CONFERENCE EDITION THE WEB OF TRANSPORT CORRIDORS IN SOUTH ASIA THE WEB OF TRANSPORT CORRIDORS IN SOUTH ASIA THE WEB OF TRANSPORT CORRIDORS IN SOUTH ASIA © 2018 International Bank for Reconstruction and Development / The World Bank © ADB. Copyrighted material from the following ADB working papers is used with permission from ADB in this work: P. Warr and A. Kohpaiboon. 2017. “Thailand’s Automotive Manufacturing Corridor.” ADB Economics Working Paper Series. No. 519. Manila: Asian Development Bank. P. Athukorala and S. Narayanan. 2017. “Economic Corridors and Regional Development: The Malaysian Experience.” ADB Economics Working Paper Series. No. 520. Manila: Asian Development Bank. Some rights reserved This work is a co-publication of The World Bank, ADB, JICA, and DFID. The findings, interpretations, and conclusions expressed in this work do not necessarily reflect the views of The World Bank, its Board of Executive Directors, or the governments they represent; or those of ADB, its Board of Governors, or the governments they represent; JICA; or DFID. The World Bank, ADB, JICA, and DFID do not guarantee the accuracy of the data included in this work. This publication follows the World Bank’s practice in references to member designations, borders, and maps. The boundaries, colors, denominations, and other information shown in any map in this work do not imply any judgment on the part of The World Bank, ADB, JICA, or DFID concerning the legal status of any territory or the endorsement or acceptance of such boundaries. In the Asian Development Bank, China is referred to as the People’s Republic of China and Vietnam is referred to as Viet Nam, Hanoi as Ha Noi, Belgique as Belgium, Deutschland as Germany, and Nederland as the Netherlands. Nothing herein shall constitute or be considered to be a limitation upon or waiver of the privileges and immuni- ties of The World Bank, ADB, JICA, or DFID, which privileges and immunities are specifically reserved. Rights and Permissions This work is available under the Creative Commons Attribution 3.0 IGO license (CC BY 3.0 IGO) http:// creativecommons.org/licenses/by/3.0/igo. Under the Creative Commons Attribution license, you are free to copy, distribute, transmit, and adapt this work, including for commercial purposes, under the following conditions: Attribution—Please cite the work as follows: Asian Development Bank, Department for International Development, Japan International Cooperation Agency, and the World Bank. 2018. The WEB of Transport Corridors in South Asia. Washington, DC: World Bank. doi: 10.1596/978-1-4648-1215-6. License: Creative Commons Attribution CC BY 3.0 IGO Translations—If you create a translation of this work, please add the following disclaimer along with the attribution: This translation was not created by The World Bank, ADB, JICA, or DFID and should not be considered an official World Bank, ADB, JICA, or DFID translation. The World Bank, ADB, JICA, and DFID shall not be liable for any content or error in this translation. Adaptations—If you create an adaptation of this work, please add the following disclaimer along with the attribution: This is an adaptation of an original work by The World Bank, ADB, JICA, and DFID. Views and opinions expressed in the adaptation are the sole responsibility of the author or authors of the adaptation and are not endorsed by The World Bank, ADB, JICA, and DFID. Third-party content—The World Bank does not necessarily own each component of the content contained within the work. The World Bank therefore does not warrant that the use of any third-party-owned individual component or part contained in the work will not infringe on the rights of those third parties. The risk of claims resulting from such infringement rests solely with you. If you wish to re-use a component of the work, it is your responsibility to determine whether permission is needed for that re-use and to obtain permission from the copyright owner. Examples of components can include, but are not limited to, tables, figures, or images. All queries on rights and licenses should be addressed to World Bank Publications, The World Bank Group, 1818 H Street NW, Washington, DC 20433, USA; e-mail: pubrights@worldbank.org. ISBN (paper): 978-1-4648-1215-6 ISBN (electronic): 978-1-4648-1216-3 DOI: 10.1596/978-1-4648-1215-6 Cover design: Bill Pragluski/Critical Stages, LLC. Library of Congress Cataloging-in-Publication data has been requested. Contents Acknowledgments........................................................................................................................... xiii Abbreviations................................................................................................................................... xv Overview ................................................................................................................................1 From Transport to Wider Economic Benefits...........................................................2 Achieving Wider Economic Benefits: FIT-2-Deeds...................................................9 Main Takeaways for Practitioners.........................................................................15 References.............................................................................................................22 PART I GOING BEYOND JUST INFRASTRUCTURE Chapter 1 Insights into Regional Integration from Three Historical Transport Corridors in South Asia........................................................................................27 Mughal India, circa 1700: A Corridor Linking South and Central Asia.................28 Bengal and the East India Company, 1745−1813: The Triangular Trade between Britain, India, and China...............................................................31 Railroads of the British Raj, 1853−1929: Changing the Technology of Trading on the Subcontinent.................................................................................38 The Three Historical Cases Share Commonalities with One Another and with Modern Transport Corridors..................................................................43 Notes.....................................................................................................................44 References.............................................................................................................45 v vi   Contents Chapter 2 Insights into Regional Integration from Two Contemporary Transport Corridors in East Asia...........................................................................................49 Vietnam’s National Highway No. 5, Industrial Anchors, and Local Spillovers....... 50 Malaysia’s Experience with Integrating Rural Areas and Industrialized Trade Hubs...........................................................................................................59 Notes.....................................................................................................................69 References.............................................................................................................70 PART II FRAMEWORK AND ANALYTICS Chapter 3 Can Transport Corridor Projects Produce Wider Economic Benefits? Evidence from International Development Organizations.................................... 75 Conceptual Framework.........................................................................................78 Sample Selection and Summary Statistics...............................................................79 Maximizing the Wider Economic Benefits of Transport Corridors........................83 Conclusion............................................................................................................91 Notes.....................................................................................................................92 References.............................................................................................................93 Spotlight 1 Financing Priority Transport Corridors in South Asia.........................................95 Spotlight 2 Private Investment in Corridor Infrastructure.................................................... 107 Spotlight 3 The Role of Public-Private Partnerships in Developing South Asia’s Corridors...................................................................................................... 113 Chapter 4 Learning from the Literature about the Estimated Wider Economic Benefits of Transport Corridors..........................................................................125 Framework for Structuring the Literature Review...............................................127 Methodology.......................................................................................................129 Descriptive Statistics............................................................................................131 Meta-Regression Analysis....................................................................................141 Main Insights and Directions for Future Research...............................................146 Notes...................................................................................................................149 References...........................................................................................................151 Spotlight 4 The Impact of Highways on Micro, Small, and Medium Enterprises: Anecdotal Evidence from Bhutan, Sri Lanka, and India .................................... 155 Spotlight 5 Do Highways Help Women?............................................................................... 161 PART III APPRAISING CORRIDOR PROJECTS Chapter 5 Appraising Transport Corridors in Japan, Europe, and Thailand...................... 167 Appraising the Pacific Belt Zone Initiative: The Engine for Japan’s Industrialization..................................................................................................168 C o n t e n t s   vii Understanding the Wider Economic Benefits of Europe’s High-Speed Train Projects......................................................................................................176 Thailand, the Detroit of the East?........................................................................191 Notes...................................................................................................................204 References...........................................................................................................206 Spotlight 6 Appraising Proposed Transport Corridors Using Spatial Econometrics ..................209 Spotlight 7 Agriculture Finance and Technical Assistance to Enhance the Wider Economic Benefits of Transport Connectivity for Rural Areas ......................... 219 Chapter 6 An Illustrative Appraisal of Complementary Interventions to Enhance the Wider Economic Benefits of Transport Corridors..............................................223 A Multi-Step Methodology Was Used to Estimate the Wider Economic Benefits of Highways in India..............................................................................224 The Wider Economic Benefits of Highways Are Estimated Using District-Level Data..............................................................................................226 Estimation Results for the Two Existing Highways in India................................232 The Simulated Impacts of Two Proposed Corridors: The China-Pakistan Economic Corridor and the Kolkata-Dhaka Corridor.........................................240 Conclusion..........................................................................................................246 Notes...................................................................................................................248 References...........................................................................................................249 Spotlight 8 Cross-Border Infrastructure Projects: Challenges and Lessons Learned from the Unrealized Sava Waterways Rehabilitation Program in Southeast Europe................................................................................................ 251 Spotlight 9 The Influx of Workers and Followers in a Transport Project: Lessons in Gender Risks from a Road Project in Uganda................................................257 BOXES O.1 Categorizing transport corridor connections.......................................................................3 O.2 Defining transport corridors at the project level.................................................................5 2.1 Meeting the manpower requirements of an export hub through a public-private partnership: The Penang Skill Development Centre..........................................................62 3.1 The uncertain wider economic impacts of the Maputo Corridor in Southern Africa....................................................................................................76 S1.1 Financing the Oresund link between Denmark and Sweden...........................................104 S1.2 The Silk Road Fund: A regional transport corridor strategic investment fund................105 4.1 Financing and facilitating urban development along corridors by “capturing” rising land values.............................................................................................................143 6.1 Estimating the average impacts of the highways.............................................................226 6.2 Estimating conditional impacts: Do highway impacts depend on market conditions?........ 228 6.3 Simulating the effects of a transport corridor in another country ..................................241 viii   Contents FIGURES O.1 WEB are achieved through various transmission channels and intermediate outcomes....... 10 O.2 The corridor program can include trade facilitation measures, as well as soft complementary policies.................................................................................................10 O.3 A corridor intervention package triggers a hierarchy of impacts...................................12 O.4 India’s market for corridor PPPs fragmented as it expanded—and quality suffered...... 13 O.5 Corridors can increase incomes, but these gains come at the expense of environmental quality...................................................................................................17 O.6 Private sector involvement in corridor projects has been limited, and its impact has been low.................................................................................................21 2.1 In the early 1990s, Vietnam needed a set of complementary efforts to promote industrialization and exports, improve transport, and upgrade labor............................51 2.2 Foreign direct investment around Hai Phong Harbor has soared..................................51 2.3 Vietnam’s growth model centered on “anchor tenants” and associated firms jointly developing local supply chains......................................................................................52 2.4 The number of enterprises surged along NH-5 from 2000 to 2004, especially in Hanoi........................................................................................................54 2.5 Production along NH-5 grew significantly more than the national average from 2003 to 2006...........................................................................................55 2.6 The Vietnamese economy underwent a vast structural transformation from 1997 to 2002........................................................................................................55 2.7 Traffic volume has soared in Vietnam on NH-5............................................................56 2.8 The NH-5 upgrade generated significantly wider economic benefits for income and poverty.......................................................................................................56 2.9 Traffic accidents have soared on NH-5, greatly outpacing the national trend as well as those of other national highways.........................................................57 2.10 The NH-5 Corridor generated many expected and unexpected wider economic impacts..........................................................................................................58 3.1 The design of transport corridor projects respects initial conditions and can involve three levels of interventions........................................................................78 3.2 Overall approach of the study.......................................................................................79 3.3 Most of the reviewed projects are in the road sector.....................................................80 3.4 The sample of reviewed projects captures more recent projects, 1984–2011.................81 3.5 Only a small share of the projects in the sample are international.................................82 3.6 Most projects in the sample do not have a good theory of change................................82 3.7 The private sector is often not involved in designing projects........................................82 3.8 The private sector is often not involved in the operation or management of transport infrastructure.................................................................................................82 3.9 Only about half the projects occurred in countries that had undertaken World Bank–sponsored institutional reforms in the five years before the project was approved ...................................................................................................83 3.10 Well-thought-out theory of change can help corridor projects succeed, the expected benefits of consulting or involving the private sector have not been realized.................................................................................................................86 C o n t e n t s   ix 3.11 Complementary financial, and industrial and trade reforms undertaken before a corridor project is implemented could help the project succeed.......................87 3.12 Initial conditions of a smaller, flatter, less developed, and coastal country could increase the likelihood for WEB..........................................................................88 3.13 Investing at the border and in connectivity to border crossings shows smaller increases in local economic activity: inherent complexity could be the challenge.............89 3.14 Policies promoting trade openness and institutions boosting governance effectiveness appear the most robust complementary interventions to help corridor investments spur local economic activity.........................................................89 S1.1 There are three main forms of infrastructure finance, plus hybrid financing.................97 S3.1 A growing share of the loans to ports and highways that are backed by the State Bank of India are nonperforming.......................................................................115 S3.2 The time needed to enforce contracts varies from less than a year to more than four years in some South Asian countries....................................................................117 S3.3 Most countries in South Asia compare well to their low- and middle-income peers in the share of domestic credit provided by their financial sectors......................118 S3.4 Overall foreign direct investment is relatively low in South Asian countries................119 S3.5 India’s market for corridor PPPs fragmented as it expanded—and quality suffered.......120 S3.6 Debt-to-equity ratios of four publicly traded developers in India varied considerably from 2010 to 2015.................................................................................121 4.1 The web of WEB: The final outcomes of a corridor intervention are achieved through many transmission channels and various intermediate outcomes ..................128 4.2 Direct and indirect impacts determine the overall impact of a corridor intervention........................................................................................................... 129 4.3 The overall balance between beneficial or detrimental impacts of a corridor intervention package depends on a hierarchy of impacts.............................................130 4.4 The literature search yielded 78 papers, of which 9 percent were unpublished............131 4.5 Rigorous evaluations of the impacts of large-scale transport projects have surged recently............................................................................................................131 4.6 Evaluations of road projects and new systems dominate the literature........................133 4.7 The final outcome most often evaluated is economic welfare, while the intermediate outcomes studied are split between trade/productivity and population/assets.........................................................................................................134 4.8 Studies at the subnational level and ex post evaluations predominate.........................134 4.9 Two-thirds of studies are motivated by a theoretical model, and 70 percent rely on reduced-form estimation.................................................................................135 4.10 To improve reduced-form estimations, the literature employs rigorous identification strategies...............................................................................................136 4.11 The literature has focused on estimating corridor impacts on welfare, much less on the inclusion, environment, and equity............................................................136 4.12 Environment and equity outcomes could be suffering due to corridor placements......138 4.13 While land values, population, and trade can increase in some areas as a result of corridor connectivity, they may decrease in other areas in parallel..............................139 4.14 The trend to analyze heterogeneous impacts is growing, 2000–17..............................139 x  Contents 4.15 Geographic location is the most studied aspect of heterogeneous impacts...................140 4.16 Income and population are the most studied aspects of reported evidence of absolute losers........................................................................................................141 4.17 Most studies do not estimate the impacts of complementary policies..........................142 4.18 The distributions of t-statistics are skewed to the right, suggesting that the transport infrastructure projects yield positive and significant impacts.......................144 4.19 Average net impacts of corridors on final outcomes estimated by the literature should alert policy makers..........................................................................................144 4.20 Varied impacts of corridors estimated by the literature (by the factor of variation) can inform complementary interventions to support losers from corridors...................................................................................................................146 5.1 Private investment/public investment were higher in the Pacific Belt Zone than in other regions of Japan, 1960−70.....................................................................170 5.2 Demographic shift to the three major metropolitan areas and the income disparities in the region...............................................................................................172 5.3 Japan’s rapid economic growth was supported by a virtuous cycle.............................176 5.4 The journey time between major stations in northwestern Europe fell by about half on most routes between 1989 and 2009.....................................................183 5.5 Transport, utilities, and investment incentives ranked high in companies’ decisions to operate on the Eastern Seaboard..............................................................194 5.6 Thailand’s automotive output and exports have soared since the mid-1990s..............196 5.7 The net value added of Thai automotive exports has increased greatly since the late 1990s.....................................................................................................197 5.8 Thailand outperformed Malaysia and Indonesia in number of vehicles produced and their export value, thanks to good policies...........................................198 5.9 Labor productivity inside and outside the improved infrastructure regions.................200 5.10 Simulated effects on the incidence of poverty (headcount measure).............................203 6.1 The difference-in-difference method is used to estimate the impact of the two existing Indian highways......................................................................................225 6.2 The triple difference-in-difference method is used to estimate how the impact of the highway depends on complementary factors.........................................228 6.3 One in four districts was “treated”.............................................................................232 6.4 The economic restructuring induced by the highways shifted employment from the farm to the nonfarm sector...........................................................................233 6.5 While the average income grew, the other benefits were not widely shared, and environmental quality deteriorated.......................................................................234 6.6 Access to finance for firms and households had different impacts on spreading WEB...................................................................................................... 236 6.7 Education boosted the impact of the two highways ...................................................237 6.8 Land market conditions had mixed impacts on spreading WEB..................................238 MAPS O.1 Many large transport investments are proposed across South Asia.................................2 BO.1.1 Possible types of transport corridors (connections).........................................................3 C o n t e n t s   xi BO.2.1 An illustration of linear kilometers..................................................................................5 O.2 The trade-offs generated by the Pacific Ocean Belt in Japan yield valuable lessons.........7 O.3 Some farmers along the Delhi-Mumbai Corridor fear being uprooted............................8 O.4 Where should South Asia focus its limited resources—toward the Middle East and Europe, toward Central Asia, or toward East Asia?...............................................16 O.5 Spatial impacts around major South Asia corridors vary significantly...........................18 O.6 The simulated impacts on per capita household consumption across districts along the Kolkata-Dhaka Transport Corridor (Bangladesh segment) call for complementary interventions........................................................................................19 1.1 Mughal Empire, 1690...................................................................................................28 1.2 Territories of the British East India Company, 1765 and 1805......................................32 1.3 Hindustan, 1864...........................................................................................................33 1.4 Completed and planned railway lines in India, 1871.....................................................39 1.5 Rail in India, 1909........................................................................................................40 2.1 Vietnam’s National Highway No. 5 connects Hanoi to the port of Hai Phong.....................................................................................................................50 2.2 Industrial parks have flourished along NH-5................................................................53 3.1 How a linear kilometer is defined.................................................................................80 5.1 The Pacific Belt Zone runs through Japan’s industrial core.........................................168 5.2 Fifteen new industrial cities were established to spread benefits throughout Japan............................................................................................... 173 5.3 Health damage related to air pollution soared in the Pacific Belt Zone.......................175 5.4 The Trans-European Transportation Network connects most of Western, Northern, and Southern Europe, the United Kingdom, and Ireland............................177 5.5 High-speed rail connects the core of North-West Europe and the United Kingdom.................................................................................................184 5.6 The Rhine Alpine Rail Freight Corridor: “From sea to sea without barriers”.............187 5.7 The four components of the Eastern Seaboard Development Plan acted as an integrated unit........................................................................................................193 5.8 A network of industrial parks, industrial estates, and industrial zones spread from the Deep-Water Port to Bangkok.............................................................201 6.1 Two highways in India are used to illustrate the appraisal framework........................224 6.2 The data set includes a range of measures of wider economic benefits........................230 6.3 Distance of districts in India from the GQ and NSEW highways is measured from the district’s centroid..........................................................................................231 6.4 The simulated relative impact on mean per capita household consumption expenditures shows disparities along the Kolkata-Dhaka Transport Corridor (Bangladesh segment) ...........................................................................................242 6.5 The simulated relative impact on the reduction in poverty (headcount measure) varied along the Kolkata-Dhaka Corridor (Bangladesh segment)................................243 6.6 The simulated relative impact on the share of regular wage jobs in women’s employment shows major disparities along the Kolkata-Dhaka Transport Corridor (Bangladesh segment)...................................................................................244 xii  Contents 6.7 The simulated relative impact on air pollution (aerosol optical thickness) should alert policy makers about needed action in some districts along the Kolkata-Dhaka Transport Corridor (Bangladesh segment) .........................................245 S8.1 The SAVA Waterway Rehabilitation Project Traversed Several Former Yugoslav Republics.....................................................................................................252 PHOTOS S4.1 Spice producers use the A9 Highway in Sri Lanka to dry spices..................................156 S4.2 The World Bank team speaks with a textile shop owner in Thane, India.....................157 S5.1 Focus group in Khasadrapchu, Bhutan........................................................................162 S5.2 Focus group in Naulla, Sri Lanka................................................................................162 TABLES O.1 Combining the layers of a corridor program with the hierarchy of impacts gives a useful tool to screen for comprehensive project design..................................................12 O.2 The trade-offs between different econometric approaches to appraise proposed corridors give options ..................................................................................................20 1.1 The five major industries in Bengal created almost one million jobs from 1795 to 1829, thanks to growth in both precolonial and new industries.......................37 1.2 Agriculture continued to dominate the economy, despite the huge investment in rail.................................................................................................................. 41 2.1 Investment and employment in Malaysia’s five corridors, 2011−14..............................66 3.1 Most of the projects in the sample are in Asia ..............................................................81 S1.1 Who pays for transport corridors?................................................................................96 S3.1 Roads in India dominate the number and share of private participation in infrastructure in South Asia .......................................................................................114 5.1 The Eastern Seaboard had the second-best set of macroeconomic indicators among regions of Thailand during its development period of 1981−96.......................200 5.2 Simulated macroeconomic effects................................................................................202 S6.1 There are trade-offs between different econometric approaches to appraise proposed corridors .....................................................................................................215 Acknowledgments T he preparation of this report was led by Martin Melecky (World Bank), together with Arjun Goswami (ADB), Akio The JICA core team comprised Akio Okamura, Takayuki Urade, Yumi Ito, and Miwa Hiasa. Okamura (JICA), and Duncan Overfield Contributors to the spotlights and boxes (DFID). Annette Dixon (World Bank) and included Yan (Sarah) Xu, Martin Humphreys, Juan Miranda (ADB) played a key role in Dominic Pasquale Patella, Daniel Sebastian putting this partnership effort in motion. Perea Rojas, Martin Melecky, Arnaud Dornel, Martin Rama (World Bank) provided techni- Michel Noel, Catherine Asekenye Barasa, cal guidance throughout the process. Stephen Muzira, Toshiaki Ono, Juan The World Bank core team comprised Buchenau, Panos Varangis, Mark Roberts, Martin Melecky, Matias Herrera Dappe, Vasudha Thawakar, and Roland White. Mark Roberts, Siddharth Sharma, Muneeza We thank for suggestions and comments Alam, Theophile Bougna, Yan (Sarah) Xu, Akihiko Nishio, Marianne Fay, William Robin Carruthers, Hari Subhash, Ruifan Maloney, Robert J. Saum, Baher El-Hifnawi, Shi, Akib Khan, Nathalie Barboza, and Hun Kim, Yasuyuki Sawada, Juzhong Esther Bartl. Zhuang, Ronald Butiong, Safdar Parvez, The ADB core team comprised Arjun Alfredo Perdiguero, Ying Qian, Kenichi Goswami, Jayant Menon, Premachandra Yokoyama, Xiaohong Yang, Rana Hasan, A t h u k o r a l a , P e t e r Wa r r, A r c h a n u n Jong Woo Kang, Cuong Minh Nguyen, Utsav Kohpaiboon, Suresh Narayanan, Anna Kumar, Kavita Iyengar, Kristian Rosbach, Cassandra Melendez, Victoria Corpuz, and Elizabeth Jones, Jaya Singh Verma, Arielle Ruth Francisco. Dove, Koki Hirota, Naohiro Kitano, Toru The DFID core team comprised Peter Arai, Shuntaro Kawahara, Kazumasa Sanui, Frankopan, Manish Vasistha, Roger Akiko Sanada, Koji Yamada, Tomohide Vickerman, Richard Bullock, and Duncan Ichiguchi, Katsuo Matsumoto, Kunihiro Overfield. Nakasone, Takuro Takeuchi, Tatsuya Asami, xiii xiv   Acknowledgments Charles Kunaka, Debora Revooltella, Adam Kulathunga, Aphichoke Kotikula, Ugyen Storeyard, Stephane Straub, Homi Kharas, Lham, Nagasubramanian Kodimangalam Jean Francois Arvis, Hatem Chahbani, Uwe Sundaram, Mohamed Asan Saleem, Jayati Deichmann, Ejaz Ghani, Somik Lall, Harris Sethi, Ashutosh Tandon, Neelam Chowdhry, Selod, Forhad Shilpi, Luc Lecuit, Gerald Paul and Yann Doignon. Ollivier, Andrew Beath, Vincent Palmade, The report benefitted from the stimulating Sanjay Srivastana, Janardan Prasad Singh, discussion of its main messages by the panel Simon Alder, Karla Gonzalez Gravajal, Ming comprising Ahsan Iqbal (Minister, Pakistan), Zhang, Catalina Marulanda, Esperanza Harsha de Silva (Deputy Minister, Sri Lanka), Lasagabaster, Samuel Maimbo, and Toshiya Yasser Rizvi (Alliance Holding, Bangladesh), Masuoka. and Fatema Sumar (Regional Deputy Special thanks go to the following task Vice President, Millennium Challenge team leaders of transport projects, who were Corporation). interviewed during the preparation of the The report was edited by Nancy Morrison. report, for their valuable responses: Martin Barbara Koeppel edited the spotlights. Humphreys, Olivier le Ber, Jacques Bure, Financial support from Australia’s Rodrigo Archondo-Callao, Ben Eijbergen, Department of Foreign Affairs and Trade is Simon Ellis, Diep Niguyen-Van Houtte, Martha gratefully acknowledged. Lawrence, Henry Des Longchamps, Jean- In producing this report, the World Bank Francois Marteau, Graham Smith, Andreas emphasizes that regional cooperation and Schliessler, and Xiaoke Zhai. connectivity initiatives and projects shall Help with logistics and organizing focus respect the sovereignty of the countries groups interviews in Bhutan, India, and Sri involved, and notes that the findings and con- Lanka was provided by Rathnija Arandara, clusions in the report may not reflect the Poorna Bhattacharjee, Varsha Marathe views of individual countries concerned by Dayal, Shanuki Gunasekera, Anoma these initiatives and projects. Abbreviations ABM agent-based modeling ADB Asian Development Bank ALM asset-liability management APTC Asia Pacific Transport Consortium ASEAN Association of South East Asian Nations BKIP Batu Kawan Industrial Park BOI Board of Investment BOOT build-own-operate-transfer BOT build-operate-transfer CAS Country Assistance Strategy CBA cost-benefit analysis CEO chief executive officer CGE computable general equilibrium CPEC China-Pakistan Economic Corridor CPIA Country Policy and Institutional Assessment CSO civil society organization CTRL Channel Tunnel Rail Link DBFM design-build-finance-maintain DFC Dedicated Freight Corridor DID difference-in-difference DPA Dayton Peace Accord DPO Development Policy Operation DSCR debt service coverage ratio EC European Commission ECER East Coast Economic Region EE electronics and electrical EIA environmental impact assessment EIB European Investment Bank EIC East India Company ELA Empowerment and Livelihoods Assistance xv xvi   A b b r e v i at i o n s EPU Economic Planning Unit ESDP Eastern Seaboard Development Plan ESIA Environmental and Social Impact Assessment ESMP Environmental and Social Management Plan EU European Union FASRB Framework Agreement on the Sava River Basin FBH Federation of Bosnia and Herzegovina FDI foreign direct investment FVS financial viability support GBV gender-based violence GE general equilibrium GIPR Great Indian Peninsula Railway GKC Greater Kamunting Conurbation GPI Governance Performance Index GQ Golden Quadrilateral GSDP Global Supplier Development Programme HSR high-speed rail IBRD International Bank for Reconstruction and Development ICR Implementation Completion and Results IDA International Development Association IEAT Industrial Estate Authority of Thailand IEG Independent Evaluation Group IIDF India Infrastructure Debt Fund IIGF Indonesia Infrastructure Guarantee Fund IM Iskandar Malaysia IOTC International Organization of Tuna Council IPA Instrument for Pre-Accession ISRBC International Sava River Basin Commission JFCU Joy for Children Uganda JICA Japan International Cooperation Agency KDTC Kolkata-Dhaka transport corridor KHTP Kulim Hi-Tech Park KSTP Kedah Science and Technology Park LCR London Continental Railway LPG liquified petroleum gas LVC land value capture M&E monitoring and evaluation MCC Maputo Corridor Company MCLI Maputo Corridor Logistics Initiative MIDA Malaysian Industrial Development Authority MIGA Multilateral International Guarantee Agency MRA meta-regression analysis MSME micro, small, and medium-sized enterprise MTR Mass Transit Railway NCER Northern Corridor Economic Region NCIA Northern Corridor Implementation Authority NEG new economic geography NEN National Expressway Network NHAI National Highways Authority of India A b b r e v i a t i o n s   xvii NIIF National Investment and Infrastructure Fund NPA nonperforming asset NPDC Nord-Pas-de-Calais NRB Nepal Rastra Bank NSEW North–South–East–West NSS National Sample Survey NTSC National Traffic Safety Committee ODA official development assistance OECD Organisation for Economic Co-operation and Development OHR Office of the High Representative OLS ordinary least squares PE private equity PIP Perlis Inland Port Project PPA Project Preparation Advance PPI Private Participation in Infrastructure PPP public-private partnership PSDC Penang Skill Development Centre R&D research and development RA research assistant RDDBFI Recovery of Debts Due to Banks and Financial Institutions RF reduced form RGC regional growth center RS Republika Srpska SARFAESI Securitization and Reconstruction of Financial Assets and Enforcement of Security Interests SBI State Bank of India SCORE Sarawak Corridor of Renewable Energy SDC Sabah Development Corridor SDI spatial development initiative SIF Strategic Investment Fund SME small or medium-sized enterprise SOE state-owned enterprise SPSEE Stability Pact for Southeastern Europe SPV special purpose vehicle SRF Silk Road Fund STD sexually transmitted disease TFP total factor productivity TSDP Transport Sector Development Project TVET Technical and Vocational Education Training UNDP United Nations Development Programme UNRA Uganda National Roads Authority VAC Violence Against Children VAR vector autoregression VAT value-added tax VOC vehicle operating cost VRA Vietnam Road Administration WEB wider economic benefits WGI Worldwide Governance Indicator WITS World Integrated Trade Solution Overview D eveloping large transport infrastruc- ture projects—transport corridors—is increasingly seen as a way to stimulate regional benefit more than the transit regions and countries it crosses. Along the corridor itself, people with different skill endowments may integration and economic growth. Countries, benefit differently, and farming households often with the help of the international com- along the c­ orridor may be at a disadvantage if munity, invest in these corridors in the hope of their land tenure is not secure. creating large economic surpluses that can In China, for instance, the construction of spread throughout the economy and society. the National Express Network (NEN) But if the corridors do not generate the increased real income across its prefectures expected surpluses, they can become wasteful by nearly 4 percent, on average—but still white elephants—transport infrastructure decreased real wages in many prefectures in without much traffic. either the urban or rural sector (Roberts et al. Recent international experience reveals 2012). Only when transport corridors share both successes and failures. Vietnam, for prosperity widely, not only creating winners instance, developed National Highway No. 5 but also not leaving behind losers, can they in the 1990s to establish an effective transport spur equitable growth and help reduce poverty. corridor between the capital Hanoi and the Many corridor initiatives are under way or port of Hai Phong. This investment, on the are being proposed, both in South Asia and back of reforms in the business environment around the world (map O.1). One ambitious and investments in human capital, fostered proposal is to revive the Grand Trunk Road prosperity in the populous Red River Delta from Kabul, Afghanistan, to Chittagong, region. From 1995 to 2000, the poverty rate Bangladesh, connecting areas that are home fell by an impressive 35 percent, outpacing to a significant share of the world’s poor. the national average of 27 percent (JICA Even more ambitious could be the plan for 2009). In contrast, traffic volume has the New Silk Road Economic Belt. This very remained low on the multi-country Greater large transport corridor connects Beijing all Mekong Subregion (GMS) corridor (ADB the way to Brussels. It also branches out into 2008; Srivastava and Kumar 2012). South Asian countries, as is the case with the Even when corridors generate aggregate China-Pakistan Economic Corridor initiative. surpluses, a relevant question is whether the The investments associated with these and net benefits are fairly distributed across other proposed initiatives could require tril- the population. If they are not, corridors lions of dollars. Such an amount exceeds the risk becoming inequitable investments. financial resources available in the foresee- The end-nodes of a transport corridor may able future to support corridors. Moreover, it 1 2  The WEB o f T r a n s p o r t C o r r i d o r s i n S o u t h A s i a MAP O.1  Many large transport investments are proposed across South Asia Source: Corridor Study Team. risks crowding out other public investment about how to prioritize proposals for corri- critical areas such as education, water and in ­ dor investments, and specifically, how to sanitation, or energy. By a conservative esti- select the more promising ones over the less mate, from 2014 to 2020 South Asia needs promising or potentially wasteful ones. A to invest at least $1.7 trillion in infrastruc- clear understanding of what constitutes ture (Andrés, Biller, and Herrera Dappe a transport ­ c orridor is an important 2013). This is just to catch up with develop- first step in this direction (box O.1). Even ing country peers, whose infrastructure may more important is to focus on economic also be below “optimal” levels. analysis, and not only on geopolitical considerations. Decisions concerning transport corridors FROM TRANSPORT TO WIDER are understandably influenced by strategic ECONOMIC BENEFITS aspirations, as well as by concerns about secu- Given the huge resource cost and the high rity. Various scholars offer interpretations that stakes, national governments and the inter- emphasize this geopolitical dimension. national community need to think clearly For example, according to Chatterjee and O v e r v i e w   3 BOX O.1  Categorizing transport corridor connections Transport connectivity can be categorized by the •  Rural-gateway connections, in practice comprise a types of location that the transport infrastructure combination of Rural-urban and Urban-gateway connects. Four main types of connections can be connections. considered: •  Purely rural-rural connections are typically not considered part of transport corridors, as they do •  Urban-urban transport corridors connect two or not link to the trunk transport investments. more urban centers or cities. •  Urban-gateway corridors connect urban ­ centers For some countries, most notably the landlocked with international gateways (an ­international port, ones, cross-country corridors cutting through their a major land border crossing, or an airport that territory could run the risk of serving mostly as tran- provides a gateway to international markets). sit connections. To make the most of the cross-coun- •  Rural-urban transport connections link rural areas try connectivity, “transit” countries may need to to urban centers. These types of transport connec- focus on the feeder network and complementary tions are typically auxiliary (feeder roads, rails) reforms to generate and spread local socioeconomic linking to trunk (main) transport investments. benefits from such transit transport corridors. MAP BO.1.1  Possible types of transport corridors (connections) Source: Corridor Study Team. Note: The size of the bubble corresponds to the size of the economy. 4  The WEB o f T r a n s p o r t C o r r i d o r s i n S o u t h A s i a Singh (2015), India’s Foreign Trade Policy for and Kolkata—would go through some of the 2015−20 has highlighted the importance of most densely populated and most dynamic the International North-South Transport areas in the world, stoking hopes of large eco- Corridor in expanding India’s trade and nomic impacts and many positive spillovers. investment links with Central Asia. Ordabayev This report aims to balance the political (2015, 2016) points out the competing economy and geopolitics of transport corri- ­ geopolitical interests of India, China, the dors with sound empirical analysis. Its goal is Russian Federation, Europe, the United States, to empower key stakeholders—policy m ­ akers, Turkey, and Iran in Central Asia, and how economic advisors, businesses, and civil soci- they relate to decisions on transport corridor ety organizations—in assessing socioeco- connectivity, trade, and energy. Palit (2017) nomic benefits when considering the idea of discusses India’s uneasy economic and strate- investing in a transport corridor. gic perceptions of China’s Maritime Silk Road To that end, this report provides a con- Initiative. And Shephard (2017) analyzes ceptual framework to think about eco- whether India and Japan have joined forces to nomic ­ c orridors and to enable a holistic counter China and build their own New Silk appraisal of program and project proposals. Road, in another geopolitical move. The framework extends beyond the immedi- While geopolitical considerations are cer- ate effects of transport corridors—such as tainly legitimate, they also bring risks. Some reducing vehicle operating costs—and of the proposed corridors could become either focuses on the ultimate goals of having a white elephant investments or inequitable positive impact on local economic activity, public investments that benefit narrow groups jobs, gender equality, and poverty reduction, at the expense of national taxpayers. Risks among other desirable socioeconomic out- could be even higher for smaller landlocked comes. It also helps think about negative countries that could bear the cost of sizeable impacts such as traffic congestion, regressive investments but see themselves become mere redistribution, social exclusion, environmen- transit passageways. In South Asia, the land- tal degradation, and other risks or unin- locked countries are Afghanistan, Bhutan and tended consequences. An important building Nepal. Several Central Asian countries also block in this framework is to define ­transport belong in that group. corridors in an operational way (box O.2). The framework extends to consider how well different markets—such as capital, labor, Only when transport corridors generate wider local land, and product markets—function around benefits can they be considered economic corridors. transport corridors. In doing so, it helps determine which policy interventions might An equally important risk is missing the be needed to strengthen markets and maxi- “right” corridors that yield the greatest wider mize the potential for larger gains from trans- economic benefits (WEB) just because they do port corridors. Moreover, the framework not fit the prevailing geopolitical views. seeks to advance the understanding of the Arguably, the transport corridor with the institutions that can help the economy adjoin- greatest economic potential is the surface link ing transport corridors evolve into prosper- between Shanghai and Mumbai. These two ous and inclusive communities. It examines cities are the economic hubs of China and how incentives could be aligned to promote India respectively, two emerging global pow- healthy business competition and create ers. The distance between them, about 5,000 ­better jobs. kilometers, is not much greater than the dis- Using this framework, the report studies tance between New York and Los Angeles. the conditions under which large-scale invest- But instead of crossing a relatively empty con- ments in transport infrastructure can generate tinent, a corridor from Shanghai to positive spillovers on local household income, Mumbai—via Kunming, Mandalay, Dhaka, jobs, equity, and poverty reduction. It also O v e r v i e w   5 BOX O.2  Defining transport corridors at the project level For the purpose of this Report, only investments •  The value of the project at appraisal is greater than in roads, rail, or inland waterways are considered $50 million and the investment is financing a criti- as possible transport corridor investments. In gen- cal link in the corridor (such as a new bridge or eral, roads, railways, and waterways are distinctly tunnel) that connects at least two economic geographically defined. They also have clearer eco- centers. nomic spillovers along their stretches compared with, for instance, airways or maritime routes. The transport corridor projects considered in the However, the findings and recommendations of the analyses for this Report satisfy at least one of these Report could extend to air, sea, and other types of two criteria. These criteria are designed to identify projects that transport corridors. have the potential to enhance the regional connectivity At a project level, an investment is classified as a of a country or a set of countries. The projects do not transport corridor if: need to be geographically continuous, but they need to be part of a set of road, rail, or waterway routes •  The transport investment is being made on a route connecting two locations. As an example, the roads that is creating, upgrading, or rehabilitating at marked in red and pink in map BO.2.1 qualify as lin- least 100 linear kilometers, or ear kilometers, but the roads marked in blue do not. MAP BO.2.1  An illustration of linear kilometers National Capital Roads Moradabad NEPAL State Capitals NEW DELHI Major Cities Bareilly Churu Bikaner Lakhimpur PAKISTAN Sikar Nagaur Agra Jaisalmer Jaipur Jaipur Lucknow INDIA INDIA Bhind Kanpur Jodhpur Beawar Gwalior Sawai Madhopur Barmer Balotra Orai Bhilwara Kota Jhansi Baran Shivpuri Chhatarpur Chittaurgarh Guna Lalitpur Udaipur Ashoknagar Palanpur Mandsaur 0 100 200 Kilometers IBRD 43391 | JANUARY 2018 Source: Corridor Study Team. Note: The roads marked in red and pink are linear kilometers, but the roads marked in blue are not. 6  The WEB o f T r a n s p o r t C o r r i d o r s i n S o u t h A s i a studies potential trade-offs that policy makers vehicle operating costs that are the focus of could face and discusses how to manage them— traditional cost-benefit analyses. Transport such as increasing income at the expense of corridor interventions have the potential to deteriorating environmental quality—using affect broader socioeconomic outcomes. complementary policies and institutions. Their economic impacts work through To establish a comprehensive knowledge agglomeration effects, increased trade and base for its recommendations, the Report migration, and changes in the local economic includes case studies of past and recent corri- structure, among other areas. These long- dor initiatives—such as Japan’s Pacific Ocean term impacts ultimately yield WEB—such as Belt Initiative, Europe’s High-Speed Rail the growth of income and consumption, new Network, and Vietnam’s National Highway jobs, and greater equity. No. 5. These are combined with a systematic The economic boom of 1920−29 in India review of the existing literature on the spatial provides an illustration of these WEB. By then, impacts of corridors, and assessments of cor- more than 66,000 kilometers of railway lines ridor investment projects supported by inter- were in operation, carrying over 620 ­ million national development organizations. passengers and 90 million tons of goods each Building on this integrative approach, and year. Some academics have argued that the relying on spatially granular data, the Report impact of the railways during this period was conducts illustrative appraisals of three major mixed because of the limited development of transport corridors in South Asia. These are other infrastructure (Chandra 2006). But a the completed Golden Quadrilateral and recent study with a rigorous methodology North-South-East-West highway systems in finds that the extension of the railroad net- India; the planned China-Pakistan Economic work increased real agricultural income by Corridor in Pakistan; and the anticipated about 16 percent (Donaldson 2010). Kolkata-Dhaka transport corridor between Not all impacts of a transport corridors India and Bangladesh. and their spillovers may be beneficial, how- A series of spotlights focus on selected ever. While improving one development out- ­ facets of transport corridors. Some concern come (such as income), these investments impacts on specific groups, such as micro, could worsen another one (such as environ- small, and medium-sized enterprises mental quality). (MSMEs); women in Bhutan, India, and An example is Japan from 1960 to 1970, Sri Lanka; and groups seeking finance to at the time of the major effort to double the take advantage of the opportunities gener- country’s income. A key component of this ated by better connectivity to markets. Other initiative was the development of the Pacific spotlights discuss the challenges in engaging Belt Zone, which increased the total shipment the private sector to co-invest in corridors at value of goods almost fourfold (from ¥24.7 different stages of their development, or trillion to ¥94.4 trillion, in 2005 prices). through different financial vehicles, in both However, rapid industrialization also resulted South Asia and around the world. One spot- in serious negative impacts in the form of light assesses in more detail the potential of environmental degradation and pollution modern spatial econometrics to rigorously (map O.2). The incidence of bronchial asthma appraise investment proposals for transport and other diseases soared in the Pacific Belt corridors. Two other spotlights focus on the Zone. Following several major disease out- lessons learned from implementing cross- breaks caused by water contamination, Japan border transport corridors and on the risks was recognized worldwide as a “Paradise of that the influx of construction workers pres- Pollution.” In the early 1970s, litigation ent for local women. favored plaintiffs for all four of the major This Report recognizes that transport pollution-related diseases. Since then, Japan ­ corridor interventions can generate a chain of has undertaken a full-scale effort to protect multiple impacts beyond the travel time and the environment. O v e r v i e w   7 MAP O.2  The trade-offs generated by the Pacific Ocean Belt in Japan yield valuable lessons Source: Corridor Study Team. Photos by World Bank (upper left); Kanagawa Environmental Research Center, Japan (upper right). Used with permission; further permission required for reuse. At the same time, well-designed corridor There, the impact was different across districts interventions can generate synergies that miti- depending on the educational attainment of gate these potential trade-offs. For instance, by their population. More educated districts con- switching to more fuel-efficient and pooled nected to the transport corridor enjoyed both transportation, income and environmental higher income growth and a lower level of air quality can improve together in some pollution (as measured by the density of aero- locations. sol particles) than their less educated counter- Evidence from the Golden Quadrilateral parts (Melecky, Sharma, and Subhash, highway system in India supports this ­ rationale. forthcoming). 8  The WEB o f T r a n s p o r t C o r r i d o r s i n S o u t h A s i a Corridors may create both winners and losers, but has put them at a disadvantage (Kumar ­ well-designed corridor programs can help share the 2015). Benefits for this group are bound to be benefits more widely higher in a context where agribusiness is more developed, and farmers can connect to value- Across locations and population groups, added chains boosting their income and help- initial conditions can affect the predisposi- ing them diversify their production. tion to benefit from better transport connec- Geographically, the Report focuses on South tivity. Even on a single development outcome Asia—not just as one of the world’s most pop- (such as jobs), some households and some ulous and poorest regions—but prominently as places could win, and others could lose. For a hinge between East Asia, Central Asia, the instance, more educated and skilled people Middle East, and Europe. This focus derives can migrate to obtain better jobs in growing from a grander vision for South Asia that also urban areas that are benefiting from corridor presents a challenge. South Asia has the poten- connectivity, while unskilled workers may be tial to be the world’s next middle-income left behind in depopulated rural areas with region. Connecting it to East Asia (including few economic prospects. Corridors can thus through Southeast Asia) can transform it into create both winners and losers. an engine of global growth. But South Asia In India, for example, farmers along the could also stand to lose if it were circumvented, Delhi-Mumbai corridor have voiced unhappi- or the “right” corridors were missed. ness about the structural transformation from Assessing the potential for functioning inter- agriculture to manufacturing and services that national connectivity requires a ­ forward-looking, is taking place along the way (map O.3). holistic perspective. The future success of It has been argued that this transformation smaller corridors critically depends on their MAP O.3  Some farmers along the Delhi-Mumbai Corridor fear being uprooted Source: Corridor Study Team for maps and calculations of connectivity. O v e r v i e w   9 ability to integrate into cross-border, transre- many aspects of the economy at the same gional, and transcontinental corridors. time. These complementary factors could comprise initial conditions in local product markets, such as the degree of business ACHIEVING WIDER ECONOMIC competition. They can also relate to the BENEFITS: FIT-2-DEEDS operation of markets for land, labor, and Policy makers should aim to maximize the capital. For instance, they may refer to land- wider economic benefits of a proposed trans- use restrictions, the availability of skilled port corridor investment, considering benefits labor, and access to credit. This direct net of costs. This policy-maker’s problem impact can vary from beneficial to detri- can play out at different levels: local (for sub- mental across individual outcomes, and national units such as states, provinces or dis- vary considerably across locations and pop- tricts), national, and international. ulation groups. The framework proposed by the Report A corridor intervention can also affect to think through the policy-maker’s prob- the final outcomes indirectly, by modifying lem, and guide holistic appraisals of pro- the complementary factors. For instance, posed programs (projects) on transport if the corridor reduces commuting and corridors, builds on six elements. For ease of migration costs, it also reduces frictions in communication, these six elements are sum- the labor market, which in turn can increase marized in the acronym “FIT-2-Deeds.” The the local availability of skilled labor, overall “FIT” in this acronym refers to the Flow of employment, and household income. These expected results, the Intervention design, and indirect changes of complementary factors the Typology of impacts. The “2” stands for are called intermediate outcomes . The two sorts of complementary public interven- impact of the corridor on intermediate out- tions, and the twin Deeds are financing and comes can also vary from beneficial to detri- implementing the corridor. mental across the individual outcomes considered. Knowledge of the direct and indirect “F”: The “Flow” of expected results— impacts, the trade-offs they could produce, The chain from corridor to benefits and the complementary policies that could Several potential transmission mechanisms help manage these trade-offs, can all improve and associated intermediate outcomes help the design of a corridor intervention. predict the ultimate impact of a corridor intervention on a relevant set of final out- “I”: The “Intervention” design— comes. The more favorable these outcomes Supporting a fairer distribution of are, the larger is the ratio of benefits to cost. greater benefits Large WEB, net of costs, should be the policy-​ makers’ ultimate objective. The framework Policy makers deliberate about corridor in this Report considers five categories of ­ features such as location, length, and mode of WEB: economic welfare, social inclusion, transport infrastructure for the transport cor- equity, environmental quality, and eco- ridor project. But they may also consider nomic resilience. The potential transmission complementary policies and institutions that mechanisms from a corridor intervention can help amplify the expected WEB. These package through intermediate outcomes to complementary interventions need to account wider economic benefits can be summa- for the constraints imposed by initial condi- rized in a Flow (chain) of expected results tions, including geography, population den- (figure O.1.). sity, market imperfections, and inequality of A corridor intervention can directly affect opportunity. the final outcomes (WEB net of costs) given The chosen set of interventions can be other complementary factors that affect characterized by different levels of ambition, 10  The WEB o f T r a n s p o r t C o r r i d o r s i n S o u t h A s i a FIGURE O.1  WEB are achieved through various transmission channels and intermediate outcomes Intermediate Wider economic benefits outcomes (WEB) (final outcomes) • Land value • Economic welfare • Migration • Income • Population • Consumption Corridor • Agglomeration • Assets intervention • Firm location • Social inclusion package • Investment/FDI • Jobs • Structural change • Gender • Productivity • Inequality • Trade • Poverty • Environmental quality • Air pollution • Deforestation • Economic resilience Source: Corridor Study Team. Note: FDI = foreign direct investment. FIGURE O.2  The corridor program can include trade facilitation from the most basic to the most comprehen- measures, as well as soft complementary policies sive (figure O.2): 1.  Investments in trunk transport corridors. This entails building entirely new transport infrastructure, such as roads, rail, or inland waterways. It may also involve upgrading Soft complementary interventions individual links or entire segments of exist- (policies and institutions) ing ones. 2.  Transport and trade facilitation ser- vices. Benefits from narrow investments in Transport and trade facilitation a trunk transport corridor can be enhanced (ports, warehouses, border crossings, by simultaneous investments and reforms auxiliary infrastructure, logistics) in enabling transport services and policies. It may also be supported by trade facilita- tion measures. The regulation of the truck- Trunk transport corridor ing industry belongs in the first group; the (road, rail, waterway) establishment of warehouses and border crossings belongs in the second. 3.  “Soft” complementary interventions. Initial conditions Benefits from improved regional connectivity (geography, population, market imperfections) can be further enhanced if the project design also addresses the most binding ­ m arket imperfections and institutional failures. Policy interventions could aim to ensure effi- Source: Corridor Study Team. cient functioning of capital, labor, land and O v e r v i e w   11 product markets, or improve institutions better jobs, and see their incomes grow faster. such as public sector governance, contract Thus impacts can be relative. But alongside enforcement, or access to social services. winners, corridor interventions could also produce losers in absolute terms. For Policy makers and other stakeholders may instance, by increasing efficiency, better trans- want to know which particular features of port connectivity could lead to simultaneous corridors and which complementary poli- job creation and destruction. This transfor- cies and institutions need to receive greater mation could require massive shifts in occu- weight. The choice of the mode of transporta- pations. People with more fungible skills may tion, or the length, location, and nodal con- be able to shift more easily, but others may nections of the corridor, could be its decisive face a harder time. characteristics. But land market reforms, Having clarity on the hierarchy of impacts improved access to finance, and regulatory across all these dimensions may help policy improvements in product markets could also makers in achieving multiple WEB, in manag- make the greatest difference. The answer is ing trade-offs, and in supporting possible ­ losers likely to depend on initial conditions, from from the corridor intervention (figure O.3). population density to the certainty of land titling or the extent and state of any pre-­ existing transport infrastructure. “2”: The “2” sorts of complementary interventions—Policies and institutions “T”: The “Typology” of impacts— Merging the layers of project design (the I in Organizing multiple impacts into a FIT) and the hierarchy of multiple impacts hierarchy from transport corridor interventions (the T) A transport corridor intervention has potential provides a useful tool to screen the quality of socioeconomic impacts across multiple WEB. project design. This tool can be illustrated in In some cases, these impacts may be positively the form of a simple matrix, where the rows correlated. For instance, the corridor could are the different layers of intervention design, boost both incomes and job ­ creation. In other and the columns present the overarching words, the corridor intervention could create ­ policy objectives (table O.1) synergies in development impacts, producing This simple screening tool could help beneficial effects for both economic welfare ensure that the fair distribution of economic and social inclusion. However, in other cases, benefits from transport corridor investments the impacts may be negatively correlated. For remains a priority in the design of the inter- instance, economic welfare impacts may be vention. Applying the matrix could discipline environmental impacts may be beneficial, but ­ policy makers by having them answer how detrimental. This leads to trade-offs between each layer of the design addresses the three different outcomes. policy objectives. For instance, cell (1,1)— Impacts may also be heterogeneous. That that is row 1, column 1—of the matrix asks: is, for a given outcome, they could vary sig- What are the expected impacts of the trunk nificantly across different geographic areas, infrastructure on the multiple WEB? Cell segments of the population, economic sec- (2,2) asks: Is the design of transport and trade tors, and the like. These varied impacts may facilitating interventions likely to generate benefit everybody, but vary in size depending trade-offs across individual WEB? Cell (3,3) on the beneficiaries’ greater or lesser predis- asks: Are there likely losers (relative and position to gain. For instance, more educated absolute), and what are the most effective and skilled population groups could benefit complementary market policies and institu- more from economic restructuring, take tions to support them? 12  The WEB o f T r a n s p o r t C o r r i d o r s i n S o u t h A s i a FIGURE O.3  A corridor intervention package triggers a hierarchy of impacts Multiple wider economic impacts Heterogeneous Impacts across effects individual WEB Relative Absolute Trade-offs Synergies (greater or smaller (winners and (income or (income and job predisposition) losers) environment) growth) Source: Corridor Study Team. Note: WEB = wider economic benefits. TABLE O.1  Combining the layers of a corridor program with the hierarchy of impacts gives a useful tool to screen for comprehensive project design Hierarchy of multiple impacts Layers of project design Achieve multiple WEB Manage trade-off impacts Support possible losers Trunk infrastructure ? ? ? Transport and trade facilitation ? ? ? Complementary market ? ? ? policies and institutions Source: Corridor Study Team. “Deed”: The “Deed” of devising a viable the expected returns from the corridor financing strategy for a given design investment could be monetized and how much could be in the form of social returns Ultimately, the cost of corridor investments is of a nonmonetary nature. The monetized borne by taxpayers and possibly by fee-­paying portion can be recovered directly, through users of the new infrastructure. A critical fees, or indirectly, through taxes. Because decision concerns the distribution of this bur- taxes and fees from corridor investments den across society. The decision depends on accrue only over time, whereas project costs social preferences, but also on who actually must be paid during the preparation and benefits from the corridor intervention. And construction phases, large corridor projects one product of the decision is the financing typically face a funding mismatch. The strategy to fill the gaps between costs and domestic or international financial systems returns over time. can help bridge the gap by mobilizing Developing a viable and efficient financ- resources nationally or globally. ing strategy starts by assessing how much of O v e r v i e w   13 Policy makers devising a financing strategy spend—by passing carefully selected risks and for a given design of corridor intervention can financing requirements onto the private in principle consider the following options: sector. However, PPPs must be devised and man- •  Increasing tax revenues or reallocating aged realistically and carefully (figure O.4). public spending that is currently funded They require adequate capacity in the public by existing tax revenues sector—usually at the level of the central gov- •  Increasing sovereign borrowing from con- ernment—to assess which risks can be passed cessional lenders, including multilateral onto the private sector, which ones need to be and bilateral agencies, and other providers retained, and how the retained and passed of concessional finance and grants risks are interconnected. The assessment is •  Increasing sovereign borrowing from even more complex in the case of corridor ­ private financial institutions, or through investments that cross borders. capital markets Risk management capacity varies across •  Leveraging public capital to mobilize countries, as it is affected by their institu- private equity and debt at the project level, ­ tional, legal, and practical contexts. When or through an investment fund (special risk management capacity is low, contingent purpose vehicle). liabilities from shifted risks may ultimately Several of these options could include public shrink fiscal space. As a result, countries may guaranties and other explicit or implicit be left with incomplete transport infrastruc- potential liabilities—including within public- ture projects, a large stock of bad loans in the private partnerships (PPPs). These contingent domestic banking system, and additional con- (potential) liabilities are originated with the tingent liabilities resulting from the need to intention of saving fiscal space—the space for recapitalize troubled systemically important the fiscal authority to further borrow and or state-owned banks. FIGURE O.4  India’s market for corridor PPPs fragmented as it expanded—and quality suffered a. Until 2004 (by volume of contracts) b. In 2012 (by volume of contracts) Others, Gamuda L&T, MSK projects 8% Malaysia, (India) Ltd., 8.28% IL & FS, 13% 3% 5.23% IL & FS, KMC Construction, 5% CIDBI 5.20% Malaysia-Malaysian, Gammon 13% Infrastructure ISOLUX Corsan, Ltd., 5.15% 7% 63 others less than 3% each, IVRCL, 4.50% Jaiprakash 50% IRB Infrastructure Industries Ltd., Developers Ltd., 9% 3.86% GMR Infrastructure, Navayuga Engineering MSRDC 11% Co. Ltd., 3.83% Mumbai, 10% Pacific Gammon Infrastructure Alliance Inc., Transtroy, Ltd., 3.70% GVK, Oriental Structural 3.51% 11% GMR Infrastructure, 11% Engineers, 3.51% 3.13% Source: Corridor Study Team of 2014 National Highways Authority of India data. 14  The WEB o f T r a n s p o r t C o r r i d o r s i n S o u t h A s i a Strengthening institutional and legal central governments is needed to frameworks, as well as well as the capacity strengthen administration capacity and for the public sector to carry out structured enhance the legitimacy of interventions. A finance tailored to the projects, is thus a key complementary action at the local priority when embarking on sizeable corridor ­ level is to upgrade land administration investments. Increasing the ability of the and cadastral records, so as to protect ­ government to transfer risk using complex property rights and enhance tax revenue legal and corporate entities can be an integral collection. Other local interventions may part of the corridor intervention package, or include skills upgrading programs, the be treated as a stand-alone governance development of economic zones, or stron- reform. ger environmental protection. This col- laboration may involve temporary increases in local budgets, as well as staff “Deed”: The “Deed” of successfully transfers from the central to the local gov- managing the implementation of the ernment, to boost local corridor interven- program tions. Possible tensions involving local governments that do not directly benefit Successfully managing the implementation from the corridor intervention must be process presents its own set of challenges. For anticipated, and managed early on. large cross-border corridors, four challenges •  Leveraging the private sector in delivery. stand out: Engaging the private sector in delivering •  Integrating expertise across sectors. To be commercially viable parts of complex cor- efficient and fair, corridor interventions ridor projects can help achieve WEB. need to keep in mind that the WEB are the However, successfully engaging the private ultimate objective, and they span sectors sector during implementation requires in nature. Therefore, the expertise that strong governance arrangements, suffi- needs to be mobilized for project design cient administration capacity, and clear and implementation extends beyond knowledge of private sector preferences. ­ transport infrastructure and its financing. Tenders must be transparent and efficient, A good understanding of urbanization, and not be plagued by conflicts of interest. trade, competition and the environment, The sequencing of implementation phases among other relevant areas, is rarely must be carefully planned; announced found in just a few staffers. Mobilizing timelines, and expectations should be such expertise and effectively integrating it adhered to; and uncertainty should be could be a challenge for both the country managed through effective communica- governments and the international organi- tion, among other requirements. For the zations that support corridor projects at private sector to engage, a regular dia- their different stages. Governments and logue with the business community at all international organizations may want to levels must be established. Without these build on the experience of other institu- foundations, the private sector could hold tions that need to quickly mobilize and back its investments until uncertainty integrate diverse expertise, such as the dissipates. U.S. space agency, NASA. This approach •  Managing cross-border complexity. This could lead to the identification of potential challenge could be tackled by establishing members of “Corridor Tiger Teams” to effective institutions, such as suprana- deploy well-integrated cross-sectoral tional committees. These standing or tem- expertise rapidly. porary bodies need to decide how to •  Boosting local delivery. Because transport partition the overall corridor program corridors concern specific locations, effec- into “implementable” projects. In doing tive collaboration between local and so, they must take into account regional O v e r v i e w   15 and national financing constraints, as well Generating large enough economic sur- as agency capabilities across borders. An pluses for the country requires a smart place- appropriate sequencing of subprojects ment of the corridors (map O.4). Policy must be embraced, with effective tracking makers can tackle this problem by prioritizing of the overall program delivery, and corridor placements that demonstrate gains in enforceable accountability for perfor- “centrality” from the new or upgraded con- mance. If not adequately empowered, nection. This is achieved through enhanced these supranational coordination bodies connectivity not only to important economic could fail. Alternatively, international hubs, but also to smaller cities whose market organizations could drive the coordination access is highly relevant for local producers. groups while playing the role of an honest But the “centrality” of cities along the corri- broker toward all countries involved, and dor could change over time as global demand, providing them with technical assistance. preferences, and trade evolve. Given the long life of corridor investments, policy makers need to account for such dynamics when MAIN TAKEAWAYS FOR making placement choices. PRACTITIONERS Another problem is that economic sur- pluses generated by transport corridors are When thinking about using transport corri- often unevenly distributed among the popula- dors as an anchor investment to promote tion. Realizing the WEB of a corridor is trade, regional integration, and broader imperative not only to maximize economic socioeconomic activity, policy makers and returns but also to ensure the socioeconomic practitioners should keep in mind four main and political sustainability of the investments. takeaways. Even if the returns from these investments are 1. Understand the challenge at hand large enough to repay the incurred debts and before proceeding with large transport remunerate the private investors, there can investments. still be resistance to their implementation if Corridors have a great potential as a develop- the distribution of the WEB is perceived as ment tool, but they are extremely costly—not unfair. only in terms of their direct outlays, but in Corridor investments may benefit some terms of other forgone development opportu- people more than others, depending on their nities, such as greater investments in educa- skills, security of land titling, access to tion, or in water and sanitation. Because they finance, or proximity to the trunk infrastruc- require large funding up front, with often ture, among other factors. Some people may commensurately large borrowing, corridor even lose in absolute terms. Failure to iden- investments can jeopardize fiscal sustainabil- tify who the relative and absolute losers may ity and macroeconomic stability. be, to enhance their chances to gain from the Risks are especially high when the WEB corridor investment, and to support them as from corridors are highly uncertain and their needed could instigate a political backlash contingent (potential) liabilities are badly against further corridor investments. Similar managed. Therefore, prudence is advisable to the backlash against globalization in some to make the most from the significant public advanced countries, this push would amount spending on transport corridor develop- to a missed opportunity to further promote ment, and to avoid the risk of white elephant trade, economic growth, and job creation. investments. Working with the private sector and the potential large users of the transport 2. Focus on WEB in designing corridor pro- infrastructure—such as local corporations— ­ grams right from the start. could help mitigate these risks by ensuring A narrow program scope could increase risks. enough traffic through the transport If the design of transport corridor programs is corridor. focused only on reducing vehicle operating 16  The WEB o f T r a n s p o r t C o r r i d o r s i n S o u t h A s i a MAP O.4  Where should South Asia focus its limited resources—toward the Middle East and Europe, toward Central Asia, or toward East Asia? Major Cities Black Sea Almaty Cas Tbilisi Road Istanbul Bishkek pian Yerevan Tashkent Baku Sea Ankara Ashgabat Dushanbe Kashgar Tehran Mediterranean Kabul Sea Shanghai Lahore New Delhi Kathmandu Thimpu Gwadar Silchar Karachi Dhaka Kunming Kolkata Chittagong Mandalay Arabian Sea Mumbai Bay of Bengal INDIAN OCEAN IBRD 43336 | NOVEMBER 2017 Source: Corridor Study Team. costs or saving time, important benefits from integrated within a comprehensive corridor the investment may be missed. Whichever intervention package right from the start. benefits are attained may not be widely Across the different local contexts covered shared, and negative impacts may arise. The in the Report, the most promising comple- corridor intervention may then fail to be mentary interventions appear to be upgrading transformational, and could even undermine skills and strengthening public sector gover- sustainable growth and social inclusion. nance around the corridors. Improving educa- Corridor investments affect multiple WEB tion coverage and quality, and upgrading land simultaneously, and can lead to trade-offs. administration systems are examples of poten- The analyses in this Report reveal that the tially important initiatives in this respect. most likely trade-offs arising is between Other promising complementary policies income growth and environmental degrada- include increasing openness to foreign trade tion, as well as between rising incomes and and promoting industry and trade worsening social inclusion (particularly lower competitiveness. labor force participation by women). The review of literature on spatial impacts Policy makers should give priority to of corridors covering 78 recent studies is tell- amplifying WEB, managing trade-offs, and ing in this respect. On average across coun- reducing the scope for absolute and relative tries, impacts on economic welfare and on losers to emerge. Various policies and institu- equity are positive and statistically significant, tional reforms (so-called “soft” interventions) but impacts on the environment are negative can be used to this effect, complementing the and even more significant. The effect on social investment in “hard” (physical) infrastruc- inclusion is positive, but only marginally sig- ture. But to identify the most promising nificant. Hence, trade-offs between welfare “soft” interventions, the potential WEB, and environmental quality are likely. But those trade-offs, and losers should be identified between welfare and social inclusion may early on. These “soft” interventions must be arise as well. The findings are similar for the O v e r v i e w   17 Golden Quadrilateral highway system in The importance of complementary inter- India. Comparing with control districts, GDP ventions is also revealed by simulations for grew significantly faster (by about 4 percent- the prospective China-Pakistan Economic age points) near the corridors, but so did the Corridor (CPEC) in Pakistan and the Kolkata- density of aerosol particles—a measure of air ­ Dhaka corridor between Bangladesh and pollution (figure O.5). Impacts on equity India. These simulations suggest that the pro- and social inclusion measures appeared posed corridors could have widely diverse insignificant. impacts on household expenditures, poverty, In a similar vein, the study of India’s the inclusion of women in the labor market, Golden Quadrilateral and North-South-East- and air pollution. West highway systems shows the importance To ensure that the WEB of these invest- of improving women’s access to private ments will be fairly distributed, complemen- finance. Easing their budget constraint tary interventions could be embedded in the enables them to move from farm to non-farm corridor intervention packages. For jobs in the districts crossed by the corridors instance, districts located in the central leg (map O.5). of the proposed Kolkata-Dhaka corridor FIGURE O.5  Corridors can increase incomes, but these gains come at the expense of environmental quality a. WEB from large corridor projects around the world b. WEB from India’s Golden (average significance of effects, in terms of t-statistic) Quadrilateral corridor (average of significant effects, in percent) 3.55 3.55 4 35 1.71 –1 25 Percentage change t-statistics –6 15 –11 5 0 –16 –15.02 –5 Economic Equity Social Environment Economic welfare Air quality-smog welfare inclusion (log of per (aerosol capita GDP) optical thickness) Effect of the GQ Control districts Source: Corridor Study Team based on Bougna, Melecky, Roberts, and Xu (forthcoming), and Melecky, Sharma, and Subhash (forthcoming). Note: In panel a, the t -statistic of 3.547 indicates a positive and reliable impact (at the 1 percent level) of transport infrastructure on various welfare outcomes (income, consumption, assets). The average impact on equity is not significantly different from the average impact on economic welfare. For social inclusion, the impact is smaller and/or less certain, but still, on average, marginally ben­ e ficial (t -statistic of 1.711). The impact of transport infrastructure on environmental quality is significantly detrimental: −15.023, on average. Panel b shows that the GQ highway increased GDP per capita growth over 2001−11 by 4 percentage points, over and above the baseline increase of 27 percent in control districts. However, the GQ highway also increased particulate pollution (measured by “aerosol optical thickness”) by an extra 0.02 points relative to a baseline increase of 0.06 points, indicating a significant trade-off between economic benefits and pollution. 18  The WEB o f T r a n s p o r t C o r r i d o r s i n S o u t h A s i a MAP O.5  Spatial impacts around major South Asia corridors vary significantly a. GDP per capita b. Female regular wage employment TAJIKISTAN TAJIKISTAN 0 to 200 0 to 16 GDP PER CAPITA 200 to 400 FEMALE REGULAR 16 to 33 CHINA 400 to 800 CHINA 33 to 49 AFGHANISTAN AFGHANISTAN WAGE EARNERS 800 to 1,200 49 to 66 1,200 to 1,600 66 to 82 PAKISTAN PAKISTAN >1,600 82 to 100 NEPAL HU H AN Road HUTA BHUTAN UTA NEPAL BHUTAN TAN TA AN Road BANGLADESH BANGLADESH INDIA MYANMAR INDIA MYANMAR Arabian Bay of Sea Bengal SRI LANKA SRI LANKA MALDIVES MALDIVES INDIAN OCEAN IBRD 43337 | DECEMBER 2017 Source: GDP per capita: India: Directorate of Economics and Statistics, Planning Commission, Government of India 2011; Bangladesh: Bangladesh Bureau of Statistics 2011; Pakistan: World Bank 2017a. The GDP per capita estimation is based on calculations based on the World Development Indicators and the South Asia Spatial Database (Li et al. 2015). Female regular wage employment: India: Census of India–Primary Census Abstract (PHC–PCA), Office of the Registrar General and Census Commissioner, India; Bangladesh: Labor Force Survey 2005; Pakistan: Social and Living Standard Measurement Survey 2012−2013. Note: Both panels display the Golden Quadrilateral highway, the North-South-East-West highway, and the anticipated Kolkata-Dhaka corridor. Panel a also includes the China- Pakistan Economic Corridor. Disclaimer: This publication follows the World Bank’s practice in references to member designations, borders, and maps. The boundaries, colors, denominations and other information shown in any map in this work do not imply any judgment on the part of The World Bank, ADB, JICA or DFID concerning the legal status of any territory or the endorsement or acceptance of such boundaries. and the northern end of the CPEC may The available technical approaches to experience lower increases in women’s appraise the potential of transport corridors employment in regular-wage jobs when to yield WEB have different strengths and there are restrictions on the use of land, lim- weaknesses. iting women’s ability to move from farm to The main trade-off is between the effi- non-farm jobs. Zoning constraints could ciency and reliability of an econometric prevent the establishment of manufacturing approach. Efficiency (precision) in this con- businesses, while other land ownership text means that the applied econometric restrictions could slow down the establish- approach can deliver answers (estimates, pre- ment of agro-processing businesses. Another dictions) with high confidence providing that factor behind the diversity of simulated the underlying problem is well understood effects of the CPEC and Kolkata-Dhaka and completely specified. Reliability (robust- corridor on household consumption is the ness), in turn, means that, even if the underly- varying share of private firms in total firms, ing problem is incompletely specified, the an indicator of market contestability at the econometric approach delivers broadly cor- local level (map O.6). rect answers or answers that point into the right direction. 3. Appraise the potential for WEB with Techniques that are more explicit about s patial data and reliable econometric ­ the economic structure and transmission methods. mechanisms behind economic corridors can O v e r v i e w   19 MAP O.6  The simulated impacts on per capita household consumption across districts along the Kolkata-Dhaka Transport Corridor (Bangladesh segment) call for complementary interventions Source: Corridor Study Team based on Melecky, Sharma, and Subhash (forthcoming). Note: The map shows the Kolkata-Dhaka Transport Corridor (KDTC) (Bangladesh segment). Units refer to average per capita household consumption expenditure, in logs. The red line indicates the highway. produce more precise estimates, while yield- or growth potential that could have influ- ing valuable insights about the transmission enced corridor placement in the first place. channels and mechanisms at play. But this is Reduced-form estimates are also unable to the case only if the underlying model is cor- shed light on indirect effects and precise trans- rectly and completely specified. If the model is mission mechanisms. even partially mis-specified, such techniques This trade-off between efficiency and reli- can produce misleading results. ability applies both to the identification of the In contrast, reduced-form econometrics are optimal placement of a transport corridor flexible and less prone to big mistakes in and to the optimal intervention design to model specification. However, this approach maximize its WEB (table O.2). struggles with the problem of endogenous From a policy perspective, the most prac- placement. That is, reduced-form estimations tical approach to appraise a proposed trans- have great difficulty distinguishing economic port corridor may be to combine network growth due to the corridor placement from analysis focusing on the centrality of corri- growth that would have happened regardless, dor connections with reduced-form 20  The WEB o f T r a n s p o r t C o r r i d o r s i n S o u t h A s i a TABLE O.2  The trade-offs between different econometric approaches to appraise proposed corridors give options Current reliability in Econometric potential Further research needed determining: in determining: to attain the potential: Technical approach Placement WEB Placement WEB Placement WEB Network analysis High Lowest High Low High Low Reduced-form econometrics Low Highest Low Medium Low Medium Structural general equilibrium Low High Medium High Low High Source: Corridor Study Team based on Melecky (forthcoming). regressions to capture the effects of comple- for private investment. Policy makers thus mentary interventions. need to ensure that the private sector under- The comparative advantages of differ- stands the corridor program, takes ownership, ent techniques make some of them better suited and is not overwhelmed by the risks. for appraisals that must be done r ­ apidly— Private investment can happen at the level when expertise, data, or funding are limited— of trunk infrastructure—the main arteries of and others better suited for appraisals that can road, rail, or waterway transport systems. It be more strategic and comprehensive, but take can also be at the level of auxiliary infrastruc- more time. Because the process of deciding on ture, such as feeder roads and rails, irrigation the placement and design of a transport corri- systems for agriculture, and industrial zones. dor program is dynamic, it would be advisable However, to maximize the WEB of economic to utilize more reliable techniques first, before corridors, private investments at the level of moving to more efficient but more demanding individual firms are as important as direct and possibly risky ones. participation in corridor-related investments. In practice, pre-feasibility studies and Private investment can help increase the assessments could be conducted using only financial viability of corridor projects and the network analysis focused on the concept of sustainability of public finances. More impor- centrality, combined with reduced-form tantly, private sector confidence is vital to econometrics to assess how the local context spur economic activity along the corridors affects the expected WEB. Later on, feasibility and generate the intended positive spillovers. studies and project appraisal could consider However, the experience with private more efficient techniques such as network sector involvement in the design, co-financing, ­ analysis accounting for varied responses of and implementation of transport corridors cities and markets on the corridor, and struc- has been mixed. For instance, with 466 pri- tural general equilibrium econometrics. This vate participations in infrastructure, India more elaborate approach would allow policy- may well be home to the largest public-­ private makers to think through possibly richer partnership (PPP) program in the world dynamics, and to identify direct and indirect (World Bank 2017b). However, by 2017, the effects of the proposed corridor intervention. program has produced a large pool of non- performing assets and created troubles for the 4. Engage the private sector better, consider- banking system (Singh and Brar 2016). ing disparities in regional development. Further, data on a set of transport corridor The success of a transport corridor interven- projects supported by international develop- tion could depend on how well the private ment organizations suggests that private sector and civil society organizations sector engagements have been modest in vol- ­ are ­i ntegrated into project design and ume, while questions arise about their quality. implementation. The involvement of industry ­ Impact estimates using various measures of associations, strategic firms, private financiers, WEB indicate that so far the engagement of cooperatives, and grassroots organizations the private sector might not have contributed can increase the effectiveness of the public to project success, but rather the opposite resources deployed. It can also pave the way (figure O.6). O v e r v i e w   21 FIGURE O.6  Private sector involvement in corridor projects has been limited, and its impact has been low a. Private sector involvement in corridor projects supported by international organizations, in percent Public-private partnership, 15% Contractor, 33% Not involved, 52% b. The effect of private sector consultation and involvement on the success of corridor projects, in average change in ratings, and average change in nightlight intensity and human settlement build-up 0.00 0.000 0.000 0.000 –0.05 –0.10 –0.15 –0.20 –0.25 –0.30 –0.302 –0.317 –0.35 –0.349 –0.40 Rating of WEB potential: Rating of WEB potential: Nightlight intensity and human private sector involved private sector consulted settlement data: private sector involved Parsimonious model Extended model Source: Corridor Study Team based on Alam, Herrera Dappe, Khan and Melecky (forthcoming). Note: Panel a reports the results of 60 transport corridor projects around the world. Panel b presents regression estimates using institutional ratings of “WEB potential” [1–5] and pooled nightlight intensity and human settlement data [−0.2; 4.8] along the corridor project alignments. Bars with a zero value indicate that the estimated impact is statistically insignificant. The parsimonious model groups complementary policy interventions across sectors, while the extended model disaggregates them by sector. WEB = wider economic benefits. 22  The WEB o f T r a n s p o r t C o r r i d o r s i n S o u t h A s i a Private and public sector incentives in rela- development and create opportunities in tion to the development of corridors may dif- previously lagging regions using public fer. The private sector has a relatively short investment and support (Government of investment horizon and is primarily profit- Malaysia 2006). oriented, not development-oriented. Private sector actors may not enter into corridor pro- *** grams by themselves because of the high risk of coordination failures, and they may be Transport corridor investments can be trans- especially reluctant in the case of cross-border formational and pave the way for greater eco- initiatives. It may also be difficult for them to nomic and social welfare. But they need to be mobilize the sizeable, long-term financial designed holistically to enhance and spread resources needed to bridge the gap between the gains while reducing the probability of investment costs up front and monetary negative outcomes. The examples, method- returns much later. ological framework, and results in this Report Thus, the public sector must take the lead are presented as a guide for policy makers to on corridors, listening to the private sector maximize the WEB of transport corridors in and trying to understand the implications of South Asia and beyond. its profit orientation, but keeping the devel- opment objectives in mind. One example of misaligned incentives concerns the tendering REFERENCES out and allocation of build-operate-transfer ADB (Asian Development Bank). 2008. Greater (BOT) contracts for transport corridors with Mekong Subregion: Maturing and Moving construction firms that are interested only in Forward. Operations Evaluation Department, the building phase of the project. Separating ADB. the building and operating stages of transport Alam, M., M. Herrera Dappe, A. Khan, and corridors can be a way to address this mis- M. Melecky. Forthcoming. “Corridor Project alignment of incentives. Design, Success Rating, and the Potential for Because the private sector is profit-­ Wider Economic Benefits.” Policy Research oriented, it will tend to cluster its investments Paper, World Bank, Washington, DC. around fast-developing growth centers on or Andrés, L., D. Biller, and M. Herrera Dappe. 2013. near the corridors. Public investment may Reducing Poverty by Closing South Asia’s need to correct the ensuing disparities in spa- Infrastructure Gap. Washington, DC: World Bank and Australian Aid. tial development, rebalancing the rapid Bougna, T., M. Melecky, M. Roberts, and Y. Xu. growth of corridor nodes with the slow devel- Forthcoming. “Transport Corridors and their opment in less productive regions. Wider Economic Benefits: A Critical Review of For example, in developing the Pacific the Literature.” Policy Research Paper, World Belt Zone, the government of Japan recog- Bank, Washington, DC. nized the need to address spatial divergence. Chandra, B. 2006. “Economic Nationalism and Various efforts to enhance the capital of the Railway Debate, circa 1880−1905.” In Our prefectures outside the Zone were promoted. Indian Railway. Themes in India’s Railway Urban functions such as commerce, produc- History, edited by R. Srinivasan, N. Tiwari, and tion, and education, were enhanced, trans- S. Silas. Delhi: Foundation Books. portation networks were improved, and Chatterjee, B., and S. Singh. 2015. “An Opportunity for India in Central Asia: The support was provided for private enterprises International North-South Transport Corridor to locate production sites outside major Belt Is a Chance for India to Build Links with Zone cities. Such efforts helped significantly Central Asia and Eurasia.” The Diplomat, expand local market size and increase May 4. https://thediplomat.com/2015/05​ an​ /­ employment opportunities in lagging -opportunity-for-india-in-central-asia/. regions (JICA 2009). Malaysia has also Donaldson, D. 2010. “Railroads of the Raj: strived to correct disparities in regional Estimating the Impact of Transportation O v e r v i e w   23 Infrastructure.” Working Paper 41, Asia ———. 2016. “Transport Corridors of South Asia Research Centre, Murdoch, Australia. and Caucasus.” Working Paper, The Institute of Government of Malaysia. 2006. Ninth Malaysia World Economics and Politics (IWEP). http:// Plan, 2006−2010. Putrajaya: Economic iwep.kz/files/attachments/article/2016-09-23​ Planning Unit. /­t ransport_corridors_of_south_asia_and​ JICA (Japan International Cooperation Agency). _­caucasus.pdf. 2009. “Ex Post Evaluation of Japanese ODA Palit, A. 2017. “India’s Economic and Strategic Loan Project: National Highway No. 5 Perceptions of China’s Maritime Silk Road Improvement Project (I) (II).” Report prepared Initiative.” Geopolitics 22 (2): 292−309. by Masumi Shimamura, Mitsubishi UFJ Research Roberts, M., U. Deichmann, B. Fingleton, and and Consulting Co., Ltd. https://www2​ .jica​ .go​ T. Shi. 2012. “Evaluating China’s Road to .jp/en/evaluation/pdf/2009_VNV -5_4.pdf. Prosperity: A New Economic Geography Kumar, R. 2015. “Delhi Mumbai Corridor: How Approach.”  Regional Science and Urban the World’s Largest Infrastructure Project Is Economics 42 (4): 580–94. Uprooting Indian Farmers.” The Guardian, Shepard, W. 2017. “India and Japan Join Forces September 15. to Counter China and Build Their Own New Li, Y., M. Rama, V. Galdo, and M. F. Pinto. 2015. Silk Road.” Forbes, July 31. https://www​ “A Spatial Database for South Asia: Paper.” .forbes​ . com/sites/wadeshepard/2017/07/31​ World Bank, Washington, DC. http://www. /­india​-and​-japan-join-forces-to-counter-china​ worldbank.org​/­spatialdatabase​-southasia -and​ - build​ - their-own-new-silk-road/#668​ M e l e c k y, M . F o r t h c o m i n g . “ A p p r a i s a l bac174982. Econometrics for Transport Corridors: Optimal Singh, C., and J. S. Brar, 2016. “Stressed Assets Placement, Intervention Design, and Wider and Banking in India.” IIMB-WP No. 507, Economic Benefits.” MPRA Paper, University Indian Institute of Management Bangladore Library of Munich, Munich. (IIMB). https://iimb.ac.in/research/sites/default​ Melecky, M., S. Sharma, and H. Subhash. /­files/WP%20No.%20507.PDF. Forthcoming. “Wider Economic Benefits of Srivastava, P., and U. Kumar, eds. 2012. Trade and Investments in Transport Corridors and the Role Trade Facilitation in the Greater Mekong of Complementary Policies.” Policy Research Subregion. Asian Development Bank. Paper, World Bank, Washington, DC. World Bank. 2017a. “Growth Out of the Blue.” Ordabayev, A. 2015. “The Geopolitics of South Asia Economic Focus, Fall 2017. World Transport Corridors in Central Asia.” Working Bank, Washington, D.C. https://openknowledge​ Paper, The Institute of World Economics and .​worldbank.org/handle/10986/28397. Politics (IWEP). http://iwep​ .kz/files/attachments​ ———. 2 0 1 7 b . P r i v a t e P a r t i c i p a t i o n i n /article/2015-07-05​/­geopolitics_of_transport​ Infrastructure (PPI) database. https://ppi​ _­corridors_in​_­central_asia.pdf. .­worldbank.org/. Accessed June 2017. Part I Going Beyond Just Infrastructure 1 Insights into Regional Integration from Three Historical Transport Corridors in South Asia T his chapter looks at three transport corridors in South Asia that have played important roles in history. The pur- significant in and of themselves, but rather as windows through which to explore effects of the deepening ties between two or more loca- pose is not to explore the key reasons and tions of major socioeconomic activity. drivers for change within this region and Unfortunately, attempts to build a detailed beyond it since 1600 or to shed light on the picture of the past are limited by gaps in long-term development of South Asia. Rather, source materials. The kinds of written materi- it is to assess the causes for three moments in als that might be helpful to understand com- time when rising exchange of goods, capital, plex issues such as gender, literacy, and social people, and ideas, and expanding patterns of change are not often available. Detailed data communication and trade, radically increased regarding topics like demographic growth, connectivity between regions, and to discuss urbanization levels, infant mortality rates, the consequences. and rises (or falls) in real wages are not The three corridors have been chosen to be always easy to come by. Furthermore, the distinct geographically, chronologically, and analysis must rely on partial, biased, or mis- causally. Each has been assessed separately. leading eyewitness accounts that focus on Consideration has been given to the historical what the commentator(s) of the time thought context under which an existing corridor was was interesting, rather than on what observ- broadened or a new corridor was opened. ers might pick up on today. It is worth stress- Careful attention has been paid to the impact ing the shortcomings of the evidence, for care that each had locally, regionally, and beyond must be taken not to overstate conclusions (if appropriate). The three moments and not to use the gaps and problems of chosen—1700, 1800, and 1900—are viewed the material that is available to oversimplify not as precise points in time that are complex issues. 27 28  The WEB o f T r a n s p o r t C o r r i d o r s i n S o u t h A s i a It is also important to recognize that trans- the empire’s first century, when the first portation corridors and the infrastructure Mughal emperor, Babur (1526−30), arrived that is put in place to establish or deepen in South Asia from Central Asia, and he and them can change considerably over time. his heirs created and expanded a unified terri- New means of travel and shipment of goods tory. Rather, the focus is on the 100 years that can have a dramatic impact over a very short followed, until about 1700. This was a period period. Conversely, roads and even railway that was broadly stable in terms of the domes- lines degrade over time because of a lack of tic history of South Asia. It was also a time use and/or investment. This can lead to sig- when the Mughals enjoyed peaceful relations nificant change, including weakening with their neighbors in Persia as well as of regional connectivity and the emergence of beyond the northern and eastern frontiers. unforeseen challenges (Kumar 2015, 42S). The case looks at how domestic reforms— coupled with an expanding economy that was fueled at least in part by changes in global MUGHAL INDIA, CIRCA 1700: trading patterns—deepened connections with A CORRIDOR LINKING SOUTH Central Asia. AND CENTRAL ASIA In the 150 years after the death of Babur, South Asia went through a period of transfor- The first case study examines changes to mation and consolidation. A series of military South Asia in the wake of the creation of the conquests, first by Babur, and then by his Mughal Empire. The period of interest is not heirs, created a realm that by 1650 stretched from Gujarat on the west coast of India to the Bay of Bengal in the east, from Lahore in the MAP 1.1  Mughal Empire, 1690 Punjab deep into central India, as far as the Godavari River. As new territories were brought under the control of the emperor and his court, as shown in map 1.1, administra- tive structures were first introduced and then streamlined to allow the single and growing polity to be managed effectively—although the pattern of change was not uniform across the region (Laroque 2004). Standardization of administrative, commercial, and judicial norms played an important role in enabling connections to deepen, and allowing for wider circulation and movements of goods, people, and services. These changes not only streamlined exchange within the Mughal territory but also stimulated rising levels of exchange. Such outcomes are also reported in work done by modern scholars on the impact of reducing or removing the costs of trade (De 2007; Anderson and van Wincoop 2001). The connectivity and socioeconomic vibrancy intensified following the European discovery of sea routes across the Atlantic and the nearly simultaneous successful rounding of the southern tip of Africa by Vasco da Source: Based on sheet 16A of Irfan Habib's An Atlas of the Mughal Empire (Habib 1982). Gama, which opened up maritime links I n s i g h t s i n t o R e g i o n a l I n t e g r a t i o n f r o m T h r e e H i s t o r i ca l T r a n s p o r t C o r r i d o r s i n S o u t h A s i a    29 between Western Europe and Africa, the produced and the available number of Arabian Gulf, the Indian Ocean, and beyond. patrons, donors, and buyers seeking to com- By the early 1600s, increasing volumes of mission and acquire art. India’s cultural trade were beginning to move between the golden age followed a similar trajectory (van Americas and Asia via the Pacific Ocean, fun- der Woude 1991). damentally changing global trade networks Many artistic forms had their origins in (Flynn and Girláldez 1995). Large numbers of Central Asia, where the Mughal Dynasty merchants were drawn to South Asia. Levels began. Indeed, Timurid influences not only of maritime traffic reaching ports in Bengal affected art but also popularized new pas- and in Gujarat rose sharply. The port of times, such as wrestling and pigeon racing Surat, in particular, not only boomed as a (Foltz 2007). 1 Fruits and foodstuffs from result but also became part of an interconti- this region also began to be traded in nental trading network (Moosvi 1990). increasing volumes, with reports that By the early 1600s, the East India fruits from “Persia, Balkh, Bukhara, and Company and the Dutch East Indies Samarkand” could be found in markets in C o m p a n y ( Ve r e n i g d e O o s t - I n d i s c h e Delhi, while the Emperor Jahangir received Compagnie) had both established major foot- apples and grapes from Central Asia holds in Gujarat (Jha 2005). These trading (Alam 1994). However, the expansion of a companies situated themselves in key loca- transport corridor running north from the tions to supply goods, products, and com- Indian subcontinent owed less to the nostal- modities from South Asia—and beyond—with gic desire to (re)connect to a former home- substantial markups in Europe as well as in land than to intensification of exchange Africa and the Americas. Central to these based in rising spending on one resource in arrangements were the technological break- particular: horses. throughs in shipbuilding that enabled ocean- The interest in horses coincided with rising going vessels to travel further, faster, and more investment in the military. Inflows of capital safely than ever before. This, coupled with from outside South Asia, alongside the rising vast expropriation of silver and gold from the levels of interaction and exchange across a Americas in the sixteenth and early seven- highly sophisticated, monetized, and unified teenth centuries, fundamentally transformed region, resulted in a greater ability to spend global bullion flows, and had effects that on the military. Spending on the army was a were dramatic and far-ranging on China and vital means of strengthening internal control, Southeast Asia, as well as South Asia (Atwell as well as offering the opportunity to expand 1982; Barrett 1990). the territory still further, if and when the The influx of hard currency into South opportunity arose. It also, of course, provided Asia had a profound impact on its domestic an important means of defense against oppor- economy. The textile industry, in particular, tunistic or ambitious rivals. A considerable grew at a staggering rate to keep up with amount of money was spent buying horses demand. By the late 1600s, the East India for the military. As many as 100,000 were Company was exporting more than 1.5 mil- bought each year by the start of the eigh- lion pieces of textiles per year, and the Dutch teenth century (Gommans 1991). East Indies Company almost the same amount The most desirable horses were bred in (Prakash 2007). Rising levels of wealth flow- Central Asia. They were more robust, stron- ing into the imperial court led to a golden age ger, and bigger than those reared in South of building, visual arts, and calligraphy, as Asia, which were “by nature so small that, wealthy patrons sought out the best artisans, when a man is upon them, his feet nearly and gave them funds to produce their finest touch the ground” (Gommans 2002, works. As demonstrated in European art in 112–13). With the economy in South Asia this period, there is a direct correlation booming, vast profits could be made from between the quantity and quality of art that is selling the best steeds, as a surge in demand 30  The WEB o f T r a n s p o r t C o r r i d o r s i n S o u t h A s i a outstripped supply. So great were the poten- substantially and grew rapidly in the tial returns on horse sales in South Asia that decades before 1700. Lahore flourished as Europeans not only became actively grand new monuments were built and open involved in equestrian trade but also began spaces and gardens were created to adorn to innovate in ship designs, in an attempt to the city (Westcoat, Brand, and Mir 1993). maximize margins by bringing in stallions Other locations like Jaipur, Jodhpur, and from Arabia and Persia (Jardine and Brotton Udaipur grew exponentially, fueled by simi- 2005). lar migratory patterns that were the hall- The corridor running north from north- mark of labor force movements in western ern India was not new. Often referred to as India in this period (Washbrook 1993). the Grand Trunk Road, it formed part of an Rapid urbanization along the corridor put ancient network that connected Kabul to pressure on water and food supplies. India’s Peshawar and Lahore, via Delhi through unpredictable monsoon climate, and there- Bengal, as far as Chittagong in what is now fore its harvests, exacerbated the difficulties Bangladesh. Improvements had been made posed by a period of rapid urban growth and in the early sixteenth century, thanks to the relationship between towns and their Emperor Sher Shah Suri (1530−45),2 who hinterlands (Kolff 1991). Although the agri- not only oversaw a major expansion of this cultural landscape was transformed as a (as well as other routes) but also set in place result of demographic change, what hap- an ambitious program to develop and pened in urban settlements affected what ensure its long-term future by setting up was grown and where—rather than the other hostels where travelers could rest on their way round (Habib 1969). journeys along the new arteries (Singh Social structures in the Mughal-controlled 1991, 173−203). By the late 1600s, the territory in South Asia underwent consider- scale of the horse trade prompted major able change as connections deepened and eco- investment into infrastructure. Bridges were nomic interaction accelerated, with a rapidly built, caravansaries (inns that could accom- expanding class of city merchants, traders, modate caravans) were upgraded, and the shopkeepers, and moneylenders swelling security of the roads running north was alongside the small groups of rich elites who improved (Dale 1994). had previously dominated (Mukerjee 1968). One result of the increasing levels of The elites, traditional or otherwise, played a exchange, both within what is now north- central role in shaping the cities, as well as the ern India and the Punjab and regions lying trade routes of the empire (Naqvi 1968). beyond, was the surge in civic buildings in Improved communication networks enabled many of the major urban centers along the faster transmissions of information to the transport corridor running north—and in political center, which resulted in a greater some cases, even of the foundation of ability if not to rule the empire, then at least entirely new cities. Gateway cities that were to receive more accurate news in a timely well located to host markets, and particu- manner, consolidating the power of elites larly horse markets, flourished, with the (Mehta 1979, 303−4). prestige and power of their governors and It is difficult to accurately assess the impact rulers rising accordingly (Gommans 2002). of high levels of mobility, given the records Cities like Delhi and Kabul benefited and historical material available. However, it is worth noting that competition for labor and skills was intense between traders and merchants, whether foreign or otherwise. The Even the earliest corridor studied was clearly linked rising investment in cities, and the prospects to global trading patterns and to rhythms of and opportunities that they offered, had a sig- supply and demand hundreds, if not thousands, of nificant effect on social structures. Goods, miles away. skills, and properties could be bought and I n s i g h t s i n t o R e g i o n a l I n t e g r a t i o n f r o m T h r e e H i s t o r i ca l T r a n s p o r t C o r r i d o r s i n S o u t h A s i a    31 sold by an increasing number of people, BENGAL AND THE EAST INDIA which meant that housing, farmland, and COMPANY, 1745−1813: THE assets that were better located and closer to TRIANGULAR TRADE BETWEEN marketplaces became more and more valu- BRITAIN, INDIA, AND CHINA able, in turn increasing commercialization The second case assessed analyzes the routes and promoting further growth. The result, emanating from the booming city of Calcutta inevitably, was to create surges that increased (now Kolkata) from the middle of the regional imbalances rather than correcting eighteenth century onward. The corridor them (Washbook 2007). resulted from the efforts of the London- The transport corridor that linked South based East India Company (EIC) to establish and Central Asia did not peter out and come and intensify connections across the Bay of to an end. Like all corridors, as well as carry- Bengal throughout the 1700s to create a ing traders and merchants, it could also dominant mercantile and political position enable those with less benign intentions to in South Asia. spread easily. In the 1730s, the Persian ruler From 1745 to 1813, the EIC reached full Nader Shah used the route linking maturation, faced numerous crises, and Afghanistan and Central Asia with Delhi to morphed into a colonial state in all but name. strike at the very heart of the Mughal Empire. The city of Calcutta was integral to the rise The attack provides a useful reminder of the and continuing prosperity of the EIC in fact that corridors can sometimes prove to be Bengal and throughout South Asia.3 The city’s conduits for unexpected problems and trou- pivotal role in international trade and its bles (Daryaee and Aghaie 2012). imperial might stemmed from its sophisti- The threat to the routes running north in cated inland and maritime connectivity across the eighteenth century put urban life under regions. Calcutta became the wealthiest and pressure, as did power shifts within the most important center of the EICs’ activities Mughal world and the rise of urban centers by the early eighteenth century. As a result, elsewhere in South Asia. According to eye- heavy fortifications were built to protect com- witnesses, some towns and elites fell on hard mercial interests (Lawson 1993, 46−48). times, with the result that artisans, traders, The subject of continued fascination across and craftsmen moved away in search of the globe, the EIC is remembered as a formi- opportunities elsewhere (Bayly 1983, dable organization, at once respected, feared, 112−13). Then and now, labor forces, espe- and loathed. An English and later a British cially collective groups of skilled workers, can joint-stock company, formed to pursue trade exercise considerable autonomy when it with the East Indies, the company rose to comes to relocating to more fertile pastures account for half the world’s trade in commodi- (Lucassen 2006). ties, ranging from cotton, silk, indigo dye, and Nevertheless, another burst of life for the salt to saltpeter, tea, and opium (Roy 2012). corridor was not long in coming. In the first The EIC’s interests turned from trade to half of the nineteenth century, major invest- territory during the eighteenth century, as the ments went into road building, which was Mughal Empire declined in power and the overseen first by the East India Company and EIC struggled with its French counterpart, then by British imperial officials across South the French East India Company (Compagnie Asia, including the first corridor under con- Française des Indes Orientales) during sideration here. “As a consequence,” noted the Carnatic Wars of the 1740s and 1750s. the Morning Chronicle, “the map of India The British defeat of the Indian power at the began to shrink.” Communities, towns—and Battle of Plassey and Battle of Buxar—led by India itself—were becoming more closely Robert Clive—left the EIC in control of connected than ever before. Then, as now, Bengal and a major military and political investment in infrastructure often yields long- power in India. In the decades that followed, paying dividends. 32  The WEB o f T r a n s p o r t C o r r i d o r s i n S o u t h A s i a it gradually increased the extent of the terri- until 1858, when—following the Indian tories under its control, ruling either directly Rebellion of 1857—the British Parliament or indirectly via local puppet rulers under the passed the Government of India Act 1858. threat of force by the armies of its three presi- The British Crown assumed direct control of dencies, much of which were composed of India in the form of the new British Raj. Indian privates (sepoys) (Heathcote 1995). Maps 1.2 and 1.3 highlight this expansion of The EIC offered its own services in the Indian British rule on the subcontinent. “military bazaar” (Bayly 1988, 48). The EIC was undoubtedly extractive, but The EIC eventually came to rule large it also brought Britain and India into close areas of India with its own private armies, connection before the development of the exercising military power and assuming British Empire in Asia, in ways that empow- administrative functions. Company rule in ered both British and local Indian interests— India effectively began in 1757 and lasted from channeling financial capital into MAP 1.2  Territories of the British East India Company, 1765 and 1805 Source: Edinburgh Geographical Institute, Imperial Gazetteer of India, 1907, volume 26, page 190, Atlas Published Under the Authority of the Secretary of State for India in Council by Oxford University Press (https://en.wikipedia.org/wiki/Company_rule_in_India#/media/File:India1765and1805b.jpg). I n s i g h t s i n t o R e g i o n a l I n t e g r a t i o n f r o m T h r e e H i s t o r i ca l T r a n s p o r t C o r r i d o r s i n S o u t h A s i a    33 MAP 1.3  Hindustan, 1864 Source: A. Johnson’s “Hindostan” map (http://westerncivguides.umwblogs.org/2013/12/06/the-british-in-india/). First published in 1860 in New York (http://www.davidrumsey. com/luna/servlet/detail/RUMSEY~8~1~2540~320020:Hindostan-Or-British-India-). domestic investment within India to the con- new ports to augment trade and ensure com- struction of infrastructure projects that pany dominance. An important example of expanded connections across the subconti- this was the EIC’s instrumental role in devel- nent to shaping processes of state formation oping Calcutta in the region of Bengal from and the realms of social and cultural life. the 1690s onward. Over the next century, Nevertheless, the role played by the EIC is the city was transformed not only into a highly controversial, and one that is not commercial hub that linked regional, inter- always easy to assess objectively. regional, and even global trade but also into This international connectivity was made an outstanding center of learning and a con- possible not only by the EIC’s technological tact zone between different people, faiths, development and exploratory endeavor in languages, and ideas (Raj 2011). global maritime travel but also by its private With a network of inland and coastal investment in creating and expanding vital transport corridors cultivated around them, 34  The WEB o f T r a n s p o r t C o r r i d o r s i n S o u t h A s i a port cities like Bombay (Mumbai), Madras in the lifespan of the EIC. Robert Clive, origi- (Chennai), and Calcutta (Kolkata) acted as nally sent to reform the company’s civil and transport hubs for the regions that the EIC military services, to end wars, and to return dominated. The economic importance of the company to a stable footing, had instead Calcutta came to dictate the geography and opened a Pandora’s Box by accepting and timing of transport regionally, nationally, and promulgating the diwani. internationally. News of the diwani and their value circu- By 1760, Bengal had an Indian ruler lated through printed media, and Clive’s backed by the EIC. In 1764, the British boasting brought immediate public scrutiny to defeated the Mughal emperor and his ally, the conduct of the EIC’s servants (overseas the Nawab of Oudh, at Buxar (Dyson 2002, employees) in India and the behavior of direc- 8). The British used this victory to gain con- tors and shareholders at home. Upon hearing siderable political leverage over Bengal. The the news, and against the directors’ advice, the EIC, under the guidance of Robert Clive, EIC’s shareholders voted to increase the divi- obtained the diwani—the rights to collect dend on their investments from 6 percent to revenues—in Bengal, Bihar, and Orissa 10 percent. This was collected by raising the (Odisha). In the guise of extracting a “divi- taxes imposed on traded goods across Bengal dend” for its services, the EIC now effec- and elsewhere. The move precipitated tively taxed the local authorities, as well as Parliament’s investigation into the EIC’s all traded goods. It never denied the author- affairs. In London, the possibility that share- ity of the Mughal emperor, but began to holders—passive members of the public— manage the administration of these three might benefit from the spoils of Indian regions—acting like other independent territories while excluding government “successor states.”4 Thus by 1765, the EIC spawned deep resentment. Parliament imposed was the de facto sovereign in Bengal and new rules on the election of directors, levied a collected territorial revenues estimated to fine of £400,000 per year on the company, be worth £2–4 million per year in the prov- and regulated the size of the dividend inces of Bengal, Bihar, and Orissa (Bowen (15 percent of gross sales in 1767/68). Thus 1989, 187). the benefits of the diwani remained specula- The EIC benefited from this time of flux tive, while the costs for the company were and opportunity. As the volume of textiles in tangible. the triangular trade between Britain, China, After 1773, the EIC’s commercial ethos and India increased greatly, Bengal, and spe- came under increased scrutiny and pressure cifically Calcutta, became extremely impor- in South Asia. This followed a catastrophic tant to British interests (Bayly 1988). The and devastating famine in Bengal—which lines between the interests of the EIC, its was directly linked to the EIC’s misadminis- shareholders, its officers, and the British gov- tration, the greed of its officers, and its gro- ernment in controlling Bengal (and other tesque incompetence—led to the death parts of India) were often blurred—no doubt through starvation of millions in Bengal sometimes deliberately (Marshall 1987). (Arnold 1999). The EIC’s officials in Bengal After the diwani were introduced in continued to devote greater attention to Bengal, the scope and scale of regulatory revenue administration than to acquiring actions were unprecedented. All major groups piece goods, raw silk, or saltpeter. of insiders—shareholders, directors, and Particular attention was paid by the com- employees—increasingly found key decisions pany and its officers to rationalizing and subject to Parliamentary fiat. From 1784, simplifying taxation, both on the value of the EIC’s political imperatives largely over- land and on transactions in towns and mar- shadowed the profit motive. The diwani of kets across Bengal and beyond (Travers Bengal were therefore a fundamental moment 2004). I n s i g h t s i n t o R e g i o n a l I n t e g r a t i o n f r o m T h r e e H i s t o r i ca l T r a n s p o r t C o r r i d o r s i n S o u t h A s i a    35 The creation of a new government Procurement of cotton textiles was there- department, the Board of Control, under fore the crucial export underpinning the EIC’s William Pitt’s India Act (1784) deepened venture into Bengal. This venture was made and institutionalized the ministerial over- possible by the thousands of weavers dis- sight that had begun after the diwani were persed across the countryside. By the 1750s, introduced. The act required servants (the textiles accounted for more than 80 percent EIC’s overseas employees) returning to of the value of EIC’s exports from Bengal. Britain to account for the source of their Cotton products, in turn, constituted more fortunes. The company had greatly inflated than 88 percent of the value of textile prod- salaries across all ranks of servants, even as ucts (Chaudhury 1995, 182, 192). Before the it continued to add to its workforce. diwani period, Bengal thrived in the export of Profiteering from private trade was no lon- cotton textiles—that is to say, in the sale of ger essential for recruiting and keeping finished products, as opposed to the raw young men in the covenanted service, with material itself, which came mainly from the the attraction of good salaries and high sta- Deccan Plateau (at least in the eighteenth tus proving sufficiently attractive to swell century). the ranks. Between 1784 and the passage of the Charter Act of 1813, which ended the EIC’s monopoly of trade to India,5 the com- The mobility of the labor force has been a clear factor pany became a company in name only. But in the ability not only to carry out infrastructure did this shift in authority alter the network projects but also in the expansion of cities and the of transport corridors emanating from development of interconnected economies. Calcutta? What lessons can be extrapolated from the case of the late-eighteenth-century Bengal? The EIC primarily procured these tex- tiles using the Agency System. The EIC hired local employees—agents—to transact with The growth of Bengal’s textile industry weavers. Typical agreements with weavers ­ was accompanied by new procurement specified a loan for working capital, the qual- patterns and labor force and legal ity and quantity of cloth to be produced, and arrangements prices contingent on quality. But the system Bengal’s inland trade routes facilitated the did not work so well. It was fraught with growth of its textile industry. With the “corruption”—or opportunistic behavior— ascendancy of the diwani system and by the agents, the weavers, and EIC officials. the EIC’s shift to combine politics and busi- Agents often did not uphold payment agree- ness, the textile industry in Bengal changed ments and cheated the weavers, and weavers greatly. Before political interests emerged, often sold output to other buyers and thereby Bengal was a major textile-producing did not repay their debts to the EIC (Kranton region on the subcontinent, alongside and Swamy 2008).6 Gujarat and the Coromandel Coast, spe- From the outset, there were unrealistic cializing in the production of luxury cotton, expectations of what taxes the local popula- silk, and mixed textiles and therefore domi- tions would be able to pay, as well as the nating textile exports to the European mar- speed and efficacy of plans put in place to ket. At the turn of the eighteenth century, transform the agrarian society and better as much as 40 percent of the total cargo extract surpluses. Although the limitations exported to Europe from Asia by the Dutch were recognized by some at the time, a as well as the EIC originated in Bengal, and strong cavalier, get-rich-quick attitude pre- consisted largely of textiles and raw silk vailed. It took time for this attitude to change (Prakash 1998). (Travers 2005). 36  The WEB o f T r a n s p o r t C o r r i d o r s i n S o u t h A s i a It was no coincidence, therefore, that as quite conform to the evidence available the regional and interregional exchange of (Prakash 2007). It was only in the second half goods accelerated, so did the services of the nineteenth century that the impact of and ideas needed to move forward. This the British cotton textile industry was felt in process of rationalization occurred not full force by the Indian handloom sector. only in Bengal but also in other towns, However, shifting characteristics along locations, and territories where the EIC these inland corridors over the period had interests. immediately before and after the diwani era Nowhere was this clearer than with the provide interesting insights about the far- spread of English law and even of the law reaching effects of international trade on courts themselves. By the early nineteenth regional networks. The EIC set up many trad- century, Hindu and Sikh merchants who felt ing outposts, employed a massive army of that they had not been served fairly or justly Indian agents and dealers, and advanced in the courts in India brought their cases all money to producers through these channels the way to Westminster and even the Privy (Dasgupta 2000). The EIC oversaw the Council. The expansion of the transport cor- creation of a supply chain dependent on ridor may not have been responsible for jus- the British. tice being delivered to all, but it opened new Indian agents, dealers, and intermediaries possibilities for those seeking to resolve dis- therefore played a major role in spreading, putes (Fraas 2014). This shift also occurred first, the EIC’s economic power and, then, its in urban centers, including Calcutta, where political power through their extensive net- the increasing velocity of transactions led to work in both urban and rural areas. The rising levels of disputes over tax collecting inland corridors of Bengal and beyond paved (Sengupta 2015). the way for the establishment of colonial The political intrusion into mercantile rule. Some scholars have suggested that with practice hampered the growth of the textile its acquisition of political power in Bengal, industry in Bengal. However, overseas politi- the EIC did away with the services of these cal involvement was not altogether disas- intermediaries, who were ruined in the pro- trous. The volume and value of the textile cess (Bagchi 1981). However, more recent trade carried on by the Europeans from critiques argue that the company considered Bengal in the second half of the eighteenth the services of these intermediary merchants century was much larger than the first half, indispensable (Chakrabarti, 1994, 1). thanks to the substantial increase in the textile Although the relationship between the trade carried on by the EIC and private EIC and the Indian merchants was essentially English traders in the wake of its Dutch and one of collaboration, it was often marked by French competitors. competition and conflict. The Indian mer- Furthermore, Indian merchants continued chants stubbornly resisted all attempts by the to control the bulk of the textile trade between EIC to ensure subordination solely on its Bengal and other parts of the subcontinent own terms. The traders of eastern India after and Asia. Large quantities of cotton were the Battle of Plassey were very different from being carried from the Deccan Plateau to the medieval maritime merchants who had Bengal, partly for use as raw material in the once dominated the coastal port towns of textile sector in the province. Interregional India in Surat, Calicut, Masulipatnam, and corridors were maintained, while interna- Hughli (Das Gupta 1967). tional maritime routes helped connect con- During the period after diwani, the mer- sumers in Europe, Africa, and even North chant princes—such as Khemchand and America. The picture of a ruined textile Chintaman Shah in Balasore—disappeared industry in Bengal during this period that is from the scene, but they were replaced by a portrayed in some of the literature does not new mercantile order constituted of men from I n s i g h t s i n t o R e g i o n a l I n t e g r a t i o n f r o m T h r e e H i s t o r i ca l T r a n s p o r t C o r r i d o r s i n S o u t h A s i a    37 disparate social groups. The disappearance of Therefore, while Bengal’s industrial decline merchant princes did not necessarily mean a started in the early 1830s and continued in decline in the fortune of Bengal’s merchants. the decades that followed, the transport The flourishing foreign trade had provided an routes created during the EIC’s life as a pri- excellent opportunity for the indigenous vate firm continued to play a role during the intermediary merchants to amass wealth and early colonial period. Textiles remained a lay the foundations of the commercial aristoc- major export to the European market. The racy in Calcutta. Further, the demand created industry’s cross-caste opportunities to connect by European traders lured large numbers of weavers with the EIC were expanded to indig- rural workers—whose traditional caste occu- enous intermediary merchants, and the Bengal pations were different—to the new opportu- economy was propped up by the introduction nities (Chaudhury 1995, 4). of new industry. Consider the impact on employment of Consequently, while Calcutta and its five major manufacturing industries in Bengal transport arteries inland and overseas were during early colonial rule: silk textiles, cotton monopolized by the EIC, stifling international textiles, salt manufacturing, ship-building, trade for the French and Dutch colonial econ- and indigo dye. Tracing the precolonial roots omies in South Asia, the institutionalized of each, Ray (2011) assessed the roles of mar- ministerial oversight brought about by the ket forces and state policies on their perfor- diwani and Board of Control under Pitt’s mance. The data on employment are shown India Act provided opportunities for those in table 1.1. left out under the rule of market forces. Table 1.1 shows that far from being Industrial decline is traditionally thought as deindustrialized, Bengal enjoyed industrial being a consequence of the EIC’s turn from prosperity from 1795 to 1829. The level of business to politics. But this conclusion is ill- employment in the five major industries founded. In some cases, market forces instead increased overall by almost 1 million people. initiated and perpetuated the event. This was due to the prosperity of the preco- Nevertheless, the transition in mind-set and lonial industries like cotton and silk textiles priorities that came with the EIC’s mutation as well as salt manufacturing on the one were both aided and driven by the continued hand, and the emergence of new industries use of transport corridors throughout like shipbuilding and indigo dye on the other. Calcutta and Bengal. TABLE 1.1  The five major industries in Bengal created almost one million jobs from 1795 to 1829, thanks to growth in both precolonial and new industries Number of jobs Year range Silk Cottona Salt Shipbuilding Indigo Total employment Change in employment 1795−99 88,775 179,905 88,020 928 460,080 817,708 — 1800−04 84,040 198,931 90,303 4,508 522,478 900,260 +82,552 1805−09 97,255 141,798 108,567 2,400 833,419 1,183,439 +283,179 1810−14 155,536 126,745 113,639 5,400 868,826 1,270,146 +86,707 1815−19 158,109 210,218 114,655 5,589 994,757 1,483,238 +213,092 1820−24 202,242 145,589 123,785 2,341 1,040,878 1,514,835 +31,597 1825−29 219,267 56,856 121,212 1,429 1,364,060 1,762,824 +599,068 1830−34 188,460 −21,616 149,887 1,074 1,230,295 1,584,100 −565,803 1835−39 237,786 −53,537 93,947 1,626 1,146,199 1,425,985 −122,115 Source: Ray 2011. a. Cotton figures represent changes in employment opportunities. 38  The WEB o f T r a n s p o r t C o r r i d o r s i n S o u t h A s i a RAILROADS OF THE BRITISH growing Indian population and more rap- RAJ, 1853−1929: CHANGING THE idly deploy troops. TECHNOLOGY OF TRADING ON The appealing prospect of railroad trans- THE SUBCONTINENT portation in India was discussed as early as 1832 (Sanyal 1930), though it was not until The introduction and expansion of rail- 1853 that the first track was laid between ways was a key determinant in the develop- Bombay and Thane—the start of what ment of India leading up to and following would form the Great Indian Peninsula 1900. From drastically altering the distri- Railway (GIPR). The importance of the bution of the food supply to accentuating inauguration can be gauged by the fact that social differences, the railroad network Bombay’s ­ government declared it a public designed and built by the British Raj holiday. brought dramatic change to the technology With the East Indian Railway Company, of trading on the subcontinent. The trans- the Carnatic Railway Company, and the port corridors of this network are the focus Great Southern India Railway Company of this section. (later the South Indian Railway Company) Before the railroad age, bullocks carried established between 1845 and 1864, tracks most of India’s commodity trade on their started to spread from the larger cities on the backs, traveling no more than 30 kilometers subcontinent. These included links from the (km) per day along India’s sparse network of Madras Presidency in 1859, the 80-km link dirt roads (Deloche 1994). By contrast, rail- from Trichinopoly to Negapatam in 1861, roads could transport these same commodi- and the Madras-Arakkonam-Kancheepuram ties 600 km in a day, and at much lower line in 1865. However, the most influential per-unit distance freight rates. Therefore, as day in the history of Indian rail was arguably rail penetrated inland districts, local adminis- March 7, 1870. On this date, the Great Indian trative regions were brought out of near- Peninsula Railway and the East Indian autarky, and previously isolated provinces Railway were linked at Jubbulpore (now were connected to the rest of India and Jabalpur). This 2,127-km railway line con- the world. necting Kolkata and Mumbai via Allahabad To give a brief overview of the expansion is now known as the Howrah–Al–Mumbai of rail in India over this period, it is critical line. It resulted in a combined network of to trace the roots of rail. The impetus was 6,400 km (4,000 miles), allowing goods and arguably a major failure of the cotton crop passengers to travel directly from Bombay in in America in 1846. Following this disrup- the west to Calcutta in the east via Allahabad tion, textile merchants in Manchester, in the northeast (where the Ganges, Yamuna, England, and Glasgow, Scotland, had to and “invisible” Saraswati rivers meet).7 This seek alternative markets. It was then that connectivity over such a grand scale was traders in the United Kingdom turned their unprecedented, providing the inspiration for attention to the cotton crop in India—a French writer Jules Verne’s book, Around the British colony rich in cotton. However, cot- World in Eighty Days. ton was produced in various parts of the A year later, the lines from Bombay and Indian subcontinent. It took days to bring it Madras met at Raichur in Karnataka, as to the nearest port to transport it to shown in map 1.4. By 1875, British com- England by ship. A link was needed from panies had invested about £95 million in the hinterlands to India’s major ports to rail in India. By 1880 the network extended speed the transport of cotton and other about 14,500 km (9,000 miles), mostly goods as demand soared. This need radiating inward from the three major port impelled the British to introduce a railway c ities of Bombay, Madras, and Calcutta. ­ in India, alongside the belief that rail could By 1895, India had started building its help the British organize and disperse the own ­ locomotives. In 1896, it began sending I n s i g h t s i n t o R e g i o n a l I n t e g r a t i o n f r o m T h r e e H i s t o r i ca l T r a n s p o r t C o r r i d o r s i n S o u t h A s i a    39 engineers and locomotives to help build the MAP 1.4  Completed and planned railway lines in India, 1871 Uganda railways. At the start of the twentieth century, the Great Indian Peninsula Railway had become a government-owned company. The network spread to the modern-day states of Assam and Rajputhana. Soon, various autonomous kingdoms began to have their own rail systems. The expansion of Indian rail from 1905 to 1929 can be followed through a string of key moments. In 1905, an early Railway Board was constituted, serving under the Department of Commerce and Industry. This was partly a result of the railways finally becoming commercially viable; until the turn of the twentieth century, the railways were unable to pay interest charges on the capital that had been invested in their construction. The fact that they eventually turned a net profit was not just symbolically important but also a key reason to justify further invest- ment (Chandra 2006). The first electric loco- motive appeared in 1908. The milestone of reaching 30,000 miles of track occurred in 1910, just behind Germany and Russia,8 as can be seen in map 1.5. By 1920, the network had expanded to 38,040 miles (61,220 km). With the arrival of World War I, the rail- ways were used to meet the needs of the British outside India. Rail transported troops Source: Wikipedia.org. and grain to the ports of Bombay and Karachi en route to Britain, Mesopotamia, and East Africa. With shipments of equipment and railways was estimated at £687 million. parts from Britain curtailed, maintenance This decade of prosperity was brought to became much more difficult. Critical workers an abrupt end with the onset of the Great entered the army, workshops were converted Depression in 1929. For the next ten years, to making artillery, and some locomotives the railways suffered economically, and and cars were shipped to the Middle East. were later crippled by the onset of the The railways could barely keep up with the Second World War. increased demand. By the end of the war, the railways had deteriorated badly. In 1923, Railroads had limited impacts on overall both the Great Indian Peninsula Railway and productivity and sectoral composition East Indian Railway Company were nationalized. The construction of railways in India was The period between 1920 and 1929 was closely linked to the idea of improving an era of economic boom. Some 41,000 British control, boosting production—and miles (66,000 km) of railway lines served galvanizing economic and social change. As the country, carrying over 620 million pas- Lord Dalhousie, governor-general in the sengers and approximately 90 million tons mid–­ n ineteenth century put it, railways of goods each year. The capital value of the “would encourage enterprise, multiply 40  The WEB o f T r a n s p o r t C o r r i d o r s i n S o u t h A s i a MAP 1.5  Rail in India, 1909 Source: Wikipedia.org. production, facilitate the discovery of latent economy into context. Per capita income resources, increase national wealth and grew slowly during the late nineteenth cen- encourage progress in social improvement” tury, but stagnated during the first half of (Kerr 2007, 18). This rosy view was born the twentieth century, Gupta (2012) finds. from seeing the impact that railways had The Indian economy grew by about had in other parts of the world—including 1 percent per year from 1880 to 1920, in Britain itself. The railways, wrote while the population also grew by 1 percent Dalhousie, “will lead to similar progress in (Tomlinson 1996, 5). As a result, India did social improvement that has marked the not enjoy a long-term change in income lev- introduction of improved communication in els during this period. Agriculture was still various Kingdoms of the Western World” dominant, with most peasants farming at (Headrick 1988, 64) the subsistence level. Extensive irrigation Assessing to what extent these aspira- systems were built, providing an impetus tions were fulfilled is helped by the fact that for growing cash crops for export and for detailed historical national accounts were raw materials for Indian industry, especially compiled during the period of British Raj jute, cotton, sugarcane, coffee, and tea rule after 1871.9 This body of data can be (Tomlinson 1975). Agricultural income had used to put the impact of rail on the Indian the strongest effect on GDP. Agriculture I n s i g h t s i n t o R e g i o n a l I n t e g r a t i o n f r o m T h r e e H i s t o r i ca l T r a n s p o r t C o r r i d o r s i n S o u t h A s i a    41 grew by expanding the land frontier it is not difficult to understand India’s disap- between 1860 and 1914. The price of culti- pointing overall productivity performance vable land rose sharply after 1914 (Baten during this period. 2016, 255). The role of rail can then be factored in Using output and employment data from to these national economic patterns. historical national accounts, Broadberry Recently, Bogart and Chaudhary (2015) and Gupta (2016) calculate indexes of analyzed these national accounts and dis- Indian labor productivity by major sector. covered that Indian railways experienced They show that labor productivity grew high labor and capital productivity growth fastest in industry, as modern industry from the 1880s to the 1910s, but fueled developed in India, and slowest in services, lower productivity growth in many sectors, despite the modernization of the transport notably agriculture. A composite produc- network. During both the first and second tivity measure, known as total factor pro- halfs of the twentieth century, although ductivity (TFP), provides a key metric of labor productivity grew respectably in economic performance. Bogart and industry and services, labor productivity Chaudhary (2015) estimate the TFP for the growth in the economy as a whole was held period before 1913 and find a healthy back by slow productivity growth in growth rate of 2.3 percent per year. This is agriculture. in stark contrast to the rest of the Indian However, to fully understand the contri- economy, such as agriculture, which had bution of the three main sectors to compara- virtually no productivity growth in the late tive productivity performance, it is necessary nineteenth century. India’s TFP growth rate to track their shares in economic activity as in railways also exceeded that of Britain in well as their comparative productivity lev- the nineteenth century. Even more remark- els. Table 1.2 shows the changing sectoral ably, only a small part of the TFP growth distribution of employment in India from was due to capacity utilization, such as 1875 to 2000. running trains more frequently. It seems The most striking finding is the domi- that Indian railways successfully adopted nance of agriculture as an employer of labor. new technologies. For the century after 1870, agriculture’s Less is known about the causes of the pro- share of the labor force was around ductivity slowdown after World War I. The 75 percent. Even by the end of the twentieth collapse of world trade in the 1920s is prob- century, agriculture still accounted for nearly ably the most immediate explanation. The 65 percent of Indian employment. Given this productivity of railway tracks closely aligns commitment of resources to a sector with with demand, which dropped substantially inherently low value added, and the poor as international markets went through tur- productivity performance within that sector, bulent times. The sharp price convergence in British TABLE 1.2  Agriculture continued to dominate the Indian wheat and rice markets between economy, despite the huge investment in rail 1861 and 1920 has been attributed to the Indian labor force, by sector, 1875−2000 (percentage of introduction of rail. However, recent total employment) authors have countered this assumption. Years Agriculture Industry Services Andrabi and Kuehlwein (2008) carried out 1875 73.4 14.5 12.1 tests examining relative price differences 1910/11 75.5 10.3 14.2 between individual districts, and found that 1929/30 76.1 9.1 14.8 while the expansion of the rail network 1950/51 73.6 10.2 16.2 clearly mattered, other factors were more 1970/71 73.8 11.1 15.1 important. In fact, their analysis showed 1999/2000 64.2 13.9 21.9 that the extension of railway lines, the Source: Broadberry and Gupta 2010. opening up of new agricultural areas, and 42  The WEB o f T r a n s p o r t C o r r i d o r s i n S o u t h A s i a the capabilities to deliver produce to towns However, the analysis does not examine and cities were less significant than climate, the volatility of real incomes over time. As in types of soil, and indeed wider changes in much of the developing world today, colonial pricing patterns—locally, regionally, and India’s precarious monsoon rains and its rain- even internationally. fed agricultural technologies made real Prices were converging during this time income volatility extremely high. Famines almost as rapidly between those districts were a perennial concern. An important ques- without railways as those with railways. tion for future research concerns the extent to Andrabi and Kuehlwein’s fixed-effect estima- which transportation infrastructure systems, tion, using yearly dummies, points out that like India’s railroad network, can help regions the pattern of price convergence between a smooth the effects of local weather extremes given district pair was no different before they on local well-being. got railways than after. One reason may be Railways may have reduced the severity of that India was already a well-established, par- famines, as argued by McAlpin (1983) and tially integrated economy at the time railroads more recently by Burgess and Donaldson were being expanded. Some lines were built (2010). However, the ability of railways, as along existing trade routes where commerce infrastructure, to curb famine pales in between districts was already significant. The comparison to that of irrigation. There were impact of railroads on districts not obviously several reasons for this. For example, one linked by trade was about five times larger prominent critic of railway expansion noted than the impact on districts that were previ- that railways were used to justify higher land ously well linked, Andrabi and Kuehlwein payments, which meant that while the prices (2008) find. of grain and wheat rose, growers saw little— if any—of the benefits (Sweeney 2008, 148). Meanwhile, railways received a dispropor- The railways of the British Raj tionate share of the colonial budget. For brought mixed economic benefits example, the year-to-year allocation of expen- and questionable social benefits ditures within the Famine Insurance Fund to India varied between famine relief, protective This analysis highlights several issues sur- railways, protective irrigation works, and rounding the introduction of a rail corridor. programs to reduce debt.11 Irrigation never When a rail corridor is built along an accounted for a large share of the budget; ­ already-well-trodden trade route, the increased protective irrigation works averaged 5 percent capacity may not lead to marked intensifica- of the Famine Relief and Insurance Fund tion of trade along the route. Importantly, in between 1893/94 and 1903/04.12 terms of economic development and poverty Overall, while the government of India reduction, the largest benefactors of the intro- was active in developing railways, irrigation duction of rail may be those areas that had was shortchanged. Railway expenditures little or no previous trading ties. reached 23 percent of the overall colonial However, using a new panel of district- budget in 1894, declining to 15 percent in level data collected from archival sources, 1935. During this time, no reallocation of Donaldson (2010) finds stronger evidence that money was made to irrigation. Of the that railroads caused an increase in the level total expenditures on railways and irrigation of real incomes in India. Specifically, the works that were expected to cover the bor- study concludes that railroads reduced rowing costs (“productive works”), more the cost of trading, reduced interregional than 85 percent was devoted to railways price gaps, and increased trade volumes. between 1894 and 1919. When the railroad network was extended, Some argue that this disproportionate real agricultural income rose by approxi- colonial emphasis on railways was socially mately 16 percent.10 inefficient because of the strong link between I n s i g h t s i n t o R e g i o n a l I n t e g r a t i o n f r o m T h r e e H i s t o r i ca l T r a n s p o r t C o r r i d o r s i n S o u t h A s i a    43 irrigation and economic development. 13 marvelous contradiction. Although the According to this view, railways better served train made traveling a possibility for more the strategic interests of the Raj; the differen- people, it also accentuated social differ- tial expenditures were yet another example of ences because passengers were separated colonial exploitation. However, when consid- by the kinds of tickets they could afford to ering governmental investments in infrastruc- purchase, as well as the whims of railway ture in the twentieth century, it is worth officials. For instance, in 1891, toilet facili- remembering that the Raj was a conservative ties were introduced in first class, adding state; it did not tax much, due to political to the huge gap in “comforts and conve- concerns, and it did not spend much. The Raj niences” between first- and lower-class was not an active development state, either in passengers (Vaidyanathan 2002, 21). agriculture or elsewhere. Decades later, Mahatma Gandhi would However, the exploitative nature of the take up the lower-class passengers’ cause. East India Company and later colonial rule The social benefits the British investment is evident in the patterns of employment and in rails brought to the Indian population procurement during the expansion of rail in were questionable, at best. Economic bene- India. Until the 1930s, both the Raj lines fits were mixed. And the investments were and the private companies hired only accompanied by a partisan belief that European supervisors, civil engineers, and investment in rail would prove to be a more even operating personnel, such as locomo- efficient counter to famine than investment tive engineers (Headrick 1988, 8−82). The in irrigation. Nevertheless, the main bur- Stores Policy of the British Crown required den on the Indian economy during this that bids on railway contracts be made to period was low productivity in agriculture. the India Office in London, shutting out It may be unrealistic to expect railways most Indian firms. The railway companies to compensate for the poor performance of purchased most of their hardware and parts agriculture. in Britain. While railway maintenance work- shops operated in India, they were rarely THE THREE HISTORICAL CASES allowed to manufacture or repair locomo- SHARE COMMONALITIES WITH tives. The domestic steel supplier, TISCO ONE ANOTHER AND WITH Steel (Tata Iron and Steel Company), could MODERN TRANSPORT CORRIDORS not obtain orders for rails until the 1920s. Little was invested or reinvested in the local The three transport corridors that have been populations affected by rail. assessed are entirely different from one Setting aside the questionable economic another in terms of how and when they took benefits of the railroads to the British Raj, shape, who and what they affected, and other scholars debate the social and cul- what their consequences were over the short, tural advantages of the expansion of rail in medium, and long terms. India. For example, Marian Aguiar (2011) Nevertheless, they share some notable draws upon a rich archive of colonial cor- features. The acceleration of commercial respondence, travel narratives, and striking exchange had a direct and obvious impact on illustrations to unpack the colonial rheto- urban centers, socioeconomic change, and ric surrounding the railway space. She change to the physical landscape. The invest- finds that not only were railway lines seen ment in transport corridors—whether as a powerful technology with which to through physical expenditures on roads, exert control over the Indian colony, but bridges, and train lines; on administrative colonial rhetoric also imagined the shared measures that widen, improve, and expand traveling space of the railway as a leveler tax collection and central spending; or on of social differences between Indians. As rationalizing judicial proceedings and the res- the author rightfully points out, this was a olution of disputes—all served to further 44  The WEB o f T r a n s p o r t C o r r i d o r s i n S o u t h A s i a strengthen connections that had already collection and central spending; rationalizing began to take shape. judicial proceedings and resolution of dis- As each of the three corridors shows, the putes) all enabled economic connections that causes and consequences of the deepening of were already emerging. Nevertheless, the networks cannot simply be seen in a local or chapter notes that, at times, corridors were even a regional context. Indeed, even the ear- also conduits for unexpected problems and liest corridor under investigation was clearly troubles. For instance, in the 1730s, the linked to global trading patterns and to Persian ruler Nader Shah used the route link- rhythms of supply and demand hundreds, if ing Afghanistan and Central Asia with Delhi not thousands, of miles away. to strike at the very heart of the Mughal In the context of transport corridors in the Empire. twenty-first century, it is striking to note that Fast forwarding to more recent times, the mobility of the labor force has been a chapter 2 takes East Asia as one possible clear factor in the ability not only to carry out role model for South Asia and discusses the infrastructure projects but also in the expan- mixed successes with using transport corri- sion of cities and the development of inter- dors to spur regional development in connected economies. Vietnam and Malaysia. The first case study Although top-down investment was key describes the impacts of Vietnam’s National for immensely expensive infrastructure proj- Highway No. 5 (NH-5) on the development ects in the past—such as train lines and of private industrial zones and their signifi- roads—and will doubtless remain so in cant local spillovers. The second case study the future, important by-products included examines Malaysia’s experience with inte- the expansion of small-scale business, the gal- grating rural areas and industrialized trade vanization of intermediaries, and the growth hubs in the northern part of Peninsular of the middle class. As modern scholarship Malaysia. has clearly demonstrated, increasing levels of commercial exchange and economic activity NOTES have not only promoted regional integration 1. The Timurid Empire, a Sunni Muslim dynasty but have also reduced poverty. or clan of Turco-Mongol lineage, descended from the warlord Timur, Babur’s great-great- Summing up and moving to the great grandfather. It comprised modern-day next chapter Iran, the Caucasus, Mesopotamia, Afghanistan, and much of Central Asia, as well as parts of This chapter has mapped three past trans- contemporary Pakistan, Syria, and Turkey port corridors that spanned South Asia and (Marozzi 2004). has assessed their impacts in three eras: circa 2. Emperor Sher Shah Suri, a Pathan from 1700, 1800, and 1900. The three transport Bihar, founded the Suri Empire in northern corridors differed considerably in their emer- India. He was contemporaneous with the gence, life cycle, and impacts, but even the Mughals. earliest corridor was clearly linked to global 3. “Bengal” spans the regions that eventually trading patterns and to rhythms of supply became Bangladesh and three states in inde- and demand hundreds, if not thousands, of pendent India—West Bengal, Bihar, and Orissa. miles away. Not only has labor mobility 4. A successor state is a totally new state. It is determined the ability to execute infrastruc- distinct from the previous state but maintains the same legal personality and possesses ture projects, but it has also affected the all the existing rights and obligations of the dynamism with which cities have expanded ­previous state. and interconnected economies have devel- 5. Except for the tea trade and the trade with oped. The investments in transport corridors China. (roads, bridges, and train lines) and their 6. Kranton and Swamy (2008) build a model of complementary interventions (improving tax this procurement system and highlight the I n s i g h t s i n t o R e g i o n a l I n t e g r a t i o n f r o m T h r e e H i s t o r i ca l T r a n s p o r t C o r r i d o r s i n S o u t h A s i a    45 problems they observe throughout the his- in the Age of Enlightenment , edited by torical record. A. Johns. New York: Routledge. 7. In the twenty-first century, this route would Atwell, W. 1982. “International Bullion Flows and comprise part of the Golden Quadrilateral the Chinese Economy circa 1530−1650.” Past that connects Delhi (North), Chennai and Present 95 (1982): 68−90. (South), Kolkata (East), and Mumbai Bagchi, A. 1981. “Merchants and Colonialism.” (West)—a highway system examined in Occasional Paper 38, Centre for Studies in chapter 6. Social Sciences, Calcutta. 8. The United States, however, was far ahead in Barrett, W. 1990. “World Bullion Flows, terms of miles of laid rack. Railway density is 1450−1800.” In The Rise of Merchant perhaps more important than length. Density Empires: Long-Distance Trade in the Early was lower in India than in the United States Modern Worlds, 1350−1750, edited by J. and Europe, but higher than in South Tracy. Cambridge: Cambridge University Press. America, Africa, the rest of Asia, Canada, Baten, J. 2016. A History of the Global Economy Russia, and Australia (Prakash 1998, 165). from 1500 to the Present . Cambridge: 9. See, for example, Heston (1983) and Cambridge University Press. Sivasubramonian (2000). Bayly, C. 1983. Rulers, Townsmen and Bazaars. 10. While it is possible that railroads were delib- North Indian Society in the Age of British erately allocated to districts on the basis of Expansion, 1770−1870. Cambridge: time-varying characteristics unobservable to Cambridge University Press. researchers today, Donaldson (2010) finds ———. 1988. Indian Society and the Making of little evidence for this potential source of the British Empire ( The New Cambridge bias with respect to the results in three sepa- History of India ) . Cambridge: Cambridge rate placebo checks. University Press. 11. The Famine Insurance Fund was established Bogart, D., and L. Chaudhary. 2015. “Off the by the Government of India in the 1880s Rails: Is State Ownership Bad for Productivity?” after a devastating famine in Bengal, Orissa, Journal of Comparative Economics 43 (4): Madras, and Bombay (Chaturvedi, Kumar, 997–1013. and Singh 2005, 444). Bowen, H. V. 1989. “Investment and Empire in the 12. Calculations are based on Statistical Late Eighteenth Century: East India Abstracts, India, 1905. Stockholding.” Economic History Review 13. See, for example, Sweeney (2008). 42 (2): 186−206. Broadberry, S., and B. Gupta. 2010. “The Historical Roots of India’s Service-Led Development: A REFERENCES Sectoral Analysis of Anglo-Indian Productivity Aguiar, M. 2011. Tracking Modernity: India’s Differences, 1870–2000.” Explorations in Railway and the Culture of Mobility. Economic History 47 (3): 264−78. Minneapolis: University of Minnesota Press. ———. 2016. “Indian Economic Performance and http://www​.jstor.org/stable/10.5749/j.ctttsw7t. Living Standards: 1600−2000.” Chapter 2 in Alam, M. 1994. “Trade, State Policy and Regional A New Economic History of Colonial India, Change: Aspects of Mughal-Uzbek Commercial 15−32. New York: Routledge. Relations, c. 1550−1750.” Journal of the Brunner, H.-P. 2013. “What Is Economic Corridor Economic and Social History of the Orient Development and What Can It Achieve in 37 (3): 202−27. Asia’s Subregions?” Working Paper 117, Series Anderson, J., and E. van Wincoop. 2001. “Borders, on Regional Economic Integration, Asian Trade and Welfare.” In Brookings Trade Development Bank, Manila, Philippines. Forum, edited by S. Collins and D. Rodrik. Burgess, R., and D. Donaldson. 2010. “Can Washington, DC: Brookings Institution. Openness Mitigate the Effects of Weather Andrabi, T., and M. Kuehlwein. 2008. “Railways Shocks? Evidence from India’s Famine Era.” and Price Convergence in British India.” American Economic Review 100 (2): 449−53. Journal of Economic History 70 (2): 351−77. Chakrabarti, S. 1994. “Collaboration and Arnold, D. 1999. “Hunger in the Garden of Plenty: Resistance: Bengal Merchants and the English The Bengal Aminre of 1770.” In Dreadful East India Company, 1757−1833.” Studies in Visitations: Confronting Natural Catastrophe History 10 (1): 105−29. 46  The WEB o f T r a n s p o r t C o r r i d o r s i n S o u t h A s i a Chandra, B. 2006. “Economic Nationalism and ———. 2002. Mughal Warfare: Indian Frontiers the Railway Debate, circa 1880−1905.” In Our and High Roads to Empire, 1500−1700 . Indian Railway. Themes in India’s Railway London: Routledge. History, edited by R. Srinivasan, N. Tiwari, and Habib, I. 1969. “Potentialities of Capitalistic S. Silas. Delhi: Foundation Books. Development in the Economy of Mughal India.” Chaturvedi, H., D. Kumar, and R. Singh. 2005. Journal of Economic History 29 (1): 32–78. India Insurance Report, Series I. New Delhi: ———. 1982. An Atlas of the Mughal Empire. Allied Publishers Pvt. Ltd. Delhi: Oxford University Press. Chaudhury, S. 1995. From Prosperity to Decline: Headrick, D. R. 1988. The Tentacles of Progress: Eighteenth Century Bengal. Delhi: Manohar. Technology Transfer in the Age of Imperialism, Dale, S. 1994. Indian Merchants and Eurasian 1850–1940 . New York: Oxford University Trade, 1600−1750. Cambridge: Cambridge Press. University Press. Heathcote, T. 1995. The Military in British India: Daryaee, T., and K. Aghaie. 2012. The Afghan The Development of British Land Forces in Interlude and the Zand and Afshar Dynasties South Asia. Manchester: Manchester University (1722–95). Oxford: Oxford University Press. Press. Das Gupta, A. 1967. Malabar in Asian Trade. Heston, A. W. 1983. “National Income.” Cambridge: Cambridge University Press. In  The Cambridge Economic History of Dasgupta, B. 2000. “Trade in Pre-Colonial India, Volume II, c. 1757−c. 1970, edited by Bengal.” Social Scientist 28 (May–June): 47−76. M. Kumar and M. Desai, 463−532. Cambridge: De, P. 2007. “Impact of Trade Costs on Trade: Cambridge University Press. Empirical Evidence from Asian Countries.” Jardine, L., and J. Brotton. 2000. Global Interests: Working Paper 2707, Asia-Pacific Research and Renaissance Art between East and West . Training Network on Trade. London: Reaktion. Deloche, J. 1994. Land Transport . Vol. 1 in Jha, M. 2005. “The Mughals, Merchants and the Transport and Communications in India Prior European Companies in 17th-Century Surat.” to Steam Locomotion . Oxford: Oxford Asia Europe Journal 3 (2): 269−83. University Press. Kerr, I. 2007. Engines of Change: The Railways Donaldson, D. 2010. “Railroads of the Raj: That Made India . Westport, CT: Praeger Estimating the Impact of Transportation Publishers. Infrastructure.” Working Paper 41, Asia Kolff, D. 1991. Naukar, Rajput and Sepoy: The Research Centre, Murdoch, Australia. Ethno-History of the Military Labour Market Dyson, K. K. 2002 (reprint of 1978 work). in Hindustan, 1450–1850 . Cambridge: A Various Universe: A Study of the Journals Cambridge University Press. and Memoirs of British Men and Women in the Kranton, R., and A. Swamy. 2008. “Contracts, Indian Subcontinent, 1765−1856. New Delhi: Hold-Up, and Exports: Textiles and Opium in Oxford University Press. Colonial India.” American Economic Review Flynn, D., and A. Giráldez. 1995. “Born with 98 (3): 967−89. a ‘Silver Spoon’: The Origin of World Trade Kumar, N. 2015. “Potential and Prospects of in 1571.” Journal of World History 6 (2): Strengthening Transport Connectivity for 201−21. Regional Economic Integration in Southern Foltz, R. 2007. “Cultural Contacts between Asia.” South Asia Economic Journal Central Asia and Mughal India.” In India and 16 (2-Supplement): 39S−54S. Central Asia, edited by S. Levi. New Delhi: Laroque, B. 2004. “Trade, State, and Religion in Oxford University Press. Early Modern India: Devotionalism and the Fraas, M. 2014. “Making Claims: Indian Litigants Market Economy in the Mughal Empire.” PhD and the Expansion of the English Legal World thesis, University of Wisconsin at Madison. in the Eighteenth Century.” Journal of Lawson, P. 1993. The East India Company: Colonialism and Colonial History 15 (1). doi: A History. London: Routledge. 10.1353/cch.2014.0006. Lucassen, J. 2006. “Brickmakers in Western Gommans, J. 1991. “Mughal India and Central Europe (1700−1900) and Northern India Asia in the Eighteenth Century: An Introduction (1800−2000).” In Global Labour History. to a Wider Perspective.” Itinerario 15 (1): A State of the Art, edited by J. Lucassen. Bern: 51−70. Peter Lang. I n s i g h t s i n t o R e g i o n a l I n t e g r a t i o n f r o m T h r e e H i s t o r i ca l T r a n s p o r t C o r r i d o r s i n S o u t h A s i a    47 Marozzi, J. 2004. Tamerlane: Sword of Islam, Singh, C. 1991. Region and Empire: Punjab in the Conqueror of the World. London: Da Capo Seventeenth Century. Delhi: Oxford University Press. Press. Marshall, P. 1987. “Private British Trade in the Sivasubramonian, S. 2000. The National Income Indian Ocean before 1800.” In India and the of India in the Twentieth Century . Delhi: Indian Ocean, 1500−1800, edited by A. Das Oxford University Press. Gupta and M. Pearson. Calcutta: Oxford Sweeney, S. 2008. “Indian Railways and Famine University Press. 1875−1914: Magic Wheels and Empty McAlpin, M. 1983. Subject to Famine: Food Crises Stomachs.” Essays in Economic and Business and Economic Change in Western India, History 26 (2008): 147−58. 1860−1920. Princeton, NJ: Princeton University Tomlinson, B. R. 1975. “India and the British Press. Empire, 1880–1935.” Indian Economic and Mehta, J. 1979. Advanced Study in the History of Social History Review 12 (4): 337–80. Medieval India. Delhi: Sterling. ———. 1996. The Economy of Modern India, Moosvi, S. 1990. “Mughal Shipping at Surat in the 1860–1970. Cambridge: Cambridge University First Half of the Seventeenth Century.” Press. Proceedings of the Indian History Congress Travers, T. 2004. “‘The Real Value of the Lands: 51 (1990): 308−20. The Nawab, the British and the Land Tax in Mukerjee, R. 1968. The Economic History of Eighteenth-Century Bengal.” Modern Asian India. London: Longmore. Studies 38 (3): 517−58. Naqvi, H. 1968. Urban Centres and Industries ———. 2005. “Ideology and British Expansion in in Upper India, 1556−1803 . London: Asia Bengal, 1757−72.” Journal of Imperial and Publishing House. Commonwealth History 33 (1): 7−27. Prakash, O. 1998. “European Commercial Vaidyanathan, K. 2002. 150 Glorious Years of Enterprise in Pre-Colonial India.” Chapter 4 in Indian Railways . Mumbai: English Edition Volume II.5 of The New Cambridge History of Publishers and Distributors. India. Cambridge: Cambridge University Press. Van der Woude, A. 1991. “The Volume and ———. 2007. “From Negotiation to Coercion: Value of Paintings in Holland at the Time of Textile Manufacturing in India in the Eighteenth the Dutch Republic.” In Art in History. Century.” Modern Asian Studies 41 (6): History in Art: Studies in Seventeenth 1331−68. Century Dutch Culture, edited by J. de Vried Raj, K. 2011. “The Historical Anatomy of a and D. Freedberg. Santa Monica, CA: Contact Zone.” Indian Economic and Social Getty Center for the History of Art and the History Review 48 (1): 55−82. Humanities. Ray, I. 2011. Bengal Industries and the British Washbrook, D. 1993. “Land and Labour in Industrial Revolution (1757−1857). London: Late Eighteenth-Century South India: The Routledge. Golden Age of the Pariah?” In Dalit Roy, T. 2012. The East India Company: The Movements and the Meanings of Labour in World’s Most Powerful Corporation. Delhi: India , edited by P. Robb. Delhi: Oxford Allen Lane. University Press. Sanyal, N. 1930. “The Development of Indian ———. 2007. “India in the Early Modern World Railways.” Calcutta: University of Calcutta. Economy: Modes of Production, Reproduction https://archive.org/details/Development​ and Exchange.” Journal of Global History OfIndianRailways. 2 (1): 87−111. Sengupta, K. 2015. “Bazaars, Landlords and Westcoat, J., M. Brand, and N. Mir. 1993. “The the Company Government in Late Eighteenth- Shedara Gardens of Lahore: Site Documentation Century Calcutta.” Indian Economic and Social and Spatial Analysis.” Pakistan Archaeology History Review 52 (2): 121−46. 25: 333–66. 2 Insights into Regional Integration from Two Contemporary Transport Corridors in East Asia S hifting from the eras of South Asia’s historical corridors to recent times, this chapter discusses the somewhat mixed suc- could build upon in forming local supply chains. The ultimate ambition was to take advantage of the global flows of FDI and, cesses that large transport infrastructure over time, introduce local companies to and investment in Vietnam and Malaysia have integrate them in global trade and global sup- had in boosting regional integration and ply chains. Much of that ambition has been generating wider economic benefits. achieved—but not without trade-offs in terms The Vietnam case study focuses on the negative impacts to the environment and of ­ developments around National Highway safety, among other areas. Some of these No. 5 and the anchor infrastructure invest- trade-offs occurred immediately, and others ment that aimed to connect the capital city of have been more gradual. Hanoi with the port of Hai Phong, an inter- The Malaysia case study reviews efforts by national gateway for exports. The anchoring the national government to integrate lagging transport investment followed earlier pro- regions with well-industrialized hubs and business reforms and the policy of increasing boost development that is more spatially openness to trade. These policy interventions (geographically) equitable. Because of were nationwide and not geographically spe- the large amounts of investment and number cific to the envisaged corridor. The investment of projects, Malaysia opted for a top- aimed to substantially upgrade one of the down approach by establishing a dedicated national highway arteries and—along with coordinating institution, along with an invest- the probusiness environment, educated labor ment fund to back up its recommendations force, and better connectivity to the world— with adequate funding. Although it is too attract foreign direct investment (FDI). early to assess the impact of the Malaysia’s Such an infusion of f ­oreign investment was initiative, this case study points to some inter- sought to create geographic hot spots around esting challenges. While the public sector can National Highway No. 5 that local business help resolve the failure of private sector 49 50  The WEB o f T r a n s p o r t C o r r i d o r s i n S o u t h A s i a stakeholders to coordinate their investments VIETNAM’S NATIONAL HIGHWAY and to build needed large-scale infrastructure NO. 5, INDUSTRIAL ANCHORS, themselves, it can face coordination chal- AND LOCAL SPILLOVERS lenges of its own. Notably, the challenges National Highway No. 5 (NH-5), a could arise because coordination between the 106–­kilometer (km)–long national road con- central and the local governments is needed in necting Hanoi with the international port of the case of investments that are geographi- Hai Phong, is an important socioeconomic cally specific, and the existing institutional artery for the northern part of Vietnam frameworks may not always facilitate such (map 2.1). It was upgraded in the mid-1990s, coordination effectively. when traffic for both freight and passengers was expected to increase ­dramatically, along MAP 2.1  Vietnam’s National Highway No. 5 connects Hanoi to the with Vietnam’s rapid ­economic growth stem- port of Hai Phong ming from its open reform policies (figure 2.1). The corridor intervention was backed by complementary policies Developing NH-5 was part of the effort to improve the overall transport network. Supported by Japanese official development assistance (ODA) in the mid-1990s, the devel- opment aimed to strengthen trade and indus- try, as well as improve living standards in the northern part of Vietnam. For that to happen, the transport capacity needed to be expanded to become much more efficient. To develop NH-5, Japan extended ODA loans totaling $187 million in three phases over 10 years (see table 2A.1 in annex 2A). As a result of the project, the transport time between Hanoi and Hai Phong has been cut from 5 hours to 2 hours, while the average speed of the vehicles has almost doubled, from 24−30 km per hour to 50−60 km per hour, according to Ministry of Transport officials and field interviews conducted with highway users. Traffic has become much smoother, except in a few bottlenecks, such as central Hanoi and the port area in Hai Phong, ex post Source: JICA. evaluations show. Widening roads, separating Note: Official World Bank country names are used. regular and light vehicles, upgrading intersec- tions with other trunk roads, and making traf- fic management more efficient resulted in vast Greater transport capacity attracted significant improvements compared to the prior chaos, when pedestrians and varying types of vehi- foreign direct investment. The presence of global cles jostled in only one or two lanes. brands—including Canon, Honda, and Panasonic— In addition to improving NH-5 as the core has accelerated industrial concentration and physical infrastructure, institutional reforms economic growth in northeastern Vietnam. complemented this trunk road investment I n s i g h t s i n t o R e g i o n a l I n t e g r a t i o n f r o m T w o C o n t e m p o r a r y T r a n s p o r t C o r r i d o r s i n Ea s t A s i a   51 FIGURE 2.1  In the early 1990s, Vietnam needed a set of complementary efforts to promote industrialization and exports, improve transport, and upgrade labor Initial Conditions Necessary Modifications Measures Shift from • Division of labor in production socialist to • Physical infrastructure around the market • Fairly operated markets Export-oriented market economy industrialization Industrialization for growth through labor shift, calling for: Transport • Promotion of labor-intensive industries network Shift from (to absorb urban population) labor-intensive and • Improvements in rural education agro-based state (to provide better-quality labor) Education to industrialized • Improvements in agricultural productivity to: state Release labor from farming Expand markets with a greater variety of products Increase income to be invested in industry development, etc. Unbalanced Growth Strategy • Southeast Area Highly elastic demand to new products (centered on Ho Chi Minh City) FDI-friendly environment Export potential • North Triangle Potential for industry concentration/spillover (including Hanoi and Hai Phong) Good access to high-quality labor Source: JICA. Note: FDI = foreign direct investment. FIGURE 2.2  Foreign direct investment around Hai Phong Harbor has soared 800 600 $ million 400 200 0 2000 2001 2002 2003 2004 2005 2006 Hung Yen Hai Duong Hai Phong Source: JICA from Provincial Statistical Yearbook. and played an important role in Vietnam’s making it easier to collaborate with foreign industrial development. Prime examples partners. include the Amended Law on Foreign The project attracted considerable Investment (2000), the U.S.-Vietnam Bilateral FDI, which has grown almost exponentially Trade Agreement (2001), and the Company (figure 2.2). The region was blessed with a Law (2002) to ensure a smoother flow of good-quality but relatively low-wage work- goods and capital, as well as underpinning force and was already exposed to global com- the legal status of private sector companies, petition, having already attracted FDI. 52  The WEB o f T r a n s p o r t C o r r i d o r s i n S o u t h A s i a Meanwhile, the national consensus to reduce One of the central features of Vietnam’s poverty was building. These initial conditions industrial development in Vietnam was a were strengthened by later interventions, such model in which “anchor tenants” were as industrial zones, incentives to attract FDI, invited to locate in the industrial parks. To and vocational training. establish local supply chains, these major The road linked four existing industrial firms have been encouraged to invite “kei- parks (one in Hanoi and three in Hai Phong). retsu” companies, affiliates, and local ven- In addition, eight new parks have been devel- dors/suppliers, linked by cross-shareholdings oped (three in Hung Yen Province and five in to form a robust corporate structure. This Hai Duong Province). Several industrial clus- model, named after the company that initi- ters have emerged along NH-5, centered on ated it (Canon, the Japanese manufacturer of the garment, food processing, machinery, and imaging and optical products) is called the electronics industries. Japanese companies Canon Effect (figure 2.3). Vietnam has been contributed substantially to the development successful in creating this kind of virtuous of industrial parks and clusters. By 2016, 652 cycle. The presence of global brands—­ Japanese companies were operating in Hanoi, including not only Canon, but also Honda Hung Yen, Hai Duong, and Hai Phong. and Panasonic—has accelerated industrial concentration and economic growth in the northeastern part of the country. FIGURE 2.3  Vietnam’s growth model centered on “anchor tenants” Not only core cities such as Hanoi and Hai and associated firms jointly developing local supply chains Phong expanded, but so did production sites along the corridor (map 2.2). The surround- ing provinces (Hung Yen, Hai Duong, and ˘ Long Industrial Park Thang Hai Phong) and the port city have developed Initial Nomura Hai Phong Industrial Park Conditions industrial parks hosting numerous foreign firms. New urban areas were also created, [Panasonic ODA] NH-5, Hai Phong and local roads were built in Hai Phong near Port Institutional development NH-5 to connect to villages that had been (streamlining administrative works, tax incentives) established before the highway was con- Inviting anchor structed (JICA 2008). = Canon, Honda, Panasonic tenants While the development of NH-5 and sur- rounding areas has brought about outstanding Canon ripple effects economically as reviewed above, Effect there still remains a view, especially from the Developing local Keiretsu companies, affiliates, viewpoint of foreign companies thinking supply chains local vendors/suppliers about expanding local operations, that the supporting industries are not well developed, with local companies not fully capable of sup- plying quality products. In 2003, Vietnam and Industry Economic Japan started a joint effort (“Vietnam-Japan development in concentration Joint Initiative”) with the aim of improving northern Vietnam the investment climate, in which various mea- sures for fostering supporting industries are Source: JICA. examined and put into practice. Note: ODA = official development assistance. On the Vietnamese side, the government has introduced a policy package for support- ing industry development (Decree 12, The areas surrounding the highway underwent a February 2012), while Japanese side has been rapid and vast structural change, from the agricultural supporting the policy formation, human to the industrial sector. resource development, SME finance, etc. I n s i g h t s i n t o R e g i o n a l I n t e g r a t i o n f r o m T w o C o n t e m p o r a r y T r a n s p o r t C o r r i d o r s i n Ea s t A s i a   53 MAP 2.2  Industrial parks have flourished along NH-5 Source: JICA from Forval Corporation, Overseas Expansion Advisors. Through these efforts, the share of local soared by 221 percent, while in the other procurement by Japanese companies has seen nodal city, Hai Phong, the number of firms a drastic increase from 22.4 percent in 2010 increased by 141 percent. The connected to 34.2 percent in 2016. On the other hand, provinces of Hung Yen and Hai Duong also the share of local suppliers in local procure- grew substantially (figure 2.4). The main rea- ment (41.1 percent) is still lower than those in son that many of the firms chose to locate in India (73.4 percent), China (59.4 percent), all three provinces was good accessibility via and Indonesia (47.2 percent). (JETRO 2016). NH-5, according to the beneficiary survey From this aspect as well, further development (JICA 2009). Hanoi became a center of grav- of local players is expected to extract WEB. ity for firms, accounting for 16 percent of the total number of firms nationwide in 2004— up from 11 percent in 2000. While the expan- The area near the highway underwent sion in the number of firms was impressive far-reaching economic and structural for Hai Phong and the other two provinces, it change matched the growth in the rest of the country With better connectivity, the number of enter- along major highways. Thus, these areas did prises has surged in the two nodal cities and not move up the ranks for the share of local the provinces along NH-5 (figure 2.4). From firms in all firms in Vietnam. 2000 to 2004, after the project was com- The nodal cities and connected provinces pleted, the number of firms in these areas all enjoyed double-digit economic growth grew faster than the national average from 2003 to 2006—much higher than the (117 percent). The biggest growth occurred in national average of 7.9 percent. The fastest Hanoi, where the number of enterprises growth in production occurred in the 54  The WEB o f T r a n s p o r t C o r r i d o r s i n S o u t h A s i a industrial sector (figure 2.5), ranging from agricultural and livestock production declined 13.3 percent in Hai Duong to 20.5 percent in from 65 percent to 46 percent of total GDP Hung Yen. The service sector also grew by between 1997 and 2002. double digits during this three-year period. In contrast, the agriculture sector declined in Transport impacts Hanoi and grew by only single digits in other In line with the structural change and produc- areas around NH-5. Overall, the areas sur- tion boom, traffic volume increased during rounding the highway underwent a vast the same period. Freight volume almost structural change, from the agricultural to the ­ tripled between 1994 and 2005, while passen- ­ industrial sector, from 2003 to 2006. ger traffic increased from 400 million to This structural change affected the produc- 1.1 billion (figure 2.7, panel a). By 2006, pas- tion structure of the entire country. The shares senger car units per day for NH-5 for all four of “commerce” and “factory” in total major types of vehicles—cars, vans, buses, national production increased more than and trucks—outpaced the targeted value for other sectors (figure 2.6). In contrast, 2008 (figure 2.7, panel b). Wider economic impacts FIGURE 2.4  The number of enterprises surged along NH-5 from The accelerating growth of industry in the 2000 to 2004, especially in Hanoi areas along NH-5 has been reflected in the growth of overall income. The wider economic a. Percentage growth in the number of enterprises, by location impacts of the development of NH-5 have been 250 very beneficial, particularly for rural farmers. 200 The improved transport corridor provided farmers with bigger and more lucrative local 150 markets (both larger existing markets and Percent higher value-added m ­ arkets), thanks to income 100 and population growth in the area. It also pro- vided opportunities for nonfarming businesses, 50 thanks to easier access to the major cities. This 0 transformation has been further stimulated by National Hanoi Hung Yen Hai Duong Hai Phong complementary interventions, such as the b. Change in the share of total number of enterprises in the country development of a feeder road network and pro- 6 vision of public transport services. The newly opened factories along NH-5 5 drew a large-scale labor force from the local Percentage change 4 farmer population and from migrants flowing into the cities along NH-5. In 2006 alone, 3 83,453 workers and 134,846 workers, respec- 2 tively, were employed in Hung Yen (in the man- ufacturing sector) and Hai Duong provinces, 1 accounting for 14 percent and 19 percent of the 0 total working populations of these provinces. Hanoi Hung Yen Hai Duong Hai Phong This efficient shift in the labor force was made possible by the growing population of Source: JICA. young people and the high quality of educa- tion in Vietnam. Education is mandatory The highway improvements increased production through the lower secondary level. The com- and trade, generated factory jobs, expanded pletion rate for lower secondary studies for supporting businesses, and activated the the whole country was 96 percent as of 2004. procurement of local inputs. Furthermore, the upper secondary completion I n s i g h t s i n t o R e g i o n a l I n t e g r a t i o n f r o m T w o C o n t e m p o r a r y T r a n s p o r t C o r r i d o r s i n Ea s t A s i a   55 FIGURE 2.5  Production along NH-5 grew significantly more than the national average from 2003 to 2006 a. Total production b. Industrial sector 13 23 12 21 11 19 17 Percent Percent 10 15 9 13 8 11 7 9 6 7 Hanoi Hung Yen Hai Duong Hai Phong Hanoi Hung Yen Hai Duong Hai Phong c. Agriculture sector d. Service sector 5 17 4 15 3 13 Percent Percent 2 11 1 9 0 7 –1 5 Hanoi Hung Yen Hai Duong Hai Phong Hanoi Hung Yen Hai Duong Hai Phong Local National Source: JICA. Note: The growth of production is measured by the growth of the respective city, province, and sectoral GDP. “Growth” is annualized real GDP growth from 2003 to 2006. NH-5 = National Highway No. 5. FIGURE 2.6  The Vietnamese economy underwent a vast structural transformation from 1997 to 2002 45 40 35 30 Percentage of GDP 25 20 15 10 5 0 Commerce Factory Handicraft Real estate Remittance Agriculture Livestock Fishery Others Industrial Agricultural 1997 2002 Source: JICA. 56  The WEB o f T r a n s p o r t C o r r i d o r s i n S o u t h A s i a FIGURE 2.7  Traffic volume has soared in Vietnam on NH-5 a. Vietnam b. NH-5 1,200 250 25,000 Target for 2008 1,000 200 20,000 Million passengers Million tons 800 PCU/day 150 15,000 600 100 10,000 400 200 50 5,000 0 0 0 Baseline Actual 94 00 01 02 03 04 05 1993 2006 19 20 20 20 20 20 20 Passenger Cargo Truck Bus Van Car Source: JICA. Source: JICA.  Note: NH-5 = National Highway No. 5; PCU/day = passenger car units per day. FIGURE 2.8  The NH-5 upgrade generated significantly wider economic benefits for income and poverty a. Annual percentage growth of GDP per capita, b. Percentage change in the number of households living in 1995−2000 (interquartile range) poverty, 1998−2000 (interquartile range) 14 70 12 60 10 50 8 40 6 30 4 20 2 10 0 0 A B The rest A B The rest Source: JICA. Note: In panel a, the maximum and minimum numbers are the reported numbers. The quartile numbers were calculated using the Excel command “quartile. inc.” Area “A” (near NH-5) includes Hanoi, Hai Phong, Hai Duong, and Hung Yen. Area “B” (further from NH-5) includes Bac Ninh, Vinh Phuc, and Quang Ninh. “The rest” includes Ha Tinh, Ha Nam, Nam Dinh, Thai Binh, Ninh Binh, Thai Nguyen, Phu Tho, Bac Giang. GDP = gross domestic product; NH5 = National Highway No. 5. rate for the four provinces along NH-5 was International Student Assessment (PISA), a about 97−100 percent, which is much higher worldwide assessment of educational attain- than the country average (90.5 percent) as of ment conducted by the Organisation for 2004.1 Tertiary education started to be rolled Economic Co-operation and Development out in Vietnam in the 1990s. The gross enroll- (OECD). In its first year of testing, Vietnam’s ment rate grew from 2.3 percent in 1990 to PISA ranking exceeded the OECD average in 9.4 percent in 2000 to 18.4 percent in 2007.2 all three categories (reading, math literacy, In 2012, Vietnam joined the Programme for and scientific literacy). 3 High-quality I n s i g h t s i n t o R e g i o n a l I n t e g r a t i o n f r o m T w o C o n t e m p o r a r y T r a n s p o r t C o r r i d o r s i n Ea s t A s i a   57 education has increased the quality of labor, FIGURE 2.9  Traffic accidents have soared on NH-5, greatly and allowed Vietnam to take full advantage outpacing the national trend as well as those of other national highways of the corridor as a driver of growth. Among the provinces in the Red River a. Number of traffic accidents in Vietnam Delta region,4 those along NH-5 have ranked 30,000 higher in the pace of income increase and pov- erty reduction (figure 2.8). From 1995 to 25,000 2000, GDP per capita grew by 6.1 percent in the Red River Delta region, compared with 20,000 the national average of 5.7 percent. The num- 15,000 ber of households living in poverty dropped by an impressive 35 percent. The cities closer 10,000 and further away from NH-5 (marked “A” and “B” in figure 2.8) both experienced higher 5,000 income growth per capita as well as faster 0 reduction in the poverty than “the rest” of the 1994 2000 2001 2002 2003 2004 2005 country. The country as a whole lowered its poverty rate by 27 percent during this period, b. Number of traffic accidents per kilometer of road thanks to broader positive spillovers from 4.5 NH-5 to other regions. 4.0 The wider impacts also included unintended 3.5 negative consequences. Traffic accidents 3.0 soared. The highway has become more danger- 2.5 ous for light vehicles and pedestrians as wider 2.0 1.5 roads have allowed the highway to carry larger 1.0 and heavier motor vehicles, as well as a greater 0.5 volume of traffic traveling at higher speeds. 0 The lack of feeder roads has increased traffic 1999 2003 2004 2005 2005 jams around the industrial parks. And the highway has generated negative environmental NH-3 NH-5 NH-10 NH-18 impacts such as dust, noise, and vibrations. Source: JBIC 2003 and National Traffic Safety Committee (NTSC). In the 2000−06 period, traffic accidents Note: NH-3 = National Highway No. 3; NH-5 = National Highway No. 5; NH-10 = National Highway were more frequent on NH-5 than other No. 10; NH-18 = National Highway No. 18. national highways (figure 2.9). Excessive speed accounted for nearly all (93 percent) of the traffic accidents along NH-5, according to promote traffic safety, including the Study on the Ministry of Transport’s Vietnam Road National Road Traffic Safety Master Plan Administration (VRA). This share is substan- (2008, JICA), the Northern Vietnam National tially higher than the average of 60−70 Roads Safety Improvement Project (2008−13, percent for other highways. This high per- JICA), the Traffic Education Center (1999− centage of accidents can partly be attributed present, Honda), and the Safety Driving to the fact that residents have been slow to Program (2014−­ present, Toyota). adapt to the huge increase in traffic volumes As for environmental issues, NH-5 was that accompanied the improvements in NH-5. developed before Vietnam’s Environmental In recent years, the National Traffic Safety Impact Assessment (EIA) system was imple- Committee (NTSC) has spearheaded various mented. Thus, at the time the highway was policy measures, including safety campaigns upgraded, the feasibility study did not go and legislation requiring all motorcycle riders through the EIA process. Environmental con- to wear helmets. Bilateral donors and private cerns were raised only later. To deal with sector companies have supported activities to these issues, policy measures (such as planting 58  The WEB o f T r a n s p o r t C o r r i d o r s i n S o u t h A s i a trees, and locating new schools and hospitals interactions between specific policy measures away from NH-5) have been taken. The nega- and expected outcomes. tive environmental impacts have been gradu- Initial conditions, including geographical ally mitigated, although not totally removed. advantages and constraints (such as distance from the major cities and ports, landlocked regions, surrounding countries, and markets) Lessons learned and demography (size, age, and the gender For a corridor intervention to be designed and make-up of the labor force and population) implemented in a holistic fashion, policy mak- must inform the choice of the geographical ers must think through the potential chain location for corridors. (sequencing) of results originating from a par- Merely providing transport infrastructure initial ticular design. They need to appraise the ­ for the corridor area does not automatically conditions, suitable policy measures, various trigger the wider economic impacts. For posi- expected outcomes, potential unintended con- tive spillovers to materialize from industrial sequences, and additional measures needed to concentration, economic and structural address them (figure 2.10). Further, the design change must be supported by complementary can more effectively achieve its primary policy measures—such as steps to provide an goals—such as encouraging the clustering of appropriate incentive structure, a legal frame- industries—when policy makers duly consider work to reduce operational costs for the FIGURE 2.10  The NH-5 Corridor generated many expected and unexpected wider economic impacts Initial Conditions • Geography • Demography Labor quality Policy Measures (HR facilitation) (Planning) Improving education / public health Identifying the site (Physical facilitation) (Technical transfer and • Supply of infrastructure ↑ encouraging • Better utilities Maintenance works innovations) • Legal framework (Institutional Rural (Dealing with Improving • Incentives facilitation) Remedies productivity empowerment side effects) Expected Outcomes Entrepreneurship ↑ Factory Household Local agro- Consumption ↑ ↑ jobs ↑ income ↑ production ↑ Transport cost ↑ Industry Operational cost concentration Industry output ↑ FDI ↑ Technological progress Farm jobs along corridor ↑ Side Effects Population shift Jobs in non- Environmental corridor areas impacts Source: JICA. Note: NH-5 = National Highway No. 5. I n s i g h t s i n t o R e g i o n a l I n t e g r a t i o n f r o m T w o C o n t e m p o r a r y T r a n s p o r t C o r r i d o r s i n Ea s t A s i a   59 companies, high-quality education to supply create corridors of broader economic activity enough entrepreneurs and factory workers, spurred by transport connectivity. and ensuring the quality of public health to Malaysia is widely considered a develop- maintain the labor supply for factories. ment success story. This multiethnic nation Policy makers must also consider some of has achieved rapid growth while reducing the spillover effects from the clustering of poverty and improving equality through affir- core industries on nonindustrial sectors— mative action policies. Since achieving its inde- such as the potential increase in local income pendence in 1957, Malaysia has leapfrogged that can help support local production. As from low-income to upper-middle-income sta- local production rises, it can further boost the tus. Economic growth has been accompanied income of local suppliers and continue to by rising living standards and improvements strengthen local demand in a virtuous cycle. in the distribution of income, ameliorating the In the case of NH-5, the increasing consump- twin problems of poverty and racial imbal- tion of higher-value-added commercial crops ances (Snodgrass 1980; Athukorala and in urbanized areas motivated farmers to Menon 1999; Lucas and Verry 2001; Faaland, change their production portfolio from tradi- Parkinson, and Saniman 2003; Lim 2011). tional to high-value crops. However, since about the late 1990s, During the process of industrialization Malaysian policy makers have been concerned along NH-5, the farming population and the about the widening of income disparities amount of farm land have decreased. But in among urban and rural areas and among parallel, agricultural production and thus ethnic groups. Had rapid growth been accom- ­ productivity have increased to a sustainable panied by a more equitable income distribu- level. This increased productivity has been tion, domestic demand would have had a far associated with booming sales of farm prod- more important role in fueling growth and ucts to the food processing industry, as well as would have reduced the economy’s reliance on in exports by land and sea (see chapter 6 for a exports as the engine of growth (Ariff 2012). detailed analysis of such economic restructur- These concerns have propelled Malaysian pol- ing in India). icy makers to focus on policies to redress the The original rationale for NH-5 to connect widening divide between rural and urban Hanoi as a regional economic center with the areas and different states in the country. sea gateway delivered more than expected. As The idea of corridor development as a industry concentration accelerated along the vehicle for achieving balanced growth was corridor, there were robust spillovers to rural first mooted in the Ninth Malaysia Plan, areas surrounding it—mostly through greater 2006−10, launched in 2006. It proposed local demand stimulated by rising local income. developing “new regional growth centres The inflow of FDI played an important role in (RGCs) and growth corridors …. together by establishing an “ecosystem” in the regional the Government and the private sector to pro- economy, not only producing local spillovers mote investment, create jobs and encourage but also integrating the NH-5 corridor into rapid development in the areas concerned” greater regional and global supply chains. (Government of Malaysia 2006, 28). But in the Mid-Term Review of the Ninth Plan (Government of Malaysia 2008), five corri- MALAYSIA’S EXPERIENCE WITH dors were identified, each with an established INTEGRATING RURAL AREAS AND or growing city as its core: INDUSTRIALIZED TRADE HUBS As part of its national development strategy 1. The Northern Corridor Economic Region to redress regional economic disparities and (NCER) in northern Peninsular Malaysia, narrow the divide between rural and urban with Georgetown as its center. areas, the Malaysian government is in the 2. Iskandar Malaysia (IM) in the south, process of implementing five programs to with Johor Bahru as its center. 60  The WEB o f T r a n s p o r t C o r r i d o r s i n S o u t h A s i a 3. East Coast Economic Region (ECER) on distribute prosperity more equitably among the east coast of the peninsula, with Kuan- the four states. The proximity of the region to tan as its center. Thailand and access to the Strait of Malacca 4. Sarawak Corridor of Renewable Energy are viewed as added advantages of the NCER (SCORE), with Kuching as its center. in a regional context. The four states are 5. Sabah Development Corridor (SDC), with within the Indonesia-Malaysia-Thailand Kota Kinabalu as its center. Growth Triangle (IMT-GT) (ADB 2012).5 Several road projects have been proposed The proposed five corridors cover almost to connect northern Malaysia and southern 70 percent of the country’s landmass. Thailand. The Songkhla–Penang–Medan The rationale for the model of interstate corridor is one of the five corridors identified ­ corridors with interlinked economic activities is under the IMT-GT implementation blueprint the belief that states could forgo the benefits of (2012−16). It aims to rejuvenate the long- joint and complementary growth if they pursue standing trade routes centered on Penang that their individual interests, concentrating on their are important for bulk shipping of goods own growth. By providing connections, both within the region and for providing outward physical and economic, to potential or growing shipping services to East Asia and Europe. economic nodes or hubs within the corridor, The Songkhla-to-Penang subcorridor is very and eventually through global economic inte- important for Thai exports, particularly gration, the corridor will help boost the eco- rubber and related products, wood products, ­ nomic potential of areas that are lagging. and agro products. The Northern Corridor Economic Region The NCER was officially launched by the in the northern part of Peninsular Malaysia is fifth prime minister of Malaysia, Ahmad an interesting and ambitious effort to develop Badawi, at Kedah and Perlis on July 30, 2007, a corridor that extends across several states. and in Penang and Perak on July 31, 2007. As The four member states of the corridor form originally envisaged, the NCER encompassed a natural bloc for economic cooperation, the northern states of Perlis, Kedah, and given their many complementarities. Kedah, Penang, and five districts of northern Perak. Perak, and Perlis are predominantly agricul- The geographical coverage was expanded in tural, hinterland states, endowed with abun- 2016 to include the remaining districts of dant land and rich natural resources that Perak, and thus to include the entire state. remain to be fully exploited. Penang (with its The region now spans more than 32,500 port and airport) functions as the gateway to square km. the other three states. Compared with the other four corridors, the NCER is at a relatively advanced stage in The NCER’s economic profile implementing the development program at varies greatly the subregional level. It provides an interest- The four NCER states account for about ing case study of how a corridor of broader 16 percent of Malaysia’s total national out- economic activity can link an agricultural put. Penang contributes the largest share hinterland with a “modern” sector of the (6.6 percent in 2010−15), followed by Perak economy. Thus the rest of this section focuses (5.4 percent), Kedah (3.3 percent), and Perlis on the NCER, with a specific analysis (0.5 percent). Kedah is the poorest. Penang’s of the role of Penang. per capita income is about 16 percent higher than the national average, while Kedah’s is only about 47 percent of the national average. The northern corridor economic region Penang has the lowest incidence of poverty in links four Malaysian states the NCER region and nationally. Poverty The NCER’s ultimate aims are to strengthen rates in the other states range from 3.5 percent the country’s overall competitiveness and to 6.0 percent. I n s i g h t s i n t o R e g i o n a l I n t e g r a t i o n f r o m T w o C o n t e m p o r a r y T r a n s p o r t C o r r i d o r s i n Ea s t A s i a   61 Penang is the most industrialized of the the major gateway for trade to and from four NCER states, with manufacturing Asia, the Strait is one of the world’s busiest directly contributing to over 46 percent shipping lanes. The port is well placed to of GDP, compared to the NCER average of act as the logistic hub for the NCER and 31.6 percent and the national average of southern Thailand. It is already the third- 23.2 percent during 2010–15. Kedah, Perak, biggest seaport in Malaysia (based on total and Perlis are predominantly agricultural, throughput). with abundant land and rich natural resources Air cargo. The international airport at Bayan with ample prospects for further development Lepas, Penang, is the second-largest airport for (Spinanger 1986; Faaland, Parkinson, and air cargo in Malaysia (after Kuala Lumpur), Saniman 2003). and the third-busiest passenger airport after Penang has a much more diversified manu- Kuala Lumpur and Kota Kinabalu. Penang air- facturing base than the other three states. port enhances Penang’s role as a major produc- Electronics, electrical goods, and related tion center within global ­ production networks. products account for a larger share of manu- Over the past four decades, it has served as a facturing in Penang, whereas processed food major outlet for high-value-to-weight electric and other resource-based products are more and electrical goods from the surrounding Free important in the other three states. Trade Zone industrial areas and high-value-to- Penang’s manufacturing sector accounts weight electronic components from Kulim for about one-third of manufacturing High-Tech Park in Kedah, which is 44 km employment in the NCER. Penang’s labor away. Over 80 percent of the electronics and productivity of manufacturing is much higher electrical goods exported from Penang are than the other three states. This suggests that shipped by air. Penang has a relatively well-developed skill Land cargo. The Padang Besar Inland base, which the NCER can potentially draw Depot in the northern part of Perlis on the on for regional development. The wage per border with Songkhla Province is the entry worker in Penang is also much higher, indi- point for land cargo destined to Penang’s port cating that there is room for the spread of from southern Thailand. It has the potential relatively more labor-intensive production to be developed further to serve as an inte- processes away from Penang to other parts of grated logistics center for handling bulk the NCER, provided that other preconditions products exported from the region through ­ (such as logistics, infrastructure, and skill Penang’s port from the NCER’s agricultural development) are met. hinterland as well as from southern Thailand. The potential for subregional The business ecosystem is mature development is strong Penang is a mature manufacturing export hub Harmonious links between the private sector that has grown, widened, and deepened over and government agencies already prevail in the past four decades (Narayanan 1999; Gill the NCER (Narayanan 1999). The NCER can and Kharas 2007; Athukorala 2014, 2017). leverage three other core strengths to bridge There are now over 200 branch plants of mul- the development divide between Penang and tinational enterprises (MNEs), employing over the other three states: physical connectivity, a 250,000 workers, operating in Penang. They mature business ecosystem, and a skilled and include major global players such as Intel, industry-ready workforce (Sime Darby 2007). Motorola, AMD, Osram, and Hitachi. The partnership between MNEs and local firms has Connectivity is strong in penang and strengthened over time, resulting in the growth promising for the other three states of a large pool of local tooling and equipment Sea cargo. Penang’s port is situated along the manufacturing firms. By 2005, 1,956 local Strait of Malacca, which runs between firms were operating in Penang, of which 144 Indonesia, Malaysia, and Singapore. Long had graduated to “large-firm” status. 62  The WEB o f T r a n s p o r t C o r r i d o r s i n S o u t h A s i a Ancillary industries that evolved around The talent pool is large and deep the major electronics and auto firms have More than four decades of growth of manu- expanded rapidly, enhancing the network’s facturing and related activities in Penang have cohesion. Some Penang firms have become created a ready pool of talent (box 2.1). A suppliers to other high-technology firms, broad range of engineering-based expertise operating both locally and overseas, in addi- has developed to support the expansion of tion to supplying their MNE partners. After new growth sectors in the region, such as starting as small backyard workshops, many LEDs, automotive, aerospace, machinery/ local vendors and some of these firms have automation, medical devices, and biotechnol- achieved the status of full-fledged service ogy. Many of the businesses in Penang can providers with substantial research and devel- ­ meet the world-class delivery standards that opment (R&D) and design capabilities. MNEs demand. Furthermore, the continuous Others have become global players with pro- engagement between industry and academia, duction bases in other countries. along with the setting up of centers of excel- In 2005, exports made up over 40 percent lence, ensures the availability of a workforce of the total gross output of large local firms. with relevant skills to meet the demands of Local firms accounted for one-third of manu- the various sectors in the region. facturing output and over half of total The NCER accounts for about 20 percent employment in the manufacturing sector in of the nation’s workforce, of which 68 percent Penang. are working age. In 2013, an estimated 50,000 With wages and rental costs increasing graduates and 25,000 workers with technical rapidly, and available space limited in Penang, and vocational education and training (TVET) there is potential for expanding the manufac- were available in the region. TVET trainees turing base to neighboring states by further are projected to increase to 187,000 between developing infrastructure and the workforce. 2014 and 2020 (Hasri 2016). BOX 2.1  Meeting the manpower requirements of an export hub through a public-private partnership: The Penang Skill Development Centre By the late 1980s, when skill shortages began to and precision engineering skills. Over the years, hamper expansion of the electronics industry in Pen- the scope and breath of the organization have ang, the Penang Development Corporation joined expanded, influenced by technological advances and with multinational enterprises (MNEs) to establish the changing operational environment. PSDC has the Penang Skill Development Centre (PSDC). Its been successfully conducting a vendor development first training program began in July 1989. PSDC has program, known as the Global Supplier Develop- since played a pivotal role in meeting the manpower ment Programme (GSDP), to help local companies requirements of the export hub. It has attracted become world-class global suppliers by developing worldwide attention as an example of a successful their capabilities through training and by forging public-private partnership in human capital devel- linkages with MNEs. The training is divided into opment (UNIDO 2009). PSDC officials have visited two streams: manufacturing and services. Once a many developing countries to provide expertise on small or medium-sized enterprise (SME) has been how to establish similar organizations. through basic training, it is selected to enter an Initially, the PSDC training programs focused MNE coaching and mentoring program. After an on creating a large pool of technicians to meet the agreed-on period of coaching and mentoring, the immediate needs of rapidly expanding electron- MNE decides whether to accept the SME as part of ics firms, particularly just-in-time measurement (Ruffing 2006). its supply chain ­ I n s i g h t s i n t o R e g i o n a l I n t e g r a t i o n f r o m T w o C o n t e m p o r a r y T r a n s p o r t C o r r i d o r s i n Ea s t A s i a   63 To coordinate activities across the regulation, as well as the approval and con- four states, an interstate authority was trol of all buildings and building operations. established The NCIA also assists/facilitates investments by assisting investors in meeting investment The task of achieving coordination between requirements and acquiring the necessary the planning agencies of four states can be approvals. Additionally, it acts as the princi- formidable. When one or more states within pal coordinating agent monitoring the prog- the corridor have different interests or priori- ress of such projects. ties, the challenges to achieving a consensus A public-private partnership unit (UKAS) are magnified further. Malaysia therefore was created in the Prime Minister’s established a formal coordination mecha- Department to encourage private sector par- nism, the Northern Corridor Implementation ticipation as prime movers in the implementa- Authority (NCIA), to support collective deci- tion of the program. UKAS is the core agency sion making and implementation. that has been given the responsibility to coor- The NCIA is supervised by the NCIA dinate the privatization and PPP projects that Council, which is headed by the prime are eligible for funding from a facilitation ­ minister. Members include the chief ministers fund operated by UKAS. The NCIA, for its of the four states, as well as the deputy prime part, helps identify such companies or proj- minister, the chief secretary to the federal gov- ects and assists them in gaining access to these ernment, and other individuals appointed by funds. the federal government. The NCIA’s chief executive serves as secretary to the council. Apart from the chief ministers of the four The NCER blueprint guides the interstate states, all other members are from the federal authority (NCIA) government or are federal appointees (includ- The NCIA is broadly guided by a blueprint ing the NCIA chief executive). The NCIA prepared for the region by a government- operates under the Economic Planning Unit linked conglomerate, Sime Darby Berhad, (EPU) of the Prime Minister’s Department, which has interests in multiple sectors, includ- which is the coordinating/monitoring body ing plantation agriculture and the automotive for all five corridor programs. and health care industries. It was charged to The NCIA is tasked with fostering the cor- recommend “commercially sustainable mea- ridor’s growth as a whole, while minimizing sures and programs.” the tendency of member states to prioritize The Northern Corridor Economic Region’s their needs over the needs of the region, and Socioeconomic Blueprint, which was pre- fostering private sector engagement in imple- pared by Sime Darby, was released in July menting the NCER’s programs. It receives 2007 (Sime Darby 2007). It covers an 18-year both financial resources and infrastructural period from 2007 to 2025. It initially focused support from the federal government and fed- on three main economic areas—agriculture, eral agencies. manufacturing, and services—to transform The NCIA draws its authority from an act the region into a “world-class region by of Parliament, the NCIA Act 2008 (Act 687). 2025.” “Services” was later subdivided into Under the act, the NCIA has the power to tourism and logistics.6 These four areas are to require/obtain particulars and information as be developed largely by leveraging the exist- may be specified by the authority from all ing strengths in the NCER and by developing government entities, companies, and corpora- potential areas of strength. tions, and other bodies and persons operating Sime Darby eventually withdrew as the within the NCER. It also can make recom- major sponsor of the project, though its mendations to state and local authorities on ­ representative serves on the NCIA Council. local government functions and services, The Blueprint now functions as the basic including local planning, control, and guide to the NCIA, though the agency has 64  The WEB o f T r a n s p o r t C o r r i d o r s i n S o u t h A s i a departed from the Blueprint as and when it Phase II (2015−25): Broadening and was felt necessary. Deepening Private Sector Involvement in the Region The implementation of the NCIA program The focus in the second (current) phase is on has been compressed into two phases the predominantly Malay states of Perlis and Implementation of the Blueprint was initially Kedah and the newly added regions of Perak. divided into three phases. The first phase Despite budgetary cuts, the allocation for cor- (2007−12) aimed to lay the foundation by con- ridor development in the recently launched structing “priority infrastructure” and securing Eleventh Malaysia Plan, 2016−20 anchor investors. The second (2013−15) was (Government of Malaysia 2016) remains sub- devoted to broadening and deepening private stantial. Included are the following major sector involvement in the region and fostering investment initiatives. foreign and domestic business network and 1. Kedah Rubber City Project. The Rub- linkages. The third phase (2016−25) aims at ber City project aims to elevate the global achieving regional market leadership through competitive position of the Malaysian rub- sustainable market-led growth. Given the time ber industry. It is based in the heart of the lag involved in initiating and implementing n atural rubber belt that lies close to the ­ large projects, the NCIA has combined the first Malaysia-Thai border in Padang Terap, in and second phases into a single phase. Kedah. This massive investment—the con- struction of an entirely new city—is a fed- Phase I (2007−14): Securing Anchor eral government initiative, implemented Investors and Constructing “Priority by Kedah state, and backed by the NCIA. Infrastructure” When fully operational in 2025, the 1,500- During this period, the Second Penang Bridge acre (607-hectare) city hopes to attract was built, linking the emerging industrial area RM10 billion in investment and generate of Batu Kawan in Seberang Perai on the between 15,000 and 20,000 jobs directly, mainland portion of Penang state with Batu while increasing the earnings of rubber Maung on Penang Island, close to the airport. smallholders and rubber tappers. Incentive The 24 km–long bridge is the longest bridge packages for investors are being prepared, in Malaysia and southeast Asia. It helped including corporate tax exemptions, waivers encourage the expansion of the Batu Kawan of the import duty on machinery, and a sub- Industrial Park (BKIP) by providing direct sidy for the training of workers. access to the firms located there to the Penang Malaysia’s well-established base in rubber airport and facilitating the movement of man- product manufacturing is expected to give the power between the two parts of the state. Rubber City a head start and an eventual Because the 1,500-acre area of BKIP is now niche in the upstream, midstream, and down- fully occupied, the state government is plan- stream activities of the rubber industry. Local ning to develop a second industrial nearby. innovation capabilities related to rubber and The first Penang Bridge was also widened. rubber products are also envisioned. The international airport in Penang was upgraded. The railway line that runs through 2. The Kedah Science & Technology Park. the four NCER states was electrified. The In the face of shortages of land and rising NCIA supported private sector investments, rental costs, several MNE subsidiaries that including firms in the growing medical devices had long operated in Penang relocated some industry and the more mature electronics and of their operations to the Kulim Hi-Tech electrical (EE) sector in Penang. More recently, Park (KHTP) in Kedah. This paved the way a small but growing aerospace industry has for new investors to come directly to Kulim. begun developing. These companies draw their Since 2002, the park has reportedly attracted workforce from the universities in the north nearly RM32 billion in investment and gener- and rely on service-based industries in Penang. ated over 30,000 high-income jobs.7 I n s i g h t s i n t o R e g i o n a l I n t e g r a t i o n f r o m T w o C o n t e m p o r a r y T r a n s p o r t C o r r i d o r s i n Ea s t A s i a   65 Science A second industrial park, the Kedah ­ 5. Greater Kamunting Conurbation. The & Technology Park (KSTP), reserved for Greater Kamunting Conurbation (GKC) industries with advanced technologies, in Perak is aimed at transforming and is planned for a 1,950-acre site in Bukit strengthening economic sectors such as Kayu Hitam. Plans include well-equipped, tourism, manufacturing, and agriculture high-end research laboratories; business in Kamunting and Taiping, leveraging the incubation centers; and technology busi- pivotal position of GKC within the NCER. ness incubators. The emphasis will be on New infrastructure will be provided, along c ollaboration between academia, govern- ­ with initiatives to train and educate the ment, and industry to lead research and workforce. The projects are projected to commercialization projects. It is antici- ­ create 90,263 jobs by 2030. pated that the park will create 23,244 jobs 6. South Perak Development Region. This by 2030 and have “an economic impact of is made up of new areas added to the NCER. RM57.4 billion” (Hasri 2016).8 It covers 449,252 hectares and had a popu- 3. Chuping Valley Development Area. lation of 424,700 in 2010. It is expected The Chuping Valley Development Area is that the area will attract public and private in Padang Besar, Perlis, the northernmost investments totaling RM30.5 billion by state on the peninsula. Encompassing 2,482 2030 and generate 109,000 jobs. acres, it will promote three clusters—solar energy generation; green materials or It is difficult to separate the achievements technologies in manufacturing; and halal of the corridor from general progress in the industries—to transform the area into a region “Green Valley.” The effort is expected to Of the RM307 billion committed by the fed- generate 12,674 jobs and contribute around eral government to implement the five corri- RM2.58 billion to gross national income dors, only 57 percent has been utilized by 2025. (table 2.1). The NCIA is the only one of the 4. Perlis Inland Port. Another major proj- five corridor authorities to fully utilize all the ect in the Chuping Valley and alongside the funds allocated to it. The NCER has three industrial clusters is the Perlis Inland accounted for 63,500 of the 427,000 jobs Port Project (PIP), which was envisioned to created within the five corridors. ­ supplement Padang Besar Cargo Terminal, ­ The first phase has attracted investment which is struggling to cope with current worth about RM113 billion (including cargo loads and facing further increases RM71.63 billion in federal funds). This includes in volume. Once completed, the PIP is individual efforts by the NCIA and efforts expected to become the largest and busiest made in cooperation with state and federal inland (dry) port on the peninsula, serving agencies (Hasri 2016). According to the NCIA’s not only in Perlis but also Kedah and its chief executive, from its launch in 2008 through Rubber City. the end of 2016, the NCIA attracted RM79.92 The PIP, spanning 200 hectares, is a RM1.5 billion in public and private investment to the billion project that includes railway lines and NCER and created 103,600 jobs. The com- roads linked to the Chuping Valley area. It bined GDP of the four states grew at an average will feature a large-capacity container yard, annual rate of 5.8 percent between 2010 and extensive warehouse facilities, and a Web- 2014, much faster than the 3.5 percent average based computer system. To increase its con- during the 2005−09 period.9 Income distribu- nectivity to the rest of the peninsula, a tion improved considerably. The poverty rate designated highway from the border to the declined from 2.83 percent in 2007 to port and a railway extension track linking it 0.45 percent in 2014, while the median monthly to the existing double-track rail network are household income increased from RM2, 112 to also in the pipeline. RM3,797 (Hasri 2016). 66  The WEB o f T r a n s p o r t C o r r i d o r s i n S o u t h A s i a TABLE 2.1  Investment and employment in Malaysia’s five corridors, 2011−14 Investment (RM billion) Corridor Committeda Realized Jobs created 1. Northern Corridor Economic Region (NCER) 51.7 51.7 63,500 2. Iskandar Malaysia 90.4 47.1 320,100 3. East Coast Economic Region (ECER) 55.4 22.9 23,000 4. Sarawak Corridor of Renewable Energy (SCORE) 12.9 8.3 5,300 5. Sabah Development Corridor (SDR) 96.7 44.5 15,200 Total 307.1 174.5 427,100 Source: Economic Planning Unit (2016), Eleventh Malaysia Plan, 2016−2020 (based on data provided by Regional Corridor Authorities). a. The data relate to private investment supported by the NCER. These figures, while impressive, are not related to the other two components—­ logistics adequate for assessing the growth and equity reforms, and business/­ entrepreneurial devel- outcome of the NCER programs. Without opment with private sector ­ involvement—will detailed project-level data, it is not possible to presumably follow in the subsequent phases. separate the impact of NCER projects from Meanwhile, cross-border logistics issues the general process of economic/industrial with Thailand are impeding the potential to development in the region. Moreover, the increase trade with Thailand. The volume of NCIA does not maintain investment and shipments from southern Thailand has not employment data at the state level, even been fully exploited because of the failure to though the prime objective of this corridor combine the development of Penang’s port project is to narrow growth and income dis- and railways to the Thai border with initia- parities among the four constituent states. tives to improve customs clearance proce- More important, the available data do not dures at the entry point at Buki Kayu Hitam permit an assessment of whether NCER pro- on the Perlis-Thai border. Thai producers grams have contributed to narrowing the have been diverting shipments from Penang to divide between rural and urban regions and the Bangkok and Songkhla ports in Thailand between the four states. to avoid the border tax of the 0.005 cent per The impact of NCER projects is not yet kilogram on canned seafood shipments from discernible in the state-level national accounts Thailand that Malaysia recently imposed. data for 2010−15. The collective share of the These cross-border logistical issues are also four NCER states in total national GDP directly relevant to the operation of the dry (15.7 percent) remained virtually unchanged port currently under construction in Perlis. during this period, as did the shares of each of Admittedly, this is a bilateral trade issue, the four states in national GDP and their per which is outside the purview of the NCIA, but capita income relative to the national average. the NCIA can play a useful advocacy role in Data relating to the sectoral composition of highlighting the urgency of resolving it. GDP for the four states also do not indicate The business/entrepreneurial development any structural changes in their economies. initiatives thus far have not directly This persistence over a relatively short period addressed the rural/urban divide—particu- may reflect the fact that it often takes a long larly the goal to raise the living standards of time to achieve the expected gains from long- people living in rural agricultural areas. The term investment projects. programs implemented so far, as well as those proposed for the second stage, seem to have been driven by the traditional view that Limitations in efforts to date point to agriculture needs to take a backseat in the new directions process of economic development because The NCER initiatives have so far focused real income levels can be lifted only by mov- mainly on heavy infrastructure. Actions ing rural workers to modern sectors. I n s i g h t s i n t o R e g i o n a l I n t e g r a t i o n f r o m T w o C o n t e m p o r a r y T r a n s p o r t C o r r i d o r s i n Ea s t A s i a   67 The only proposed initiative in the NCER food safety standards. Export success with program that has direct implications for rais- these products has become increasingly chal- ing rural income levels relates to promoting lenging in recent years because of formal halal food. There is potential to expand the food safety standards and the quality require- halal food industry in Malaysia, but halal ments of demanding buyers (Fujita 2008; food products account for only a small share Page 2012). of world trade in processed food. Attention All four NCER states have unexploited should shift to processed food in general, potential for expanding seafood processing, including food falling under the halal in addition to agro-based, high-value category. p roducts. The International Organization ­ of Tuna Council (IOTC) has approved Processed food offers great potential Penang’s port as an outlet for tuna exports. The emergence of processed foods in world However, exports of tuna still account for trade is a structural (rather than a “pass- only a small share of products exported ing”) phenomenon that is deeply embodied from Penang’s port. According to a Penang in the ongoing process of global economic port official, trawlers from China and integration. The new export opportunities nearby islands, fish tuna in the surrounding in processed food offer several advantages seas. Their catch is exported in canned form for countries rich in agricultural resources, because it is so costly to refrigerate fresh like Malaysia. First, the final stages of fish during the long voyage. The NCER food processing are labor intensive. Second, states have potential to develop a fish pro- ­ processed food appears superior to conven- cessing industry. They could also use “min- tional manufactured exports in terms of ing ponds” (abandoned tin mines filled potential net export earnings and thus the with water) in Perak for fish farming, trans- impact on national income (GNP). Third, forming them from their common use as the processed food industry naturally has a illegal land-fill sites. strong rural base. Based on these consider- ations, the expansion of processed food The kehah rubber city project faces strong exports is a powerful vehicle for linking competition from Thailand the rural economy in a positive way The Kedah Rubber City project is a massive with the ongoing process of economic investment project, largely driven by the globalization. availability of natural rubber as an input for Neighboring Thailand is one of the main rubber-based products. There is no evidence success stories of processed food exports to suggest that the role of entrepreneurship in the developing world. Processed food and market links, and potential competition accounts for more than one-fifth of Thailand’s from Thailand, have been considered in merchandise exports (Athukorala and designing the project. In resource-based Jayasuriya 2003, 1401). Given the similarities industries, the availability of a strong raw in terms of agricultural resource endowment material base is not the sole determinant of and climatic conditions, the agricultural the development of downstream industries. In ­ hinterland of the NCER seems to have signifi- this era of falling shipping costs, raw materi- cant potential for emulating the Thai als can be transported to production loca- experience.10 tions elsewhere that meet the other Processed food production is a class of preconditions required for competitive indus- economic activity in agriculture that closely trial production. resembles manufacturing. It requires capa- Thailand has a well-established rubber- bilities to keep products fresh and deliver based manufacturing industry making tires, them from farm to processing plants and gloves, condoms, rubber-based apparel, and then to shop shelves with proper packaging rubberwood furniture. Drawing on these and labeling, while meeting international existing capacities, Thailand began work on 68  The WEB o f T r a n s p o r t C o r r i d o r s i n S o u t h A s i a its own Rubber City in southern Thailand. true, it would be important to foster dialogue The first phase is expected to be fully opera- to fully explore possible long-term ramifica- tional in 2017. The project has already tions for Penang, including the proposed attracted Michelin, while investors from Kulim International Airport, which is likely to China have also expressed interest in the affect the status of Penang Airport. Rubber City. While Thai laws prohibit for- eigners from owning land in their own name, The way forward the Rubber City is an exception.11 There is little reason to expect that Thai businesses The four member states of the NCER form a will be drawn to Kedah when similar facilities natural bloc for economic cooperation, are available in southern Thailand. given their many complementarities. Kedah, Perak, and Perlis are predominantly Institutional coordination challenges agricultural hinterland states, endowed with hamper the corridor abundant land and rich natural resources The corridors have a top-down coordination that remain to be fully exploited. Given its structure. They are federal government initia- strategic location and high global economic tives, and the statutory bodies and implemen- integration, Penang can perform the role of ­ tation authorities created to oversee their the gateway and knowledge hub for the development are also federal creations. This is NCER to bridge the development gaps understandable in the Malaysian federal across the constituent states. ­system,12 where the most important powers While it is too early to assess the full out- remain concentrated in the hands of the fed- come of the NCER initiative, a potential eral government. 13 Nevertheless, further problem looms in the future. The NCIA, engagement with state-level officials could be which is charged with implementing projects helpful given that the NCER cuts across four in the NCER, has no formal positions allotted states and that these states retain sole jurisdic- to planning officials from the member states. tion over land matters within their boundar- This will not pose problems in implementing ies. This is a powerful tool when determining projects that are in broad alignment with the the location of investments and other infra- interest of the states. Conflicts may arise, structure development.14 This may also help however, when the states’ and the NCIA’s when priorities across states regarding proj- views on projects differ. ects for the corridor may differ. The goals of the plan can be met only The original Blueprint was designed with through consultations and coordination with Penang as the NCER’s regional integrated the states and the private sector, as originally logistic hub (Sime Darby 2007; Lim 2007). envisioned. This concern is of national impor- About half the federal funds allocated to the tance because the corridor program is here to NCIA (RM71.62 billion) during 2009−15 stay. It was a key theme in the past three five- was channeled to Penang, while the rest was year development plans. The latest (eleventh) divided among the other three states. Since plan has considerably increased the commit- then, the emphasis has shifted to development ments of federal funding to the corridors, projects in the other three states. Some of the even in the face of budgetary constraints. infrastructure projects that were part of the Moving forward, representatives from key original plan to develop Penang as the logistic state implementation agencies from all mem- hub of the NCER have been postponed. ber states could be included in discussing, The justification for the emphasis on planning, and implementing NCIA projects. increased investment in the other three states is A transparent and nonpartisan stance should that Penang is already well developed in terms be adapted, as has been done by some other connectivity, of industrial maturity and physical ­ agencies established by the federal govern- while the other three states are not and there- ment. For instance, the Malaysian Industrial fore deserve more attention. While this may be Development Authority (MIDA) has an I n s i g h t s i n t o R e g i o n a l I n t e g r a t i o n f r o m T w o C o n t e m p o r a r y T r a n s p o r t C o r r i d o r s i n Ea s t A s i a   69 economy-wide focus based on the compara- Turning to transport projects supported by tive advantage and complementarity of vari- international development organizations, ous states in promoting and approving foreign chapter 3 employs a quantitative approach to direct investment nationwide and in its analyze which features could boost a corridor regions. project’s potential for generating wider eco- nomic benefits. It takes advantage of a unique new data set based on a survey of large trans- Summing up and moving to the port infrastructure (corridor) projects sup- next chapter ported by the Asian Development Bank Chapter 2 has presented two case studies of (ADB), the Japan International Cooperation government programs to develop transport ­ Agency (JICA), and the World Bank Group, corridors with positive spillover to wider and summarizes the results of an extensive economy and regional integration. In meta-analysis. Vietnam, greater transport capacity has attracted significant foreign direct investment. As a result, the areas surrounding Vietnam’s NOTES National Highway No. 5 have undergone a 1. Statistical Yearbook of Vietnam 2005. rapid and vast structural change from the 2. World Bank, World Development Indicators. agricultural to the industrial sector. The pres- 3. The Red River Delta region, home to over ence of global brands—including Canon, 17 million people, is Vietnam’s most densely Honda, and Panasonic—helped establish a populated area. It has many large industrial zones, including those in Hanoi and Hai booming “ecosystem” in the local economy, Phong, and accounts for one-fifth of the not only producing local spillovers but also country’s rice crop. integrating the NH-5 corridor into regional 4. The Indonesia-Malaysia-Thailand Growth and global supply chains. But these positive Triangle (IMT-GT) subregional program developments were not without trade-offs. aims to stimulate economic development in Accidents per km of highway significantly 32 of these three countries’ less-developed outpaced all other national highways from states and provinces, which are home to over 2000 to 2006. Environmental problems grad- 54 million people (Asian Development Bank, ually surfaced and were mitigated only later https://www.adb.org/countries/subregional​ by additional measures. -programs/imt-gt.) In Peninsular Malaysia, the development of 5. See, for example, http://www.mycorridor​ .malaysia.gov.my/IC/NCER/Pages/default​ the Northern Corridor Economic Region .aspx. (NCER) has been less successful so far, partly 6. See http://www.kulimhitechpark.com/kedah​ because of a greater bias toward pure infra- -to-set-up-more-technology-parks/. structure investment compared to the original 7. It is unclear what is meant by “economic blueprint for the initiative. The involvement of impact.” the private sector, as originally envisioned, 8. Malaysian Digest, January 25, 2017. See http:// failed to fully materialize. An uneven presence w w w. m a l a y s i a n d i g e s t . c o m / f r o n t p a g e​ of local representatives in central planning /29-4-tile/655159-ncia-to-achieve​-­accumulated​ bodies—namely, the Northern Corridor -investment-of-rm87-3-bln-by​-end-2017.html. Implementation Authority (NCIA)—resulted 9. Whether the existing land tenure system is a in diversion of some investments toward other constraint on promoting the production of high-value food production is an important subregions in Malaysia. The chapter suggests issue that is beyond the scope of this study. reforming the NCIA to help the NCER meet For an authoritative analysis of the tenure the original goals. A transparent and nonpar- system in Malaysia, see appendix A in tisan stance could be adopted, such as the Faaland et al. 2003. model used by the Malaysian Industrial 10. Star Online, January 30, 2016. See http://www​ Development Authority (MIDA), which has .thestar.com.my/business/business​ - news​ also an economy-wide focus. /2016/01/30/thai-govt-seeks-investors​ - for​ 70  The WEB o f T r a n s p o r t C o r r i d o r s i n S o u t h A s i a -­rubber-city/ and http://globalrubbermarkets​ Fujita, M. 2008. “Spurring Economic .com/40439/malaysia-thailand-in-friendly​-­com Development by Capitalising Brand Agriculture: petition-for-rubber-city-fdi.html. Turning Development Strategy on Its Head.” In 11. See Hutchinson (2014) for a succinct back- A n n u a l Wo r l d B a n k C o n f e r e n c e o n ground on Malaysia’s federal system. Development Economics: Rethinking 12. These include the power to collect all major Infrastructure for Development, edited by taxes, determine the allocation of develop- F. Bourguignon and B. Pleskovic, 205−30. ment funds to states, and provide defense, Washington, DC: World Bank. security, and transport infrastructure. Only Gill, I., and H. Kharas. 2007. An East Asian the federal government has the power to Renaissance: Ideas for Economic Growth. borrow funds from external sources. ­ Washington, DC: World Bank. 13. The states also retain powers over matters Hasri, H. 2016. “Economic Corridor pertaining to Islamic religious matters. Sabah Development for Competitive and Inclusive and Sarawak have additional powers to Asia: The Experience of Northern Corridor control immigration and the issuance of ­ Economic Region” (PowerPoint presentation). work permits in their respective states. Northern Corridor Implementation Authority, Penang. Hutchinson, F. E. 2014. “Malaysia’s Federal System: Overt and Covert Centralisation.” REFERENCES Journal of Contemporary Asia 44 (3): 422−42. ADB (Asian Development Bank). 2012. JBIC (Japanese Bank for International Indonesia-Malaysia-Thailand Growth Cooperation), with International Development Triangle (IMT-GT): Implementation Blue Center of Japan. 2003. “Impact Assessment of Print. Manila: ADB. Transport Infrastructurse Projects in the Ariff, M. 2012. “Development Strategy under Northern Vietnam.” Final Report, July. https:// Scrutiny.” Preface to Malaysia’s Development www.jica.go.jp/english/our_work/evaluation​ Challenges: Graduating from the Middle , /­oda_loan/post/2003/pdf/1-03_full.pdf. edited by H. Hill, T. S. Yean, and R. H. M. Zin, JETRO (Japan External Trade Organization). xvii– xxiii. London: Routledge. 2016. “2016 JETRO Survey on Business Athukorala, P. 2014. “Growing with Global Conditions of Japanese Companies in Asia and Production Sharing: The Tale of Penang Export Oceania.” JETRO. Hub, Malaysia.” Competition & Change 18 (3): JICA (Japan International Cooperation Agency). 221–45. 2008. “National Highway No. 5 Improvement ———. 2017. “Global Productions Sharing and Project.” Vietnam-Japan Joint Evaluation Team Local Entrepreneurship in Developing 2007. https://www.jica.go.jp/english/our_work​ Countries: Evidence from Penang Export Hub, /­e v a l u a t i o n / o d a _ l o a n / p o s t / 2 0 0 8 / p d f / e​ Malaysia.” Asia & the Pacific Policy Studies. _­project29_full.pdf. doi: 10.1002/app5.171. ———. 2009. “Ex Post Evaluation of Japanese Athukorala, P., and S. Jayasuriya. 2003. “Food ODA Loan Project: National Highway No. 5 Safety Issues, Trade and WTO Rules: Improvement Project (I) (II).” Report prepared A Developing Country Perspective.” World by Masumi Shimamura, Mitsubishi UFJ Economy 26 (9): 1395−1416. Research and Consulting Co., Ltd. https:// Athukorala, P., and J. Menon. 1999. “Outward www2.jica.go.jp/en/evaluation/pdf/2009_VNV​ Orientation and Economic Development in -5_4.pdf. Malaysia.” World Economy 22 (8): 1119−39. Lim, W. S. 2007. “Penang as an Integrated Athukorala, P., and S. Narayanan. 2017. “Economic Logistics and Transportation Hub under Corridors and Regional Development: The Northern Corridor Economic Region.” Penang Malaysian Experience.” Economic Monthly 9 (8): 1−11. Faaland, J., J. Parkinson, and R. Saniman. 2003. Lim, D. 2011. “Economic Development: Growth and Ethnic Inequality: Malaysia’s A Historical Survey.” In Malaysia: Policies and New Economic Policy , 2nd edition. Kuala Issues in Economic Development, edited by Lumpur: Utusan Publications & Distribution H. Hill, 1–38. Kuala Lumpur: Institute of Sdn Bhd. Strategic and International Studies. I n s i g h t s i n t o R e g i o n a l I n t e g r a t i o n f r o m T w o C o n t e m p o r a r y T r a n s p o r t C o r r i d o r s i n Ea s t A s i a   71 Lucas, R., and D. Verry. 2001. Restructuring the Ruffing, L. 2006. “Deepening Development Malaysian Economy: Development and Human through Business Linkages.” United Nations Resources. New York: Macmillan. Conference on Trade and Development Malaysia, Government of. 2006. Ninth Malaysia (UNCTAD). http://unctad.org/­e n/Docs​ Plan, 2006−2010. Putrajaya: Economic /­iteteb20067_en.pdf. Planning Unit. Sime Darby. 2007. Northern Corridor Economic ———. 2008. Mid-Term Review of the Ninth Region Socioeconomic Blueprint, 2007−2025. Plan, 2006 − 2010 . Putrajaya: Economic Kuala Lumpur: Sime Darby Berhad. Planning Unit. Snodgrass, D. R. 1980. Inequality and Economic ———. 2016. Eleventh Malaysia Plan, 2016−2020. Development in Malaysia. Kuala Lumpur: Putrajaya: Economic Planning Unit. Oxford University Press. Narayanan, S. 1999. “Factors Favouring Spinanger, D. 1986. Industrialization Policy Technology Transfer to Supporting Firms in and Regional Economic Development in Electronics: Empirical Data from Malaysia.” Malaysia. Singapore: Oxford University Asia Pacific Development Journal 6 (1): 55−72. Press. Page, J. 2012. “Can Africa Industrialise?” Journal UNIDO (United Nations Industrial Development of African Economics 21 (Supplement 2): Organization). 2009. Industrial Development ii86−ii125. Report 2009. Vienna: UNIDO. Part II Framework and Analytics 3 Can Transport Corridor Projects Produce Wider Economic Benefits? Evidence from International Development Organizations I nterest in studying the wider economic benefits (WEB) (economic welfare, social inclusion, reduction in inequality, environ- any positive spillovers to the wider economy. In assessing these risks, and designing better transport corridors with wider economic mental benefits, and economic resilience) of benefits, there is much to learn from the large infrastructure projects is growing design and implementation of transport (Behrens, Brown, and Lonla 2016; Redding investment projects and from practitioners. and Turner 2015).1 Attention is turning to Such a knowledge base could complement economic benefits that go beyond trade and the rich academic literature, which typically efficiency gains at the aggregate level to con- assumes a uniform project design and imple- sider benefits that reach not only large mentation in its investigations. national and international economic actors This chapter assesses which characteristics (corporations, foreign traders) but also of transport corridor project design and smaller and more localized ones (micro, implementation are correlated with wider small, and medium-sized enterprises), as well economic benefits of transport corridor proj- as poorer people. This interest has been ects. To accomplish this, a survey tool was propelled in part by the increasing number developed and used to collect and catalog of proposals for transport corridors, as well investments in large transport projects sup- as uncertainties about their gains for regional ported by three international development integration and the wider population that organization: the Asian Development Bank lives and works along these corridors. There (ADB), the Japan International Cooperation is a risk that some corridors will fail to Agency (JICA), and the World Bank. The attract the expected traffic, and thus justify data were used to analyze the characteristics the investment, while others will simply of project design and approaches to imple- become fare-throughs, and fail to generate mentation over time, across countries, and 75 76  The WEB o f T r a n s p o r t C o r r i d o r s i n S o u t h A s i a across the three organizations. Further, the which project attributes in modules 2 through data were used to run meta-regressions of 4 (basic project characteristics, project design, project performance as it relates to project and economic analyses of projects) are most design characteristics and implementation significant in explaining the project perfor- approaches. mance measures considered. The project survey tool has six modules: The analysis focuses on answering several (1) screening for eligible corridor projects, main questions. What country characteristics (2) cataloging basic project characteristics, are associated with WEB? Which design and (3) surveying project designs, (4) reviewing implementation characteristics of large economic analyses of projects, (5) transport projects are associated with WEB? summarizing project monitoring and evalua- Which complementary policies and institu- tions, and (6) assessing project performance tions (such as reforms to improve government using ­ institutional indicators as well as expert effectiveness and development policy) are opinions. The data collected cover 60 projects associated with WEB?2 Practitioners consider in 23 countries approved from 1984 to the Maputo corridor in Southern Africa 2011. The data set covers 13 ADB projects, (see box 3.1) to be an example of a corridor 22 JICA projects, and 25 World Bank that produced WEB. Is the Maputo corridor projects, totaling US$16.6 billion (in 2016 ­ an exception, or one of many successful dollars). The meta-regressions investigate corridors that have produced WEB? ­ BOX 3.1  The uncertain wider economic impacts of the Maputo Corridor in Southern Africa In the mid-1990s, after the peace agreement in The government of South Africa developed a spa- Mozambique and the collapse of apartheid in tial development initiative (SDI) strategy in the mid- South Africa, both countries embarked on reha- 1990s, which was later scaled up to the whole of bilitating the Maputo corridor, which connects Southern Africa. The strategy’s rationale is that pri- the Gauteng, Limpopo, and Mpumalanga prov- vate sector investments are the key to unlock unreal- inces of South Africa with Maputo, which is a port ized economic potential. The first step is to remove and the capital of Mozambique. a The corridor had the bottlenecks that are hindering investments. The been severely damaged during Mozambique’s civil most common bottlenecks in the region covered by war. Before Mozambique’s independence, around the SDI are transport and logistics. The second step 40 percent of South Africa’s industrial exports were is to identify strategic investment opportunities in transported along the Maputo corridor. The corri- agribusiness, industry, tourism, and other sectors dor traverses a vast cross-border region with high with high demand for transport and logistics ser- potential for industry, trade, agriculture, tourism, vices that can trigger additional upstream or down- and mining. The goal of Mozambique and South stream development. Africa was to unlock the economic potential of the The Maputo corridor was the first regional spatial region through private-sector−led investments in development initiative. Restoring the Maputo cor- transport infrastructure and economically viable ridor as a credible route for regional trade required anchor projects. To do that, the governments of a combination of reforms and investments in the Mozambique and South Africa made the political transport system. Investments in the EN4/N4 high- and economic environment favorable to attract pri- way, the Ressano Garcia railway, the Lebombo-Res- vate sector investments. sano Garcia border post, and the Port of Maputo (box continued next page) C a n T r a n s p o r t C o r r i d o r P r o j e c t s P r o d u c e W i d e r Ec o n o m i c B e n e f i t s ?    77 BOX 3.1  The uncertain wider economic impacts of the Maputo Corridor in Southern Africa (continued) became a prerequisite for reviving the corridor. After the civil war in Mozambique, repairing and The anchor investment projects underpinning the rebuilding the transport infrastructure required major Maputo SDI included the mega-projects of MOZAL capital investments, which the country could not (the second largest aluminum smelter in Africa), afford with its own public resources. The best alter- the Beluluane Industrial Park, the Pande and Temane natives were concessional financing from international gas fields, and a petrochemical cluster. A pipeline and a development organizations and private sector involve- power line were also built to support these investments. ment. At the same time, the trend in Africa and other Key aspects of the design of the Maputo corridor developing regions was to promote the involvement were the upstream work, and the role of the public of the private sector in the provision and management and private sectors. As part of the SDI, the develop- of transport infrastructure and services. However, the ment potential and infrastructure needs along the cor- demand that would make the private sector invest- ridor were scoped out and mapped. This created the ments viable was uncertain. The anchor projects, foundation for a holistic approach to revive the cor- which were all transport-intensive, played the addi- ridor. Soon after the agreement between the govern- tional role of supporting the viability of private sector ments of South Africa and Mozambique, the Maputo involvement in improving road, rail, and port services. Corridor Company (MCC), a public-­ private body to The Maputo corridor seems to be a successful cor- facilitate investments on the corridor, was created. ridor in many aspects, even though no robust anal- MCC was dominated by South African ­ public agen- ysis of the wider economic benefits of the corridor cies, with limited involvement from Mozambique and exists. Rehabilitating the corridor not only improved the private sector. In the early 2000s, MCC ceased the connectivity between South Africa and Mozam- to exist, and the Maputo Corridor Logistics Initia- bique, as demonstrated by increases in road and rail tive (MCLI) was created. MCLI is a nonprofit orga- traffic, but also boosted transit trade flows and bilat- nization consisting of infrastructure investors, ser- eral trade between the two countries. The increase in vice providers, and stakeholders from Mozambique, road traffic has been mainly in passenger traffic—and South Africa, and Swaziland focused on promoting thus an increase in people-to-­ people interactions. and further developing the Maputo corridor as the The Maputo corridor led to more than US$5 billion region’s primary logistics transportation route. worth of investments, and 15,000 direct jobs in the Private sector involvement was key in improving construction and operation of transport, logistics, the transport components of the Maputo corridor. energy, and industrial ventures along the corridor. Source: Sequeira, Hartman, and Kunaka 2014. a. Swaziland and Zimbabwe, both landlocked countries, also rely on the corridor for access to the sea. To our knowledge, this study is the first determinants of success for selected projects attempt to analyze transport corridor proj- of other international development organi- ects supported by international development zations, such as the African Development organizations since 1980, and link the proj- Bank (Mubila, Lufumpa, and Kayizzi- ect investments to project performance using Mugerwa 2000). However, these studies do meta-analysis. Various studies have analyzed not focus on the investments in transport macro and micro determinants of World corridors and their particular challenges, Bank project performance (Denizer, such as geographic specificity and the net Kaufman, and Kraay 2013), and have economic benefits in their vicinity as well as offered explanations for the success of ADB further away. This study thus aims to con- projects and programs (Feeny and Vuong tribute to the existing literature by highlight- 2017). Other studies have analyzed the ing the aspects of project design and 78  The WEB o f T r a n s p o r t C o r r i d o r s i n S o u t h A s i a implementation that could matter the most low trade barriers, and accessible and quality for the success of large transport corridor health care and education. investments. As globalization and innovation change the way people and goods move around the world, regional connectivity is becoming This study highlights the aspects of designing increasingly fluid and disruptive. Historically, and implementing corridor projects that could matter developing countries have sought transport the most for the success of transport corridor improvements to increase trade, catalyze programs. growth, and create jobs. While better trans- port connectivity is necessary for regional integration, achieving widespread gains from CONCEPTUAL FRAMEWORK regional integration requires more—a holis- tic approach to transport corridor develop- Trunk transport corridors play a crucial role ment to maximize the wider economic in connecting people with goods and services benefits that accrue from building trunk and fostering regional connectivity. The aca- corridors. demic literature links improved transport In general, three different types (levels) of infrastructure to wider economic benefits transport project/program designs can through four key mechanisms: reducing trans- enhance regional integration and generate port and production costs; expanding pro- wider economic benefits (see figure 3.1): ductive capacity; improving access to markets and basic services like health care and educa- 1. Investments in trunk transport corridors. tion; and reducing the prices of final goods This entails building entirely new trans- and services. These benefits depend on sup- port infrastructure (roads, rail, or inland portive conditions in other sectors, such as waterways) or upgrading existing links. access to credit, functioning land markets, 2. Investments in trunk transport corridors, as well as transport and trade facilitation services. Benefits from a trunk transport FIGURE 3.1  The design of transport corridor projects respects initial conditions and can involve three levels of interventions corridor investment can be enhanced if simultaneous investments are made in transport services (trucks, rail, and port services) and trade facilitation services (warehouses and border crossings). 3. Investments in trunk transport corri- Soft complementary policies dors, transport and trade facilitation (capital, labor, land, product markets) services, and other “soft” policies. Ben- efits from improved regional connectivity can be expanded even further if the project designs also account for market conditions Transport and trade facilitation and simultaneously make complementary (ports, warehouses, border crossings) interventions in other sectors—to ensure that capital, labor, land, and product mar- kets function smoothly. Trunk transport corridor The wider economic benefits of a trunk (road, rail, waterway) transport project also depend on the initial conditions. These conditions can include geog- raphy, population, development level, and Initial conditions market imperfections. These initial conditions (geography, population, market imperfections) can magnify or reduce the wider economic benefits of transport corridor investments for all three levels described in figure 3.1. C a n T r a n s p o r t C o r r i d o r P r o j e c t s P r o d u c e W i d e r Ec o n o m i c B e n e f i t s ?    79 Therefore, any assessment of potential whether they would qualify as a transport gains from trunk transport infrastructure and corridor. A transport project is classified as a services projects should also consider the state transport corridor if: of the initial conditions and interactions with •  The transport investment is being made on complementary markets to identify comple- a route that is creating or upgrading/reha- mentary interventions that are needed to bilitating at least 100 linear kilometers maximize the benefits that can accrue from (km), or the investment. However, the analysis of such •  The loan at appraisal is greater than interactions—assessing how and when trans- US$50 million and the investment is in a port infrastructure projects can be designed to new bridge/tunnel connecting at least two maximize the wider economic benefits—is economic centers. largely missing from the literature, leaving significant knowledge gaps across the spec- These criteria were designed to identify trum of transportation settings. projects that have the potential to enhance regional connectivity of a country or a set of countries. This is measured either using length SAMPLE SELECTION AND (restricting the length to at least 100 linear SUMMARY STATISTICS kilometers) (see map 3.1) or by identifying The study selected a sample of 60 closed investments on critical links of transport net- transport corridor projects from the Asian works (large bridges and tunnels), measured Development Bank, Japan International using loan amounts. The analysis included Cooperation Agency (JICA), and the World those projects that satisfied at least one of Bank. The study team administered a survey these two criteria. (comprising 68 questions) on these projects to The screening resulted in a sample of 60 create a database of key project characteris- projects—13 ADB projects, 22 JICA projects, tics, design, monitoring and evaluation, and and 25 World Bank projects. For the most ex post assessments. This database was cre- part, the selected projects were approved ated by reviewing project documents and between 1984 and 2011. interviewing individuals involved in the proj- The sample of selected projects consists of ects. The study team then performed a meta- investments made in both domestic and inter- analysis using the collected data and other national corridors. In 32 of the 60 projects, spatial data sets to assess the conditions and the investment is made on a route that is part features of corridor projects that can help of a larger regional corridor. Seven projects produce wider economic benefits (see make an investment at the border. For 9 proj- figure 3.2). ects, the investment is made in landlocked The sample of transport projects was countries only. While the thrust of the study is ­ l imited to those involving investments in very much toward regional integration, roads, rail, or inland waterways. Projects domestic transport corridors tend to have an were subjected to screening criteria to assess impact on regional connectivity as well and FIGURE 3.2  Overall approach of the study Key project Corridor success Systematic characteristics Overview Meta-analysis measures review and design 80  The WEB o f T r a n s p o r t C o r r i d o r s i n S o u t h A s i a MAP 3.1  How a linear kilometer is defined National Capital Roads Moradabad NEPAL State Capitals NEW DELHI Major Cities Bareilly • The investment is on the Churu Bikaner Lakhimpur same—or a handful of— PAKISTAN road, rail, or waterway Sikar routes connecting the same locations. Nagaur Agra Jaisalmer Jaipur Jaipur Lucknow • The investment does not Bhind INDIA INDIA have to be geographically Kanpur contiguous. Jodhpur Beawar Gwalior Sawai Madhopur Barmer Orai • As an example, the roads Balotra marked in red and pink are linear kilometers, but Bhilwara Kota Jhansi Baran Shivpuri the roads marked in blue are not. Chhatarpur Chittaurgarh Guna Lalitpur Udaipur Ashoknagar Palanpur Mandsaur 0 100 200 Kilometers IBRD 43391 | JANUARY 2018 Source: Corridor Study Team. FIGURE 3.3  Most of the reviewed projects are in the road sector All (n = 60) 40 15 2 3 World Bank (n = 25) 18 3 2 2 JICA (n = 22) 13 8 0 1 ADB (n = 13) 9 4 0 0 10 20 30 40 50 60 70 80 90 100 Percent Road Railways Inland waterways Multimodal Source: Corridor Study Team. are thus included in the sample. Much can be development organizations are based on roads. learned from domestic corridors in large Overall, the sample spans 40 road projects, countries (India, China, Brazil, Russian 15 rail projects, 2 inland waterways projects, Federation). Furthermore, in landlocked and 3 multimodal projects (road and railway, countries, people usually live at the center of and railway and port). The cumulative num- the country rather than at the borders. In bers and loan amounts of corridor projects in such cases, the WEB of regional connectivity the sample follow similar trends (see figure 3.4). are expected to accrue where people live, The regional distribution of the selected rather than at the border. projects is largely consistent with the As shown in figure 3.3, most the projects regional portfolio of the international devel- selected for each of the three international opment organizations (see table 3.1). C a n T r a n s p o r t C o r r i d o r P r o j e c t s P r o d u c e W i d e r Ec o n o m i c B e n e f i t s ?    81 The portfolios and samples of projects for Just over one-third of the corridor projects ADB and JICA are largely based in Asia, are greenfield. Slightly more than half while the World Bank portfolio spans five of (55 percent) focus on rehabilitating infrastruc- the six regions of the world. ture. The remainder (10 percent) expand the Figure 3.5 summarizes the geographic capacity of existing infrastructure. scope of the sample projects. Most projects In just over 40 percent of the projects, appear to have an objective of improving the private sector was either consulted regional connectivity within a country. or involved before, during, or after the design Less than one-fifth of the projects have an of the transport investment ­ ( figure 3.7). international scope. However, the level of private sector All projects in the sample are underpinned ­ participation in the operation and manage- by a theory of change. However, the quality ment of transport infrastructure in the sam- of that theory varies. A panel of experts pled ­ projects remains below 50 percent. The assembled for this study rated the quality of most common mode of engagement was the theory in 51 percent of these projects as contractor(s), followed by public-private being fairly poor or poor (see figure 3.6). partnerships (PPPs) ­(figure 3.8). FIGURE 3.4  The sample of reviewed projects captures more recent projects, 1984–2011 a. Cumulative counts of projects b. Cumulative loan amounts 60 18 50 15 US$ billion (2016 dollars) Number of projects 40 12 30 9 20 6 10 3 0 0 84 87 90 93 96 99 02 05 08 11 84 87 90 93 96 99 02 05 08 11 19 19 19 19 19 19 20 20 20 20 19 19 19 19 19 19 20 20 20 20 Appraisal year Appraisal year ADB JICA World Bank All Source: Corridor Study Team. TABLE 3.1  Most of the projects in the sample are in Asia Number of projects Region ADB JICA World Bank All East Asia and the Pacific 10 17 9 36 Europe and Central Asia 1 1 4 6 South Asia 2 2 3 7 Sub-Saharan Africa 0 0 3 3 Middle East and North Africa 0 2 0 2 Latin America and the Caribbean 0 0 6 6 Total 13 22 25 60 Source: Corridor Study Team calculations. 82  The WEB o f T r a n s p o r t C o r r i d o r s i n S o u t h A s i a FIGURE 3.5  Only a small share of the projects in FIGURE 3.7  The private sector is often not the sample are international involved in designing projects Localized, 5% International, 18% Consulted/involved before/during design, 28% Not consulted/involved, 59% National, Regional within 23% Consulted/involved country, after design, 54% 13% Source: Corridor Study Team. Source: Corridor Study Team. FIGURE 3.8  The private sector is often not FIGURE 3.6  Most projects in the sample do not involved in the operation or management of have a good theory of change transport infrastructure Public-private partnership, Good, Poor, 15% 30% 32% Contractor, 33% Not involved, Fairly good, Fairly poor, 52% 19% 19% Source: Corridor Study Team. Source: Corridor Study Team. To measure the presence or absence of Almost half the projects in the sample are complementary policies before the project in countries that had a DPO. The three most started, the study ascertained whether the common forms of DPOs in the sample are host country had at least one World Bank public administration, industry-trade-services, Development Policy Operation (DPO) (loan, and financial sector. Of the 29 projects with grant, or credit)in the five years before the DPOs, only 8 had a DPO relating to the project was approved.3 transport sector (figure 3.9). C a n T r a n s p o r t C o r r i d o r P r o j e c t s P r o d u c e W i d e r Ec o n o m i c B e n e f i t s ?    83 FIGURE 3.9  Only about half the projects occurred in countries that had undertaken World Bank– sponsored institutional reforms in the five years before the project was approved Any DPO in the past 5 years 29 Public administration 22 Industry-trade-services 21 Financial sector 17 Energy 17 Social protection 15 Health 12 Agriculture 11 Education 10 Transportation 8 ICT 6 Multisector 4 Water-sanitation 2 0 5 10 15 20 25 30 Source: Corridor Study Team. Note: DPO = Development Policy Operation; ICT = information and communication technology. MAXIMIZING THE WIDER interest are initial conditions, project design ECONOMIC BENEFITS OF and characteristics, and complementary pol- TRANSPORT CORRIDORS icies and institutions. In addition, the analy- sis controlled for characteristics specific to Hypothesis and analytical framework the international development organization The analysis tested four hypotheses. Are proj- and the type of rating (through interna- ects more or less successful when: tional development organization fixed effects and rating type fixed effects). The •  They are initiated in countries that are estimations were performed using ordinary richer, bigger, and easier to connect (initial least squares. conditions)? Four metrics of project success were used: •  They have a large geographic scope and two that capture the likelihood that a project include a well-thought-through theory of produced WEB, and two that proxy for change (project design and characteristics)? WEB. This allows the analysis to assess •  They engage and/or involve the private sec- which characteristics are linked with the tor (project design and characteristics)? likelihood that the project will generate •  They proceed against a backdrop of WEB and which characteristics are linked complementary policies and institutions with WEB. (measured through openness, government Both the likelihood that a project will pro- effectiveness, and DPOs)? duce WEB and the nature and extent of the WEB themselves need to be considered The study tested the hypotheses by esti- because the monitoring and evaluation (M&E) mating multivariate regression models that frameworks of projects tend to be based on use a set of project success measures as concrete outputs (such as number of/km of outcome variables. Explanatory variables of roads built or rehabilitated, travel time, 84  The WEB o f T r a n s p o r t C o r r i d o r s i n S o u t h A s i a number of accidents, daily ridership, and so °° Rating = 2. Achievement of basic out- on) rather than focusing on outcomes, which comes of reduced operating costs and are broader and harder to measure. It is also travel times. more difficult to determine cause and effect °° Rating = 3. In addition to the basic out- (attribution): how the project affects specific comes of (2), lower transport fares (for outcomes. All projects have monitoring and passengers) and/or tariffs (for freight). evaluation indicators that measure basic out- °° Rating = 4. In addition to the lower- comes, but only some of them have indicators level outcomes of (3), a higher level of that capture intermediate outcomes or final trade and/or economic growth than outcomes (WEB).4 In the sample, 46 percent of would have been achieved without the projects have indicators that relate only to project. basic outcomes, 30 percent have at least one °° Rating = 5. In addition to the outcomes indicator that relates to an intermediate out- of 4, some increase in levels of personal come, and 24 percent have at least one indica- incomes and or reductions in poverty tor that relates to WEB. levels. Against this backdrop, measuring the suc- In the sample, the average official comple- cess of projects based on institutional ratings tion rating for projects is 4.08, while or even experts’ assessments can only indicate the ­ a verage expert’s rating is 3.53. Why whether a project was successful in setting were the expert ratings lower? Some projects the conditions for WEB to arise. In other have objectives that are very easy to achieve, words, these measures of success are only while other projects that are very similar in indicative of the likelihood that a project pro- content could have objectives that are much duced WEB. more difficult to achieve. Ranking projects on The likelihood of a corridor project pro- the extent to which they achieve their own ducing WEB is measured using two ratings stated objectives would run the risk of assess- that range from 1 (lowest) to 5 (highest): ing a project that achieved objectives that were not very ambitious more highly than one that •  Official completion rating. This is the achieved the same lower-level objectives, but average of the internal rating when the did not completely achieve its more ambitious project is completed and the independent objectives. evaluation rating (if both ratings are avail- To measure WEB, the study uses two prox- able) or the rating that is available.5 ies for spatial economic activity near transport •  Expert assessment. Official completion corridors: nightlights data (as a proxy for the ratings assess projects only according to intensity of economic development) and human project objectives and monitoring and settlement data (as a proxy for the density of evaluation measures. For this reason, development).6 In both cases, the outcome this study produced a more comprehen- variable is created using a difference-in-­ sive (although somewhat subjective) difference approach. The first difference com- assessment that captures the extent to pares average nightlights and human settlement which the projects contributed to or levels observed in areas close to the corridor established the conditions to achieve (treated areas) to areas far away from the cor- higher-level outcomes. These assess- ridor (control areas). Thus, it is a comparison ments were performed by knowledge- across space. The second difference compares able experts from each organization. the first difference across time, subtracting the The assessment was based on existing level before the project started from the level data and analysis on the projects. Five after the project was completed to ascertain the objective criteria were used to create the effect of the project. The outcome variable is expert assessment: the difference of these two differences (that is, °° Rating = 1. Failure to achieve even sim- the difference in difference itself). ple objectives, such as reduced trans- For the first difference, the control group port operating costs and travel times. consisted of areas far away from the corridor C a n T r a n s p o r t C o r r i d o r P r o j e c t s P r o d u c e W i d e r Ec o n o m i c B e n e f i t s ?    85 (90−100 km away), while the treatment °° Rating = 1. Project objectives are lim- group consisted of areas “on” the corridor) ited to transport operating costs and (0−10 km from the corridor) and “by” the times. corridor (0−20 km from the corridor). For °° Rating = 2. The project documents the second difference, for the initial year, the mention some higher level of objec- study used the latest year before the project tives (such as an increase in trade or was appraised for which the data on night- economic growth), but is silent on lights and human settlement data were the conditions needed to achieve available. For the final year, the study allowed these objectives (for example, com- for the longest gestation period possible for petitive markets to translate reduc- WEB to emerge after the project was com- tions in transport operating costs and pleted. Thus, the last year for which data times into lower tariffs). were available was used: 2012 in the case of °° Rating = 3. In addition to 2, the proj- nightlights, and 2014 in the case of human ect documents mention conditions in settlement data. The analysis pooled the data the transport sector needed to achieve from the two data sets (nightlights and human the higher-level objectives. settlement) and pooled the data for the two °° Rating = 4. In addition to 3, the proj- treatment and control groups (0−10 km, ect documents mention conditions 90−100 km) and (0−20 km, 90−100 km). beyond the transport sector needed Unfortunately, human settlement data were to achieve the higher-level objectives. available for only 45 projects and nightlights °° Rating = 5. In addition to 4, the proj- data were available for only 34 projects. Thus ect documents include an assessment the sample of projects the analysis could use of whether the conditions have been to estimate WEB was much smaller than the met, and if not, what actions will be total sample of 60 projects. taken to create them. In terms of potential correlates, the analy- °° Private sector consultation [0−1]. This sis considers the following set of variables: is a dummy that is 1 if the private sector was consulted before, during, or after •  Initial conditions are captured through the project was designed, and 0 if the three variables: private sector was not consulted. °° Log land area. This is the logarithm of a °° Degree of private sector involvement country’s area, in 1,000 hectares.7 [0−3]. This captures the extent of pri- °° Terrain ruggedness index. This is mea- vate sector involvement in the operation sured in terms of the average difference or management of the transport infra- in elevation (measured in hundreds of structure. The variable is 0 if no private meters of elevation) for points 30 arc- sector actor is involved; 1 if the involve- seconds apart (that is, 926 meters on ment is only as a contractor; 2 if the the equator or any meridian) not cov- involvement is as part of a public-pri- ered by water. vate partnership; and 3 if the private °° Log GDP per capita at appraisal. This sector actor is the owner: is the log of GDP per capita at the time °° Degree of connectivity increase [1−3]. the project is appraised (in constant This variable captures the degree of 2010 US$million). increase in connectivity, ranging from •  Project design and characteristics are 1, for the highest increase in connec- captured through seven variables: tivity (greenfield projects), to 2 °° Quality of theory of change. This is an (expanding the capacity of an existing expert rating of the theory of change, link), to 3, for the lowest increase in with ratings of 1 (poor), 2 (fairly poor), connectivity (rehabilitation of an 3 (fairly good), 4 (good), and 5 (excel- existing link). lent). The following objective criteria °° Investment at border [0−1]. This is a were used to create these rankings: dummy that is 1 if the project makes 86  The WEB o f T r a n s p o r t C o r r i d o r s i n S o u t h A s i a any investment at a country border, and It ranges from approximately −2.5 0 if it does not. (weak) to +2.5 (strong). °° Geographic scope [1−4]. This variable °° DPOs. Dummies are included for captures the geographic scope of the whether, in the five years before the proj- investment, ranging from 1 (local), to ect was approved, the World Bank had 2 (subnational or regional within any Development Policy Operations in the country), to 3 (national), to 4 the country. Various dummies are used (international). in relation to DPOs. The general dummy °° Landlocked-ness [0−1]. This is a dummy is 1 if any DPO had been undertaken in that is 1 if all countries involved in the the country in the past five years, and project are landlocked, and 0 if they 0 otherwise. Similarly, separate dummies are not. are included for DPOs in various sectors, •  Complementary policies and institutions including transport, agriculture, financial are captured through three variables: sector, and industry-trade-services. °° Openness at approval. This measure of trade openness consists of the sum of Regression analysis: potential for WEB imports plus exports as share of GDP. °° Government effectiveness at approval. The estimated regression pools data for both This is the Worldwide Governance metrics capturing the likelihood of achieving Indicator (WGI) on government effec- WEB (institutional project completion rat- tiveness. It reflects perceptions of the ings and expert success ratings). This pooling quality of public services, the quality of helps enhance the sample size for the regres- the civil service and the degree of its inde- sion model and averages the pros and cons pendence from political pressures, the of the two success measures. Based on this quality of policy formulation and imple- pooling, a regression analysis of 120 obser- mentation, and the credibility of the gov- vations should be possible. However, because ernment’s commitment to such policies. some project success ratings and some values in the explanatory variables were missing, the sample that results includes 55 projects FIGURE 3.10  Well-thought-out theory of change can help corridor and 107 observations. Figures 3.10 and 3.11 projects succeed, the expected benefits of consulting or involving the private sector have not been realized present the ordinary least squares estima- tions results. Two models were estimated: a 0.5 parsimonious version that estimates only an average effect for development policy opera- 0.4 tions, and an extended version that breaks 0.3 down the development policy operations 0.2 by sector. Some findings from the regression 0.1 analysis are consistent across different specifi- 0 cations, despite the small sample size. –0.1 Initial conditions do not seem to be signifi- –0.2 cantly associated with the success of a project in setting the conditions for WEB. –0.3 Characteristics of the project and project –0.4 design process are significant (figure 3.11). In Consultant’s assessment Private sector Private all the specifications, a well-thought-out the- of the theory of change involved sector consulted ory of change is consistently associated with Parsimonious model Extended model the success of projects in setting the conditions Source: Corridor Study Team. for WEB. These findings suggest that projects Note: Significance level = 10 percent. with a clear link between the intervention, C a n T r a n s p o r t C o r r i d o r P r o j e c t s P r o d u c e W i d e r Ec o n o m i c B e n e f i t s ?    87 FIGURE 3.11  Complementary financial, and industrial and trade reforms undertaken before a corridor project is implemented could help the project succeed 3 2 1 0 –1 –2 –3 Openness Education DPO Financial sector Industry-trade- Multisector Public at appraisal DPO services DPO DPO administration DPO Parsimonious model Extended model Source: Corridor Study Team. Note: Significance level = 10 percent. DPO = Development Policy Operation. intermediate outcomes, and WEB are more association with project success. The World thorough in their design and more cognizant Bank supports public administration reforms of the design features needed to be successful.8 through DPOs when a country has weak gov- The degree of private sector involvement ernment capacity. Hence, this result might be (contractor, public-private partnership, or picking up continuing weaknesses in the coun- ownership) and consultation appear to have try’s implementation capacity. It could also be a somewhat negative association with proj- the case that reforms of public administration ect success. This could be interpreted as the are highly centralized and draw public need to improve the quality of engagement resources away from the transport sector that with the private sector, not necessarily to is implementing the corridor projects. limit it. Some complementary reforms imple- Regression analysis: selected WEB mented before a corridor project has been (economic activity) approved appear to alter the likelihood of a project producing WEB (figure 3.11). The estimated regression pools data across When using reforms supported by World two different measures of economic activity: Bank DPOs as proxies for sectoral comple- nightlights and human settlement. In addition, mentary reforms, the analysis finds that finan- the regression pools data for two treatment cial, industrial, and trade reforms undertaken groups: 0−10 km and 0−20 km away from the before a project is implemented have a posi- corridor project. These measures are aimed tive association with project success. Projects to capture areas both “on” and “by” the cor- implemented in countries where public admin- ridor to varying degrees (to varying distances). istration reforms took place not long before The corridor impact (treatment effect) is esti- the project was approved have a negative mated against a control group of areas 88  The WEB o f T r a n s p o r t C o r r i d o r s i n S o u t h A s i a 90−100 km “away” from the corridor project. It indicates that, on average, building a large The variance of the two measures of economic transport corridor project is associated with a activity is standardized. This double pooling 0.7 standard deviation increase in economic helps enhance the sample size and robustness activity (measured using nightlights and human of the estimated regression, thanks to the settlement data across projects in that country). information delivered by two complementary Using the sample estimated standard deviation measures of economic activity—that is, the and assuming 0.3−0.4 correlation of nightlights intensity of development (nightlights data) with GDP (based on Henderson, Storeygard, and the extension of development (human and Weil 2012), a US$100-million-plus trans- settlements data). The transport corridor port corridor intervention could, on average, treatment becomes a common factor for the be associated with a GDP increase of about two measures of economic activity. Thus, the 0.78−1.04 percentage points near the corridor. estimated regression can be seen as a sim- The analysis captures a partial effect; that is, ple common factor model. As before, two the increased economic activity near a trans- models were estimated: a parsimonious ver- port corridor could just be a reallocation of sion that estimates only an average effect for economic activity from another area, instead of development policy operations, and an a newly generated activity. extended version that breaks down the devel- Several conditional effects accompany the opment policy operations by sector. The average change in economic activity near a results are reported in figures 3.12 to 3.14. given transport corridor project. The condi- The estimations reveal, on average, a tional effects are associated with initial condi- ­ significant increase in economic activity in tions, project design and characteristics, and ­ connected regions (those areas 0−10 km and complementary policies and institutions. Four 0−20 km away from the corridor), after trans- initial conditions are significantly associated port corridor projects have been built, as mea- with the average increase of economic activity sured by human settlement and nightlights near transport corridor projects (figure 3.12). data. This average positive change is gauged The increase in economic activity near the relative to the data for control regions corridor could be significantly greater in 90−100 km away from the alignment. smaller countries. Further, economic activity FIGURE 3.12  Initial conditions of a smaller, flatter, less developed, and coastal country could increase the likelihood for WEB 0 –1 –2 –3 –4 –5 –6 –7 Log land area Terrain ruggedness Log GDP per capita Landlocked country/ index (100 meters) at appraisal countries only Parsimonious model Extended model Source: Corridor Study Team. Note: Significance level = 10 percent. GDP = gross domestic product; WEB = wider economic benefits. C a n T r a n s p o r t C o r r i d o r P r o j e c t s P r o d u c e W i d e r Ec o n o m i c B e n e f i t s ?    89 FIGURE 3.13  Investing at the border and in connectivity to border crossings shows smaller increases in local economic activity: inherent complexity could be the challenge 0.4 0.2 0 –0.2 –0.4 –0.6 –0.8 –1.0 Consultant’s Degree of Investment Geographical Private sector assessment of the connectivity at the scope involvement theory of change increase border [1–4] [0–3] Parsimonious model Extended model Source: Corridor Study Team. Note: Significance level = 10 percent. FIGURE 3.14  Policies promoting trade openness and institutions boosting governance effectiveness appear the most robust complementary interventions to help corridor investments spur local economic activity 6 4 2 0 –2 –4 –6 –8 –10 Openness at WGI at appraisal: Any DPO in Agriculture Health DPO Industry-trade- Public appraisal Government the past 5 years DPO services DPO administration effectiveness DPO [–2.5, 2.5] Parsimonious model Extended model Source: Corridor Study Team. Note: Significance level = 10 percent. DPO = Development Policy Operations; WGI = Worldwide Governance Indicators. 90  The WEB o f T r a n s p o r t C o r r i d o r s i n S o u t h A s i a near a transport corridor project could economic activity around the transport corri- increase more in countries with flatter or dor (figure 3.14). The estimation results sug- smoother terrain. In less developed countries gest that increases in economic activity near (with lower per capita GDP), the increases in corridor projects could be much greater in economic activity near the transport corridor countries with greater trade openness. intervention could be relatively larger—­ Similarly, the average increases in economic perhaps spurring the income convergence activity near transport corridor projects are process in the connected regions. Also, when associated with greater governance effective- the corridor intervention happens in coastal ness. For instance, a 1-point increase in the countries, the positive association with eco- index of governance effectiveness of a country nomic activity is much greater than in land- (for example, from zero (India) to 1.0 locked ones. (Malaysia) or −0.7 (Pakistan) to 0.3 (South Some project designs and characteristics of Africa)is associated with a more than fourfold transport corridors are estimated to be signifi- increase in economic activity.10 cantly associated with changes in economic Reforms, however, could be also disrup- activity (figure 3.13). Adding new infrastruc- tive or take a long time to complete. In some ture or upgrading existing infrastructure is cases, broad-based reforms—approximated associated with a larger increase in economic by the presence of World Bank Development activity than rehabilitating trunk infrastruc- Policy Operations (DPOs)—undertaken ture. Investment at the border and connectiv- before a corridor project is implemented ity to border crossings are associated with could interact negatively with transport cor- significantly smaller changes in economic ridor interventions. Estimations containing activity—perhaps because of the additional greater detail on the sectoral composition of complexities that border crossing interven- these reform programs reveal that the effect tions entail. However, even this estimate is not varies substantially by sector. First, on aver- robust against the inclusion of more detailed age, the effect of the DPOs becomes statisti- complementary policies. The estimated asso- cally insignificant. Second, the estimation ciation of geographic scope suggests that suggests that some complementary reforms changes in economic activity near national could have significantly positive associa- and international projects could be larger tions, while others could have significantly than near subnational or local transport infra- negative associations with changes in eco- structure projects—perhaps because of their nomic activity near transport corridor proj- greater potential for regional integration. ects. As expected, policy reforms that Interestingly, only the estimation accounting promote trade and industry competitiveness for detailed complementary policies picks up are associated with significantly higher a possible negative association between pri- economic activities along transport corri- ­ vate sector involvement in the transport cor- dors. A similar positive association is deliv- ridor projects and economic activity. This ered by health sector reforms.11 In contrast, finding, if confirmed by future studies, could reforms of public administration and the suggest that, so far, public-private partner- agricultural sector are associated with ships have hindered widespread economic smaller changes in economic activity. One activity rather than spurred it. One explana- explanation could be that reforms of public tion could be that it is difficult to align the administration are highly centralized and private sector’s focus on monetary profits draw policy attention and public resources with society’s focus on generating wider socio- to the aggregate, national level rather than economic benefits.9 leaving them freely available to support eco- The cross-sectoral complementary policies nomic development along transport corri- and institutions, such as trade openness and dors. Further, if transport corridors aim at governance effectiveness, have robust and inducing structural transformation from positive association with the generation of agriculture to manufacturing, reforms to C a n T r a n s p o r t C o r r i d o r P r o j e c t s P r o d u c e W i d e r Ec o n o m i c B e n e f i t s ?    91 boost agricultural ­ performance could pro- evaluate its success—typically right after the long the needed reallocation of resources implementation stage has been completed— from agriculture to manufacturing. are “getting it right.” In sum, smaller developing countries with smoother terrain and access to the sea could experience a greater boost in economic activ- Improvements in economic activity near corridor ity near transport corridors projects if they projects could be greater in smaller, less-developed are open to foreign trade and have effective governance. More tentatively, the quality of countries with smoother terrain and direct access to engagement with the private sector needs to the sea. improve for corridor projects with private widespread participation to be associated with ­ increases in economic activity. Finally, com- This analysis found that initial condi- plementary policies for trade and industry tions do not seem to be associated with the competitiveness, as well as better performance success of a corridor project in setting of the health sector, could help spread the ­ p reconditions for WEB (with “success” economic benefits and increase economic rated by international development organi- activity along the transport corridors. zations and experts). In contrast, increases in economic activity near the corridor proj- ect are significantly associated with initial CONCLUSION conditions, such as the country size, rug- This study conducted a survey of large gedness of the terrain, level of development, transport corridor projects supported by and geography (whether landlocked or the Asian Development Bank, the Japanese open to the coast). Specifically, the estima- International Cooperation Agency, and the tions suggest that increases in economic World Bank. It used the survey data to cata- activity near corridor projects tend to be log project characteristics and relate them to larger in smaller, less-developed countries success measures. To this end, the study per- with smoother terrain and direct access to formed multivariate meta-­ regressions link- the sea. ing project characteristics to the likelihood As expected, the project design and that a project would produce wider eco- characteristics do influence the success in ­ nomic benefits (WEB) as well as achieve ­ setting conditions for WEB. That is, some specific WEB (specifically, economic activ- project designs and characteristics are ity). The likelihood that a project will systematically perceived by the international ­ ­ produce WEB is measured through institu- development organizations as well-established tional ratings specific to the international preconditions for development success. For development organization, along with instance, international development organiza- expert assessments of project success. The tions are more likely to rate projects with a achievement of WEB is measured through well-­ developed theory of change as successful, spatially disaggregated data on nightlights perhaps because these projects have clearer and human settlement near and further designs and matching monitoring and evalua- away from the project location. tion frameworks that better demonstrate It can be revealing to compare the link the achievement of development objectives. from the project design to broad-based insti- Some project designs and characteristics seem tutional measures of success with the link to be more associated with positive effects on from the project design to changes in economic activity—such as national and selected measures of economic activity actu- international projects, as opposed to subna- ally achieved. This exercise can indicate tional and local projects. However, the effect whether the international development is not robust and significant across different organizations that design the project and e s t i m a t i o n s . I n t e r e s t i n g l y, f o r b o t h 92  The WEB o f T r a n s p o r t C o r r i d o r s i n S o u t h A s i a international development organization suc- around the corridors and thus enriches the cess ratings and economic activity, greater pri- evidence from the academic literature, which vate sector involvement in the operation and does not work with such detailed project management of transport infrastructure (for design characteristics. example, through public-private partnerships) The next chapter aims to inform policy seems to have been associated with negative makers by summarizing the growing body of effects in some cases. This tentative finding academic and policy literature on the impact could point to a challenge of obtaining the of large transport investments on WEB. expected benefits from the collaboration of Chapter 4 conducts a quantitative summary public and private sectors in development by using a large new data set of studies in dif- investments. ferent country contexts and meta-regressions. Implementing certain complementary It aims to integrate current knowledge and reforms before undertaking the corridor proj- point to knowledge gaps of which policy ect could make the project more successful in makers must be aware when appraising the setting the preconditions for WEB and actually design of proposed corridor projects. spurring greater economic activity (a proxy for Knowledge of the direct and indirect impacts, WEB). Using reform programs supported by the trade-offs they could produce, and the World Bank Development Policy Operations complementary policies that could help man- as proxies for complementary reforms in dif- age these trade-offs can all help shape the ferent sectors, the analysis found that reforms optimal policy design of future corridor promoting industrial and trade competitive- interventions. ness are associated with increases in both the perceived potential for WEB and increased economic activity near the corridor. NOTES If the corridor intervention package 1. For a review of the literature, see Bougna includes complementary reforms to increase et al. (forthcoming). openness to foreign trade, improve gover- 2. For a detailed discussion of complementary nance effectiveness, and boost industry and policies, see chapter 6. trade competitiveness to the levels of higher- 3. World Bank assistance to its clients can be performing developing countries, the host provided in the form of a Development Policy country could expect GDP to increase Operation (DPO). This could be a loan, grant, 9.3−12.4 percentage points near the new or or credit that rapidly disburses financing to upgraded corridor. help a borrower address actual or anticipated development financing requirements. The World Bank’s use of DPOs in a country is Summing up and moving to the determined in the context of a comprehensive next chapter Country Assistance Strategy (CAS) drawn up by the government and the World Bank. The This chapter took advantage of a unique new World Bank makes the funds available to the data set based on a survey of large transport country when the country maintains an infrastructure (corridor) projects supported ­ adequate macroeconomic policy framework by the Asian Development Bank (ADB), (as determined by the World Bank, with the Japan International Cooperation Agency inputs from assessments by the International Monetary Fund); the overall reform program (JICA), and the World Bank Group. It sum- is being implemented in a satisfactory man- marized the results of an extensive meta-­ ner; and the World Bank and the client have analysis aiming to identify which features completed a set of critical, mutually agreed- could boost a corridor project’s potential for on prior policy and institutional actions (prior generating WEB. Chapter 3 links the detailed actions). See http://siteresources​ .worldbank​ project characteristics to institutional success .org/PROJECTS​/­Resources/40940-12447326 ratings and measures of economic activity 25424/Q&Adplrev.pdf. C a n T r a n s p o r t C o r r i d o r P r o j e c t s P r o d u c e W i d e r Ec o n o m i c B e n e f i t s ?    93 4. Basic outcomes include transport access, trans- jobs. For instance, greater transport connec- port governance/management, road safety, and tivity helps spread diseases (Tatem, Rogers, increased competitiveness in the transport sec- and Hay 2011). Health sector reforms may tor. Intermediate outcomes include trade, tour- boost the preparedness and coping mecha- ism, and access to basic services such as health nisms of areas near the transport corridors. and education. Final outcomes (wider eco- nomic benefits, or WEB) include changes in economic growth, aggregate/­ spatial equality, REFERENCES social inclusion, resilience (such as food secu- Behrens, K., W. M. Brown, and T. Bougna Lonla. rity), and environmental quality (such as defor- 2016. “The World Is Not Yet Flat: Transport estation and carbon dioxide emissions). Costs Matter!” Working Paper 7862, World 5. For the World Bank, the internal rating is Bank, Washington, DC. the rating reported in the Implementation Bougna, T., M. Melecky, M. Roberts, and Y. Xu. Completion and Results report (ICR). The Forthcoming. “The Estimated Wider Economic independent evaluation rating is the assessment Benefits of Transport Corridors: A Critical by the Independent Evaluation Group (IEG). Review of the Literature.” Policy Research 6. The nightlights data measure the intensity of Paper, World Bank, Washington, DC. human-made light on a scale of 0 to 63 using Denizer, C., D. Kaufmann, and A. Kraay. 2013. satellite imagery. These annual data are “Good Countries or Good Projects? Macro and high resolution (they have a resolution of Micro Correlates of World Bank Project 0.86 square kilometer at the equator). The Performance.” Journal of Development data correct for cloud coverage and report the Economics 105: 288–302. average visible and stable lights. This study Feeny, S., and V. Vuong. 2017. “Explaining Aid uses the annual nightlights data that are com- Project and Program Success: Findings from parable across time and space. These data are Asian Development Bank Interventions.” available from 1992 to 2012. The Global World Development 90: 329–43. Human Settlement measures the intensity of Henderson, J. V., A. Storeygard, and D. N. Weil. human settlements through satellite on a scale 2012. “Measuring Economic Growth from from 0 to 1. These data are available for four Outer Space.” American Economic Review years: 1975, 1990, 2000, and 2014. 102 (2): 994−1028. 7. One hectare is equivalent to 10,000 square Mubila, M. M., C. L. Lufumpa, and S. Kayizzi- meters. Mugerwa. 2000. “A Statistical Analysis of 8. In addition, the regression analysis also finds Determinants of Project Success: Examples that landlocked-ness, the geographical scope from the African Development Bank.” of the project, and the type of corridor Economic Research Paper 56, African ­ intervention project alignment (greenfield, Development Bank, Abidjan, Côte d’Ivoire. upgrade, or rehabilitation) do not affect the Redding, S. J., and M. A. Turner. 2015. success of a project. “Transportation Costs and the Spatial 9. Most of the projects in the sample utiliz- Organization of Economic Activity.” Chapter 20 ing public-private partnerships invest in in Handbook of Regional and Urban Economics, roads. In these cases, the private sector has Vol. 5, edited by G. Duranton, J. Vernon ­ limited or no power to increase revenue Henderson, and W. C. Strange, 1339–98. through changes in the volume or compo- Amsterdam: Elsevier. sition of traffic. The efficiency added by Sequeria, S., O. Hartman, and C. Kunaka. 2014. the private sector comes from construction “Reviving Trade Routes: Evidence from the and maintenance, which produce hardly Maputo Corridor.” Discussion Paper 4, Sub- any WEB. Saharan Africa Transport Policy Program 10. Using Worldwide Governance Indicators (SSATP), Washington, DC. for 2015. Tatem, A. J., D. J. Rogers, and S. I. Hay. 2011. 11. This is perhaps because healthier people can “Global Transport Networks and Infectious better face risks and seize opportunities cre- Disease Spread.” https://www.ncbi.nlm.nih​ ated by better connectivity to markets and .gov/pmc/articles/PMC3145127/. Spotlight 1 Financing Priority Transport Corridors in South Asia T ransport corridors support trade flows and economic activity along certain routes. They may be domestic (such maintain, and operate the infrastructure but also to develop the supply chains along the routes. as India’s Golden Quadrangle, connecting Mumbai, Delhi, Kolkata, and Chennai) or WHO BENEFITS AND INVESTS IN cross-border/regional (such as the West THE CORRIDORS? Bengal Corridor, serving Bangladesh, the states of Sikkim and Assam in India, and Transport corridors can produce large socio- landlocked Nepal and Bhutan). economic benefits, but they also carry large The corridors involve transport links (rail, costs. Depending on the beneficiaries’ level of roads, and waterways on which passengers income, the operators’ ability to control and freight travel), nodes (interconnecting access, and the quality of services provided, transport and other services, such as ports and some of the costs can be recovered from users. logistics centers), and gateways (access roads, Other important economic benefits are less allowing traffic to enter or exit the corridor to direct—such as increased trade, commercial reach the hinterland) (World Bank 2005). development, and job opportunities along the In addition, they entail large infrastructure routes—and may be monetized only by the investments and multiple assets—some state, through general tax revenues. already completed, and others built from Ultimately, the cost of corridors will be cov- scratch, called greenfield projects. They also ered either by individual taxpayers, commer- involve different stakeholders, which some- cial taxpayers, or both. However, tax and tariff times have conflicting objectives. Because of revenues accrue over time, while project costs the magnitude, externalities, and monopolis- are paid at the preparation, development, and tic aspects of transport corridor investments, construction stages. Small projects can be paid as well as the need to coordinate across differ- for from current cash resources, but large ones ent stakeholders, governments are crucial to face a mismatch in the maturity of funding, the planning, controlling, and funding of which can be addressed by financing from the these projects. Given the fiscal constraints domestic or international financial system of the host or participating countries, the cor- (financial intermediation). This could involve ridors require considerable financing on a sovereign borrowing (from multilateral, commercial basis, not only to construct, b ilateral, or other financial institutions), ­ 95 96   THE WEB OF TR A NSPORT C ORRIDORS IN SOUTH A SI A TABLE S1.1  Who pays for transport corridors? Type of funding Individual taxpayers Commercial taxpayers Funding from Budget transfer. Investment undertaken Internal accruals. Commercial investment current revenue by the state is funded from current tax is funded from current tariff revenue. revenue. Taxpayers prefinance future Current taxpayers prefinance future infrastructure users. infrastructure users. Funding from State financing. Investment is financed Commercial financing. Investment is future revenue by sovereign borrowing or contingent financed by project equity and debt, which liabilities,a ultimately repaid from future tax are ultimately remunerated and serviced revenue. Future taxpayers pay for future from future tariff revenue. Future taxpayers infrastructure users. pay for future infrastructure users. Source: Corridor Study Team. a. Potential liabilities that may occur, depending on the outcome of an uncertain future event. or equity/debt financing on a commercial basis financial returns and produce cash flows. (commercial financing), where equity is remu- Arrangements for private sector participation nerated and project debt are serviced from tar- range from management contracts (with no iff revenues (see table S1.1). commercial investment), to leasing, build-­ operate-transfer (BOT) arrangements, conces- sions, and full private investment, among PROJECTS THAT CAN BE others. Projects in the most commercial sectors COMMERCIALLY FINANCED (those that allow full recovery of operating and Commercial finance combines three elements: investment cost) can be financed on a “user- commercial investors or lenders, equity or pays” basis. These projects obtain revenues debt finance on commercial terms, and com- from retail end users (such as fares paid by mercial infrastructure borrowers and other train p­ assengers or tolls paid by vehicle driv- obligors (those with contractual obligations, ers); from anchor users by way of off-take con- including payment obligations). Since the tracts, in which major sponsors guarantee a state plays such an extensive role in infra- minimum use of the infrastructure (with ship- structure, commercial finance often occurs in ping companies in the case of ports, or rolling- a hybrid rather than pure form. For example, stock operators or commodity shippers in the private lenders may lend to state utilities, and case of railway lines); or from dedicated taxes state-owned financial institutions may facili- paid by users (surcharges on railway freight or tate equity and debt financing to private utili- fuel taxes allocated to road funds). Projects ties on commercial or concessional (favorable) involving social infrastructure (such as public terms. This is especially true with transport schools or fire stations) that generate no fee corridors, given their socioeconomic exter- revenue, or p ­ rojects involving economic infra- nalities and coordination challenges. structure with low cost recovery that do not The concept of commercial finance does not generate sufficient tariffs (as is often the case fully overlap with private finance, in the sense for water supply) can be commercially financed that public sector utilities or financial institu- on a “government-pays” basis. Projects that tions may operate commercially, despite their create large economic benefits but face uncer- ownership by the state. tain demand can be financed based on state The extent and channels of cost recovery availability payments (a payment for perfor- directly affect the way investments are financed. mance made irrespective of demand)— Commercial investors will only become although these create long-term liabilities with involved in projects that can generate sufficient a fiscal impact. Financing Priority Transport Corridors in South Asia    97 Transactions related to South Asian trans- FIGURE S1.1  There are three main forms of infrastructure finance, port corridors that have been commercially plus hybrid financing financed in recent years include: Ports and airports. These projects are pre- Public Finance ferred by commercial investors and lenders, Sovereign (subnational) especially if the financing is backed by long- borrowing from donors/ term off-take actions (such as for ports), and commercial banks, access to foreign exchange revenues (such as for supplementing tax revenue international airports). Transactions financed in 2016 were predominantly in India, and included the Gangavaram Port in Visakhapatnam, a Structured SOE Public project in Andhra Pradesh, and the JSW Jaigarh Finance Finance Port in Ratnagiri, Maharashtra. Bridges, tunnels, and toll roads. Many of Corporate Finance Project Finance these transactions have been financed in India Debt issuance on balance Debt without recourse Structured sheet and direct equity and equity investment in (and to a lesser extent, Pakistan). In view of investment (supplementing Corporate project SPVs, repayment the mixed outcomes of past transactions, Finance internal cash accruals) relies expected project investors and lenders are now wary of traffic cash flow risks, especially on greenfield projects. Railways. A few transactions have been financed in India, but mostly in the form of Source: Corridor Study Team. annuities (where the debt service is secured by Note: SOE = state-owned enterprise; SPV = special-purpose vehicle. payments from Indian Railways) and a hand- ful of last mile projects incorporated in inter- national ports, with off-take from major agencies or state-owned enterprises (SOEs) industrial users. with sovereign guarantees. It is convenient for When user fees do not provide full cost large infrastructure projects, and is used by recovery and the state must bear part of the default to finance assets that do not produce investment, it tends to invest in basic infra- distinct cash flows (such as electrification, sig- structure (such as civil works, like ports), while naling, or gauge conversion, in the case of private financiers invest in superstructure and with railways), or projects for which there is logistics (like jetties and cranes for ports, roll- no private sector interest, due to their market ing stock or sidings for railways, and toll pla- risk or long payback horizons. The state con- zas and operations for toll roads). tribution to these projects—particularly in fragile, low-income, and lower-middle-income countries—may be supported by concessional POSSIBLE FINANCING FOR financing from multilateral and bilateral CORRIDOR ASSETS financial institutions (donors). Since many The three main ways in which infrastructure countries face challenges to their fiscal space can be financed by the domestic or interna- (their flexibility in spending choices), the tional financial system are public finance, cor- amounts available for public finance may porate finance, and project finance, as well as be limited by the need to keep the state’s debt their hybrid forms, as shown in figure S1.1. service obligations and overall indebtedness at Each mode is defined by the structure of its sustainable levels, and/or by limiting govern- debt financing (although equity is also ment-contingent liabilities resulting from required in the case of corporate and project sovereign guarantees. finance). Corporate finance supports investments Public finance involves borrowing by cen- by established corporate entities, which carry tral and local governments, or by state both the projects as well as associated equity 98   THE WEB OF TR A NSPORT C ORRIDORS IN SOUTH A SI A and loans on their balance sheets. Corporate project to its debt service obligation. Project investments may be financed by a mix of finance requires careful structuring of con- equity, quasi-equity, and/or debt issuance. tracts so that risks are well identified and In their most basic form, corporate finance allocated to the parties best placed to bear loans are plain (vanilla), with a final maturity them. A well-structured transaction with usually of not over five years. Lenders assess abundant, predictable cash flows acceptable the creditworthiness of borrowers based on to commercial lenders is said to be bankable past financial results and available collateral. (and investable, to investors). Due to its rela- Alternately, these loans can be structured tive complexity, project finance tends to be (through financial covenants, commercial uneconomic for deals that are less than undertakings, assignment of project con- US$50 million−US$100 million. tracts, escrow accounts and other limitations Hybrid financing forms include equity and/ on the use of future cash flows, and the like) or debt finance by a state utility (SOE finance). to allow for larger amounts, grace periods They are a mix between public and corporate during construction, and longer repayment finance, depending on the extent to which the maturities. Smaller loans can be bilateral borrower derives its credit strength from its (self-standing credit facilities separately own cash flows, or from state support. extended by each bank), while larger ones If financially viable, SOEs may attract private need to be syndicated—collectively extended sector equity investment (such as through a by a group of banks sharing the same loan stock exchange listing) as part of privatization documents and security. The key metrics that efforts. Hybrid project/corporate debt can be determine the obligors’ borrowing capacity used to finance expansions, with repayment are the debt/equity ratio (the traditional secured by a mix of existing and planned cash approach, more relevant for vanilla corporate flow. Borrowings by public sector entities loans) as well as the ratio of debt/cash flow backed by commercial receivables, such as from operations (a modern approach, espe- state-owned toll roads, can be viewed as struc- cially relevant for structured corporate tured public finance. finance). In general, corporate finance is more suitable for well-established commercial enti- FINANCING SOURCES FOR ties that make investments of a moderate size CORRIDOR PROJECTS relative to the balance sheet. Larger commercial projects are handled on Commercial infrastructure, including trans- a project finance basis through equity and port corridor projects, is financed by a mix of debt finance, allowing maturities up to 10−15 equity (usually 25 percent to 30 percent in the years, and even 20 years (depending on case of project finance, depending on the risk ­ market conditions and the transaction’s risk profile of the project, and 35 percent to profile). Equity investments in ad hoc project 50 percent in the case of corporate finance) companies (special-purpose vehicles, SPVs) and debt. are made by project developers, infrastruc- Traditionally, project developers provide ture private equity (PE) funds, and/or strate- the core of equity finance in greenfield proj- gic investment funds (SIFs). Project finance ects. However, in recent years, particularly in loans are extended to project SPVs with lim- India, they have become increasingly stretched ited (or no) recourse to the balance sheet of as their balance sheets have become highly the project sponsors; instead, loans are leveraged. Thus, project cost overruns had to expected to be repaid from a project’s future be funded by equity, since added debt was not cash flows. The key metric to determine a forthcoming. As a result, developers in India project’s borrowing capacity is its ­ projected have a limited appetite for further risk since debt service coverage ratio (DSCR), the ratio they have already scaled back their ambitions of operating cash flow generated by the substantially. Financing Priority Transport Corridors in South Asia    99 In recent years, PE funds have emerged as a The first challenge relates to the nature of new source of equity for infrastructure proj- the underlying transactions. Corridor projects ects. For example, in India, over 15 PE infra- generate valuable socioeconomic benefits, but structure funds operate, with total assets under not necessarily high financial returns that can management of US$1.7 billion. The top three be captured by commercial investors and lend- funds are Reliance Diver Power, UTI ers. Greenfield transport corridor projects may Infrastructure, and ICIC Prudential have risk profiles (particularly related to the Infrastructure Fund, representing 41.8 percent acquisition of rights of way, construction, of the total infrastructure PE industry. coordination with other infrastructure, and Historically, these PE funds rarely took traffic volume) that may require public sup- development risks. However, given limited port through various instruments, including project pipelines, several infrastructure PE concessional equity and partial guarantees. In funds have shifted focus earlier in the project India, some leading domestic banks have lent life cycle and increasingly are entering at the considerable amounts to toll roads and other development stage. While some funds build transport projects that have not performed up internal development teams, they more often to expectations. In recent years, this has meant invest in one or more portfolio companies that large equity losses for investors, as well as do project development and investment man- large volumes of impaired loans for commer- agement work on their behalf. Portfolio com- cial lenders, which have dampened the appetite panies are paid an annual retainer for their of commercial financiers for such projects. services and are entitled to a share of carried Other challenges stem from the character- interest. istics of the host countries’ banking sectors, Debt financing can be sourced from the especially related to asset-liability manage- domestic financial system or from interna- ment (ALM). In particular, infrastructure tional financial institutions. With respect to loans require very long maturities, but bank domestic debt, banks remain the major source resources consist mainly of short-term depos- for commercial infrastructure projects in its. Thus, infrastructure loans expose banks South Asia. Although the specifics of pruden- to maturity mismatches. Another issue relates tial regulations differ from country to country, to the structure of interest rates. Banks typi- banks may generally lend up to 20 percent of cally lend on the basis of variable rates, rather their regulatory capital to a nonstate borrower. than fixed ones, since the bulk of their So far, in the largest economies in South Asia resources (short-term deposits) involve vari- (India, Pakistan, and Bangladesh, and to some able rates. Lacking a sufficiently deep market extent Sri Lanka), this credit limit has been to hedge local currency interest rates, infra- sufficient for the domestic banking sector to structure borrowers are exposed to the risk finance all but the largest commercial infra- that their debt service obligations may structure projects. In practice, banks may become unsustainably high if market rates decide to lend less than the amount allowed by rise in the future. In other cases, borrowers the prudential limit, due to other consider- would rather obtain loans in foreign curren- ations. But these limits do not apply to loans cies to avoid paying the higher interest rates guaranteed by the central government. on local currency loans. Even if this does not However, in practice, the financing of cor- create a currency mismatch for domestic ridor projects faces steep hurdles—some spe- banks (to the extent that they have access to cific to the transaction, some related to the matching foreign exchange resources), it characteristics of the financial sector, and some exposes borrowers to a high foreign exchange contingent on the broader macroeconomic debt service obligation in case of devalua- environment. These challenges translate into tions. While the impact of asset-liability man- growing c ­ onstraints on traditional bank lend- agement reaches beyond infrastructure and ing to infrastructure projects in the region. affects the financing of other types of 100   THE WEB OF TR A NSPORT C ORRIDORS IN SOUTH A SI A investments, it is strongly preferred for infra- countries except for India, which is rated structure investments, including transport BBB- by Fitch and Standard & Poor’s), inter- corridor projects, given the large amounts national banks are reluctant to extend long- and long maturities. term infrastructure loans—except for Cross-border commercial debt financing transactions for which one of their core cli- for projects related to the South Asia corri- ents is a project sponsor and for which they dors has mostly come from two sources: are extensively covered against country risk the commercial arms of multilateral through an export credit provided by a mem- donors (such as the World Bank Group’s ber country of the OECD (Organisation for International Finance Corporation, IFC), Economic Co-operation and Development) and state policy banks (such as China Exim with a high credit rating or a guarantee from and China Development Bank). the World Bank Group’s Multilateral IFC and similar commercial arms of inter- International Guarantee Agency (MIGA), or national donors are well suited to finance where debt repayment is secured by offshore infrastructure transactions due to their strong revenue, as with international airports or structuring capability and their focus on commodity exports. emerging markets. China has earmarked very Besides bank lending, corporate bonds large resources to finance infrastructure proj- are another major source of infrastructure ects in South Asia under the One Belt One finance. India has the deepest financial Road Program. In particular, with respect to investor base and largest corporate bond Pakistan, China has committed over US$45 volume in South Asia. Bonds subscribed to billion to construct infrastructure projects by an investor such as a life insurance under the China-Pakistan Economic Corridor company have maturities extending up to (CPEC) Agreement, with funding from 30 years for issuers with the strongest matu- Chinese banks and corporations focused on rity profile. Most listed corporate bonds energy (75 percent of the total committed) relate to infrastructure, either directly in the and roads/ports/railways (25 percent of the form of bonds issued by infrastructure utili- total). Usually, the loans are based on a sov- ties, or indirectly in the form of bonds issued ereign guarantee (for example, in the case of by specialized financial institutions, such as province-level transport projects); on a India Railway Finance Corporation, Ltd repayment guarantee for project loans from (IRFC). domestic banks with an acceptable credit rat- In other South Asian countries, the corpo- ing guaranteeing the repayment of project rate bond market is much more limited, and loans; or on a repayment guarantee from a in some cases, it is nonexistent (as in Chinese sponsor that holds majority owner- Afghanistan). However, in countries such as ship in the infrastructure project. To date, Pakistan, there is significant untapped poten- few such loans have been extended on a project tial. Bond issuers could include toll roads and finance basis. other brownfield (existing or upgraded) com- In recent years, international corporate mercial infrastructure assets suitable for and investment banks have had relatively lim- Islamic issuances (equity-based certificates of ited involvement in financing South Asia’s investment, or sukūk). Currently, several infra- transport corridors and related infrastructure. structure utilities are listed on the Pakistan International banks, in principle, have a Stock Exchange (PSX), including Pakistan strong comparative advantage in financing International Container Terminal. infrastructure due to their very large size, and Beyond corporate bonds, there is also a their competence and long-standing leader- global market for project bonds. However, ship in global project finance. But with coun- in practice, such transactions mostly occur tries that have sub-investment-grade credit in investment-grade countries and relate to ratings (currently the case for all South Asian the refinancing of transactions once a Financing Priority Transport Corridors in South Asia    101 project is complete and has a satisfactory are too uncertain to attract private sector track record in terms of operations and cash investors. Such schemes exist in both devel- flow guarantees. But market interest appetite oped and emerging markets, and are typically is very limited for international project incorporated in the host country’s PPP frame- bonds in non-investment−grade countries, work. Examples are found in Chile, India, and even more so for greenfield transport Mexico, Peru, India, and South Africa. These projects. schemes can take different forms, such as via- In recent years, private debt funds have bility gap funds (targeting priority projects emerged as a new source of long-term whose economic returns are good but do not finance for infrastructure projects globally. attract private sector investors due to insuffi- New infrastructure debt funds raised US$34 cient financial returns) and availability pay- billion globally from 2013 to 2016. In India, ment schemes that support projects exposed to the India Infrastructure Debt Fund (IIDF), demand risk, such as transport corridor proj- sponsored by commercial banks and non- ects, for which operating cash flows cannot be bank financial corporations, acts as a vehicle reliably predicted. for refinancing the existing debt of infra- A second type of instrument involves structure SPVs, thus providing headroom for establishing a guarantee fund for the payment new lending by commercial banks, including obligations and other key contractual under- for infrastructure. takings to the project company and related project financing, where the state counter- parts in major contracts are not strong COMMERCIAL FINANCING FOR enough to support commercial financing. An TRANSPORT CORRIDOR PROJECTS example is the Indonesia Infrastructure Beyond its necessary role as a direct investor, Guarantee Fund (IIGF). A comparable the state also has a key role in facilitating scheme is the African Trade Insurance investment and financing from the private sec- Agency, which was established in 2001 and is tor (or the commercial public sector) for prior- headquartered in Nairobi, which covers ity corridor projects. Broadly, state policies infrastructure investors and lenders against a need to provide an enabling environment, variety of risks, including possible default by including macroeconomic stability (in terms of African member states. interest rates and sovereign credit rating), via- Other interventions promote financial ble infrastructure tariffs, good governance, intermediation for viable infrastructure proj- professional management of state utilities, ects. State interventions in the financial sector transparent public-private partnership (PPP) may follow a wholesale or retail route. procurement and contract enforcement, credi- Wholesale development finance schemes foster tors’ rights, capital market regulations, and lending from domestic financial institutions by equitable tax regimes. providing them with long-term liquidity, shar- In addition, to attract private investors to ing selected financial hedging risks, and shar- infrastructure projects, including transport ing credit risks on the financing. Wholesale corridor projects, authorities may consider schemes do not require the creation of a new specific state interventions through various institution. instruments. Wholesale financing schemes. Instruments The first type of instrument aims to designed to address market gaps related to strengthen the project structure to make it long-term liquidity include mirror (matching) suitable for commercial investors and lenders. refinancing, where domestic banks (primary Such instruments include financial viability or retail lenders) may obtain long-term funds support (FVS) schemes, which support priority from a state agency (acting in a secondary or projects that are economically viable but wholesale capacity) that matches the maturi- whose financial returns are not high enough or ties of their long-term loans to eligible 102   THE WEB OF TR A NSPORT C ORRIDORS IN SOUTH A SI A infrastructure obligors (borrowers). One Retail financing schemes. In contrast to example is the IPFF project in Bangladesh, secondary or wholesale instruments, the pri- which is managed by the central bank on mary or retail approach involves direct behalf of the Ministry of Finance and is backed financing of commercial infrastructure obli- by World Bank financing. Other options gors. Direct intervention typically involves include take-out facilities, such as those offered establishing a new financial institution dedi- by the U.S. Export-Import Bank, which gives cated to infrastructure finance. Countries banks the option to sell (at a predetermined where the World Bank and IFC have sup- price) eligible long-term loans back to a state ported primary development finance institu- agency—thereby freeing up their balance tions include India (with IIFCL), Colombia, sheets so they can finance new projects. and Indonesia, and an institution is planned Liquidity schemes can also share maturities, for Pakistan. While these institutions can help where banks finance earlier maturities, while a kick-start market development, they may also state agency funds the longest ones. face pitfalls such as competition with the pri- Besides long-term liquidity, there is also vate sector, cronyism (where financial support scope for state agencies to act as market mak- is exchanged for some privilege or benefit), ers, along with pension funds or other com- political capture, and a lack of financial disci- mercial financial institutions. They can pline or of development impact. To avoid provide hedging that mitigates the interest- such pitfalls, development finance institutions rate mismatch carried by primary lenders, need to adopt good international practices, extending fixed-rate, long-term loans to infra- notably a clear mandate that focuses on structure obligors. In practice, state agencies well-identified market gaps, as well as strong ­ carry exchange rate fluctuation risk when management, governance, and supervision— they on-lend in local currency the foreign cur- ideally, with majority private sector owner- rency resources they obtain from international ship. Because the resources of these financial institutions. Exposure to either inter- institutions are limited compared to the scale est rate or exchange rate fluctuation risk of infrastructure needed by host countries, needs to be carefully monitored and managed, their impact should be measured in terms of given a country’s macroeconomic and fiscal the total financing they facilitate, rather than situation. the amount they provide or the size of their Another type of wholesale instrument balance sheet. focuses on sharing credit risk. Under such In recent years, a number of countries schemes, the risk participation of the second- have established strategic investment funds ary financial institution can be pro-rata, as in (SIFs) to serve as anchor investors for inter- the case of partial credit guarantees, or national and domestic investors in projects in unfunded risk participations. Alternately, the key sectors of the domestic economy. SIFs are risk sharing may be in the form of long-term created as investment funds or corporations contingent facilities, such as those pioneered and are sponsored and/or fully or partly by the European Investment Bank (EIB) as ­ capitalized by a government, by several gov- part of its Project Bond Initiative, which ernments, or by government-owned global or supports the Trans–European Network– regional finance institutions. They invest for Transport (TEN–T) (see chapter 5). Under financial as well as economic returns, accord- this scheme, a secondary financial institution ing to a double-bottom-line objective. SIFs extends a contingent credit line, which can be aim to attract private capital to invest in key drawn on if the cash flows generated by the economic sectors; operate as expert investors project are not sufficient to ensure senior on behalf of their sponsors; provide long- bond debt service or to cover construction term “patient” capital (from those who are costs overruns. willing to invest with no expectation of Financing Priority Transport Corridors in South Asia    103 turning a quick profit), primarily as equity, FINANCING CROSS-BORDER AND and may also invest in quasi-equity or debt; TRANS-REGIONAL TRANSPORT and are created as investment funds or cor- CORRIDORS porations (see Havard, Noel, and Tordo Cross-border infrastructure projects—such as 2016). Over the past 15 years, at least international canals; regional ports; and inte- 30 SIFs have been were created around the grated water management systems, including world. Another 12 are planned. Examples of waterways, railway lines, roads, and associ- existing SIFs are Bahrain’s Mumtalakat ated border and customs logistics—are impor- (2006), Italy’s Strategic Investment Fund tant for developing regional and transregional (2011), Kazakhstan’s Baiterek (2013), trade, and can wield immense benefits for ben- Mexico’s Macquarie Mexico Infrastructure eficiary countries. But these projects face fund, Morocco’s Ithmar Capital, Senegal’s financing challenges that are even more daunt- Fonds Sénégalais d’Investissements ing than domestic transport corridor projects. Stratégiques (FONSIS), and the Africa Much of the international experience in Renewable Energy Fund (2014) (see Havard, financing cross-border infrastructure relates Noel, and Tordo 2016). In South Asia, India to either telecommunications (submarine created the National Investment and cables, fiber-optic backbones) or energy trade, Infrastructure Fund (NIIF) as an SIF in 2015, such as power generation and transmission with US$3 billion of seed capital from the projects in Latin America, the Mekong River government of India. Delta subregion, and Sub-Saharan Africa; and The structures of SIFs vary, from the pri- gas pipelines in Eastern Europe, Western and vate management of public capital through Central Asia, the ASEAN (Association of hybrid funds with private and public share- Southeast Asian Nations) region, and the holding, to fully state-owned direct invest- Southern Cone of Latin America. Examples ment funds. Private management of public in South Asia include hydropower plants in capital occurs when the government invests in Bhutan and Nepal, and the associated trans- a private fund on terms that reflect policy pri- mission lines for export to India. orities, or when a public entity shares risk as a The two most prominent historical exam- limited partner in a hybrid fund. In this ples of cross-border infrastructure projects model, investment decisions are made inde- are related to greenfield transport corridors: pendently by the private sector general part- the Suez Canal (opened in 1869) and the ner that manages the fund, or by an Panama Canal (opened in 1914). The Suez independent investment committee that may Canal was undertaken on a commercial basis include government representatives. The by a concession company responsible for its overall investment policy is set by the fund’s construction and operation. This company board, which is usually controlled by limited listed its shares on the Paris Bourse and raised partners. The fund manager and general part- debt financing through a bond issue (which ner may be required to put up some share of ultimately ended in default). A similar the total capital. In funds that are fully owned approach was attempted for the Panama and/or operated by the government, market Canal, but failed. Eventually, construction of validation may come from limiting the own- the Panama Canal was completed by the U.S. ership share in each investment, and limiting Army Corps of Engineers. the SIF’s investments to minority participa- Cross-border transport corridor projects tion of a certain size. For these funds, man- have involved three types of investment and agement is frequently provided by a financing vehicles: government-owned entity operating at arm’s National public finance. Each state is length from the government (see Havard, responsible for investment in its territory. Noel, and Tordo 2016). 104   THE WEB OF TRANSPORT CORRIDORS IN SOUTH ASIA This method is the simplest and most Private joint ventures, which, given their ­ common. However, it may lead to subopti- very large size, tend to be structured along the mal outcomes (projects with a strong lines of project finance (as was the case for economic rationale that are not undertaken ­ Eurotunnel), rather than corporate finance. or are postponed, ­ particularly where costs Private joint ventures in transport corridors and benefits do not arise proportionally in may be supported by a regional SIF sponsored the respective states—as may be the case by one or more governments in the region, for the Canal Seine Nord, a 107-kilometer with the goal of being the anchor investor for waterway project intended to connect regional or international investors in infra- France, Belgium, and the Netherlands). structure. The most prominent example of a Public joint ventures involve the public sec- regional corridor SIF is the Silk Road Fund tors of the respective countries, as illustrated (SRF), which was established by the by the Oresund Link between Sweden and Government of China in 2014 with US$40 bil- Denmark (see box S1.1). lion in capital (see box S1.2). BOX S1.1  Financing the Oresund link between Denmark and Sweden Oresund (Øresund in Danish, Öresund in expected to be repaid within 34 years from the Swedish) is a 16-kilometer (km) fixed rail-road date when the project was begun. The project link connecting Copenhagen in Denmark and company is exposed to a moderate level of cur- Malmö in Sweden. It also serves the Copen- rency fluctuation risk (loans are denominated hagen airport, which is halfway between the in three currencies—euros, Swedish kronor, two cities. The project was constructed from and Danish kroner) and interest rate fluctua- 1995 to 2000 and includes an 8-km two-deck tion risk (parts of the loans are on a floating bridge, an artificial island, and a 4-km tun- rate basis, and the fixed-rate bonds may vary nel, at a cost of approximately €4 billion. The at the time they are retired for refinancing). infrastructure is owned and operated by a The project company dynamically hedges consortium formed by the Danish and Swed- these financial market risks to optimize the ish states. The two national governments, risk/cost profile of its financing. along with respective local governments, sepa- The solution of commercial financing rately financed the ancillary infrastructure guaranteed by the two states worked well for and access roads in their countries. Denmark and Sweden, given their close rela- To simplify the financing and reduce its tions and low likelihood of default. But it cost, the project was financed by loans and could require adjustments in countries with bond issues guaranteed by the two govern- lower credit ratings and a higher likelihood of ments through the Swedish National Debt default, commanding different loan prices, as Office and Denmark’s National Bank, both could be the case for transport projects across rated AAA (the highest possible sovereign neighboring South Asian countries. credit rating). The financing was entirely ser- viced from the proceeds received from auto- Sources : Øresundsbron website (https://w w w​ mobile tolls and railway access charges and is oresundsbron.com/en/info/finance); Painvin 2009. .­ Financing Priority Transport Corridors in South Asia    105 BOX S1.2  The Silk Road Fund: A regional transport corridor strategic investment fund As part of its One Belt One Road Initiative and financial cooperation in countries/regions (a strategic development program designed involved in the One Belt One Road Initiative to to enhance the links and cooperation among ensure medium- and long-term financial sus- Eurasian countries), China established the tainability and reasonable returns on invest- Silk Road Fund Co., Ltd. (SRF) in Beijing ment. SRF invests in equity, debt, and other in December 2014. The initial capitalization funds. It can work with international develop- was US$40 billion from the State Administra- ment organizations, along with domestic and tion of Foreign Exchange, China Investment overseas financial institutions, to jointly set up Corporate, China Development Bank, and funds, manage assets entrusted to it, and com- Export-Import Bank of China. The aim is to mission others to invest. promote the common development and pros- SRF was established as an investment fund perity of China and other countries/regions with complete corporate governance, which involved in the One Belt and One Road initia- created a board of directors, a board of super- tive. Besides financial returns, SRF focuses on visors, and a management term according to environmental and sustainable development the Company Law of the People’s Republic of issues, and actively works to address its social China. It has also recruited high-caliber pro- responsibilities. fessionals from different backgrounds and SRF is a medium- to long-term develop- disciplines, and set up an effective, efficient ment and investment fund. Through various corporate governance mechanism. forms of investment and financing (mainly equity investment), SRF is dedicated to sup- Source : Silk Road Fund website (http://www​ porting infrastructure, resource and energy . ­s ilkroadfund.com.cn/enweb/23775/23767/index​ development, industrial capacity cooperation, .html). REFERENCES Painvin, N. 2009. Large Projects, Giant Risks? Lessons Learned: Suez Canal to Boston’s Big Havard, H., M. Noel, and S. Tordo. 2016. Dig. Fitch Rating Transportation Global Special “Strategic Investment Funds: Opportunities Report (May 18). and Challenges.” Policy Research Working World Bank. 2005. “Best Practices in Corridor Paper 7851, World Bank, Washington, DC. Management.” World Bank, Washington, DC. Spotlight 2 Private Investment in Corridor Infrastructure P rivate sector sponsors vary in their involvement in developing transport corridors. Their role can be as minimal as service (a debt service coverage ratio equal to or greater than 1.5 is frequently required). 2. The private sponsors have the expertise helping plan these projects, or it can extend and the financial ability to implement the through the financing and/or operations of project and provide lenders with an ade- the infrastructure and services. The most quate works’ completion guarantee. successful schemes are usually developed as 3. The host governments can regulate private part of a public master planning exercise sponsors’ operations and, more important, because the private sector cannot be expected can provide lenders with a creditworthy to address issues surrounding the provision early termination guarantee as well as an of public goods. Private sponsors’ involve- appropriate security covering currency ment cannot be understood without examin- convertibility and an offshore debt repay- ing the views of the lenders and equity ment structure. investors that make their participation pos- 4. Lenders can secure political risk guaran- sible. Both types of participants usually have tees that are appropriate for the project’s similar views when analyzing a project’s risk profile. risks. However, their opinions and interests From the sponsors’ perspective, the cor- diverge about a project’s upside potential. ridor projects’ requirements are the same as When returns on equity are higher than those for the first and third requirements returns on debt, equity investors benefit, for lenders. However, sponsors’ require- while lenders do not benefit (that is, higher ments differ and/or are contradictory profitability does not translate into higher regarding the size and duration of the guar- debt repayments). antees they need to offer their debt provid- From the lenders’ perspective, there are ers, as well as the public service obligations, four basic requirements to support a corridor including those that imply the added project: upfront investment they need to provide to local communities. 1. The level of projected cash flows under a Private investment projects usually create base-case scenario will cover operations and a special-purpose vehicle (SPV) that is maintenance costs, concession fees, and debt responsible for developing, managing the 107 108   THE WEB OF TR A NSPORT C ORRIDORS IN SOUTH A SI A finances of, constructing, and operating a greenfield projects—smaller projects can also project. The SPV, as the holder of the debt, be structured in a way that allows them to caps the exposure of private sponsors to their raise commercial debt through public-private equity contribution if, and when, it is linked partnerships. In these cases, a project’s suc- to a nonrecourse type of project structure. If cess rests on the sponsors’ ability to design a project is not performing and the debt is shared-use regimes (among themselves) to not being served, the lenders’ recourse to pri- maximize revenues from multiple clients, vate sponsors is limited to the amount of without exposing them to local authorities’ equity invested in the SPV. accusations about perceived or real abuses When authorities seek private sector par- with respect to market access and/or price. ticipation to develop transport corridors, This is particularly true when uncertainties the plans must consider various bankability exist about the interest and credit quality of factors. Thus, public–private sector dia- prospective clients at the time a project is logue on risk allocation/mitigation must designed. Sponsors must then ensure that occur before project planning commences lenders and regulatory authorities have given and private investors’ participation is them enough contractual and oversight flexi- secured. This dialogue is essential to deter- bility to adapt to changing demand. While mine whether private sponsors will assume lenders may accept that sponsors provide public service obligations, which can access to new and unidentified clients (at the include enlarging their investments to time of financial closing), they will try to ensure wider benefits to local communities. maintain the right to block sponsors’ deci- Dialogue is also needed to clarify if they sions to expand their client pool if this weak- will scale up their investments or make part ens their debt service capacity. Indeed, sharing of their infrastructure available to other end infrastructure investments across multiple users to motivate them to contribute to the sponsors and/or multiple clients raises lend- project. ers’ perception of a project’s risks. Thus, this type of public-private partnership (PPP) structure makes sense to private sponsors CO-INVESTING IN CORRIDOR when their participation/investment comple- INFRASTRUCTURE: ATTRACTING ments investments in a corridor’s backbone DEBT AND EQUITY PROVIDERS infrastructure made by other private or pub- Before investing in corridor infrastructure, lic investors. private sponsors must secure funding by attracting more equity holders and/or mobi- PRIVATE INVESTMENT AND THE lizing debt from lenders. Their ability to do so BUSINESS ENVIRONMENT will be determined by the nature and/or degree of the market’s volatility. That is, will Corridor projects are often structured the infrastructure be competitive or noncom- as PPPs, in which regulations are clear, petitive; promoted by one or more products; that can attract private sector participation. serve one or more clients; and /or rely on an However, private equity investment in off-take agreement?1 The key to bankability infrastructure depends particularly on the in this case is the creditworthiness of the proj- private equity holders being offered attrac- ect sponsors and the producer’s competitive- tive concession terms. This is critical since ness (both in cost and value) in relevant the legal and regulatory risks can easily markets (specifically, if its price is among exceed the tolerance of project sponsors those in the lowest quartile). and/or lenders due to potential losses from While most lenders are prepared to assume government actions and/or exposure to the risk on large projects that rests on a single ­ d iscriminatory practices regarding equal anchor client—especially when dealing with access. P r i v a t e I n v e s t m e n t i n C o r r i d o r I n f r a s t r u c t u r e    109 Laws and regulations that foster corridor country and is promoted by the government’s infrastructure PPPs cover the following negotiating team. This contractual approach is aspects, among others: easier to implement in common law systems or where no PPP regulations exist. The contract •  Responsibilities shared among public and must be legally acceptable and have a clear private parties; precedent over any conflicting regulations. •  Contractual guidelines for PPPs and the Where laws are unclear, investors can con- scope of infrastructure concession sider the process (to define a legal regime contracts; through a contract) too long and risky. •  Contractual arrangements to accommo- In such cases, a complete review and/or mod- date multiple private parties; ernization of PPP laws might be warranted. •  Status and ownership of assets—including Institutional framework. The second set of when the PPP ends; issues regards the institutional framework. •  Securities; Establishing a PPP requires interacting with •  Taxes and customs, setting and revising of various public authorities responsible for fees, royalties, and tariffs; finances, sector policies, labor and social •  Service and performance obligations, espe- security, and safety and the environment, as cially regarding public services; well as other authorities whose jurisdiction •  Rules on terminating PPPs and dispute relates to the PPP. resolution; To ensure predictability and stability, the •  Authorizations, licenses, and rules regarding private sector seeks to deal with as few actors competition; as possible, and it assumes that the public •  Rules regarding labor, safety, and environ- authority has already consulted with other mental and social commitments; and public institutions. •  The nature and role of institutions In countries where institutions are unstable, involved at the various stages of the PPP private sector stakeholders are concerned that (inception, conclusion/negotiations, and new institutions may appear and claim the supervision). right to intervene under the PPP. To avoid this risk, the host governments should early on For investors (sponsors and lenders), these identify all public institutions that need to be issues can be considered either “systemic” consulted. Where this is ambiguous, they need (pertaining to all corridor PPPs in the jurisdic- to clarify each party’s role under the PPP. tion) and/or “project specific.” Enforcement of laws and contracts. The A transparent legal framework is often a third set of issues pertains to how well laws prerequisite for a PPP. Investors require clear, and contracts are enforced. This requires stable, and secure laws and institutions that appropriate laws, along with a reliable, inde- support the proposed operation. These pendent, and functioning judiciary. Often, involve three sets of issues. because private investors do not have enough Host country laws. The first set of issues confidence in local jurisdictions, PPP contract pertains to the host country’s laws, particu- disputes are submitted to arbitration. Thus, it larly those related to corporations; securi- is important that the host countries see arbi- ties; foreign investment and taxes; sector and tration as a valid method to resolve disputes PPP regulations; lease and real estate rights; and will enforce it. procurement laws; and labor, safety, and environmental regulations. TWO CASE STUDIES The absence of clear laws in any of these areas can represent a significant risk. This risk Many lessons emerge from attempts to can be mitigated through a contractual involve the private sector in developing corri- approach that is strongly supported by the host dor infrastructures. Interactions among the 110   THE WEB OF TR A NSPORT C ORRIDORS IN SOUTH A SI A different stakeholders are complex, and inter- The financing plan consisted of US$32 ests are diverse in projects, both with and ­ million in equity, senior debt of US$27 mil- without an anchor economic activity (where lion, and subordinated debt of US$8.5 mil- the major function is served by the infrastruc- lion. Senior debt consisted of two term ture). Two cases are considered next. loans: a US$14 million 12-year term loan The Maputo Corridor in Mozambique. from the Standard Corporate and This effort was part of a government strategy Merchant Bank, which included political to launch PPP programs to develop transport and partial commercial guarantees from infrastructure that had deteriorated during a the Swedish International Development period of low economic development and civil Agency; and a US$13 million, 10-year loan war (see box 3.1). The Maputo Corridor from the Development Bank of South promised the potential to contribute to Africa and the Netherlands Development domestic and regional economic development. Finance Company. Subordinated debt con- Since it ended at the Port of Maputo, the sec- sisted of a 10-year loan from FinnFund ond-largest deepwater port in Mozambique, and the Nordic Development Fund. this made it a natural gateway to export goods Because the port was developed simulta- from Mozambique’s neighbors and domestic neously with the railway, the port concession mines. Mining was considered an anchor lasted nearly two years, and was delayed activity for the corridor, which provided a until an operator chosen for the rail conces- critical mass for the railway and port. sion. Although the process collapsed in The government planned the PPPs to reha- 2005, the rail PPP relaunched in 2006, when bilitate the port and restore and develop the Mozambique Ports and Railways and railway to connect the port with domestic and Spoornet, a South African state-owned foreign customers. Although the two were enterprise, signed a cooperation agreement. structured and organized as separate projects, The initial rail PPP failure did not signifi- they were to be developed simultaneously. cantly affect the port. Instead, miners used toll The PPP’s structure and financing consisted roads and heavy trucks to deliver products to of the public sector, represented by the govern- the port. Volume increased to 11.8 million ment (49 percent of the concession)—includ- tons per year in 2011, and reached nearly full ing Mozambique Ports and Railways (which capacity in the following years. has 33 percent of the concession)—and by a The Maputo Port PPP is an example of a private group with interests in the region, rep- multiclient/multimodal project designed on resented by Mersey Docks Group, Skansa, an unbundled basis, where the railway (a key and Liscont (51 percent), which united in a part of the infrastructure) was designed and joint venture to form the concessionaire. The executed separately. It ultimately revealed the Maputo Port Development Co. (MPDC) problems attached to unbundling corridor agreed to finance, develop, and restore the infrastructure that shared an anchor eco- port, in return for receiving the right to oper- nomic activity (in this case, mining). Because ate it for 15 years in a classic build-operate- different pieces of the infrastructure played transfer (BOT) structure. Fifteen years were integral roles in the entire supply chain, the added to the concession in 2010, with an unbundled organization and execution of option for another 10 years after 2033. each part added complexity and time, and The concession included fixed annual pay- possibly had negative impacts. ments of US$5 million indexed to the U.S. This example also shows the role inter- Consumer Price Index. MPDC was required national finance institutions and develop- to pay the Public Port Authority 10 percent of ment finance agencies can play in bringing its gross income during the first five years, mining-related infrastructure PPPs to mar- 12.5 percent from years 6 to 10, and ket by addressing funding gaps. Indeed, 15.5 ­percent from years 11 to 15. without the institutions’ funding and It was expected to take three years to guarantees, commercial lenders may not restore the port, and cost US$67.5 million. have considered the project bankable. P r i v a t e I n v e s t m e n t i n C o r r i d o r I n f r a s t r u c t u r e    111 Without credit financing, it is unlikely that indigenous property titles, and for resolving private sector sponsors would have been environmental concerns. It was projected to interested. add b ­ illions to the GDP of the federal and The Alice Springs–Darwin railway project regional governments and create many jobs. in Australia. This project aimed to provide a However, it faced various economic issues. new transport corridor linking the northern There were concerns about whether it could and southern coasts, offering an alternate get shippers of lower value and/or less time- route for goods from Australia’s industrial sensitive cargoes to switch to the faster but center to Asia, and improved passenger service costlier rail alternative. Also, it was thought leading to more tourism. In this case, the proj- that even without a secured customer arrange- ect did not have an anchor activity. ment, it could attract domestic and interna- In 1995, the Northern Territory and the tional freight with mixed loads. However, the state of South Australia signed a memorandum banks’ financing model suggested that the to establish a joint approach to create a north- project needed to capture approximately south continental railway. By 1996, the 45 percent of the market share to break even, regional parliaments had created the and repay its senior debt within 12 years of AustralAsia Railway Corporation (AARC), commencing operations. which was a state-owned enterprise designed Unfortunately, this rail project did not attract to award a build-own-operate-transfer enough customers or make an operating profit (BOOT) concession and hold the titles to the during its first three years. By October 2006, land for railway construction. The Asia Pacific APTC could not meet the debt service on the Transport Consortium (APTC)—which was senior debt. Thus, in December 2006, APTC led by Kellogg Brown & Boot and included the entered into a nonaction agreement with senior John Holland Group, Barclay Mowlem, debt holders to restructure the business or Macmahon Holdings, and the Australian ­ refinance the senior loans. In December 2010, Railroad Group—was awarded the concession the assets and business were sold to the opera- in June 1999. The concession involved the tor-contractor, Genesee & Wyoming Inc. design, construction, and operation of the rail- At this time, APTC had liabilities of nearly way for 50 years. Construction began in July $A900 ­ million, a ratio of total debt to capital of 2001 and was completed in September 2003. 256 ­ percent, and interest coverage of 0.3. Rail traffic started in 2004. The Alice Springs–Darwin rail PPP illus- The total cost of this rail line was $A1.32 trates the level of complexity and risk, and rel- billion. APTC provided $A842 million, and ative lack of clarity, of multiuser/multimodal AARC provided $A478 million. The financing projects. Although the project had many bene- plan consisted of equity of $A238 million, fits, and was completed on time and almost on senior debt of $A491 million, subordinated budget, it failed because revenue projections debt of $A112 million, and government fund- were significantly overestimated. ing of $A478 million. Senior debt consisted of a $A150 million, five-year bullet loan; 2 NOTES a $A261 million, 12-year loan; and a $A80 mil- lion, 12-year rolling stock loan. Subordinated 1. An off-take agreement serves as a guarantee to debt included a tier-one m ­ ezzanine loan of the lender that the client has secured the sale of $A86 million and a tier-two mezzanine loan a specified amount of the goods/services to be of $A26 million. Finally, government funding produced provided by an anchor client. The simplest case is when a corridor serves a consisted of a $A50 million loan and single customer that makes a single product $A428 million in works’ contribution. The and the project’s sponsors can rely on an off- contract was very complex, with over 300 doc- take guarantee from the corridor’s main user​ / uments and 112 separate signatories. anchor client that is low risk. The rail project won immediate praise for 2. With this type of loan, interest is paid during its efficient construction, for creatively resolv- the course of the loan and the principal is paid ing challenges that arose regarding land and at the end. Spotlight 3 The Role of Public- Private Partnerships in Developing South Asia’s Corridors SOUTH ASIA’S EXPERIENCE The experiences of general transport PPPs WITH PPPs FOR TRANSPORT in South Asia are also mixed. In India, INFRASTRUCTURE which has more PPPs than the other coun- tries, it appears that the insolvencies of Corridors offer unique many highway PPPs are challenging the opportunities for PPPs banking sector and impeding physical prog- If corridor projects that aim for wide eco- ress on construction. These problems have nomic benefits are to succeed, funding and not yet been addressed, which affects financing arrangements must serve the goals both infrastructure assets and their wider that underlie a corridor. In South Asia, economic benefits. The issues with the PPP where corridor projects are difficult to highways are linked to a heavy reliance on deliver (regardless of the financing mode), public sector banks for PPP lending, com- the goals of public-private partnerships bined with speculative behavior in contracts (PPPs) and those for developing corridors for short-term gains while the projects were are often misaligned. However, they could being constructed. Similarly, in Nepal, PPP be resolved if the region’s PPP models evolve models have failed to deliver on major cor- in ways that meet expectations for what ridor projects such as the Fast Track high- they can deliver. way, despite six attempts since 1996. South Asia has huge financing needs to Conversely, in Pakistan, PPPs for ports develop its infrastructure. In 2014, the World and container terminals appear to have per- Bank estimated that Bangladesh, India, formed better. For example, Gwadar Port Nepal, Pakistan, and Sri Lanka would require entered into an agreement with the Port of US$408–US$685 billion in investment at Singapore Authority to manage operations, 2010 prices to bridge gaps in the transport and Karachi Port engaged Hutchison Port sector alone (Andrés, Biller, and Herrera Holdings to construct and operate the Dappe 2013). Moreover, many of the projects Karachi Deep Water Port. It also appears that are complex and risky because of the political-­ India’s relatively small but important railway economy environment. This is particularly PPP projects are succeeding. However, the true for major corridor projects. Thus, PPPs positive examples of delivering better results are an enormous challenge because unstable in the highways sector projects are limited. project environments can easily compromise Generally, the ability of PPPs to consistently bankability. deliver the large-scale, complex, greenfield 113 114   THE WEB OF TR A NSPORT C ORRIDORS IN SOUTH A SI A corridor infrastructure projects South Asia indicative, rather than definitive. For example, needs is still unproven, despite significant the road project in Nepal (the Kathmandu- support from the international community Kulekhani-Hetauda Tunnel Highway) has and the region’s governments. struggled to obtain financing for more than In considering the future role of PPPs in four years, and its future is still uncertain. developing South Asia’s main transport cor- Similarly, when analyzing the volume of ridors, it is important to explore past expe- investment, different sources offer different riences. First, this entails examining cash figures. flows and risk profiles of projects where India’s road PPPs are concentrated on investors were asked to tolerate both calcu- national highways. The program is large lated risks and complete uncertainty. (US$32 billion−US$47 billion), depending Second, this involves understanding the on whether the developer’s or India’s South Asian context, where financial sec- National Highway Authority calculate the tors, construction industries, laws, and gen- total project costs. Importantly, roughly eral governance are still developing. Third, 80 percent of the debt to finance this port- it involves addressing the issue of whether folio has come from public sector banks private finance should be considered a pri- in which the government owns more than mary solution or if expectations should be 50 percent of capital shares. Since 2013, lowered. This last point might mean a projects have faced mounting financial and greater focus on expanding public finance to operating challenges that have been passed facilitate infrastructure investment, either in on to the banks and developers supporting support of the private sector or corporatizing them and have contributed to the rising public sector entities that operate on commer- number of nonperforming assets (NPAs) in cial principles. India’s banking sector. The State Bank of India (SBI), which holds the greatest nomi- nal amount of Indian highway-related debt, South Asia’s mixed experiences reported that about 20 percent of loans to To date, private participation in South Asia’s ports and highways were in nonperforming transport infrastructure has mainly been in status by the end of 2016, with the India’s road subsector, according to the World trend increasing in 2016 (figure S3.1). Bank’s Private Participation in Infrastructure Developers, several of which are India’s (PPI) data. Table S3.1 calculates the transac- national construction industry champions, tions for airports, ports, railways, and roads also show signs of stress, slowing invest- by country. These data should be seen as ment growth. (This dual stress among TABLE S3.1  Roads in India dominate the number and share of private participation in infrastructure in South Asia Total number, Percentage of Country Airports Ports Railways Roads by country total, by country India 9 48 10 399 466 95.5 Pakistan 1 13 0 0 14 2.9 Bangladesh 1 2 0 2 5 1.0 Sri Lanka 0 2 0 0 2 0.4 Nepal 0 0 0 1 1 0.2 Bhutan 0 0 0 0 0 0.0 Total 11 65 10 402 488 100 Percentage of total, by subsector 2.3 13.3 2.0 82.4 100.0 Source: World Bank, Private Participation in Infrastructure data. T H E R O L E O F P U B L I C - P R I V A T E P A R T N E R S H I P S I N D E V E L O P I N G S O U T H A S I A ’ S C O R R I D O R S    115 FIGURE S3.1  A growing share of the loans to ports and highways that are backed by the State Bank of India are nonperforming 25 NPAs (% of disbursed loans) 20 15 10 5 0 :4Q :1Q :2Q :3Q :4Q :1Q :2Q :3Q :4Q :1Q :2Q :3Q :4Q 13 14 14 14 14 15 15 15 15 16 16 16 16 20 20 20 20 20 20 20 20 20 20 20 20 20 Power Telecommunications Roads and ports Source: Corridor Study Team analysis of Basel III disclosures. Note: NPAs = nonperforming assets; Q = quarter. public and private lenders and developers is Under the second, Indian Railways handles called the “twin balance sheet problem.”) acquisition. Regardless of the model, any land Recapitalization of the banks and projects acquired for the railway may be used only for appears inevitable, but the government’s railway purposes, so broader corridor devel- ability to recapitalize public sector banks opment in the PPP project is limited. and ensure that the projects are completed India’s railway PPPs have experienced in time has been doubtful. The announced strong demand and support from off-takers or bank recapitalization announced on groups of them, as they have a serious stake in October 24, 2017, could be a step in the the service to be delivered, as well as financial right direction. However, many have criti- stakes in the project SPV. For example, the cized it for not addressing moral hazard Angul-Sukinda New Railway Line Project in issues, cooperate governance of banks, and Odisha benefited steel plants in the Angul the negative fiscal impact.1 region—as well as ports, mines, and alumi- num smelting operations—providing them with better transport of bulk commodities. Railway PPPs in India have Several of these off-takers have contributed fared better than road PPPs capital to the SPV and have representatives India’s experience with railway PPPs has been sitting on the board of Angul Sukinda Railway less extensive than with roads but still offers Ltd., wearing multiple hats as investors, cus- valuable lessons. These PPPs have adopted an tomers, and governors of the railway. unbundled approach to infrastructure and The large, concentrated nature of commer- service delivery, where investment and mainte- cial stakes in a project’s railway service have nance are handled by special-purpose vehicles created strong security for investment because (SPVs), while the rolling stock and operations the off-takers’ own demand secures the debt are handled by Indian Railways. Revenues are and equity to finance the project. In addition based on a formula that nets out the operating to lenders, project sponsors also understand cost of railway services from an access charge the importance of their demand for the ser- payable to the SPV. Two variations of the vices, as well as their control over it and the model exist with respect to land acquisition. service. In contrast, road PPPs are usually Under the first, the SPV acquires the land. secured against a broader revenue base, 116   THE WEB OF TRANSPORT CORRIDORS IN SOUTH ASIA where revenues from a single road user or for investment and associated repayment small group of them would be unlikely to would derive from the cash flows that these account for most of project cash flows. Thus, agreements help ensure. Countries in demand risk in India’s railway PPPs has been South Asia (except for Bhutan) fall below the largely controlled by groups with strong links average among comparable country groups to the SPV. As a result, financing for the rail- regarding the time needed to enforce con- way PPPs has been more stable than for tracts, the World Bank Group’s 2016 Doing India’s roads. Several SPVs have reportedly Business survey shows (see figure S3.2). retired debt sooner than expected, with robust For example, in Bangladesh, India, and project cash flows. Sri Lanka, it takes about four years to enforce Two factors affect the replicability of the a contractual claim. This uncertainty creates a experiences with India’s railway PPPs. First, challenge for investors that is difficult to meet the scope of the nine projects in this portfolio through project-specific mechanisms that are was limited and targeted to serve certain ben- also acceptable to governments. eficiaries. Second, the scale of individual PPPs are addressed in laws and precedents transactions was generally smaller than for that relate to corporations and partnerships, road projects, ranging from US$17 million to competition, property rights, insolvencies, US$300 million. The potential for expanding regulation of public services, safety, environ- ­ the model to larger corridors is limited by mental protection, labor, sector-specific questions about the long-term model for ­ provisions, and the like. These arrangements India’s railway services and the role of Indian are complex, particularly for one-off projects Railways. The World Bank–supported that may have atypical structures. Eastern Dedicated Freight Corridor (DFC) To reduce complexity and facilitate PPP conducted a pre-​­feasibly analysis for using a projects in transport and other sectors, India PPP to develop the DFC lines. This exercise has sought to standardize agreements. But it showed that larger projects with more com- has still not addressed the consistency and plex demand and land acquisition risks, along time required to enforce provisions that with more complex access charges, may secure investment. For example, India’s laws not simply be viable in the current context for enforcing security are relatively devel- as PPPs. oped but ineffectively applied. The way insolvencies are treated to recover nonper- forming loans is particularly challenging. Contracts and investment The 1993 Recovery of Debts Due to Banks security are weak and Financial Institutions (RDDBFI) Act cre- The enforceability of contracts—the glue for ated debt recovery tribunals to help collect project finance—is a key concern in South Asia. bad debts. The 2002 Securitization and The region’s context has challenged the foun- Reconstruction of Financial Assets and dations on which transport PPPs depend, spe- Enforcement of Security Interests cifically the use of contracts as the backing for (SARFAESI) Act strengthened the lenders’ investment. Typically, these would include a ability to collect. But the enforcement pro- project agreement with government authori- cess is slow, inconsistent, and subject to out- ties (such as a concession agreement), as well side influences. India’s Supreme Court as off-take contracts, partnership agreements, recognized this, noting, “It is a matter of seri- supplemental project support agree- ous concern that despite the pronouncements ments, construction contract(s), financing of this Court, the High Courts continue to agreements, credit enhancements/​­ guarantees, ignore the availability of statutory remedies supply agreements for inputs of raw materi- under the RDDBFI Act and SARFAESI als, leases, and potentially many others. In a Act.”2 Moreover, the Reserve Bank of India traditional project finance structure, security noted that large borrowers—the same ones T H E R O L E O F P U B L I C - P R I V A T E P A R T N E R S H I P S I N D E V E L O P I N G S O U T H A S I A ’ S C O R R I D O R S    117 FIGURE S3.2  The time needed to enforce contracts varies from less than a year to more than four years in some South Asian countries 1,800 Time required to enforce a contract (days) 1,600 1,400 1,200 1,000 800 600 400 200 0 es ia h ka l om w om le an an n pa es ta inc L o Ind inc idd iv an ut ist Ne kis e e lad ald Bh iL an M Pa ng Sr M gh Ba Af South Asian countries Comparator country income groups Source: 2018 Doing Business Survey. Note: Low- and middle-income countries are as defined by the World Bank. entrusted to deliver on India’s PPP projects— these have had floating interest rates with rate are able to manipulate the enforcement pro- resets of 1 to 2 years. Such rates and limited cess in their favor.3 Laws across South Asia hedging instruments expose projects to added are still underdeveloped. For example, in risks due to fluctuating interest rates. Project Bhutan and Nepal, laws regarding transpor- delays, often attributed to public authorities tation are likely to undergo revision in the and project companies, can exacerbate these near future. risks—particularly during a project’s early years, when debt service is often most sensi- tive. The rate on India’s PPP debt has also Financial markets are underdeveloped been an issue, with highway-related PPP debt South Asia’s financial markets are at different generally at 12 percent to 16 percent in nomi- stages of development, and governance varies nal terms. Perhaps most important, these PPP from country to country. South Asian coun- programs have shown that the ability to tries, except for Bangladesh, rank above aver- source debt on amenable terms is based on age compared to their low- and middle-income relationships. On account of this, many proj- peers, while the depth of credit provided by ects have not stood on sound technical or domestic sources is comparable, according to financial fundamentals. Thus, financial sec- the World Bank’s Country Policy and tors’ may not be able to support such PPPs Institutional Assessment (CPIA) of financial and large, high-risk projects. sector governance (figure S3.3). However, the countries have limited ability to meet PPP The role of underdeveloped markets investment needs in transport corridors. For example, India’s highway PPP projects The substantial size of corridor PPP projects have struggled to source long-term debt. relative to the banking sectors in South Asia While loans up to 15 years have been made, makes transactions potentially unwieldy. 118   THE WEB OF TRANSPORT CORRIDORS IN SOUTH ASIA FIGURE S3.3  Most countries in South Asia compare well to their low- and middle-income peers in the share of domestic credit provided by their financial sectors 3.60 India Pakistan Sri Lanka Nepal 3.40 IBRD only CPIA financial sector rating (1 = low to 6 = high) 3.20 IDA blend 3.00 Bhutan Middle income 2.80 IDA only Low income 2.60 Bangladesh 2.40 2.25 22.25 42.25 62.25 82.25 102.25 122.25 142.25 Domestic credit provided by financial sector (% of GDP) Source: World Bank, World Development Indicators. Note: Low- and middle-income countries are as defined by the World Bank. “IBRD only” refers to countries that are eligible only for near-market-based financing from the World Bank Group’s International Bank for Reconstruction and Development (IBRD). IDA blend refers to countries that are eligible for a blend of concessional financing from the World Bank Group’s International Development Association (IDA) and other financing, including IBRD financing. For example, Nepal’s central bank, the Nepal These considerations have contributed to Rastra Bank (NRB), reported that the total six unsuccessful attempts to pursue PPPs Tier I and II capitalization of the domestic related to Nepal’s Fast Track project since banking sector was NPR 255.7 billion in 1996, although these efforts have had inter- January 2017 (US$2.3 billion).4 To put that national support from bilateral and multilat- figure in context, most recent cost estimates eral development partners. There have been for the Fast Track highway project, which at least 10 different studies and technical would provide the transport backbone of a assistance engagements since 1974, including key corridor with improved trade links to from the World Bank / Public-Private India (Nepal’s largest trading partner), exceed Infrastructure Investment Facility, the Asian US$1.3 billion. Nepalese banks face signifi- Development Bank, the United Nations cant risks in taking on high levels of exposure Development Programme (UNDP), and the relative to their capitalization against a single governments of Finland, Japan, Switzerland, high-risk project. However, while it may be and the United Kingdom. These efforts, possible to obtain international financing, the along with Nepal’s own actions, have been geopolitical implications of tapping significant constructive despite challenges that have levels of cross-border debt in a region with a made delivery of the Fast Track project elu- history of instability and conflict are trou- sive. Major infrastructure projects are chal- bling. Whether these considerations outweigh lenging—even in developed countries. the benefits of using PPPs to develop a corri- However, the complexities related to Fast dor may raise sovereignty issues more than Track as a PPP within a difficult country considerations about finance and project context may have contributed to delays and delivery. Similarly, political economy consider- distracted from efforts to address project ations of cross-border lending also introduce fundamentals. Specifically, PPP attempts high levels of uncertainty for investors that are involving Fast Track seem to have led to the difficult to price. view that the private sector would do the THE ROLE OF PUBLI C - PRIV A TE P A RTNERSHIPS IN DEVELOPING SOUTH A SI A ’ S C ORRIDORS   119 engineering to prepare the project, which is a An important consideration when seeking common misconception. Despite its long his- finance from outside each South Asian coun- tory of technical assistance, the Fast Track try and outside the region is whether macro- project still does not have a credible and prudential regulations of domestic banks buildable design or enough engineering due could be one way to encourage foreign lend- diligence to warrant the design being ing to PPP projects in South Asia. India has produced. used central bank regulations to provide special treatment for loans to infrastructure, and this effort seems to have encouraged Finance from outside the bank lending. For example, the Reserve Bank region is low in general, of India previously allowed banks to risk- and especially low for weight their infrastructure loans at a level corridor PPPs that was more favorable than for standard Given the constraints on domestic financial corporate debt. However, sector policies that sectors, a key question is whether South Asian target PPPs and central bank regulations of countries can obtain international capital for domestic banks can have both positive and programmatic projects for corridor PPPs, if negative implications for the projects. not for one-off projects. In answering this Specifically, India’s experience is that the question, it is important to consider the actual risk of loans, particularly in the high- potential of cross-border infrastructure way subsector, was materially different than investment relative to the overall attractive- the risk-weightings allowed for capital provi- ness of foreign investment in South Asia. Data sioning in general. Challenges in highways on inflows of foreign direct investment (FDI) have occurred along with high rates of to South Asian countries offers a mixed view nonperforming loans across other sectors as (figure S3.4). Most of the countries lag peer well. The public sector banks in India are groups in the amount of FDI inflows they also facing a need to increase capitals in capture relative to GDP. order to meet Basel III capital requirements. FIGURE S3.4  Overall foreign direct investment is relatively low in South Asian countries 4.50 Foreign direct investment, net 4.00 3.50 inflows (% of GDP) 3.00 2.50 2.00 1.50 1.00 0.50 0 e ly e ia nd sh ka n an l pa om om ta on Ind e an ble ut Ne kis lad inc inc Bh RD iL Pa A ng Sr IB ID w le Ba Lo idd M Source: World Development Indicators. Note: Low- and middle-income countries are as defined by the World Bank. “IBRD only” refers to countries that are eligible only for near-market- based financing from the World Bank Group’s International Bank for Reconstruction and Development (IBRD). IDA blend refers to countries that are eligible for a blend of concessional financing from the World Bank Group’s International Development Association (IDA) and other financing, including IBRD financing. 120   THE WEB OF TR A NSPORT C ORRIDORS IN SOUTH A SI A The impact of unstable projects and market fragmentation. In the program’s later programs on national champions years, developers were no longer required to disclose their financing plans or provide evi- Except for Pakistan’s PPPs for ports, develop- dence that net cash accruals would not be less ers in South Asia’s corridor PPP programs than 10 percent of a project’s total cost (which have been dominated by domestic players. was previously required). Eligibility criteria India’s highway program, which has evolved were also reduced, which allowed firms, over several years, offers a look at the deepest regardless of their capabilities, to qualify as and most developed of such markets in the bidders using an associate’s credentials. region. Two interesting observations emerge Changes to the qualifications removed nearly from this analysis (figure S3.5). First, as the all barriers to entry for would-be developers. volume of India’s PPP transactions expanded New market entrants included traditional during the boom years, the market for poten- contractors-turned-concessionaires, as well tial developers fragmented. In 2004, there as financiers-turned-concessionaires. were 10 major players in the developer mar- Anecdotal interviews from the World Bank’s ket for India’s highway PPPs. However, the 2013/14 research noted that the market frag- number of prospective developers increased mentation led to particularly aggressive bid- dramatically, and the market fragmented. By ding against speculative factors. Further, new 2012, no single player held more than a and underqualified bidders had fewer techni- 10 percent market share and developers with cal capabilities to implement projects. market shares of under 3 percent each held Another important finding from the analy- roughly half of all contracts. sis is that stress in the portfolio of PPP proj- Less stringent eligibility criteria and a ects seems to have contributed to financial reduced focus on bid quality helped drive weakness in several of India’s leading FIGURE S3.5  India’s market for corridor PPPs fragmented as it expanded—and quality suffered a. Until 2004 (by volume of contracts) b. In 2012 (by volume of contracts) Others, Gamuda L&T, MSK projects 8% Malaysia, (India) Ltd., 8.28% IL & FS, 13% 3% 5.23% IL & FS, KMC Construction, 5% CIDBI 5.20% Malaysian, Gammon 13% Infrastructure ISOLUX Corsan, Ltd., 5.15% 7% 63 others less than 3% each, IVRCL, 4.50% Jaiprakash 50% Industries Ltd., IRB Infrastructure 9% Developers Ltd., 3.86% GMR Infrastructure, Navayuga Engineering MSRDC 11% Co. Ltd., 3.83% Mumbai, 10% Pacific Gammon Infrastructure Alliance Inc., Transtroy, Ltd., 3.70% GVK, Oriental Structural 3.51% 11% 11% Engineers, GMR Infrastructure, 3.51% 3.13% Source: Corridor Study Team analysis of 2014 National Highways Authority of India data. T H E R O L E O F P U B L I C - P R I V A T E P A R T N E R S H I P S I N D E V E L O P I N G S O U T H A S I A ’ S C O R R I D O R S    121 domestic construction firms. Publicly traded nonperforming assets in the banks that sponsors provide insight regarding how these lent to individual projects. factors matter. For example, while Larson & There are two views concerning developer Toubro appeared to hold the largest portfolio and project stress. The first considers devel- of stressed national highways contracts in oper stress as the self-induced outcome of 2013/14, the total amount of these contracts speculation in contracts for short-term gains was only 28 percent of the firm’s total share- during early construction stages of projects. holder capital and reserves in the 2014 finan- This may be accurate and justified to some cial year. In contrast, GMR’s portfolio of extent; evidence shows that developers played stressed assets was equal to 110 percent of the a short-term game when bidding for highway firm’s shareholder capital and reserves, while contracts. An opposing view is that these Gammon Infrastructure’s portfolio of high- actions responded to risk allocation frame- way assets was roughly equal to 518 percent works, tender strategies, and financing mod- of the firm’s reported shareholder capital and els that invited speculation under an unsound reserves in the same year (figure S3.6).5 PPP model. Delays in land acquisition, clear- The accumulation of stress in India’s ances, and other public sector obligations also national highway PPP program also coin- played a large role in further stressing the cided with an accumulation of debt by highway PPP portfolio. This fact, combined many (but not all) developers. Publicly with the need for strong domestic firms that traded developers provide an indication of can develop India’s infrastructure, may mean how this occurred. The level of indebted- considering ways to relieve stress in overlever- ness for developers also believed to hold aged developers that could be financially via- large portfolios of stressed national high- ble in the future. It may also mean w a y P P P s i s n o t e w o r t h y, g i v e n t h e reconsidering traditional notions of PPP increased risk of insolvency and illiquidity risk transfer, given the dilemma countries will resulting from proportionately large face if they also want to prevent their domes- l iabilities and debt service payments. ­ tic construction industry from being exposed These factors are behind the buildup of to added financial risks. MOVING FORWARD FIGURE S3.6  Debt-to-equity ratios of four publicly PPPs and private finance both have a role in traded developers in India varied considerably developing South Asia’s transport corridors. from 2010 to 2015 However, it is unclear whether they can 8 deliver at the scale and on the types of proj- ects such development requires. Where PPPs Debt to equity (book values) 7 6 may be a viable approach, it is important to 5 also consider how best to accommodate 4 South Asia’s context. International partners 3 working in this area may need to change their 2 approach and adjust expectations about what 1 PPPs can reasonably achieve. 0 2010 2011 2012 2013 2014 2015 L and T GMR Corporate finance modes IVRCL Gammon infrastructure Ltd. need to be modified Source: Corridor Study Team analysis of publicly available audited South Asian countries might consider financial statements. Note: In 2014, Gammon Infrastructure Ltd. executed a private placement departing from the traditional project of shares to increase shareholder capital and reserves. finance (non-recourse) mode for financing 122   THE WEB OF TR A NSPORT C ORRIDORS IN SOUTH A SI A PPPs—​ p articularly where lending is often with very little engineering due diligence and recourse-based and relationship-based. with a proposed preliminary design that would Models commonly found in utility sectors have created unsustainable slope cuts into could offer some possibilities. In the case of unstable terrain. An alternate approach could major road corridors and their tributaries, entail extensive engineering due diligence and alternatives could include placing toll-­ developing tender strategies that first focus on supporting geographical areas under a sin- extracting efficiencies and commoditizing the regulated-​ gle corporate entity, using a ­ asset private sector’s role(s) to drive value. Enhanced base approach, and having financing preparations could then help shape project secured through that entity’s balance sheet structures that would be more targeted. To be (Helm 2012). These entities would be man- clear, these preparations should not only be dated to maintain the existing network and aimed at PPPs, and would likely benefit both carry out a program of incremental invest- publicly financed and privately financed ment, as agreed upon with its regulator, undertakings. using cash flows from tolls and any potential subsidies to meet costs and pro- The capacity of public funding and vide returns to capital. This model would finance needs to be expanded be analogous to that applied to a regu- lated utility operating under defined tariffs A case can be made for bringing in more and investments to meet service targets. beneficiaries to fund corridor projects If the model could demonstrate stability, it (whether or not financed under PPPs), con- would simplify potential financing arrange- sidering the broad nature of the wider eco- ments and allow for an incremental nomic benefits the corridor investments approach to the overall development of ideally will produce. This expansion could networks around the key corridors. A major take the form of incremental taxes or challenge would be the governance envi- charges levied on wider beneficiary groups. ronment of the entity managing the These revenues could be sequestered for ­network assets. projects or programmatic approaches. India previously uses a tax on fuel to capi- talize its central road fund, which has fueled Preparing projects to the National Highways Development commoditize the role of Program. Outside of South Asia, the funding private sector partners approach to London’s Crossrail project It is difficult for an authority to effectively out- (which is not a PPP) is meeting approxi- source what it does not understand. A key mately one-third of its costs with wider ben- notably a constraint on the broader use of PPPs in South e f i c i a r y c o n t r i b u t i o n s — ​ Asia is the common expectation that the pri- supplemental tax on London businesses and vate sector will compensate for the public sec- added taxes on land development. An tor’s failings in project preparation. important result is that it has helped the In particular, the design-build elements that Crossrail project survive budget cuts in constitute many transport PPPs may have led the United Kingdom because canceling the authorities to underprepare projects with a project would have forgone much larger view that private sector engineering or due investment (and revenues) than the treasury diligence would compensate. In the Indian might have hoped to save (World Bank and example, highway PPPs were created on the Imperial College London, forthcoming). basis of what could be considered a pre-​ Efforts to adopt PPPs more widely to feasibility study. With Nepal’s Fast Track proj- develop South Asia’s corridors may benefit ­ ect, attempts to award PPP contracts proceeded from approaches that seek to augment the THE ROLE OF PUBLI C - PRIV A TE P A RTNERSHIPS IN DEVELOPING SOUTH A SI A ’ S C ORRIDORS   123 funding mix using public sources derived with deficient oversight according to specula- from a wider range of beneficiaries. tive models, the likelihood of destroying pub- lic value is high. This has often been the case Expectations should be moderated when PPPs in South Asia have led to unin- tended negative outcomes. The international South Asia’s positive and negative experiences development community should consider how offer valuable lessons for the international its actions or inactions may have contributed community. Some of these lessons are techni- to such outcomes, and what adjustments to cal, relating to risk allocations, the need for country engagements must be made to technical due diligence, contract management, improve the chances of success in the future. and managing implementation challenges. Others are more strategic, relating to institu- tional behavior, the relationship between cen- NOTES tral authorities and states, oversight of the 1. See Shenoy, D. “Explained: The Great Indian banking sector, and general governance. Bank Recapitalisation Push.” The Wire. However, the main lesson may be that South h t t p s : / / t h e w i r e . i n / 1 9 0 5 5 4 / e x p l a i n e d​ Asia offers a window on how governments -​great-indian-bank-recapitalisation-push/. and the international community can 2. Supreme Court of India, Union of India approach PPPs and what they should expect v. DRT Bar Association, January 22, 2013. from them. 3. Raghuram G. Rajan, Governor, Reserve Bank of There is often confusion about the differ- India, at the Third Dr. Verghese Kurien Memorial Lecture at IRMA, Anand, November 25, 2014. ence between funding and financing 4. Nepal Rastra Bank, “Key Financial Indicators,” i nfrastructure. PPPs are not a funding ­ January 2017. mechanism, in that they do not create new ­ 5. World Bank analysis of publicly available money to pay for infrastructure investments. financial reports for the companies m ­ entioned. They do not fill vast funding gaps—despite all The end of the financial year varies between hopes and hype that they may. Rather, they companies. offer value for money only to the extent that they capture incremental efficiencies. The REFERENCES cash that supports privately financed infra- structure investments must ultimately come Andrés, L., D. Biller, and M. Herrera Dappe. 2013. from one of three sources: user or other bene- “Reducing Poverty by Closing South Asia’s Infrastructure Gap.” World Bank and ficiary charges; transfers from general tax rev- Australian Aid. enues; or foreign donor grants. The financing Helm, D. 2012. “What to Do about the Roads?” that the private sector can provide is a net September. http:/www.dieterhelm.co.uk/assets​ consumer of cash, which can just as easily /­s ecure/documents/What-to-do-about-the​ destroy value as create it. When private -­roads.pdf. financing is deployed on poorly prepared World Bank and Imperial College London. projects in weak institutional environments Forthcoming. “The Operators’ Story.” 4 Learning from the Literature about the Estimated Wider Economic Benefits of Transport Corridors C urrent initiatives and proposals to build transport corridors could demand trillions of dollars in the near prospects, and misunderstandings between governments and among international organizations. future—much more than the capacity of public finances (Bougna et al., forthcoming). By reducing transport costs, such initiatives The literature offers insights into prioritizing typically aim to unleash efficiencies in investment proposals for large transport projects and regional and transregional trade. But they the proper methodology to use to appraise projects. also strive to generate net benefits for house- holds and firms that extend beyond sav- ings of travel time and vehicle operating In parallel with rising interest among pol- costs. These net benefits stem from the posi- icy makers in the prioritization and appraisal tive spillovers and increased economic of transport corridors, there has been a surge opportunities that transport corridors could of academic interest in empirically evaluating generate. However, to help realize these ben- the economic, social, and environmental efits, national governments, the private sec- impacts of large transport infrastructure proj- tor, and the global development community ects (Redding and Turner 2014; Berg et al. need clear economic thinking about how to 2017; Laird and Venables 2017). This surge prioritize investment proposals for large has partly been driven by improved tech- transport projects and a proper methodol- niques for rigorously evaluating the economic ogy to appraise projects. Without these, impacts of these investments and recently there are risks of uncoordinated efforts that completed large-scale investments in trans- could result in missed opportunities, wasted portation networks by countries such as public and private funds, geopolitical China and India. The academic interest ­covers considerations prevailing over economic these recent investments as well as historical 125 126  The WEB o f T r a n s p o r t C o r r i d o r s i n S o u t h A s i a large-scale investments—such as the nine- infrastructure to include reforms and teenth-century investments of the United policies that amplify the net benefits of States in its railway network and the con- this infrastructure in the wider process of struction of the colonial Indian railway net- economic and social development. The opti- work in the nineteenth and early twentieth mal package then becomes the one that a centuries. It also covers investments at a vari- benevolent social planner would choose to ety of scales, ranging from the evaluation of maximize overall net benefits. the impacts of individual links within a preex- In reviewing the literature and thinking isting or a new system to the evaluation of about the wider economic impacts of trans- entire national or even continental systems. port corridors, this study is interested not so Some of these investments may not adhere to much in the direct benefits (immediate out- what is today considered a transport corridor comes), such as savings of travel time and investment. Nevertheless, the large academic vehicle operating costs (VOC), that remain the literature estimating the impacts of these focus of most cost-benefit analyses for trans- investments could inform policy makers’ port projects.1 Although their importance thinking on corridors and help develop a should not be dismissed and their accurate comprehensive methodology to appraise measurement remains an issue for cost-benefit corridor projects. analysis, these savings are unable to capture the full economic benefits of a transport proj- ect in anything other than a hypothetical Knowledge of the direct and indirect impacts, the world of perfectly competitive and complete trade-offs they could produce, and the complementary markets (see, for example, Vickerman 2007 policies that could help manage these trade-offs all on this point).2 Rather, this report’s interest is in the wider economic impacts of large trans- can help shape the optimal policy design for the port projects, which include impacts on devel- corridor intervention. opment outcomes such as economic welfare (monetary measures of well-being, including income, wages, and consumption), social This chapter aims to provide a rigorous inclusion (jobs, gender), equity (poverty, review of the empirical literature that esti- inequality), environmental quality (pollution mates the impacts of large transport invest- and deforestation), and economic resilience ments. In doing so, it also seeks to improve (unexpected losses from socioeconomic and policy makers’ understanding of the multi- other shocks). 3 Furthermore, the study is ple impacts of transport corridor invest- interested in the potential trade-offs that may ments, which may have varied effects on occur between these different types of wider certain economic sectors, locales, and popu- economic outcomes—for example, boosting lation groups. This includes improving income at the expense of increasing pollution understanding of the potential trade-offs ­nequality—as well as the varied impacts of or i that the investments may give rise to, both transport projects on a given outcome across between different development outcomes different places and economic agents, such as and between different groups of economic economic impacts that differ across subna- actors. The review also seeks to extend pol- tional regions, industries, and segments of the icy thinking to considering the types of com- population. These heterogeneities are hidden plementary policies and institutions that by estimates of average impacts. In some may be needed to attain wider economic cases, “winners” and “losers” may only be benefits. The aim is to help determine which relative. For instance, when all subnational project elements constitute an optimal cor- areas along the route of a transport project ridor intervention package—defined as a set gain, some areas gain more than others. But of interventions that extends beyond the in other cases, the losses for some areas may investment in the main (trunk) transport be absolute. LE A RNING FROM THE LITER A TURE A BOUT THE ESTIM A TED WIDER   127 To achieve greater rigor, this chapter There is strong evidence that transport corridors adopts a quantitative approach to reviewing generate significant positive impacts on economic the literature. The chapter is based on a welfare, but  environmental quality and possibly review (a meta-analysis) of 78 papers that have been published since 1999 and that meet social inclusion could suffer in parallel. a series of qualifying criteria. Taken together, these papers provide 243 individual results, FRAMEWORK FOR STRUCTURING which constitute the sample for the quantita- THE LITERATURE REVIEW tive analysis. This sample includes both descriptive analysis and formal econometric To provide a conceptual structure for the meta-­ r egression analysis. The literature literature review, this chapter proposes the review covers papers published in peer- following simple canonical model. Policy reviewed academic journals and recognized makers try to maximize the wider economic working paper series, as well as “gray” litera- benefits of a potential corridor, considering ture that has not (yet) been published through benefits net of costs. They deliberate about a formal outlet and that is available only for corridor features such as location, length, and download on authors’ personal websites.4 mode of transport, as well as accompanying The review excludes literature that is solely complementary policies that can help maxi- concerned with the evaluation of transport mize the targeted wider economic benefits. infrastructure projects within an urban area The cost of the new transport corridor inter- (intra-urban transport).5 It also excludes lit- vention includes trunk infrastructure (such as erature in which more general transport- highways), transport and trade facilitation related variables appear in the empirical (such as improving border crossings and estimation, but these are either not related to removing obstacles to foreign trade), as well evaluating the impacts of a particular large- as the costs associated with the design and scale transport infrastructure project and/or implementation of complementary policies they appear as one of several general vari- and institutions. This cost increases with the ables of interest.6 gap between the new set of policies and the A few recent reviews of literature on the preexisting set of policies. This policy maker’s economic impacts of large-scale transport problem can play out at different levels of infrastructure projects have been published. aggregation: local (subnational units such as The most notable are the reviews by Redding districts), national, and transregional. and Turner (2014) and Berg et al. (2017). The final outcomes (net wider economic In both cases, the authors present qualitative benefits) can be directly affected by a corridor reviews of the literature. In contrast, this intervention and its features, depending on chapter attempts more of a quantitative other complementary factors (initial market review of the literature, which includes conditions, policies, and institutions) that the use of formal meta-regression analysis affect many aspects of the economy at the techniques. It also considers a broader set of same time, such as initial conditions in local potential economic impacts than is consid- product markets and factor markets for land, ered by Redding and Turner (2014). Finally, it labor, and capital. This direct impact can vary provides a review that connects more directly from beneficial to detrimental across individ- to the deliberations of policy makers. The ual outcomes. Furthermore, the structural framework for the review is designed to reflect factors can themselves be affected by the the problem faced by the policy maker who is corridor intervention. Thus, the corridor trying to maximize the net wider economic intervention can affect the final outcomes benefits of a proposed transport corridor indirectly. For instance, if the corridor reduces (the policy maker’s problem). This framework commuting and migration costs, it also also helps highlight important research gaps reduces frictions in the labor market, which in that remain from the policy perspective. turn can increase employment and income. 128  The WEB o f T r a n s p o r t C o r r i d o r s i n S o u t h A s i a These indirect changes on structural factors The corridor intervention package can be are called intermediate outcomes. The impact characterized by: of the corridor on intermediate outcomes can •  Placement of the trunk infrastructure, con- also vary from beneficial to detrimental across nection type, transport mode, intermodal the individual outcomes. Knowledge of the transport, resilience and redundancy of direct and indirect impacts, the trade-offs they the system to ensure its resilience to shocks could produce, and the complementary poli- and minimize transport disruptions cies that could help manage these trade-offs (backup redundancies). all can help shape the optimal policy design •  Transport and trade facilitation, including for the corridor intervention. auxiliary transport network such as feeder Some features of the optimal corridor roads or railways. package may be needed regardless of the ini- •  Regulatory and institutional reforms (land tial conditions. Other features will work best market, labor market, capital market, and or will be needed only under certain initial product market conditions, performance conditions—including the preexisting trans- of crosscutting institutions). port network. For instance, if land markets are functioning poorly, making it difficult or Several potential transmission mechanisms costly to acquire land for a transport corri- and their associated intermediate outcomes dor or harming certain groups, such as help determine the ultimate impact of a corri- smallholders, then land market reforms may dor intervention on the set of final outcomes. be needed to attain all the wider economic These can be summarized in a chain of benefits and distribute them fairly. Such expected results (a results chain). A stylized reforms are unlikely to be needed if land version is shown in figure 4.1. markets are functioning well. Similar consid- Policy makers and other stakeholders would erations apply to the need for reforms in like to know which particular features of cor- other key markets—including product, ridors (mode of transportation, length, loca- labor, and capital markets—as well as cross- tion, nodal connections, and so on) and which cutting institutions (such as public sector complementary policies (land market reforms, governance). improved access to finance, regulatory reforms FIGURE 4.1  The web of WEB: The final outcomes of a corridor intervention are achieved through many transmission channels and various intermediate outcomes Intermediate Wider economic benefits outcomes (WEB) (final outcomes) • Land value • Economic welfare • Migration • Income • Population • Consumption Corridor • Agglomeration • Assets intervention • Firm location • Social inclusion package • Investment/FDI • Jobs • Structural change • Gender • Productivity • Inequality • Trade • Poverty • Environmental quality • Air pollution • Deforestation • Economic resilience Source: Corridor Study Team. Note: FDI = foreign direct investment. LE A RNING FROM THE LITER A TURE A BOUT THE ESTIM A TED WIDER   129 FIGURE 4.2  Direct and indirect impacts determine the overall impact of a corridor intervention Overall impact of Direct Indirect impacts corridor intervention = (unconditional) + (conditional on initial frictions on WEB impact in markets and institutions) Source: Corridor Study Team. Note: WEB = wider economic benefits. in product markets, and so on) need to receive welfare and social inclusion. However, in other greater weight under different sets of initial cases, the impacts may be negatively corre- conditions (unclear land titles, labor market lated. For instance, economic welfare impacts frictions, financial market imperfections, may be beneficial, but environmental impacts extent and state of any pre-existing transport may be detrimental. This leads to trade-offs infrastructure, and so on). between different outcomes. Although an optimal policy rule to design Impacts may also be heterogeneous. That corridor interventions might be desirable, it is, for a given outcome, they could vary sig- may not be fully attainable with current nificantly across different geographic areas, knowledge (see Bougna et al., forthcoming). segments of the population, economic sectors, Less ambitiously, this chapter thus focuses on and the like. These heterogeneous impacts understanding how different corridor packages may be beneficial to all, but vary in size. Thus, affect the net wider economic benefits, consid- the impact will be relative, depending on the ering both direct (unconditional) and indirect beneficiaries’ greater or smaller predisposition (conditional) effects. For instance, how does to benefit. Losses can also be absolute. For changing the types of nodes linked by a corri- instance, if people migrate in search of better dor, or the mode of transportation within the jobs, towns lose population and become more corridor, affect development outcomes—both fragile. Certain communities with bleak pros- directly and indirectly, by improving access to pects and poor endowments get left behind. markets, land use, and migration patterns? Figure 4.3 presents a hierarchy of multiple Figure 4.2 illustrates the relationship impacts across all these dimensions. between the direct (unconditional) and indirect (conditional) impacts. Direct (unconditional) METHODOLOGY impacts captures the average impact of differ- Selection of papers ent aspects of a corridor intervention package on the set of final development outcomes with A three-step process was used to identify which a policy maker may be concerned (such papers for inclusion in the literature review. as economic welfare, social inclusion, equity, In step one, a long list of papers for possible environmental quality, and resilience). Indirect inclusion was identified by performing a series (conditional) impacts captures how aspects of of Google Scholar searches based on different a corridor intervention package might interact permutations of three categories of keywords: with different initial conditions to influence the transport-related keywords; keywords relat- set of final development outcomes. ing to variables of interest regarding out- A transport corridor has potential impacts comes, including intermediate outcomes and across multiple outcomes. In some cases, these wider economic impacts (both beneficial and impacts may be positively correlated—that is, detrimental); and other keywords designed to the corridor boosts both incomes and job identify empirical or econometric studies ­ creation. In other words, the corridor interven- (for a full list of keywords, see table 4A.1 in tion creates synergies in development impacts, annex 4A). These searches were confined to producing beneficial effects for both economic papers published since 1999. 130  The WEB o f T r a n s p o r t C o r r i d o r s i n S o u t h A s i a FIGURE 4.3  The overall balance between beneficial or detrimental impacts of a corridor intervention package depends on a hierarchy of impacts Multiple wider economic impacts Heterogeneous Impacts across effects individual WEB Relative Absolute Trade-offs Synergies (greater or smaller (winners and (income or (income and job predisposition) losers) environment) growth) Source: Corridor Study Team. Note: WEB = wider economic benefits. In step two, the papers were subjected to The standard is indifferent as to whether a two further screening criteria designed to nar- paper was published through a formal aca- row down the list. First, the paper must con- demic outlet—such as in a peer-reviewed aca- sider a “large” transport infrastructure demic journal, a recognized working paper investment. “Large” is defined two ways: series, or in an edited book volume—or on an •  A “large” investment could result in a sig- author’s personal website. This allows the lit- nificant improvement in connectivity erature review to capture unpublished litera- between at least two major nodes of eco- ture ­(so-called gray literature). In formal nomic activity, such as cities or ports. This meta-regression analysis (MRA), the inclusion condition is designed to reflect this study’s of such literature is considered important to underlying interest in transport corridors. detect and avoid publication bias (see, for •  A “large” investment could also entail example, Card, Kluve, and Weber 2009). extensive improvements in last-mile con- In step three, an attempt was made to cap- nectivity, such as widespread improvements ture any relevant papers that might have been in a rural road network that con- missed from the list produced by steps one nects directly to a national network of and two. This was done by analyzing the ref- highways and primary roads. This condi- erences in the papers from the list, consulting tion is designed to capture papers that con- with experts within the World Bank in areas sider transport infrastructure investments where the search had found little coverage that may be complementary to investments (such as on the subject of resilience), and by in the trunk infrastructure of corridors. visiting the personal websites of authors of papers on the list in order to identify addi- The second screening criterion requires tional papers that were “works in progress.” a paper to meet an “academic stan- In addition, any obviously or only marginally dard”; that is, the paper must present the relevant literature was weeded out. results from original applied research involv- The final sample consists of 78 papers ing statistical or econometric methods. (see annex table 4A.2 in annex 4A). LE A RNING FROM THE LITER A TURE A BOUT THE ESTIM A TED WIDER   131 Figure 4.4 shows the breakdown of papers by publication details, intervention details, meth- mode of publication. Of the 78 papers, odology, results, and “additional.” To ensure 9 percent are unpublished (that is, not pub- consistency in the collection of information lished in a formal academic outlet). The across papers, the value for each variable was papers yielded a total of 243 separate results.7 restricted to a certain number of multiple- choice options. Annex 4.D (online) provides a Tagging of papers detailed list of these options, as well as descriptions of the variables. The information A tagging system was applied to extract rele- collected was compiled in a database. vant information for the purposes of this To maximize accuracy in the data collec- review. Information was collected from each tion process, the tagging system was imple- paper for five basic categories of variables: mented through a double-blind review system. Two appropriately trained research assistants FIGURE 4.4  The literature search yielded (RAs) reviewed and independently tagged 78 papers, of which 9 percent were unpublished papers in batches of ten. Then the study team would meet to compare the results between Unpublished papers, the two RAs and to reconcile any differences. 9% Sometimes, the tags chosen by both RAs would be overridden based on the chapter authors’ own reading of the papers.8 As the papers were tagged, the system for tagging Other working was also refined to reflect the study team’s papers, evolving understanding of the literature. For a 19% Edited book Journal articles, sample entry, see table 4A.3 in annex 4A. volume, 59% 1% DESCRIPTIVE STATISTICS NBER working papers, 4% Publication time-trend World Bank working papers, The trend in terms of number of papers pub- 8% lished per year has been strongly upward, as Source: Corridor Study Team. seen in figure 4.5. While only one paper in the Note: NBER = National Bureau of Economic Research. FIGURE 4.5  Rigorous evaluations of the impacts of large-scale transport projects have surged recently 20 18 16 Number of papers published 14 12 10 8 6 4 2 0 00 01 02 03 06 07 08 09 10 11 12 13 14 15 16 17 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 Source: Corridor Study Team. Note: The sample consists of 78 papers with 234 tagged results (analyzed outcomes). In the case of unpublished manuscripts, the year of publication corresponds to the date on the manuscript. 132  The WEB o f T r a n s p o r t C o r r i d o r s i n S o u t h A s i a sample was published in 2000, 19 papers and Wheeler 2015; and Dasgupta and were published in 2016 (the last full year of Wheeler 2016). A number of the papers on data). More generally, 71 percent of papers in Africa focus explicitly on transport networks the sample were published between 2011 and that cut across national boundaries and 2016. This strong upward trend reflects the therefore cover more than one country (for growing academic focus on rigorously evalu- example, Buys, Deichmann, and Wheeler ating the impacts of major transport infra- 2006; Jedwab and Storeygard 2016). structure projects. Intervention details More than three-quarters (78 percent) of the Focus of the “typical” paper 234 reported results in the sample evaluate the Geographical focus impacts of some sort of road transport infra- Of the 78 papers in the sample, 33 papers—or structure (figure 4.6, panel a). The only other 42 percent—focus on transport infrastructure transport mode that receives significant atten- projects in only three countries: the United tion in the literature is rail (17 percent of States (13 papers), India (11 papers), and results). The overwhelming focus on road net- China (9 papers). This overwhelming focus on works partly reflects the focus on evaluating these three countries reflects the number of the impacts of the NEN in China and the GQ extremely large-scale transport infrastructure in India, as well as the impacts of the Interstate projects that they have undertaken either Highway System in the United States. All these recently or historically. Over the past two systems were newly constructed networks. decades, both China and India have con- Accordingly, 63 percent of results in the sam- structed major new highway networks—the ple focus on the evaluation of new network National Expressway Network (NEN) in systems (figure 4.6, panel b). About 21 percent China (constructed between 1992 and 2015),9 of results evaluate the upgrading of a preexist- and the Golden Quadrilateral (GQ) network ing transport system, while 8 percent of results in India (constructed between 2001 and 2012) assess the impacts of upgrading an individual (see chapter 6).10 In historical analyses, several link within a wider transportation network. papers have evaluated the impacts of exten- In terms of the type of connection consid- sive national railway networks constructed in ered, the typical paper focuses on projects both the United States in the nineteenth cen- that are designed to link urban centers tury and in India in the nineteenth and early (45 percent), although a significant propor- twentieth centuries (Haines and Margo 2006; tion of the literature (22 percent) looks at the Donaldson 2010; Atack and Margo impacts of urban–rural connections. Typically, 2011; Herrendorf, Schmit, and Teixeira 2012; the focus is on evaluating connections Donaldson and Hornbeck 2016). Crucially, between places that fall within national these evaluations have been facilitated by the boundaries.11 However, a number of studies existence of high-quality historical data on (8 percent) focus on evaluating the impacts of both the networks themselves and relevant connection between one or more urban cen- outcome variables. Other papers have exam- ters and an international gateway (a port or ined the Interstate Highway System that the an airport).12 The preoccupation of the litera- United States built in the mid–­ twentieth cen- ture with urban–urban connections that fall tury (Chandra and Thompson 2000; Baum- wholly within national borders is, again, Snow 2007; Michaels 2008). related to the dominance of China, India, and Numerous papers focus on countries in the United States as countries of study. either Africa (18 papers) or Latin America (13 papers). Several focus on environmental Methodology outcomes, particularly on roads’ impacts on The most frequently analyzed category deforestation (see, for example, Pfaff et al. of final outcomes is “economic welfare.” 2007; Weinhold and Reis 2008; Damania The categories of intermediate outcomes LE A RNING FROM THE LITER A TURE A BOUT THE ESTIM A TED WIDER   133 FIGURE 4.6  Evaluations of road projects and new systems dominate the literature Percentage of reported results a. By transportation mode b. By construction type 90 70 63 78 80 60 70 50 60 40 Percent Percent 50 40 30 30 21 17 20 20 8 10 6 10 2 2 1 1 0 0 Road Rail Waterway Rail, Rail, New Upgrade Upgrade Other New link road, road, ports, system system link existing waterway airports Source: Corridor Study Team. Note: The sample consists of 78 papers with 234 tagged results (analyzed outcomes). analyzed most often are “trade and produc- before and after the transport infrastructure tivity” and “population and assets” (fig- investment occurs (see chapter 6). Such esti- ure 4.7). The typical paper focuses on ex post mation focuses on the identification of evaluation and uses subnational geographical impacts in the immediate geographic vicinity regions—such as Indian tehsils (Khanna of the investment without explicitly seeking 2016; Alder, Roberts, and Tewari, forthcom- to identify the mechanisms through which ing), Chinese counties (Banerjee, Dufo, and such impacts occur (see, for example, the Qian 2012), U.S. counties (Donaldson and papers by Ghani, Goswami, and Kerr Hornbeck 2016), and electoral wards in the 2016, 2017). United Kingdom (Gibbons et al. 2016)—as Meanwhile, papers that use a structural the unit of analysis (figure 4.8).13 In particu- model focus on a particular mechanism, which lar, 80 percent of sampled papers rely on data is normally related to internal trade. Examples for subnational regions, while 15 percent of include Alder (2015) and Asturias, García- studies utilize micro data for individual Santana, and Magdaleno (2016). Finally, 9 households. By contrast, a mere 2 percent of percent of sample results use a computable papers use micro data for firms (Martincus general equilibrium (CGE) model. These and Blyde 2013; Gertler et al. 2014; results are invariably from papers that under- Martincus, Carballo, and Cusolito 2014). take ex ante impact evaluation analysis (such The typical paper is also explicitly moti- as Arman and Izady 2015). Structural and vated by some underlying theoret model— CGE-based approaches to evaluation possess quite often, an economic geography model of the advantage over reduced-form regressions trade—and relies on reduced-form estimation that, in principle, they can separate out the (figure 4.9). 14 Specifically, 70 percent of creation of new activity from the redistribu- results employ reduced-form estimation, while tion of existing activity. However, the down- 21 percent evaluate the impacts of transport side of these approaches is that they make the infrastructure using a structural model. strong assumption that the true underlying Reduced-form estimation tends to rely on structure of the economy is known.15 the use of a difference-in-difference (DID) Finally, the typical paper pays strong atten- estimator. The impacts of the treatment (for tion to the fact that the locations that policy example, on the treated subnational regions) makers chose to link with transport infrastruc- are evaluated against those of a set of com- ture are not random. In principle, this can lead parators (untreated subnational regions) standard ordinary least squares (OLS) 134  The WEB o f T r a n s p o r t C o r r i d o r s i n S o u t h A s i a FIGURE 4.7  The final outcome most often evaluated is economic welfare, while the intermediate outcomes studied are split between trade/productivity and population/assets Percentage of reported results a. Intermediate outcomes b. Final outcomes Environment, 11% Population and assets, 50% Economic welfare, 61% Social inclusion, 18% Trade and Equity, productivity, 10% 50% Source: Corridor Study Team. Note: The sample consists of 78 papers with 234 tagged results (analyzed outcomes). FIGURE 4.8  Studies at the subnational level and ex post evaluations predominate Percentage of reported results a. Unit of analysis b. Type of analysis Countries, 3% Ex post, 86% Micro firm-level data, 2% Ex ante, 14% Micro household-level data, 15% Micro household-level data, micro firm, 1% Subnational regions, 80% Source: Corridor Study Team. Note: The sample consists of 78 papers with 234 tagged results (analyzed outcomes). LE A RNING FROM THE LITER A TURE A BOUT THE ESTIM A TED WIDER   135 estimation to either overestimate or underesti- A recent trend in the “strongest” papers has mate the impacts of the infrastructure. For been to employ multiple strategies—for exam- example, if the outcome variable of interest is ple, both IV and placebo strategies—to con- a measure of local economic activity, overesti- vince the reader that estimated impacts are mation is likely to occur if policy makers accurately identified (see, for example, Ghani, deliberately target the linking of locations that Goswami, and Kerr 2016). The papers that they expect to grow quickly. By contrast, focus on the economic welfare category of underestimation is likely to occur if policy ­ utcomes are most advanced in terms of final o makers deliberately prioritize connections to their attention to endogeneity concerns. lagging regions with low underlying economic By contrast, papers that focus on, for example, potential. environmental outcomes tend to pay r ­ elatively It is fair to say that addressing the issue of little attention to employing identification endogenous placement has become the over- strategies designed to address concerns about whelming empirical concern in the l ­iterature endogenous placement. (Berg et al. 2017; Redding and Turner 2014). In an attempt to address endogeneity con- Descriptive analysis of results cerns, more than 70 percent of sample results adopt an explicit identification strategy Estimated average impacts (figure 4.10, panel a). Such identification strat- The analysis of the 234 tagged results reveals egies most frequently involve a strategy based that the literature provides evidence of sta- on instrumental variables (IV).16 Additional tistically significant impacts of wider eco- strategies have emerged in recent years—­ nomic benefits (at the 5 percent level or particularly, the use of a placebo, following greater) on economic welfare, social inclu- Donaldson (2010) (figure 4.10, panel b). sion, equity, and environmental quality FIGURE 4.9  Two-thirds of studies are motivated by a theoretical model, and 70 percent rely on reduced-form estimation Percentage of reported results a. Underlying theoretical model b. Methodologies used 80 70 70 60 No, 34% 50 Pcercent 40 30 20 13 10 9 8 0 Yes, 66% Reduced-form Structural model– Computable Structural model- estimation estimation + general system calibration equilibrium mode estimation Source: Corridor Study Team. Note: The sample consists of 78 papers with 234 tagged results (analyzed outcomes). 136  The WEB o f T r a n s p o r t C o r r i d o r s i n S o u t h A s i a FIGURE 4.10  To improve reduced-form estimations, the literature employs rigorous identification strategies Percentage of reported results a. Reported results that employ an identification b. Identification strategies employed strategy to address endogeneity concerns Others, 10% No, 29% Placebo, 12% IV/2SLS, 47% Difference-in-difference, 15% Matching, Yes, 71% 6% GMM, lag, 11% Source: Corridor Study Team. Note: The sample consists of 78 papers with 234 tagged results (analyzed outcomes). GMM = generalized method of moments; IV = instrumental variable; 2SLS = two-stage least squares. FIGURE 4.11  The literature has focused on estimating corridor impacts on welfare, much less on the inclusion, environment, and equity Percentage of reported results a. WEB b. Intermediate outcomes 120 40 35 100 Number of reported results Number of reported results 30 80 25 29 60 20 33 82 15 40 10 20 22 5 18 16 0 0 Welfare Social inclusion Environment Equity Population and assets Trade and productivity Not significant Significant Source: Corridor Study Team. Note: The sample consists of 78 papers with 234 tagged results (analyzed outcomes). WEB = wider economic benefits. LE A RNING FROM THE LITER A TURE A BOUT THE ESTIM A TED WIDER   137 outcomes (figure 4.11, panel a). For a given both environmental quality and equity. The type of outcome, the amount of confidence existence of such trade-offs suggests that can be attached to this evidence depends an important potential role for complemen- on both the number of results reported and tary policies. the proportion of those results that are signifi- In terms of intermediate outcomes, the evi- cant. In this sense, the evidence of significant dence suggests that investment in transport impacts on economic welfare outcomes is infrastructure tends to reduce prices and fairly strong. Almost 100 results relate to eco- boost levels of investment, trade, and produc- nomic welfare outcomes, and more than 80 tivity, while promoting industrial diversifica- of these results are statistically significant. For tion and improvements in the efficiency of social inclusion, equity, and environmental allocating resources. However, in some cases, quality outcomes, the proportion of reported reported impacts on levels of trade, popula- impacts that are statistically significant is even tion, and land values are detrimental greater than for economic welfare. However, (figure 4.13). Care should be taken in inter- in each case, the number of results is far fewer preting what constitutes a detrimental impact (22 for social inclusion, 16 for equity, and 18 in terms of population and land values from for environmental quality outcomes). The both a welfare and productivity perspective. sample contains no results for the wider eco- For example, Baum-Snow (2007) provides nomic benefit of resilience, which highlights a evidence to show that the construction of new prominent omission in the literature. As for limited-access highways in the United States intermediate outcomes, there is considerable between 1950 and 1990 contributed to sub- evidence that transport infrastructure has sig- urbanization by inducing people to move nificant impacts on both population and away from central cities. However, while this assets (such as land values), as well as on lev- study tags this impact as “detrimental” in els of trade and productivity (figure 4.11, terms of the population of a central city, it is panel b). For population and assets, 29 out of not clear that it is detrimental from the per- 37 reported impacts are statistically signifi- spective of overall social welfare. cant, while for trade and productivity, 33 out of 37 are statistically significant. Evidence of heterogeneous impacts Most reported impacts on levels of real There has been a growing trend in the litera- income, poverty, consumption, and jobs are ture to move beyond simply estimating average beneficial (in the sense that they can be impacts of transport infrastructure on the vari- expected to enhance social welfare). 17 ous final outcomes (WEB) and intermediate However, there are some notable excep- outcomes, to analyze how these impacts vary tions. All reported results for environmental across the different units of analysis considered quality outcomes are detrimental (in the in a study (figure 4.14). Most papers analyze sense that they can be expected to diminish impacts across space (subnational regions) or social welfare). Likewise, transport infra- economic sector (figure 4.15). A few studies structure projects had significant detrimen- have analyzed how impacts vary over time.18 tal effects on reported levels of equity However, the literature provides very little evi- among geographical units (places) and peo- dence (in terms of number of reported results) ple. In about one-quarter of cases, the proj- on whether, and how, impacts may vary across ects increased levels of inequality between different types of firms and differently subnational regions (spatial inequality) endowed individuals. (figure 4.12). This implies that major trans- The evidence of heterogeneous impacts in port infrastructure projects can entail trade- the literature implies that some geographic offs between different types of outcomes. areas or groups of households or firms gain While average impacts on economic well- more from transport infrastructure improve- being tend to be positive, this can come at ments than others. In some cases, however, the potential cost of worse outcomes for the losses may not be relative but absolute. 138  The WEB o f T r a n s p o r t C o r r i d o r s i n S o u t h A s i a FIGURE 4.12  Environment and equity outcomes could be suffering due to corridor placements Welfare (real income) Equity (poverty) Social inclusion (gender) Welfare (consumption growth) Welfare (consumption) Welfare (nominal income) Equity (spatial distribution) Social inclusion (job creation) Welfare (income growth) Equity (overall income distribution) Environment quality (deforestation) Environmental quality (CO2) 0 10 20 30 40 50 60 70 80 90 100 Percent Significant and beneficial Significant and detrimental Insignificant Source: Corridor Study Team. Note: The sample consists of 78 papers with 234 tagged results (analyzed outcomes). CO 2 = carbon dioxide. Major transport infrastructure projects can entail impact on real wages in either the urban or policy trade-offs not only between different types rural sector. of  WEB (such as economic welfare versus To account for this type of variation in outcome, this study reports evidence of abso- environmental quality and social inclusion) but also lute losers alongside evidence of absolute win- between different geographic areas and subgroups ners for the final outcomes of both economic of the population. welfare and social inclusion, as well as the intermediate outcomes of population and A fair amount of evidence of absolute losses trade (figure 4.16).19 The conclusion to be can be found in the subsample of results that drawn is that major transport infrastructure look for evidence of heterogeneous impacts— projects can entail policy trade-offs not only even where overall average impacts tend to be between different types of WEB (such as eco- overwhelmingly positive. nomic welfare versus environmental quality A good example is provided by a study of and social inclusion) but also between differ- the impacts of the construction of the ent geographic areas and subgroups for a National Expressway Network (NEN) in given outcome. China (Roberts et al. 2012). Although the authors find that the construction of the NEN increased real income across Chinese prefec- Complementary policies tures by slightly under 4 percent, on average, Complementary policies can mitigate the in many prefectures, the NEN had a negative trade-offs between different types of WEB LE A RNING FROM THE LITER A TURE A BOUT THE ESTIM A TED WIDER   139 FIGURE 4.13  While land values, population, and trade can increase in some areas as a result of corridor connectivity, they may decrease in other areas in parallel Price Allocation efficiency Education Investment (FDI) Entry into export Diversification (number of products) Trade Assets Productivity Population Land value Population (growth) Firm (entry/exit) Agglomeration 0 10 20 30 40 50 60 70 80 90 100 Percent Significant and beneficial Significant and detrimental Insignificant Source: Corridor Study Team. Note: The sample consists of 78 papers with 234 tagged results (analyzed outcomes). FDI = foreign direct investment. FIGURE 4.14  The trend to analyze heterogeneous impacts is growing, 2000–17 16 Number of papers that address heterogeneity 14 12 10 8 6 4 2 0 00 02 06 07 08 09 10 11 12 13 14 15 16 17 20 20 20 20 20 20 20 20 20 20 20 20 20 20 Source: Corridor Study Team. Note: The sample consists of 78 papers with 234 tagged results (analyzed outcomes). 140  The WEB o f T r a n s p o r t C o r r i d o r s i n S o u t h A s i a FIGURE 4.15  Geographic location is the most studied aspect of heterogeneous impacts Geographic Sector Temporal Market size Land Age Gender Labor Initial urbanization Infrastructure type Size (firm) Literacy Protected area Crop type Assets Production technology Urban/rural Groundwater depth Population Temperature Initial clearing level Poverty Congestion 0 20 40 60 80 100 120 Number of reported results Source: Corridor Study Team. Note: The sample consists of 78 papers with 234 tagged results (analyzed outcomes). Spatially informed complementary policies and on labor market policies and adopt a more institutions could help mitigate policy trade-offs in structural modeling approach.20 An impor- tant exception are papers that focus on envi- transport corridor investment, but their effects ronmental outcomes, especially deforestation. require more research. Several papers consider the role of protected areas in mitigating the impacts of roads and among different subgroups for a given on deforestation (Cropper et al. 2001; WEB. Unfortunately, of the 234 results ana- Damania and Wheeler 2015; Dasgupta and lyzed, only 17 percent seek to investigate Wheeler 2016). how complementary policies might influence One interesting example of a complemen- a transport infrastructure project’s impact on tary policy that needs to be studied more by a given outcome (figure 4.17). This lack of the literature is land value capture, which analysis of the role of complementary poli- attempts to “capture” the rising land values cies may, in part, reflect the literature’s preoc- around a corridor. This mechanism can help cupation with isolating the effects of a finance urban development along corridors, transport infrastructure project from other, as well as complementary interventions to potentially “confounding,” influences for the mitigate the varied effects of corridors across purposes of clean identification. When papers locales and population groups—such as do explicitly analyze the role of potentially those who do and those who do not own complementary policies, they tend to focus land that is appreciating in value, or those LE A RNING FROM THE LITER A TURE A BOUT THE ESTIM A TED WIDER   141 FIGURE 4.16  Income and population are the most studied aspects of reported evidence of absolute losers Income (nominal) Social inclusion (job creation) Income (growth) Income (real) Income (consumption) Final outcomes Environment (deforestation) Poverty Assets Equity (spatial distribution) Social inclusion (gender) Social inclusion (education) Equity (overall income distribution) Population Trade Intermediate outcomes Price Productivity Allocation efficiency Land value Investment (FDI) Firm (entry/exit) Entry into export Diversification (number of products) 0 5 10 15 20 25 30 35 40 45 Number of reported results Winners and losers No evidence of absolute losers Source: Corridor Study Team. Note: The sample consists of 78 papers with 234 tagged results (analyzed outcomes). FDI = foreign direct investment. who have and those who lack skills to take from multiple papers into a single statistical advantage of the development. Using land analysis, it allows primary effects to be distin- value capture can thus help fully realize guished from background variation and con- the wider economic benefits of transport taminating influences (Stanley, Doucouliagis, corridors. This particular complementary and Jarrell 2008). policy is discussed in box 4.1. The MRA undertaken for this chapter is intended to shed light on two questions. First, can the variations in the estimated impacts of META-REGRESSION ANALYSIS large-scale transport infrastructure projects This section presents the results of the formal that are found in the literature be “explained” meta-regression analysis (MRA) of the litera- by variations in the characteristics of the ture. MRA is a quantitative tool that helps projects themselves, and by methodological synthesize findings from diverse empirical variations across papers? Second, do policy- studies of a particular phenomenon. It has relevant insights emerge from the literature? become a commonly used methodology in the For example, is there evidence that certain social sciences. By combining the results features of projects (such as the mode of 142  The WEB o f T r a n s p o r t C o r r i d o r s i n S o u t h A s i a FIGURE 4.17  Most studies do not estimate the impacts of complementary policies None Labor Trade Protected area Land Financial market, migration Property rights, financing of infrastructure Market access Last mile Labor, land 0 20 40 60 80 100 120 140 160 180 200 Number of reported results Source: Corridor Study Team. Note: The sample consists of 78 papers with 234 tagged results (analyzed outcomes). transportation or the type of locations they treatment effect of the transport infrastruc- connect) are, on average, associated with ture or a more precisely estimated (that is, a better (more beneficial) outcomes? more certain) treatment effect.21 The MRA The MRA must contend with a major is implemented using the sample of results problem. The impacts of interest span a for final outcomes (wider economic wide set of outcomes and are derived from impacts) only, leaving formal meta-analysis a diverse set of treatment variables for of results for intermediate outcomes for different countries based on a nonuniform ­ future work.22 Annex table 4B.1 in annex set of modeling approaches. Thus, esti- 4B presents the results of the MRA. The mated coefficients on the treatment vari- model is summarized in annex 4C. able are not comparable across different Before turning to the estimation results, it papers. In some cases, they are not even is important to mention that the estimated comparable across different regression impacts of large-scale transport infrastruc- results reported within a paper. To over- ture projects vary widely in the literature. come this issue, the MRA uses the reported This variation is captured in figure 4.18, t -statistic for the coefficient on the treat- which presents the distributions (kernel ment variable as the measure of impact densities) of the estimated coefficients and ­ rather than the actual treatment coefficients the t-statistics. As the vertical lines (mean themselves. This allows the analysis to estimated coefficient and t-statistic) show, standardize results across different papers both the distributions are skewed to the right. and model specifications. It is important to This is consistent with the fact that the mean note that a higher reported t -statistic can paper finds a positive (beneficial) and signifi- result from either a higher estimated cant impact. Annex 4E (online) presents LE A RNING FROM THE LITER A TURE A BOUT THE ESTIM A TED WIDER   143 BOX 4.1  Financing and facilitating urban development along corridors by “capturing” rising land values While the construction of major transport corri- generators such as shops, leisure facilities, and dors between cities and other nodes of economic parking facilities. activity is expected to provide a stimulus for accel- Successful LVC schemes have four common erated urban development in locations along these denominators. They return land value increases corridors, such development needs to be supported generated by public investments to communities, by complementary investments in infrastructure, thereby allowing the benefits of those increases to be including in transport networks that feed into the more widely shared; they tap the combined poten- corridors. Stimulating the growth of cities along tial of both public and private land; they are based the corridors through increased market access also on sound planning principles; and they benefit from increases demand for wider and more efficient trans- efficient coordination among governmental bodies port systems within these cities. and agencies. The question arises of how to finance such com- An excellent example of LVC that helps recoup plementary investments in urban transport and not only the investment costs but also covers opera- development. One particularly promising option is tion and maintenance expenses is the Mass Transit “land value capture” (LVC). LVC takes advantage Railway (MTR) in Hong Kong, China. Over half of the fact that improvements in connectivity gen- of MTR revenue comes from property development erated by public investments in transport systems and asset management. MTR Corp. uses land lease tend to boost land ­values—whether along the cor- conditions to partner with developers based on the ridor or in a network c ­ onnecting urban areas that full market value, which assumes, at the time when feeds into the corridor. Rather than allowing the the rail line project is only about to start, that the increase in land value to be entirely captured by new rail line is completed and in place (Murakami whoever happens to own the land—even though 2015). In Japan, the Tokyo Metropolitan Area they have undertaken no investment to increase uses land readjustment to finance transit systems its value—LVC seeks to capture some of the gain in built-up and peripheral areas (Sorensen 2000). to finance public investments. More generally, the It does so through master plans at the economy, purpose of LVC is to “mobilize for the benefit of provincial, and metropolitan levels that provide a the communities some or all of the land value incre- clear vision and promote polycentric regional devel- ments (unearned income) generated by actions other opment. The system is characterized by excellent than the landowner’s, such as public investments in stakeholder coordination across public and private infrastructure” (Smolka 2013). entities. New York City has used transferable devel- LVC instruments can be broadly classified into opment rights to preserve landmarks and increase two groups: tax- or fee-based instruments such the density of activities in specific locations (Suzuki as land and property taxes, betterment charges, et al. 2015). and tax increment financing; and development- By facilitating the financing of complementary based instruments such as land sale or lease, joint urban investments that are required to help fully d evelopment of areas by government and private ­ catalyze the wider economic benefits of major parties, land pooling, and urban redevelopment transport corridors, land value capture can itself schemes. The implementation of development- be viewed as a complementary policy. LVC also has based instruments normally involves joint action the potential to help channel the urban develop- by the government and private parties. Such instru- ment that a major transport corridor can stimulate ments have the advantage of providing increased in a well-organized and planned manner. Essential resources to finance capital-intensive investments to the successful application of LVC instruments with limited fiscal distortion and public opposi- are adequate capacity and coordination among key tion. They are also able to generate sustainable rev- institutions and clear earmarking of revenues for enue streams through the construction of revenue infrastructure investment. 144  The WEB o f T r a n s p o r t C o r r i d o r s i n S o u t h A s i a FIGURE 4.18  The distributions of t -statistics are skewed to the ­ istributions for each of the four major WEB: d right, suggesting that the transport infrastructure projects yield economic welfare, equity, social inclusion, positive and significant impacts and environmental quality. Kernel density estimate of the t-statistics Estimation results 1.5 For estimations, the dependent variables are transformed so that a positive value always indicates a beneficial outcome and a nega- tive value denotes a detrimental outcome. The discussion that follows focuses on the 1.0 results from the preferred specifications (for details and further results, see Bougna et al., Density forthcoming). Main results 0.5 The literature finds, on average, that transport infrastructure has a significant beneficial effect on economic welfare (figure 4.19). On aver- age, the literature estimated a t-statistic of 0 3.547 for the impacts of large transport infra- structure on economic welfare, indicating that –10 0 10 20 30 the beneficial impact of large transport infra- t-statistics structure could be reliable, especially on income. This result is statistically significant at Source: Corridor Study Team. the 1 percent level. The literature also finds, on FIGURE 4.19  Average net impacts of corridors on final outcomes estimated by the literature should alert policy makers 3.547 3.547 4 1.711 –1 t-statistics –6 –11 –16 –15.023 Economic welfare Equity Social inclusion Environment Source: Corridor Study Team. Note: The average impact on equity is not significantly different from the average impact on economic welfare. LE A RNING FROM THE LITER A TURE A BOUT THE ESTIM A TED WIDER   145 average, that the transport impact on equity is specifically, relative to identification strategies not significantly different from the estimated using instrumental variables and/or differ- beneficial mean impact on economic welfare. ence-in-difference methods, estimations using By contrast, the impacts on social inclusion ordinary least squares (OLS) yield a t-statistic and environmental quality are significantly on the treatment variable that is, on average, smaller than the average impact on welfare. 3.81 higher.23 For social inclusion, the impacts are smaller Fourth, a failure to consider varied impacts and/or less certain, but still, on average, ben- across geographic units and the population is eficial (the net estimated coefficient on social not only a gap in itself; it leads to an overesti- inclusion is 1.711 (that is, 3.547 minus mation of either the wider economic benefits 1.836). The less certain impacts on social of transport infrastructure and/or the level of inclusion could also indicate possible trade- confidence about these impacts (figure 4.20). offs between boosting economic welfare and Compared with studies that ignore varied social inclusion through large transport infra- impacts, studies that do report evidence of structure in some contexts. The effect of varied impacts along several different dimen- transport infrastructure on environmental sions (such as across subnational geographic quality is detrimental. The net coefficient is areas or across different economic sectors) −15.023 (that is, 3.547 minus 18.57). These also tend to report significantly smaller findings are consistent with those reported in t-statistics. the descriptive analysis of the literature in the As an additional level of analysis, earlier section titled “Descriptive Statistics.” this chapter examines studies that report The results also suggest that several “policy some evidence of absolute losers alongside variables” are important and could inform winners—that is, the existence of geographic the design of future corridor intervention units, economic sectors, or population groups packages. First, the transport mode matters. that incur absolute losses of a develop- Compared with roads and waterways, the ment outcome. For instance, Chandra and reported impacts of investments in railways Thomson (2000) studied the impact of are significantly smaller and/or less certain. U.S. interstate highways on economic growth. Second, the type of connection also They found that interstate highways signifi- matters. The t-statistics are, on average, sig- cantly increased earnings growth in services, nificantly smaller for investments that focus retail, manufacturing, and transport and on connecting urban centers to international ­ public utilities industries in counties through gateways such as ports, land border crossing which highways run, but adjacent counties points, and airports. This result could suggest without highways experienced significantly that the ability of such connections to gener- decreased earnings in the government and ate beneficial outcomes depends on other fac- retail trade sectors. This analysis found that tors, such as the efficiency of port operations, these studies tend to report larger t-statistics. tariff and nontariff trade barriers, and the However, this result is not very robust. functioning of markets, including those to To further investigate the consideration which the gateways connect. of winners and losers as an important aspect Third, policy makers could end up priori- of estimating the wider economic benefits of tizing enhanced connectivity between loca- transport corridors, an alternative method tions that would, in any event, be expected to for the metaregression analysis was also yield more beneficial outcomes. This “endog- employed: the ordered-probit regression. enous placement” of transport corridors is a Specifically, a variable was created that takes real and important issue when trying to esti- a value of 0 when a paper reports a signifi- mate corridor development impact. Hence, cantly beneficial impact; a value of 1 when a the estimated coefficients for studies that do paper reports an insignificant impact (despite not use any identification strategy is signifi- positive or negative estimated values); and a cantly higher (more positive). More value of 2 when a paper reports a significantly 146  The WEB o f T r a n s p o r t C o r r i d o r s i n S o u t h A s i a FIGURE 4.20  Varied impacts of corridors estimated by the literature (by the factor of variation) can inform complementary interventions to support losers from corridors 0 –1 –2 –3 t-statistics –4 –3.797 –4.029 –5 –6 –5.681 –6.17 –7 –7.58 –8 Geographic Sectoral Geosectoral Land Labor Source: Corridor Study Team calculations. detrimental impact. This method categori- that ignores the possibility of losers alongside cally separated the significant results in the winners could deliver much smaller wider literature from the insignificant ones. economic benefits and much larger wider eco- It is reassuring that most results from this nomic costs—possibly including those for alternative estimation are consistent with the social inclusion (jobless GDP growth) or envi- baseline regression results for the t-statistics— ronmental quality (smog and greenhouse gas although some results do change (for details, emissions). s e e B o u g n a e t a l . , f o r t h c o m i n g ) . 24 Importantly, the analysis found significant MAIN INSIGHTS AND DIRECTIONS results for the consideration of winners and FOR FUTURE RESEARCH losers in the evaluation of transport corridor impacts on wider economic benefits. Once an This chapter has conducted a quantitative evaluation (study) considers the possibility of review and formal meta-regression analysis losers alongside winners, the probability of of studies estimating the wider economic finding beneficial wider economic impacts impacts of large-scale transport infrastruc- decreases by 34.7 percentage points. ture projects. Its main objective is to help Similarly, the same consideration makes the policy makers chose the optimal features and probability of finding detrimental results design of policies and options to promote increase by 18.3 percentage points. Both transport corridors. The core of this package findings are statistically significant at the of interventions are investments in the trunk 1 percent level. transport infrastructure (roads, rails, water- It is possible that this finding on evaluation ways). A set of complementary policies and design can also apply to the design of the institutional reforms can extend and deepen transport corridor intervention. That is, cor- the net wider economic benefits of the trunk ridor intervention packages designed in a way infrastructure by boosting average impacts LE A RNING FROM THE LITER A TURE A BOUT THE ESTIM A TED WIDER   147 and mitigating trade-offs. To assist policy institutions that optimize a corridor inter- makers, the chapter yields three main vention package will depend on two aspects: insights. imperfections in product, capital, labor, and First, when designing a transport corridor land markets that existed before the inter- package, policy makers must account for vention; and the initial endowments of dif- potential trade-offs in different development ferent locations and economic agents that outcomes and across different (sub)sets of are affected by the transport intervention. economic actors. This follows from two key This point is explored further in chapter 6. findings of the literature review: Third, the placement and design of the transport infrastructure matter, and so •  While for economic welfare and equity, does the mode of transportation. The size and the average estimated impacts of the wider certainty of estimated impacts depends on the economic outcomes tend to be beneficial, type of locations being connected, as well as for social inclusion and environmental the mode of transportation. The estimated quality, they are often detrimental. wider economic benefits of sampled rail proj- •  Even for economic welfare and equity, ects are smaller/less certain than those of road while the average impacts may be benefi- projects. However, this finding should be cial, these impacts can vary considerably interpreted with care because the literature across locales, economic sectors, and pop- mainly studies historical rail projects—such ulation groups. Some locales or popula- as the colonial railway network in India and tion groups may lose in absolute terms. the transcontinental railway network in the The existence and nature of these trade- United States. It is possible that the impacts of offs should drive the choice of complemen- modern freight and passenger railway net- tary interventions, such as compensation works—including modern high-speed railway policies for the identified losers from trunk networks, which have not been studied infrastructure investments. much—differ considerably from those of the Second, the set of complementary policies historical networks. and institutions that form part of the opti- mal transport corridor package could depend on the nature of the transport infrastructure Directions for further research intervention that forms the core of this The literature review also highlights many package. This follows from the finding of the important areas where further research is metaregressions that estimated impacts needed to better inform the optimal design of depend on the type of locations being con- transport corridor packages. Five areas are nected. Specifically, transport projects that most pressing: seek to enhance connectivity between urban centers and international gateways (such as 1.  Trade- offs . Much more research is ports, land border crossings, and airports) required to clarify the nature and extent yield significantly smaller or less certain of the trade-offs in the average impact of impacts than projects that target purely transport corridors on various development domestic enhancements of connectivity (such outcomes and the varied impacts across var- ­ as enhancements between two domestic cit- ious subsets of economic actors. While the ies). Realizing the full benefits of better con- impacts of large transport projects on mea- nectivity to international gateways requires sures of economic welfare have been stud- tackling such impediments as inefficient port ied extensively, the evidence of impacts on operations or prohibitively high tariff and other types of outcome is more limited and nontariff barriers to trade. The canonical less rigorous. A big omission is the impact model of the policy maker’s problem devel- on economic resilience—defined as the resil- oped by Bougna et al. (forthcoming) suggests ience of economic agents to various types that the set of complementary policies and of shock, rather than the resilience of the 148  The WEB o f T r a n s p o r t C o r r i d o r s i n S o u t h A s i a transport infrastructure itself. This study librium modeling. As the review shows, the found a lack of papers about resilience that literature is dominated by reduced-form met the criteria for inclusion in the sample. 25 estimations. Perhaps there is some skepti- Although environmental impacts are bet- cism about structural general equilibrium ter studied, the literature is not as advanced modeling because it assumes that the model as the literature focusing on economic welfare structure is correct at capturing the fac- in addressing possible identification issues. tual transmission mechanisms and behav- A particularly useful area of research ioral rules of economic agents when linking would be to study the impact of a single policy interventions to outcomes, as well as transport project on the outcomes of eco- when drawing conclusions about the mul- nomic welfare, social inclusion, equity, and tiple and second-round impacts. If the model environmental quality simultaneously. This structure is correct, the structural general would provide more direct insight into the equilibrium models could be the best tool to nature of the trade-offs between these out- use for policy decision making. But if incor- comes, rather than the more indirect infer- rect, it could be an inferior tool to use. By ence this chapter made by looking across contrast, standard reduced-form, difference- papers. Likewise, more research is required in-difference estimation is purely empirical to explore trade-offs across different and requires little knowledge of the under- segments of the population for each devel- lying mechanisms. Thus, reduced-form opment outcome. Although the analysis of estimation is generally seen as being more heterogeneous impacts and winners/losers flexible and reliable. But the flexibility comes has become more common in recent years, at the cost of little consideration of second- it is still relatively rare for papers to consider round effects. Further research is required to varied impacts along any dimension other improve the quality and accuracy of struc- than geographic location. tural general equilibrium models, and the identification and comprehensiveness of 2.  Complementary policies. The literature reduced-form regressions (see also spotlight has done little to analyze the role of other 6, “Appraising Proposed Transport Corri- policies in shaping the impacts of transport dors Using Spatial Econometrics”). infrastructure. Hence, this chapter found virtually no evidence on the appropriate 4.  Modern rail. The literature has been con- design of complementary policies in differ- cerned mainly with estimating the impacts ent circumstances. An important technical of road projects. When it has evaluated rail reason for this neglect may be that the lit- projects, it has mainly focused on historical erature has been striving to clearly identify projects. More research is needed to evalu- the impacts of the transport infrastructure ate modern rail projects, including both itself. This requires isolating the impacts high-speed rail and freight railway corri- of infrastructure from the impacts of other dors, such as those that India is currently factors, including other policy changes and constructing. More research is also required reforms. While this is desirable from the on the impacts of multimodal transport perspective of academic rigor, such rigor projects—especially because multimodal may limit the relevance of findings to policy, freight movements are likely to be critical to which is necessarily concerned with the full the design of many transport corridor proj- range of impacts from a corridor interven- ects in the future. tion package. 5.  C o u nt r y c ove rage . T he l iterat u re 3.  Structural general equilibrium modeling. surveyed covers a wide variety of coun- In principle, a promising way to study trade- tries. Nevertheless, much of the research offs and the interaction of transport infra- (42 percent of papers in the sample) has structure investments with other policies focused on only three countries: China, could be through structural general equi- India, and the United States. The ­ literature LE A RNING FROM THE LITER A TURE A BOUT THE ESTIM A TED WIDER   149 would benefit from research on other economic agents to various types of large ­ c ountries to gain insights into impacts in shocks (such as climate change, natural disas- different country contexts. ters, or commodity price shocks)—as a devel- opment outcome. However, virtually no examples of empirical literature that rigor- Summing up and moving to the ously examine the impacts of transport infra- next chapter structure on economic resilience were found. Note that economic resilience is different This chapter has followed a quantitative from the resilience of the transport infra- approach—including formal meta-regression structure itself to shocks. Literature on this analysis techniques—to review the growing topic exists (see, for example, Gapanovich et literature studying the impact of large trans- al. 2015), but it is not a direct subject of port investments on wider economic benefits. interest for this review. The review collected a large new data set of 4. The inclusion of such literature is important studies in different country contexts. The to both assessing and, at least partially, framework for the review is designed to addressing possible issues of publication bias. That is, formal outlets (particularly reflect policy makers’ objectives to maximize peer-reviewed journals) are more likely to the net wider economic benefits of a proposed publish studies with statistically significant transport corridor. In addition to integrating results (see, for example, Card, Kluve, and current knowledge, this chapter has pointed Weber 2009). Even if publication bias can be to gaps in the knowledge base that policy perfectly addressed, a serious potential issue makers must be aware of when appraising the of “writing-up” bias cannot be addressed. design of proposed corridor projects. That is, authors may refrain from writing Building on the findings from chapters 3 up results that are statistically insignificant, and 4, the next chapter starts a new section, even in an initial, unpublished draft version illustrating how both qualitative and, more of a paper. important, quantitative appraisals of large- 5. For a discussion of the literature on the impacts of intraurban infrastructure, see scale transport investments could be per- Redding and Turner (2014). formed following the methodological 6. Thus, the literature review excludes a large framework developed by the report. empirical New Economic Geography (NEG) literature on the relationship between mar- ket access and outcomes of either wages or NOTES GDP per capita. For a meta-analysis of this lit- erature, see Bosker and Garretsen (2010). It 1. Savings on transport costs also form the focus further excludes a large empirical literature in of the social savings approach to assessing macroeconomics that builds on the work of the impacts of transport infrastructure invest- Aschauer (1989) and analyzes the relationship ments. This approach was originally pio- between a country’s stock of infrastructure, neered by Fogel (1964) in his assessment of including its stock of transport infrastructure, the benefits of the construction of the U.S. and its level of development. For a review of railroad system in the nineteenth century. this literature, see Straub (2011). 2. It is sometimes argued that, even when the 7. The 78 papers reported more than 243 assumptions that underpin traditional cost- results. Typically, any given paper will report benefit analyses break down, the exercise results from numerous regressions. This study remains useful in providing a lower-bound extracted only those results from the most estimate of the net benefits of a transport relevant regressions, which typically consti- infrastructure project. However, this line of tute the authors’ preferred specification(s). argument implicitly assumes that all the wider 8. Before being applied formally, an initial impacts not captured by a traditional cost- dummy run of the double-blind tagging sys- benefit analysis are, on net, positive. This may tem was undertaken using a sample of ten or may not be true. papers. This dummy run was used to both 3. This study attempted to consider economic help train the RAs in the system and resilience—defined as the resilience of 150  The WEB o f T r a n s p o r t C o r r i d o r s i n S o u t h A s i a identify any initial obvious areas for 17. Increases in levels of real income and jobs are improvement. The results from this dummy considered to be beneficial, while increases in run were discarded, and the papers were rates of poverty and inequality are consid- reviewed again. ered to be detrimental. 9. Papers that focus on the evaluation of 18. A particularly interesting paper in this regard is China’s National Expressway Network (NEN) that by Pereira, Hausman, and Pereira (2014). include Roberts et al. (2012); Faber (2014); The authors analyze the overall impact of rail- and Bosker, Deichmann, and Roberts (2015). road investment on economic growth in the China has also been building an extensive United States in the mid-1800s. Their method- high-speed railway network, which has ological approach—a bivariate, dynamic time attracted some attention in the literature. See, series, based on the use of a vector autoregres- for example, Wang and Wu (2015). sive (VAR) model—is very different from the 10. Papers that focus on India’s GQ include cross-sectional approach used in most of the Alder (2015); Ghani, Goswami, and Kerr literature. This approach allows the authors to (2016, 2017); and Khanna (2016). distinguish short-term, demand-side effects on 11. With the exception of a few papers for Africa, economic growth from longer-term, supply- such as Buys, Deichmann, and Wheeler (2006); side effects. and Jedwab and Storeygard (2016). 19. Again, some care in interpretation is required 12. Examples of such studies are Cosar and Demir here. Although a subnational region may (2016); and Martincus and Blyde (2013). experience a decline in population as a result 13. A very recent innovation in the literature has of a transport project that improves its con- been to use subnational “grid cells” of uni- nections to other regions, it is not necessarily form area as units of analysis (Ali et al. 2015; clear that this makes the region a “loser” Jedwab and Storeygard 2016). from a welfare and productivity perspective. 14. Economic geography models of trade include A loss in population may stimulate an overall both “new economic geography” models increase in productivity and real wages for (Krugman 1991a, 1991b; Fujita, Krugman, the region if, for example, it helps ease con- and Venables 1999) and Ricardian models of gestion forces within the region. internal trade of the Eaton-Kortum (2002) 20. A good example is the paper by Bosker, variety. Examples of papers that are moti- Deichmann, and Roberts (2015), which ana- vated by a NEG model include Roberts et al. lyzes both the national and spatial impacts of (2012); and Bosker, Deichmann, and Roberts the National Expressway Network (NEN) in (2015). Papers motivated by an Eaton- China on levels of real income, while simulta- Kortum–style Ricardian trade model include neously considering how these impacts have Donaldson (2010); Donaldson and Hornbeck been influenced by restrictions on migration (2016); Alder (2015); and Alder, Roberts, associated with the country’s permanent and Tewari (forthcoming). household registration (Hukou) system. 15. Baum-Snow et al. (2016) provide a compari- 21. Ideally, we would like to run our meta-­ son of results for the impacts of the analysis by looking only at average effects on construction of China’s NEN obtained using subsets of papers with more comparable out- reduced-form estimation and a structural comes and independent variables. However, model—specifically, the Eaton-Kortum model. the huge heterogeneity observed in papers In doing so, they demonstrate that “tech- and interventions limits this avenue. nique matters”—that the results obtained 22. Thus, estimates of treatment effects based on depend fundamentally on the approach CGE models are dropped from the sample adopted. They also provide evidence to show because these estimates lack accompanying that the Eaton-Kortum model misses some t-statistics. quantitatively important features that are 23. This could also be attributable to a relatively evident in the data. greater precision of OLS estimates, especially 16. Redding and Turner (2014) identify three if valid (exogenous) instruments are less rel- main different types of IV-strategy that are evant (weak). employed in the literature: the “planned 24. The full results from the estimation ordered- route IV,” the “historical route IV,” and the probit model are presented in annex 4E, “inconsequential units approach.” annex table 4E.1 (online). For comparison LE A RNING FROM THE LITER A TURE A BOUT THE ESTIM A TED WIDER   151 with the OLS results, the signs of the esti- Baum-Snow, N. 2007. “Did Highways Cause mated coefficients were flipped, and only Suburbanization?”  Quarterly Journal of results for the model with final outcomes and Economics 122 (2): 775–805. policy variables are reported. Baum-Snow, N., J. V. Henderson, M. A. Turner, 25. Revoltella et al. (2016) provide evidence of Q. Zhang, and L. Brandt. 2016. “Highways, the catalytic role that transport infrastructure Market Access and Urban Growth in China.” plays when it comes to linking local busi- Working Paper C-89114-CHN-1, International nesses with global growth opportunities dur- Growth Centre, London School of Economics ing periods of both positive and negative and Political Science. economic shocks. This paper was brought Berg, C. N., U. Deichmann, Y. Liu, and H. Selod. to the authors’ attention only after the analy- 2017. “Transport Policies and Development.” sis on which this chapter is based had been Journal of Development Studies 53 (4): 465–80. completed. Bosker, E. M., U. Deichmann, and M. Roberts. 2015. “Hukou and Highways: The Impact of China’s Spatial Development Policies on Urbanization and Regional Inequality.” Policy REFERENCES Research Working Paper 7350, World Bank, Alder, S. 2015. “Chinese Roads in India: The Washington, DC. Effect of Transport Infrastructure on Economic Bosker, E. M., and H. Garretsen. 2010. “Trade Development.” Working Paper, University of Costs in Empirical New Economic Geography.” North Carolina, Chapel Hill. Papers in Regional Science 89 (3): 485–511. Alder, S., M. Roberts, and M. Tewari. Forthcoming. Bougna, T., M. Melecky, M. Roberts, and Y. Xu. “The Effect of Transport Infrastructure on Forthcoming. “Transport Corridors and Their India’s Urban and Rural Development.” Wider Economic Benefits: A Critical Review of Ali, R., A. F. Barra, C. N. Berg, R. Damania, the Literature.” Policy Research Working Paper, J. D. Nash, and J. D. Russ. 2015. “Transport World Bank, Washington, DC Infrastructure and Welfare: An Application to Buys, P., U. Deichmann, and D. Wheeler. 2006. Nigeria.” Policy Research Working Paper 7271, “Road Network Upgrading and Overland World Bank, Washington, DC. Trade Expansion in Sub-Saharan Africa.” Arman, S. A., and A. T. Izady. 2015. “Design of a Policy Research Working Paper 4097, World CGE Model to Evaluate Investment in Bank, Washington, DC. Transport Infrastructures: An Application for Card, D. E., J. Kluve, and A. M. Weber. Iran.”  Asian Economic and Financial 2009. “Active Labor Market Policy Review 5 (3): 532–45. Evaluations: A Meta-Analysis.” CESifo Aschauer, D. A. 1989. “Is Public Expenditure Working Paper 2570, Center for Economic Productive?” Journal of Monetary Economics Studies, Munich. 23: 177–200. Chandra, A., and E. Thompson. 2000. “Does Asturias, J., M. García-Santana, and R. Ramos Public Infrastructure Affect Economic Activity? Magdaleno. 2016. “Competition and the Evidence from the Rural Interstate Highway We l f a r e G a i n s f r o m Tr a n s p o r t a t i o n System.”  Regional Science and Urban Infrastructure: Evidence from the Golden Economics 30 (4): 457–90. Quadrilateral of India.” CEPR Discussion Cos¸ ar, A. K., and B. Demir. 2016. “Domestic Road Paper DP11283, Centre for Economic Policy Infrastructure and International Trade: Research, London. E v i d e n c e f r o m Tu r k e y. ”   J o u r n a l o f Atack, J., and R. A. Margo. 2011. “The Impact of Development Economics 118: 232–44. Access to Rail Transportation on Agricultural Cropper, M., J. Puri, C. Griffiths, E. B. Barbier, and Improvement: The American Midwest as a Test J. C. Burgess. 2001. “Predicting the Location of Case, 1850–1860.” Journal of Transport and Deforestation: The Role of Roads and Protected Land Use 4 (2): 5−18. Areas in North Thailand.” Land Economics Banerjee, A., E. Duflo, and N. Qian. 2012. “On the 77 (2): 172–86. Road: Access to Transportation Infrastructure Damania, R., and D. Wheeler. 2015. “Road and Economic Growth in China.” Working Improvement and Deforestation in the Congo Paper 17897, National Bureau of Economic Basin Countries.” Policy Research Working Research, Cambridge, MA. Paper 7274, World Bank, Washington, DC. 152  The WEB o f T r a n s p o r t C o r r i d o r s i n S o u t h A s i a Dasgupta, S., and D. Wheeler. 2016. “Minimizing Economic Policy Research, London School of Ecological Damage from Road Improvement in Economics and Political Science. Tropical Forests.” Policy Research Working Haines, M. R., and R. A. Margo. 2006. “Railroads Paper 7826, World Bank, Washington, DC. and Local Economic Development: The United Donaldson, D. 2010. “Railroads of the Raj: States in the 1850s.” NBER Working Paper Estimating the Impact of Transportation 12381, National Bureau of Economic Research, Infrastructure.” NBER Working Paper 16487, Cambridge, MA. National Bureau of Economic Research, Herrendorf, B., J. A. Schmitz Jr., and A. Teixeira, Cambridge, MA. A. 2012. “The Role of Transportation in Donaldson, D., and R. Hornbeck. 2016. U.S. Economic Development: 1840– “Railroads and American Economic Growth: 1860.” International Economic Review 53 (3): A ‘Market Access’ Approach.” NBER Working 693–716. Paper 19213, National Bureau of Economic Jedwab, R., and A. Storeygard. 2016. “The Research, Cambridge, MA. Heterogeneous Effects of Transportation Eaton, J., and S. Kortum. 2002. “Technology, Investments: Evidence from Sub-Saharan Africa Geography, and Trade.” Econometrica 70 (5): 1960–2010.” Working Paper, Oxford 1741–79. University and London School of Economics Faber, B. 2014. “Trade Integration, Market Size, and Political Science. and Industrialization: Evidence from China’s Khanna, G. 2016. “Road Oft Taken: The Route to National Trunk Highway System.” Review of Spatial Development.” Presented at the Economic Studies 81 (3): 1046–70. Michigan, the Indian Statistical Institute, Fogel, R. W. 1964. Railroads and American NEUDC (MIT), Center for Global Economic Growth: Essays in Econometric Development. History. Baltimore: Johns Hopkins University Krugman, P. 1991a. “Increasing Returns and Press. Economic Geography.” Journal of Political Fujita, M., P. R. Krugman, and A. J. Venables. Economy 99 (3): 483–99. 1999.  The Spatial Economy: Cities, Regions ———. 1991b. Geography and Trade. Cambridge, and International Trade. Cambridge, MA: MA: MIT Press. MIT Press. Laird, J. J., and A. J. Venables. 2017. “Transport Gapanovich, V. A., I. B. Shubinsky, E. N. Investment and Economic Performance: A Rozenberg, and A. M. Zamyshlyaev. 2015. Framework for Project Appraisal.” Transport “System of Adaptive Management of Railway Policies 56: 1–11. Transport Infrastructure Technical Maintenance Martincus, C. V., and J. Blyde. 2013. “Shaky (Urban Project).” Reliability: Theory & Roads and Trembling Exports: Assessing the Applications 10 (2): 30−41. Trade Effects of Domestic Infrastructure Using Gertler, P. J., M. Gonzalez-Navarro, T. Gracner, a Natural Experiment.” Journal of International and A. D. Rothenberg. 2014. “The Role of Economics 90 (1): 148–61. Road Quality Investments on Economic Martincus, C. V., J. Carballo, and A. Cusolito. Activity and Welfare: Evidence from Indonesia’s 2014. “Routes, Exports, and Employment in Highways.” Unpublished. Developing Countries: Following the Trace of Ghani, E., A. G. Goswami, and W. R. Kerr. 2016. the Inca Roads.” http://econweb.umd​ “Highway to Success: The Impact of the Golden .edu/~carballo/Routes_Incas_Exports.pdf. Quadrilateral Project for the Location and Michaels, G. 2008. “The Effect of Trade on the Performance of Indian Manufacturing.” World Demand for Skill: Evidence from the Interstate Bank Economic Review 126: 317–57. Highway System.” Review of Economics and ———. 2017. “Highways and Spatial Location Statistics 90 (4): 683–701. within Cities: Evidence from India.” World Murakami, J., 2015. “Rail Plus Property Program, Bank Economic Review 30 (Supplement 1): Hong Kong SAR, China.” Chapter 3 in S97–S108. Financing Transit-Oriented Development with Gibbons, S., T. Lyytikäinen, H. Overman, and Land Values: Adapting Land Value Capture in R. Sanchis-Guarner. 2016. “New Road Developing Countries, edited by H. Suzuki, Infrastructure: The Effects on Firms.” CEPR J. Murakami, Y.-H. Hong, and B. Tamayose. Discussion Paper DP11239, Centre for Washington, DC: World Bank. LE A RNING FROM THE LITER A TURE A BOUT THE ESTIM A TED WIDER   153 Pereira, R. M., W. J. Hausman, and A. M. Pereira. Sorensen, A. 2000. “Land Readjustment and 2014. “Railroads and Economic Growth in the Metropolitan Growth: An Examination of Antebellum United States.” Working Paper Suburban Land Development and Urban 153, Department of Economics, College of Sprawl in the Tokyo Metropolitan Area.” William & Mary, Williamsburg, VA. Progress in Planning 53 (4): 217−330. Pfaff, A., J. Robalino, R. Walker, S. Aldrich, Stanley, T., C. Doucouliagis, and S. B. Jarrell. 2008. M. Caldas, E. Reis, S. Perz, C. Bohrer, E. Arima, “Meta-Regression Analysis as the Socio- W. Laurance, and K. Kirby. 2007. “Road economics of Economics of Research.” Journal Investments, Spatial Spillovers, and of Sociology-Economics 37 (1): 276−92. Deforestation in the Brazilian Amazon.” Journal Straub, S. 2011. “Infrastructure and Development: of Regional Science 47 (1): 109–23. A Critical Appraisal of the Macro-Level Redding, S. J., and M. A. Turner. 2014. Literature.” Journal of Development Studies “Transportation Costs and the Spatial 47 (5): 683–708. Organization of Economic Activity.” NBER Suzuki, H., J. Murakami, Y. H. Hong, and Working Paper 20235, National Bureau of B. Tamayose. 2015. Financing Transit-Oriented Economic Research, Cambridge, MA. Development with Land Values: Adapting Revoltella D., P. Brutscher, A. Tsiotras, and Land Value Capture in Developing Countries. C. Weiss. 2016. “Linking Local Business with Washington, DC: World Bank. Global Growth Opportunities: The Role of Vickerman, R. 2007. “Cost-Benefit Analysis and Infrastructure.” Oxford Review of Economic Large-Scale Infrastructure Projects: State of the Policy 32 (3): 410–30. Art and Challenges.” Environment and Planning Roberts, M., U. Deichmann, B. Fingleton, and B: Planning and Design 34 (4): 598–610. T. Shi. 2012. “Evaluating China’s Road to Wang, Y., and B. Wu. 2015. “Railways and the Prosperity: A New Economic Geography Local Economy: Evidence from Qingzang Approach.”  Regional Science and Urban Railway.”  Economic Development and Economics 42 (4): 580–94. Cultural Change 63 (3): 551−88. Smolka, M. O., 2013. Implementing Value Capture Weinhold, D., and E. Reis. 2008. “Transportation in Latin America: Policies and Tools for Urban Costs and the Spatial Distribution of Land Development . Cambridge, MA: Lincoln Use in the Brazilian Amazon.”  Global Institute of Land Policy. Environmental Change 18 (1): 54–68. Spotlight 4 The Impact of Highways on Micro, Small, and Medium Enterprises: Anecdotal Evidence from Bhutan, Sri Lanka, and India U pgraded or reopened highways have successfully linked major commercial centers and trade corridors, and have eased the team visited the outskirts of Delhi and Mumbai and spoke with MSME owners operating along the major highways. congestion in South Asia. These include the Participants included women and men 5,800-kilometer (km) Golden Quadrilateral who grow and sell fruits, vegetables, plants, (GQ) highway in India, which links four and flowers, or make sweets, yogurt, and major metropolitan areas; the A9 highway in coconut oil. Others weave or sew fabrics and Sri Lanka, which connects the Sri Lankan cit- clothes, or make handbags, shoes, mosquito ies of Kandy in the south to Jaffna in the nets, furniture, and handcrafts. Some own north (photo S4.1); and the Thimphu- small and medium-sized transport, construc- Phuntsholing and Thimphu-Gelephu high- tion, and food/clothing export companies. ways in Bhutan, which connect the country’s capital to the Indian border. BENEFITS To study the impact of these highways on micro, small, and medium-size enterprises Saving time. While their stories vary, the (MSMEs), the World Bank held a series of entrepreneurs all agreed that the highways, focus group meetings and interviews. In Sri which are now reliable, save them time in Lanka, the team travelled over 150 kilometers obtaining inputs and delivering final goods. along the A9 route and interviewed With shorter travel times, business owners 52 MSMEs in the towns of Dambulla, Naulla, can also travel further and gain access to a Ipalogama, and Vavuniya. In Bhutan, it held greater variety of inputs than before. Some three sessions in the towns of Hongtsho, said the improved highways have reduced the Jemina, and Babesa, where 36 entrepreneurs costs of transporting inputs, and brought new took part (21 men and 15 women). In India, customers and business opportunities. 155 156   THE WEB OF TR A NSPORT C ORRIDORS IN SOUTH A SI A PHOTO S4.1  Spice producers use the A9 Highway in Sri Lanka degree, who inherited a bronze-casting busi- to dry spices ness, said, “I use the Thimphu-Phuntsholing highway to get my raw materials from India. Since I can use larger trucks and place big- ger orders, I lower unit costs and at the same time meet my clients’ growing demand.” The owner of a flower shop in Dambulla (Sri Lanka) reported that the flower pots and flowers she ships are no longer damaged. Tashi Dorji, a carpenter from Hongtsho (Bhutan), said some of his products used to break en route to his cus- tomers; now this rarely happens. Saving transaction costs. Some entrepreneurs have profited in unexpected ways. For example, rice growers in Vavuniya (Sri Lanka) say buyers Source: Martin Melecky, World Bank. now come directly to their farms, and their profits have increased because they can super- vise the weighing of the rice—which eliminates The highways also shifted the geographic cheating by middlemen! locations of firms to some extent—with firms Expanded variety of inputs and products. often clustering along the highways. The improved highways have enabled more For example, Choki Wangmo, who pro- types of products to be brought into the local duces wooden bowls and cups in Hongtsho markets. For example, more varieties of seeds (Bhutan), noted, “Before, we hired horses and and textile inputs are available in Dambulla walked days to get raw materials. Now we (Sri Lanka) and entrepreneurs can more easily do this in a few days with public transport.” modify and invent new products with the new Mr. Kabir, who manages the Ongdi Wood inputs. Industry, said the upgraded highway has Accessing markets more easily and getting greatly helped businesses in Khasadrapchu better prices. The improved roads also make (Bhutan). “We can now hire an 18-wheeler to it easier to access markets. One man in transport raw materials as well as our end Ipalogama (Sri Lanka) has been producing products.” He also has more control over the and selling curd in his local area since 2006. time he spends obtaining inputs and can Now, he uses public transport on the A9 to deliver goods just in time. deliver his product as far as 60 kilometers Saving fuel and vehicle maintenance costs. away. As his business grew, he hired four Better roads increase fuel efficiency. Six of the more employees. In Bhutan, the highways eight participants in one focus group in have also cut marketing and distribution Hongtsho, who own their vehicles, said their costs. Tashi Dorji, a carpenter in Hongtsho, private vehicles cost less to maintain and run reported, “I now have customers coming because they have fewer problems on the from the capital of Thimphu to my produc- road and use less fuel. Four of the eight said it tion site to buy their products.” Equally now costs less to transport raw materials, important, the improved access from the vil- although most cannot calculate their exact lages to towns means some women have elim- savings and profits. inated the middlemen and now take their Saving input costs. The improved trans- vegetables directly to the market, instead of portation system also helps some entrepre- having a distributer pick them up at their neurs save on input costs. Pema Dorji, a farms. By arriving at the market early, they 29-year-old Bhutanese with a graduate can sell when prices are the highest. THE IMP A C T OF HIGHW A Y S ON MI C RO , SM A LL , A ND MEDIUM ENTERPRISES   157 Spurring new business ventures. Business PHOTO S4.2  The World Bank team speaks with a textile shop opportunities have also grown. In India, once owner (second from right) in Thane, India the GQ highway was improved, new housing was constructed along the corridor. One women from the Mumbai area said that after the houses were built, her son started a busi- ness installing CCTVs (closed circuit televi- sion) in them. In Bhutan, building on local traditional craft techniques, new business lines such as bronze casting, saw mills, and furniture making started in Khasadrapchu. Expanding financing. Some financial insti- tutions have opened branches in the towns and helped locals start businesses. In the Vavuniya (Sri Lanka) area, one exporter of rice-based products said a new local bank helped him develop his business, lent him funds, and connected him to clients. Today, he gets inputs from 3,000 farmers, has 30 Source: Yan (Sarah) Xu, World Bank. employees, has wholesale dealers who export his products to Germany, and has opened new production sites in Jaffna, Colombo, and demand, so I want to expand my workshop Kandy. ‘Without the financial institution and am negotiating with the landlord for moving to the area, I wouldn’t have started more space.” He also plans to hire a few more my business,” he noted. employees from his area. Attracting more customers. In addition, the better roads have attracted more business. PROBLEMS AND POLICIES Some firms have located their shops/stalls closer to the highway, where they attract Besides the positive impacts, focus group par- more customers in transit. As one entrepre- ticipants also described some problems. neur in Vavuniya (Sri Lanka) reported, “the Constraints on land. For example, since closer you are to the A9, the more prosperous demand for land near the highway has your business, since people driving on the grown, less land is available for the MSMEs’ highway stop and buy your products.” needs. Dechem M., a carpenter in Bhutan, Expanding the scope of business activities. said that because the highway increased his The improved roads have also changed the business, he needs more land to expand, but geographic scope of various companies. it is in short supply. Parking spaces are in An Indian bank manager said the improved short supply along A9 in Dambulla GQ has attracted more economic activities to (Sri Lanka). Small shop owners there say that Thane, which is near Mumbai and is close to more travelers would stop to buy goods if the airport and cheaper land (photo S4.2). parking was better. In Bhutan, of 29 focus group participants, Traffic congestion and environmental 7 moved their businesses closer to the high- trade-offs. For example, in Dambulla way. Dechen Zangpo, who makes religious (Sri Lanka), traffic was greater than expected statues in Hongtsho and who relocated closer and the road became quite congested. to the highway, said, “From the time I shifted Travelers started to use another parallel my business location, my client base has highway to avoid traffic. The local shops increased. Five to seven more clients place ended up getting less transit customers than orders every week. I can’t keep up with the they used to have. In Bhutan, the surge in 158   THE WEB OF TR A NSPORT C ORRIDORS IN SOUTH A SI A regional tourists—who, unlike high-end portability and taste. He designed the pack- international tourists, do not use guides and age himself and produced the products at drive their own vehicles—increased conges- home with his family’s help. However, to tion.1 Local MSMEs must spend more time expand the business beyond local markets, travelling for their businesses, which is he still needed to improve his marketing diminishing their productivity. Moreover, skills. Some firms participate in training pro- communities are raising concerns about vided by the government and private institu- environmental pollution due to unguided tions. A lady from Iplogama (Sri Lanka) and unregulated regional tourism. who supplies plants to an export company Landslides and other natural disasters. was selected to attend a training program MSMEs also need the government to help run by a community bank, along with 14 them cope with natural disasters. For exam- other people. The community bank issued a ple, in Bhutan, landslides are still an issue. certificate to her firm after she completed Before the highway was upgraded, the road the program and connected her firm to sup- was narrow, uneven, and full of potholes. pliers. Of the 15 trainees, 10 started planta- Participants said the soil along the roads often tion businesses with loans provided by the eroded, which caused landslides (particularly community bank. Despite the effectiveness in summer), blocking the roads and stranding of such training programs, they are available commuters, sometimes for days. Sonam only to a few people. More programs are Jamtsho, who runs a mid-sized factory in needed. Khasadrapchu (Bhutan) that makes uncoated Difficulty connecting to more distant mar- calcium carbonate, said that in the past, land- kets. While the improved roads make it easier slides damaged his shipments. While he enjoys to obtain inputs, they do not necessarily link the benefits from the improved road, the issue entrepreneurs to clients in other parts of the of landslides has not yet been resolved. country or world. Most focus group partici- Need for improvements in feeder roads. pants said it was still difficult to enter new Some focus group participants say that besides markets in the major cities and supply large the highway upgrade, more needs to be done exporters. At the local level, many would like to improve the poor conditions on rural feeder to have a town market where they can sell roads. One entrepreneur, who runs a whole- products and attract more customers. Rural sale fish business and employs 35 people in MSMEs do not seem to be involved in e-com- the Kurunegala area (Sri Lanka), says while merce—which could help them find new more customers now take the A9 to reach clients and expand their businesses to markets him, fishermen who supply his shop must still that are further way in their own country and use the rural roads to bring him their catch. across borders. During the rainy season, some roads are not Limited access to finance. Many focus passable. The rough conditions cause vehicles group participants said that with more capital, to break, which raises transport costs. Poor they could buy inputs and stock inventories in feeder roads make it difficult for firms to get larger quantities, increase production/sales, inputs, distribute goods, and reach clients. and lower unit costs. One woman who runs a Need for more training . While large tourist bus service in Sri Lanka said could export firms, the government, and some pri- increase her business if she could buy another vate institutions offer training, the overall bus to meet the demand—but lacks the credit level of technical, management, and market- to do so. ing skills among MSMEs is low and prevents Increased competition. In some areas, the some firms from expanding. For example, a improved transport links have allowed new producer of spice powder producer from businesses to enter the local markets, which Naulla (Sri Lanka) changed the traditional has increased competition. In the worst cases, spicy paste into powder to improve its smaller businesses have been crowded out and THE IMP A C T OF HIGHW A Y S ON MI C RO , SM A LL , A ND MEDIUM ENTERPRISES   159 prices have been driven down. For example, a While the MSMEs benefitted from the rice-based product exporter said that after the improved highways to some extent, comple- A9 highway reopened in Sri Lanka, more busi- mentary polices need to be pursued to help nesses competed for the local market and sold them grow. Improving the availability of land, at lower prices. A caterer reported that competi- putting in place measures to recover from nat- tors moved into the area, selling food services at ural disasters, and upgrading rural feeder a lower rate, and he has lost market share. roads can boost their productivity further. Need to attract foreign investment. The Warehouse receipt financing can be used as improved highways, by themselves, do not collateral so farmers can borrow to meet attract foreign investment. Governments will their business financing needs. Training, need to develop smarter policies to promote it. commerce, and town markets are crucial to e-­ For example, when exports increase, ware- help MSMEs reach more clients. houses are needed for storage and value-added services such as packaging and consolidation. State-level governments have offered incen- NOTE tives, such as tax credits, although such poli- 1. See http://www.kuenselonline.com/need-to​ cies need to undertaken carefully. -regulate​-regional-tourism/. Spotlight 5 Do Highways Help Women? D awa Zam, who used to be an accountant for an Indian construc- tion company at the hydropower construc- and now pass through the smaller towns and villages and extend to India’s northeastern border. tion site in Wangdue (Bhutan), has now Before, roads were narrow, uneven, and full moved to Gidagom and has a shop at the of potholes. Sometimes they were blocked, capital city of Thimphu. She said the upgrade stranding commuters for days (in Bhutan). of the Thimphu-Phuntsholing highway Public transport from the towns and villages to “offered new opportunity for me to earn a the larger cities was limited. Thus, most women decent living from the grocery shop in did not travel to urban centers or commute to Thimphu and grow vegetables on my moth- jobs, which limited them to work that was er’s farmland at Kasadrapchu. I am able to available nearby. If they did make the trips, save rent in Thimphu by staying at Gidagom, travel was long and tiring, due to poor road which is adequately linked by public trans- conditions. port services.” She noted, “I am now able to To study the impact of the highways, the start my own family and be independent.” World Bank held focus group discussions Several thousand kilometers of recently from February to May 2017 with women opened or upgraded highways in Bhutan, Sri (many of whom are microentrepreneurs) in Lanka, and India have dramatically changed several communities, mainly in Bhutan and the lives of women like Dawa Zam and their Sri Lanka. families. In Bhutan, 27 women attended the groups In Sri Lanka, the A9 highway, which was that focused on women’s issues (photo S5.1), closed during the civil war, was reopened and while in Sri Lanka, 48 women attended now links towns and villages along a route (photo S5.2). To compare women’s responses from the city of Kandy in the south to Jaffna with those of men, the focus group organizers in the north. invited 12 men to participate in Bhutan. In India, the Golden Quadrilateral (GQ), a Although almost all the women interviewed four-to-six lane highway (with two and three were housewives, they also work in unpaid lanes in each direction), now connects the family businesses or run small shops to earn country’s four major metropolitan areas— an income. The women included small Delhi, Mumbai, Chennai, and Kolkata. ­ f armers, weavers, dressmakers, furniture In Bhutan, the Thimphu–Phuntsholing and makers, vegetable sellers with stalls along the Thimphu–Gelephu highways were upgraded highways, and civil servants, or those who 161 162   THE WEB OF TR A NSPORT C ORRIDORS IN SOUTH A SI A PHOTO S5.1  Focus group in Khasadrapchu, Bhutan Most of the benefits that the women report have enriched their lives are similar in the three countries. The women said that it is eas- ier and takes them less time to travel, conduct business, shop for food and household items, and take their children to schools and hospi- tals. This means they have more time to do household chores and engage in income-pro- ducing work, or to work more hours. For example, in Bhutan, the trip from Phuntsholing to Thimphu, which used to take eight hours, now takes five, while the trip from Hongsho to Thimphu, which took nearly an hour, has been cut in half. In Sri Lanka, some of the focus group par- ticipants live in a village that is 8 kilometers from the town of Vavuniya. The trip, which used to take 1½ hours, is now just 30 min- Source: Martin Melecky, World Bank. utes. The trip from the town of Ipalogama (on the A9 highway) to Colombo was 5 hours each way and is now just 3½ hours. PHOTO S5.2  Focus group in Naulla, Sri Lanka The reduced travel times also mean women can leave their children with a neighbor, do errands, and return quickly; in addition, they can make the trips and return home the same day. Since these trips would have involved an overnight stay in the past, the women did not make them. With the better road conditions, public buses and taxis run more frequently. For example, in Dambulla (Sri Lanka), buses, which formerly operated only during the day, now also run at night. Thus women travel more often, taking buses locally for jobs, obtaining social services, or riding to cities to attend school or find better-paying jobs. For example, Dema, who lives in Hongtsho vil- lage (Bhutan), says she now takes public transport to Thimpu, since buses run twice a Source: Martin Melecky, World Bank. day, when there were none before. Tshering Yangchen said the frequency of travel has greatly increased: “About 16 residents from rent space in their homes to tenants. Some of Hongtsho travel every day to Thimphu for the focus group participants had no educa- work and return in the evening on a city bus. tion, while others had completed either Even students travel to Thimphu schools on primary or secondary school. ­ public transport and return in the evening,” Almost all said the improved highway had because neither Hongtsho nor Kasadrapchu helped them significantly—although they have schools that go beyond level 10. noted some negative effects and offered sug- The women report that they now travel to gestions for improvements to the authorities. the cities for entertainment or to visit their D o H i g h w ay s H e l p W o m e n ?   163 families and friends more often, which Vavuniya (Sri Lanka), vendors transported improves the quality of their lives. They have water to towns and villages on unpaved greater access to social services, since travel to roads. Dirt often got into the water supply, so the cities is easier, and in Sri Lanka, local gov- the government banned the transportation of ernments now offer some services in the the water into the area. Over time, villagers smaller towns. developed kidney disease from drinking the Safety is another plus. The roads are polluted water from wells. But since the wider, and children can walk to school more A9 reopened, water has been transported on easily. In Sri Lanka, women said their chil- better roads and is no longer contaminated by dren were sometimes attacked because the dust or dirt. Further, with the reopening of roads were isolated. Now, because of the the A9, the government now sends public increased activity along the highways, fami- health staff to the rural areas to inform resi- lies are more comfortable sending their chil- dents about the importance of clean water, dren to school alone. In Dambulla (Sri the effects of polluted water, and how to Lanka), where the road was dark and empty purify untreated water. after 6 p.m., women now feel safe traveling The improved road links have also alone at night because of the increased traffic increased some land values, particularly of on the road and extended night-time bus property located closer to the highway. This schedules. Teachers and other government benefits women who own land because it workers also benefit from the safer and faster helps them generate rental income and travel. For example, female government increases their accessibility to credit—as they workers from Jaffna (Sri Lanka) were reluc- can use their property as collateral. In tant to take jobs in Vavuniya due to safety Bhutan, although inheritance practices vary concerns (with not many people on the by region, in the western part of the country, road). Now, with the more efficient travel, where Thimphu is located, matrilineal prac- female teachers not only are more willing to tice is common; thus, the oldest daughter work there but also can get there faster and usually inherits the agricultural land and arrive on time. other property (such as livestock). Several In general, the women are involved in women say they own property that they or microbusinesses, such as growing and selling their mothers converted into apartments or vegetables and fruit, or raising chickens to commercial buildings to rent or lease. produce and sell eggs near the highways. Some benefits are unique to certain The increased traffic from the highway has locales. For example, in Sri Lanka, private brought more customers to their businesses. businesses and the government moved into Although few in the focus groups own the Vavuniya area after the road reopened, vehicles, those who do say they spend less on which created new jobs. One woman said oil and maintenance because of the improved this allowed her daughter to find a good job, road conditions. Women usually do not drive locally. themselves or drive only within the city Besides these positive effects, partici- where they live. Instead, their husbands take pants also described some negative impacts. them to places that are farther away. Thus, They note that there is more traffic, more while it seems that the availability and qual- congestion, more noise, more air pollution ity of public transport have improved the (from exhaust) due to the big increase in mobility of women more than men, given vehicles, more litter caused by drivers social norms, this mobility is still limited to throwing trash out their windows, and local commutes and to public transport for more deforestation, which has led to a gen- longer distances. eral loss of wildlife habitat. In addition, The improved and reopened roads are also Manlice, who practices traditional medicine improving villagers’ health and the environ- in Naulla (Sri Lanka) and the surrounding mental quality of the area. Previously in areas, reported that since the A9 reopened, 164   THE WEB OF TR A NSPORT C ORRIDORS IN SOUTH A SI A she finds that more children have respira- the feeder roads were upgraded, this tory problems because pollution from the would improve the quality of life and busi- traffic has increased. ness operations, and allow people greater People also drive more rapidly (especially access to their temples, which are an truck drivers, at night) than when the roads important part of the culture. In Bhutan, were in poor condition and they had to be focus group participants noted the impor- cautious. Thus, more accidents are occur- tance of allowing drivers to make U-turns ring—two or three a month, mainly from in certain spots. heavy vehicles hitting small cars—and some- Some said that to help working mothers, times individuals and animals along the roads the government should offer child care in its and at crossings are killed. Sonam Pelden, a offices or centers. Tshering Lham told the resident of Hongtsho (Bhuta), says that group “I have a 22-month old daughter, and because of the speeding, she worries about as a working mother, it’s hard to take care of her children, who walk to school on the high- my child and work at the same time. But if way. Rinzin, who lives in Babesa (Bhutan), there were child care centers in offices or says there are more drunk drivers and crime near highways, this would make the infra- at night along the highways. structure projects more valuable, particularly Some prices have also increased. For exam- for working women like me.” At present, ple, Tshering Lhaden, a civil servant who there are only a few centers, which are far rents an apartment in Babesa (Bhutan), says from the mothers’ jobs, or the children are that because the highway has made travel sent home before the women finish work. between her town and the city easier, “my Tshering Lham added that workplaces rent has increased twice in the past two should offer flexible hours for mothers, years.” which would encourage more to join the When asked what the authorities could do workforce. to resolve the new problems, the groups had Often, women in both countries said they various suggestions, depending on where they would like to be trained with more skills. lived. For example, many sew garments while car- In Sri Lanka, to reduce speeding and ing for their children at home. However, few accidents, the women said that speed acquire technical or marketing skills, and bumps should be built into the roads to thus cannot sell to exporters. Although most slow the traffic, especially near schools. of the housewives want to increase their Ts h e r i n g Ya n g c h e n f r o m H o n g t s h o income-producing activities, they do not (Bhutan) suggested that the government have the skills needed to expand their busi- build sidewalks and footpaths along the nesses, nor do they know how to obtain highway to make it safer for pedestrians, them. The government and private institu- particularly children. tions such as banks are providing such train- To make travel even easier, one Sri Lanka ing in Sri Lanka, but it is available to only a focus group noted that the government limited number of women. should also improve local roads. Although Finally, on the basis of the focus group the A9 highway is in good condition, sec- responses, it appears that despite the eco- ondary roads are still poor, which means it nomic improvements and gains in quality of is difficult for women to travel to and life generated by the highway upgrades or from villages. Buses still do not travel reopening, the women in Sri Lanka and more than 2 kilometers off the main road Bhutan, in general, are not more socially because of the poor surface conditions. If empowered. Part III Appraising Corridor Projects 5 Appraising Transport Corridors in Japan, Europe, and Thailand D rawing on the framework that this report proposes, chapter 5 appraises the impact of three transport infrastructure infrastructure investment do not automati- cally trickle down to the more disadvantaged locales and groups of the population. programs in Japan, Europe, and Thailand on Complementary investments and reforms wider economic benefits (WEB). It does so must be implemented in tandem with these by means of a narrative supported by anec- investments to ensure that economic benefits dotal evidence and selected data. are spread most widely. The case study on Japan reviews the Pacific Along with examining the challenges of Belt Zone Initiative proposed by the govern- financing and implementing cross-border ment as an engine of industrialization to dou- rail corridors, the appraisal of Europe’s ble Japan’s national income between 1960 three rail corridors reveals that the place- and 1970 in an equitable way. The case study ment of rail corridors and their alignment on Europe reviews three selected high-speed with important and relevant population rail corridors of the Trans-European and productive centers could be a major Network–Transport (TEN-T), and the diverse determinant of their wider economic ways in which spillovers from these corridors impacts. These impacts take more than 10 helped—or did not help—generate WEB in years to realize. Again, transport infra- their vicinity. The final case study on Thailand structure alone does not deliver WEB unless appraises the role that investments in large- it is part of a comprehensive package of scale transport infrastructure played in help- measures to support wider economic and ing develop a booming automotive industry social development. on the back of several complementary reforms. The appraisal of Japan’s Pacific Belt Zone Initiative illustrates the importance of focusing Transport infrastructure alone does not deliver wider on an equitable diffusion of benefits from economic benefits unless it is part of a comprehensive large-scale investment right from the begin- package of measures to support wider economic and ning. The positive effects of these large social development. 167 168  T h e WEB o f T r a n s p o r t C o r r i d o r s i n S o u t h A s i a The appraisal of Thailand’ success in APPRAISING THE PACIFIC BELT developing a booming automotive industry ZONE INITIATIVE: THE ENGINE FOR portrays infrastructure investments as the JAPAN’S INDUSTRIALIZATION centerpiece of a broader set of proactive mea- The Pacific Belt Zone served as a driving force sures to diversify overly concentrated produc- for Japan’s rapid economic growth in the 1960s. tion near Bangkok. A new deep-sea port, an It extends approximately 1,000 k ­ ilometers from energy production complex, and public utili- the district of Kanto in the southwest to the ties were part of the anchor infrastructure northern part of the district of Kyushu, span- investment. But it was only the investment ning the cities of Fukuoka, Osaka, Nagoya, and incentives and the government’s flexible pol- Tokyo (map 5.1). Industrialization, initially in icy stance on access to finance and foreign the form of recovery from the devastation of direct investment that provided the sufficient World War II, began advancing in the four conditions for the shift to heavy, automotive major industrial zones in this area in the 1950s.1 industry. Environmental trade-offs again However, concerns spread that excessive cen- surfaced and were dealt with only after the tralization and dependence on these major infrastructure was put in place. industrial zones could impair further economic growth. To avoid this, the government of Japan Investments in transport infrastructure—including sought to link the four major industrial zones to high-speed rail, roads, and ports—were part of a form the Pacific Belt Zone, to treat this zone as a policy package ranging from power to health care corridor of broader economic activity, and to and education. encourage development in the entire zone. The Pacific Belt Zone initiative was pro- posed while Japan was deliberating a plan to MAP 5.1  The Pacific Belt Zone runs through Japan’s industrial core double its national income from 1960 to 1970.2 To achieve this goal, Japan needed to triple its national industrial production. Intensive development of the Pacific Belt Zone was considered essential to achieve high growth within this limited period of 10 years. At that time, ports and harbors, highways, railroads, and other infrastructure were being developed in large cities across the Pacific Belt Zone. However, the Income-Doubling Plan concluded that the process of industrial loca- tion should not widen regional disparities, and should underscore efficiency in industrial distribution. The term “Pacific Belt Zone” did not appear in the final version of the plan, allegedly because of political sensitivities. Some districts located outside the zone objected strongly to prioritizing development of the zone because they were concerned they might remain undeveloped under the initiative. The need for an integrated development plan covering the entire country was recog- nized soon after World War II. The Comprehensive National Development Act was established in May 1950. This act aimed Source: Annotations by the Corridor Study Team. to promote social welfare and ensure the A p p r a i s i n g T r a n s p o r t C o r r i d o r s i n J a p a n , E u r o p e , a n d T h a i l a n d    169 comprehensive utilization of national land pavement rate for them went from 18.5 and proper location of industry. Although the percent (1956) to 75.1 percent (1970).4 The Act stipulated that a comprehensive national Meishin Expressway and the Tomei development plan, so called long-term spatial Expressway were fully opened in 1965 and plan to show the ideal state of territory, 1969, respectively, to connect regional should be drawn up. Though this plan also agglomerations in the Pacific Belt Zone. In avoided specific descriptions about regional addition, high-speed railways (Shinkansen, or priorities for industrial location, the Pacific bullet trains), which enabled high-speed mass Belt Zone initiative was considered as a back- passenger transport, were launched between ground of intensive capital formation to spur Tokyo and Osaka in 1964, between Osaka rapid economic growth for the nation. and Okayama in 1972, and between To promote industrialization in the Pacific Okayama and Hakata (Fukuoka) in 1975. Belt Zone, large-scale infrastructure projects Along with public investments in infra- to develop expressways, high-speed railway structure, capital investment by private enter- systems, power plants, ports and harbors, and prises increased. Despite a brief dip in other facilities were intensively promoted. 1965−66 following turmoil in Japan’s securi- Funding this massive investment solely with ties market, capital investments by private domestic sources was not possible because enterprises soared. At the same time, the stock Japan’s level of savings was insufficient. It was of social capital increased rapidly, even on a also difficult for private businesses to make net basis. The stock of social capital was massive capital investments by borrowing remarkably concentrated in the Pacific Belt from financial institutions in Japan. The World Zone.5 Alongside an effort to actively intro- Bank became an important catalyst. World duce foreign technologies, Japan’s crude steel Bank loans totaling US$930.4 million covered production more than quadrupled, from 31 projects, ranging from expressways, ship- 22.1 million tons in 1960 to 93.3 million tons yards, and high-speed railway systems (the in 1970. In 1965, Japan surpassed what was Shinkansen bullet trains) to electric power, pri- then West Germany to become the world’s vate sector industrial plants and factories, and second-biggest crude steel producer after the agricultural development. Virtually the entire United States. Japan’s oil refining volume amount (96.4 percent) lent was used to pro- increased sixfold from 1960 to 1970. This mote industrialization in the Pacific Belt Zone. surge in manufacturing materials strongly Compared to Japan’s annual public invest- supported the manufacture of industrial prod- ment, which ranged from US$1.514 ­ billion in ucts, including cars and machinery. 1955 to US$8.142 billion in 1965, World Overall, capital investment by private Bank lending Japan was not huge. However, enterprises far exceeded public investment. the impact was significant. The loans demon- This trend was more apparent in the Pacific strated the Japanese government’s strong com- Belt Zone than any other region in Japan mitment to develop large-scale infrastructure (figure 5.1). The concentration of petro- projects, while World Bank support signaled chemical, steel, car manufacturing, and other the feasibility of these projects and gave the industries in the Pacific Belt Zone indicated private sector confidence in making long-term that many private enterprises in these indus- investments. tries considered capital investment in the The needs were dire. As the Watkins Pacific Belt Zone to be more economically Report reported in 1956, “The roads of Japan rational than in other parts of Japan. are incredibly bad. No other industrial nation The development of infrastructure invigo- has so completely neglected its highway rated economic activities, which in turn ­s ystem.” 3 The investments dramatically increased the volume of freight transporta- improved road conditions. The improvement tion by truck 6.5 times (from 20.8 billion rate (ratio roads with more than 5.5 m width) ton-​­ kilometers in 1960 to 135.9 billion ton-​ for national roads went from 38.4 percent ­ kilometers in 1970); the volume of cargo han- (1956) to 77.7 percent (1970), while the dling at ports and harbors 4.2 times (from 170  T h e WEB o f T r a n s p o r t C o r r i d o r s i n S o u t h A s i a FIGURE 5.1  Private investment/public investment were higher in the (now known as JFE Steel Corporation); the Pacific Belt Zone than in other regions of Japan, 1960−70 expansion of production of four-wheel vehi- cles by Toyota Motor and Nissan Motor; and 240 the launch of production of four-wheel vehi- cles by Honda Motor and Suzuki Motor. 220 Both the government and the private sector propelled economic development, sometimes 200 competing for limited financial resources such as World Bank loans and governmental 180 finance, and sometimes cooperating to intro- duce foreign technologies or promote efforts Percent to develop technology. 160 High economic growth yielded rapid and huge positive outcomes. While the policy 140 objective had been to double national income within ten years, from 1960 to 1970, 120 this goal was achieved in only seven years. Japan’s economy surpassed what was then 100 West Germany in 1968 to become the second-largest economy in the world. This ­ 60 61 62 63 64 65 66 67 68 69 70 19 19 19 19 19 19 19 19 19 19 19 Pacific Belt Zone Outside the Pacific Belt Zone income increase was constantly led by vast demand of private consumption to improve Source: Calculations by the Corridor Study Team based on Cabinet Office, Government of Japan. people’s standard of living, which was satis- Note: The Pacific Belt Zone consists of 19 prefectures. fied by enhanced production capacity. A ­virtuous cycle was established, in which robust private consumption spurred invest- ment, investment improved technologies, Remarkably, economic and social disparities technological improvements enhanced the between the Pacific Belt Zone and other regions did quality of industrial products and lowered not widen much. Benefits were widely shared. their prices, and consumption increased, boosted by an increase in income. By repeat- ing this cycle, Japanese industries increased 440 million tons in 1960 to 1.852 billion tons their export competitiveness, which in 1970); and the volume of power genera- increased tax revenues and savings in Japan. tion 2.9 times (from 100.1 billion kwh in Since imports exceeded exports through- 1960 to 293.9 billion kwh in 1970).6 out the entire period of high economic The total value of shipments of products growth, the contribution of exports to the manufactured or fabricated in the Pacific Belt increase in income was seemingly limited. Zone increased 3.8 times (from JPY24.7 tril- However, vigorous domestic private con- lion (1960) to JPY94.4 trillion (1970), in 2005 sumption was backed by an increase in prices). The rates of increase in the traditional imports of raw materials, and the financing centers of Tokyo (291.2 percent), Osaka for such imports was underpinned by steadily (327.4 percent), and Fukuoka (251.6 percent) increasing exports. During the 1960s, exports were lower than the average for the entire of machinery, steel, and cars expanded greatly. Pacific Belt Zone. This suggests that industrial This expansion was sustained by the global activity was dispersed by such measures as economy and trade, which grew strongly and regulating factory locations and promoting steadily during the 1960s. development projects across the entire zone. Social indicators also improved signifi- Cases of dynamic private initiative include cantly. For example, Japan’s infant mortality the construction of a steel plant in Chiba rate declined from 30.7 to 13.1 for every Prefecture by Kawasaki Steel Corporation 1,000 births from 1960 to 1970. A p p r a i s i n g T r a n s p o r t C o r r i d o r s i n J a p a n , E u r o p e , a n d T h a i l a n d    171 A growth cycle was established that industrialization in the Pacific Belt Zone supported improvements in living caused rapid population inflows into this standards for most Japanese citizens area. As a result, the total population as well as the working-age population in the Remarkably, during this period of high zone increased greatly. In 1962, 43.7 percent ­ economic growth, economic and social dis- of job-seeking junior high school graduates parities between the Pacific Belt Zone and and 35.2 percent of job-seeking high school other regions did not widen much.7 While graduates in the other regions (235,000 some gaps between the Pacific Belt zone youths) found employment in the Pacific and the other regions persisted because the Belt Zone. This population migration Pacific Belt Zone had always fared better in increased the number of households in terms of indicators such as per capita Japan and strengthened consumer demand income and infant mortality, even rapid for durable goods and other goods and economic growth in the Zone did not services. expand such gaps. As both the Pacific Belt Zone and the other regions made rapid Enhancement of agricultural and improvements, indicators in the other nonagricultural productivity regions reached the same level as the Pacific In parallel with the nation’s industrialization, Belt Zone in several years. the ratio of agricultural and forestry output The public also perceived that the income to the total production in all industries gap was not widening. According to a public declined steadily in both the Pacific Belt Zone opinion poll conducted at the time, the ratio and in the other regions. Meanwhile, agricul- of respondents who considered themselves to tural productivity improved significantly be middle class increased from 76.2 percent because of mechanization, the introduction of in 1958 to 89.2 percent in 1967.8 The pene- chemical fertilizers, and other efforts. As a tration rates of consumer goods such as elec- result, agricultural and forestry output tric washing machines, electric refrigerators, increased from 1960 to 1970. In the other and television sets soared from 21 percent, regions, nonagricultural earnings increased 5 percent, and 24 percent, respectively, to for two reasons. First, job opportunities in approximately 90 percent after the economic industries other than agriculture increased, growth of the 1960s. Then people switched especially in the prefectural capitals, where to buying color television sets, air- urbanization provided a sales base that conditioners, and cars. boosted consumer spending. Second, the pool Three main factors accounted for such bal- of migrant workers employed in agricultural anced development: the demographic divi- off seasons in the Pacific Belt Zone increased, dend and internal migration; enhancement of raising the ranks of part-time farmers. In the agricultural productivity, along with growth other regions, the ratio of full-time farmers in employment opportunities in industries decreased from 36.5 percent (1960) of house- other than agriculture; and government poli- holds to 16.9 percent (1970) and to cies to correct regional disparities. 13.4 percent (1975). With this change, the ratio of part-time farmers whose nonagricul- The demographic dividend and internal tural earnings exceeded 50 percent of their migration respective total earnings increased from During the reconstruction period following 29.6 percent (1960) to 46.6 percent (1970) World War II, the number of births and pop- and to 58.3 percent (1975). In addition, the ulation growth increased remarkably, peak- maintenance of producer prices for rice under ing in the three-year period from 1947 to the foodstuff control system stabilized and 1949. The period of high economic growth sustained farmers’ incomes. However, period coincided with a demographic divi- enhancement of production technologies and dend when the citizens in this population favorable weather conditions greatly bulge reached working age. Moreover, increased the rice harvest, especially in 1967 172  T h e WEB o f T r a n s p o r t C o r r i d o r s i n S o u t h A s i a and 1968. The rice surplus became a financial succeeding Comprehensive National burden at the national level.9 To deal with the Development Plans. Particular attention was surplus, the national government imple- paid to government support for the regions mented a rice paddy reduction program and where productivity was lower. The govern- other measures. ment of Japan has sought to develop infra- structures, transfer fiscal resources and redistribute industries to rural areas. As a Government policies aimed to correct result, the disparity of household income in regional disparities and share benefits regions and excess demographic shift to the The population shift from the rural areas to three major metropolitan areas have been metropolitan areas is largely related to income eased. The almost linear decreasing trend of and employment disparities. Figure 5.2 shows in-migrants ratio and income disparity can be high correlation between the excess in- found for the era of rapid economy growth. migrants ratio in three major metropolitan As explained, the total investment amount in areas in the Pacific Belt Zone and the dispari- the Pacific Belt Zone, including capital invest- ties in income. ments by private enterprises, exceeded invest- The need to correct regional disparities ment in the other regions. However, the stock was emphasized in the first Comprehensive of per capita social capital (on a net basis) in National Development Plan and the concept the Pacific Belt Zone and the other regions of “Well-balanced development between was similar during the period of high eco- regions” continued to be core concept for the nomic growth up to 1970.10 Then the other FIGURE 5.2  Demographic shift to the three major metropolitan areas and the income disparities in the region 3.0 1.6 Correlation coefficient with excess in-migrants ratio: 1955–2011 1955–89 1990–2011 2.5 Income Disparity 0.96 0.97 0.04 1.5 1963–2013 1963–89 1990–2013 (excess in-migrants ratio, %) 2.0 Employment disparity, rate Income disparity, rate Employment Disparity 0.55 0.34 0.93 1.4 1.5 1.0 1.3 0.5 1.2 0 –0.5 1.1 19 4 19 6 19 8 19 0 19 2 19 4 19 6 19 8 19 0 19 2 19 4 19 6 19 8 19 0 19 2 19 4 19 6 19 8 19 0 19 2 19 4 19 6 20 8 20 0 20 2 20 4 20 6 20 8 20 0 20 2 14 5 5 5 6 6 6 6 6 7 7 7 7 7 8 8 8 8 8 9 9 9 9 9 0 0 0 0 0 1 1 19 Income disparity Employment disparity Excess in-migrants ratio Source: White Paper on Land, Infrastructure, Transport and Tourism in Japan 2015. Note: The excess in-migrants ratio is calculated as follows: “(in-migrants – out-migrants in three major metropolitan areas) / population of Japanese.” Income disparity is calculated as “the three major metropolitan areas average of the income of one prefectural resident / national average (excluding the three major metropolitan areas),” Employment disparity is calculated as the ratio of active job openings to active job applicants, which is “the three major metropolitan areas average of the ratio of active job openings to active job applicants / national average (excluding the three major metropolitan areas).” The numbers of the graph are the correlation coefficient of the excess in-migrants ratio for each period and the disparity index. A p p r a i s i n g T r a n s p o r t C o r r i d o r s i n J a p a n , E u r o p e , a n d T h a i l a n d    173 regions far surpassed the Pacific Belt Zone. districts that included prefectural capitals.12 This result suggests that policies to prevent One conceivable reason for the lack of success widening regional disparities worked. to achieve the goal intended by the launch of the new industrial cities was the failure to Allocation of government spending to foresee private enterprises’ preference to different regions expand business within areas where they were Approximately half the revenue for local gov- already located, based on criteria of economic ernment depended on the national treasury, rationality. While the development of the new including local allocation tax grants, national industrial cities is presumed to have contrib- treasury disbursements, and local govern- uted to curbing the expansion of disparities ment bonds. Local taxes covered only about between the Pacific Belt Zone and other 30 percent of local budgets. This level was regions, there are arguments for and against criticized as “30 percent autonomy” by those the value of supporting new industrial cities. who favored more self-determination. The Comprehensive National Development Nevertheless, significant resources from the Plan proposed creating cities as regional devel- national government played an important opment hubs that would implement develop- role in securing a national minimum in the ment policies in their respective ­ districts. nonprofit sector and in redistributing income Various efforts to enhance the capital of each among regions. Across all regions in the prefecture were promoted, including efforts to country, per capita fiscal revenue was almost improve urban functions (such as commerce, the same. production, education, and culture) and to New industrial cities One example of a policy to eliminate regional MAP 5.2  Fifteen new industrial cities were established to spread disparities was a measure to designate benefits throughout Japan 15 new industrial cities. Most of these cities were ­ outside the Pacific Belt Zone (map 5.2). This effort aimed to avoid the excessive con- centration of people and industries in preex- isting large cities. Under the policy, some existing cities were expanded—such as pre- fecture capitals like Sendai Bay, Akita Bay, and Tokushima—and some entirely new ­ c ities were built—such as Doou, Joban- Koriyama, and Hyuga-Nobeoka—as hubs of industrial siting and regional development. In all, JPY72 trillion was invested to develop infrastructure in the new industrial cities dur- ing the 34-year period from 1964 to 1998.11 In these new industrial cities, infrastruc- ture developments were followed by the con- struction of factories, focused mainly on the heavy and chemical industry sector. However, during this period, Japan’s industrial struc- ­ ture was undergoing a rapid shift from manu- facturing to services. Thus, the launch of new industrial cities was not enough to form regional hubs capable of generating many new job opportunities based on the expansion of manufacturing. The exceptions were the Source: 1962 Law for Promoting Establishment of the New Industrial Cities. 174  T h e WEB o f T r a n s p o r t C o r r i d o r s i n S o u t h A s i a enhance regional development without over- Education dependence on megacities like Tokyo, Osaka, Several laws were passed to ensure that equal and Nagoya. Transportation networks were opportunities in education (especially in improved, and private enterprises were sup- compulsory education) would be provided ported in effort to expand their production and educational standards would be upheld sites beyond headquarters in the major cities and improved. These included the Basic Act to other parts of the country. Along with on Education (1947), the Act on National structural shift of manufacturing from mate- Government Contribution to Compulsory rial industry such as chemical and metal Education (1952), and the Act on National industry to mechanical and electronic one, Government Contribution to Equipment industrial location increased in rural areas, for and Facilities of Public Schools (1953). example, along newly developed expressways Concrete measures included the allocation of and production share of conventional coastal national budget resources to local govern- industrial areas along the Pacific Belt Zone ments to achieve a specified number of pupils has relatively decreased. Such efforts and or students and of assigned teachers per industrial location shift contributed to the class, implementation of unified curricula, expansion of market size and the increase of and operation of a textbook authorization employment opportunities in the other system. In addition, candidates for principal regions. and assistant principals of elementary schools and junior high schools were required to serve as teachers in remote districts before Health care they could be appointed. Because of these The government also tapped the nonprofit efforts, it makes little difference where one sector to enhance balanced growth across the receives compulsory education in Japan. country. In the area of health and medical Japanese citizens have broadly supported care, one of the major goals was universal equal opportunities and uniform quality in health insurance coverage. This goal was compulsory education. This equality and achieved in 1961. It enabled Japan’s health uniformity is considered to have played a care infrastructure to offer equal opportuni- major role in facilitating the increase of ties for access to medical care. Other policies employment opportunities and the internal with a focus on local areas included efforts to migration that accompanied the rapid indus- increase and improve public medical institu- trialization of Japan. tions and medical personnel; measures to secure medical services in remote districts; The social costs of pollution were and health and hygiene activities carried out not adequately considered during the through coordination with relevant personnel planning and implementation stages (such as activities by life improvement instruc- tors to improve nutrition). Water-related Rapid industrialization also had a serious infectious diseases were curtailed by increas- negative impact, in the form of environmen- ing the coverage of the piped water supply tal deterioration and pollution (map 5.3). nationwide, which increased from 53.4 Air pollution worsened. Sulfur oxide emis- percent (in 1960) to 80.8 percent (in 1970) to sions increased ninefold from 1955 to 1971 91.5 percent (in 1980). While the industrial- in the three major metropolitan areas of ization and associated development of infra- Japan, for instance. Sometimes it was diffi- structure advanced rapidly in the Pacific Belt cult to stay outside during daytime hours in Zone and other regions, some improvements these cities. The incidence of bronchial in social capital took longer. For example, asthma and other diseases reached high lev- coverage of sewerage systems increased from els in many areas in the Pacific Belt Zone.14 at 8 percent in 1965 to 23 percent in 1975 to Following outbreaks of several serious dis- 36 percent in 1985 to 51 percent in 1995.13 eases caused by water contamination, A p p r a i s i n g T r a n s p o r t C o r r i d o r s i n J a p a n , E u r o p e , a n d T h a i l a n d    175 including Minamata disease and Itai-Itai MAP 5.3  Health damage related to air pollution soared in the disease, Japan was ­ recognized worldwide as Pacific Belt Zone a “Paradise of Pollution.” Rapid urbaniza- tion also worsened garbage problems. In 1971, the so-called Tokyo Garbage War broke out. While the City of Koto had been disposing of garbage for 23 special wards of Tokyo solely by using landfills, the Koto City Authority used force to prevent gar- bage from being deposited from the City of Suginami, which was reluctant to build a new waste treatment plant.15 In 1964, a project to develop petroleum industry complexes in the eastern part of Shizuoka Prefecture, located in the Pacific Belt Zone, was canceled because of local residents’ opposition to exposure from pollutants. Concerns about pollution and ­ environmental deterioration grew, starting in the mid-1960s. According to a public opin- ion poll conducted in 1970, the majority of respondents considered economic growth to be negative. The ordinary session of the Diet in 1970 was known as the Pollution Diet because many bills related to the envi- ronment were enacted. In the early 1970s, Source: Priority Relief Area, Environmental Restoration and Conservation Agency. litigation concerning all four major pollu- Note: The red triangles indicate Priority Relief Areas designated under 1973 Act on Compensation, tion-related diseases was concluded in favor etc., of Pollution-Related Health Damage. of the plaintiffs. 16 Since then, Japan has made a full-scale effort to pursue pollution control measures. for the vast majority of Japanese citizens. No significant disparity between those regions undergoing rapid industrialization and other Effective development of transport regions occurred. Figure 5.3 presents a con- corridors requires active and ceptual framework for the cycle that led to responsible participation by many Japan’s high economic growth. actors On the other hand, the social costs of Japan’s high economic growth with a focus on pollution were not adequately considered industrialization in the Pacific Belt Zone during the planning and implementation achieved extremely rapid and significant stages. Thus, appropriate preventive mea- results in terms of increasing income. This pos- sures could not be taken. This oversight itive impact is attributed to large-scale indus- offers a lesson for economic development trial development by leveraging the dynamism through rapid industrialization. To ensure of private enterprises in the transport corridor, the effective development of transport cor- the formation of human resources through ridors, all relevant players must fulfill their improvement of education and health and respective functions, including the national medical services, effective utilization of such government (which is responsible for formu- human resources through internal migration, lating development plans, allocating budget and the establishment of a growth cycle sup- resources, and improving the legal system); ported by improvements in living standards local ­governments (which are responsible 176  T h e WEB o f T r a n s p o r t C o r r i d o r s i n S o u t h A s i a FIGURE 5.3  Japan’s rapid economic growth was supported by a virtuous cycle World Bank loans Capex Innovation Capital accumulation Production cost↓ Labor income↑ Savings rate↑ HHs’ affordability↑ Consumption↑ Agro- Urban HH number↑ Rural HH number↓ productivity↑ Quality education and health Labor shift (rural→urban) Labor demand Productivity-related Capital-related Labor-related Source: JICA, with reference to Yoshikawa 2012. Note: capex = capital expenditures; HH = household. The alignment of the transport corridor is a major UNDERSTANDING THE WIDER determinant of the wider impacts. ECONOMIC BENEFITS OF EUROPE’S HIGH-SPEED TRAIN PROJECTS for promoting investment by private enter- Western Europe has an extensive transport prises, and adequately monitoring enter- network, with a number of rail corridors prises to ensure their compliance with laws, designated as components of the Trans- regulations, and standards); enterprises European Network–Transport (TEN-T) (which pursue entrepreneurship, and meet (map 5.4). This section analyzes the context, requirements for compliance, disclosure of design, implementation, and wider economic ­ information, and the like); citizens; commu- benefits (WEB) of three corridors that repre- bodies; and the media. Such nities; judicial ­ sent the diversity of rail corridors in Europe: functions include considering and imple- menting measures to minimize or avoid neg- 1. LGV Sud-Est connecting Paris and Lyon ative impacts such as pollution. in France, a national passenger High Speed Rail (HSR) line. 2. N orth West Europe High Speed Rail The wider economic benefits of rail corridors take at (NWE HSR) connecting Paris, Brussels, least 10 years to emerge. Cologne, A msterdam and London. A p p r a i s i n g T r a n s p o r t C o r r i d o r s i n J a p a n , E u r o p e , a n d T h a i l a n d    177 MAP 5.4  The Trans-European Transportation Network connects most of Western, Northern, and Southern Europe, the United Kingdom, and Ireland Belgrade Sarajevo Podgorica Ploce Pristina Tirana Bar Durres Skopje Source: European Commission, 2017, Trans-European Transport Network: TEN-T Core Network Corridors, http://ec.europa.eu/transport/infrastructure​ /­tentec/tentec-portal/site/maps_upload ​/ SchematicA0_EUcorridor_map.pdf. NWE HSR is a transnational high speed case reviews the implementation of the cor- passenger rail network. ridors, emphasizing the type of financing, 3. The Rhine-Alpine Corridor is a north- including the challenges of using private south multi-modal corridor that runs capital for rail infrastructure. Finally, it through the core of western Europe. The examines the wider economic benefits, iden- focus is on the rail freight component. tifying main impacts that can be documented The first two examples explore the with peer-reviewed literature. broader context within which the corridors The third example takes a slightly different developed to demonstrate how the economic approach. Many of the core rail sections of and political environment was conducive to the Rhine-Alpine Corridor were developed in investment in such large-scale and complex the latter half of the twentieth century, transport infrastructure. The discussion then although significant replacement, upgrading, turns to the rationale and design of the cor- and modernization has occurred since then. ridors, including key decisions on the align- The WEB of the Rhine-Alpine Corridor have ment of the corridor, which is a major emerged over a longer time line—a fact this determinant of the wider impacts. Next, the study seeks to analyze. 178  T h e WEB o f T r a n s p o r t C o r r i d o r s i n S o u t h A s i a LGV Sud-Est: pioneering high-speed commissioned a review of public enterprises, rail in Europe which was released in 1967. This influential report (Nora Report 1967) advocated for France was the pioneer of HSR in Europe.17 public enterprises to adopt a commercial, LGV Sud-Est was the first line in France’s financially sustainable approach and operate high-speed rail network (LGV, Ligne à more like private companies. A key recom- Grande Vitesse, high-speed line). The line mendation was to differentiate potentially links the capital city of Paris with France’s competitive market services from govern- second-richest city, Lyon, with high-speed ment-subsidized services provided by public TGV (Train à Grande Vitesse, high-speed enterprises (Albalate and Bel 2012). train) services operated by the national rail For SNCF, this meant that rail services that operator, SNCF. were not subject to a public service contract LGV Sud-Est is 409 kilometers (km) long would need to compete with other modes. and has two intermediate stations between Unsurprisingly, there was considerable reluc- Paris and Lyon, Le Creusot and Mâcon- tance to embrace the report’s recommenda- Loché. Operations began in 1981. The tions, which represented a fundamental financial success and popularity of LGV cultural change for the railways. However, Sud-Est reinforced support for high-speed external factors, including the economic crisis rail in France, though later lines did not induced by the 1970s oil shocks, reinforced achieve the same financial and socioeco- the need to reduce the financial losses of pub- nomic returns. Over the past 30 years, lic enterprises (Gilbert and Perl 2012, 43). a dditional high-speed lines have been ­ The development of the TGV therefore built, developing a hub-and-spokes network occurred within the context of a state-directed ­ centered on Paris, with further lines under policy of commercial viability. The oil shocks construction. The total length of the high- also reinforced the need for a transport sys- speed passenger rail network in France is tem that was independent of oil supplies. now over 2,000 km. The political and institutional environ- High-speed rail is a relatively new mode of ment of France in the 1970s was also condu- transport in Europe. As one of the oldest cive to the development of the TGV high-speed rail corridors, the LGV Sud-Est (Albalate and Bel 2012). Political power was line was chosen as a case study because it is stable, and g­ overnment authority was highly possible to draw upon ex post studies that centralized, with decision making concen- look at the wider economic benefits over a trated in Paris. As is discussed later in this longer period. This longer time frame reflects section, this power structure allowed the a broader issue in assessing the wider eco- central government to issue a special instru- nomic benefits of rail corridors: the impacts ment (a public declaration) for the construc- of these investments take a long time to tion of LGV Sud-Est, which permitted emerge. A good rule of thumb is at least significant expropriation without consulta- 10 years. tion (Crozet 2013, 13). Context Rationale and design By the 1950s, rail was losing modal share to Before LGV Sud-Est was opened, the Paris- road, and from the 1960s rail was losing Lyon conventional rail line was already the modal share to air in France (Crozet 2013, busiest rail corridor in France. This line con- 13). More broadly, the financial position of nected the Île-de-France, the region centered the French National Railways (SNCF), like on Paris, with Rhône-Alpes, the region cen- many other public enterprises in France at the tered on Lyon. These were (and are) the two time, was deteriorating. The railways regions of France with the highest GDP and were seen to be in terminal decline (Dunn and the largest metropolitan populations. The Perl 1994). The French government existing conventional rail line, which traveled A p p r a i s i n g T r a n s p o r t C o r r i d o r s i n J a p a n , E u r o p e , a n d T h a i l a n d    179 through Dijon, faced significant capacity con- Implementation straints and was losing modal share to air The line was developed in two phases. The (Gilbert and Perl 2012, 44). This is a key first phase—the 274-km southern section point for this corridor: the high-speed rail link between Saint-Florentin (174 km south of was connecting two prosperous regions of Paris) and Sathonay (8 km north of Lyon)— France, with existing levels of traffic reaching opened in 1981, reducing the journey time capacity constraints on the route. from Paris to Lyon from 4 hours to 2 hours SNCF sought to develop a long-distance, 40 minutes. The second phase—the 116-km high-frequency rail service linking Paris and northern section from Combs-La Ville Lyon within two hours, which would make (27 km south of Paris) and Saint-Florentin— the return trip attractive as a day trip, through opened in 1983, reduced the journey time the construction of a largely dedicated high- further to less than 2 hours. The line was speed rail line (Meunier 2002, 6). SNCF’s implemented as a public project, with the objective was to make rail the preferred mode a2 billion infrastructure cost (2005 prices) of transport for travel between the two cities, financed entirely through SNCF borrowing competing with air. The alignment avoided (Vickerman 1997). The rolling stock fleet of intermediate cities, with two purpose-built 107 TGV-PSE (Train à Grande Vitesse—Paris out-of-town stations to provide access Sud-Est) sets was manufactured by Alstom, a (Le Creusot and Mâcon). Avoiding intermedi- French multinational company operating in ate cities reduced the rail distance between rail markets worldwide. The unit cost for the Paris and Lyon by more than 100 km infrastructure construction of LGV Sud-Est (Vickerman 2015). Therefore, the TGV was was approximately a5 ­ million per km. LGV designed to target long-distance travelers— Sud-Est had the lowest cost per km for high- generally, business and leisure travelers— speed rail infrastructure construction in a rather than reducing journey times for global comparison of high-speed rail devel- commuters or other short- to middle-distance opment (Campos, de Rus, and Barron 2009). travelers (Crozet 2013, 4). Again, this deci- This low cost structure reflects a range of sion aligns with the wider objective for the external and design factors. First, the popula- service to compete with air. tion density of France is relatively low, allow- The focus on minimizing end-to-end jour- ing for a relatively straight alignment. There ney times and the lack of intermediate sta- are no tunnels on LGV Sud-Est. Similarly, the tions led to concerns in the 1970s that the decision to avoid intermediate cities reduced TGV would result in further urbanization in the length of the line, which is significantly Paris, at the expense of the underdeveloped shorter than the conventional rail line countryside. There had long been anxiety between Paris and Lyon. The TGV was also about economic concentration in Paris, par- designed to be interoperable with the con- ticularly since Gravier’s famous thesis, Paris ventional rail network. The existing urban and the French Desert (1947), which argued lines in Paris and Lyon are used to reach for regional investment to support the devel- the terminal stations, further reducing the opment of a balanced national economy. The overall cost. concerns extended to Lyon, which, it was The lack of public consultation also con- feared, could turn into a “low rent suburb” of tributed to the low cost of construction. In Paris (Meunier 2002, 6). However, although 1976 a formal public declaration was issued the project was declared to be in the public for the construction of LGV Sud-Est, an interest, there was very limited public consul- action by the French state that diminished the tation and the alignment planning was largely importance of private interests. SNCF then conducted internally at SNCF. Essentially, ­ identified a “broad swathe of right-of-way” SNCF made a strategic decision to prioritize in the French countryside within which land m aximizing modal share over regional ­ was acquired through ­ eminent domain, with development. consultation limited to a public inquiry that 180  T h e WEB o f T r a n s p o r t C o r r i d o r s i n S o u t h A s i a was largely a legitimization of the existing However, the growth in business travel design (Meunier 2002, 6). was not uniformly distributed across sectors. The TGV has particularly benefited the ser- Economic analysis vice sector in Lyon. Expansion was greatest In general, LGV Sud-Est was successful in for high-value services in Lyon, including technical, commercial, and economic terms. market research, advertising, and consulting The line generated traffic volumes and reve- firms. Business travel for the sale or purchase nues beyond the levels forecasted by SNCF of services grew by more than 100 percent. (Melibaeva, Sussman, and Dunn 2011). The The impact on manufacturing was far more estimated financial internal rate of return of limited. The result from surveys of 10 cities in LGV Sud-Est was 12 percent, the minimum Rhône-Alpes and Burgundy showed that return for a project to be considered finan- proximity to TGV services was not a deter- cially viable by SNCF. The expected financial mining factor in the choice of location for internal rate of return was comfortably industrial companies, but was a factor when exceeded in ex post studies, at 15 percent industrial companies were choosing between (Vickerman 1997). A key objective of LGV alternative locations that were otherwise sim- Sud-Est was to compete with air by provid- ilar (Bonnafous 1986). ing a high-frequency long-distance service Specialist businesses, such as public rela- between Paris and Lyon. The ex ante tions, no longer needed to relocate to Paris to appraisal estimated that the reduction in expand beyond the local market. Service busi- journey time from high-speed rail would gen- nesses could maintain relationships within the erate a significant modal shift from air trans- company and carry out sales activities in Paris port (Crozet 2013, 19). By 1997, 16 years while remaining based in Lyon, which offered after the line opened, the TGV had captured a higher quality of life. The line introduced a 70 percent of the Paris-Lyon market, and air fundamental shift in the way distance was transport had fallen from 31 percent to perceived between the two largest population 6 percent of total passenger traffic on this centers in France, from spatial to “temporal route (Campos and Gagnepain 2009). In distances” (Melibaeva, Sussman, and Dunn commercial terms, the line also exceeded 2011). The perception of Lyon as being “two expectations. LGV Sud-Est was financed by hours away from Paris” influenced the behav- SNCF through borrowing, and the debt was ior of users, inducing a significant level of fully repaid 12 years after operations com- business travel, and eventually altered the menced (Vickerman 1997). broader economic and social relationship An early and often referenced study on the between the cities (Melibaeva, Sussman, and wider impacts of LGV Sud-Est was carried Dunn 2011). out five years after the service began The differential impact of high-speed pas- (Bonnafous 1986). The study highlighted the senger rail on services and manufacturing growth in business travel on the corridor, reflects the respective sectors’ different needs which increased by 56 percent between 1980 for mobility. High-speed passenger rail and 1985. Business trips from Lyon to Paris largely benefits companies that rely more on grew at over double the rate of those from face-to-face contacts (Albalate and Bel Paris to Lyon, suggesting that the improved 2012). These companies tend to be involved accessibility increased the market size for in higher value-added service activities and Lyon-based firms. cluster in larger cities to access agglomera- tion benefits (Chen and Hall 2012). Another sector that grew was tourism. Rail infrastructure alone does not deliver the Day trips to Lyon increased significantly economic impacts envisioned. Rail infrastructure through TGV tourism packages, as well as must form part of a comprehensive package of conference-related travel. This shift, however, measures for economic and social development. also reduced the volume of overnight stays, A p p r a i s i n g T r a n s p o r t C o r r i d o r s i n J a p a n , E u r o p e , a n d T h a i l a n d    181 which fell from an average of 2.3 days in rail tends to reinforce the position of large 1980 to 1.7 days in 1992 (Melibaeva, regional cities, such as Lyon (Chen and Sussman, and Dunn 2011). This required a Hall 2012). restructuring of the hospitality sector to adapt A similar pattern of concentration to new groups of customers. occurred at Mâcon-Loché, one of the two These findings contradict the earlier fears intermediate stops on the line. While of the Lyon business community of a central- employment grew by 13.5 percent in Mâcon ization of economic activities in Paris. Lyon from 1999 to 2006—which has been attrib- grew rapidly. Demand for office space around uted to the TGV—it fell in nearby cities the TGV station, Gare de la Part-Dieu, in par- (Melibaeva, Sussman, and Dunn 2011). ticular, increased substantially. Between 1983, Mâcon was already a major employment when the TGV became operational on the area for the region. The arrival of the TGV whole line from Paris to Lyon, and 1990, helped secure its position at the expense of office space around the station increased by nearby cities, from which businesses relo- 43 percent (Melibaeva, Sussman, and Dunn cated. Also, the pattern of increasing intrare- 2011). Further commercial development in gional disparities between Lyon and the the old city center of Lyon was physically Rhône-Alpes region has been replicated restricted, and the arrival of the TGV sup- at the intermediate city Mâcon and in the ported the creation of the new Part-Dieu busi- Saône-et-Loire region. Interesting, the same ­ ness district. A number of Paris-based pattern did not occur in the second interme- technology companies moved their back diate city, Montchanin-Le Creusot. Despite office operations to Lyon to capitalize on the an 85-minute journey time from Paris, only lower cost base. two new companies had been established On balance, the TGV has generated a net around Le Creusot’s TGV station six years economic benefit for both Paris and Lyon, ­ after its opening, with no discernible local with no strong evidence to suggest Paris has economic impact in terms of jobs or com- gained at the expense of Lyon (Cheng, Loo, mercial development. It is likely that other and Vickerman 2015). However, there is factors of Le Creusot’s location, such as the some evidence that increased access to Paris isolated station and poor road access, lim- redistributed economic activity to Lyon from ited its development (Melibaeva, Sussman, surrounding subregions, such as Bourgogne and Dunn 2011). Furthermore, the TGV sta- and Rhône-Alpes, widening disparities within tion serves a group of smaller towns rather regions (Vickerman and Ulied 2009). Some than a single urban area, which has likely areas within Lyon fared better than others. impaired the TGV’s consolidating effect Notably, the traditional downtown area around Montchanin-­ Le Creusot. lost significant numbers of companies to the ­station area (Melibaeva, Sussman, and North-West Europe high-speed rail: Dunn 2011). Connecting five major European cities These shifts highlight the difference between the interregional and intraregional North-West Europe High Speed Rail (NWE impacts of high-speed passenger rail corri- HSR) connects five large European cities, dors. Reducing the journey time between Paris-Brussels-Cologne(Köln)/Frankfurt- Paris and Lyon to two hours does not appear Amsterdam-London (PBKAL). It is more to have exacerbated interregional disparities accurate to refer to NWE HSR as a network between Paris and Lyon. However, there is rather than as a corridor because there are evidence that Lyon has benefited from a multiple connections between the cities, centralization of economic activities, particu- ­ rather than a linear route. Nonetheless, this larly in the services sector, to the detriment of high-speed network connects the metropoli- neighboring cities in the region. This aligns tan core of Europe and offers insights into with a broader conclusion that high-speed the impacts from cross-border passenger 182  T h e WEB o f T r a n s p o r t C o r r i d o r s i n S o u t h A s i a high-speed rail. NWE HSR is 932 km long of high-speed railways in Western Europe and has four cross-border links, including (Peters 2005). The project to develop the the Channel Tunnel. The network is opera- North-West Europe high-speed rail network tional, with international services being run was launched in 1989 with an official agree- by a number of operators, including Eurostar ment between France, Belgium, Germany, and Thalys. The network is also used for fast Netherlands, and the United Kingdom. intercity services within the five participating countries. Rationale and design NWE high-speed rail was conceived as a Context network offering substantial reductions in ­ The development of NWE HSR was spurred journey times between the five countries. by the growing congestion of air and road Figure 5.4 shows the reductions in journey links between the cities in the network, and times on the NWE high-speed rail network the need to improve transport between these from 1989, when the network was planned, metropolitan areas (Vickerman 2015). to 2009, when the whole network was oper- Following the economic crises of the 1970s, ational. These reductions are between one European governments cut capital spending and three hours, representing a 50 percent on infrastructure significantly. The gap time saving for most routes. It is a city-to- between the level of infrastructure investment city ­ network, largely ignoring intermediate and the growth in transport demand widened. towns, except for a few, including Lille and By 1990, real spending on infrastructure was Calais-Frethun in France and Ashford and only at the levels of the mid-1970s. The infra- Ebbsfleet in England. The impact of high- structure deficit was growing, and transport speed rail passenger connections on these needs could not be met (Johnson and Turner intermediate towns is discussed later in the 1997, 45). Environmental concerns also section. drove the shift to rail, as the negative impacts An underlying rationale for the network of mass motorization were being felt. There was to compete with air, reflecting both the was a consensus among governments to pro- increasing congestion in the aviation market mote a modal shift from air and road to rail in northwestern Europe at the time, as well as (Peters 2005). the growing awareness of air transport’s envi- The drive for improved transport connec- ronmental impact. The first LGV line between tivity came not just from governments. Paris and Lyon, discussed earlier, had demon- European industry also played an important strated how high-speed rail could capture air role in promoting transnational transport net- market share. The PBKAL cities are less than works. An influential lobby group, the 500 km apart, falling within the optimal European Round Table of Industrialists (ERT), medium distance range of 400 km to 600 km produced a series of reports in the 1980s and for competition with air (Chen and Vickerman early 1990s advocating for the European 2017) (see map 5.5). Union (EU) and national governments to At the international level, the develop- invest in capital-intensive projects to improve ment of NWE high-speed rail was one of transport infrastructure. ERT membership the Priority Projects of the Trans-European consisted of around 40 chief executive officers Transport Network, endorsed by the (CEOs) of major transnational corporations, European Commission in 1994, designed to including Total, BP, Volkswagen, and Rolls promote accessibility across the European Royce. ERT’s landmark report, Missing Links: Union and improve the functioning of the Upgrading Europe’s Transborder Ground ­ single market (map 5.4). Alongside the objec- Infrastructure, published in 1984, argued for a tives to enhance the competitiveness of rail number of core projects, including a tunnel transport and improve the wider sustainabil- under the English Channel linking the United ity of the transport network, it was also envi- Kingdom and France and a general network sioned that the PBKAL network would release A p p r a i s i n g T r a n s p o r t C o r r i d o r s i n J a p a n , E u r o p e , a n d T h a i l a n d    183 capacity on existing lines, freeing up tracks for FIGURE 5.4  The journey time between major stations freight traffic, and improving the quality rail in northwestern Europe fell by about half on most routes between 1989 and 2009 freight services. Implementation London Amsterdam In 1993, operations commenced on the first section of the network, between Paris and the 1:51 3:18 4:52 1:53 2:39 Channel Tunnel (which was still being built), 5:16 2:55 2:40 via Lille. This section is also part of the French TGV network, referred to as LGV 2:15 Brussels Cologne (Köln) 1:47 Nord, and was developed as a public project 5:12 2:30 1:22 1:10 with significant funding from the European 2:25 3:15 2:10 3:14 Commission (EC) via the European 5:05 5:00 Investment Bank (EIB). The original route 3:48 Frankfurt 6:13 proposed by SNCF avoided the city center of 1:50 Lille, with a station at the point where the Paris 3:55 Saarbrücken lines diverged to London and Brussels. However, political mobilization by the mayor 2:20 HSR (2009) 4:00 of Lille at the time (and future French prime Ordinary rail minister) Pierre Mauroy led to the route being Strasbourg Journey time 2009 diverted through Lille city center (Chen and 1989 Hall 2012). The impacts of the TGV on Lille are discussed in the following section. Source: Chen and Vickerman 2017, citing EU 2010. It is worth discussing two components of the NWE high-speed rail network, the Channel Tunnel and Channel Tunnel Rail Eurotunnel was itself a partnership Link (also known as High Speed 1 in the between the UK and French concessionary United Kingdom), in more detail because they companies. The contractor TML was also a illustrate some of the challenges of ­ private “joint venture of joint ventures” between UK sector provision of high-speed rail infrastruc- and French construction companies ture. In both cases, the governmental agree- (Vickerman 1995). The capital structure of ments to develop the projects stipulated the the project was 20 percent equity and use of private capital for financing. 80 percent debt. In 1987, around 50 percent The most technically challenging section of of the shares were offered to the general pub- NWE HSR was the Channel Tunnel. A treaty lic (which lost much of its investment in the between the United Kingdom and France for later restructuring). Some 200 banks were a 50-km twin tunnel for rail was signed in involved in the loan syndication. The EC sup- 1986. The Treaty of Canterbury specified ported the project through substantial financ- how the project would be financed, “without ing through the EIB. recourse to government funds”; that is, on the The anticipated cost of constructing basis of project risk. The project was struc- the Channel Tunnel in 1985—broadly, tured as a design-build-finance-maintain the time the concession was awarded—was (DBFM) private sector concession. The £4.7 ­billion (2004 prices). The actual cost was c oncession was awarded to the private ­ about twice that: £9.5 billion (2004 prices). Eurotunnel consortium (Eurotunnel) under a The cost overrun for the Channel Tunnel has concession agreement with both govern- often been used as a case study to illustrate ments. The concession runs until 2086. Even the difficulty of controlling costs on mega- for a large infrastructure concession, the con- projects (see, for example, Priemus, Flyvbjerg, tractual structure for the Channel Tunnel was and van Wee 2008). Alongside the escalation unusually complex. of capital costs, the actual traffic volumes fell 184  T h e WEB o f T r a n s p o r t C o r r i d o r s i n S o u t h A s i a MAP 5.5  High-speed rail connects the core of North-West Europe and the United Kingdom Amsterdam High-Speed Rail Connections Schiphol Airport UNITED The Hague NETHERLANDS North KINGDOM Sea Rotterdam Breda 's-Hertogenbosch Stratford Tilburg London Ebbsfleet Helmond Marsgate Medway Towns Ramsgate Brecht Eindhoven Sittingbourne Ostend Brugge Berlin Canterbury Antwerp Ashford Dover Dunkirk Ghent Calais Mechelen Cologne Calais-Fréthun Tourcoing Aachen Boulogne-sur-Mer Saint-Omer Roubaix Brussels Liège Hazebrouck Lille Béthune Namur Mons Lens Charleroi Arras Douai Valenciennes BELGIUM English Frankfurt Channel TGV Haute-Picardie GERMANY Luxembourg FRANCE TGV Champagne- Ardenne TGV Aéroport Charles-de-Gaulle TGV Lorraine Paris TGV Meuse Stuttgart Rennes, Strasbourg Nantes, Lyon, Bordeaux, Marseille, IBRD 43409 | DECEMBER 2017 Toulouse Montpellier Source: Cheng, Loo, and Vickerman 2015. ­orecasts. Revenues failed to well short of the f consortium of banks and construction and cover debt service and the concessionaire transport companies. Construction of the defaulted in one year, requiring a major finan- rail link, which became known as High cial restructuring. In 2007 there was a further Speed 1 (HS1), commenced in 1988, with restructuring, with debt-for-equity swaps phase 1 opening in 2003 and phase 2 in leaving financial institutions as the major 2007. Upon the opening of phase 2, the shareholders. In 2009, Eurotunnel paid its journey time from London to Paris was first dividend, 15 years after commencing reduced to 2 hours 15 minutes, and London operations in 1994. to Brussels to 1 hour 51 minutes. LCR was The section of PBKAL in the United besieged with financial difficulties stemming Kingdom, the Channel Tunnel Rail Link from lower-than-expected revenues. As a (CTRL), was also developed under a project result, the UK government guaranteed a finance structure. In 1996, London major bond issuance to restructure LCR’s Continental Railways (LCR) won the con- debt. In 2006, LCR was reclassified as a tract from the UK government to develop public corporation, as the company effec- ­ the rail section between London and the tively had come under the UK government’s Channel Tunnel under a design-build- control. In 2009, LCR was formally finance-­ m aintain concession. LCR was a renationalized. A p p r a i s i n g T r a n s p o r t C o r r i d o r s i n J a p a n , E u r o p e , a n d T h a i l a n d    185 These issues are not unique to private benefits, or else risk underestimating their eco- infrastructure rail projects. For example, the nomic returns. For PBKAL, the report Betuweroute, a public project to construct a ­ estimated that community benefits accounted freight railway in the Netherlands that was for more than one-quarter of the overall completed in 2007, experienced similar issues benefits. Their inclusion allowed the national ­ in terms of cost overruns and weak revenues. sections to exceed the hurdle rates for approval However, the contractual structures of the (Peters 2005). This in turn led to increased Channel Tunnel and High Speed 1 conces- funding from EC sources, particularly the sions resulted in significant disagreements European Investment Bank. By 2003, EIB had between the parties involved, which com- provided a2.4 billion in funding for PBKAL pounded the financial issues. (EIB 2003). Services on several sections of the network overlap, and are provided by competing oper- Major Cities ating companies. Thalys runs services between A recent study analyzed the wider economic Paris, Brussels, Cologne, and Amsterdam. benefits of high-speed rail by comparing These compete with TGV services on the employment rates in the PBKAL cities and French section, and with Eurostar, which runs outlying areas between 1999 and 2008 between Paris, Brussels, and London. The (Cheng, Loo, and Vickerman 2015). The German state railway, Deutsche Bahn, with- average employment growth of 10.1 percent drew from the Thalys consortium and now in the PBKAL cities was higher than the competes with Thalys between Brussels and 8.9 percent in the hinterlands. However, Cologne (and travels on to Frankfurt). This there was significant variation among the cit- plurality of service providers on the same ies. Amsterdam presented an interesting routes makes it more difficult to book train case, r­ egistering employment growth of tickets, impairing the goal of being competi- 14.1 percent—the second highest among the tive with airlines (Vickerman and Ulied core cities—even though it was not fully con- 2009). However, the different operators have nected to the NWE HSR network until 2009. taken more steps recently to increase the ease A key conclusion of the study was that the of booking for consumers. Alongside interna- wider impacts of high-speed rail were depen- tional high-speed rail services, the PBKAL dent on location, and that transport network is also used for regional high-speed improvements would not necessarily result services for commuting traffic. in local economic development—particularly on cross-border or interregional routes Economic analysis (Cheng, Loo, and Vickerman 2015). Based Before considering the wider economic bene- on this analysis, it was suggested that pas- fits for the major and intermediate cities, it is senger high-speed rail in northwestern worth reiterating how the network was Europe has not had a transformative eco- largely developed as a series of national nomic impact, particularly because most sec- ­ sections. Studies carried out at the national tions lack intermediate stations. level in the mid-1990s found relatively low Yet the same study also had a more levels of economic returns for the sections in p ositive conclusion about high-speed ser- ­ the United Kingdom, Belgium, and the vices. It suggested a tendency for cities and Netherlands. The section in France, LGV their hinterlands to converge since high- Nord, was already operational by 1993, and speed services have been introduced. This was experiencing traffic volumes that were convergence seems to be acting in the oppo- significantly lower than expected, primarily site direction from the pattern of centraliza- due to lower international passenger traffic. tion associated with passenger high-speed A report carried out by the EC Working rail. For already-congested cities that lack Group on PBKAL highlighted the need to space for expansion and where rent and include cross-border, or “community,” property prices are already high, high-speed 186  T h e WEB o f T r a n s p o r t C o r r i d o r s i n S o u t h A s i a passenger rail service can shift growth to This package illustrates the need for the connected towns. ­ omplementary investments alongside infra- c structure access in order to generate positive, Lille wider impacts. It is also important to note Lille is often held up as an example of the that while the Euralille development resulted positive benefits of high-speed rail (see, for in job creation in Lille, net employment example, Greengauge 2006). By the 1980s, growth is questionable. As was the case for the coal, steel, and cotton industries in the the LGV Sud-Est corridor, there is evidence of city were struggling against cheaper imports, centralization of economic activity in the alongside a gradual reduction in the supply regional hub (Lille) as a result of its increased of raw materials (Hickman, Bonilla, and accessibility, at the expense of smaller cities in Banister 2015, 306). As mentioned, the the region (Vickerman 2015). mayor of Lille lobbied intensively for LGV Nord to stop in the city center, making Lille The Rhine-Alpine Corridor: a major transport hub with direct train con- supporting the development of nections to London, Paris, and Brussels. Europe’s industrial core A key aspiration with the arrival of LGV Nord was the development of the knowl- The third case study focuses on the Rhine- edge-based economy in Lille. Analysis of Alpine Corridor, one of the nine core network employment patterns showed that Lille has corridors of the Trans-European Transport had the highest increase in knowledge-­ Network (map 5.6). The Rhine-Alpine intensive employment in the Nord-Pas-de- Corridor is multimodal, integrating a number Calais (NPDC) region, and significantly of corridors, including Rail Freight Corridor above the national average (Chen and Hall 1 (RFC 1). While this corridor does not have 2012). Access to Paris within one hour dedicated rail freight lines and passenger brought Lille within commuting distance, trains operate along the route, the underlying which supported growth in public services rationale for the Rhine-Alpine Corridor within Lille. The same study found that clearly emphasizes freight transport. In other other subregions in the NPDC region that words, the basis for designating this route as did not have access to TGV services had the a core transport corridor in Europe is its lowest knowledge-intensive employment. importance for the international transporting Chen and Hall (2012) suggest that, more of goods. For the purposes of this report, this broadly, the arrival of passenger high-speed corridor was selected to illustrate how the rail and transport infrastructure has sup- wider impacts of rail corridors have extended ported the economic restructuring from to influencing the spatial structure of national industry toward high-value services and the economies. knowledge economy. In Lille, the TGV station was located close Description to land that was available for commercial The Rhine-Alpine Corridor is a north-south development. The Euralille development was corridor extending from the North Sea ports a major regeneration project to create a new of Rotterdam and Antwerp to the urban center in Lille, including commercial, Mediterranean port of Genoa in northern residential, leisure, and conference facilities. Italy. The corridor is a primary artery for The first phase was built in parallel with the transporting goods in Europe. Over 1 billion construction of LGV Nord (Ureña, Phillippe, tonnes of freight are transported on the cor- and Garmendia 2009). Euralille was struc- ridor annually, of which 370 million is cross- tured as a joint venture between local authori- border. The share of rail on the corridor is ties and private investors. It included job approximately 12 percent, with 34 percent creation programs, offering training to local for road and 54 percent for inland water- unemployed people. ways (EC 2015, 11). A p p r a i s i n g T r a n s p o r t C o r r i d o r s i n J a p a n , E u r o p e , a n d T h a i l a n d    187 The corridor passes through five coun- MAP 5.6  The Rhine Alpine Rail Freight Corridor: “From sea to sea tries: the Netherlands, Belgium, Germany, without barriers” Switzerland, and Italy. A significant length of the corridor follows an inland waterway, the River Rhine, a historically significantly transporting route in Europe. Other than the first section of the river in Switzerland (the High Rhine, or Hochrhein), the corridor fol- lows the Rhine from Basel to the Rhine delta on the North Sea. The River Rhine remains an important means of transporting bulk commodities between the North Sea ports and France, Germany, and Switzerland. The Rhine-Alpine Corridor is a mature corridor, with end-to-end road and rail con- nections from Rotterdam to Genoa. Electrification is complete along the length of the corridor, and freight speeds of 100 km per hour are possible on over 90 percent of the route. The entry and exit of the corridor are maritime ports, including Antwerp and Rotterdam at the northern end. The Port of Rotterdam is the largest port in Europe, handling 466 million tonnes in 2015, and is the eleventh-­largest container terminal glob- ally. There are substantial intermodal nodes along the corridor, including at Duisburg in the Ruhr Valley in Germany. Duisburg Port is the world’s largest dry port, with road, rail, and inland water connections. Increasing demand has resulted in the need for significant investments in the corridor, primarily addressing the need for capacity enhancement and interoperability, including Source: RFC Rhine-Alpine, “From Sea to Sea without Barriers,” https://www.corridor-rhine-alpine​ 60 rail projects. .eu/home.html. Note: Map is as of 2017. Historical development developed by the Romans to facilitate trade in The history of the corridor can be traced back what is now the Netherlands. This includes a to the Roman period, when the trade route likely transhipment port at Domburg, located between Northern and Southern Europe south of Rotterdam, with seafaring vessels emerged out of the economic asymmetry traveling to the British Isles and river barges between these two regions. Lower-value goods to Cologne (Grazhdankin 2012). Transport such as salt, amber, wool, and wood were was central in the early formation of spatial being traded in the North, and higher-value goods such as spices, silk, and precious stones in the South (Drewello and Scholl 2016, 2). Key cities on the Rhine emerged during the The development of the Rhine-Alpine transnational Roman period—including Cologne, Bonn, rail freight corridor had as much to do with and Basel—and a network of land routes, the  harmonization of standards as the physical waterways, ports, and warehouses was construction of railway infrastructure. 188  T h e WEB o f T r a n s p o r t C o r r i d o r s i n S o u t h A s i a economic structures along the corridor, with is currently true for the Gotthard Base the River Rhine as the central axis. Tunnel, a new tunnel that is discussed later From the early 1800s, the countries along in this section). the corridor were engaged in industrial revo- The sources of finance for the line and tun- lutions, alongside acceleration in the develop- nel were largely outside Switzerland. ment of natural resources, particularly in the Combined, the governments of Italy and Ruhr and Saar regions (in Germany). In the Germany contributed two-thirds of the capi- Ruhr Valley, growth in coal production was tal for construction, demonstrating that the followed by the formation of large steel and benefits of the line were recognized by other chemical industries, furthering the growth of countries on the corridor (the same approach agglomerations along the corridor. The popu- is institutionalized today through the EU). In lation of the Ruhr Valley grew from 400,000 return, Italian and German rail operators in 1850 to 3.8 million in 1925 (Taylor 2015). were offered preferential tariffs for use of the The development of railway networks tunnel, and a profit sharing arrangement was contributed to the expansion of industry put in place. Essentially, for those providing along the corridor, and key sections of the capital for the project, this was a return on Rhine-Alpine rail corridor were first devel- equity in the form of dividend payments oped in this period. These included the (Anastasiadou 2011, 33). Mannheim-Karlsruhe-Basel Railway, also known as the Rhine Valley Railway, which International Cooperation was built using German public funds between Beyond the physical construction of infra- 1840 and 1855; and the “Iron Rhine,” a structure, an international regime to coordi- freight railway connecting the Port of nate and facilitate cross-border freight traffic Antwerp (Belgium) to the Rhine Basin in emerged. Germany led an earlier period of Germany via the Netherlands, which was development of the rail corridor. The German built between 1868 and 1879. The cross-­ Railway Union, formed in 1847, had signifi- border section is no longer in use. Passenger cant influence on railway development across and freight international traffic instead cur- Central and Northern Europe, and promoted rently use the Montzen line, which avoids the cooperation between the various national and Netherlands. However, Belgium is now seek- regional railway administrations. By 1879, the ing to reopen the Iron Rhine due to capacity German Railway Union had 110 railway constraints. administrations across Germany, the An interesting early example of a transna- Netherlands, Luxembourg, and Austria tional approach to rail development is the (Austro-Hungary), which collectively adminis- Gotthard line, a primary north-south rail tered a network of over 50,000 kilometers. axis in Switzerland, and a key section of the In parallel, the Swiss government led an Rhine-Alpine Corridor. A dense rail network initiative to establish agreements between had developed in Switzerland by the 1870s. governments that would facilitate interna- However, there was no line through the tional traffic. Starting in 1874, a series of con- Swiss Alps. All north-south traffic traveled ferences were organized by the Swiss east or west around Switzerland. In 1871, government, which sought to develop a com- the Gotthard Railway Company, a private mon standard for rail goods transport to enterprise, was incorporated to develop and facilitate more efficient cross-border traffic in operate a new north-south line and signifi- Europe. In 1882, a conference was held to set cantly reduce the journey times of transit international standards for rolling stock, traffic. This included the construction of the resulting in the Technical Unity on Rail first Gotthard Tunnel, a 15-km rail tunnel Transport (L’Unité Technique, UT), which under the Swiss Alps, between 1872 and included parameters such as loading gauge, 1881. The Gotthard Tunnel was at the time maximum vehicle length, and the position of the longest tunnel in the world (and the same couplings. In 1890, the International A p p r a i s i n g T r a n s p o r t C o r r i d o r s i n J a p a n , E u r o p e , a n d T h a i l a n d    189 Convention Concerning Railway Freight countries, a discontinuous, 1,500-km urban Transport was signed by representatives of corridor stretching from Manchester in north- 11 European countries, including all countries western England down to Milan in Northern on the Rhine-Alpine corridor. Italy. The corridor has a population of over The task of harmonization of standards 100 million, and spans major European cities continues to this day. The current focus is on including London, Amsterdam, Brussels, the implementation of the European Rail Frankfurt, Zürich, and Milan (the European Traffic Management System, which, along- backbone can be easily identified through side enhancing interoperability with the empirical analysis of population density). The deployment of a common signaling system, European backbone has a high number of has the ability to significantly enhance metropolitan areas with high per capita capacity. The development of the Rhine- incomes, high labor force participation in ser- Alpine transnational rail freight corridor has vices, and low unemployment relative to the had as much to do with the harmonization of rest of Europe (Heidenreich 1998). standards as the physical construction of The concept of a single, linear European railway infrastructure. backbone is highly contested. The concept omits key urban centers such as Paris, as well as the urbanization along the northern Economic analysis Mediterranean coast. Moreover, with the Corridor Level expansion of the European Union, the geo- As discussed, the origin of the Rhine-Alpine graphical center of Europe has been moving corridor was the River Rhine, a natural trans- eastward, with new growth poles in Central portation route from Lake Constance in the and Eastern Europe. Nonetheless, by global Swiss Alps downstream to the Rhine delta in standards, there is a dense agglomeration of what is now the Netherlands. The River people and economic activity along this Rhine was a major determinant in shaping urban corridor, which has survived the tran- the economic geography of a number of sition from industrial to service-based major industrialized nations in Western economies. Europe. Later, the globalization of trade, The transport corridor that is now desig- exploitation of natural resources, and the nated the Rhine-Alpine Corridor has co- Industrial Revolution spurred railway devel- evolved with this agglomeration. Originally opment, further increasing the density of peo- with the River Rhine, and later with railways ple and economic activity along the corridor. and roads, this transport corridor has pro- In 1989, the area served by the corridor vided the infrastructure to support the devel- was dubbed the “Blue Banana” by the French opment of Europe’s industrial core. It is geographer Roger Brunet, who originally estimated that the region served by the named it the Dorsale européenne (European Rhine-Alpine Corridor currently generates backbone). The European backbone refers to approximately 20 percent of the European an arc of Europe starting in northwestern Union’s GDP. England in the United Kingdom, and extend- ing through Greater London, the Randstad Gotthard Base Tunnel (Amsterdam, Rotterdam, and the other urban Some 135 years after the original Gotthard centers in the western Netherlands) and Tunnel was inaugurated, a second rail tunnel Brussels, down through the Ruhr Valley in was opened in 2016 to expand capacity for Germany, along the Rhine through southern rail freight. The Gotthard Base Tunnel, at Germany and Switzerland, and finally into 57 km in length, is the longest tunnel in the the “industrial-triangle” of northwestern Italy world. Switzerland has a long history of pro- (Milan-Turin-Genoa). viding routes for transit traffic through the This spatial concept captures what may be Alps. Indeed, the country was formed in the the largest agglomeration in the Western Middle Ages around these trade routes. 190  T h e WEB o f T r a n s p o r t C o r r i d o r s i n S o u t h A s i a The primary rationale for the construction that required to achieve the constitutionally of the Gotthard Base Tunnel was the modal mandated target of a modal shift from road to shift of transalpine traffic from road to rail. rail (Metron 2009, 34). There were growing calls to address the con- gestion caused by the volume of heavy freight Conclusion traffic on the roads in the Alps, reduce the level of road accidents, and minimize negative A number of conclusions can be drawn environmental impacts, including air and regarding the wider economic benefits of rail noise pollution. Despite the absolute growth corridors in Europe. The first is that none of in rail volumes, the rapid growth of road the passenger high-speed rail lines reviewed in freight eroded the modal share of rail. this study have had a transformative eco- In 1994, under the Swiss system of direct nomic impact. The LGV Sud-Est line and the democracy, the Swiss electorate voted to com- PBKAL network connect the most densely mit Switzerland to transferring freight traffic populated and prosperous regions in France from road to rail, resulting in a constitutional and northwestern Europe, respectively, and commitment for a modal shift to rail. In therefore have served to reinforce the existing 1998, the public voted for the construction of importance of these regions. However, it can the tunnel to be publicly funded using the be argued that these high-speed lines have Public Transport Infrastructure Fund (FinöV). helped maintain economic development, as The sources of the fund were taxes on heavy capacity constraints in the transport system vehicles, a tax on gasoline, and the value- had the potential to limit growth. In other added tax (VAT). In 1999, an act came into words, while it is not possible to accurately force setting a target of 650,000 truck jour- assess the counterfactual, these rail corridors neys per year (against 1.4 million truck jour- have helped support existing centers of eco- neys in 2000). The tunnel was to be a key nomic activity. instrument in achieving this policy objective. There is some evidence that high-speed rail The agency responsible for implementation has widened disparities within regions, fur- was AlpTransit Gotthard Ltd, an enterprise ther concentrating economic activity in cities established for this purpose and fully owned served by high-speed services. This has been by Swiss Railways. The total cost of con- observed at both the terminal and intermedi- struction of the tunnel was a 11 billion ate cities on the LGV Sud-Est line. However, (US$12.3 billion). there is less evidence that high-speed rail has A key study in 2011 estimated that the two led to an overall centralization in a core- new tunnels (Gotthard Base Tunnel and periphery structure. Lötschberg) would, combined, generate a The two passenger rail corridor case stud- 20 percent increase in rail traffic, including a ies also demonstrate how high-speed rail in 13 percent modal shift from roads. The eco- Europe primarily has affected the services nomic analysis used a 60-year appraisal time- sector, while having little or no impact on line and a discount rate of 2 percent, manufacturing. Passenger high-speed rail assessing the economic, environmental, and essentially supports face-to-face contact, social impacts. The cost/benefit ratio was cal- which is a critical factor in the development culated at approximately 1, with a caveat that of higher-value services and the knowledge regional economic and trade benefits were not economy. captured (Ecoplan and Infras 2011, 70). Timing is also critical. The PBKAL net- Another study completed before the opening of work became fully operational around the tunnel (but after construction commenced) the same time as the global financial crisis. more conservatively estimated that the tunnel The resulting economic downturn reduced would increase rail’s modal share by 2.5 passenger traffic volumes, which in turn cur- percent—an improvement, but far less than tailed wider impacts. The earlier financial A p p r a i s i n g T r a n s p o r t C o r r i d o r s i n J a p a n , E u r o p e , a n d T h a i l a n d    191 failure of the privately financed sections of (2013) dubbed Thailand the “Detroit of the the PBKAL, the Channel Tunnel, and High East,” harkening to Detroit’s glory days as Speed 1 suggest that caution should be the center of the U.S. auto industry. applied when forecasting traffic volumes. Three sets of factors have driven Thailand’s The Rhine-Alpine Corridor case study success in attracting automotive production. illustrated how a natural transport corridor The government sponsored a brand new evolved over centuries to become the core industrial cluster to address the overconcen- north-south route for the movement of tration of production around Bangkok, goods in Europe. This corridor helped including a new deep-sea port and a shape the economic geography of the ­ petrochemical complex to produce energy for region it served, including having a role in production. It also supported improvements the formation of states. In the latter half of in transport infrastructure and public utilities, the twentieth century, the development of while ­ p roviding incentives to induce railways on the Rhine-Alpine Corridor sup- investment. Meanwhile, it maintained a flexi- ­ ported the rapid growth of manufacturing ble policy stance to adjust to changing exter- and the development of natural resources, nal conditions such as macroeconomic helping to create the industrial core of conditions, access to finance, and inflows of Western Europe. This case study demon- foreign direct investment. The result was a strates the importance of rail freight in the ­ ublic-private collaboration with successful p historical economic development of both domestic and foreign investors that Europe. spurred a new direction for heavy industry in Finally, the study has shown how rail Thailand, especially in the automotive infrastructure alone does not deliver the eco- industry. nomic impacts envisioned. To do so, infra- structure must form part of a package of measures, as was the case in Lille, where the A set of integrated transport infrastructure arrival of the TGV was accompanied by mea- investments and government policies aided a shift in sures to support urban regeneration. Thailand’s industrial base to heavy manufacturing. THAILAND, THE DETROIT OF The centerpiece was a proactive set of THE EAST? infrastructure investments, beginning in the The growth of Thailand’s automotive pro- late 1980s, centered on the creation of a duction and exports is considered an eco- high-capacity deepwater port to facilitate nomic success story. Although Thailand the importing of raw materials and export- began producing motor vehicles and parts in ing of finished products. The new port, the 1960s, it catered primarily to the highly known as Laem Chabang, is located 75 km protected domestic market through the southeast of Bangkok. Unlike the historic 1990s. Production for export has been and highly congested Bangkok port, located important only since 2000. In 2015, upstream from the coast on the Chao Phraya Thailand produced more than 2 million River, Laem Chabang can receive large vehicles, making it the world’s ninth-largest oceangoing container vessels. The Laem automotive producer. Roughly half the Chabang port is integrated with major industry’s final output is exported. investments in roads, electricity, and water Employment within Thailand’s automotive supplies adjacent to the port and along the sector (final assembly plus parts) exceeds highway connected to it. The port, together 250,000 workers, about 4 percent of with the industrial area immediately adja- Thailand’s manufacturing workforce.18 In cent to it, might be considered a hub, but light of these achievements, The Economist the highway system connected to it, with 192  T h e WEB o f T r a n s p o r t C o r r i d o r s i n S o u t h A s i a The development of an efficient gateway for The plan focused on providing exports—a deep-water port connecting roads and infrastructure and incentives to promote associated manufacturers—spurred Thailand’s heavy manufacturing beyond Bangkok automotive industry. By the mid-1980s, the Bangkok port could no longer support heavy manufacturing infrastructure investments in electricity and within Thailand. The port upstream on the water located along this highway system, Chao Phraya River was incapable of han- created a corridor of broader economic dling large, oceangoing container ships and activity consisting of the outskirts of required trans-shipment of cargoes to smaller Bangkok and the seven provinces lying in a vessels. Meanwhile, the road connection to semicircle to the east and north of Bangkok, industrial areas passed through Bangkok’s all linked to the Laem Chabang port.19 This notoriously congested traffic. Japanese transport and infrastructure corridor facili- expertise and financial support were impor- tated the establishment and growth of final tant in designing a new port area, 75 km to automotive assemblers (builders of cars, the southeast of Bangkok. This scheme, the trucks, and SUVs). Crucially, this develop- Eastern Seaboard Development Plan (ESDP), ment occurred in conjunction with the centered on the new port of Laem Chabang. establishment of manufacturers of parts and It was connected by road to the large Map components along the corridor. The final Ta Phut petrochemical complex planned fur- assemblers were all foreign owned—mostly, ther to the south at Rayong and also served but not entirely, Japanese. The parts and by a deep-water port. The new port was components manufacturers included both designed specifically to encourage a shift in foreign firms (mainly Japanese) and numer- Thailand’s manufacturing base. As a deep- ous smaller Thai firms. water port, it could accommodate oceango- Along with this push for infrastructure, ing container vessels and thereby support the the Thai government introduced a set of development of heavy manufacturing within policy changes shortly after and partly in Thailand, rather than the g ­ arments, electron- response to the disastrous Asian Financial ics, and other light manufacturing that were Crisis of 1997−98. These changes for the already important within the country. The first time permitted unlimited foreign own- planning documents of the time did not ership of both final assemblers and parts anticipate that the resulting industrial devel- and components manufacturers in the auto- opment would take the form of export-ori- m o t i v e s e c t o r. T h e y a l s o a b o l i s h e d ented automotive production, though it was Thailand’s restrictive requirements for the an obvious potential candidate. The new local content of motor vehicles produced port was intended to support heavy industry within Thailand. Without the foreign in general. exchange shortage that accompanied the The Eastern Seaboard Development Plan Asian Financial Crisis, these policy changes was implemented from the 1980s through the probably would not have been politically early 1990s. It created a new industrial cluster feasible. in three provinces (Chachoengsao, Chonburi, Another factor for the success in attracting and Rayong)—the Eastern Seaboard (ESB) foreign investment was noneconomic and area—located 80 to 200 km southeast of difficult to quantify. For historical and cul- ­ Bangkok. The ESDP had four main tural reasons, Thailand is an attractive and ­c omponents: the Map Ta Phut Port and welcoming venue for Japanese firms. Japanese industrial/urban complex; the Laem Chabang executives and their families enjoy living in Port and industrial estate; railways and roads Thailand, giving it an advantage in attracting connecting these areas with Bangkok; and ­ the Japanese investment relative to most of its development of water resources and pipelines East Asian neighbors. (map 5.7). A p p r a i s i n g T r a n s p o r t C o r r i d o r s i n J a p a n , E u r o p e , a n d T h a i l a n d    193 MAP 5.7  The four components of the Eastern Seaboard Development Plan acted as an integrated unit Source: JICA. The first component, the Map Ta Phut plant. These plants serve the domestic area, was intensively developed to ensure the demands for energy and materials for the expansion of the production of heavy chemi- heavy chemical industry. Initially, the Thai cals. The port consisted of a multipurpose government identified four industrial sectors berth and two berths for liquid cargo. for potential national projects (chemical fer- Industrial estates with the necessary infra- tilizer, soda ash, reduced iron, and heavy structure were developed in a 380.8-hectare chemicals). The government finally decided to (ha) area—later expanded to 804.8 ha by the concentrate on heavy chemicals after a com- Industrial Estate Authority of Thailand prehensive examination, taking into account (IEAT). By 1998, 50 companies were operat- the macroeconomic conditions and the busi- ing there. The natural gas plant in the com- ness feasibility. plex produces ethane, propane, and liquified The second component, Laem Chabang, petroleum gas (LPG) from the natural gas leveraging on its deep-sea port for importing delivered to the port. A plant with an initial and exporting, became the center of the production capacity of 3.5 thousand cubic manufacture and assembly of automotive feet (mcf) / day was followed by a second vehicles and electronics. At that time, (250 mcf/day) and a third (350 mcf/day) Thailand was in desperate need of a 194  T h e WEB o f T r a n s p o r t C o r r i d o r s i n S o u t h A s i a FIGURE 5.5  Transport, utilities, and investment incentives ranked water source, with pipelines connecting it to high in companies’ decisions to operate on the Eastern Seaboard Map Ta Phut and Laem Chabang. Development policy, focusing on provid- Good transport infrastructure 3.84 ing incentives and relevant infrastructure, played an important role in accelerating Good public utilities 3.71 industry’s development in these regions. In BOI investment incentives 3.71 a survey of 113 companies operating on the Eastern Seaboard (figure 5.5), firms stated Proximity to Bangkok 3.01 that the three most important factors in Proximity to seaport 2.98 their decision to locate in the region were transport infrastructure; public utilities; Proximity to customers 2.31 and investment incentives, which were more Proximity to suppliers or favorable than those in Bangkok. 1.87 Supporting industries “Proximity to Bangkok” was also impor- Tax incentives in EPZ 1.09 tant, but the quality of social services and education has yet to be improved to ensure Proximity to parent company 0.58 a quality of life as high as that in Bangkok. 0 1 2 3 4 5 “Proximity to seaport” as an alternative to Bangkok also ranked high. Source: JICA. The government encouraged development Note: Responses were based on a 1-to-5 scale, with 1 = not important and 5 = very important. of privately operated and financed industrial BOI = Board of Investment; EPZ = export-processing zone. estates along the highway system connected to Laem Chabang port. Aside from a small pub- deep-sea port. The site met that need with a licly owned industrial estate adjacent to the depth (16 m) that was much deeper than port, the development of industrial estates was Bangkok’s port (8.5 m–11 m). left to the private sector. These industrial The third component focused on rail- estates were not confined to automotive- ways and roads to connect these areas with related production, but also included the full Bangkok. Their development has resulted in range of Thailand’s manufactured exports. closer and tighter connections among Within these estates, the private operators these regions. Railways are used for the provided electricity connections to the public long-distance bulk shipment of containers grid, made industrial land available for sale or from and to Laem Chabang, as well as lease, and in many cases offered standard fac- heavy chemicals from Map Ta Phut. Japan tory buildings for lease to foreign or domestic supplied official development assistance firms. (ODA) to finance three of the four sections of rail (Si Racha-Laem Chabang, Sattahip- Although Thailand offered incentives Map Ta Phut, and Kiong Sip Kao-Kaeng to promote the decentralization of Khol). The Thai government financed the manufacturing, infrastructure was fourth (Chanchoengsao-Sattahip). Japanese more important ODA also financed the construction of the Chonburi-Pattaya Highway, the Bangkok- Since the 1960s, Thailand’s Board of Chonburi Highway, and the Outer Bangkok Investment (BOI) has used a combination of Ring Road. fiscal incentives to engineer the decentral- The fourth component, the water supply ization of manufacturing production away system, has been developed to meet industrial from the immediate vicinity of Bangkok. demands, as well as the residential demands For the scheme, BOI divided the country of the growing population serving the into three zones and offered different incen- industries. Nong Pla Lao Dam is the main ­ tives in each zone, including reductions in A p p r a i s i n g T r a n s p o r t C o r r i d o r s i n J a p a n , E u r o p e , a n d T h a i l a n d    195 import tariffs and temporary exemptions The Thai automotive industry has from the corporate income tax and (after steadily increased Its production, value 2013) nontax incentives, such as permits added, and exports, thanks to good for foreigners to buy land or for skilled policies and connectivity workers and experts to work in Thailand. Zone 1 included the five provinces immedi- From 1960 to 1997, the automotive industry ately adjacent to Bangkok, including Samut pursued a program of import substitution, Prakan, where Toyota is located. Zone 2 replacing imports with domestic production. consisted of nine provinces, including During this phase, industrial output Chonburi and Ayuthaya, where Mitsubishi, fluctuated with domestic demand. During ­ Ford, Mazda, and Honda are located. the economic boom from 1987 to 1996, Zone 3 included the remaining 62 of when real GDP grew at almost 10 percent Thailand’s 76 provinces, all more distant per year (Warr 2005), the automotive indus- from Bangkok. No automotive producer try expanded rapidly, reaching an output of has ever located in Zone 3. Although a roughly half a million units in 1996. With rationale for encouraging firms to locate in the collapse of demand during and after the Zone 3 could be made based on the lower Asian Financial Crisis of 1997−98, output incomes of the provinces there, poor infra- plummeted to just over one-fifth of this level. structure prevented significant manufactur- Over the next 15 years, the policy changes ing development from occurring in those and infrastructure investments described ­ provinces. The incentives offered were not produced a resurgence of output, which sufficient to overcome this drawback. reached around 2 million units in 2015 The BOI’s decentralization policy did ­(figure 5.6, panel a).20 The export share of not succeed. this output grew from roughly zero in 1997 To some extent, the BOI’s incentive to over 60 percent in 2015 ­ ( figure 5.6, structure was at variance with the govern- panel b). In the 20 years after 1995, immedi- ment’s infrastructure policy. The Eastern ately before the crisis, sales to the domestic Seaboard scheme was explicitly intended to market grew from 0.5 to 0.8 million units concentrate scarce infrastructure resources and exports expanded from near zero to along the southeastern corridor connected 1.2 million units. to the Laem Chabang port, all within the In 2014, automotive exports earned BOI’s Zones 1 and 2. At the same time, the US$33.6 billion, 16 percent of total merchan- BOI was attempting, unsuccessfully, to dise exports and 19 percent of total manufac- encourage manufacturing firms to locate tured goods exports. Of this total, vehicle outside this area. The Eastern Seaboard exports made up just over half, while parts scheme assumed that workers would relo- and components made up the remainder. cate to where the jobs could be found. The Automotive imports totaled US$13.5 billion, BOI’s incentives assumed that industry of which only 15 percent consisted of could be induced to locate wherever work- vehicles, while the remainder was parts and ­ ers lived. The latter scheme did not work, components. Around one-quarter of all vehi- ­ and the decentralization objective was cle exports were to other ASEAN (Association abandoned in 2013. The new system is said of Southeast Asian Nations) countries to be intended to encourage high-technol- (reflecting the 1992 ASEAN Free Trade ogy, skill-intensive investments. It remains Agreement) and another quarter to Australia to be seen whether this strategy will be (reflecting the 2005 Thailand-Australia Free important for the future of the automotive Trade Agreement). Perhaps surprisingly, other industry, but experience does not suggest ASEAN countries are the largest source that the BOI’s incentives will have much for Thailand’s vehicle imports, followed effect on firms’ decisions. by the European Union (EU) and Japan. 196  T h e WEB o f T r a n s p o r t C o r r i d o r s i n S o u t h A s i a FIGURE 5.6  Thailand’s automotive output and exports have soared since the mid-1990s a. Vehicle production, 1960–2015 (million units) b. Export share of automotive output, 1995–2015 3.0 3,000 70 2.5 2,500 60 Percent of total production 50 2.0 2,000 10,000 units Million units 40 1.5 1,500 30 1.0 1,000 20 0.5 500 10 0 0 0 60 65 70 75 80 85 90 95 00 05 10 15 95 97 99 01 03 05 07 09 11 13 15 19 19 19 19 19 19 19 19 20 20 20 20 19 19 19 20 20 20 20 20 20 20 20 Production (1,000 units) left-hand side Exports (% of total production) right-hand side Sources: Thai Automotive Industry Association; Federation of Thai Industries; and Automotive Intelligence Unit, Thailand Automotive Institute, Bangkok. Other ASEAN countries are the main desti- then has been due to the high electronics nation for exports of parts and components content of vehicles, which require more exports, followed by Japan and the United sophisticated imports. The abolition of States. local content requirements apparently was From just over 5 percent of manufactur- attractive to final assemblers. The lesson is ing value added before the Asian Financial that export-oriented manufacturers will Crisis, the value-added share of Thailand’s attempt to source their parts locally when automotive industry had doubled to 10 they can—but they do not want to be com- percent by 2014 (figure 5.7, panel a). The pelled to do so. industry’s employment share within manu- Comparing Thailand’s automotive produc- facturing is estimated at roughly 4 percent. tion and export performance with those of The difference between this share and its neighboring Malaysia and Indonesia high- value-added share reflects the high capital lights the policy differences between these i n t e n s i t y o f t h e a u t o m o t i v e s e c t o r. countries ­(figure 5.8). Malaysia and Indonesia Commercial vehicles (primarily 1-ton pick- were both committed to national car policies. ups) represent about 60 percent of Thailand’s Foreign ownership was restricted and local total vehicle output. This share has declined content requirements were enforced, as they steadily, from around 70 percent in the early were in Thailand before 1997. In 1999, 1990s, replaced by passenger vehicles. Thailand’s vehicle output was only slightly A striking feature of the Thai auto indus- larger than Malaysia’s. By 2015, it was more try is that the import content of vehicles than three times Malaysia’s output. The com- produced in Thailand has declined steadily parison is even more dramatic in the case of since the early 1990s (figure 5.7, panel b). exports. Malaysia’s automotive exports have This started occurring well before the local grown only marginally compared with content requirements were abolished in Thailand’s. Indonesia has performed better 2000. The steep decline continued until than Malaysia in both respects, but still much around 2005. The moderate increase since less well than Thailand. A p p r a i s i n g T r a n s p o r t C o r r i d o r s i n J a p a n , E u r o p e , a n d T h a i l a n d    197 FIGURE 5.7  The net value added of Thai automotive exports has increased greatly since the late 1990s a. Value-added share of automotive sector to total b. Value of imported parts per locally assembled vehicle, manufacturing, 1993–2014 1988−2014 12 10 9 10 8 7 US$million/1,000 units 8 6 Percent 6 5 4 4 3 2 2 1 0 0 93 95 97 99 01 03 05 07 09 11 13 88 90 92 94 96 98 00 02 04 06 08 10 12 14 19 19 19 19 20 20 20 20 20 20 20 19 19 19 19 19 19 20 20 20 20 20 20 20 20 Sources: Panel a: National Economic and Social Development Board, Bangkok. Panel b. Calculations by Warr and Kohpaiboon (2017), based on official data sources. Data on the import value of parts are drawn from UN Comtrade. Vehicle production data are from the Thai Automotive Industry Association; Federation of Thai Industries; and Automotive Intelligence Unit, Thailand Automotive Institute, Bangkok. Different kinds of vehicle assembly tend to locate in the area surrounding car firms tend to cluster differently along the assembly plants. For example, the number of corridor and have different impacts on parts supplier plants located in Samut Prakan employment province increased from 56 (in 1996) to An industrial survey of plant-level data for 122 (in 2006) and to 144 (in 2011). These three representative years—1996, 2006, and parts suppliers have been crucial to the devel- 2011—reveals that vehicle assembly includes opment of the Thai automotive sector. two quite different kinds of firms.21 Large, Domestically owned firms are smaller and multinational car manufacturers engage in sig- more labor intensive, as measured by output nificant manufacturing within Thailand and per worker. Turnover among firms is higher produce within very large plants. Small, Thai- among the Thai-owned input suppliers. This owned assemblers produce for niche markets is indicated by the average age of plants within Thailand. The latter include firms that responding to the surveys in the three years assemble buses and certain types of trailer covered. In 1996, the average age of Thai- trucks using imported new or used engines. owned plants exceeded the average age of These firms undertake very little actual manu- foreign-owned plants, but by 2011 this differ- facturing activity within Thailand. ence had been reversed. Over the five-year The large, foreign-owned vehicle assem- interval between 2006 and 2011, the average blers are linked to many parts suppliers. age of foreign-owned input suppliers These suppliers tend to be small to medium increased by roughly five years, but the aver- sized and include both foreign- and domesti- age age of Thai-owned suppliers increased by cally owned firms. New parts supplier plants only half as much—even though the number 198  T h e WEB o f T r a n s p o r t C o r r i d o r s i n S o u t h A s i a FIGURE 5.8  Thailand outperformed Malaysia and Indonesia in number of vehicles produced and their export value, thanks to good policies a. Automobile production, Thailand, Malaysia, and Indonesia,1999−2015 2,500,000 2,000,000 1,500,000 Units 1,000,000 500,000 0 99 00 01 02 03 04 05 06 07 08 09 10 11 12 13 14 15 19 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 b. Export value of automobiles,Thailand, Malaysia, and Indonesia, 1999−2014 30,000 25,000 20,000 US$ million 15,000 10,000 5,000 0 00 01 02 03 04 05 06 07 08 09 10 11 12 13 14 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 99 19 Thailand Malaysia Indonesia Source: Compilations by Warr and Kohpaiboon (2017) from the UN Comtrade database, using the WITS (World Integrated Trade Solution) website, http://wits.worldbank.org/. of Thai firms increased only marginally. between 1996 and 2006. Assemblers and Many Thai firms had left the industry, to be parts suppliers have located along the high- replaced by others. way system connected to the Laem Chabang The location of the final assembly and auto port. Parts suppliers have tended to locate parts industries changed during the decade close to the final assemblers. A p p r a i s i n g T r a n s p o r t C o r r i d o r s i n J a p a n , E u r o p e , a n d T h a i l a n d    199 Econometric analysis suggests that the The development of the Thai automotive industry, entry of foreign input suppliers after 1997 did enabled by a set of transport infrastructure not have positive spillover effects on domestic investments, created hundreds of thousands of suppliers (Warr and Kohpaiboon 2017). The differential between the two groups in output manufacturing jobs and led to a small but significant per firm, capital intensity, and labor produc- reduction in poverty. tivity was significant in each case and did not converge over time. Did the long period of only for 2006. The “inside” mean for that local content requirements before 1997 have year is more than three times the “outside” lasting effects on the productivity of domestic mean. For local firms, the “inside” mean is at input suppliers, relative to foreign suppliers? least twice the “outside” mean in each of the The findings from the industrial surveys sug- three years. Performing similar calculations gest that the answer is no. for parts suppliers reveals similar results. The means of labor productivity are again higher “inside” than “outside,” except for foreign Public infrastructure significantly raised firms in 1996 and 2011, where the “outside” labor productivity among both foreign means are higher. and domestic final assemblers and parts Recalling that the Industrial Census is a suppliers sample survey of only selected firms rather Beginning with the development of the Laem than a true census of all firms, it makes sense Chabang port, the Thai government invested to ask whether these differences on the in infrastructure upgrades in the eight prov- ­ sample-based mean estimates of labor produc- inces close to the Eastern Seaboard scheme tivity are statistically significant. The results of (Bangkok, Samut Prakan, Nonthaburi, a significance test between the average labor Pathum Thani, Ayutthaya, Chon Buri, productivity inside and outside the regions Rayong, and Chachoengsao). The aim was to with improved infrastructure are presented in improve the investment climate for manufac- figure 5.9.22 For foreign assemblers and parts turing firms—including, but not solely, auto- suppliers in 2006, labor productivity is signifi- motive final assemblers and parts suppliers. cantly higher “Inside” than “Outside.” This is These infrastructure upgrades consisted of also true for local final assemblers and local investments in improved roads, industrial- parts suppliers. However, results for 1996 and capacity supplies of electricity and water, and 2011 are less conclusive on this significant dif- telecommunications. Infrastructure upgrades ference because of some missing data in 1996 in the other 68 provinces were significantly and a problematic response rate to the survey less extensive. If successful, the infrastructure in 2011 (as discussed earlier) that are not investments should have raised labor produc- reported in figure 5.9. tivity relative to those areas that did not receive similarly favorable treatment. The Wider economic effects of Industrial Census data can be used to investi- infrastructure improvements gate whether the intended effect was achieved. This study calculated labor productivity During the development period of 1981−95, both inside and outside the improved infra- the ESDP had a positive effect on growth of structure regions. This is done for each of the per capita GDP and population on the three years of the Industrial Census and for Eastern Seaboard. Manufacturing growth both foreign and local firms. The Industrial propelled the region to produce the second- Census records no foreign final assemblers largest value added of all regions in the coun- outside the region with improved infrastruc- try (table 5.1). ture (the eight provinces) in 1996 and 2011. Map Ta Phut flourished as a petrochemical Thus, for foreign final assemblers, the center, mainly because of the lack of domestic “inside/outside” comparison can be made competition and the growing domestic 200  T h e WEB o f T r a n s p o r t C o r r i d o r s i n S o u t h A s i a FIGURE 5.9  Labor productivity inside and outside the improved demand for products. In contrast, Laem infrastructure regions Chabang became a center for the automotive electronics industries, deviating from an ini- 3.0 tial intention of the government to make it a light industry hub. This growth was on the back of inflow of foreign direct investment from Japan, and the shift of production from Value added per worker 2.0 Bangkok. The impact was not confined to these two locations. It also extended to the areas between them and Bangkok, through a series 1.0 of industrial parks, industrial estates, and industrial zones (map 5.8). The heavy industrialization was accompa- nied by some negative side effects, including foul smells and water pollution from the 0 Foreign Local Foreign Local heavy chemical plants. A Joint Committee consisting of the Industrial Estate Authority Final assemblers Parts suppliers of Thailand, the Office of Environmental Inside Outside Policy and Planning, the Pollution Control Source: National Statistical Office, Industrial Census 1997, 2007, 2012. Department, and the Department of Industrial Note: Data are as of 2006. Works was established to deal with these TABLE 5.1  The Eastern Seaboard had the second-best set of macroeconomic indicators among regions of Thailand during its development period of 1981−96 Bangkok and Nation­w ide Vicinity Eastern (ESB) Central Western North­eastern Northern Southern GDP per capita 1981 20,278 63,198 26,212 17,845 18,610 7,860 12,402 15,740 (35,564) 1995 49,514 149,592 80,232 48,558 37,295 16,631 23,681 31,735 (121,376) GDP per capita (% yoy) 1981–86 3.4 2.2 5.8 (7.6) 2.5 3.5 3.7 3.5 3.0 1986–91 9.3 11.0 8.4 (8.5) 9.5 5.5 6.2 5.3 7.2 1991–95 7.3 6.0 11.5 (12.1) 11.2 6.6 7.0 5.5 5.2 Manufacturing value added (ratio to the whole country) 1981 100.0 72.2 11.2 (10.6) 3.3 3.1 3.9 3.5 2.7 1995 100.0 63.2 15.8 (14.9) 6.5 3.6 5.0 3.8 2.1 Population growth (% yoy) 1981–86 1.9 2.9 2.2 (1.5) 1.1 1.6 1.7 1.4 2.5 1986–91 1.5 2.6 2.1 (2.5) 1.3 0.8 1.3 1.1 1.6 1991–96 1.4 0.6 2.3 (1.9) 0.7 1.9 1.4 2.0 1.0 Source: Japan International Cooperation Agency. Note: ESB = Eastern Seaboard; yoy = year on year. A p p r a i s i n g T r a n s p o r t C o r r i d o r s i n J a p a n , E u r o p e , a n d T h a i l a n d    201 environmental problems. It identified seven MAP 5.8  A network of industrial parks, industrial estates, and factories for guidance on improvements. industrial zones spread from the Deep-Water Port to Bangkok What role did the ESDP play in the wider economic benefits, such as GDP growth? Answering this question is problematic. More rigorous evaluation of the likely impacts of better infrastructure on the auto- motive sector and economic benefits is needed. To do so, this study drew on an existing general equilibrium model of the Thai economy, known as JamlongThai. Using this model, this study analyzed the effect infrastructure improvements had on reducing the production and distribution costs of automotive firms. The purpose is to ascertain the effect that infrastructure devel- opment on the Eastern Seaboard had on automotive output, exports, and employ- ment. This analysis, in turn, can be used to examine the effect that these industry out- comes have had on the substantial reduction of poverty incidence that has occurred in Thailand in recent decades. The JamlongThai model is a 65-sector gen- eral equilibrium model with a highly disag- gregated household structure. This disaggregation of households permits the model to produce estimates of the effects that economic events have on poverty incidence within Thailand. The model is documented in Warr (2010). Based on the interviews with industry Source: Tokyo Development Consultants (Thailand) Co., Ltd. participants, this study estimated that infra- Note: BOI = Board of Investment; I.E. = industrial estate; I.P. = industrial zone; I.Z. = industrial zone. structure improvements reduced the costs of export-oriented firms by around 15 percent.23 The database of the JamlongThai as ­a ssembly—​ r eal GDP would have been model relates to 2007. Thus, it incorporates 0.9 percent lower (table 5.2). Put differently, the impact of the cost-reducing infrastruc- the estimates indicate that the 15 percent ture investments that had occurred before productivity gain in the automotive sector then. The simulations estimate the unob- raised the level of annual real GDP in served counterfactual in which these cost Thailand by just under 1 percent. It is reductions did not occur. The estimated important to recognize that this is a perma- impact of the cost reductions is thus the dif- nent increase in the level of GDP per year ference between the observed value of the and not just a temporary one-year increase variables and the simulated counterfactual in the level of GDP. The cost reductions led value of these variables in which the cost to an increase in the level of skilled real reductions did not happen. wages of 1.15 percent and an increase in The simulation indicates that in the unskilled real wages of 0.75 percent. The absence of the productivity gain in the whole high skill intensity of the automotive sector automotive sector—including parts as well explains the differential in wage effects. 202  T h e WEB o f T r a n s p o r t C o r r i d o r s i n S o u t h A s i a TABLE 5.2  Simulated macroeconomic effects population of 60 ­ million people, this means Simulation: Final that 0.12 ­million (120,000) people moved Measure assembly plus parts from levels of real consumption below the Real GDP −0.907 poverty line to levels above it. This number Real household consumption −1.537 includes both poverty reduction among those GDP price index −0.428 directly employed in automotive parts and Consumer price index −0.667 Wages −1.147 assembly firms and among those receiving   Paid skilled −0.749 remittances from these workers. Investments   Paid unskilled in infrastructure that raised productivity   Unpaid skilled −5.071 enabled the automotive sector to contribute   Unpaid unskilled −1.021 to poverty reduction in Thailand. However, Average capital rental −1.601 when viewed from within the parameters of Output of petroleum −1.254 the simulation, those infrastructure invest- Government revenue −0.820 ments do not appear to have been a major Government expenditures −0.992 driver of the huge reductions in poverty inci- Government budget balance (million baht) 871.308 dence that have occurred. It is important to note that these calculations do not measure Source: Warr and Kohpaiboon (2017) calculations, using the JamlongThai model of the Thai economy. the total poverty-reducing effects of the Note: Values are percent change, unless stated). ­ i nfrastructure investments described, but only those that operated via the automotive sector. Thus, the full effects would be much These wage effects occurred through an larger. increase in the demand for labor resulting from the growth of automotive output. In the absence of productivity gains, output would have been lower, producer and con- Lessons from the Thai case study sumer prices would have been higher, Development of the infrastructure supporting exports would have been lower, imports an efficient export gateway (Laem Chabang would have been higher, and the domestic port and the associated Eastern Seaboard cor- consumption of automobiles would have ridor) was a necessary condition to achieve been considerably lower. the Thai government’s goal to reorient the The estimated effects on the incidence of country to heavy manufacturing in the late poverty are shown in figure 5.10. Without the 1980s and early 1990s. The policy reorienta- increases in automotive productivity, poverty tion built on the choice that major manufac- incidence would have been higher, especially turers around the world were making to in urban areas. Relative to the massive reduc- relocate production internationally to lower- tions in poverty incidence that have occurred cost venues. Financial support and intellectual in Thailand, these estimated impacts are sig- inputs from Japanese sources played an nificant, but small. Holding the real value of important role in developing the Eastern the poverty line constant, between 1986 and Seaboard c ­ orridor, which paved the way for 2014, poverty incidence in Thailand declined Thailand to become a new manufacturing from 67 percent to 11 percent of the total center for Japanese automotive companies. population. This remarkable achievement Nevertheless, Thailand’s infrastructure devel- implies that over 28 years, the incidence of opment was not specific to the automotive poverty declined by 56 percent of the popula- industry, and the growth of the automotive tion, at an average rate of decline of 2 percent sector was not anticipated by the planners of the population per year. The estimates in concerned. figure 5.10 mean that the reductions in auto- Thailand’s automotive sector has not motive costs reduced poverty incidence by an become the “Detroit of the East” because estimated 0.2 percent of the population. In a the industry is largely foreign owned. A p p r a i s i n g T r a n s p o r t C o r r i d o r s i n J a p a n , E u r o p e , a n d T h a i l a n d    203 FIGURE 5.10  Simulated effects on the incidence of poverty (headcount measure) 1.9 1.8 1.77 1.7 1.6 Percentage point change 1.5 0.5 0.4 0.3 0.32 0.27 0.2 0.18 0.17 0.1 0.06 0.05 0.05 0 National Urban Rural Bangkok Central North Northeast South and southeast Regional Source: Calculations by Warr and Kohpaiboon (2017), using the JamlongThai model of the Thai economy. Note: Positive numbers mean increases in the simulated level of poverty incidence. Changes in poverty incidence are reported as the difference between the simulated level of poverty incidence (post-shock) and the initial level (pre-shock), both expressed in percentage form. For example, in 2007 (the year of the model’s database) the initial (pre-shock) level of poverty incidence for the total population was 14.36 percent. The reported change in national poverty incidence above is 0.18 percentage point, meaning that the simulated level (post-shock) is 14.36 + 0.18 = 14.54 percent. That is, it is estimated that without the effect of the productivity-enhancing effect that infrastructure investments had on the automotive industry, in 2007 poverty incidence would have been 0.18 percentage point higher, at 14.58 percent of the population. Nevertheless, the growth of the automotive components directly to them (tier-1 input sector generated hundreds of thousands of ­ suppliers). Not many of the existing indige- manufacturing jobs that would not otherwise nous input suppliers survived this period, but have existed. Thailand avoided the failed those that did mainly became tier-2 suppliers, “national car” policies of some of its supplying inputs to the tier-1 suppliers. ­ neighbors. It permitted full foreign ownership Thailand’s investments in transport and of vehicle manufacturing, but did not elimi- other infrastructure reduced costs for manu- nate its high rates of protection of final facturers and thus contributed greatly to the vehicles. These tariffs were largely irrelevant to ­ development of the Thai automotive sector. the development of the export-­ oriented com- This development led to a small but signifi- ponent of the automotive industry. Thailand’s cant reduction in poverty. steps to liberalize input supplies by abolishing local content requirements, and to become an Summing up and moving to the next export platform, ironically facilitated higher, chapter not lower, local content. Following the relax- ation of restrictions on foreign entry of input This chapter has conducted a qualitative suppliers (1997), multinational enterprise final appraisal of major transport corridors around assemblers often preferred domestically the world by means of narrative critiques. located but foreign-owned firms that supplied The three transport corridor projects 204  T h e WEB o f T r a n s p o r t C o r r i d o r s i n S o u t h A s i a examined were Japan’s Pacific Belt Zone 7. Because accurate statistical data specific to Initiative, Europe’s high-speed train projects, the Pacific Belt Zone do not exist, data for and the emergence of Thailand’s automotive economic and social statistics by prefecture industry with the help of robust investment in have been used. Specifically, this analysis uses data for 19 prefectures: Ibaraki, Saitama, transport infrastructure. Chiba, Tokyo, Kanagawa, Shizuoka, Aichi, Building on this qualitative assessment, Gifu, Mie, Osaka, Hyogo, Wakayama, chapter 6 performs a quantitative appraisal of Okayama, Hiroshima, Yamaguchi, Kagawa, two contemplated transport corridors in the Ehime, Fukuoka, and Oita. These prefectures eastern and western parts of South Asia. Using make up 27.7 percent of Japan’s total land the holistic appraisal methodology developed area and accounted for 79.3 percent of the by the report, it draws on spatial data, econo- value of Japan’s total exports of products metrics, and simulations to appraise the pro- manufactured or fabricated in 1970. posed China-Pakistan Economic Corridor in 8. Public Opinion Survey on National Life Pakistan and the Kolkata-Dhaka transport (Cabinet Office), 1958−70. corridor in Bangladesh. 9. Japan’s Staple Food Control Act required the Government of Japan to purchase rice at the fixed price determined considering labor NOTES costs in urban areas. The act was passed in 1942 to secure stable food supplies and stabi- 1. The four major industrial zones are Keihin lize inflation. It was abolished in 1995. Industrial Area, which includes Tokyo; 10. The figures are the totals for the following Chukyo Industrial Area, which includes 15 domains: roads, ports and harbors, avia- Nagoya; Hanshin Industrial Area, which tion, public rental housing, sewerage and includes Osaka; and Kita-Kyusyu Industrial drainage services, waste treatment, water Area, which includes Fukuoka. supply services, urban parks, education and 2. The so-called Income Doubling Plan was pro- cultural facilities, flood control, soil saving, posed by the Industrial Location Subcommittee coast administration, agriculture/forestry/ of the Economic Council established under fisheries, national forestry administration, the then–Ministry of International Trade and and industrial water management. Industry. The Cabinet approved the Plan in 11. To put this total in context, the national 1960. budget totaled ¥3.3 trillion in 1964, and ­ 3. The Report on the Economic and Technical ¥88.0 trillion in 1998. Feasibility of the Proposed Kobe-Nagoya 12. Analysis by Shimokobe Atsuchi, a former Expressway. senior bureaucrat who played a central role in 4. The improvement rate is the ratio of the drawing up and revising the Comprehensive length of road improved (in conformity with National Development Plan, cited in Economist the standards specified in the Road Structure Hensyu-bu (1999, 102−16). Ordinance) to the total road length. Data 13. The figures for sewerage system coverage from the Government of Japan’s Road come from data that cover all of Japan. For Statistics Annual Report. areas where a sewerage system was not avail- 5. Social capital refers to capital related to able, downsized septic tanks for combined roads, ports/harbors, rail, airports, subways, treatment were developed to treat miscella- public apartment houses, sewage, waste neous wastewater and sewage together. A disposal, water supply, urban parks, schools, ­ subsidy program encouraged households social education/culture, river control, forestry across Japan to adapt the septic tanks. conservancy, coastal protection, agriculture, 14. In all, 41 districts—all in the Pacific Belt forestry, fishery, national forests, industrial Zone—were designated as areas suffering water, and postal services. from severe air pollution and a high inci- 6. All of these figures are total values on a nation- dence of bronchial asthma and other pollu- wide basis. Because of limitations of data on tion-related diseases under the Compensation transportation and power, they cannot be cal- System for Pollution-Related Health Damage culated separately for the Pacific Belt Zone. launched in 1974. The aggregate number of A p p r a i s i n g T r a n s p o r t C o r r i d o r s i n J a p a n , E u r o p e , a n d T h a i l a n d    205 certified patients as of December 31, 1988, from 2006 to 2011. In actuality, no firm left was 108,489 (Ministry of the Environment). the industry, but many did not respond to the 15. A plant was finally built in Suginami in 1982. survey. For this reason, the comparison 16. Japan’s four major pollution-related diseases between the 1996 and 2006 surveys is possi- are Minamata disease (Kumamoto Prefecture), bly the most instructive. Niigata Minamata disease (Niigata Prefecture), 22. The analysis assumes that the sample is an Yokkaichi asthma (Mie Prefecture), and Itai- unbiased random sample from the overall Itai disease (Toyama Prefecture). population of relevant firms. The null 17. LGV Sud-Est is often regarded as the first hypothesis is that true labor productivity for high-speed line in Europe. However, the first the full population is the same inside and high-speed line was actually the “Direttissima” outside the improved infrastructure regions. line in Italy connecting Florence and Rome, The alternative hypothesis is that “inside” which opened in 1978. Unlike LGV, the productivity is higher. Can the null hypothe- Direttissima was not designed exclusively for sis be rejected in favor of the alternative high-speed passenger services, which partly hypothesis? This can be tested by calculating explains its relative obscurity. the t-statistic for the estimated mean differ- 18. The total employment is based on industry ence and comparing it with the critical values sources. Thailand’s manufacturing work- from a one-tail t-test. The t-test is one-tailed force totals 6.5 million. because the alternative hypothesis is that 19. The eight provinces constituting this corridor labor productivity is higher inside than out- are Bangkok, Samut Prakan, Nonthaburi, side the improved infrastructure region, not Pathum Thani, Ayutthaya, Chonburi, Rayong, just that it is different, which would corre- and Chachoengsao. Within the corridor, spond to a two-tailed test. industrial clusters link final manufacturers 23. In June and July 2016, the authors con- and parts suppliers. For example, a major ducted interviews with representatives from cluster in Samut Prakan province centers on major automotive manufacturers, current Toyota, one in Ayutthaya centers on Honda, and former government officials who were and another in Rayong centers on a commer- closely involved with the development of cial alliance between Ford and Mazda. Thailand’s automotive industry, and other 20. Output surged temporarily in 2012 and experts. Government agencies and officials 2013. An initiative of the populist government interviewed included the Ministry of of Prime Minister Yingluck Shinawatra Industry and Industrial Estate Authority of (2011−14) to provide households with tax Thailand (IEAT), Bangkok; National rebates for the purchase of new passenger vehi- Economic and Social Development Board, cles stimulated domestic demand by more than Bangkok; Siriruj Chulakaratana, chief, half a million units a year in the next two Office of Industrial Economics, Ministry of years, leading to total output of 2.5 million Industry, Bangkok; Sanoh Unakul, TDRI units. Output contracted correspondingly founder, former deputy prime minister, for- when a military coup in July 2014 led the mer director-general, NESDB; IEAT, Port policy to be abandoned. Authority of Thailand, and representatives 21. The industries of interest in this study are from the automobile industry, Laem Cha ISIC 3410 (manufacture of motor vehicles) Bang Industrial Estate, Chonburi Province. and ISIC 3430 (manufacture of parts and Representatives from manufacturers accessories for motor vehicles and their included Paiboon, Isuzu Sales (Thailand), engines). Although the data are intended to Bangkok; Suparat Sirisuwanangkura, senior cover all producers in these industries, the president, Toyota Motor Thailand Co., vice ­ actual response rate has varied widely across Ltd; Trevor Nugus, CEO, Auto Alliance, years. This was particularly important in the Rayong Province; and Asanee Kulakow, case of the 2011 census, which was severely Mitsubshi Motors, Chonburi Province. disrupted by flooding in central Thailand. Kanit Sansuprann, Thai Airways, Bangkok, Many firms did not respond. For example, and Ruth Banomyong, Thammasat the data indicate that the number of large University Business School, were also final assembly plants declined substantially interviewed. 206  T h e WEB o f T r a n s p o r t C o r r i d o r s i n S o u t h A s i a REFERENCES http://ec.europa.eu​/­transport/sites/transport​/­files​ /­t h e m e s​ /­i n f r a s t r u c t u r e / n e w s / d o c / 2 0 1 5​ Albalate, D., and G. Bel. 2012. The Economics -05-28-coordinator-work-plans/wp_ra_final​ and Politics of High-Speed Rail: Lessons from .pdf. Experiences Abroad. Lanham, MD: Lexington Economist . 2013. “Thailand’s Booming Car Books. Industry−Detroit of the East.” April 4. https:// Anastasiadou, I. 2011. Constructing Iron Europe: www.economist.com/blogs/schumpeter/2013/04​ Transnationalism and Railways in the /­thailands-booming-car-industry. Interbellum. Amsterdam: Amsterdam Ecoplan and Infras. 2011. “Wirtschaftlichkeitsstudie University Press. NEAT 2010–Hauptbericht.” https://www.bav​ Bonnafous, A. 1986. “The Regional Impact of the .admin.ch/dam/bav/de​/­d okumente/aktuell​ TGV.” Transportation 14 (2): 127−37. -startseite/berichte​/­wirtschaftlichkeitsstudieneat Campos, J., G. de Rus, and I. Barron. 2009. 2010.pdf.download.pdf​/­wirtschaftlichkeitsstudi “A Review of HSR Experiences around the eneat2010.pdf. World.” In Economic Analysis of High Speed EIB (European Investment Bank). 2003. “EIB Rail in Europe, edited by G. de Rus, 19−32. £400 Million Loan for Completion of Channel Informes 2009, Economía y Sociedad. Bilbao, Tunnel Rail Link.” http://www.eib.org​ Spain: Fundación BBVA. /­i nfocentre/press/releases​ / a ­ ll​ / 2003​ / 2003​ Campos, J., and P. Gagnepain. 2009. “Measuring -050-eur-560-mio-for-completion​-of-channel​ the Intermodal Effects of High Speed Rail.” In -tunnel-rail-link-ctrl.htm. Economic Analysis of High Speed Rail in ERT (European Round Table of Industrialists). Europe, edited by G. de Rus, 71−88. Informes 1984. Missing Links: Upgrading Europe’s 2009, Economía y Sociedad. Bilbao, Spain: Transborder Ground Infrastructure. Paris: ERT Fundación BBVA. Secretariat. Chen, C.-L., and P. Hall. 2012. “The Wider EU (European Union). 2010. High Speed Europe: Spatial-Economic Impacts of High-Speed A Sustainable Link between Citizens. Trains: A Comparative Case Study of Luxembourg: Publications Office of the Manchester and Lille Sub-regions.” Journal of European Union. Transport Geography 24: 89−110. Gilbert, R., and A. Perl. 2012. Transport Chen, C.-L., and R. Vickerman. 2017. “Can Revolutions: Moving People and Freight without Transport Infrastructure Change Regions’ Oil. New York: Routledge. Economic Fortunes? Some Evidence from Gravier, J.-F. 1947. Paris and the French Desert. Europe and China.” Journal of Regional Paris: Partulan. Studies 51 (1): 144−60. Grazhdankin, A. S. 2012. “The Role of the River Cheng, Y.-S., B. P. Y. Loo, and R. W. Vickerman. Rhine in the Formation of the Spatial Structure 2015. “High-Speed Rail Networks, Economic of the Economy of European Countries Integration and Regional Specialisation in (1st Century BC−19th Century AD).” Baltic China and Europe.” Travel Behaviour and Region 2: 100−08. Society 2: 1–14. Greengauge. 2006. “High Speed Trains and Crozet, Y. 2013. “Performance in France: From the Development and Regeneration of Cities.” Appraisal Methodologies to Ex-Post http://www​.­g reengauge21.net/wp-content​ Evaluation.” OECD Discussion Paper /­uploads/hsr​-­regneration-of-cities.pdf. 2013−26. Prepared for the Roundtable on the Heidenreich, M. 1998. “The Changing System of Economics of Investment in High Speed Rail, European Cities and Regions.” Journal of December 18−19, New Delhi. European Planning Studies 6 (3): 315−32. Drewello, H., and B. Scholl. 2016. Integrated Hensyu-bu. 1999. Kodo Seicho-ki eno Shogen S pa tia l a n d Tr a n s p o r t I n fr a str uc tur e (Jyo) (Ge) (Testimony to the High Economic Development. London: Springer. Growth Period). Tokyo: Nihon Keizai Dunn Jr., J. A., and A. Perl. 1994. “Policy Hyouronsha Inc. Networks and Industrial Revitalization: High- Hickman, R., D. Bonilla, and D. Banister. 2015. Speed Rail Initiatives in France and Germany.” International Handbook on Transport and Journal of Public Policy 14 (3): 311−43. Development. Cheltenham, United Kingdom: EC (European Commission). 2015. “Rhine Alpine: Edward Elgar Publishing. Work Plan of the European Coordinator.” A p p r a i s i n g T r a n s p o r t C o r r i d o r s i n J a p a n , E u r o p e , a n d T h a i l a n d    207 Japan, Cabinet Office, Government of Japan. Points for China . Institute for Industrial National Accounts of Japan. Tokyo. Productivity. http://www.iipnetwork.org​ ———. Public Opinion Survey Concerning /­I ndustrial%20Restructuring%20in%20 People’s Lifestyle. Tokyo. the%20Ruhr%20Valley.pdf. JICA (Japan International Cooperation Agency). Ureña, J. M, M. Philippe, and M. Garmendia. 1999. “Eastern Seaboard Development Plan 2009. “The High-Speed Rail Challenge for Big Impact Evaluation.” https://www.jica.go.jp​ Intermediate Cities: A National, Regional and /­e nglish/our_work/evaluation/oda_loan​ Local Perspective.” Cities 26 (5): 266−79. /­post/2000/pdf/01-01.pdf Vickerman, R. 1995. “The Channel Tunnel: The Johnson, D., and C. Turner. 1997. Trans-European Case for Private Sector Provision of Public Networks: The Political Economy of Infrastructure.” In Transport and Urban Integrating Europe’s Infrastructure. London: Development, edited by D. Banister. London: Macmillan. Chapman and Hall. Melibaeva, S., J. Sussman, and T. Dunn. 2011. ———. 1997. “High-Speed Rail in Europe: “Comparative Study of High-Speed Passenger Experience and Issues for Future Development.” Rail Deployment in Megaregion Corridors: Annals of Regional Science 31: 21−38. Current Experiences and Future Opportunities.” ———. 2015. “High-Speed Rail and Regional Paper No. JRC2011-56115, Proceedings of the Development: The Case for Intermediate 2011 Joint Rail Conference, Pueblo, Colorado, Stations.” Journal of Transport Geography 42: March 16−18, 541−61. 157−65. Metron. 2009. Effect of the Gotthard Base Tunnel on Vickerman, R., and A. Ulied. 2009. “Indirect and the Transfer of Freight from Road to Rail. http:// Wider Economic Impacts of High Speed Rail.” www​.­alpeninitiative.ch/dms/­alpentransitboerse​ In Economic Analysis of High Speed Rail in /­doc​/­atb_studien/09-Studie-Metron_en/09%20 Europe, edited by G. de Rus, 89−103. Informes Studie%20Metron%20englisch.pdf. 2009, Economía y Sociedad. Bilbao: Fundación Meunier, J. 2002. On the Fast Track: French BBVA. Railway Modernization and the Origins of the Warr, P. 2005. “Boom, Bust, and Beyond.” In TGV. Westport, CT: Praeger. Thailand Beyond the Crisis, edited by P. Warr, Nora Report. 1967. Rapport sur les Entreprises 1−65. London: Routledge. Publiques. Paris: La Documentation Française. ———. 2010. “Thailand.” In Agricultural Peters, D. 2005. “Networking for Trans-national Price Distortions, Inequality and Poverty, ‘Missing Links’: Tracing the Political Success of edited by K. Anderson, J. Cockburn, and European High Speed Rail.” In The Network W. Martin, 283−99. Washington, DC: Society: A New Context for Planning, edited by World Bank. L. Albrechts and S. Mandelbaoum. New York: Warr, P., and Archanun K. 2017. “Thailand’s Routledge. Automotive Manufacturing Corridor.” ADB Priemus, H., B. Flyvbjerg, and B. van Wee. 2008. Economics Working Paper 519, Asian Decision-Making on Mega-Projects: Cost–Benefit Development Bank. https://www​ . adb.org​ Analysis, Planning, and Innovation. Cheltenham, /­p u b l i c a t i o n s / t h a i l a n d - a u t o m o t i v e​ United Kingdom: Edward Elgar Publishing. -manufacturing-corridor. Taylor, R. 2015. A Review of Industrial Yoshikawa, H. 2012. Kodo Seicho . Tokyo: Restructuring in the Ruhr Valley and Relevant Chuokoron-Shinsha. Spotlight 6 Appraising Proposed Transport Corridors Using Spatial Econometrics T his spotlight discusses the rela- tive a ­ bility of current econometric approaches to help policy makers appraise can motivate both the application of RF and GE econometrics using spatial data. Similarly, both the theories of trade gravity and NEG could the proposed placement and design of motivate application of least square ­ corridor intervention packages. Faced with regressions. many proposals for transport corridors with Several recent studies have summarized uncertain wider economic benefits (WEB) and systematized the approaches and find- and limited funding, policy makers are call- ings in the literature on the impacts of trans- ing for more rigorous appraisals of the place- port infrastructure investments. These studies ment and intervention design for transport use both more qualitative (Bosker and corridors. Adding urgency to the task, the Garretsen 2010; Straub 2011; Redding and tolerable margin for error in appraisals is Turner 2014; Berg et al. 2017) and more declining, including because of the high alter- quantitative (Melo, Graham, and Brage- native costs (for example, forgone opportu- Ardao 2013)1 approaches. Although they dis- nities from not investing more in other areas cuss the comparative advantages of different with social benefits, for example, education). modeling approaches (Redding and Turner Therefore, the approaches d ­ iscussed go 2014), they do not emphasize the ex ante beyond conventional cost-benefit analyses. perspective of project appraisals and the pos- Although many of them have been applied sible range of econometric approaches policy in evaluations after the project is completed makers could consider as more reliable. For (ex post), the emphasis here is on rigorous instance, Bakkera, Koopmans, and Nijkamp evaluation while the project is being designed (2010) take the project appraisal perspective, (ex ante) (Melecky, forthcoming). This spot- but focus mainly on cost-benefit analyses light groups the econometric approaches into (CBA) in the urban context. Laird and three groups: network econometrics; single- Veneables (2017) also take the project equation, reduced form (RF) econometrics; appraisal view, proposing disciplining struc- and structural general equilibrium (GE) ture for more rigorous CBA. But their useful econometrics. think piece stops at intermediate develop- The focus is not on the theories motivating ment outcomes, apart from employment. these applied econometric methods, because This spotlight takes an alternative perspective more than one theory could motivate the use to fill a possible gap in the literature, inform of a single econometric method. The theory of policy makers, and provide some direction new economic geography (NEG), for instance, for future research. 209 210   THE WEB OF TR A NSPORT C ORRIDORS IN SOUTH A SI A This spotlight does not go as far as review- network, accounting for road costs, typogra- ing the different objectives and preferences phy, and existing traffic flows (Gastner and that policy makers seek to optimize and that Newman 2006). The engineering and logis- could be empirically derived from data. tics optimization typically has not linked the To that end, Bougna et al. (forthcoming) pres- transport network optimization to broader ent a simple policy maker’s objective function development objectives such as trade, income, that considers multiple WEB. or environmental quality. But recently, the transport network optimizations that have penetrated economics have begun to examine NETWORK ECONOMETRICS such development objectives. Consequently, Network econometrics could be a useful hybrid papers that bridge the disciplines of approach to estimate how investments in engineering, logistics, and economics have transport corridors could trigger positive been emerging. spillovers to the wider economy, spur broader The idea of optimizing logistics is close to economic activity, and generate net WEB. The the idea of maximizing market access. It comparative advantage of network econo- makes sense for a transport corridor to be metrics could be in appraising the proposed placed in such a way as to maximize the placements of corridors—rather than the access to markets for both exporters and design of interventions to spread the eco- importers. In turn, this should create eco- nomic surpluses generated by the corridors nomic opportunities for those agents that from the aggregate level across spatial units become connected to markets for the first (such as state districts) and ultimately to time or in better ways. But how can the mar- people. ket access be relevant for higher develop- In its basic form, network econometrics ment objectives, such as household income endorses the revival of historical corridor and consumption, best be measured? One routes such as the Silk Road and the Grand simple way to measure market access has Truck Road.2 These proposals are driven by been to divide the sum of the population in the idea that the historical, rich networks of all domestic and foreign cities above X mil- agents have not completely disappeared and lion to which tradable goods and services still hold potential. The proposals assume can be transported from a given domestic that—despite new borders and other obsta- city by the travel distance from the domestic cles for movements of trade, capital, and city to all other domestic and foreign cities. labor—these rich networks have persisted in a The benefit of connectivity to markets of big restrained form, maintaining ties in terms of cities very far away is thus discounted to culture, business, and trust (Starr 2007). By zero by the increasing transport distance improving trunk infrastructure and providing (cost) (Burgess et al. 2015). For instance, the modern transport and trade facilitation, these policy maker’s objective when deciding “tried and true” corridors are expected to where and how to place a transport corridor produce greater surpluses—and perhaps even could be situating a trunk highway across diffusion effects—than a greenfield project country borders in a way that maximizes driven by a competing idea. access to markets for all cities with popula- Optimization of a road network in terms tions of more than X million in the countries of logistics and engineering is another aspect cofinancing the venture. of network econometrics used in more local- Network econometrics based on simple ized contexts (urban, regional, national), market access measures could miss impor- rather than in the context of transnational tant aspects concerning the placement of or transcontinental transport corridors. transport corridors. Two such aspects are Within a given budget, this type of network the importance and relevance of potential econometrics primarily optimizes the road market access or trade connectivity. Page A PPR A ISING PROPOSED TR A NSPORT C ORRIDORS USING SP A TI A L E C ONOMETRI C S   211 et al. (1999) highlight importance and ­ elevance, but without the focus on the wider r r elevance as components of centrality in ­ economic benefits for households along trans- the context of Google’s system of ranking port corridors (Baniya et al. 2017). The prob- pages for searches (PageRank). The crite- ability of these obstacles being removed in the rion “importance” considers whether the future also affects the relevance of the poten- new (transport) connection from city A tial new connections. is to a city B, which is “well connected” As the number of considerations increases, with many other connections that provide their relevance and importance—and thus indirect connections for city A, or to not- centrality—could be summarized by spatial so-well-connected city C, which does not propensity scores that capture the likelihood provide these indirect markets with access that exporters and importers in each location, connections. For instance, Donaldson and such as a district through which the potential Hornbeck (2016) adjust the simple measure transport corridor will pass—will benefit of market access for importance. from the new connectivity and market access. The relevance of the potential direct and But even such rich and spatially disaggregated indirect connections that a transport corri- propensity scores could fall short of the future dor can provide also needs to be considered. reality because of changing global value The potential of connecting to big but chains, urbanization, and city development already highly competitive markets may not strategies. be as useful for city A in a developing coun- In this dynamic context, potential cities to try as connecting to another big city with be connected by transport corridors could be markets that are less competitive, where thought of as agents with varied decision country A’s exporters can more easily pene- rules (heterogenous response functions). So, trate this new market with old or new prod- too, can corporations that shape the growth ucts. For instance, Baum-Snow et al. (2016) of these cities (for instance, if they decide to consider the relevance of market access and reshape their global supply chains as market adjust for market competition in their mar- dynamics change). Suppose that, in the next ket access measure. But relevance could be a 20 years, one city wants to become an inter- much richer concept, including matching national financial center, another city wants demand preferences and the structure of the to attract research-and-development clusters supplying industries, as well as distinguish- and become the next Silicon Valley, and yet ing between the relevance of market access another city wants to become a major manu- for exporters and importers (Duranton et al, facturing hub. These three cities would be 2013). For instance, in calculating market expected to respond to potential connectivity access, a possible new transport connection in different ways. Further, knowing or of a district in country A that exports pork observing the responses of other cities, a city to a city with a large population that can reinforce its strategy or change it. observes Jewish kosher or Muslim halal rit- Considering such varied decision rules of uals (and thus does not eat pork) in neigh- agents could be important for determining boring country B can be viewed as a new which two cities are the priority to connect potential market. However, this market through transport corridors and via which could be irrelevant for the pork-exporting mode of transport (road, rails, waterways, district and should be heavily discounted, if airways).3 not discarded. To exploit the full potential of network Existing transport and trade obstacles econometrics and incorporate historical, (prohibitive border crossings, tariff and non- existing, and forward-looking perspectives, tariff trade barriers) also need to be consid- agent-based modeling (ABM) could be help- ered for relevance. The literature on trade ful. Like GE econometrics and RF regression connectivity considers such aspects of analysis, ABM could provide greater 212   THE WEB OF TR A NSPORT C ORRIDORS IN SOUTH A SI A structure to network econometrics to help unique characteristics of locales and popu- identify how a corridor intervention will lation groups to study the conditional (het- affect different cities, countries, and conti- erogenous) effects of a transport corridor nents directly and indirectly through second- and the effects of its complementary polices round and systemic interactions. But like GE (see chapter 6). Carefully selecting control econometrics, ABM needs many assump- groups and finding both a valid and relevant tions. Its highly nonlinear nature makes it instrument—one that is both exogenous prone to big errors if mis-specified (even to the dependent variable and able to suf- partly). Moreover, simpler market access ficiently explain the treatment variable—are measures could be readily linked to develop- probably the biggest technical struggles for ment outcomes such as income growth, at the RF approach. least in an RF manner (Donaldson and The main issue for policy decision making Hornbeck 2016; Duranton, Morrow, and is that RF regressions do not enable structural Turner 2013). This second step could be identification of the transmission mechanisms prohibitive for more complex network and the analysis of system-wide (GE) effects— econometric approaches such as ABM. including indirect (second-round) effects and Given the state of development in network effects beyond the immediate vicinity of the econometrics, a combination of approaches transport corridor. Moreover, the traditional could be useful to help determine the priorities RF setups are constrained to analyzing the for placing transport corridors. Computations corridor effect on one development outcome based on the notion of c ­ entrality could be the at a time. most reliable currently. More research into Some of the shortfalls of single-equation, developing ABM that views cities as heteroge- RF econometrics could be remedied by read- nous agents could be the most promising ily available extensions of the approach. For approach to help determine corridor place- instance, RF regressions with a common fac- ments in the future. tor (a corridor intervention) for a panel of development outcomes can help analyze the impact of transport infrastructure invest- REDUCED-FORM ECONOMETRICS ment on multiple development outcomes Single-equation, reduced-form (RF) econo- (such as income, equity, and environmental metrics is the most popular method in the quality) together. 5 Likewise, extending literature to evaluate the ex post impact of ­ equation, RF regressions to a system single-​­ transport infrastructure on development of RF equations such as vector auto- regres- outcomes. The most rigorous setups involve sions (VARs) can help analyze GE effects in difference-in-difference estimation based on an RF manner. For instance, the New a carefully selected control group and instru- Economic Geography theory could be used mental variables (IVs) to address possible to motivate the vector of endogenous vari- biases due to endogenous placements. To ables to be included in the (spatial) VAR further sharpen identification, such a setup without imposing the possibly problematic is often complemented by a set of condi- structural restrictions (functional form, tioning variables, matching techniques, or cross-equation coefficient restrictions) placebo treatments (see chapter 4 in this implied by the theory. One example of an volume; and Redding and Turner 2014).4 effort in this direction is the work of With proper IV identification and controls, Revoltella et al. (2016) on transport connec- RF regressions are readily applicable to tivity and economic resilience. (ex post) evaluations. Further, their flexible To date, RF regressions have typically functional form helps in adding terms that focused on effects near transport corridors interact transport corridor connectivity with and not those further away. While using A PPR A ISING PROPOSED TR A NSPORT C ORRIDORS USING SP A TI A L E C ONOMETRI C S   213 macroeconomic data and simulations lacks future policy making and abandoned. locational specifics compared to spatial data, To the contrary, structural GE econometrics it could be a good complement to capture is probably the most promising tool for more system-wide (GE) effects when using RF appraising the WEB of transport corridor econometrics. interventions. More effort is needed to develop its potential. These efforts should focus on covering all important markets, het- STRUCTURAL (GENERAL erogeneous impacts, and system estimations EQUILIBRIUM) ECONOMETRICS that could help improve the data fit of both Using calibrated or estimated parameters, models.7 structural GE and other structural ­ structural general equilibrium (GE) economet- The models also need to be back-tested, rics is the second most popular type of including simulating moments of the model econometrics for appraising the impact of ­ variables and comparing them to the transport connectivity on development out- moments of actual data. comes (Bosker, Deichmann, and Roberts 2015; Burgess and Donaldson 2012). It helps analyze USEFUL DEVELOPMENTS several development outcomes jointly, while IN THE LITERATURE identifying clear transmission channels, linking model parameters to behaviors of economic A number of papers are trying to bridge the agents founded in microeconomic theory, and different econometric methods to advance considering feedback effects and counterfactu- understanding of the effect of large infrastruc- als. These models hold great potential. But the ture projects on socioeconomic development. current applications are rather far from Several studies have linked measures of exploiting it. m arket access (network econometrics) ­ From the current policy-making perspec- with income growth using RF regressions—­ tive, the existing applications of structural GE including estimated equations motivated by a econometrics have some shortfalls. Typically, structural model (Donaldson and Hornbeck the structural application considers only a 2016; Duranton, Morrow, and Turner 2013). limited number of transmission mechanisms This approach, however, suffers from endoge- and clearing markets (such as labor mobility neity and recursive problems between market only, ignoring land and capital market access and growth. Endogenizing market c hannels). Neglecting some transmission ­ access (introducing a first-stage regression) or mechanism can distort the predicted dynam- using instrumental variables could help. But, ics, including because of possibly omitted using this approach, the literature links mar- variables and the resulting estimation bias ket access only to a single final outcome when the structural models are estimated (income) at a time, but not multiple out- without some important markets. Further, the comes. Also, it has not considered GE effects, assumed functional forms and cross-equation including the possibility of losers left behind coefficient restriction (see, for instance appen- in regions with low connectivity to markets dix A in Bosker Deichmann, and Roberts when the majority of the population emi- 2015) could help increase the precision grates to regions that are served by the trans- ­ (efficiency) of the model. However, if they are port corridor. It also has not taken into not in line with the process that generated the account the role of complementary interven- data in reality, these restrictions could result tions (such as access to finance to scale up the in low fit to the data (reality) and low predic- production of local tradable goods and ease tive ability.6 emigration). This is not to imply that structural Some academics and practitioners point to estimations should be viewed as inferior for ­ the methodological tension between RF and 214   THE WEB OF TR A NSPORT C ORRIDORS IN SOUTH A SI A GE econometrics. Several recent papers capital and the frictions in land use could examine this possible tension by comparing lead to incomplete specifications when these two econometric approaches quantita- trying to match the model to the processes tively. For instance, Baum-Snow et al. (2016), generating the actual data. Hence, despite Rothenberg (2013), and Alder, Roberts, and the potential leap in theoretical rigor and Tewari (forthcoming) compare the results of precision (efficiency) of this economic syn- RF and GE econometrics applied to the India thesis, its reliability in practice is open to Golden Quadrilateral (GQ) Highways. While question. the RF and GE results differ in all studies, only Baum-Snow et al. (2016) reject the CONCLUSION results of GE modeling as inferior and possi- bly misleading. There are trade-offs in using a single econo- Even more recently, some papers have metric approach to appraise a large transport attempted to combine network and GE infrastructure project. The main trade-off econometrics (Alder 2017; Fajgelbaum and could be between efficiency and the reliability Schaal 2017). Because the task is computa- of the chosen approach. Econometrics that is tionally very intensive, heuristic network more explicit about the economic structure rules and algorithms from operational and transmission mechanisms behind eco- research are used to implement such nomic corridors could produce much more approaches more easily. Although the two precise and intelligent estimates if correctly applications do not go as far as modeling and completely specified. But if misspecified cross-border agent networks—for instance, in (even partly) or incompletely specified, such trade and global value chains—they are quite econometrics can produce misleading results. promising. For instance, Alder (2017) com- In contrast, RF econometrics struggles with bines a network algorithm for highway place- endogenous placement issues and ignoring ment with GE econometrics to estimate a indirect effects. This trade-off could apply to counterfactual for the Indian policy on the estimating both the optimal placement of a Golden Quadrilateral (GQ) highways system transport corridor and the optimal interven- (see chapter 4 and 6) by following China’s tion design to diffuse its overall net economic policy for the National Express Network (see benefits. chapter 4). He finds that the GQ construction Conditional effects and interactions of cor- generated aggregate income surpluses, but ridor placement with other local factors— they were shared unequally across regions— such as market conditions, policies, and including because the counterfactual net- institutions—are important but rarely work would have been larger than the GQ, addressed. For RF regressions, the struggle to s upporting greater sharing with lagging ­ separate the effects of transport infrastructure regions, and thus stimulate further aggre- from other parallel effects—that is, to achieve gate gains. Fajgelbaum and Schaal (2017) clean identification—makes the inclusion of combine operations research methods and interactive (conditional) effects problematic. a ­n eoclassical trade model with mobile Despite possible biases due to issues with labor to estimate the GDP gains from an reversed causality (endogeneity), these inter- ­ optimal expansion of current networks in active effects need to be applied more (see 25 European countries. They find large gains chapter 6). Because nonlinear structural GE from such an optimal expansion. models are commonly log-linearized using Cross-border applications that account only the first-order Taylor series expansion for corresponding obstacles and imperfec- (not a higher order one), interaction terms tions are still scarce. Also, neglecting the typically disappear from the structural role of and dynamics in the movement of GE econometrics. But they do not need to. A PPR A ISING PROPOSED TR A NSPORT C ORRIDORS USING SP A TI A L E C ONOMETRI C S   215 TABLE S6.1  There are trade-offs between different econometric approaches to appraise proposed corridors Current reliability in Econometric potential in Further research needed to determining: determining: attain the potential: Econometrics approach Placement WEB Placement WEB Placement WEB Network High Lowest High Low High Low Reduced form Low Highest Low Medium Low Medium General equilibrium Low High Medium High Low High Source: Melecky (forthcoming). Note: WEBs = wider economic benefits. Hence, GE econometrics has a great potential could introduce more efficient methods (net- for credibly researching the conditional effects work econometrics based on the integration of transport corridors. of heterogeneous agents and structural GE Table S6.1 illustrates the trade-offs econometrics) to think through possibly between different econometric approaches. It richer dynamics of interactions, as well as the roughly sorts the corridor econometrics dis- direct, indirect, and second-round effects of cussed in this spotlight by their current ability, the corridor intervention. econometric potential, and amount of work Finally, the appraisal econometrics that is needed (“distance’) to attain this potential in currently available does not estimate the link helping policy makers reliably determine cor- between the potential for market access and ridor placement and the potential of a corri- the actual use of this access by firms (in terms dor intervention package for spreading of trade and travel) and people (in terms of economic benefits. For instance, network travel for jobs, education, health, and the econometrics probably has the greatest ability like). The econometrics largely assumes that currently to appraise the optimal placement the link exists and is on average the same in of cross-­border transport corridors. However, every context. But this assumption can be its ability to appraise the design of corridor problematic. For instance, if complementary intervention package is rather small because it factors are needed so that firms and people focuses more on immediate and intermediate can actually use the market access potential outcomes than on final development out- provided by transport connectivity, this link comes—the WEB that households receive. cannot be assumed. It can be broken in many Current structural GE econometrics could be countries, or vary greatly across different eco- efficient in appraising expected WEB, but it is nomic actors and/or geographic units. The less reliable. Structural GE econometrics has appraisal econometrics for transport corri- the greatest potential to appraise WEB effi- dors and related research should seek to shed ciently and reliably, but considerable research more light on the empirical dimensions of this is needed to meet this potential. microeconomic mechanism—including by Since the process of deciding on the place- building on existing anecdotal evidence ment and design of transport corridor inter- (see spotlights 1 and 2). vention is dynamic, sequencing more reliable and more efficient econometric methods could be also useful. Prefeasibility studies and assessments could be deployed, using only NOTES simpler and more reliable econometrics (net- 1. Also see chapter 4 in this book. work econometrics focused on the concept of 2. See https://en.wikipedia.org/wiki/Silk_Road; centrality, combined with RF regressions). https://en.wikipedia.org/wiki/Grand_Trunk​ Later, feasibility studies and project appraisals _Road. 216   THE WEB OF TR A NSPORT C ORRIDORS IN SOUTH A SI A 3. Railway and waterways are more suitable for Working Paper C-89114-CHN-1, International bulk transport, while roads and airways Growth Centre, London. could be more suitable for fragmented supply Berg, C. N., U. Deichmann, Y. Liu, and H. Selod. chains (boutique production, such as Italian 2017. “Transport Policies and Development.” shoes) or high-value perishables (for exam- Journal of Development Studies  53 (4): ple, Maldivian tuna to Japan). 465–80. 4. Other empirical, “reduced-form” methods Bosker, E. M., U. Deichmann, and M. Roberts. include semiparametric and nonparamet- 2015. “Hukou and Highways: The Impact of ric econometrics (Pagan and Ullah 1999; China’s Spatial Development Policies on Graham, McCoy, and Stephens 2014). Urbanization and Regional Inequality.” Policy 5. See chapter 4 for application of this idea in Research Working Paper 7350, World Bank, the context of a metaregression analysis. Washington, DC. 6. For the comparison to the reduce form Bosker, E. M., and H. Garretsen, H., 2010. “Trade model to be fair, one must compare the log-­ Costs in Empirical New Economic Geography.” linearized version of the nonlinear GE mod- Papers in Regional Science 89 (3): 485−511. els to the RF models. But in this case, the Burgess, R., and D. Donaldson. 2012. “Can predicted number of relevant variables by Openness to Trade Reduce Income Volatility? GE models to be included in the economet- Evidence from Colonial India’s Famine Era.” rics is also typically smaller than the number Preliminary draft. ­ typically used by RF models. Also, the cross- Burgess, R., R. Jedwab, E. Miguel, and A. Morjaria. equation restrictions are binding and 2015. “The Value of Democracy: Evidence from increasing the efficiency/robustness trade- Road Building in Kenya.” American Economic offs, especially if the log-linearization Review 105 (6): 1817−51. were to be done at a higher-order Taylor Donaldson, D., and R. Hornbeck. 2016. expansion. “Railroads and American Economic Growth: A 7. See Lall, Wang, and Deichmann (2010) for ‘Market Access’ Approach.” Quarterly Journal the three-stage least-squares method; and of Economics 131 (2): 799–858. Fan and Chan-Kang (2008) for the full-­ Duranton, G., P. M. Morrow, and M. A. Turner. information, maximum-likelihood method. 2013. “Separate Appendices with Supplemental Material for: ‘Roads and Trade: Evidence from the US.’” http://real.wharton.upenn.edu​ REFERENCES / ~ d u r a n t o n / D u r a n t o n _ P a p e r s / C u r r e n t​ Alder, S. 2017. “Chinese Roads in India: The _­Research/Supp_FunTrade.pdf. Effect of Transport Infrastructure on Economic Fajgelbaum, P. D., and E. Schaal. 2017. “Optimal Development.” Working Paper, University of Transport Networks in Spatial Equilibrium.” North Carolina, Chapel Hill. Working Paper 23200, National Bureau of Alder, S., M. Roberts, and M. Tewari. Economic Research, Cambridge, MA. Forthcoming. “The Effect of Transport Fan, S., and C. Chan-Kang. 2008. “Regional Road Infrastructure on India’s Urban and Rural Development, Rural and Urban Poverty: Development.” Unpublished manuscript. Evidence from China.” Transport Policy 15 (5): Forthcoming. 305−14. Bakker, P., C. Koopmans, and P. Nijkamp. 2010. Gastner, M. T., and M. E. Newman. 2006. “Shape “Appraisal of Integrated Transport Policies.” and Efficiency in Spatial Distribution In  Integrated Transport: From Policy to Networks.”  Journal of Statistical Mechanics: Practice , edited by D. Banister, 117−36. Theory and Experiment 2006 (1): P01015. Abingdon, UK: Routledge. Graham, D. J., E. J. McCoy, and D. A. Stephens, Baniya, S., S. Murray, N. Rocha, and M. Ruta. 2014. “Quantifying Causal Effects of Road 2017. “The Trade Effects of the New Silk Network Capacity Expansions on Traffic Volume Road.” Presentation, World Bank, Washington, and Density via a Mixed Model Propensity Score June. Estimator.” Journal of American Statistical Baum-Snow, N. B., J. V. Henderson, M. A. Turner, Association 109 (508): 1440−49. Q. Zhang, and L. Brandt. 2016. “Highways, Laird, J. J., and A. J. Veneables. 2017. “Transport Market Access and Urban Growth in China.” Investment and Economic Performance: A A PPR A ISING PROPOSED TR A NSPORT C ORRIDORS USING SP A TI A L E C ONOMETRI C S   217 Framework for Project Appraisal.” Transport Bringing Order to the Web.” Stanford InfoLab, Policy 56 (2017): 1–11. Stanford, CA. Lall, S. V., H. G. Wang, and U. Deichmann. Redding, S. J., and M. A. Turner. 2014. 2010. “Infrastructure and City Competitiveness “Transportation Costs and the Spatial in India.” Working Paper 22/2010, WIDER Organization of Economic Activity.” Working (World Institute for Development Economics Paper 20235, National Bureau of Economic Research), United Nations University, Tokyo. Research, Cambridge, MA. Melecky, Martin. Forthcoming. “Appraisal Revoltella D., P. Brutscher, A. Tsiotras, and Econometrics for Transport Corridors: C. Weiss. 2016. “Linking Local Business with Optimal Placement, Intervention Design, and Global Growth Opportunities: The Role of Wider Economic Benefits.” Forthcoming as Infrastructure.” Oxford Review of Economic MPRA Paper, University Library of Munich, Policy 32 (3): 410–30. Munich. Rothenberg, A. D. 2013. “Transport Infrastructure Melo, P. C., D. J. Graham, and R. Brage-Ardao. and Firm Location Choice in Equilibrium: 2013. “The Productivity of Transport Evidence from Indonesia’s Highways.” Working Infrastructure Investment: A Meta-analysis of paper, under revision. Empirical Evidence.” Regional Science and Starr, S. F., 2007. The New Silk Roads: Transport Urban Economics 43 (5): 695−706. and Trade in Greater Central Asia. Baltimore: Pagan, A., and A. Ullah. 1999. Nonparametric Johns Hopkins University Press. Econometrics (Themes in Modern Econometrics). Straub, S. 2011. “Infrastructure and Development: New York: Cambridge University Press. A Critical Appraisal of the Macro-Level Page, L., S. Brin, R. Motwani, and T. Winograd. Literature.” Journal of Development Studies 1999. “The PageRank Citation Ranking: 47 (5): 683−708. Spotlight 7 Agriculture Finance and Technical Assistance to Enhance the Wider Economic Benefits of Transport Connectivity for Rural Areas F or several decades, Asian, African, and Latin American countries have improved highways, secondary roads, To seize the opportunities brought by improved transport connectivity, accessible business services—such as different agricul- waterways, and railroads, with some tural finance instruments, insurance, and impressive results. Farmers and others in warehouses—are essential for micro, small, rural areas who are close to these corridors and medium-sized farmers. have used these newly reliable transport Consider the case of Raju, a small farmer links to get inputs and take their products in an Indian village, who produced rice for his to market. family and sold his surplus at a local market. For instance, farmers near new corridors in When a highway was built that connected his one Nepal district sold more of their products village to the town of Chhindwara, his family and increased their incomes by 20 percent income dropped dramatically. New traders because they accessed markets more easily and and producers moved into his area, which lowered their transport costs, the Asian forced small farmers like him to lower their Development Bank found (ADB 2009). prices or crowded them out altogether. Thus Similarly, 42,000 small farmers in Peru Raju had to find new business opportunities. improved their productivity and incomes He learned that producing tomatoes was because of the newly established linkages with profitable, now that transport costs were processors and traders in the corridors, accord- lower and the time to take the crop to market ing to the U.S. Agency for International was reduced. But he needed start-up financing Development (USAID 2008). In some areas, to grow the new crop. While his family and the benefits have helped reduce poverty. For friends lent him some funds, the amount was example, in Bangladesh, after the Jamuna not enough to buy the inputs. He also needed Bridge was built in 1998, the western part of to learn how to market his crop. the country was linked to the main consump- For this, he needed some way to access tion centers in the east. With the arrival of the finance and information. Traditionally, bridge, agricultural techniques improved faster when small farmers such as Raju tried to in the west than in the east (Gautam 2016). expand their crops or start new farm-related 219 220   THE WEB OF TR A NSPORT C ORRIDORS IN SOUTH A SI A activities, they could not get the credit they mandatory lending quotas for certain sectors needed to buy seeds and fertilizer or cover they see as priorities. the costs of harvesting, processing, market- In South Asia, lending quotas are the tools ing, and transporting their products. Or they most often used and account for the largest could not provide banks with the information volume of agriculture finance. For example, needed to prove they were viable borrow- in India, where farmers were mostly financed ers. Or they did not have the necessary col- by money lenders until 1950, the government lateral. Thus they turned to informal, local introduced a quota scheme 40 years ago. moneylenders, who often charged steep inter- Since then, financial institutions have been est rates. With the improved transport links, required to lend to specific sectors, such as some government agencies and nongovern- agriculture. The policy seems to have chan- mental organizations have moved into previ- neled more commercial bank loans to the ously unserved rural areas, along with private agriculture sector (Nathan Associates 2013). financial institutions. India also introduced subsidized crop Organizations such as these remain a bet- insurance linked to credit, which increased ter option for rural dwellers like Raju to get banks’ willingness to make agricultural loans. credit and other financial and technical ser- In addition, the country has a range of finan- vices than online banking, which does not cial institutions—state, cooperative, and com- provide larger credit amounts and long-term mercial banks—that can reach different types credit. Moreover, online lending often of farmers and micro, small, and medium- requires credit seekers to apply in person at sized enterprises (MSMEs). It also uses whole- financial institutions—which are located in sale-retail links that allow commercial banks cities or towns that are so far away that the to purchase loan portfolios from local finan- potential borrowers do not make the trip. cial institutions to meet their priority lending Improved transport connectivity has requirements. However, this is a system brought financial institutions to more remote unique to India. Its many types of public and areas, which can help farmers gain access to private financial institutions and wholesale- credit, increase their productivity, expand their retail arrangements would be hard to dupli- businesses, and add value to their products. In cate elsewhere. India, for example, once the country invested Although lending quotas have expanded in roads and markets, some commercial banks farmers’ access to loans, they have also had moved into rural areas, which in turn provided some negative effects. They have increased farmers with the funds they needed to meet the costs of lending for banks. The quality of their increasing demand for fertilizer, tractors, the banks’ loan portfolios has dropped water pumps, and livestock, and thus improve because they were required to lend to riskier their output, a World Bank study found clients. (Binswanger, Khandker, and Rosenzweig In Bangladesh, the use of another tool— 1989). Governments also encourage financial interest rate caps on agricultural credit institutions to enter areas where commercial (from commercial banks) (Bangladesh Bank finance for agriculture is limited, particularly 2017)—may have squeezed banks’ profit where transport corridors were improved. margins and discouraged them from lending Governments have also sought various to the sector. Caps may lead banks to ways to promote lending to agriculture. These exclude more marginal clients, especially include using credit guarantees, interest rate those who need smaller loans and/or live in subsidies and caps, and warehouse receipts as rural areas. collateral for loans. They have extended sub- In Sri Lanka, in an attempt to motivate sidized crop insurance, offered grants that public and private banks to lend to agricul- match loans from financial institutions, ture, the government introduced partial ­ p rovided venture capital funds, and set credit guarantees to share the credit risk A g r i c u l t u r e F i n a n c e a n d T e c h n i ca l A s s i s t a n c e t o E n h a n c e t h e W i d e r Ec o n o m i c B e n e f i t s    221 with lenders. In Mozambique, the Berira in these areas. This approach has been repli- Catalayatic Fund (BCR), a $20 million ven- cated in other Latin American countries ture capital fund, helps supply goods and ser- (Gálvez Nogales 2014). vices to small farmers located along a Efforts have also been made to create particular transport corridor. Managed by farmer cooperatives, which can help small AgDevCo and financed by the governments farmers achieve a critical mass to obtain of Norway, the Netherlands, and the United loans. The cooperatives can also act as inter- Kingdom, it offers debt and equity loans mediaries between the farmers, public ware- (from $50,000 to $500,000) and technical houses, agribusinesses, and banks. In India, assistance to the investor companies. By for example, Frito Lay, a subsidiary of 2013, BCF had invested more than $3 million PepsiCo, helped establish cooperatives to con- in 12 projects that had helped over 10,000 tract with the company and provided them farmers (Gálvez Nogales 2014). with inputs and technical assistance. More In Tanzania, the government introduced a than 14,000 potato farmers participated matching grant scheme, called the Catalytic (Dutta, Dutta, and Sengupta 2016). Trust Fund, to support agribusiness compa- Elsewhere, cooperatives are involved in logis- nies that create jobs and income for small tics, such as transporting products to ware- farmers. In Sri Lanka, the government also houses and silos, and arranging for the credit has extended matching grants (linked to the the producers need to buy inputs, scale up farmers’ own funds and loans) to encourage their operations, and improve postharvest groups of small farmers to invest in agricul- marketing. tural technologies and improve postharvest While some of the approaches work well practices. These improvements leverage the in some countries, they are not a one-size- bank loans and increase the likelihood of fits-all solution. Each country’s situation repayment. must be examined to determine which of Farmers’ ability to get loans can also the measures is needed or feasible. Where increase when producers, processors, and possible, the private sector should be traders can use their warehouse inventories included, even if its involvement extends (called warehouse receipts) as collateral to the time needed to achieve results. secure loans. In India, the government also Moreover, where the government offers subsidizes storage infrastructure—specifi- financial support or subsidies, these efforts cally, the construction of rural warehouses— should be temporary and be confined to the which encourages farmers to use the start-up phase of creating sustainable finan- warehouse receipts scheme to obtain credit. cial services. Otherwise, they become long- When governments or their agencies have term budget outlays. Finally, the risks of improved the transport links in Southeast overexposure in a portfolio from expanding Asia, Africa, and Latin America and the targeted loans to a certain sector or geo- Caribbean through economic and transport graphic area along a transport corridor corridor projects, they have often offered must be carefully assessed. technical assistance on various aspects of pro- duction—such as marketing, information technology, management practices, market REFERENCES research, postharvest handling, agro-­ ADB (Asian Development Bank). 2009. “Asian processing, and connecting smaller agri-­ Development Bank’s Contribution to Inclusive business to larger firms—which promoted Development through Assistance for Rural agricultural finance and investment. For Roads.” Evaluation Study. https://www.oecd​ example, a USAID project to reduce poverty .org/derec/adb/47166759.pdf. in Peru focused on developing infrastructure Bangladesh Bank. 2017. “ACD Circular No. 02: and businesses, which promoted investment Agricultural & Rural Credit Policy & Program 222   THE WEB OF TR A NSPORT C ORRIDORS IN SOUTH A SI A for the FY 2017-18.” https://www.bb.org.bd​ Agribusiness and Food Industries Series No. 4, /­mediaroom/circulars/circulars.php# Food and Agriculture Organization of the Binswanger, H. P., S. R. Khandker, and United Nations, Rome. M. R. Rosenzweig. 1989. “How Infrastructure Gautam, M., and R. Faruqee. 2016. Dynamics of and Financial Institutions Affect Agricultural Rural Growth in Bangladesh: Sustaining Poverty Output and Investment in India.” World Bank, Reduction. Washington, DC: World Bank. Washington, DC. Nathan Associates. 2013. “Re-Prioritizing Priority Dutta, A., A. Dutta, and S. Sengupta. 2016. Sector Lending in India: Impact of Priority “A Case Study of Pepsico Contract Farming for Sector Lending on India’s Commercial Banks.” Potatoes.” IOSR Journal of Business and Economic Impact Analysis. Nathan Associates, Management. http://www.iosrjournals.org/iosr​ Arlington, VA. -jbm/papers/ICSE%20Conference/14.75-85. USAID (U.S. Agency for International pdf. Development). 2008. “The Peru Poverty Gálvez Nogales, E. 2014. “Making Economic Reduction and Alleviation (PRA) Program.” Corridors Work for the Agricultural Sector.” http://pdf.usaid.gov/pdf_docs/Pdacn111.pdf. 6 An Illustrative Appraisal of Complementary Interventions to Enhance the Wider Economic Benefits of Transport Corridors P olicy makers are increasingly interested in appraising large transport infrastruc- ture proposals using rigorous methods This chapter develops an illustrative framework to appraise large transport infra- structure projects in South Asia and beyond. beyond a simple cost-benefit analysis. They As an ­ i llustration, it focuses on India’s want these appraisals to be comprehensive Golden Quadrilateral (GQ) and North- and cover the “wider economic benefits” South-East-West (NSEW) highway systems (WEB) of these investments—that is, benefits (map 6.1).1 First, it shows how to estimate that go beyond savings of travel time and the impacts these two systems have on WEB vehicle operating costs. They would like to (the overall net spatial impact on develop- ensure that such transport infrastructure ment outcomes, such as economic welfare, projects benefit both the big firms involved in equity, social inclusion, and environmen- foreign trade and the small enterprises that quality) by applying the “difference-in-​ tal ­ could take advantage of the increased con- ­ difference” (DiD) method to district-level nectivity and market access to pursue new data from 1994 to 2011. Second, it dis- economic opportunities. They are also inter- cusses how to examine the dependence of ested in mitigating the possible negative these impacts on initial conditions in input impact of corridors, whether on environmen- ­ (capital, labor, land) markets, output (prod- tal quality, income inequality, or gender uct) markets, and governance by using inclusion. They would like to identify any interaction terms in the DiD regressions. potential losers the corridor investments Third, it shows how to use the ­ e stimated might produce and offset those losses to the DiD model to simulate an appraisal of two extent possible. proposed highway systems in the region: the 223 224  The WEB o f T r a n s p o r t C o r r i d o r s i n S o u t h A s i a MAP 6.1  Two highways in India are used to illustrate the appraisal and female employment in regular wage framework jobs); and environmental quality (thickness of aerosol particles, carbon dioxide emis- sions, and nitrogen dioxide emissions). In addition, the estimation looks at the structural transformation in employment (the share of farm versus non-farm employ- ment), an intermediate outcome of interest (see chapter 4). The change in connectivity is measured by the distance from the GQ (NSEW) highway and is identified follow- ing the method developed by Ghani, Goswami, and Kerr (2016) to estimate the impact of the GQ corridor.2 Capital market conditions are measured using indicators of household and firm-level access to banking. Labor market conditions are captured with rates of literacy and secondary schooling. Land markets conditions are measured using indicators of land use patterns. Product markets conditions are captured with indicators of industrial composition and private ownership. Governance is mea- sured using a state-level index that looks at infrastructure; social services; fiscal perfor- mance; justice, law, and order; and the qual- ity of the legislature. Source: National Highways Authority of India, http://www.nhai.org/gqmain_english.htm. Disclaimer: This publication follows the World Bank’s practice in references to member designations, borders, and maps. The boundaries, colors, denominations and other information shown in any map in this work do not imply any judgment on the part of The World Bank, ADB, JICA or DFID concerning the legal status of any territory or the endorsement or acceptance of A MULTI-STEP METHODOLOGY such boundaries. WAS USED TO ESTIMATE THE WIDER ECONOMIC BENEFITS OF HIGHWAYS IN INDIA The study uses a differences-in-differences The efficiency of land, labor, financial, and product methodology to estimate the impact of the markets and the effectiveness of local policies and highways on the district-level outcomes of institutions can help determine how widely the interest (figure 6.1). This method compares economic benefits are shared. the change in the outcome of interest after the highway was constructed in districts located close to the new highways (the China-Pakistan Economic Corridor (CPEC) “treatment districts”) to those located far in Pakistan, and the Kolkata-Dhaka corri- away from them (the “control districts”). dor in Bangladesh (Melecky, Sharma, and The computed differences capture the Subhash forthcoming). change in the outcome before and after the The WEB include measures of welfare highway was built. It controls for the con- d istrict-level GDP and household con- (­ founding effect of unobserved factors that sumption); inequality (district-level pov- do not vary over time (that is, factors that erty); social inclusion (total employment are time-invariant). For instance, districts A n I l l u s t r a t i v e A p p r a i s a l o f C o m p l e m e n t a r y I n t e r v e n t i o n s   225 FIGURE 6.1  The difference-in-difference method is used to estimate the impact of the two existing Indian highways Impact of the highways = B – A A B Impact 2001 2011 2001 2011 Control districts Treatment districts Source: Corridor Study Team. Note: The difference-in-difference method assumes that the change (without intervention) of control and treatment districts is the same between 2001 and 2011 (the parallel path assumption). Therefore, the average change without intervention (A) is obtained by subtracting the results of 2001 from the results of 2011 for control districts. The change in treatment districts between 2001 and 2011 is B. The impact of the treatment (in this case, building the GQ or NSEW highway system) is obtained by subtracting A from B. GQ = Golden Quadrilateral; NSEW = North-South-East-West. that are near the highways could have been wider economic benefits only when certain relatively productive even before the high- market conditions are present that allow ways were built. The set of second differ- people to exploit opportunities opened up ences compares the change across treatment by the improved access to markets. That is, and control districts. That is, it controls for highway and other large transport invest- the confounding effect of unobserved fac- ments may need to be complemented with tors common to control and treatment dis- other policy and institutional reforms to tricts that do vary over time (factors that support the transport corridor intervention are time-varying). For instance, it is possi- in specific locales. For example, highways ble that reforms undertaken around the could have a bigger impact on firms in areas same time as the highway was built caused with better access to finance or to skills, or a general rise in productivity across India. in areas with better functioning land Thus, the identification assumption behind markets. this approach is that unobserved, time-­ The study undertakes this assessment by varying factors had the same impact across adding interactive labor, land, and capital both control and treatment districts (see market conditions to the regression frame- box 6.1).3 work (figure 6.2). The impact of comple- The analysis further assesses the possibil- mentary factors is the difference of the ity that investments in highways generate impact of highway in districts with low 226  The WEB o f T r a n s p o r t C o r r i d o r s i n S o u t h A s i a l ­evels of complementary factors and the difference of the differences therefore impact of highway in districts with high makes the method a ­triple-difference (DDD) levels of complementary factors. The estimation (see box 6.2). impact of the highway was obtained using the DiD method (see figure 6.1). Taking the THE WIDER ECONOMIC BENEFITS OF HIGHWAYS ARE ESTIMATED USING DISTRICT-LEVEL DATA While public investments can increase net economic The analysis uses a district-level panel data set benefits in the aggregate, sharing those benefits to estimate the impact of the highways. widely across locations and specific population Districts are the primary administrative unit groups is more  difficult, and depends on the of India below the state level. In 2010−11, the effectiveness of complementary policies and last year in the data set, there were about 640 institutions. districts in India. BOX 6.1  Estimating the average impacts of the highways Formally, the underlying regression specification important to control for the presence of the other can be described as follows: one. Second, the two networks could have had dif- ferent impacts. In other words, there were two sets Yi ,t = β × Highwayi × PosttHighway (6.1.1) + φi + ϕ t + ε i , t . of treatment districts: those close to the GQ, and those close to the NSEW. This regression is estimated on a district-level Following Ghani, Goswami, and Kerr (2016), panel data set. Here, Yi,t is an outcome of interest in districts were assigned to distance bands based on district i and year t. The dummy variable PosttHighway proximity of the district centroid (the geometric is equal to 1 in years after the highway has been center of each district) to the GQ. The bands are: completed, and 0 in the years before that. The more than 100 kilometers (km) from the nearest dummy variable Highwayi is equal to 1 in districts GQ point; 40−100 km from the GQ; 0−40 km close to the new highways (the treatment districts) from the GQ; and nodal districts. a ­I ndicators were and 0 otherwise. The variable f i is a set of district then interacted for each district distance band fixed effects that controls for time-invariant district- with a variable ­i ndicating the years after the com- level factors, and j t is a set of year dummies that pletion of the GC. This process was then repeated controls for unobserved time-varying ­ factors com- for the NSEW. Thus, the specification estimated is mon to all districts. The impact of the highways as follows: is estimated by b  , the c ­ oefficient on the treatment term (the interaction Highwayi × Post tHighway ), mea-  suring how the change in the outcome after the Yi ,t = β GQ × GQi × Post tGQ + β NSEW × NSEWi (6.1.2) highway was built differed across control and treat- × PosttNSEW + φi + ϕ t + ε i ,t . ment districts. This basic specification is adjusted to account Here, GQ i (or NSEW i) is a vector of dummies for the fact that the estimation is simultaneously indicating the distance band from the GQ (or estimating the impacts of two highway networks, NSEW) to the centroid of district i , while PosttGQ the G Q and the NSEW. T here are t wo fac- tors to consider in this regard. First, in estimat- (or PosttNSEW ) is a dummy equal to one in the ing the impact of either highway network, it is years after the GQ (or NSEW) is completed. The (box continued next page) A n I l l u s t r a t i v e A p p r a i s a l o f C o m p l e m e n t a r y I n t e r v e n t i o n s   227 BOX 6.1  Estimating the average impacts of the highways (continued) omitted distance band dummy corresponds to dis- estimation is controlling for NSEWi × Post tNSEW , tricts more than 100 km from the highway (GQ b GQ ,0 − 4 0 in effect measures how the post-GQ or NSEW). The variable f i is a set of district fixed effects, and j t is a set of year dummies (or state- change in the outcome differed between districts year dummies). 0−40 km from the GQ highway (the GQ treatment Depending on the outcome, the panel dataset group) and districts more than 100 km from both c overs either four of the data spells (1994−95, ­ highways (the control group). Similarly, the impact 2000−01, 2004−05, and 2010−11) or just two spells of NSEW is measured by the b NSEW, corresponding (2000−01 and 2010−11). Since the GQ network was to the 0−40 distance band from the NSEW, denoted by  b NSEW,0−40 . The main results tables thus report largely complete by 2005, the indicator PosttGQ is set these two b s. equal to 1 in the years 2004−05 and 2010−11, and 0 The preferred specification replaces the year fixed otherwise. Work on NSEW started after 2005. Thus effects with more flexible state-year fixed effects. PosttNSEW is set equal to 1 only in 2010−11.b This controls for unobserved state-level differences The impact of the GQ is measured by the b GQ, in growth patterns, which is important, given the corresponding to the 0−40 distance band from the documented divergence in economic growth across GQ, which is denoted by b GQ,0−40 hereafter. Since the Indian states. a. As in Ghani, Goswami, and Kerr (2016), the estimation assigns nodal districts to a separate category, and does not treat them as treated districts. Nodal districts correspond to major metropolitan areas and their peripheries, and thus are distinct from the average Indian district. b. Certain segments for the NSEW were not complete by 2011. The analysis takes this into account when assigning districts to distance bands around the NSEW. The main source for the district-level data The analysis supplemented the spatial database is the South Asia Spatial Database, which is variables with additional district-level mea- being developed by the World Bank’s Office sures of welfare and labor market conditions of the Chief Economist for South Asia.4 derived for this study from various rounds of This database is intended to bring together India’s National Sample Survey (NSS) of data on India (and other South Asian econo- Employment and Unemployment. The NSS mies) from a range of sources—such as official labor force data are available for 1994−95, censuses, administrative records, surveys, sat- 1999−2000, 2004−05, and 2010−11. Thus, ellite imagery, and official maps—into a single the panel data set covers four years if the out- source of spatial platform data. It covers two come variable is derived from the NSS data, points in time (2000−01 and 2010−11) and and two years otherwise. has four administrative levels (ranging in spa- Table 6A.1 in online annex 6A lists the tial detail from state or province, district, and main outcome variables used in the study town or village to hundreds of thousands of and their primary data sources. Measures gridded cells, or “tiles”). are broken down by district. The outcome This study selected those district-level vari- ables from this database that measure specific WEB outcomes of interest and initial market The use of detailed spatial data and econometric conditions that could have interacted with analysis can add more rigor and depth to the the highways to generate varied impacts. appraisal of transport corridors. 228  The WEB o f T r a n s p o r t C o r r i d o r s i n S o u t h A s i a FIGURE 6.2  The triple difference-in-difference method is used to estimate how the impact of the highway depends on complementary factors Impact of Complementary Factors = Impact B – Impact A Impact B Impact A 2001 2011 2001 2011 2001 2011 2001 2011 Control Treatment Control Treatment Districts with low levels of complementary factors Districts with high levels of complementary factors Source: Corridor Study Team. BOX 6.2  Estimating conditional impacts: Do highway impacts depend on market conditions? A further estimation also tests the hypotheses that Here, Zi is a vector of initial conditions of inter- the impact of the highways depended on condi- est in district i. The effect of initial conditions on tions in factor and product markets. This estima- the impact of the highways is estimated by the tion exploits variation in initial market conditions d s, the coefficients on the triple interaction term across districts, and adopts a triple-­ d ifference between Highwayi × Post tHighway and the vector of (DDD) approach by interacting the treatment initial conditions Zi. (Highwayi × Post tHighway ) with variables capturing For illustration, suppose that the Zi in question is a variable measuring the efficiency of land markets, initial conditions in districts. The exact specification with higher values indicating more ­ efficiency. The for the conditional impacts is as follows: corresponding d  GQ (or d  NSEW) coefficient measures how the impact of the GQ (or the NSEW) depends Yi ,t = β GQ × GQi × Post tGQ markets. A positive estimate on the efficiency of land ­ of this d GQ would imply that the impact of the GQ + β NSEW × NSEWi × Post tNSEW on the outcomes of interest was more positive in dis- + δ GQ × GQi × PosttGQ × Zi tricts with more efficient land markets. (6.2.1) Many potential Z i variables are available for + δ NSEW × NSEWi × Post tNSEW × Zi each factor and product market. The estimation + γ 1Post tGQ × Zi + γ 2 Post tNSEW × Zi of interaction effects is also c ­ omplicated by the + φi + ϕ t + ε i , t . fact that potential Zi variables could be correlated. (box continued next page) A n I l l u s t r a t i v e A p p r a i s a l o f C o m p l e m e n t a r y I n t e r v e n t i o n s   229 BOX 6.2  Estimating conditional impacts: Do highway impacts depend on market conditions? (continued) This makes it important to check if an estimated full set of interaction variables and progressively interaction term is robust to controlling for inter- drops interaction terms with low p -values, until actions with other potential Zi variables. Hence, a only a small set of interaction terms is left. As a practical issue arises of choosing a sparse regres- procedure, the set of Z i variables in result of this ­ sion specification while not ignoring the impor- the final specification varies across outcome vari- tance of controlling for correlated Z i variables. ables. The procedure ensures that the results being The estimation addresses this issue by employing shown are robust when controlling for other inter- a simple iterative algorithm that starts with the action terms. ­ ariables correspond to four WEB catego- v product. The main labor market variables ries: conomic welfare, inequality, social are measures of human capital as of 2001: inclusion, and environmental quality. The the literacy rate and the percentage of measures of economic welfare are the GDP those with a secondary school or higher per capita and mean household per capita educational qualification. The land market expenditures for each district, while inequal- variables measure the extent of land that is ity is measured using rural and urban pov- suitable/available for agriculture, as well erty headcount rates. The measures of social as the capacity of the district for mineral inclusion are the percentage of the working- production. Thus, they are measures of the age population that is working and the per- nature of land endowments in districts. centage of regular-wage (“good”) jobs to The capital market variables measure total employment, both disaggregated by household access to bank services (that is, gender (map 6.2). The second measure is the to formal bank accounts), and firms’ preferred measure of social inclusion. The access to bank loans, as of 2001. The measures of environmental quality include product market variables include a mea- measures of particulate matter (smog) in the sure of product diversification, as well as a air, as well as measures of air pollution (car- measure of the share of private firms in bon dioxide and nitrogen dioxide levels). A industrial establishments; both are measure of deforestation was discarded intended to proxy for competition in the because data were not sufficiently varied product market. A third measure is the between districts near the highways (the share of agroprocessing in manufacturing. “treated” districts) and those further away This variable proxies for the initial level of (the “control” districts). In addition to these opportunity for factory work available to final outcome variables, the analysis also low-skilled workers, who constitute the estimates impacts on the breakdown of total majority of the workforce, particularly in employment by farm and nonfarm jobs, which is an intermediate outcome of interest. Annex table 6A.2 lists the main interac- The design of corridor investments and their tion (initial market condition) variables complementary policies should be based on a better and their sources. They are categorized by understanding of the underling mechanisms that type of market: labor, land, capital, and generate development outcomes. 230  The WEB o f T r a n s p o r t C o r r i d o r s i n S o u t h A s i a MAP 6.2  The data set includes a range of measures of wider economic benefits Sources: GDP per capita: India: Directorate of Economics and Statistics, Planning Commission, Government of India 2011; Bangladesh: Bangladesh Bureau of Statistics 2011; Pakistan: World Bank 2017a. The GDP per capita estimation is based on calculations based on the World Development Indicators and the South Asia Spatial Database (Li et al. 2015). Female regular wage employment: India: Census of India–Primary Census Abstract (PHC–PCA), Office of the Registrar General and Census Commissioner, India; Bangladesh: Labor Force Survey 2005; Pakistan: Social and Living Standard Measurement Survey 2012−2013. Disclaimer: This publication follows the World Bank’s practice in references to member designations, borders, and maps. The boundaries, colors, denominations and other information shown in any map in this work do not imply any judgment on the part of The World Bank, ADB, JICA or DFID concerning the legal status of any territory or the endorsement or acceptance of such boundaries. rural areas. The hypothesis is that a large the GPI is used as an initial interaction condi- agroprocessing sector also signals better tion in the estimations.5 This is shown in supply chain infrastructure (such as ware- annex table 6A.3. houses, cold chains, and other logistical A major challenge in putting together the facilities) in rural areas. final district database was the matching and The state-level measure of governance is harmonization of districts across time. Many drawn from a recent study (Mundle, new districts have been created over time, Chowdhury, and Sikdar 2016) that scores the leading to changes in district boundaries and 19 largest Indian states along five dimensions: names. Currently, there is a total of 707 dis- infrastructure; social services; fiscal perfor- tricts in India, compared to 640 in the 2011 mance; justice, law, and order; and the quality Census of India and 593 recorded in the of the legislature. The scores are based largely 2001 Census of India (for a list of sources, on “output” measures, such as proportion of see annex table 6A.4). The analysis addressed trials completed in fewer than three years this issue by mapping newly created districts (one of the indicators for the justice, law, and back to their unique parent district in 1999.6 order dimension) and development expendi- For instance, if district X in 1999 was split tures as a share of total expenditures (one of into districts Y and Z by 2010, the 2010 the indicators for the fiscal performance data for Y and Z were combined to recreate dimension). The overall “Governance the parent district X in 2010. In addition to Performance Index” (GPI) combines the score aggregating new districts up to their 1999 on these five dimensions. The 2001 value of parent district, districts in the remote areas A n I l l u s t r a t i v e A p p r a i s a l o f C o m p l e m e n t a r y I n t e r v e n t i o n s   231 of Jammu and Kashmir and northeastern identify the “treatment” and “control” dis- states of India have been dropped. This is tricts in the differences-in-differences esti- standard practice in district-level studies on mation. Specifically, the 0−40 km distance India, including previous studies of the band from the GQ (or the NSEW) identifies impact of the GQ highway. After these steps, the GQ (or NSEW) treatment districts, the data set consists of around 425 districts while the control districts are those more per year. than 100 km away from either highway. Finally, the geocoded data on the location Figure 6.3 shows that the sample contains of GQ and NSEW were merged into the dis- about 70 GQ treatment districts and about trict database.7 This information was used to 40 NSEW treatment districts. There are calculate the distances of the district centroid about 200 districts in the common control from the nearest points on the GQ and NSEW group. networks. The districts were then grouped One concern about the analysis is that into four distance bands from each highway: districts close to the GQ could also be close nodal district; 0−40 km from the highway; to the NSEW, which would make it hard to 40−100 km from the highway; and more distinguish the impacts of each highway than 100 km from the highway (map 6.3).8 system. However, annex table 6A.5 shows Thus, there are eight distance bands in total. that there is little overlap across districts in Annex table 6A.5 lists the joint distribution terms of nearness to these highways. Most of the sample according to these distance of the districts that are close to the GQ bands. highway (0−40 km from the GQ network) As explained in the methodology section, are more than 100 km from the NSEW net- the study uses these distance bands to work, and vice versa. This increases our MAP 6.3  Distance of districts in India from the GQ and NSEW highways is measured from the district’s centroid Source: World Bank’s Spatial Database for South Asia (SARCE), https://spatialdatabase-southasia.herokuapp.com/. Note: The red line GQ/NSEW highway systems, the blue dots signify district centroids, and the green lines signify the linear distance from highways to the district centroids. 232  The WEB of Transport Corridors in South Asia confidence in being able to distinguish km from the GQ network) and control dis- between the impacts of the GQ and NSEW tricts (those more than 100 km away). highways. Likewise, the impact of NSEW is given by b NSEW,0−40, the coefficient on the NSEW treat- ment term NSEWi0− 40 × PosttNSEW . ESTIMATION RESULTS FOR The first result of note is the impact of THE TWO EXISTING HIGHWAYS the highways on the movement of labor IN INDIA from farm to nonfarm work: that is, on Average impacts of the highways structural transformation (figure 6.4). This happened during a period in which the Annex table 6A.6, panels a through d, shows share of non-farm employment in total the results of estimating equation 6.1.2 by employment was generally increasing in ordinary least squares (OLS). This is the base- India; specifically, the share of nonfarm line differences-in-differences specification employment rose by 2.5 percentage points measuring the average impact of the high- in the baseline (the control districts), fol- ways. The panels report estimates from lowing the construction of the GQ. regressions that include a full set of state-year According to the point estimate of b GQ,0−40, fixed effects. This is the preferred specification which is statistically significant at the 1 because it controls for unobserved state-level percent level, the GQ highway increased the variables that could have been changing sig- share of nonfarm employment by an addi- nificantly during the study period. The impact tional 1.6 percentage points. Thus, the of the GQ is given by b GQ,0−40, the coefficient impact was large. The GQ also increased on the GQ treatment term GQi0− 40 × PosttGQ , the share of nonfarm employment among which measures how the change in outcomes females by 2.4 percentage points. The esti- varied across treatment districts (those 0−40 mate of b  NSEW, 0−40 is also positive and FIGURE 6.3  One in four districts was “treated” a. Number of districts, by distance to GQ b. Number of districts, by distance to NSEW 450 500 409 459 400 450 350 400 350 300 300 250 250 200 200 150 150 100 85 84 100 73 49 50 18 50 15 0 0 Nodal 0–40 km 40–100 km >100 km Nodal 0–40 km 40–100 km >100 km districts districts Source: Corridor Study Team. Note: GQ = Golden Quadrilateral; NSEW = North-South-East-West. A n I l l u s t r a t i v e A p p r ai s a l o f C o m p l e m e n t a r y I n t e r v e n t i o n s   233 statistically significant. Overall, the NSEW FIGURE 6.4  The economic restructuring induced by the highways highway appears to have raised the share of shifted employment from the farm to the nonfarm sector nonfarm employment by about 2.5 percent- age points in the general population as well 10 as among females. The positive impacts of the highway con- struction on the farm to nonfarm transition 5 suggests that limited access to markets is a reason for India’s slow structural transfor- Percentage change mation. The lack of market access has reduced employment opportunities beyond 0 the farm, while keeping workers trapped in low-­ p roductivity agriculture by shielding the sector from competition. This hypothe- –5 sis, however, does not imply that highways should lead to a movement away from farm work in every district. In districts with a –10 strong inherent comparative advantage in Total farm Female farm Total nonfarm Female nonfarm agriculture, it could theoretically lead to a employed employed employed employed movement into the agricultural sector. The Control districts Effect of the GQ results only suggest that the former direc- tion of movement is the case for the average Source: Corridor Study Team. treatment district. Note: GQ = Golden Quadrilateral. Moving to final outcomes of interest, the regressions suggest that the GQ highway had a statistically significant positive impact (Ghani, Goswami, and Kerr 2016). The for- on district output per capita (figure 6.5). mal manufacturing sector still constitutes Looking at the results for the GDP per cap- only a small share of output, and an even ita (in logs) variable, the point estimate of smaller share of employment, in most b GQ,0−40 implies that the highway increased Indian districts. It is possible that increased GDP per capita growth over 2001−11 by market access from the highways mainly 4 percentage points. This is of significant benefited formal manufacturing and the rel- magnitude, given the baseline increase of atively small ­ n umber of skilled workers 27 percent. However, the corresponding employed in that sector. This was enough to estimated impact of the NSEW corridor is cause a significant increase in total (and per statistically insignificant. capita) district GDP. But because the growth The regressions do not find statistically formal ­ in ­ manufacturing started from such a significant impacts of either highway on small base, it did not lead to a detectable other measures of welfare, such as mean increase in the total number of regular-wage household per capita consumption jobs or in the incomes of low-skilled indi- ­ expenditures. Nor do the regressions detect viduals. The results thus suggest that the significant impacts on inequality (poverty economic benefits of the highways were not headcount rates) and the main measure of widely shared. labor market inclusion, the share of regular- Interestingly, the results also suggest that wage (“good”) jobs.9 the structural change caused by the high- These results are somewhat unexpected, ways was not in itself enough to lead to a given the positive impact on GDP per cap- significant increase in “good” (that is, regu- ita, and the prior finding that the GQ had a lar-wage) jobs. Nor was it enough to reduce significant impact on output, productivity, poverty significantly. The opening up of and wages in formal manufacturing market access drove workers off the farm, 234  The WEB o f T r a n s p o r t C o r r i d o r s i n S o u t h A s i a FIGURE 6.5  While the average income grew, domestic migration in India is within a dis- the other benefits were not widely shared, and trict (Kone et al. 2016). But it bears further environmental quality deteriorated investigation, possibly with district-level migration data. 35 Finally, the regressions examine the high- ways’ impacts on measures of environmen- tal quality. While they do not detect a significant impact on carbon dioxide and 25 nitrogen dioxide levels in the air, it appears that the GQ highway led to an increase in air pollution (smog) related to particulate Percentage change matter. Specifically, the GQ is estimated to 15 have increased particulate pollution (as measured by “aerosol optical thickness”) by 0.02 point relative to a baseline increase of 0.06 point. This signals a significant trade- 5 off between economic benefits and pollution. 0 How the impacts could have depended –5 on product and factor market conditions: Log of GDP per Aerosol optical estimates of interaction effects capita thickness This section presents estimates of the DDD Effect of the GQ Control districts specification spelled out in equation 6.2.1. The main results, which look at interac- Source: Corridor Study Team. tions with product and factor ­ market con- Note: GDP = gross domestic product; GQ = Golden Quadrilateral. ditions, are shown in annex table 6A.7, panels a through d. The panels are arranged by outcome categories: welfare outcomes in but not necessarily into better-paying jobs. It panel a, distributional outcomes (inequal- could be that because most farm workers ity) in panel b, labor market outcomes are uneducated, they could not shift into (inclusion and structural change) in panel c, well-paying jobs in other sectors. Other and environmental quality outcomes in labor market frictions, such as barriers to panel d. rural-urban migration or restrictive labor As noted, a simple iterative procedure was regulation of formal jobs, could also be at used to reduce the number of extraneous play here. interaction variables. Thus, the set of interac- Another potential explanation for tion variables in the final specification varies these results is the possibility of migration across outcomes. These regressions include between the districts in response to the state-year dummies to control for state-level highways. That is, the highways did increase shocks. the demand for nonfarm labor significantly, Panel e shows the interaction results when but immigration into the treated districts a measure of state-level governance perfor- was sizable enough to wipe out any equilib- mance was included among the interaction rium wage increase. This explanation is at terms. These regressions use year dummies odds with the well-established facts that instead of state-year dummies because the India’s rates of domestic migration are low governance variable does not vary within by international standards and that most states. A n I l l u s t r a t i v e A p p r ai s a l o f C o m p l e m e n t a r y I n t e r v e n t i o n s   235 The broad hypothesis is that gaining the areas where the access to finance was better full benefits of market access could have initially. There is some support for this depended on certain factor endowments, hypothesis in recent research on the impact such as skills (which are immobile in the of connective infrastructure in China short to medium terms), and on the effi- (Banerjee, Duflo, and Qian 2012). However, ciency of product and factor markets. If this the results of this study do not support hypothesis is correct, then low average lev- this hypothesis. Firms’ access to finance els of factor endowments or market effi- enhanced the impacts of the NSEW ciency could explain why the average district ­(figure 6.6). When looking at the share of did not experience widespread benefits of nonfarm jobs among females, the regres- the highway construction. Identifying such sions find that the estimated coefficient on complementary factors can point to how the the interaction between the NSEW treat- construction of highways could be com- ment and a measure of firms’ access to for- bined with complementary public invest- mal financing is positive and significant ments or policy reforms to maximize their (annex table 6A.7, panel c). The estimated benefits. interaction effect of firms’ access to formal financing and the NSEW on the share of female farm employment is consistently Capital markets: households’ access ­n egative (annex table 6A.7, panel c). 11 to bank accounts and firms’ access to This suggests that firms’ access to formal loans loans could have enhanced the impact of the The results for capital markets are intrigu- highways on structural transformation. ing, and hint at a complex relationship between market access and access to formal Labor markets: skills banking (figure 6.6). Consider the impact of the GQ on mean household consumption The results indicate a complementarity expenditures per capita. The estimated coef- between highways and skills. Regarding ficient on the interaction between the GQ the impact on the share of nonfarm employ- treatment and a measure of households’ ment, the coefficient on the triple interaction access to bank accounts is negative and sig- between a measure of secondary schooling nificant (annex table 6A.7, panel a).10 This (the initial percentage of 15+-year-olds implies that the impact of the GQ on house- with secondary school education) and hold expenditures was less positive in dis- GQi × PosttGQ is positive and statistically sig- tricts where households had better access to nificant at the 1 percent level (annex formal savings accounts. Given that the table 6A.7, panel c; fi­ gure 6.7). This implies main use of these formal accounts is to save, that the GQ had a significantly more positive one potential explanation is that even impact on nonfarm employment in districts though the highways increased household with more secondary education. The point earnings, a larger fraction of that additional estimate of the d GQ implies that relative to income was saved in locations where house- a district at the 10th percentile of the second- holds had better access to formal channels of ary schooling measure, the impact of the GQ saving. This could reflect an unmet demand on the share of nonfarm employment was among households for channels to diversify 7 ­percentage points higher in a district at assets to build resilience to shocks (“save for the 90th percentile. Similarly, the interaction a rainy day”). of the ­ s econdary school variable with Another potential explanation is that NSEWi × PosttNSEW is positive and statistically highways help bring capital to unbanked significant.12 The same pattern is also observed locations; this would explain why the high- with regard to the impact on the share of non- ways have less impact on some outcomes in farm employment among females. 236  The WEB o f T r a n s p o r t C o r r i d o r s i n S o u t h A s i a FIGURE 6.6  Access to finance for firms and households had different impacts on spreading WEB Firm’s access to finance Household access to banking did not enhanced the impact of the NSEW enhance the impact of the GQ interaction term: capital market interaction term: capital market private nonfarm enterprise finance household access to banking services 10 0.16 8.79 8 0.11 6 4 0.06 2 Percentage change 0.01 0 Unit change –2 –0.04 –4 –6 –0.09 –8 –0.11 –0.14 –10 –9.32 –0.15 –12 –0.19 Nonfarm Farm Female non Female farm GDP per Household Rural Regular Aerosol employed employed farm employed capita consumption poverty wage optical employed rate employed thickness GQ NSEW Source: Corridor Study Team. Note: The figure shows only those results that are significant at 1 percent. Household consumption and GDP per capita are in log terms. GDP = gross domestic product; GQ = Golden Quadrilateral; NSEW = North-South-East-West; WEB = wider economic benefits. Education also seems to have enhanced the Results for GDP per capita as an out- impact of the NSEW on regular-wage jobs— come also hint at a complementarity although in this case, it is basic literacy that between connectivity and skills. While the seems to have mattered. The estimated coef- results for the GQ are not statistically sig- ficient on the interaction between the literacy nificant, in the case of the NSEW, the coef- rate and NSEWi × PosttNSEW is positive ficient on the triple interaction between and significant when the outcome being basic education (the i ­nitial literacy rate examined is the share of regular-wage jobs among 7+-year-olds) and NSEWi × PosttNSEW in total employment (annex table 6A.7, is positive and statistically significant at the panel c). It implies that in moving from the 5 percent level (annex table 6A.7, panel b). 10th ­percentile of the literacy measure to the The implied complementarity is large in 90th percentile, the impact of the NSEW on magnitude; moving from the 10th percen- the share of regular-wage jobs would increase tile of the literacy measure to its 90th per- by 12 percentage points. centile, the impact of NSEW on per capita A n I l l u s t r a t i v e A p p r a i s a l o f C o m p l e m e n t a r y I n t e r v e n t i o n s   237 FIGURE 6.7  Education boosted the impact of the two highways Education enhanced the impact of the GQ Basic education enhanced the impact of the NSEW interaction term: labor market interaction term: labor market secondary education completed (% of population) literacy rate (% of population 7 years of age or older) 15 12 10 8.4 7.69 5 Percentage points 0.17 0 –5 –7.52 –8.49 –10 Nonfarm Farm Female Female farm GDP per Household Rural Regular Aerosol employed employed nonfarm employed capita consumption poverty wage optical employed rate employed thickness GQ NSEW Source: Corridor Study Team. Note: The figure shows only those results that are significant at 1 percent. Household consumption and GDP per capita are in log terms. GDP = gross domestic product; GQ = Golden Quadrilateral; NSEW = North-South-East-West. GDP would increase by additional As shown in annex table 6A.7, panel d, the 17 percentage points. This increase is large, negative impact of highways (the GQ) on par- even compared to the 27 percent increase in ticulate matter air pollution (aerosol optical baseline growth in GDP per capita during radius) is mitigated by having a better-educated 2001−11. population (as measured by the share of the Consistent with these patterns, the esti- population that has completed secondary mated coefficient on the interaction between schooling). This could be because more edu- the literacy rate and NSEWi × PosttNSEW is cated individuals buy higher-quality, environ- positive, although marginally insignificant, mentally cleaner vehicles as their incomes rise. for two other economic welfare outcomes: Another possible explanation for this interac- mean household consumption expenditures tion is that the way in which economic activity per capita, and the reduction in urban poverty changes after the highway is built is more envi- headcount (annex table 6A.7). ronmentally friendly in more educated 238  The WEB o f T r a n s p o r t C o r r i d o r s i n S o u t h A s i a districts. Notably, districts with higher levels of As for the impact of the GQ on the share secondary schooling experience a greater of nonfarm jobs among females, and on the impact of the GQ on the shift from farm to share of regular-wage jobs among both males nonfarm jobs. This reduced pollution from the and females, the coefficients on the interac- burning of straw on farms, which is a major tion of the GQ treatment with the cropland particulate pollution in India. contributor to ­ measure are negative and statistically signifi- cant (annex table 6A.7, panel c). For exam- ple, the estimates of the interaction term Land markets imply that moving from the 10th to the 90th The results suggest that the impact of the high- percentiles of cropland share would reduce ways also depended significantly on the share the impact of the GQ on nonfarm employ- of cropland in a district’s total land area ment among females by 5 percentage points, (figure 6.8). This interaction variable can be and the share of regular wage jobs by 0.1 seen as a proxy for a district’s comparative percentage point. This is consistent with the advantage in agriculture. A higher share of hypothesis that the extent to which improved cropland also signals a bigger constraint on the market access shifted people out of farm jobs availability of land for industrial purposes. was negatively related to comparative FIGURE 6.8  Land market conditions had mixed impacts on spreading WEB Having more cropland reduced the impact of the In contrast with the GQ, the availability of cropland GQ on nonfarm jobs enhanced the impact of the NSEW interaction term: land market interaction term: land market cropland (% of area) cropland (% of area) 8 0.19 0.20 6 0.15 4 0.10 0.07 2 0.05 Percentage change Unit change 0 0 –0.15 –2 –0.05 –4 –0.10 –6 –5.33 –0.15 –7 –8 –0.20 Nonfarm Farm Female Female farm GDP per Household Rural Regular Aerosol employed employed nonfarm employed capita consumption poverty wage optical employed rate employed thickness GQ NSEW Source: Corridor Study Team. Note: The figure shows only those results that are significant at 1 percent. Household consumption and GDP per capita are in log terms. GDP = gross domestic product; GQ = Golden Quadrilateral; NSEW = North-South-East-West; WEB = wider economic benefits. A n I l l u s t r a t i v e A p p r a i s a l o f C o m p l e m e n t a r y I n t e r v e n t i o n s   239 advantage in agriculture (or depended posi- of agroprocessing in local industry and the tively on the availability of industrial land). impact of the GQ in reducing rural poverty The impacts on household consumption per (annex table 6A.7, panel b). There could be capita are also consistent with this hypothe- two reasons for this complementarity. First, sis. Namely, the coefficient on the interaction as discussed, for the structural change of the GQ treatment with the cropland mea- wrought by the GQ to have translated into sure is negative and statistically significant widespread benefits, the availability of suit- (annex table 6A.7, panel a). The GQ districts able jobs for the vast reserves of low-skilled gained less from market access if they had workers leaving the farm could have been less land available for nonfarm usage. critical. Agroprocessing could be an impor- The results also suggest that the nature of tant source of suitable jobs for rural workers. the interaction with land conditions differed Second, the size of the agro­ processing sector across districts near the GQ and the NSEW. could be acting a proxy measure for the qual- First, more cropland is not significantly ity of “soft infrastructure” (such as ware- associated with a lower positive effect of the houses and cold chains) in rural areas. This is NSEW on nonfarm jobs, which suggests that indicative of a complementarity between unlike the case of the GQ, a larger share of “hard” infrastructure like the highway and cropland did not impede the structural soft connective infrastructure. change caused by the NSEW (table 6A.7, panel c). Second, looking at GDP per capita, Governance the coefficient on the interaction of the NSEW treatment with the cropland measure In addition to specific conditions in factor is positive and statistically significant: hav- and product markets, the impact of connectiv- ing more cropland seems to have enhanced ity could also depend on crosscutting the impact of the NSEW highway on per institutional factors such as the quality of ­ capita output (annex table 6A.7, panel a). A ­ g overnance. Better-governed areas, for potential explanation for these results is that instance, might have been better at enacting some districts in the areas connected by the complementary policies in locations affected NSEW began to specialize in high-value, by the new highways. The regressions test this tradable farm products and agroprocessing. idea by including a measure of state-level gov- Among these districts, having more crop- ernance performance in 2001 as an additional land increased the gains from market access, interaction term. As noted, state-year fixed while not necessarily impeding the creation effects cannot be included. Hence, the results of non-farm jobs in agro-related industries. with governance interactions should be inter- preted with the caveat that they may not be robust to controlling for unobserved state-level Product markets ­ resented shocks. Accordingly, these results are p The regressions also tested the hypothesis that separately in annex table 6A.7, panel e, and the gains from highways depend on efficiency are not included in the main results shown in and competition in product markets. This annex table 6A.7, panels a through d. ­ roduct part of the analysis relies on proxies for p The statistical significance levels and signs market competition at the district level, such of the product and labor market interaction as the share of the private sector in formal effects shown in annex table 6A.7, panels a manufacturing. No statistically significant through d, survive the replacement of state- and consistent interaction was observed year fixed effects with year fixed effects, and between these measures and the impact of the the addition of a governance interaction highways; perhaps it reflects the crude quality term. Thus, the estimated interaction effects of the available measures. of product and labor market conditions There is, however, a positive and statisti- remain unchanged in this alternative specifi- cally significant interaction between the share cation. Second, the newly included 240  The WEB of Transport Corridors in South Asia governance and highways interaction terms output markets in the places located along are not statistically significant for most out- the corridor. If the GQ were to be repli- comes, with the important exception of cated in a different location, such as the labor market inclusion measures. These proposed paths of the CPEC or the Kolkata- results are shown in annex table 6A.7, Dhaka corridor, it would have a different panel e. 13 Regarding the share of regular impact because the initial market condi- wage jobs in total employment—the pre- tions in those locations were different (see ferred measure of the creation of “good” box 6.3). jobs—the interaction of governance with the Of the three proposed arteries for the GQ is positive and significant. The same CPEC, the easternmost one was chosen effect is observed in the case of regular-wage because it resembles the GQ the most, passing jobs share among females. These findings through well-populated areas and several cities suggest that better governance enabled a and towns, and connecting major metropoli- wider sharing of the impacts of highways in tan nodes and ports (international gateways). creating more numerous, more inclusive, The simulation exercise focused on assess- and better-quality jobs. ing the spatial variation in the expected impacts of the proposed corridors along their prospective paths, and not on assessing their THE SIMULATED IMPACTS OF average impact. The simulated corridors are TWO PROPOSED CORRIDORS: only roughly comparable to the GQ, and THE CHINA-PAKISTAN ECONOMIC their average impact could differ from that of CORRIDOR AND THE KOLKATA- GQ for several reasons. However, if the mech- DHAKA CORRIDOR anism of impact is the same, the way in which it depends on initial conditions is expected to The simulation method be similar. This section simulates and discusses the The simulation for the CPEC and the expected impacts of two corridors that have Kolkata-Dhaka corridor used district-level not yet been built: the China-Pakistan data from Pakistan and Bangladesh on the Economic Corridor (CPEC) in Pakistan, and distance to the proposed corridor and rele- the Kolkata-Dhaka corridor in Bangladesh. vant market conditions. The exercise mea- These simulations are based on the estimated sured the distance of district centroids to the regressions describing the impacts of the GQ nearest point on the proposed highway, and in India, and are intended to illustrate how identified districts within 40 km of the pro- micro data and econometric modeling can be posed highway. Following the estimated used to better assess the varied impacts of model for India, these districts were consid- proposed transport investments. ered the “treatment” districts; that is, the dis- The simulation uses estimates of the GQ tricts that would be affected by the proposed instead of the NSEW, for two reasons. First, corridor. The impacts were simulated for only we are more confident about the estimates of these districts.14 Also, following the Indian GQ because they are based on data for a lon- estimation, major metropolitan districts that ger span of time after the highway was built. lie on the proposed corridor path, like Dhaka Second, there are similarities in the market in Bangladesh and Karachi in Pakistan, were access effects of the GQ and the simulated treated as “nodal” districts and excluded corridors; like the GQ, both the CPEC and from the simulation. Kolkata-Dhaka corridor will connect places Data on market conditions were estimated along their paths to major external markets, from household- or firm-level surveys, or ports, and/or metropolitan areas. sourced from satellite data. Because the high- The estimated regression model for India ways have not yet been built, the exercise assumes that the impact of the GQ depends used the most recent values of the variables to on a set of initial conditions in input and simulate “initial” market conditions. As much A n I l l u s t r a t i v e A p p r ai s a l o f C o m p l e m e n t a r y I n t e r v e n t i o n s   241 BOX 6.3  Simulating the effects of a transport corridor in another country Based on the estimated regression model for GQ, the impact of GQ on Y depends on initial market the expected impact of a proposed corridor in coun- conditions Z in a district with upgraded connectiv- try c is simulated using data on initial market con- ity. The vector Zd,c−Z India measures how initial con- ditions of districts d that lie along the path of that ditions in location d in country c differ from their corridor: average values in locations in India along the GQ. This formula expresses the basic idea behind the ∆Yd ,c = β Y + δ Y (Zd ,c − ZIndia ), (6.3.1) simulation: the impact of a GQ-like corridor on where DYd,c is the expected impact of the c ­ orridor location d in country c will differ from the average on outcome Y in location d of country c. b  Y is the impact of the GQ because initial market conditions estimate of the average impact of GQ on outcome Y in location d are different from their averages along in India. The vector d Y consists of estimates of how the path of the GQ. as available data allow, the simulations used results for the Kolkata-Dhaka corridor. The sources and definitions of indicators equiva- focus is on the spatial variation in impacts, lent to those used in India. For example, the and not the mean impact. This discussion is share of cropland in land area was based on therefore limited to those outcomes for MODIS satellite data, the same source as in which there were statistically significant India. In some cases, it was not possible to interaction effects with one or more initial avoid minor differences in the source type or market conditions in the case of the GQ. variable definition. For example, variables on Thus, even though output per capita is an educational attainment were based on census important outcome variable, it is not dis- data for India and Bangladesh, and household cussed here because the simulations could survey data for Pakistan. Annex tables 6B.1, not estimate how the impact depends on 6B.2, and 6B.3 in annex 6B present the sum- market conditions with sufficient precision. mary statistics and data sources for Pakistan, We can predict that the proposed corridors while annex tables 6C.1, 6C.2, and 6C.3 in will have a positive impact on output per annex 6C present corresponding information capita, but not how it will vary spatially.15 for Bangladesh. Map 6.4 depicts the simulated impact on Because comparable measures of subna- per capita consumption expenditures of tional governance for Pakistan and Bangladesh households (in logs) through “heat maps.” are unavailable, the exercise excluded the The district-level predicted impact was nor- effects of governance from the simulations. malized by its mean value across all the treat- Hence, the simulations are based on the base- ment districts, thus showing its relative value line estimations for India that did not include across districts. the governance variable as an interaction term The average impact of the GQ on house- (the estimation results presented in annex hold expenditures was measured imprecisely table 6A.7, panels a through d). (that is, it is statistically insignificant). However, its dependence on land and out- put market conditions was statistically The simulation results significant—namely, the dependence on the ­ Annex table 6B.4 in annex 6B presents the share of cropland in total area, and the per- simulated impacts of the CPEC on WEB, centage of firms that are privately owned. The while annex table 6C.4 in annex 6C presents positive impact on household expenditures 242  The WEB o f T r a n s p o r t C o r r i d o r s i n S o u t h A s i a MAP 6.4  The simulated relative impact on mean per capita household consumption expenditures shows disparities along the Kolkata-Dhaka Transport Corridor (Bangladesh segment) Source: Corridor Study Team based on Melecky, Sharma, and Subhash (forthcoming). Note: Units refer to average per capita household consumption expenditure, in logs. The red line indicates the highway. could be expected to be smaller in districts than in the share of private ownership. with a larger share of cropland. As discussed, Moreover, the predicted impact on household one explanation is that districts with more expenditures is more sensitive to the share of land devoted to farming could find it harder cropland, based on the estimated model for to reallocate land to industrial uses and the India. The simulation therefore suggests that, service sector, limiting their ability to benefit for both proposed corridors, the spatial varia- from improved market access. In contrast to tion in impacts on household expenditures will cropland, greater private ownership of firms be mostly influenced by variation in the share is expected to enhance the positive impact of of cropland—that is, by the variation in land corridors on household expenditures—most market constraints. For example, districts likely because of greater market contestability close to Karachi (for CPEC) and Chittagong in those districts. (for the Kolkata-Dhaka corridor) could expe- The spatial variation in the share of crop- rience the biggest impacts on household land in Pakistan and Bangladesh is greater expenditures because they have more land A n I l l u s t r a t i v e A p p r a i s a l o f C o m p l e m e n t a r y I n t e r v e n t i o n s   243 available for industrial uses. Because the The hypothesis is that in such locations, a Kolkata-Dhaka corridor passes through areas relatively large share of low-skilled individu- with more diverse shares of cropland, the vari- als was already employed in the agroprocess- ation in the simulated household expenditures ing industry, a sector expected to benefit of will be greater for this corridor. increased market access. Such workers could Map 6.5 presents simulated impacts on be better predisposed to share benefits of bet- the reduction in the headcount measure of ter market access, moving up the value chain poverty. A more positive effect on this indica- to better jobs without having to acquire new tor implies a wider sharing of economic ben- skills or move to new jobs. Even poor work- efits. Although the average impact on poverty ers in agriculture could possibly benefit from reduction was not estimated precisely for the the growth of the local agro-industry. GQ, it was significantly bigger in areas with Along the proposed path for the Kolkata- a larger share of agroprocessing firms in Dhaka corridor, high variation in the share of the total number of manufacturing firms. agroprocessing in the manufacturing sector MAP 6.5  The simulated relative impact on the reduction in poverty (headcount measure) varied along the Kolkata-Dhaka Corridor (Bangladesh segment) Source: Corridor Study Team. Note: The red line indicates the highway. 244  The WEB o f T r a n s p o r t C o r r i d o r s i n S o u t h A s i a could result in significant variation in the cor- The simulated impacts on the share of ridor’s impacts on poverty reduction across regular-wage jobs in female employment—a the connected districts. The range of expected measure of women’s inclusion in the labor impacts is nearly 15 percentage points. For market—also vary significantly. The varia- the CPEC, district-level estimates of the share tion depends on the share of cropland in the of agroprocessing were not available. Instead, total land area (map 6.6). Land constraints the analysis relied on province-level data. could restrict women’s ability to benefit pro- Hence, the simulated impacts show no varia- portionately from improved market access. tion within provinces. Even the variation The effect of land constraints on the varia- across provinces is relatively small; much of tion in the expected impact on women’s the eastern arm of the CPEC passes through inclusion is even more pronounced for provinces with similar levels of agro-industry, Bangladesh. The range in simulated impacts such as Sindh and Punjab. is on the order of 7 ­ percentage points for the MAP 6.6  The simulated relative impact on the share of regular wage jobs in women’s employment shows major disparities along the Kolkata-Dhaka Transport Corridor (Bangladesh segment) Source: Corridor Study Team. Note: The red line indicates the highway. A n I l l u s t r a t i v e A p p r a i s a l o f C o m p l e m e n t a r y I n t e r v e n t i o n s   245 CPEC and 9.5 percentage points for the Pakistan and Bangladesh (map 6.7). The spatial Kolkata-Dhaka corridor. Several districts variation is driven by differing levels of higher near the central leg of the Kolkata-Dhaka education; districts with higher rates of second- corridor could miss out on the benefits of ary school completion could see a smaller rise increased female labor market inclusion in air pollution. Two explanations are possible. because of severe land constraints. Along the First, while the advent of highways should CPEC, the land market constraint could increase road traffic in every treatment district, dampen benefits in districts nearer to the those with highly educated populations might northern leg of the corridor. tend to switch to cleaner vehicles, even if they On average, the simulated impact of corri- are relatively more expensive. Second, as the dors on air pollution (as measured by the aero- GQ estimates implied, areas with more second- sol optical thickness indicator) is negative, but ary schooling experience a larger structural spatial variation is substantial across districts in transformation toward nonfarm jobs. In those MAP 6.7  The simulated relative impact on air pollution (aerosol optical thickness) should alert policy makers about needed action in some districts along the Kolkata-Dhaka Transport Corridor (Bangladesh segment) Source: Corridor Study Team. Note: The red line indicates the highway. 246  The WEB of Transport Corridors in South Asia districts experiencing greater structural change, farmland could constrain the potential to farms might burn less straw—a major contrib- gain from a highway. utor to air pollution in South Asia. Access to financial services. The results suggest that access to credit by nonfarm enterprises amplified the shift of women to CONCLUSION better (regular-wage) jobs induced by the This chapter has illustrated how the use of NSEW, while reducing farm employment of detailed spatial data and econometric analysis women in districts connected to the NSEW by can bring more rigor and depth to nearly an equal amount. Thus, firms’ access the appraisal of transport corridors. The to credit seems to enable and amplify the ­ estimations found that the improved connec- wider economic benefits of highways. In con- tivity generated by the GQ and NSEW high- trast, the impact of the GQ on household way networks contributed significantly to expenditures was less positive in districts structural transformation of the districts. The where households already had better access to shift from farm to nonfarm employment formal savings accounts.17 was statistically significant in both cases. Labor market conditions. Labor market Each network had a significant positive conditions, as measured by the literacy impact on district-level per capita GDP. level, amplified the positive impact of the However, this impact was not widely shared. NSEW on the per capita GDP of connected The estimation did not find similar positive districts. Higher levels of literacy and sec- effects for household consumption, poverty, ondary education also seem to have lessened and employment, including those of females. the negative environmental impacts of the Moreover, the analysis found evidence of sig- highways. Similarly, a higher level of sec- nificant trade-offs in the impacts on WEB, ondary education was associated with a estimating that while GDP increased signifi- more positive impact of the GQ on regular- cantly, air quality decreased significantly. wage employment, particularly for women, and a more negative impact on farm employment. Overall, the estimated interac- Initial conditions tions suggest that the low average level of With respect to initial conditions, the analy- schooling in India may have prevented the sis found that initial local market conditions benefits of connectivity from being shared are important in determining the outcomes more widely. of highway upgrades and the ways and Land market conditions. A greater area of extent to which WEB are shared across geo- cropland constrained local firms and house- graphic units. The central mechanism holds from pursuing the higher-value nonag- through which highways affect development ricultural opportunities opened up by greater is increased trade across connected locations. market access stemming from the construc- This expansion was accompanied by a real- tion of the GQ. Nonfarm employment for location of resources to more productive women was also constrained. In contrast, firms and to sectors of comparative advan- in NSEW districts where more cropland was tage. Conditions in labor, land, capital, and available, GDP increased more. One explana- product markets could matter because tion is that some areas located near the NSEW they affect this process of reallocation. 16 have rich potential in tradable farm products For example, in districts with a comparative and agroprocessing. Their gains from market advantage in manufacturing, limited avail- access were thus boosted by the availability of ability of land for nonfarm uses could lessen cropland. the gains from a highway by ­ constraining the Product market conditions . The initial reallocation of land toward manufacturing. industrial composition or other product In a district with a comparative advantage in market conditions could be important fac- agricultural products, limited availability of tors in ensuring that transport corridors A n I l l u s t r a t i v e A p p r a i s a l o f C o m p l e m e n t a r y I n t e r v e n t i o n s   247 help reduce poverty. The estimations found The results are largely consistent with the that a higher share of agroprocessing in man- idea that transport corridors lead to wider ufacturing significantly improved the impacts economic benefits by increasing market access of the GQ on reducing poverty. A strong and trade across connected locations. This base in agro-processing seems to have helped process requires a reallocation of land, labor, unskilled rural workers moving off the farm and capital to sectors of comparative advan- find a job. Districts with more agroprocessing tage. The GQ, for example, led to a significant are also likely to have a better quality of “soft reallocation of workers from the farm to the infrastructure” (such as warehouses and cold nonfarm sector. The benefits of transport chains) in rural areas. The result indicates a ­ corridors—and how widely they are shared— complementarity between hard infrastruc- therefore depend on how quickly and effi- ture, like the highway, and soft connective ciently land, labor, and capital can move to infrastructure. new sectors of promising activity. This could Governance. Regular-wage employment be the key reason why the impacts of high- increased more in states with better gover- ways like the GQ depend on initial market nance. One explanation is that better-­ governed conditions. areas were better at foreseeing and implement- To further illustrate this complementarity, ing policies that enhanced the highways’ the analysis used the GQ impact estimates to impact on job creation. These results should be simulate the expected impacts of two pro- interpreted with caution. The governance mea- posed corridors: the CPEC in Pakistan and sure is at the state level and could be reflecting the Kolkata-Dhaka corridor in Bangladesh. other unobserved differences across states. The corridor impacts that were found to be dependent on initial conditions in the appraisal of the GQ were used to appraise the Complementary policies and expected variation of WEB across newly institutions ­ connected districts in the two countries. This The illustrative case study of highways in simulation assumed that the impact mechan- India thus suggests that public investments in ics for the proposed projects match those of transport corridors can yield net economic the GQ. The simulation aimed to illustrate benefits in the aggregate. However, widely how project appraisers could screen for sig- sharing these benefits across districts and pop- nificant variations from district to district ulations can be more difficult, and depends on (spatial variations) in the expected impacts of the effectiveness of complementary policies a corridor project, and identify the comple- and institutions. India’s GQ highway, for mentary reforms that are most needed. example, increased net output and manufac- The simulations suggest that the proposed turing sector activity, but a clear significant corridors could have significantly varied effect on reducing poverty and boosting access impacts on household expenditures, poverty, to better (regular-wage) jobs cannot be con- the inclusion of women in the labor market, firmed. This wider sharing of economic bene- and air pollution. The variation would be fits could depend on conditions in land, labor, driven by spatial differences in levels of higher financial, and product markets. For the GQ, the estimations suggest that the highway did reduce poverty significantly in areas (districts) Policy makers should focus on identifying the with a sufficiently large agroprocessing base underlying constraints in the local context of (while in the average district near the highway, prospective corridors—whether these are policies poverty did not drop). Wage employment among females increased in districts where and institutions constraining the use of key factors of land was available for nonfarm uses. For the production (land, labor, capital, or product markets) NSEW, the impact on wage employment was or other factors, such as low skills in the local labor enhanced by higher levels of education. market or poor governance. 248  The WEB o f T r a n s p o r t C o r r i d o r s i n S o u t h A s i a education, land market constraints, and socioeconomic benefits from being shared industrial composition along the paths of the more widely. But this does not imply that proposed corridors. For example, in districts agriculture-­intensive locations or areas with located in the central leg of the proposed low education and skill levels are not suitable Kolkata-Dhaka corridor and in the northern for the placement of corridors. Instead, the les- end of the CPEC, the impact on women’s son to draw is that policy makers should focus employment in regular-wage jobs should be on identifying the underlying constraints in the lower because of constraints on land use, local context of prospective corridors— which limit women’s shift from farm to non- whether these are policies and institutions farm jobs. behind the constrained use of land or the The simulation exercise highlights an ele- f actors behind low skills in the local ­ ment largely missing in the media and policy labor market. forum discussions on these proposed high- ways: the fact that their wider economic impacts could vary greatly across locations NOTES because of differences in existing market conditions. Significant reductions in pov- 1. The Golden Quadrilateral (GQ) highway, erty, for example, are likely to occur only in together with the NSEW corridor and places where a sizable number of low-skilled port connectivity highways, connect many of workers are already working in the agropro- India’s major manufacturing, commercial, and cessing sector—an industry expected to ben- cultural centers. Both roadways are four-lane highways. efit from improved market access. 2. Datta (2012) uses a similar methodology, The results also illustrated the potential although not with district-level data. environmental trade-offs from transport cor- 3. Researchers develop “identification assump- ridors. The estimates suggest that the GQ tions” or “identification techniques” to increased particulate air pollution. While not clearly isolate the effect of the intervention surprising, given the increase in traffic, this (treatment)—in this case, the highway that potential negative impact is largely absent was built—from the confounding effects of from policy decisions. other factors on the outcome of interest. The main message for the design of corri- 4. As of writing this draft, the database was still dors is that the efficiency of markets and the being tested and was not available to the effectiveness of local policies and institutions public. The authors were able to download data upon special request. The intention is to could complement investments in transport make the data set public. corridors and steer their overall spatial impact 5. The analysis could not use alternative, better- on development outcomes—that is, the net known state-level measures of governance wider economic benefits. Investing in human dimensions, such as the World Bank’s state- capital and better governance is also impor- level Ease of Doing Business indicator, tant, as are policies that could mitigate the because to our knowledge none of the other potential negative effect on environmental measures are available for the initial period of quality. The design of corridor investments our study. and their complementary policies should be 6. The reference year of 1999 was chosen based on a better understanding of the under- because a wave of new states were created lying mechanisms through which the corridor between 1999 and 2001, leading to a sharp rise in the number of districts and significant in question could affect development out- renaming of districts over the course of a few comes. Thus, more analysis is needed than a years. Mapping districts to their 1999 defini- simple extrapolation of the patterns observed tions is therefore the more conservative in the case study. For example, for the GQ, the approach to harmonizing districts. availability of land for nonfarm uses and 7. The authors are grateful to Ejaz Ghani and the level of secondary education seem to have coauthors (Ghani, Goswami, and Kerr 2016) been the main constraints preventing for sharing these data, which they compiled A n I l l u s t r a t i v e A p p r a i s a l o f C o m p l e m e n t a r y I n t e r v e n t i o n s   249 using official highway maps. The merger of the statistically significant. This is puzzling, given GQ data was relatively straightforward because that, in general, the coefficients on the inter- the network was largely complete by 2005. action with educational measures go in the This study used the final GQ network map to positive direction. However, the regression calculate distances of district centroids to the also controls for the interaction between sec- highway. The NSEW distance calculation was ondary schooling and GQi × PosttGQ , so this more complicated because some sections of the result could be an artifact of the correlation network, particularly in its eastern leg, had not between these two educational measures. been completed by 2010. The analysis ignored This puzzling pattern did not emerge for the those incomplete sections of the NSEW when NSEW. calculate district distances to the NSEW. 13. Panel e corresponds to the same set of 8. For comparability, these distance bands were outcomes as panel c, replacing state-year ­ chosen to correspond to those used in the dummies with year dummies and adding gov- study by Ghani, Goswami, and Kerr (2016) ernance interactions. Notably, the signifi- of GQ. The distance cutoffs are not the same cance and signs of the other interaction terms because they calculate distance to the high- are similar in these two panels. way from the district’s nearest edge, while this 14. The treatment group also includes all dis- study does so from the district’s centroid. tricts through which the proposed corridor 9. Night lights intensity, a commonly used would pass, regardless of how close it would proxy for economic activity, also was not come to the district centroid. Overall, the affected significantly. exercise identified 41 districts for the CPEC 10. Moreover, the estimated coefficient on the in Pakistan and 33 treatment districts for the interaction between the GQ treatment and Kolkata-Dhaka corridor in Bangladesh. the banking access measure is negative and 15. Recall that the impact estimate for the GQ is significant when the outcome being examined +4 percent. is the reduction in the poverty headcount. 16. For a recent literature review of how connec- Since the headcount measure is derived from tive infrastructure affects development and the consumption expenditures measure, these the mechanisms behind those impacts, see interaction results are consistent with each Bougna et al. (forthcoming), a companion other. background paper for this report. 11. Regarding the impact on the share of regular- 17. This conditional impact could still be desir- wage jobs in female employment, the inter- able if the access to formal payment and sav- action between firms’ access to finance and ings methods encourages people to save and the NSEW is negative. This is in apparent accumulate assets to protect against shocks contrast to the positive interaction impact of (“save for a rainy day”). However, the lack firms’ access to finance and NSEW on of household asset data prevents this study ­ nonfarm jobs. However, it reflects a positive from testing this hypothesis. interaction impact of firms’ access to finance and the NSEW on total female employment, which is the denominator for estimating the share of regular-wage jobs in female employ- REFERENCES ment. Overall, the interaction between firms’ Banerjee, Abhijit, Esther Duflo, and Nancy Qian. access to finance and the NSEW increased 2012. “On the Road: Transportation and female nonfarm employment more than it Infrastructure Growth in China.” Working decreased female farm employment. There Paper 17897, National Bureau of Economic was no significant interaction impact on Research, Cambridge, MA. the absolute number of regular-wage jobs Bougna, Theophile, Martin Melecky, Mark among females, but since the interaction Roberts, and Yan Xu. Forthcoming. “Transport impact on total female employment was Corridors and Their Wider Economic Benefits: A positive, the interaction impact on the Critical Review of the Literature.” Policy ratio of regular-wage jobs to total jobs Research Paper, World Bank, Washington, DC. was negative. Datta, Saugato. 2012. “The Impact of Improved 12. The interaction term on GQi × PosttGQ and a Highways on Indian Firms.” Journal of measure of literacy, however, is negative and Development Economics 99 (1): 46−57. 250  The WEB o f T r a n s p o r t C o r r i d o r s i n S o u t h A s i a Ghani, Ejaz, Arti Grover Goswami, and William R. Melecky, Martin, Siddharth Sharma, and Hari Kerr. 2016. “Highway to Success: The Impact Subhash. Forthcoming. “Wider Economic of the Golden Quadrilateral Project for the Benefits of Investments in Transport Corridors Location and Performance of Indian and the Role of Complementary Policies.” Policy Manufacturing.”  Economic Journal (Royal Research Paper, World Bank, Washington, DC. Economic Society) 126 (591): 317−57. Mundle, Sudipto, Samik Chowdhury, and Satadru Kone, Z. L., M. Y. Liu, A. Mattoo., C. Ozden, and Sikdar, 2016. “Governance Performance of S. Sharma. Forthcoming. “Internal Borders and Indian States: 2001−02 and 2011−12.” Migration in India. Journal of Economic Working Paper 164, National Institute of Public Geography. Finance and Policy, New Delhi. Spotlight 8 Cross-Border Infrastructure Projects: Challenges and Lessons Learned from the Unrealized Sava Waterways Rehabilitation Program in Southeast Europe T his spotlight reviews the preparation of a regional project, the Sava Waterways Rehabilitation Program, in Southeast Europe. bulk inputs and outputs to and from the heavy industry located along the way. The conflict from 1991 to 1995 was particularly hard on The project was not implemented, and the the Sava River and nearby areas because these World Bank investment was eventually with- were often the front line. Maritime infrastruc- drawn because of challenges that could not be ture and ports were destroyed, the riverbanks resolved. Still, the experience is instructive as it and surrounding areas were heavily mined, illustrates the problems faced by regional bod- and the towns and industry were devastated. ies, national governments, subnational enti- After the Dayton Peace Accord (DPA) was ties, and international organizations regarding signed in Bosnia and Herzegovina, the Sava cross-border infrastructure projects. River crossed four sovereign countries, tra- versing Slovenia and Croatia, then forming the border between Croatia and Bosnia and PRELIMINARY ATTEMPTS Herzegovina, becoming the border between TO RESTORE THE REGIONAL Bosnia and Herzegovina and Serbia, and WATERWAY finally crossing Serbian territory to join the Before 1990, the Sava River had an important Danube in Belgrade (map S8.1). Within role in regional transport in the former Bosnia and Herzegovina, the river crossed the Yugoslavia, from where it meets the Danube two entities created by the Dayton Peace River at Belgrade, extending through what Accord—the Federation of Bosnia and became four separate countries (Serbia, Bosnia Herzegovina (FBH) and the Republika Srpska and Herzegovina, Croatia, and Slovenia). (RS)—as well as the Brc ˇko Administrative Navigation was possible for much of the year District (BAD).1 Given the issues the new from Belgrade all the way to Rugvica, near countries faced after the war, the Sava River Zagreb, in what is now Croatia, along 683 was neglected, with little maintenance or river kilometers. Some 5.2 million tons of reconstruction in the ports or industry after freight was transported each year, composed of the war. By 2004, volumes carried on the 251 252   THE WEB OF TR A NSPORT C ORRIDORS IN SOUTH A SI A MAP S8.1  The SAVA Waterway Rehabilitation Project Traversed Several Former Yugoslav Republics Source: World Bank. entire waterway had dwindled to less than River Basin (FASRB), with the support of the 400,000 tons a year, reflecting limitations on Stability Pact for Southeastern Europe navigability, draft, a lack of navigational (SPSEE), the Office of the High Representative markings to indicate the channels for safe pas- (OHR) in Bosnia and Herzegovina, and the sage, and the danger from unexploded ord- European Union (EU). The Framework nance and unmarked wrecks. In some of the Agreement was signed in Kranjska Gora river’s upper sections, navigation was possible (Slovenia) in December 2002 by Slovenia, for only 100 days a year or less because of Croatia, Bosnia and Herzegovina, and the for- draft restrictions. The d­ angers were so large mer Federal Republic of Yugoslavia (as Serbia that the owners of vessels could not obtain and Montenegro were known at that time). insurance to ­operate on the waterway. The Framework Agreement came into force By the early 2000s, these countries realized on December 29, 2004. It sought to create an that the status quo was unsustainable, so they international navigation regime on the Sava created a Framework Agreement on the Sava River and its tributaries; a regimen for C r o s s - B o r d e r I n f r a s t r u c t u r e P r o j e c t s : C h a l l e n g e s a n d L e s s o n s L e a r n e d    253 sustainable water management; and various height of the bridges to allow safe passage for measures to prevent or restrict the risks and larger vessels underneath, dredge and protect eliminate the hazards to navigation. the river, introduce electronic marking/vessel The Framework Agreement also created identification and tracking systems, and remove the International Sava River Basin sunken wrecks was €87 million.4 However, Commission (ISRBC), based in Zagreb, an based on the river length or border in each efficient, capable organization that was to national territory, costs were to be split between design/implement a program to carry out the three main countries—Croatia (61 percent), these tasks and liase with the Danube River Serbia (20 percent), and Bosnia and Herzegovina Basin Commission.2 The Sava River was des- (19 percent)—while the main beneficiaries ignated an International Waterway, along (Serbia, and Bosnia and Herzegovina) were with the three ports of Belgrade (Serbia), responsible for the smallest shares, reflecting the Brcˇko (in Bosnia and Herzegovina), and Sisak part of the river within their boundaries. Serbia, (Croatia). The commission’s operating costs despite the modest amount of investment it were to be covered by the participating coun- would need, saw the Danube River and Tiso tries, which would also provide more funds River as its main priorities, reflecting the traffic for further agreed-upon activities. The ISRBC carried on those systems. commissioned a prefeasibility and feasibility Another problem was in Bosnia and study and the necessary environmental impact Herzegovina: Under the Dayton Peace assessment (EIA), which were finalized at the Accords, transport and related infrastructure end of October 2008. They recommended were the responsibility of the two entities, the that the Sava River rehabilitation should Federation of Bosnia and Herzegovina (FBH) occur in two phases: from Belgrade to Brc ˇko, and the Republika Srpska (RS) and their min- achieving Class Va status; and from Brc ˇko to istries, but international transport agreements Sisak, achieving Class IV status.3 were the responsibility of Bosnia and The program was considered regional Herzegovina’s State Ministry of Transport, because it involved a shared public good (the which believed it was responsible for these Sava River), the different parties showed owner- transport links. The Republika Srpska did not ship, and there were clear regional spillover ben- accept this. Although it supported the project, efits and a mechanism for coordination among it did not recognize the authority of the state the national governments. Moreover, it was government of Bosnia and Herzegovina. Since supported by the European Union, the Stability the government had no authority to borrow Pact for South Eastern Europe, the Office of the money and no income to repay debt, World High Representative, and the World Bank. The Bank funds had to be provided to or through private sector was interested in investing in ves- the entities. This was a formidable obstacle. sels, port structures, and the river’s operations/ Similarly, the Dayton Peace Accords placed maintenance if it was safe and reliable, and the responsibility for air transport and regula- parties agreed on the goals. What could possibly tions between the entities at the state level: go wrong? Unfortunately, a great deal. the Federation of Bosnia and Herzegovina and the Republika Srpska. 5 The accord’s Annex 9 implied that the state was responsi- INSTITUTIONAL CHALLENGES TO ble for international and interentity water WORKING ACROSS BORDERS infrastructure, such as river traffic—which While the three national governments supported the Republika Srpska also disputed. To con- the proposed program, the level of support dif- firm the mandate, a state-level law on inland fered, sometimes markedly, reflecting the asym- water transport would be needed. Although metric distribution of costs and benefits, the one was drafted, it was never adopted, countries’ domestic priorities, and the resources because of the opposition from the RS repre- available to each. The estimated cost to raise the sentatives in the State Assembly, who believed 254   THE WEB OF TR A NSPORT C ORRIDORS IN SOUTH A SI A that the Federation of Bosnia and the countries could not find common ground Herzegovina and the Republika Srpska, along on the larger issues. with BAD, were responsible for all intra- entity river traffic, along with all aspects of PRACTICAL CHALLENGES TO waterway management/operations/infra- CROSS-BORDER IMPLEMENTATION structure for the Sava River sections within their borders. The next stage of preparation required The laws to regulate vessels on the Sava the engineering designs (for river protection, River in Bosnia and Herzegovina were also dredging, and channel enhancements), unclear. On the sections where it crossed including the hydraulic modeling of water Serbia, traffic would be regulated according flow given rainfall and runoff predictions, to its Law on Internal Navigation.6 Where it and safeguards for the whole length of the crossed Bosnia and Herzegovina, traffic river. The World Bank team originally pro- would be regulated by Bosnia and posed that Bosnia and Herzegovina apply Herzegovina’s Law on Internal and for a Project Preparation Advance (PPA)— Maritime Navigation. 7 While there was which could equal up to 20 percent of the some symmetry between the laws, it was credit/loan amount—to fund the work for unclear which one regulated traffic on sec- the Bosnian and Serbian sections. Croatia, tions shared either with Croatia or Serbia, as an EU candidate country, could have or which law regulated traffic on the short sought an EU grant to cover its prepara- section of river within BAD boundaries. tions. The World Bank proposed that Bosnia This issue could have been resolved, but and Herzegovina use a PPA to pay for the again there was no political consensus. limited design work for the Serbian section While the international organizations proac- because most of the design work was on the tively helped with the drafting and signing section in Bosnia and Herzegovina, and of the Framework Agreement and created Bosnia and Herzegovina would also receive the International Sava River Basin a disproportionate share of the benefits Commission, preparations continued w h e n i m p l e m e n t e d . H o w e v e r, T h e against a backdrop of waning international Republika Srpska’s Ministry of Finance influence in the domestic affairs of Bosnia ruled against this, noting that a PPA would and Herzegovina. Without this international require the same approvals as any loan or support, obtaining the consensus of key sub- credit agreement. Given the lack of political national stakeholders (primarily the entities) consensus in the three parliaments in Bosnia was impossible, and ultimately derailed the and Herzegovina over ownership, responsi- whole process. bilities, and operations, it was unlikely to be Another problem involved which agency achieved. would maintain the river in Bosnia and The International Sava River Basin Herzegovina. If the entities were responsible Commission then brokered an agreement for the infrastructure, then their respective between the three national governments. ministries of transport would ensure it was At that level, it was agreed that Croatia would navigable and safe. But the management and be responsible for developing the design on maintenance of the river appeared to be the the section from Brc ˇko to Sisak, using finan- responsibility of the two entity’s ministries of cial assets from the available Instrument for environment and water, and the two line Pre-Accession Grant (IPA) funds and EU agencies for managing water resources in the grants. The EU was willing for the grants it basin. Again, the situation was confused but gave to one country to be used to benefit could have been resolved with a political another, as long as the design of the down- ­ c onsensus. Details about the role of the stream section was progressing. The tripartite ­ private sector were not even discussed because agreement followed the same suggestion for C r o s s - B o r d e r I n f r a s t r u c t u r e P r o j e c t s : C h a l l e n g e s a n d L e s s o n s L e a r n e d    255 the downstream section as earlier: that consul- Financing for construction was equally tants procured by Bosnia and Herzegovina problematic. Croatia and Serbia were both would be responsible for preparing the design members of the World Bank Group’s from the river mouth in Belgrade to Brc ˇko. To International Bank for Reconstruction and this end, Bosnia and Herzegovina’s Ministry of Development (IBRD); Bosnia and Herzegovina Transport applied for EU grant money under a was a member of the World Bank’s different window than Croatia had used to International Development Association (IDA), fund the design/work to remove the embedded and thus eligible for financing at favorable mines in Bosnia and Herzegovina and Serbia— (concessional) terms; and Slovenia, at that which was needed before the design/dredging/ time, had graduated from the World Bank, construction phases could begin.8 and thus was not eligible for World Bank Unfortunately, Bosnia and Herzegovina’s par- financing. The Bank committed to provide liament challenged this before the grant agree- IDA financing for Bosnia and Herzegovina’s ment was signed between the EU and Bosnia share of the regional program, while the paral- and Herzegovina because it believed the grant lel project in Croatia was to be financed money allocated to Bosnia and Herzegovina entirely with EU grants. Thus, while Bosnia would be spent to benefit Serbia, without a and Herzegovina was willing to use noncon- clear quid pro quo from Serbia. cessional funding provided by a multilateral Even without these problems, the timeline development bank to rehabilitate the ports, for approving the grant was long, and it river, and access infrastructure, Croatia and depended on the World Bank confirming that Serbia wanted to use EU grant money under its funds would be available. However, the different windows, each with a different time- Bank’s finances depended on preparations line regarding availability and the likelihood moving forward in the countries, including of being provided, reflecting their respec- funds being approved to de-mine the river— tive status as candidate and preaccession something to which the EU would not agree countries. until the World Bank committed to its pro- The contracting/tendering strategies were posed investment. This Catch-22 was difficult also challenging because the river formed the to overcome. border between two of the countries along Implementation arrangements were also much of its length. The design consultant problematic. Croatia would use its line minis- needed to prepare an implementation/contract- try as its implementing agency, while continu- ing strategy to launch works from the river ing to coordinate through the International mouth at Belgrade to Brc ˇko (234 ­ kilometers). Sava River Basin Commission. Serbia would It would have to show that the works were use Plovput, the public agency with manage- needed in three countries, two entities, and ment and maintenance responsibilities on the BAD, sometimes on different sides of the river Danube and Tiso rivers within its boundaries. and within the river in the same spot, using In Bosnia and Herzegovina, the most efficient funds from three different sources, and with option would have been to establish a small up to five implementing units. The practical coordinating unit at the state level, reflecting challenges to designing such a contract in an the status of the river and the need for inter- efficient manner proved extremely difficult. national cooperation with the other two countries and the ISRBC, with maintenance ­ THE CURRENT SITUATION contracted out to a suitable body. However, this was not acceptable to Republika Srpska, The World Bank, after a preparation period of even if Plovput was the contracted party, over four years, decided insufficient progress as it did not recognize the authority of the had been made and withdrew its potential State Ministry of Transport in Bosnia and support. This led the EU to withdraw its Herzegovina in this area. grants for the preparations and de-mining. 256   THE WEB OF TR A NSPORT C ORRIDORS IN SOUTH A SI A Even today, communities along the river and fund, administered by the World Bank, for the vessels that operate on it must contend preparing/implementing works and mainte- with challenges similar to the ones they faced nance, would have helped resolve these in 1995. obstacles, and also would have enhanced the World Bank’s convening power. MAIN MESSAGES NOTES Strong regional leadership and coordination 1. The Brc ˇko Administrative District (BAD) was were and are needed at the national level and created by the Dayton Peace Accords, since the with key stakeholders at all levels, which did competing claims of the two entities, the not happen. Agreements must involve techni- Federation of Bosnia and Herzegovina (FBH) cal decisions, as well as delineating financing and the Republika Srpska (RS), were impossible and institutional responsibilities for preparing, to overcome, given that their territories over- implementing, and operating all the program’s lapped in the Brc ˇko District. The BAD is self- phases. While the international organizations governing, and at that time was internationally proactively helped with the drafting and supervised. The DPA was later amended and ­ signing of the Framework Agreement and the BAD received the same status as the other created the International Sava River Basin ­ two entities. 2. The International Commission for the Protection Commission, preparations overlapped at a of the Danube River—based in Vienna—was time when international influence in the created to ensure the sustainable, equitable use domestic affairs of Bosnia and Herzegovina of water and freshwater resources in the Danube was waning. Without it, obtaining the consen- River Basin. sus of key subnational stakeholders (primarily 3. Class IV European inland waterways of the three entities) was impossible. While this is international importance require a draft of ­ an exceptional case, the lesson is that interna- 2.5 meters and allow vessels of 1,000 to tional organizations must ensure a consistent 1,500 tons, while Class Va waterways require level of support to the stakeholders through- a draft of 2.5−2.8 meters and allow vessels out the process. weighing 1,500−3,000 tons. While this program was never imple- 4. The cost of removing the unexploded ord- nance was to be funded by EU grants. mented, progress was made in creating the 5. Dayton Peace Accords (1995), Annex 4. Framework Agreement and the International 6. Official Gazette of RS, November 14, 2001, Sava River Basin Commission, improving the no. 58. sustainability of management in the river 7. Official Gazette of FBH, December 28, 2005, basin, and identifying the major c ­ hallenges to no. 73, year XII. introducing a navigation system. These 8. After the war in the western Balkans (from actions would not have occurred without 1992 to 1995), many areas contained mines on the initial financing, and more important, both sides of the front lines. Despite extensive technical support from the international orga- efforts to remove them, by the end of 2008, nizations working in the region. However, there were still 220,000 land mines and unex- even if the program had progressed, financing ploded ordnance in over 13,077 locations, which were killing and injuring more than the preparation and subsequent work would 30 people each year at the time the project still have been a major obstacle, given the was being prepared. A total of 1,755 square cross-border nature of the preparatory work kilometers (3.4 ­percent of the countries’ terri- and subsequent physical works interventions, tory) was still considered hazardous. The Sava both of which, if implemented in the most River’s south bank was considered particularly efficient manner, required the contractor to dangerous because the river was the front line work across national borders. A pooled trust for much of its length. Spotlight 9 The Influx of Workers and Followers in a Transport Project: Lessons in Gender Risks from a Road Project in Uganda B ank-financed projects often involve civil works on road corridors that require workers and goods/services to be construction workers that may be related to religious, cultural, or ethnic differences, or based on competition for local resources, supplied from outside the local area because and put pressure on already-overstretched local labor is not available or lacks the nec- social services. Tensions can also occur or be essary skills. In many cases, the influx of sparked among different groups in the labor workers and service providers is com- force, and preexisting conflicts in the local pounded by ­ others who follow the workers community may be exacerbated. Ethnic and to sell them goods/services or to find jobs regional conflicts may be aggravated if work- and business opportunities. This rapid in- ers from one group move into the territory of migration to the project area is called a another. Some workers coming from outside labor influx. Under c ­ ertain conditions, it the region may have higher incomes than the can negatively affect the host communities local workers, which can cause resentment, (World Bank 2016a). particularly when the newcomers have rela- tionships, with married women and girls. Increased risk of illicit behavior and crime. LABOR INFLUX AND GENDER The newcomers may increase feelings of insecu- RISKS rity in the local community. Criminal activities An influx of labor and service providers can may increase, including theft, physical assaults, pose various risks. Appraisers, project spon- substance abuse, prostitution, and human traf- sors, and other stakeholders may fail to recog- ficking both by locals and newcomers drawn to nize the issues and their effect on the project, the area by the project. Local law enforcement especially when problematic social behavior is agencies may not be equipped to deal with the culturally tolerated or even accepted, nation- temporary increase in population. ally or locally. The risks are varied. Influx of additional population ­ (followers). Risk of social conflict. Conflicts can Especially in projects with large footprints flare up between local communities and and/or a longer time frame, others may 257 258   THE WEB OF TR A NSPORT C ORRIDORS IN SOUTH A SI A migrate to the project areas, exacerbating such as unwanted aggressive advances and sex- problems of the labor influx. These people ual harassment of women/girls/minors and may look for jobs with the project. The fol- exploitative s ­ exual relations.1 In addition, it lowers could also include worker’s family may lead to an increase in human trafficking, members, as well as other traders, suppliers, where women and girls are forced into sex and service providers (including sex workers). work. In rural ­ settings, where law enforcement The influx is induced by the heightened is limited, the risk of sexual harassment, espe- ­ economic opportunities surrounding the road cially of younger females, is apt to be high. construction investment. It can become d ­ angerous for them to walk on Impacts on community dynamics. roads to and from schools, markets, jobs, and Depending on the number of newcomers and water collection points. They often face rude their engagement with the host community, stares or derogatory comments, taunting, the composition of the local community and hounding, groping, or rape. Women are espe- social dynamics can change significantly, and cially afraid to walk alone in poorly lit or iso- preexisting social conflict may intensify. lated areas. Increased burden on and competition for Sex work. The high influx of outside work- public services. The presence of workers and ers can spark or lead to more sex work. The service providers (and sometimes their fami- concentration of male workers who earn rela- lies) means more demand for public services tively higher incomes, can draw in young girls such as water, electricity, medical care, trans- and women, exacerbating the risks of gender- port, and education. This is particularly the based violence and STDs. case when the influx is not accompanied by Child labor and school dropouts. extra or separate supply systems. Increased opportunities for the host com- Increased risk of communicable diseases munity to sell goods/services can lead to and burden on local health services. The children being asked to produce and deliver influx can introduce communicable diseases, them, which, in turn, means they must drop including sexually transmitted diseases out of school. Young women, in particular, (STDs), or incoming workers may be exposed may drop out because of early marriages or to diseases to which they have low resistance. pregnancies. Both situations strain local health resources. Local inflation of prices. A significant Workers with health concerns—such as sub- increase in the demand for goods/services can stance abuse, mental issues, or STDs—may lead to price hikes and/or crowding out of not wish to visit the project’s medical facility local consumers. and instead may go to local medical provid- Increased pressure on rental accommoda- ers, further straining local resources. Health tions. Depending on workers’ income and and rescue facilities may be overwhelmed and/ their accommodations, demand for housing or ill equipped to address the industrial acci- may increase, which can also cause price dents that occur on large construction sites. hikes and crowd out local residents. Gender-based violence (GBV). Construction Increases in traffic and related accidents. workers are mainly younger males who are Transport of supplies and workers can away from home, are separated from their increase traffic, accidents, and the burden on families, and are unknown in the local areas. local facilities. Some drivers with the project They may believe that they can behave in ways may speed on the construction roads, causing they normally would not, without fear of accidents and injuries. repercussion. This typically leads to fraterniza- These effects are usually worsened by the tion—close social relations considered inappro- low capacity (at the local level) to manage and priate with those who are unrelated to one absorb the incoming labor force, especially another, typically with local females. It also when civil works are in or near vulnerable leads to unacceptable and/or illicit practices, communities and other high-risk situations. INFLU X OF WORKERS A ND FOLLOWERS IN A TR A NSPORT PRO J E C T   259 THE NEGATIVE IMPACTS OF A also worked closely with the civil society LABOR INFLUX RELATED TO A organization Joy for Children Uganda ROAD PROJECT IN UGANDA (JFCU), which supported the community in lodging the complaint during a mission in In a road project in Uganda—the Uganda May 2015. As a result, the Bank concluded Transport Sector Development Project (TSDP) that there was credible evidence that some and Additional Financing—supported by the road workers had engaged in sexual miscon- World Bank, the negative outcomes of the duct with minors. This demonstrated to the labor influx were profound. The project aimed Bank yet another side of the serious nature of to improve the links and efficiency of Uganda’s the GBV risks associated with the labor influx transport system by upgrading its national that needed to be mitigated. road network and improving the management of the roads, road safety, and transport. It was Lessons learned launched by the Uganda National Roads Authority (UNRA). Work on the 66-kilometer The World Bank prepared a detailed report stretch from Kamwenge to Fort Portal Road describing the issues and the Bank’s response begun in August 2013 and was expected to be at the corporate and project levels (World completed by January 2016. But the road Bank 2016b). Three major lessons stand out project had a number of negative environmen- from the experience in this report. tal and social impacts, such as exacerbating Lesson 1. It is necessary to create a clear existing GBV and Violence Against Children institutional architecture and understand/ (VAC) that overwhelmed the project and led address capacity constraints. This involves the World Bank to suspend and subsequently defining the roles/responsibilities of different cancel it in December 2015 because the gov- ­ parties to address gender-based risks so ernment failed to conform to certain environ- they can collaborate to mitigate them. mental and social standards/practices. The Responsibilities need to be legally and contrac- project offers several important lessons in tually binding. All parties must ensure compli- understanding the types of gender risks that ance with national laws, standards, and can arise and how they must be managed. regulations related to occupational health and The project was the subject of a World Bank safety, labor welfare/conditions, the environ- Inspection Panel investigation. The panel’s ment—including management of quarries, request for inspection, registered on September social issues (specifically gender issues), and 28, 2015, addressed complaints from commu- road safety. These parties include the imple- nities along the Kamwenge-to-Fort-Portal road menting agency (usually the road agency/ alleging a variety of negative environmental authority acting as the employer on the road and social impacts, in addition to those identi- contract); road works contractor (including fied by the project’s Environmental and Social subcontractors); the supervising engineer; dis- Impact Assessment (ESIA). The GBV impacts tricts or local governments where the roads included the road workers’ sexual relations cross local communities; central government with minor girls, as well as sexual harassment ministries responsible for gender, education, of female employees. and health; local community service organiza- Confirming the facts in the allegations tions that work on these issues (for prevention was extremely difficult (World Bank 2015)— and response); and the police. These parties partly because many community members need a joint mechanism for working together to and officials were reluctant to discuss the address gender-related risks. Financial resources issues. However, over time, the World Bank’s and arrangements also need to be clear to avoid missions gained more insight into the com- turf wars. Further, it is critical to appraise the plaints, particularly when the Bank hired different parties’ capacities to carry out their social consultants in April 2015. The Bank responsibilities. If weaknesses are observed, the 260   THE WEB OF TR A NSPORT C ORRIDORS IN SOUTH A SI A project needs to develop an action plan to over- ensure they are implemented. Best practice come them before it is too late. measures include the following: Lesson 2. It is important to understand the social and environmental context of the •  Ensure that the borrower is committed to project. More must be invested initially to addressing these issues. understand the environment where the proj- •  Address child protection risks before the ect will be implemented. This social context project begins and throughout the project is essential to identify the broad risks to cycle. poor rural communities, among others, •  Incorporate social and environmental mit- caused by large influxes of local and foreign igation measures into the civil works construction workers, and then create mea- contracts. sures to reduce them. Further, social/tribal •  Incorporate strong environmental and norms, behavior, and culture must be exam- social oversight responsibilities and staff- ined to address ­ gender-based risks. There ing needs in the supervising engineer’s should be stakeholder focus groups, inter- contract. It is critical that an independent views, and meetings. An upstream assess- third party oversee issues related to GBV. ment is also needed to identify social and •  Improve the quality of the ESMPs and environmental issues in projects, focusing ensure that labor influx concerns are on the country context, as well as the local included. context. •  Ensure that local authorities are actively Lesson 3. It is important to prepare/ engaged. launch measures to address emerging gender •  Ensure that the contractor and supervising risks. The main recommendations from engineer implement their mandates on the Environmental and Social Impact these issues. Assessments (ESIAs) and stakeholder focus •  Ensure that adequate community-­ groups, interviews, and meetings must be engagement and grievance-management translated into actions in the project’s envi- committees are created to receive, channel, ronmental and social management plans and refer or respond to complaints or and supervised ­ r egularly. A fundamental issues. change is needed to ensure that systemic •  Provide adequate resources for this work. social risks are addressed in a timely manner •  Agree on identification and reporting proto- in the ESIAs and Environmental and Social cols for GBV and violence against children Management Plans (ESMPs) of infrastruc- when these occur and ensure that informa- ture investments; appropriate mitigation tion flows from households to the responsi- measures are identified to deal with them; ble parties (police, referral service providers, enforceable construction-related mitigation institutions’ management) for timely action. measures are included in the contract; the •  Develop and implement a zero-tolerance capacity of implementing entities to manage sexual harassment policy. the contract and enforce the measures is •  Implement a workers’ code of conduct carefully appraised; and the World Bank’s that is included in their contracts and is supervision of the measures’ implementa- enforced. tion become a focus of Bank support. The •  Launch awareness campaigns for workers ESMP and mitigation measures should be an and communities through education and integral, enforceable part of construction communications materials (such as posters), contracts. The site-specific ESMP should be as well as through the local media, with approved in a defined period after construc- radio and TV talk shows, advertisements, tion begins. These activities, and the cost of or programs. the measures, should be explicit in the •  Encourage the participation of all stakehold- bidding documents and final contracts to ers including men to prevent GBV within INFLU X OF WORKERS A ND FOLLOWERS IN A TR A NSPORT PRO J E C T   261 the community and worker-led efforts on the Empowerment and Livelihoods Assistance ­ employee’s side to prevent GBV. (ELA) model for adolescent girls.2 •  Encourage the local recruitment of •  Ensure that those in power do not retaliate workers. against people who identify risks related to •  Empower women and girls with job oppor- the project. Provide opportunities for tunities through affirmative action mea- anonymous reporting through hotlines and sures during the recruitment process. coordination with the police. •  Provide gender-segregated sanitation facili- •  Collaborate with police authorities to ensure ties at all project sites. that workers’ criminal behavior is punished •  Ensure safe walking paths for women, chil- and thus deters such future behavior. dren, and the disabled by maintaining easy •  Ensure that response measures are created, access, introducing traffic management including a minimum package for survi- controls, and conducting safe infrastruc- vors of gender-based violence (referral ture and road safety campaigns. paths for health, psychosocial support, •  Ensure that sufficient background checks legal redress, reintegration into schools, are made on the workers and obtain infor- livelihood assistance and training pro- mation about them—including their grams, fighting stigma, and protection of names, places of origin, next of kin, and the identities and dignity of survivors). reference letters from previous employers or authorities from the previous places they lived. Issue identity cards so the public can NOTES easily identify them. 1. The term “minor” is defined on the basis of a •  Identify workers who will address these country’s legal framework. In many countries, issues and place them throughout the work- the age of consent (which determines who is force to act as the eyes and ears on the no longer a minor) is 18 years. The World ground. Bank, in line with UN Secretary-General’s •  Ensure that worker accommodations do Bulletin guidance of 2016, takes the cutoff not create opportunities for committing age to be 18 years (World Bank 2016a). sexual offenses by restricting housing to 2. See http://www.brac.net/search/item/723​ certain areas and collaborating with local -­empowerment​-and-livelihood​-for​-adolescents. authorities and landlords to monitor and report on behavior that violates the code of conduct or the country’s laws. REFERENCES •  Ensure strict fleet control arrangements, World Bank. 2015. “Management Response to especially so drivers do not commit crimes Request for Inspection Panel Review of the or drive recklessly. Uganda Transport Sector Development Project •  Hire an HIV/AIDS service provider to – Additional Financing (P121097).” World develop awareness, counseling, testing, Bank, Washington, DC. treatment and support for workers and ———. 2016a. “Managing the Risks of Adverse Impacts on Communities from Temporary community members, including highly vul- Project-Induced Labor Influx.” Note, Operations nerable groups like sex workers. Policy and Country Services and Environmental •  Ensure a partnership with a civil society and Social Safeguards Advisory Team, World organization(s) throughout the project if Bank, Washington, DC. risks are great. ———. 2016b. “Uganda Transport Sector •  Build the host communities’ resilience Development Project—Additional Financing: using empowerment models for children Lessons Learned and Agenda for Action.” World and youth, such as BRAC International’s Bank, Washington, DC. ECO-AUDIT Environmental Benefits Statement The World Bank Group is committed to reducing its environmental footprint. In support of this commitment, we leverage electronic publishing options and print- on-demand technology, which is located in regional hubs worldwide. Together, these initiatives enable print runs to be lowered and shipping distances decreased, resulting in reduced paper consumption, chemical use, greenhouse gas emissions, and waste. We follow the recommended standards for paper use set by the Green Press Initiative. The majority of our books are printed on Forest Stewardship Council (FSC)–certified paper, with nearly all containing 50–100 percent recycled content. The recycled fiber in our book paper is either unbleached or bleached using totally chlorine-free (TCF), processed chlorine–free (PCF), or enhanced elemental chlo- rine–free (EECF) processes. More information about the Bank’s environmental philosophy can be found at http://www.worldbank.org/corporateresponsibility. T he WEB of Transport Corridors in South Asia develops a holistic appraisal methodology to ensure that economic benefits of investments in transport corridors are amplified and more widely spread, and possible negative impacts such as congestion, environmental degradation, and other unintended consequences are minimized. It focuses on South Asia—not only as one of the world’s most populous and poorest regions—but as a hinge between East Asia, Central Asia, the Middle East, and Europe. The book is aimed at politicians, technocrats, civil society organizations, and businesses. It presents case studies of past and recent corridor initiatives, provides rigorous analysis of the literature on the spatial impact of corridors, and offers assessments of corridor investment projects supported by international development organizations. A series of spotlights examines such issues as private sector co-investment; the impacts of corridors on small enterprises and women; and issues with implementing cross-border corridors. The “WEB” in the title stands for both the wider economic benefits (WEB) that transport corridors are expected to generate and the complex web of transport corridors that has been proposed. The appraisal methodology introduced in this book shows how the web of interconnected elements around corridors can be disentangled and the most promising corridor proposals—the ones with the greatest wider economic benefits—can be selected. SKU 211215