.76 Primary 19.27 7.22 1.83 1.21 0.59 1.09 1.29 3.56 Secondary (grades 6­10) 26.55 1.82 4.54 1.08 1.22 0.12 1.49 3.93 High (grades 11­12) 20.50 12.11 3.00 2.20 1.84 1.63 0.00 4.60 Tertiary 42.50 14.44 5.92 0.59 0.00 0.76 0.00 5.61 Total 22.27 7.91 3.09 1.12 0.67 0.93 1.35 2.41 Females Illiterate 12.59 26.58 20.14 20.74 20.37 23.47 21.73 Literate 0.00 49.47 35.66 28.17 74.78 22.81 51.34 42.51 Primary 71.66 41.64 64.27 47.74 59.80 49.63 76.96 57.65 Secondary (grades 6­10) 73.18 42.91 48.93 22.06 22.57 25.48 0.00 43.17 High (grades 11­12) 62.57 69.85 17.21 12.37 46.11 0.00 0.00 42.52 Tertiary 34.60 28.41 17.56 6.94 8.28 0.00 58.19 22.61 Total 65.85 42.79 51.24 32.00 48.60 36.22 67.13 31.44 127 128 The Knowledge Economy and Education and Training in South Asia Years of potential work experience Country, gender, and level of education 0­4 5­9 10­14 15­19 20­24 25­34 > 34 Total India, 2004 Males Illiterate 2.81 4.58 3.72 2.37 2.86 2.46 2.88 Literate 9.06 3.51 1.79 2.54 3.01 2.60 3.31 Primary 8.88 8.93 5.64 3.58 2.83 2.09 2.39 4.29 Middle 12.72 9.54 6.69 3.46 2.14 2.30 2.27 5.82 Secondary 20.76 11.59 6.24 2.77 1.66 1.60 1.39 6.94 Tertiary 25.90 10.06 2.73 1.71 0.41 0.46 0.59 9.12 Total 18.66 9.76 5.51 3.07 2.18 2.37 2.36 5.00 Females Illiterate 1.20 4.03 3.75 2.83 2.62 1.90 2.58 Literate 5.44 3.16 0.13 5.13 1.48 1.77 2.55 Primary 0.24 7.61 3.53 5.16 4.42 2.16 2.69 4.27 Middle 11.56 10.86 6.86 5.96 2.78 2.85 1.21 7.23 Secondary 25.27 24.02 16.60 11.34 4.75 0.37 0.18 17.04 Tertiary 44.88 22.67 7.57 4.85 0.62 0.00 0.00 24.54 Total 26.29 12.52 6.37 4.57 3.23 2.43 1.91 5.22 Years of potential work experience Country, gender, and level of education 0­4 5­9 10­14 15­19 20­24 25­34 > 34 Total Pakistan, 2003­4 Males Illiterate 3.61 4.43 2.34 1.37 1.01 0.82 1.60 Literate 11.85 5.61 3.83 0.47 0.93 1.24 3.22 Primary 6.56 4.21 1.70 2.15 1.05 1.36 2.87 Middle 9.63 11.11 7.46 3.73 1.41 1.41 1.13 5.30 Secondary 19.94 15.89 11.17 5.52 0.71 2.24 1.95 7.92 Tertiary 22.38 12.39 8.15 2.03 1.31 1.51 0.24 6.75 Total 19.28 11.77 6.77 3.17 1.32 1.31 1.01 3.94 Females Illiterate 3.61 4.43 2.34 1.37 1.01 0.82 1.60 Literate 11.85 5.61 3.83 0.47 0.93 1.24 3.22 Primary 6.56 4.21 1.70 2.15 1.05 1.36 2.87 Middle 9.63 11.11 7.46 3.73 1.41 1.41 1.13 5.30 Secondary 19.94 15.89 11.17 5.52 0.71 2.24 1.95 7.92 Tertiary 22.38 12.39 8.15 2.03 1.31 1.51 0.24 6.75 Total 19.28 11.77 6.77 3.17 1.32 1.31 1.01 3.94 Unemployment Rates by Education and Years of Potential Work Experience 129 Years of potential work experience Country, gender, and level of education 0­4 5­9 10­14 15­19 20­24 25­34 > 34 Total Sri Lanka, 2001­2 Males Illiterate 4.88 6.10 1.30 0.44 0.00 0.88 Literate 10.32 9.40 4.56 0.58 0.63 0.59 1.64 Primary 19.10 13.92 7.74 2.61 1.15 1.23 0.44 3.36 Middle 30.82 19.71 7.15 2.73 1.52 1.47 0.83 8.79 Secondary 42.17 20.43 6.21 3.84 2.63 1.54 0.77 10.04 Tertiary 21.18 29.29 4.74 0.00 0.00 1.39 2.97 5.52 Total 35.67 19.05 7.03 3.22 1.60 1.21 0.61 6.51 Females Illiterate 8.19 3.85 0.00 0.55 0.82 1.02 Literate 13.18 6.20 8.67 2.32 2.25 0.39 2.21 Primary 25.91 23.50 3.97 6.38 3.82 3.07 0.89 5.29 Middle 31.55 22.08 12.37 10.64 4.15 1.60 0.88 12.56 Secondary 52.69 33.89 16.40 11.44 5.93 1.01 1.42 21.41 Tertiary 33.70 31.76 14.79 3.23 0.00 1.55 0.00 12.72 Total 45.57 28.27 13.23 9.62 4.22 1.85 0.72 12.30 Source: Bangladesh, household income and expenditure survey 2000; India, national sample survey 60; Pakistan, labor force survey 2003­4; Sri Lanka, labor force survey pooled 2001 and 2002. Appendix 6 Percentage of the Population Trained by Field of Training and Average Duration of Training, India, 2004 Percentage trained Weeks of Training field All Male Female training Computer trades 25.3 24.6 26.5 45.2 Electrical and electronic engineering trades 15.6 22.9 2.9 83.5 Mechanical engineering trades 12.7 19.4 1.2 95.4 Textile-related work 11.2 1.3 28.5 41.1 Health- and paramedical services-related work 6.5 4.8 9.4 94.6 Driving and motor mechanic work 4.2 6.5 0.2 40.8 Office- and business-related work 3.2 2.1 5.0 44.9 Civil engineering and building construction 3.2 4.4 1.0 107.4 Artisan, craftspersons, cottage-based production 1.7 1.2 2.7 60.9 Childcare, nutrition, preschool, and childcare 1.2 0.0 3.3 38.1 Beautician, hairdressing, and related work 0.9 0.0 2.5 47.2 Noncrop-based agricultural services 0.9 1.3 0.1 60.0 Creative arts, artists 0.8 0.9 0.7 99.3 Catering, nutrition, hotels, and restaurant work 0.8 0.7 0.9 99.8 Chemical engineering trades 0.6 0.6 0.7 68.2 Printing technology-related work 0.6 0.4 0.9 48.5 Agriculture, crop production, food preservation 0.4 0.5 0.1 85.1 131 132 The Knowledge Economy and Education and Training in South Asia Percentage trained Weeks of Training field All Male Female training Photography and related work 0.3 0.5 0.0 59.4 Leather-related work 0.2 0.3 0.1 56.0 Journalism, mass communication, media work 0.1 0.1 0.0 147.9 Other 9.5 7.4 13.2 46.9 Source: National sample survey (2004). Appendix 7 Percentage Trained by Field of Training, Bangladesh 1995 Percentage trained Training field All Male Female Transport mechanic 15.83 16.77 -- Cottage industry 10.11 10.4 5.36 Tailoring, embroidery 7.33 6.25 25.49 Polytechnic 6.47 6.69 2.65 Agriculture, livestock 6.38 5.83 15.59 Weaving 6.04 5.92 8.03 Typing, shorthand 5.52 5.59 4.33 Health, family planning 4.59 3.36 25.38 Electrical-related work 4.15 4.39 -- Computer-related work 0.87 0.92 -- Others 32.72 33.87 13.18 Source: Household income and expenditure survey 1995. Note: -- = not available. 133 Appendix 8 Percentage Trained and Duration of Training by Training Institution, India 2004 Percentage trained Weeks of Training institution All Male Female training Industrial training institutes/industrial training centers 27.31 38.87 7.25 79.2 Tailoring, embroidery, and stitch craft institutes 8.81 0.92 22.49 38.8 Polytechnics 5.85 7.62 2.77 125.1 School offering vocational courses (secondary and 5.21 5.57 4.57 67.1 higher secondary level) Hospital and medical training institutes 2.74 2.93 2.41 105.0 Institutes run by companies and corporations 2.41 2.12 2.91 67.4 Recognized driving schools 2.38 3.63 0.19 19.3 Nursing institutes 2.21 0.20 5.68 103.0 University Grants Commission (first degree level) 1.88 2.08 1.52 99.7 Small industries service institutes, district industries 1.65 1.48 1.93 35.7 centers, toll room centers Nursery teachers' training institutes 1.35 0.06 3.58 41.6 Institutes giving diploma in pharmacy 1.13 1.32 0.80 71.0 Secretarial institute 0.81 0.45 1.43 45.6 Recognized beautician schools 0.56 0.00 1.54 47.1 Institutes for journalism and mass communication 0.48 0.66 0.18 59.7 National open school 0.44 0.38 0.53 56.1 Hotel management institutes 0.41 0.28 0.64 118.1 Handloom, handicraft design training center, Khadi 0.41 0.08 0.98 47.7 and Village Industry Commission Institutes offering training for agricultural extension 0.38 0.60 0.00 73.7 135 136 The Knowledge Economy and Education and Training in South Asia Percentage trained Weeks of Training institution All Male Female training Community polytechnic 0.32 0.27 0.41 35.3 Fashion technology institutes 0.30 0.17 0.51 60.2 Rehabilitation, physiotherapy, ophthalmic, and dental 0.10 0.15 0.01 74.9 institutes Food craft and catering institutes 0.09 0.14 0.00 48.0 Training provided by carpet weaving centers 0.02 0.00 0.07 12.0 Other institutes 32.77 30.00 37.57 44.2 Source: National sample survey (2004). Appendix 9 Number of People with Vocational Education by Training Field and Year, Pakistan Type of training 1993­4 1996­7 1997­8 1999­2000 2001­2 2003­4 Computer course 173,070 370,908 261,398 287,142 131,455 273,480 Driving course 212,173 308,561 284,463 122,727 99,746 217,508 Embroidery and knitting course 79,186 116,129 193,125 62,513 34,001 200,902 Garment making 453,303 390,657 269,708 104,196 47,410 158,904 Electrician 75,506 97,096 66,012 64,974 46,762 108,231 Automobile mechanic course 67,831 57,144 54,318 59,219 48,134 64,088 Carpentry 55,249 71,792 47,208 16,393 34,386 48,684 Mason 166,263 104,353 49,010 23,125 8,778 37,394 Civil engineering technology 25,296 66,862 40,976 32,323 36,212 37,226 Weaving course 100,030 48,266 33,726 23,027 13,845 36,497 Draftsperson 6,413 32,075 26,826 12,320 24,436 34,182 Welding course 23,594 38,767 24,953 24,537 15,884 31,855 Electrical engineering technology 9,376 30,077 28,487 23,162 10,774 27,449 Typing and shorthand course 41,864 41,843 43,502 36,745 21,514 25,480 Pharmacy course 17,496 44,603 27,044 32,248 17,773 24,783 Mechanical engineering technology 19,732 25,073 23,413 17,519 14,449 17,381 Refrigeration and air conditioning 12,183 16,292 5,798 4,763 9,603 16,726 General nursing course 4,120 9,452 1,128 3,562 10,367 15,094 Plumbing and pipe fitting 5,481 13,969 10,777 13,759 12,592 14,107 Automobile and farm machinery 14,164 29,177 13,284 13,576 9,641 13,918 Laboratory technician 2,872 9,240 13,288 9,532 7,150 13,264 L.H.V. course 3,672 10,530 5,406 12,046 3,281 12,965 137 138 The Knowledge Economy and Education and Training in South Asia Type of training 1993­4 1996­7 1997­8 1999­2000 2001­2 2003­4 Diploma in radio and TV 10,562 17,866 15,024 15,159 11,167 11,603 Leather work 91,814 59,416 52,297 5,756 2,413 10,837 Textile technology 12,562 9,361 10,379 11,086 4,071 8,959 Diploma in arts 4,820 9,341 2,777 4,481 1,223 6,828 Cooking course 1,087 5,412 4,900 2,422 3,646 5,537 Midwifery course 5,706 8,835 5,168 7,616 10,840 5,269J Jewelry and embroidery 7,171 7,874 8,333 6,169 1,527 4,606 Machinery course 13,529 42,247 13,308 17,907 5,646 4,347 Woodwork 18,027 29,769 15,888 6,385 4,477 3,454 Polishing and soldering 1,497 8,849 13,128 1,176 811 2,897 Architectural technology 6,709 4,700 5,334 3,659 0 1,442 X-ray technicians 2,339 2,508 4,560 3,057 858 1,141 Livestock and poultry farming course 1,949 0 699 0 170 1,115 Metallurgy and mining technology 11,372 4,926 6,065 4,730 0 74 Ceramics technology 16,096 7,385 0 1,678 2,431 0 Foundry technology 5,880 6,584 0 2,854 0 0 Interior decoration 1,653 1,066 1,094 0 0 0 Diploma in design 668 3,508 2,411 259 3,997 0 Flower making course 0 2,084 1,988 2,069 0 0 Pattern making course 1,571 1,676 1,586 917 3,927 0 Other 0 0 0 0 139,249 196,303 Total 1,783,886 2,166,273 1,688,789 1,096,788 854,646 1,694,530 Source: Pakistan LFSs. Note: The number of people with vocational training was estimated by multiplying the frequencies reporting each field of train- ing by the population weights assigned to survey respondents. References Acemoglu, Daron, and Jorn-Steffen Pischke. 1998. "Why Do Firms Train? Theory and Evidence." Quarterly Journal of Economics 113 (February): 79­119. Barro,RobertJ.,andJong-WhaLee.2000."InternationalData on Educational Attainment: Updates and Implications." Working Paper 42, Harvard University, Center for International Development, Cambridge, MA. Barnow, Burt, Glen Cain, and Arthur Goldberger. 1981. "Issues in the Analysis of Selectivity Bias." In Evaluation Studies Review Annual, ed. William Stromsdorfer and George Farkas, vol. 5, 43­59. Beverly Hills, CA: Sage. Batra, Geeta, and Andrew Stone. 2004. Investment Climate, Capabilities and Firm Performance: Evidence from the World Business Environment Survey. Washington, DC: World Bank, Investment Climate Unit. Becker, Gary S. 1975. Human Capital, 2nd ed. New York: Columbia University Press. Bell, Martin, and Keith Pavitt. 1992. "Accumulating TechnologicalCapabilityinDevelopingCountries."Paper 139 140 The Knowledge Economy and Education and Training in South Asia presented at the World Bank "Annual Bank Conference on Development Economics," April, Washington, DC. Berman, Eli, John Bound, and Stephen Machin. 1998. "Implications of Skill-Biased Technological Change: International Evidence." Quarterly Journal of Economics 113 (4): 1245­80. Blom, Andreas, Lauritz Holm-Nielsen, and Dorte Verner. 2001. "Education, Earnings, and Inequality in Brazil: 1982­1998." Policy Research Working Paper 2686, World Bank, Washington, DC. Dahlman,Carl,andAnujaUtz.2005.IndiaandtheKnowledge Economy: Leveraging Strengths and Opportunities. Washington, DC: World Bank, South Asia Region, Finance and Private Sector Development Unit. Dearden, Lorraine, Howard Reed, and John Van Reenan. 2006. "The Impact of Training on Productivity and Wages: Evidence from British Panel Data." Oxford Bulletin of Economics and Statistics 68 (4): 305­49. De Ferranti, David, Guillermo E. Perry, Indermit Gill, J. Luis Guasch, William F. Maloney, Carolina Sanchez-Paramo, and Norger Schady. 2003. Closing the Gap in Education and Technology. Latin American and Caribbean Studies. Washington, DC: World Bank. Enos,John.1962."InventionandInnovationinthePetroleum Refining Industry." In The Rate and Direction of Inventive Activity, 299­321. Princeton, NJ: Princeton University Press. Giovagnoli, Paula Inés, Ariel Fiszbein, and Harry A. Patrinos. 2005. "Estimating the Returns to Education in Argentina: 1992­2002." Policy Research Working Paper 3715, World Bank, Washington, DC. References 141 Heckman, James. 1979. "Sample Selection Bias as a Specification Error." Econometrica 47 (1): 153­62. Lachler, Ulrich. 1998. "Education and Earnings Inequality in Mexico." Policy Research Working Paper 1949, World Bank, Washington, DC. Mincer, Jacob. 1974. Schooling, Experience, and Earnings. Cambridge, MA: National Bureau of Economic Research. Patrinos, Harry Anthony, Cris Ridao-Cano, and Chris Sakellariou. 2006. "Estimating the Returns to Education: AccountingforHeterogeneityinAbility."PolicyResearch Working Paper 4040, World Bank, Washington, DC. Savchenko, Yevgeniya, and Hong Tan. 2007. "Overview of TVET Trends in Pakistan from Labor Force Surveys." Working Paper, World Bank, South Asia Human Development Unit, Washington, DC. Tan, Hong. 1980. "Human Capital and Technological Change: A Study of Japanese Wage Differentials in Manufacturing." Ph.D. thesis, Yale University, New Haven, CT. ------. 2000. Malaysia Skill Needs Study. Washington, DC: World Bank Institute. ------.2005.TheSkillsChallengeofNewTechnology:Training, Technology, and Productivity Growth in Malaysian Manufacturing in the 1990s. Washington, DC: World Bank Institute. Tan, Hong, and Geeta Batra. 1995. "Enterprise Training in Developing Countries: Incidence, Productivity Effects, and Policy Implications." Occasional Paper 9, World Bank, Private Sector Development Department, Washington, DC. 142 The Knowledge Economy and Education and Training in South Asia Tan, Hong, and Sunil Chandrasiri. 2004. "Training and Labor Market Outcomes in Sri Lanka." Working Paper, World Bank Institute, Washington, DC. Tan, Hong and Gladys Lopez-Acevedo. 2003. "Mexico: In- Firm Training for the Knowledge Economy." Policy ResearchWorkingPaper2957,WorldBank,Washington, DC. Tan, Hong, and Yevgeniya Savchenko. 2005. "In-Service TraininginIndia:EvidencefromtheInvestmentClimate Survey." Background paper for Skills Development in India: The Vocational Education and Training System. Washington, DC: World Bank. ------. 2006. "In-Service Training in Bangladesh." Background paper for The Bangladesh Vocational Education and Training System: An Assessment. Washington, DC: World Bank. United Nations Educational, Scientific, and Cultural Organization.1976.InternationalStandardClassification of Education. Paris: United Nations Educational, Scientific, and Cultural Organization, Division of Statistics on Education. World Bank. 1997. Malaysia: Enterprise Training, Technology, and Productivity. Washington, DC: World Bank. ------. 1999. Sri Lanka: A Fresh Look at Unemployment. Report 19609-CE. Washington, DC: South Asia Region, Poverty Reduction and Economic Management Unit. ------. 2005a. Malaysia: Firm Competitiveness, Investment Climate, and Growth. Report 26841-MA. Washington, DC: World Bank, East Asia and Pacific Region. ------. 2005b. Treasures of the Education System in Sri Lanka: Restoring Performance, Expanding Opportunities, and Enhancing Prospects. Washington, DC: World Bank. References 143 ------.2006a.MalaysiaandtheKnowledgeEconomy:Building a World-Class Higher Education System. Washington, DC: World Bank, East Asia Region. ------. 2006b. World Development Indicators. Washington, DC: World Bank.