South Asia’s Turn
South Asia’s Turn Policies to Boost Competitiveness and Create the Next Export Powerhouse
Gladys Lopez-Acevedo, Denis Medvedev, and Vincent Palmade, Editors © 2017 International Bank for Reconstruction and Development / The World Bank 1818 H Street NW, Washington, DC 20433 Telephone: 202-473-1000; Internet: www.worldbank.org
Some rights reserved 1 2 3 4 20 19 18 17
This work is a product of the staff of The World Bank with external contributions. The findings, interpre- tations, 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. The World Bank does not guarantee the accuracy of the data included in this work. The boundaries, colors, denominations, and other information shown on any map in this work do not imply any judgment on the part of The World Bank concerning the legal status of any territory or the endorsement or acceptance of such boundaries. Nothing herein shall constitute or be considered to be a limitation upon or waiver of the privileges and immunities of The World Bank, all of which 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: Lopez-Acevedo, Gladys, Denis Medvedev, and Vincent Palmade, eds. 2017. South Asia’s Turn: Policies to Boost Competitiveness and Create the Next Export Powerhouse. South Asia Development Matters. Washington, DC: World Bank. doi:10.1596/978-1-4648-0973-6. 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 and should not be considered an official World Bank translation. The World Bank 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. 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. 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: [email protected].
ISBN (paper): 978-1-4648-0973-6 ISBN (electronic): 978-1-4648-0974-3 DOI: 10.1596/978-1-4648-0973-6
Cover photo: © Ismail Ferdous / World Bank. Permission required for reuse.
Library of Congress Cataloging-in-Publication Data has been requested. Contents
Foreword ...... xi
Acknowledgments ...... xiii
Abbreviations ...... xv
Overview ...... 1 With some exceptions, South Asia has reaped few benefits from global integration ...... 1 Productivity is the key to improved competitiveness ...... 2 Business environment challenges remain a constraint on firm performance ...... 3 Growth of the region’s cities and clusters offers multiple opportunities to raise firm productivity ...... 4 Increasing prominence of global value chains provides a pathway to greater efficiency .....4 Improving firm capability and leveraging technology can substantially raise firm productivity ...... 6 Faster growth of exports and jobs is within reach if productivity improves ...... 7 Policies to boost competitiveness and productivity ...... 7 Reference ...... 9
Part 1 SOUTH ASIA’S COMPETITIVENESS CHALLENGE AND OPPORTUNITY 1 The Region’s Competitiveness Potential Remains Largely Unrealized ...... 13 Pockets of excellence are evidence of vast untapped potential ...... 14 Difficulties in attracting investment and penetrating global markets ...... 14 Little progress in diversifying the merchandise export basket ...... 18 Elusive sophistication and low quality of exports ...... 19 Annex 1A ...... 26 Notes...... 28 References ...... 29
v vi CONTENTS
2 Improving Competitiveness Requires Raising Productivity Rather Than Keeping Costs Low ...... 31 Macro challenge: Contribution of TFP to growth is low and declining ...... 32 Sectoral challenge: Slow pace of structural transformation ...... 34 Firm challenge: Firm growth is low and resources are trapped in small firms ...... 35 Four case studies of important industries in South Asia ...... 42 Annex 2A ...... 44 Notes...... 46 References ...... 47
Part 2 PRODUCTIVITY PERFORMANCE: FIRMS AND LINKAGES 3 Business Environment Challenges Continue to Weigh on Firm Performance ...... 51 Industry Case Study A: Industry-specific business environment issues in agribusiness ...... 57 Note ...... 61 References ...... 61 4 Productivity-Boosting Agglomeration Economies Are Underleveraged ...... 63 Economic activity in South Asia is highly concentrated ...... 63 Agglomeration economies raise firm productivity ...... 65 Resources do not flow easily to more-productive firms ...... 68 Industry Case Study B: Better cities and trade logistics are needed for the electronics sector to thrive ...... 70 Notes...... 76 References ...... 76 5 Limited Success in Linking to Global Value Chains ...... 79 GVC participation supports productivity ...... 79 South Asia’s success in global and regional GVCs is limited to apparel ...... 82 Most policy determinants of GVC participation are lacking ...... 88 Industry Case Study C: The effect of trade barriers on the apparel sector ...... 91 Notes...... 97 References ...... 97 6 Firm Capabilities Are Constrained ...... 101 Firms lack managerial quality and do not use resources efficiently ...... 101 Adoption of knowledge and technology is low ...... 103 Innovation is widespread but novelty is limited ...... 108 Returns to innovation are high ...... 113 Annex 6A ...... 115 Industry Case Study D: Protection from global good practices limits the spread of world-class firm capabilities in the automotive industry ...... 117 Notes...... 124 References ...... 125 CONTENTS vii
Part 3 THE WAY FORWARD 7 Potential for Increased Growth through Policy Reforms ...... 131 Macro benefits: Faster export growth through higher productivity ...... 131 Sectoral benefits: More jobs, higher earnings, greater inclusion ...... 133 Firm benefits: Greater density of successful firms ...... 137 Annex 7A ...... 139 Notes...... 142 References ...... 143 8 Need for Greater Emphasis on Trade Policies, Spatial Policies, and Firm Capabilities .... 145 Policies to better connect to global value chains ...... 145 Policies to maximize agglomeration benefits ...... 149 Policies to strengthen firm capabilities ...... 153 Lessons from the four industry case studies: Drivers and constraints of competitiveness as seen by the managers of leading firms ...... 157 Notes...... 160 References ...... 160
Boxes 2.1 Barriers to competition and productivity dispersion ...... 38 3.1 Efforts to improve the investment climate in South Asia ...... 55 3.2 Promoting access to finance and backward links in Afghanistan ...... 60 4.1 Agglomeration and productivity ...... 66 4.2 How Bangladesh missed the opportunity to break into the global electronics industry ...... 73 4.3 How Tos Lanka is developing the skills of its workforce ...... 75 4.4 Foxconn enters India ...... 75 5.1 Measuring GVC participation ...... 83 5.2 Expert evaluation of the duty drawback system in Pakistan ...... 94 6.1 Data for analysis of ICT adoption in South Asia ...... 104 6.2 Determinants of ICT adoption ...... 107 6.3 Innovation activities and outputs in the Enterprise Surveys ...... 108 6.4 The Crepon-Duguet-Mairesse model ...... 114 6.5 How leading Indian automotive firms acquired their world-class skills ...... 121 7.1 Why focus on women? ...... 135 8.1 Leveraging urbanization in South Asia ...... 152 8.2 Public support for the development of Pusa-1121 basmati rice in India ...... 154 8.3 China’s approach to workforce skills ...... 156
Figures 1.1 Countries in South Asia attract less foreign direct investment than global peers .....15 1.2 South Asia’s trade integration is relatively low ...... 16 1.3 Outside of garments, South Asia’s market share growth has been low ...... 17 1.4 Most export growth in South Asia has taken place along the intensive margin ...... 18 viii CONTENTS
1.5 Average export sophistication in South Asia has been on the decline...... 20 1.6 Export quality is generally low and has declined for some countries ...... 21 1A.1 Export Orientation Index—Rankings for South Asia region countries and comparators ...... 27 1A.2 Import Orientation Index—Rankings for South Asia region countries and comparators ...... 28 2.1 Growth in real GDP and real GDP per capita in South Asia has accelerated ...... 33 2.2 South Asia’s growth has been largely driven by factor accumulation ...... 33 2.3 Agriculture has given way to industry and services in South Asia ...... 34 2.4 As incomes grow, resources shift from agriculture to industry and services ...... 36 2.5 As incomes grow, resources shift from agriculture to industry and services ...... 37 2.6 Size distribution of firms in India is heavily biased toward small firms ...... 40 2.7 Labor productivity is lower in small firms in India and China ...... 41 2.8 Firms in South Asia grow less rapidly than those in comparator countries ...... 42 2A.1 Scatterplot of firms at the 25th percentile of employment distribution and income per capita ...... 45 2A.2 Scatterplot of firms at the 50th percentile of employment distribution and income per capita ...... 45 2A.3 Scatterplot of firms at the 75th percentile of employment distribution and income per capita ...... 46 3.1 South Asia countries lag behind comparators in business environment rankings .....52 3.2 Food spending in relation to GDP per capita ...... 57 3.3 Horticulture yield ...... 58 3.4 Productivity in agro-food processing firms ...... 58 4.1 “Thick” district borders prevent efficient allocation of resources ...... 69 4.2 Electronics production is low in South Asia ...... 71 4.3 India has a cost advantage in labor-intensive processes ...... 71 4.4 Process times for information technology hardware and electronics ...... 73 4.5 Logistics issues in India ...... 74 5.1 Structure of the global value chain for apparel ...... 80 5.2 Firms that participate in GVCs also have higher productivity ...... 81 5.3 South Asia’s strength in apparel explains the region’s high GVC integration ...... 82 5.4 Apparel accounts for the lion’s share of South Asia’s GVC exports ...... 84 5.5 India’s auto and electronics exports have focused on the lower end of the market . . . 85 5.6 India relies less on foreign inputs than its comparators ...... 85 5.7 Although product sophistication has increased in some instances, market sophistication of South Asia’s textile and apparel exports has gone down ...... 86 5.8 South Asia is developing a regional GVC in apparel ...... 87 5.9 South Asia is highly integrated with East Asia in GVC products ...... 88 5.10 GVC capability gaps compared to ASEAN and SACU are concentrated in areas amenable to policy intervention ...... 90 5.11 SAR, unlike its competitors, focuses heavily on cotton ...... 95 5.12 India and Pakistan are far behind in apparel exports, 2005–12 ...... 95 6.1 Management capabilities in South Asia are relatively weak ...... 102 6.2 ICT adoption varies substantially across countries in South Asia ...... 104 B6.1.1 Components of the ICT Index ...... 105 6.3 Firms use the Internet to reach customers more than to connect with suppliers ....106 6.4 Innovation rates vary significantly across South Asia ...... 111 6.5 The majority of innovation takes place in-house rather than collaboratively ...... 112 CONTENTS ix
6.6 Share of exports in production in the automotive sector, 2014 ...... 118 6.7 Value added per worker in the automotive sector ...... 119 6.8 Overview of skills in Indian and Chinese automotive firms ...... 120 6.9 Expenditures on R&D in India and globally ...... 120 6.10 Domestic production of auto parts and nominal tariff reduction in India ...... 123 6.11 Exports of commercial vehicles and nominal tariff reduction in India ...... 123 7.1 Continued productivity growth could lead to substantial gains in South Asia’s global market share ...... 132 B7.1.1 Working in garments with fewer children ...... 136 7A.1 South Asia’s sources of growth, baseline, 2011–30 ...... 141 8.1 Tariff reduction and faster growth of auto parts production in India went hand in hand ...... 146 8.2 South Asia’s apparel input mix is not fully consistent with global demand patterns ...... 148 8.3 Labor regulations are a more important constraint in South Asia than in other regions ...... 152 8.4 South Asia invests relatively less in R&D ...... 155
Tables 1.1 Despite rapid growth, South Asia’s share of global merchandise exports remains low ...... 16 1.2 South Asia has more than tripled its market share in services since the 1990s ...... 17 1A.1 Per capita income and ratios of FDI, imports, and exports to GDP by region, 1990–2014 ...... 26 1A.2 Changes in FDI, imports to GDP, and exports to GDP by region, 1990–2014 ...... 26 2.1 South Asia lags well behind China in apparel exports despite much lower labor costs ...... 32 2.2 Labor productivity drives structural transformation ...... 38 2.3 Productivity dispersion across South Asia’s firms tends to be large ...... 39 2.4 Small firms dominate the distribution across South Asia ...... 39 2.5 Country coverage of the four industry case studies ...... 44 3.1 Firms in South Asia are at a disadvantage with respect to investment climate ...... 53 3.2 South Asia’s investment climate deficiencies constrain firm performance ...... 54 4.1 Employment across South Asia is concentrated in a few districts ...... 64 4.2 Spatial concentration of manufacturing in South Asia has not changed much over time ...... 65 4.3 Agglomeration economies are associated with higher firm productivity in India and Bangladesh ...... 67 4.4 Agglomeration economies boost firm performance in Sri Lanka ...... 67 4.5 More-productive firms derive greater benefits from agglomeration economies ...... 68 5.1 Outside of apparel, GVC exports from South Asia lag well behind East Asia ...... 83 5.2 Many GVC capabilities and endowments in South Asia are limited ...... 89 5.3 South Asia is less competitive than Southeast Asia in noncost areas ...... 92 5.4 Import tariffs in South Asia and East Asia, 2014 ...... 93 B6.1.1 ICT questions from the Innovation Module ...... 105 6.1 Knowledge capital intensity in South Asia is lower than in other regions ...... 110 6.2 Imitation is the most common form of innovation in South Asia ...... 112 x CONTENTS
6.3 Agglomeration in South Asia matters most for radical innovation ...... 113 6.4 Innovation helps increase firm productivity ...... 115 6A.1 Determinants of overall ICT adoption ...... 115 6A.2 Determinants of innovation ...... 116 7.1 Southeast Asia benefits more than South Asia from a rise in Chinese apparel prices ...... 134 7.2 Higher wages increase women’s labor force participation ...... 136 7.3 Increased apparel price competitiveness vis-à-vis China can create many jobs in South Asia ...... 137 8.1 Major constraints to competitiveness identified in the industry case studies ...... 159 Foreword
outh Asia is at a turning point. The One of these policy levers (improving the region is benefitting from a confluence of business environment) is well known, but Spositive internal and external forces. much remains to be done. The other three South Asian countries are starting to receive (policies to better connect to global value the competitiveness dividends from the eco- chains, maximize agglomeration benefits, nomic reforms and public investments in and strengthen firm capabilities) are much infrastructure and education carried over the less discussed, and we hope this report will past 25 years. Rising labor costs in East Asia help policy makers focus more on them. are steering global investors toward South The report combines a critical mass of Asia as a possible cheaper alternative. At a quantitative analysis, using the latest data and time of declining global growth and trade, tools available, with a rich set of industry and South Asia—home to a quarter of humanity— company case studies to draw new insights on has the potential to boost global growth as what South Asia needs to do to boost com- both a major exporter and consumer market. petitiveness. And it proposes a number of spe- This is good news, not only for South Asia but cific policy solutions drawn from relevant also for the world as a whole. But challenges international good practices (including from to the region’s competitiveness remain. More within the region). than one million young people are reaching We very much hope that this report will working age every month and will need jobs; help the countries of South Asia, individually firm competitiveness is low; and countries in as well as collectively, take a turn toward real- the region have not been particularly successful izing their great competitiveness potential. in integrating with each other. This book, South Asia’s Turn: Policies to Annette Dixon Boost Competitiveness and Create the Next Vice President Export Powerhouse, looks in detail at the South Asia Region drivers and constraints impacting South Asia’s World Bank Group competitiveness. It outlines the four policy levers that will help the region become Anabel González more globally competitive across a broader Senior Director spectrum of industries, accelerating growth Trade and Competitiveness Global Practice and reducing poverty, especially for women. World Bank Group
xi
Acknowledgments
his report was prepared by a team led Sousa; and chapters 7 and 8 by Gladys Lopez- by Denis Medvedev and Vincent Acevedo, Vincent Palmade, and Dominique TPalmade, both of the World Bank’s van der Mensbrugghe. Bill Shaw streamlined, Trade and Competitiveness Global Practice organized, and edited the narrative. (GTCDR), under the guidance of Esperanza The agribusiness case study (attached to Lasagabaster (practice manager, GTCDR) chapter 3) was authored by Soujanya and Martín Rama (chief economist, South Chodavarapu, Asa Giertz, and Peter Jaeger. Asia Region). Annette Dixon (vice president, The electronics case study (attached to South Asia Region), Anabel González chapter 4) was authored by Ashish Narain (senior director, GTCDR), and Cecile with support from Sriyani Hulugalle, Fruman (director, GTCDR) linked the team Tim Sturgeon, and Daria Taglioni. The to the World Bank Group’s overall strategy apparel case study (attached to chapter 5) and steered them in that direction. was authored by Gladys Lopez-Acevedo, This report presents a combination of find- Raymond Robertson, and Atisha Kumar. The ings from a cross-cutting analysis of the automotive case study (attached to chapter 6) dynamics of firms, clusters, value chains, and was authored by Priyam Saraf with support cities across the region, as well as detailed from Michel Bacher, Amjad Bashir, and case studies of the apparel, electronics, auto- Sanjay Kathuria. motive, and agribusiness industries. The full Major contributions were received from case studies are available online at www Alvaro Gonzalez, Atisha Kumar, and Siddharth .worldbank.org/SouthAsiaCompetes. Sharma, and the team benefitted from the Chapter 1 of the report was authored by advice provided by Ana Margarida Fernandes Antonio Martuscelli and Gonzalo Varela; and Sebastian Saez. Guoliang Feng, Deeksha chapter 2 by Apoorva Gupta, Antonio Kokas, and Lara Loewenstein provided invalu- Martuscelli, and Gonzalo Varela; chapter 3 by able research assistance. Rosanna Chan, Filipe Lage de Sousa, Deeksha Kokas, and Paramita Dasgupta, Tugba Gurcanlar, Luke Giuliana de Mendiola Ramirez; chapter 4 by Jordan, Bertine Kamphuis, Sanjay Kathuria, Filipe Lage de Sousa and Deeksha Kokas; Eric Manes, Anna Reva, Fatima Shah, and chapter 5 by Michael Ferrantino and Gaurav Shahid Yusuf provided critical inputs during Nayyar; chapter 6 by Xavier Cirera and the inception phase. Tanya Cubbins and Ana Cusolito, with inputs from Filipe Lage de Isaac Lawson provided production and
xiii xiv ACKNOWLEDGMENTS
logistical support. Sections of the report draw Pravin Krishna (Johns Hopkins University), and extensively from a recently prepared World Bill Maloney (Equitable Growth, Finance, and Bank report, Stiches to Riches. The peer review- Institutions, World Bank). ers were Shubham Chaudhuri (Macroeconomics The team is grateful to South Asia country and Fiscal Management, World Bank), directors and, most important, country Uri Dadush (Carnegie Endowment for authorities for their support as well as the International Peace), Navin Girishankar dozens of companies, industry associations, (Trade and Competitiveness, World Bank), and experts interviewed for this report. Abbreviations
ADF Agriculture Development Fund (Afghanistan) AEPC Apparel Export Promotion Council (India) ASEAN Association of Southeast Asian Nations ASI Annual Survey of Industries BEPZA Bangladesh Export Processing Zones Authority BPO business process outsourcing CAGR compound annual growth rate CBU completely built unit (automobiles) CDE constant differences in elasticity CES constant elasticity of substitution CGE computable general equilibrium CIF cost, insurance, and freight CSR corporate social responsibility DTRE duty and tax remission for export program EPZ export processing zone EU European Union FAO Food and Agriculture Organization (United Nations) FDI foreign direct investment FOB free on board GCI Global Competitiveness Index GDP gross domestic product GSEM generalized structural equation model GTAP Global Trade Analysis Project GVC global value chain HTGL Hi-Tech Gear Ltd. IAMC Integrated Assessment Modeling Consortium IARI Indian Agricultural Research Institute ICP International Comparison Project ICT information and communication technology IIASA International Institute for Applied Systems Analysis ILFS Infrastructure Leasing and Financial Services Ltd. (India)
xv xvi ABBREVIATIONS
ILO International Labour Organization ISIC International Standard Industrial Classification IT information technology LES linear expenditure system LPI Logistics Performance Index MAOTRI Market Access Overall Trade Restrictiveness Index MFA Multi-Fiber Agreement MFN most-favored nation MSSL Motherson Sumi Systems Ltd. NAFTA North American Free Trade Agreement n.e.s. not elsewhere specified NGO nongovernmental organization NTM nontariff measures OECD Organisation for Economic Co-operation and Development OEM original equipment manufacturer OLS ordinary least squares OTRI Overall Trade Restrictiveness Index PPP purchasing power parity R&D research and development SACU Southern African Customs Union SAR Special Administrative Region SEZ special enterprise zone SMEs small and medium enterprises TFP total factor productivity TPP Trans-Pacific Partnership TTIP Trans-Atlantic Trade and Investment Partnership USAID U.S. Agency for International Development USITC U.S. International Trade Commission WITS World Integrated Trade Solutions WTO World Trade Organization
All monetary amounts are U.S. dollars unless otherwise indicated. Overview
hich region will become the next do so in the dynamics of firms, industry value global factory? As the workforce chains, clusters, and cities across the region. Wages and labor costs rise in China and other East Asian countries, many eyes turn With some exceptions, South Asia to South Asia. South Asia’s potential is unques- has reaped few benefits from tionable: education levels are on the rise, more global integration than one million young workers enter the labor market each month, and the population Despite the recent slowdown in trade, the of the region’s mega-agglomerations and global economy is more connected than ever. sprawling cities is expanding monthly by Participation in larger regional and global roughly the same number. By 2030, more than markets offers many opportunities to raise a quarter of the world’s working adults will productivity: stronger competitive pressures live in South Asia. But despite flashes of weed out the least productive firms, while oth- brilliance across a handful of sectors, loca- ers improve by gaining access to knowledge tions, and leading firms, this potential remains and better inputs. Across the region, there are largely unfulfilled. several examples of these channels at work, South Asia ranks below both its neighbors ranging from highly successful apparel indus- and various global benchmarks in attract- tries in Bangladesh and Sri Lanka to India’s ing investment, penetrating tough markets, software and business process outsourcing diversifying and upgrading its products, and (BPO) sectors, and from productive agglom- regional integration. These forgone opportu- eration of light manufacturing firms in Sialkot, nities are also reflected in low scores across Pakistan, to Bangalore, India’s becoming a a range of measures of international competi- global research and development (R&D) hub tiveness. With the growth rate of global trade for major auto parts and electronics produc- having dramatically slowed down, how can ers. South Asia’s leading firms have risen to South Asia improve its competitiveness, standards of global excellence, demonstrating become an export powerhouse, create jobs, that world-class levels of operational perfor- reduce poverty, and boost shared prosperity? mance, efficiency, and innovation can be This report proposes that the solution lies in achieved with the right management, scale, improving productivity and looks for ways to technology, and worker training. Since 2010,
1 2 SOUTH ASIA’S TURN
on the strength of its services trade, South prices on resource-rich economies, while the Asia’s ratio of trade to gross domestic product benefits to many commodity importers have (GDP) has consistently been better than been eroded by declining remittances. New China’s, and—after controlling for size—India megaregional trade agreements, such as is more open to trade than the global average. the Trans-Pacific Partnership (TPP) and In the aggregate, however, South Asia’s the Trans-Atlantic Trade and Investment intraregional and global ties remain relatively Partnership (TTIP), may divert trade and weak. The ratios of merchandise trade to investment away from nonmembers. Against GDP and foreign direct investment (FDI) to this background, it has become even more GDP in the region are well below those of urgent for countries in South Asia to make competitor countries. From 1990 to 2014, overdue investments in boosting competitive- the region’s FDI inflows were, on average, ness to avoid falling further behind compara- between 2.2 and 2.8 percentage points of tor countries in the global marketplace. GDP below that of countries in East Asia. Moreover, countries in South Asia receive Productivity is the key to improved little FDI from within the region. Trade inte- competitiveness gration is also low. From 1990 to 2014, South Asia’s average ratio of exports to GDP What lies behind South Asia’s subdued varied from 17 to 21 percentage points below competitiveness, and what strategies can help that of East Asia, and the average ratio of the region become more competitive? Porter imports to GDP was 21 to 22 percentage (1990) has argued that different countries points lower. have become competitive with different mixes The region has also made little progress in of endowments, factor prices, and policies. diversifying its exports and moving up the Although competitiveness can be buttressed value chain. Although South Asia has had in the short term by keeping costs low, the some success in penetrating new markets, only sustainable path to improved competi- almost 80 percent of the region’s export tiveness in the long term is increased produc- growth from 2001 to 2013 came from sales tivity. Yet South Asia’s growth over the past of the same goods to the same destinations, two decades seems to have been driven mostly and the remaining 20 percent came from by an accumulation of factor quantities selling the same products in new markets. rather than by improvements in their quality, Exports remain highly concentrated in textiles efficiency, or productivity. Accelerating pro- and apparel in Afghanistan, Bangladesh, ductivity growth should be at the top of Nepal, Pakistan, and Sri Lanka; in minerals in policy makers’ agendas in the region to ensure Bhutan; and in animal and vegetable products continued and sustained progress on job cre- in Afghanistan and Maldives. Overall, the ation, growth, poverty reduction, and shared region’s export basket does not reflect a sub- prosperity. stantial transformation of production struc- Across the globe, one mechanism for long- tures or innovative activities. Although the term improvement in productivity has been sophistication of exports has increased in the movement of resources from agriculture India, it has remained low in the rest of South to higher-productivity manufacturing and Asia; and quality (as measured by the prices services. Between 1960 and 2013, the share its products fetch in international markets) of agriculture in South Asia’s GDP decreased has generally remained low and has declined from 44 percent to 19 percent, while the for some countries. share of industry rose from 18 percent to Meanwhile, the global environment is 29 percent. Similar to patterns observed in becoming tougher. The demand for develop- high-income economies, labor productivity ing countries’ exports has been limited by the differentials between agriculture and more slow recovery in the industrial economies and modern activities play an important role by the effect of the decline in commodity in explaining the movement of labor across OVERVIEW 3
sectors, with the sensitivity of labor move- The drivers of and constraints on produc- ment to productivity much higher in South tivity growth were analyzed in four case Asia than in high-income economies. studies of critical industries—agribusiness, However, the movement of labor from agri- electronics, apparel, and automotive. These culture to industry and services in South Asia case studies show the links between the exter- has not been rapid enough to substantially nal environment of firms and their behavior reduce the large differences in productivity within a well-defined industry in which indus- across sectors. In other words, the region has try dynamics (such as competition) can be significant untapped potential, compared to analyzed and performance benchmarked. As the economies of the Organisation for such, they are of great help in understanding Economic Co-operation and Development the relative importance of the external factors (OECD) for example, to reap productivity that drive or constrain a firm’s performance gains by a further reallocation of labor from and productivity. Also, and crucially, industry lower- to higher-productivity activities. case studies assess industry-specific factors Another important mechanism for produc- and policies that traditional cross-cutting tivity growth operates within sectors through analysis is ill-equipped to identify. the movement of resources from less- productive to more-productive firms. In South Business environment challenges Asia, the high dispersion of productivity levels remain a constraint on firm across firms and a strong bias of firm distri- performance bution toward small, inefficient, and slow- growing firms indicate that this mechanism On average, countries in South Asia score has a strong potential to increase efficiency. poorly on major indexes used globally to cap- Firms aged 25 years or more in Bangladesh, ture key aspects of competitiveness, such as Bhutan, India, and Sri Lanka are only the Global Competitiveness Index (GCI) pub- 50 percent to 90 percent larger than start-ups, lished by the World Economic Forum and the while in China, Indonesia, and Vietnam World Bank’s Doing Business report. In the similar firms are two to five times larger. most recent GCI rankings (2015–16), India is According to some estimates, if the distor- the only South Asian country in the top half tions that prevent the reallocation of resources of nearly 140 countries. In the World Bank’s to more-productive firms in India were 2016 Doing Business report, all the South reduced to the levels observed in the United Asian economies, with the exception of States, productivity might increase as much as Bhutan, are ranked in the bottom half. Ten 60 percent. years ago, a World Bank Investment Climate Barriers to the growth of firms can also be Assessment argued that South Asian countries found in government policy. Across the underperform comparator countries on many region, licensing and size restrictions (which investment-climate measures, including infra- have declined in importance but still exist), structure and electricity supply, access to labor regulations that increase the cost of hir- finance, employee skills, and corruption. ing and firing, financial sector regulations Similar results emerge from the most recent that favor small enterprises, and inadequate round of Enterprise Surveys, in which an bankruptcy laws may have limited the ability average firm in South Asia consistently ranks of efficient plants to grow and enabled ineffi- investment-climate constraints as more bind- cient plants to survive. Taxes or labor costs ing than does an average firm in China or that affect larger firms more than smaller Vietnam. Although performance varies sub- firms may reduce the return on investment in stantially across countries and indicators— large firms. Impediments to reaching foreign pointing to significant potential for markets, whether from trade policy or the improvement by leveraging best practices high cost of logistics, can also impede within the region—the overall gap puts South expansion. Asia’s firms at a clear disadvantage compared 4 SOUTH ASIA’S TURN
to similar firms in other parts of the world. channels: localization (firms in the same These challenges may be particularly daunt- industry locate close together to benefit from ing for the region’s high potential firms, which a specialized labor pool) and urbanization would otherwise grow more rapidly and cre- (firms from different industries locate close ate more jobs. together to benefit from a diverse supplier The agribusiness case study shows that network, common infrastructure, or large business environment issues (such as trade number of workers). Unlike in high-income barriers, and restrictions on agricultural mar- countries, firms in South Asia appear to ben- kets, products, and prices as well as blanket efit relatively more from a wide diversity of subsidies) remain the main impediments to workers available in a single location (urban- the inclusive and sustainable development of ization economies) rather than a concentra- the sector. tion of highly specialized workers (localization economies). These results suggest that cities, Growth of the region’s cities with diverse labor pools catering to a range of and clusters offers multiple industries, may currently be more effective opportunities to raise firm vehicles of supporting firm productivity in productivity South Asia than clusters, which cater to a spe- cific sector, although the two are not mutually Economic activity in South Asia is highly exclusive. For example, a number of top firms concentrated. In most countries a small num- in South Asia’s automotive sector fostered ber of districts account for a large share of innovation by locating near their customers, economic activity. In India, for example, the enabling their engineers to work with those of five largest districts account for 18 percent of their clients and gradually building their total employment. However, the degree of capacity from simpler to more complex geographic concentration of manufacturing in components. South Asia has not changed substantially The electronics case study shows that in the last two decades: the share of total South Asia needs urban areas that provide employment held by the top five districts has thick markets for skilled labor, large tracts of remained relatively constant, while which dis- industrial land, and world-class logistics to tricts are in the top five has changed. This become competitive. indicates that more-productive locations have generally not been successful in attracting Increasing prominence of global additional resources at the expense of less- value chains provides a pathway productive locations, although congestion in to greater efficiency the major economic centers has not reached sufficiently high levels to push a substantial Participation in global value chains (GVCs) share of economic activity out to the periph- and exposure to international markets more ery. In the three South Asian countries with generally are associated with higher levels of adequate data, district or state borders tend firm productivity in South Asia. Access to for- to be “thick”—impediments to efficient allo- eign markets, either through trade or the cation of resources among districts are stron- licensing of foreign technology, brings stron- ger than those within districts. ger outcomes for adoption of information Agglomeration economies—the benefits and communication technology (ICT) and that accrue to firms and workers from locat- innovation, and these in turn have a robust ing close together in cities or clusters—matter positive relationship with firm-level produc- for firm productivity: a 10 percent increase tivity. Greater exposure to international trade in district employment leads to a 0.2 to makes firms more viable participants in 0.9 percent increase in the total factor pro- GVCs, which in turn can further enhance ductivity (TFP) of the district’s firms. The productivity in a virtuous cycle. For example, effect operates primarily through two a number of South Asia’s leading firms in the OVERVIEW 5
automotive sector learned by becoming Pakistan’s exports of woven cotton denim domestic suppliers to multinationals entering were destined for Bangladesh and Sri Lanka. the region and then leveraged that experience An East Asia–South Asia regional value chain to access international markets on their own. is also emerging, especially in intermediate Although it is also true that more-productive apparel: 70 percent and 24 percent of South firms may elect to join GVCs, evidence sug- Asia’s imported apparel inputs come from gests that GVC participation and deeper East Asia and South Asia, respectively. South global integration more generally have posi- Asian GVC activity is more integrated with tive productivity effects on firms. East Asia than is that of any other region in With more than 20 percent of its exports the world, except East Asia itself. coming from GVC products, South Asia has Despite the importance of GVCs for firm the second-highest rate of GVC participation productivity and overall export growth, among developing regions. This ranking South Asia lags on many capabilities that largely reflects, however, the region’s strong matter for GVC participation. Countries performance in apparel. Bangladesh has one in South Asia are, on average, more wage of the highest GVC participation rates in competitive and closer to markets than are the world precisely because it exports little members of the Association of Southeast besides garments. India’s participation in Asian Nations (ASEAN) or members of the GVCs is low because it has a more diversi- Southern Africa Customs Union (SACU), fied range of exports. In addition, the loca- but compare unfavorably with them on tion along the value chain of each country’s policy variables such as human capital, firms varies substantially: firms in Pakistan institutions, logistics, and trade barriers on tend to be more upstream, while Sri Lankan imports of intermediate inputs. Bangladesh, and Bangladeshi firms are much further Maldives, and Pakistan charge high tariffs downstream. Final apparel producers in on intermediate apparel goods, and all Sri Lanka and Pakistan have been more suc- countries except Sri Lanka impose high tar- cessful at penetrating higher-income markets iffs on finished automobiles. Nontariff bar- than firms in Bangladesh and India, while riers are also pervasive, particularly in the firms in Pakistan and Bangladesh have automotive sector in Pakistan. Trade facili- shown a greater ability to penetrate high- tation could be substantially improved; the income markets in intermediate apparel. ability to access imported inputs in a timely Overall market sophistication declined, manner is particularly important for sectors however, between 2000 and 2010 in all four in which South Asia has already developed countries, either because of increased sales an advantage (apparel in particular) as well to middle-income markets or because of as sectors of emerging opportunity (such as more intense competition in high-income electronics). Restrictive product market markets, or both. regulations—such as limits on storage, pro- Value chains tend to cluster regionally cessing, and marketing of agricultural pro- because of transport and other transaction duce, price caps and minimum support costs and the need for timely delivery. prices on key agricultural commodities, Bangladesh and Sri Lanka have the highest fragmented approaches to food safety stan- shares of final apparel goods (86 percent and dards and their poor enforcement, and gaps 44 percent of apparel exports, respectively) in in harmonization of local norms with inter- the region and source many apparel inputs national automotive standards—inhibit from Pakistan and India, which focus rela- South Asia’s ability to increase its participa- tively less on final products (18 percent and tion in GVCs in the agribusiness and auto- 6 percent of apparel exports, respectively). In motive sectors. 2013, two-thirds of India’s exports of knit The apparel case study shows how import and crochet fabric were destined for Sri Lanka barriers affect exporters and prevent the and Bangladesh, while nearly half of region from realizing its great potential. 6 SOUTH ASIA’S TURN
Improving firm capability and computers and an Internet connection, which leveraging technology can corresponds with the average for OECD substantially raise firm productivity countries. ICT use in Pakistan is consistent with that of global peers but is very low With the exception of a few global leaders, in Bangladesh and Nepal. Despite wide- firm capabilities in the region tend to be spread Internet use, however, the adoption of limited. On average, South Asia’s firms over- e-commerce and other online business tools is employ relatively scarce capital and under- limited, with the difference particularly stark employ abundant labor. Given the prevailing in India. Size, export status, and, to a lesser wage rates and marginal products of workers, extent, import status are important determi- the optimum level of labor use in Indian and nants of ICT adoption at the firm level. Sri Lankan firms is 1.7 and 2.2 times current Complementary factors—technology and employment levels, respectively, while esti- skills—are also important determinants of mates for Nepal and Pakistan suggest under- ICT adoption. Last, access to finance and to use on the order of 14 to 16 times the existing financial institutions is critical to the adoption workforce. Thus, most firms in South Asia do of e-commerce. The region’s moderate not operate close to optimum efficiency, given achievements on many of these dimensions the prevailing factor prices, bringing down may explain the limited penetration of some aggregate productivity. Potential reasons for technologies and hint at missed opportunities this less-than-rational behavior include lim- to improve productivity performance. ited managerial capacity, labor market rigidi- Patterns of investment in innovation ties (particularly with regard to firing inputs, including ICT, managerial practices, workers), and spatial distortions that prevent and R&D, are reflected in innovation out- firms from locating close to a ready supply of puts. Within the region, close to 80 percent of workers or vice versa. firms in Bangladesh and India engage in tech- Although the region’s top firms make nological innovation, well above the average investment in creating knowledge and provid- in Eastern Europe and Africa. On the other ing skills and training to their workforces a hand, only about 20 percent of firms in Nepal priority, on average firms in South Asia and Pakistan invest in new products or pro- underinvest in knowledge. Overall public and cesses. Moreover, the acquisition of knowl- private investment in R&D in the region is edge capital (such as R&D, investments in low and is increasingly falling behind Latin equipment, and training) is highly concen- America and the Caribbean and East Asia. trated in a few firms, and mature, exporting, Investment in R&D within the region varies and foreign-owned firms tend to be the most greatly, with a higher incidence of firms con- innovative. Compared to other regions, a ducting R&D in Bangladesh and India (above much larger share of innovation in South Asia the rates observed in Africa and Eastern takes place in-house, limiting productive col- Europe and Central Asia) and a much lower laboration across firms and possibly explain- incidence in Nepal and Pakistan (below ing higher rates of imitation instead of radical the rates of Africa and Eastern Europe and innovation. Central Asia). In general, larger firms are Returns to innovation in Bangladesh, more likely to engage in R&D activities. With India, and Nepal are positive and statistically the exception of Bangladesh, access to licenses significant. A 1 percent increase in innovation to use foreign technology increases R&D, intensity increases firm productivity by 0.6 to while financial constraints significantly inhibit 1.4 percent, an impact that is two to five R&D investment. times stronger than the magnitudes com- Adoption rates of ICT also vary across monly estimated for OECD countries. Even the region. Indian firms score very high on among the leaders, however, most innovation multiple dimensions of technology use: consists of the imitation of existing products nearly 100 percent of registered firms have or processes. Few firms engage in disruptive OVERVIEW 7
innovative activities such as introducing new apparel exports to the United States. Given products to the country or to the world. Most the high labor intensity of apparel manufac- of the firms in the region tend to innovate to turing and the large sensitivity of South Asia’s upgrade the quality of their products, labor supply, especially women, to higher although the introduction of new products is wages, a 10 percent price advantage over slightly more frequent in India. And most China in the U.S. market could translate into innovation is done in-house (more so than in employment gains of 8.9 percent in Pakistan— Africa or Eastern Europe and Central Asia), by far the biggest winner— followed which may limit the potential for new by Bangladesh (4.2 percent) and India products. (3.3 percent). These would be well-paying The automotive case study shows how jobs: the wage premium of the apparel sector protections from global good practices (high over agriculture ranges from 8 percent to import tariffs and obsolete safety and emis- 27 percent, depending on the country, and is sion standards) limit the spread of world-class even higher for women. Moreover, jobs cre- firm capabilities. ated in textiles and apparel are particularly likely to attract low-skilled women. These Faster growth of exports and jobs results not only point to the critical impor- is within reach if productivity tance of implementing productivity-enhancing improves measures in the apparel sector but also cau- tion that inaction may lead to a decline in South Asia has tremendous potential to market share. Competitors that have pursued increase incomes and gain market share in more aggressive apparel-friendly policies exports through policies that enhance pro- (such as Vietnam and Cambodia) can stand to ductivity. In a scenario under which produc- gain much more than the South Asian coun- tivity growth contributes about 2 percentage tries in market access. The prospect of new points per year to increases in regional GDP megaregional trade agreements, such as TPP (consistent with South Asia’s best historical and TTIP, for which South Asian countries performance), South Asia becomes the have thus far remained on the sidelines, giving world’s fastest-growing region for exports. By an additional boost to the region’s main 2030, it could more than triple its share of competitors further underscores the urgency global exports of electronics and motor vehi- of reform. cles, and come close to doubling its already significant market share in apparel. Further Policies to boost competitiveness investments in port infrastructure, improve- and productivity ments in customs processes, and behind-the- border services (such as warehousing and In order to realize these gains and more, the transportation); more rapid implementation region’s policy makers should reexamine their of improvements in port-to-port trade and policies and priorities for competitiveness transportation costs; and a reduction in the and productivity. This report highlights not domestic cost of trade could boost export only policies to improve South Asia’s invest- growth by more than 1 percentage point per ment climate but also three policy areas that year over the projected baseline and lead to have so far been less prominent in discussions additional unskilled wage gains of as much as of competitiveness and productivity but 17 percent. that—as shown by the empirical results and Turning to job creation, the report the industry case studies—have the potential examines the intensely competitive apparel to raise productivity across the region. These market. Productivity-enhancing measures that include policies to maximize the benefits of produce a 10 percent cost advantage over agglomeration economies, better connect to Chinese apparel could lead to a 13 percent to GVCs, and boost firm capabilities. In the crit- 25 percent rise in South Asian countries’ ical case of the agribusiness value chains 8 SOUTH ASIA’S TURN
(which constitute one-third of South Asia’s Strengthening the participation of the GDP), additional policy areas include the region’s firms in GVCs calls for taking the spe- need to reform agricultural markets, price cific steps that matter most to global buyers as regulations, product standards, and blanket well as making broad-based investments in subsidies. GVC capabilities. The former include improv- Deriving the maximum productivity bene- ing fundamentals such as cost, quality, and fits from South Asia’s rapid urbanization lead times, particularly relevant to the shift requires policies that leverage agglomeration toward lean retailing and just-in-time delivery economies while minimizing the adverse in many industries. However, other factors are impacts of congestion. The removal of policy- also growing in importance. Buyers who induced distortions that limit the flexibility of attempt to reduce the complexity of their sup- labor, capital, and land markets could enable ply chains increasingly value sellers who offer firms that are more productive to grow. In accompanying services such as input sourcing, particular, policies to increase the flexibility of product development, and financing (known labor markets, especially for women (who as full package services). Buyers also take into face particularly high discrimination in South account social and (to a lesser extent) environ- Asia’s labor markets) are likely to substan- mental compliance, which has become more tially reduce the misallocation of labor and important to their bottom lines because of improve productivity. Policies directed at pressure from corporate social responsibility improving urban governance and bridging the campaigns by nongovernmental organiza- region’s infrastructure gap will ensure that tions, compliance-conscious consumers, and, firms and workers are matched more easily. more recently, the increase of accidents in Achieving these goals will require tackling apparel factories. congestion issues head-on. In particular, Improving broader-based GVC capabilities investments in roads and public transit, provi- requires policy actions such as facilitating sion of quality affordable housing and other imports for use by exporters (for example, basic infrastructure services, and reducing the through better-functioning duty drawback negative social impact of agglomeration (such schemes); reducing average rates of protection as crime) should be high on the policy mak- and harmonizing tariff schedules across inter- ers’ agenda. mediate and final goods; improving standards When large-scale solutions are difficult or and product market regulations; and strength- costly, improved infrastructure could be deliv- ening trade logistics to reduce customs clear- ered through industrial zones or clusters. ance and transit times—all areas where the Although a number of traditional approaches region falls short of its Southeast Asian com- to industrial zones in South Asia have not petitors and global benchmarks. At the firm delivered the expected benefits, there are level, improving firm capabilities to adopt encouraging examples of new approaches new technology (including better managerial from within and outside the region, such practices) and to innovate will be critical to as India’s Scheme for Integrated Textile Parks accelerating the introduction of new prod- and China’s plug-and-play industrial zones. ucts, improving product quality, and moving The location of a cluster often makes all the into higher value segments within existing or difference, and countries in the region could new GVCs. make further efforts to identify and develop For innovation, managerial capabilities, industrial areas close to ports, resolve pend- technology adoption, and worker skills, the ing issues in existing industrial zones, and report’s findings suggest different priorities ensure provisions for worker housing. across the region. In Nepal and Bangladesh, Providing access to R&D and testing facili- the focus should be on efforts to foster the ties, waste dumping, and recycling facilities adoption of the Internet and computers, would make these zones more attractive to which will require overcoming infrastructure small and medium enterprises. challenges as well as improving the provision OVERVIEW 9
of complementary skills such as technological programs that support improving firm capa- training and human capital. In India, where bilities through technology extension, mana- the use of ICT is already widespread, the gerial training, and access to consulting focus should be on improving ICT practices, services, networking, and information can particularly in e-commerce and other online have large and long- lasting productivity ben- business tools. India has a large software efits. In addition, countries with higher inno- development industry and a relatively high vation rates in the region should focus on number of information technology engineers; breaking the pattern of inward innova- access to finance and the establishment of tion development by supporting cooperation broad-based financial transaction platforms among firms and institutions to generate online could be critical to increasing the use novel and, if possible, radical innovations. of the Internet for commercialization. For countries with lower innovation rates, A possible explanation for South Asia’s policy should focus on increasing the number growing gap in R&D investments compared of firms engaged in incremental innovation in to other regions is low returns to R&D in the order to boost productivity, profits, survival absence of complementary factors: manage- rates, and sales growth. rial capabilities, worker skills, and finance. Therefore, investing in these should be a policy priority throughout the region. Reference Modernizing training institutions and Porter, M. 1990. “The Competitive Advantage of expanding access to on-the-job training can Nations.” Harvard Business Review 68 (2): lead to higher efficiency and lower costs, and 73–93.
PART 1
South Asia’s Competitiveness Challenge and Opportunity
The Region’s Competitiveness Potential Remains Largely 1 Unrealized
hich region will become the next Investment Partnership (TTIP)) promise wel- global factory? As the workforce fare gains to members but may divert trade Wages and labor costs rise in China and investment from nonmembers. Against and other East Asian countries, many eyes this background, it has become even more turn to South Asia. South Asia is still largely urgent for countries in South Asia to make rural—agriculture accounts for a large share overdue investments in boosting competi- of employment and a substantial fraction of tiveness and raising productivity to avoid gross domestic product (GDP)—and it has falling further behind comparator countries not been particularly successful in integrat- in the global marketplace. ing economically within itself or with the There are many examples of excellence in rest of the world. Yet more than one million the region. Exports of goods and services are young workers enter the labor market each higher, relative to GDP, than in China. At the month, and by 2030, 26 percent of the sectoral level, the software industry in India, world’s working adults will live in South the garment sector in Bangladesh and Asia. This is the region’s greatest opportu- Sri Lanka, and the Sialkot cluster in Pakistan nity and greatest challenge. are global success stories. And at the firm In the meantime, the global environment level, there is no shortage of global champi- is becoming tougher. The commodity boom ons (for example, US Apparel, Orient Craft, is over, putting the brakes on demand and Pacific Jeans, and MAS in apparel; Tata tightening fiscal belts in resource-rich coun- Motors, Bharat Forge, and Hi-Tech Gear Ltd. tries. Although commodity importers benefit in the automotive sector; Fauji Foundation, from improved terms of trade, many also Dilmah, and KRBL Ltd. in agribusiness; and receive reduced remittances, limiting the ben- Dixon Technologies and Micromax in elec- efits to the current account. Slowing global tronics). Yet so far the region as a whole has growth and an even more pronounced slow- made relatively little progress in integrating down in global trade make it more challeng- economically within itself or with the rest of ing for firms to enter and expand in export the world, diversifying and increasing the markets. New megaregional trade agree- sophistication of its exports, moving up the ments (such as the Trans-Pacific Partnership quality ladder, and improving its ranking on (TPP) and the Trans-Atlantic Trade and many competitiveness benchmarks. It has yet
13 14 SOUTH ASIA’S TURN
to realize the substantial benefits of economic Difficulties in attracting investment integration and achieve its full potential— and penetrating global markets both relative to its endowments and its global competitors—while the window of opportu- Trade and investment integration can increase nity may not remain open for long. The productivity. Opening up local markets to for- following discussion develops this observa- eign trade and investment increases competi- tion in more detail. tion, which encourages labor and capital to move from less-productive to more-productive firms (Melitz 2003). Further, increased com- Pockets of excellence are evidence petition may induce firms to improve their of vast untapped potential efficiency (Helpman and Krugman 1985), to South Asia, and India in particular, is already focus on their core competencies (Bernard, well known for having achieved excellence Redding, and Schott 2011) and reduce mana- and preeminence in the information and com- gerial slack (Hicks 1935), or to invest in new munication technology (ICT) industry. Less technology (Aghion et al. 2005). Finally, well known is that South Asian countries, openness facilitates access to better inputs and locations, and firms have become major play- technology (Atkin, Khandelwal, and Osman ers in important manufacturing industries. 2016), particularly important for those devel- These successes include: oping countries whose import substitution policies previously reduced firms’ ability to UÊ The apparel industries in Sri Lanka and purchase imported inputs. Bangladesh, which are as big (on a per Despite the well-known benefits of eco- capita basis) as those in China and nomic integration, South Asia’s intraregional Vietnam; and global ties are relatively weak. Foreign UÊ The light manufacturing cluster in Sialkot, direct investment (FDI), for example, is low Pakistan, which, despite all odds, has (figure 1.1). Over the period 1990 through achieved a dominant global market share 2014, the region’s FDI inflows were, on aver- in products such as soccer balls and surgi- age, between 2.2 and 2.8 percentage points cal instruments; of GDP below that of countries in East Asia UÊ Indian auto-parts firms, which are becom- (see tables 1A.1 and 1A.2). Particularly in ing global players through exports to and the latter part of this period (2011–14), as acquisitions of firms in leading markets panel b of figure 1.1 shows, FDI inflows in such as Germany; all countries (except Maldives) were substan- UÊ Global electronics and auto-parts firms tially below the average of countries at simi- that have established their global research lar levels of development. and development (R&D) centers in India; The flow of intraregional investment in and South Asia shows the lack of regional invest- UÊ Leading agribusiness firms developing, in ment integration. Globally, South–South partnership with governments, new crop cross-border investment has increased, and a varieties for the domestic and interna- recent market survey showed that multina- tional markets (such as tea in Sri Lanka tionals in other regions allocate a significant and rice and mint oil in India). share of outward investment to countries These cases were selected for the in-depth within their region. By contrast, countries in case studies in this report to analyze the driv- South Asia receive little FDI from within the ers of competitiveness as well as the con- region. In particular, despite the lower trans- straints that limit the competitiveness of firms, action costs of investing in nearby, familiar locations, industries, and countries at a time markets, Indian multinationals tend to invest when South Asia is uniquely positioned to outside the region (Gomez-Mera et al. 2015). take advantage of rising production costs in Some of the blame for low FDI inflows East Asia. can be traced to burdensome regulations THE REGION’S COMPETITIVENESS POTENTIAL REMAINS LARGELY UNREALIZED 15
FIGURE 1.1 Countries in South Asia attract less foreign direct investment than global peers
a. 1990–94 b. 2011–14 20 20
15 15 MDV 10 (%) (%) 10 GDP GDP 5 I to I to 5 D D LKA F F 0 NPL IND PAK BGD BGD IND LKA 0 AFG PAK –5 NPL
–10 –5 67891011 68910117 Log of GDP per capita (PPP) Log of GDP per capita (PPP)
Source: World Bank calculations based on World Development Indicators database. Note: FDI = foreign direct investment; PPP = purchasing power parity. Shaded area corresponds to the 95 percent confidence interval.
governing FDI. Nepal, for example, has highly developed tourism export sector, being failed to attract substantial FDI because of again the exception).1 Countries in the region the complicated process required to repatri- have become more integrated in the global ate profits, high entry barriers (with a long marketplace, however: the region’s share of negative list), and insufficient guarantees of merchandise trade increased from $1.15 of investor protection. The lack of readily avail- every $100 traded globally in 2000–04 to able land with adequate access to infrastruc- $1.82 in 2014, an increase of nearly 60 ture services has constrained foreign percent. Nevertheless, growth in the region investment, particularly in Bangladesh. For has been more inward-oriented than that in example, in 2011, Samsung’s large intended East Asia. investment in electronics in Bangladesh fell South Asian countries can be divided into through because adequate land was not two groups by their performance in merchan- available in an export processing zone. By dise exports. Although exports increased at contrast, Indian states compete for invest- double-digit rates from 2000 through 2013 in ment by major original equipment manufac- Afghanistan, Bangladesh, Bhutan, and India turers (OEMs) by offering land combined (15, 13, 16, and 14 percent per year, respec- with tax incentives, although there is some tively), growth was slower in Maldives, risk that such competition will lead to sub- Pakistan, and Sri Lanka (2, 8, and 5 percent optimal investment locations and industry per year, respectively) and exports fell fragmentation. 1 percent per year in Nepal (see table 1.1). Trade integration is also low. As shown in Regardless of whether countries belong to figure 1.2, from 1990 through 2014, South slow- or fast-growing groups, however, their Asia’s average ratio of exports to GDP varied share of the global merchandise export mar- between 17 and 21 percentage points below ket remains small. For example, India, with that of East Asia, and the average ratio of an 80 percent increase in its market share in imports to GDP was 21 to 22 percentage the past decade and a half, has just reached points lower than the countries of East Asia 1.5 percent of the global exports market. and the Pacific Region (Maldives, with a Bangladesh, with somewhat slower growth of 16 SOUTH ASIA’S TURN
FIGURE 1.2 South Asia’s trade integration is relatively low
a. Before 1990–95 b. Before 2011–14 250 250
200 200 MDV (%) (%) 150 150 GDP GDP
100 100 Trade to Trade to LKA LKA 50 50 AFG IND NPL BGD PAK NPL PAK BGD IND 0 0 67891011 68910117 Log of GDP per capita (PPP) Log of GDP per capita (PPP)
Source: World Bank calculations based on World Development Indicators database. Note: PPP = purchasing power parity. Shaded area corresponds to the 95 percent confidence interval.
TABLE 1.1 Despite rapid growth, South Asia’s share of global merchandise exports remains low
Afghanistan Bangladesh Bhutan India Maldives Nepal Pakistan Sri Lanka South Asia 2000–04 0.002 0.107 0.001 0.814 0.003 0.010 0.139 0.074 1.149 2010–14 0.004 0.162 0.001 1.463 0.001 0.004 0.135 0.057 1.827 Absolute growth 0.002 0.055 0.001 0.649 −0.002 −0.005 −0.005 −0.017 0.678 Percent growth 121 52 112 80 −61 −54 −3 −23 59
Source: World Bank calculations based on UN COMTRADE database. Note: Cell values in the first two rows indicate each country’s share in global merchandise exports, expressed in percentage points. ellC values in the third row indicate the absolute change in market share, expressed in percentage points (difference between the first two rows). Cell values in the fourth row indicate the percent change in market share.
50 percent in its market share, has passed 1 percentage point and almost fully explained Pakistan to become the second-largest mer- by increased exports from India. chandise exporter in South Asia, with nearly In services, the region has recorded much 0.2 percent of the global merchandise market. better performance. Overall, South Asia’s Afghanistan and Bhutan more than doubled share of global services exports rose from their global market share in the past decade 0.9 percent in the early 1990s to 3.6 percent and a half, but still account for less than in the early 2010s. Every country in the region 0.01 percent of the world market, combined. except Nepal increased its share of global ser- The textiles and apparel sector is one vices exports; India and, at a substantially exception to this general trend (figure 1.3). lower scale, Maldives more than doubled South Asia’s share of global exports in gar- their shares of services exports (table 1.2). ments rose from 7.4 percent in 2000–04 to Limited trade integration results from high 11.6 percent in 2010–14. More than half of trade costs. For example, Nepal charges high that increase is accounted for by Bangladesh tariffs on yarn, a key input for most of its (due in particular to effective import facilities apparel exports. In India, high tariffs on for exporters), about 40 percent by India, and inputs particularly affect the electronics sec- the remainder by Pakistan. Gains in market tor, while tariffs on manmade fibers (com- share in other sectors are almost all below bined with the problem of duty drawback THE REGION’S COMPETITIVENESS POTENTIAL REMAINS LARGELY UNREALIZED 17
FIGURE 1.3 Outside of garments, South Asia’s market share growth has been low
5.0 4.5 4.0 3.5 3.0 2.5 2.0 1.5 1.0 0.5 0 hange in export market share by country (%) C –0.5
ood etals nimal A inerals W M egetables M hemicals Footwear V Foodstuffs C Hides and skins Stone and glass Plastics and rubbers Textiles and apparel Transport equipment achinery and equipment M Afghanistan Bangladesh Bhutan India Sri Lanka Maldives Nepal Pakistan
Source: World Bank calculations based on UN COMTRADE database.
TABLE 1.2 South Asia has more than tripled its market share in services since the 1990s
Afghanistan Bangladesh India Maldives Nepal Pakistan Sri Lanka South Asia 1990–94 — 0.05 0.54 0.01 0.03 0.17 0.06 0.87 2000–04 — 0.05 1.27 0.02 0.02 0.12 0.07 1.56 2010–13 0.08 0.06 3.16 0.05 0.02 0.13 0.08 3.58 Absolute growth — 0.01 1.89 0.02 0.00 0.01 0.01 2.02 Percent growth — 26 149 107 −13 10 8 129
Source: UN COMTRADE database. Note: — = Not available. Cell values in the first three rows indicate each country’s share in global merchandise exports, expressed in percentage points. Cell values in the fourth row indicate the absolute change in market share, expressed in percentage points (difference between rows 1 and 3). Cell values in the fifth row indicate the percent change in market share.
schemes for exporters) limit exporters largely Portugal, and Shahid 2015). In Bangladesh, to garments made of domestic cotton. Cotton tariffs have declined and paratariffs have is grown and harvested predominantly during increased, making the latter the more impor- the summer, which reduces capacity utiliza- tant constraint on imports (Kathuria and tion. In addition to high tariff rates, South Malouche 2016). In contrast, Sri Lanka’s Asian countries impose high paratariffs, an garment exporters have benefited greatly extra tax on imported products that is typi- from zero tariffs on textile imports. In the ser- cally complicated, subject to arbitrary vices trades, Bangladesh, India, Nepal, and enforcement, and applied irrespective of trade Sri Lanka are substantially more restrictive preferences. The addition of paratariffs brings than high-income economies, and even China, the average import tax rates in Bangladesh according to the World Bank’s Services Trade and Sri Lanka to a level more than double Restrictiveness Index,2 although Pakistan’s that of the customs duty alone (Kathuria, service restrictiveness is relatively low. 18 SOUTH ASIA’S TURN
Steps already taken to reduce barriers to imported intermediate inputs (Goldberg et al. trade and investment in the region have paid 2010). Improved services policies after 1991 substantial productivity dividends and hint at also boosted Indian firms’ productivity. potential future benefits from further policy Policy changes that facilitated the operations efforts to improve integration. In India, for of foreign services firms—particularly in example, the reduction in tariffs on auto banking, telecommunications, insurance, and parts and electronics greatly increased com- transport—increased the productivity of for- petition in the domestic market, raising stan- eign and local manufacturing firms that used dards among firms and enabling them to those services (Arnold et al. 2015). increase productivity further by working with demanding clients. More broadly, the sharp Little progress in diversifying the fall in the level and dispersion of tariffs in merchandise export basket response to the 1991 balance of payments crisis induced firms to improve their effi- The composition of the export basket from ciency and improved their access to imported South Asian countries has changed little in inputs (Topalova and Khandelwal 2011).3 15 years, which shows limited product inno- The productivity benefits of reform were vation. Exports remain highly concentrated in smallest in sectors in which burdensome reg- textiles and apparel in Afghanistan, ulations limited firms’ ability to adopt new Bangladesh, Nepal, Pakistan, and Sri Lanka; technologies,4 greater for domestic than for- in minerals in Bhutan; and in animal and eign firms (probably because foreign firms vegetable products in Afghanistan and had already been exposed to competition), Maldives. Almost 80 percent of the region’s and largest in industries that also experienced export growth from 2001 to 2013 came from the most deregulation and FDI liberalization. the intensive margin: sale of the same set of About a third of the rise in firms’ product goods to the same destinations (figure 1.4, diversification was caused by increased access panel a). The remaining 20 percent came to better quality and a higher variety of from the extensive margin, but almost entirely
FIGURE 1.4 Most export growth in South Asia has taken place along the intensive margin
a. Export growth decomposition, 2001–13 b. Destinations, 2000–13 90 160 80 140 70 120 60 100 50 80
ercent 40 reached P 60 30 40 20 umber of export markets N 10 20 0 0 K A G U IND LK NPL F H PA BGD A B MDV
et intensive N ld products in 2000 2001 2002 2003 2004 2005 2006 ew products in ew productsold markets in O N new marketsN new markets 2007 2008 2009 2010 2011 2012 2013 Net extensive
(continues next page) THE REGION’S COMPETITIVENESS POTENTIAL REMAINS LARGELY UNREALIZED 19
FIGURE 1.4 Most export growth in South Asia has taken place along the intensive margin (continued)
d. Share of export revenues accounted for by top five c. Products, 2000–13 destinations, 2000–13
5,000 100 4,500 90 4,000 80 3,500 70 3,000 60 2,500 50 ercent P exported 2,000 40 1,500 30 umber of product varieties N 1,000 20 500 10 30 0 K A G U G U K A IND LK NPL F H F H IND NPL LK PA BGD A B MDV A B BGD PA MDV 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013
Source: World Bank calculations based on UN COMTRADE database. by selling the same set of goods to new leapfrogged both Vietnam and Indonesia on markets. Although the number of product this metric between 2000 and 2014, sophisti- varieties exported increased in all countries cation did not increase in Bangladesh and Sri except Nepal (figure 1.4, panel c), diversifica- Lanka, rose steadily but from a low level in tion into new products (either in old or new Pakistan, and declined in Bhutan (figure 1.5, markets) accounted, on average, for only 0.07 panel a). Even in India, one measure of percent of export growth. In some countries, export quality and sophistication, PRODY,6 most exports go to only a few destinations. has remained low. A recent International For example, the top five export destinations Monetary Fund (IMF) study finds that the account for 97 percent and 70 percent of average level of sophistication for India’s export revenues in Bhutan and Maldives, manufacturing exports is lower than that of respectively (figure 1.4, panel d). In contrast, the rest of Asia, in sharp contrast to India’s India’s top five markets purchase only 36 performance in the services sector (IMF percent of the country’s exports (figure 1.4 2015). The average sophistication of the panel d). countries that purchase exports from South Asia, as measured by the weighted average of Elusive sophistication and low the buyers’ per capita incomes, has declined quality of exports over time (figure 1.5, panel b), both because South Asia is moving toward relatively On average, the sophistication of merchan- poorer trading partners (figure 1.5, panel c) dise exports from South Asia (as measured and because its existing trading partners are by the Export Sophistication Index [EXPY] growing less rapidly than average (figure 1.5, indicator) is higher than expected given the panel d). region’s income level.5 With the exception of Quality, as measured by the price that India, however, countries in the region have exporters of a particular product fetch in not been successful in further increasing international markets compared with other export sophistication. Although India producers, is generally low and has declined 20 SOUTH ASIA’S TURN
FIGURE 1.5 Average export sophistication in South Asia has been on the decline
a. Export sophistication b. Weighted average of income of importers 1.4 25,000 CHN2014 BTN1990 BTN2000 1.2 CHN2000 20,000 IND2014 IDN2000 VNM2014 1.0 CHN1990 BTN2014 IDN2014 IND2000 IDN1990 15,000 VNM2000 0.8 NPL2014 MDV2014 IND1990 LKA2014 PAK2014 ophistication Index)
S NPL2000 LKA2000 0.6 10,000 VNM1990 PAK2000 median importer
xport LKA1990 0.4 E BGD2014 ( 5,000 PAK1990 PY AFG2014 Income of importers relative to X BGD2000 0.2 E BGD1990 NPL1990 0 0 0 5,000 10,000 15,000 20,000 25,000 hutan India epal anka B aldives N akistan L GDP Per Capita (PPP, constant prices) M P Sri fghanistanangladesh A B Note: PPP = purchasing power parity. 2000–04 2010–14
c. Change due to trading partners‘ incomes d. Change due to trade shares 2.0 1.4
1.2 1.5 1.0
0.8 1.0 0.6 median importer median importer 0.4 0.5 Income of importers relative to Income of importers relative to 0.2
0 0
hutan India epal Lanka hutan India epal Lanka B aldives N akistan ri B aldives N akistan ri M P S M P S fghanistanangladesh fghanistanangladesh A B A B 2000–04 2010–14 2000–04 2010–14
Note: Trade weights are fixed at the 2000–04 level. Note: Partner incomes are fixed at the 2000–04 level. Source: World Bank calculations based on UN COMTRADE database.
for some countries.7 For example, some Sri exports are in the lowest tenth (figure 1.6, Lankan apparel (such as brassieres) was at panel f). Pakistani cotton moved from the the higher end of the price spectrum at the bottom quartile to the second quartile of the turn of the century, but now fetches prices in distribution of prices, advancing six positions the lower fifth of the distribution ( figure 1.6, in the ranking (figure 1.6, panel g); trousers, panel i). In tea, for which the country has a however, are sold at the very bottom of the built-in brand name, Sri Lankan exporters distribution (figure 1.6, panel h). Key secure prices just above the median Bangladesh exports have lost ground along (figure 1.6, panel j). Nepali carpets are in the quality ladder, moving six positions down the middle of the distribution of prices the quality ladder in cotton t-shirts (figure 1.6, panel e), while mineral water between 2000 and 2013 (figure 1.6, panel b). THE REGION’S COMPETITIVENESS POTENTIAL REMAINS LARGELY UNREALIZED 21
FIGURE 1.6 Export quality is generally low and has declined for some countries
a. Afghanistan, figs, 2000 & 2013 2.5 RA F PER
2.0 F A Z X ME SR I BEL L H C ARG PER G USA
1.5 F A BRA BRA N SAU H N UN C H T AFG C X PN J ME SGP DEU K PA NLD T EGY GRC AU K GBR DN K elative unit value SYR 1.0 E SV H RA R C F UN A T NLD IT UR T USA T DEU AU EGY GBR T GRC RN I PR ESP UR T T RN G I PR F A A IT ESP SYR K F
A AFG PA 0.5 Z ARE SAU
0 10 20 30 Rank
Quality ladder (2000) Afghanistan position in 2000 Quality ladder (2013) Afghanistan position in 2013
b. Bangladesh, cotton t-shirts, 2000 & 2013
2.5
ANT
MNP
NZL
KWT
ITA
2.0 KAZ
MRT
VEN
FJI
COL
ISR
AUS
ARG
ESP
FIN
NZL
VNM
1.5 FRA
AUS
ERI CAN
OAS
BRB
FRA
GBR
FIN
CZE
DNK
ARG
SGP
PRT
CRI SWE NLD
TTO
PER
CHL
EST
USA GRC NOR CHE
UMI MYS SWE IDN EST PER NLD DEU CAN GRC PLW FJI ISR BOL ECU LKA SVN IRL GBR LUX NOR BEL SGP PRT
1.0 PHL OAS FRE IRN UKR MAC PSE ZAF MOZ ZMB TUR
DNK ECU LSO SLE TTO ZAF elative unit value TGO SAU MMR HKG SWZ
BGD DEU
ARE HKG HUN
LVA
MDG
SYR
TUN AUT MAC ROM ETH MKD BRA BGR R
LTU KOR BIH HRV ARM PHL MYS LAO PAN SVK ESP NPL POL COL TUR SHN THA BEL MDA IRL LBN BOL JOR MLT FRE MUS SVN TZA UGA BWA SER GEO NAM GRD VNM CHN KHM IDN MUS MAR IND EGY AUT HRV RUS LKA PRY IND BRN KWI BIH MKD BRN LTU CZE UKR POL LVA LAO KEN KOR MDV BGR MWT KHM HUN JOR MOZ THA SVK CHL SAU LBN MDA BGD NIC MEX RUS NPL GRD CIV TUN EGY MMR MDG MNG GTM USA BLR MAR SYR PAK OMN ROM ARM DOM ALB BEN HTI BWA TZA SLV BRA HND BGD ARE
0.5 CHN LSO LCA QAT CPV KEN SWZ PAK GTM ALB BLR MEX
UZB HND BGD NIC HTI SLV
0 50 100 150 Rank
Quality ladder (2000) Bangladesh position in 2000 Quality ladder (2013) Bangladesh position in 2013
(continues next page) 22 SOUTH ASIA’S TURN
FIGURE 1.6 Export quality is generally low and has declined for some countries (continued)
c. India, cars of 1000−1500cc, 2000 & 2013 2.0 PRT
1.5 MAC MKD LVA CAN SER CHL DEU SLE AUS BEL MEX MEX USA IDN PHL MNT LTU GBR ARG VEN NLD BEL GBR BGR ARG TJK ALB DEU CHN SVK CZE GTM FRA HRV JPN OAS SLE URY MKD ESP TUR LUX DNK ZAF AUT
1.0 IND HUN SVN IRL ITA ITA POL MYS SVK UZB UGA THA ARE HUN TUR BRA PAN CYP NZL KEN CZE OAS NLD JPN COL DNK FRA ZAF ECU ESP SVN BIH HKG TUN SWE MAR FIN LUX KOR CHN ASM ISR SEN IDN AUT BLR CHL elative unit value CAN ROM SAU MLT LSO PRK NAM MYS RUS TGO BEN HRV SGP R UZB LKA THA GRC PRT SER AUS IND UKR SAU POL CHE RUS HKG KOR NCL URY GRC IND CHE LVA UKR IRN EST ISL BRA 0.5 USA IND PRK FIN LTU ARE COL FRE
0 20 40 60 80 Rank Quality ladder (2000) India position in 2000 Quality ladder (2013) India position in 2013
d. Maldives, tilapia, 2000 & 2013 2.0 L H P V AUS I C L H P POL PN J C I UN OAS N
H V 1.5 AUS VNM OR MDV
K MD OR L K T Z N ES A FJI K L MDV MDV F X MUS PY MYS T ME SL T I MR AU V GRD RL I I SYC C SLV SEN T L H ESP M ECU DN WE C MR I T OMN MDG BEL Z G YEM T I T CAN ND DN SYC I GRC ES H ER I AU X I NOR RA RA N T CR F N UR PAN F
1.0 A ME SL GRL FI UM I T A H SEN IT LVA K ML C NLD L L ROM T Z FJI H COL A C E PR GBR WE K UR GRC E Z X Z T L USA Z ESP C CAN RUS UGA Z C BEL LU elative unit value ND I POL NOR ARG GBR RE N EN BRA K SUR F RUS USA K U R SUR RO A O T F DN C A L DEU SWE IT I K TT NLD N TZ I DN ND DEU SWE COL PER RO H CR BGR F PER GUY F NAM RL A I Z F PAN SLV ECU A RV Z ARG A BRA H SAU EN K UGA A TZ
0.5 GRL TH T CYP A A Z PR N L D H TH H BLR C M NAM ARE VNM OAS SGP
0 20 40 60 80 100 Rank Quality ladder (2000) Maldives position in 2000 Quality ladder (2013) Maldives position in 2013
(continues next page) THE REGION’S COMPETITIVENESS POTENTIAL REMAINS LARGELY UNREALIZED 23
FIGURE 1.6 Export quality is generally low and has declined for some countries (continued)
e. Nepal, carpets, 2000 & 2013
2.5 BEL GRC UN T LBN ARE
2.0 POL A IT G HK NOR EGY ARG USA SWE L UR T TK ARM GBR RUS E
1.5 Z SR C I T A PR NPL ROM NLD TH BGD G R K PAN HK U NPL A CAN EGY TH E K DEU UR E E H F G T M Z H C F A T DN K GRC A C
1.0 Z A VNM TK K SGP RN PA AU elative unit value I PER MAR T PA N R H N AU NLD C RA ND H I F RN C ARM I G GBR UN F DEU MAR T RL NOR A I M A NPL SWE ROM BEL IT TK RE F ESP T AUS K PR OR CYP K DN ND VNM F I SGP NPL A
0.5 Z ESP AUS DN I
0 1020 30 40 50 Rank Quality ladder (2000) Nepal position in 2000 Quality ladder (2013) Nepal position in 2013
f. Nepal, mineral water, 2000 & 2013 2.5 NLD
2.0 Z A K K DN EGY T E CAN H ML C
1.5 NLD AM E J H C EN K PN J N NOR USA T B ECU SWE SR I PNG GBR AUS VEN DEU SL T I A GEO LBN H ECU PR B L G A Z SR X Z T I SGP N T Z NAM U PER N BRA R A FI D W 1.0 LU L K Z T AU O K T IT GBR elative unit value H BUN PSE U G RA A USA AND P F TT K ML N DOM SAU VC BEL ARM CAN TH N ND NPL R FI RE A I C I F H ND K ARE K I BRB EN ESP M N C L COL H SLV BRB E GO K T X K OR SV RUS MAC T CR Z DN G G MDA K K I K T GRC C F UN ME UR DN NGA OR T A Z PR PAN ES H CYP J HK T SV VNM E RV Z LVA MYS Z OAS PN H SVN AU J A ARG WE A MO HTI PER BGR L Z POL NPL R TZ H K URY D UN SER H C GRD Z K T B I U PA A IH ROM M T BLR ALB UGA ND B X IT L SW L CR O H F GRC Z ND NA
0.5 I BRN LU A TT N K AM Z J MAC L A L SGP A K OR H AUS H L K P OAS C BRA LAO TH Z PAN X OMN T D K G G SLV M K PRY EGY VEN RN K T RV SAU I T IH ME N UR VC DOM BWA LBN PA U SVN UN SYR ARE HK DN NOR M VNM H POL OR ABW B T I R T W G MYS CYP T T COL H ARM J RUS LVA ESP UN ROM K K L C ES NGA H U MDA GO NPL SER ARG T BGR GEO MUS MAR NPL 0 50 100 Rank Quality ladder (2006) Nepal position in 2006 Quality ladder (2013) Nepal position in 2013
(continues next page) 24 SOUTH ASIA’S TURN
FIGURE 1.6 Export quality is generally low and has declined for some countries (continued)
g. Pakistan, cotton, 2000 & 2013 RA 2.5 F
SR COL I A
IT KOR
2.0 LVA 1.5 BRA PN J SVN ECU T AU elative unit value PER
R 1.0 N UN COL EGY H OR H C BLR E K CAN H T DEU C X PR X G RUS N ME M UN PER SLV T H ME HK E H R C ECU G Z K MYS K H C B ESP OR SV BEL PA I SLV J GRC SGP BGR A U VNM T NLD ML POL ARE RE TZ SER Z E UR A L DN BEN EGY A F I EN WE NLD U G Z T IH ARG UN IT GBR RE BRA K Z T SYR C K B T SW TH WE UR F L SVN MO BOL T Z OR HK I A T Z BEL J A L RA OAS A PA LVA F ND DN R
0.5 BOL F PR IH H I K ARG CR I TZ MUS M B ND ND B USA P T H LSO DEU MAR I A POL ESP H RN B TH D MB TK I B USA E K Z AU Z TH SYR V F I NGA M U GRC PRY A C AUS A MYS Z K OAS B K RUS AUS MAC TJK L RN Z I SGP BGD M U PA RL I NGA E SWE T TK Z BGR A SDN VNM ES K A F PA B SAU ROM TJK 0 20 40 60 80
Rank Quality ladder (2000) Pakistan position in 2000 Quality ladder (2013) Pakistan position in 2013
h. Pakistan, trousers, 2000 & 2013
2.5 LBN
QAT
AND 2.0 CYP ITA ML CAN T
BUN
DN 1.5 C H K F IRL E L RA FI RO T N U M N Z MA NO L C ES Z NLD T C R MR TK E K T O L ES R C IT GBR H A P AU BEL L AR J NLD S PN M PA G CAN LVA H SGP DN RV K N NO SW D N RU 1.0 US T PR ES UN K R Z PR F T elative unit value RA FI L S E T A DE C T N I H FJ S GRC U R R PO L HK I MAC BG U L SW GB K G LVA T R E DEUARE SE U R GRC H MU AU PE R R BR UN R PE OA T R MU BL MA A S H MYSP Z R SVN RV H G CPV BE S I ML PR L S MY R CO SG BR D PAK B T L K Q N K MO AU BLR I M H T T Z R I TH T OR A A H ME SV US U UN T L P F SER G A N S RO VNMPA TK BR P CR O Z T L T A N BG PA RE F G MMI U K A Z H NAMDO J MD PO C HK ND T N X MDG RUS L FJ M Z T M MN Z B MW M M N I BG O U R H K K L OAS F E M OMN MO G LU K CUB T W H D LAO U RE R L R 0.5 KH MDA A K C N A B I CY R D M V MAR TT LBN ARE R C ES GO AR BO AN K EGY N F EN I D G E I B MDAARM R SYR BW V H Z X P O IC G PR LSO ALB Z IH P M Z G L T A SV N WE CO A Z SYR MM Y F SAU E TH K U E HTI SLV LAO L NPL MDG T Z LU I H A B L ND R SW PR K VNM X
EGY A Y I DN Z PA DOM TJK KH HTI TK C I J RL H ME OR K M M SVN N N BGD BOL H X LSO I ND C PAK 0 50 100 150 Rank
Quality ladder (2000) Pakistan position in 2000 Quality ladder (2013) Pakistan position in 2013
(continues next page) THE REGION’S COMPETITIVENESS POTENTIAL REMAINS LARGELY UNREALIZED 25
FIGURE 1.6 Export quality is generally low and has declined for some countries (continued)
i. Sri Lanka, brassieres, 2000 & 2013 2.5 FRA
JPN 2.0 PRT DMA T ES T BEL PR RA N 1.5 F T FI K MAR AU DN DEU UN PN H J R K A U GBR IT UN SWE MMR T SWE LAO BGR AUS URY A ROM IH MAR K K L B E LVA BRA DN SER RV I Z BLR C H BRB E
1.0 A GEO MUS CR NOR H IT U GBR elative unit value A K X C T ND RL DEU I L I TH VNM ND PA COL LU R VNM I POL ESP MAC UN CAN NOR GRC M BGR H A C SR A ESP CAN MDV M I I T SVN USA K OR K SVN X ARG DOM UR OAS N J PER BRN G SLV L E RUS L L LBN ND T T KH SR Z K I H ALB H MYS ME LAO DN C N UN ES P I RV N G SV T OR FI H GRC NLD H MDG K BGD K ARG C HK T A G L LSO RL SV MYS URY I BLR F RUS LVA H VEN I USA COL PER ML A TH SGP C C AUS A I ND EGY X CR K BRA OR SLV L N H SER DOM POL U MMR K M LBN L R ARM T G GUY 0.5 ME T ALB L K H ARE G P U HK SGP DN IH N I X B H LU C ROM FJI UR MAC OAS T F MNG A WE BGD EGY Z Z PAN
0 20 40 60 80 Rank Quality ladder (2000) Sri Lanka position in 2000 Quality ladder (2013) Sri Lanka position in 2013
j. Sri Lanka, tea, 2000 & 2013 2.5 ARE SAU AUS RUS PN J CAN G HK USA 2.0 U T MYS L OAS RL I PAN ARE U T RA OAS L F DEU POL YEM RUS 1.5 PRY A OR Z K K A L K T E A SPE Z Q C LBN E NPL Z K A DEU PA G USA N A I K HK G E L H C N H A C NLD 1.0 GBR TH F BRA GBR BEL A elative unit value NLD ND I PN Z J R SVN EN E E K EN UN M Z H LSO K T A A C MAC F TK I N A I A TZ MYS H Z SGP BD RWA ND A TH C I POL K MW L WE Z RE Z BGD UGA COG UGA F I PNG RWA MO ECU A DN TH VNM BD CAN I E SGP NPL I UR A ARG TZ T K L
0.5 GEO TH W RN MW UR I E Z BGD T BRA DN I ECU PNG VNM ARG GEO RN I
0 20 40 60 Rank Quality ladder (2000) Sri Lanka position in 2000 Quality ladder (2013) Sri Lanka position in 2013
Source: World Bank calculations based on UN COMTRADE database.
Indian cars and cell phones also secure low 50 percent more than the price received by prices relative to those of competitors. There the median exporter of the same product are some exceptions in agricultural products, (figure 1.6, panel d). Similarly, figs from however. In Maldives some exports of fish Afghanistan have substantially moved up the are priced at the high end of the market— quality ladder between 2000 and 2013 tilapia exporters, for example, receive (figure 1.6, panel a). 26 SOUTH ASIA’S TURN
Annex 1A
TABLE 1A.1 Per capita income and ratios of FDI, imports, and exports to GDP by region, 1990–2014
Variables FDI/GDP FDI/GDP Imports/GDP Imports/GDP Exports/GDP Exports/GDP GDP per capita (PPP, 6.91e-05*** 0.000199*** 0.000750*** constant 2005) (1.07e-05) (2.92e-05) (2.51e-05) Europe and 0.705 0.0880 −11.63*** −12.98*** −8.802*** −13.82*** Central Asia (0.521) (0.533) (1.425) (1.432) (1.348) (1.230) Latin America and 0.622 0.985* −15.10*** −12.91*** −14.36*** −9.476*** the Caribbean (0.552) (0.562) (1.526) (1.541) (1.443) (1.324) Middle East and North −1.688*** −2.729*** −16.64*** −18.68*** −9.538*** −17.09*** Africa (0.652) (0.694) (1.820) (1.878) (1.721) (1.613) North America −2.259 −4.128*** −35.99*** −40.89*** −29.72*** −48.19*** (1.485) (1.520) (4.119) (4.159) (3.895) (3.573) South Asia −2.855*** −2.266** −21.09*** −21.21*** −23.57*** −17.62*** (0.858) (0.893) (2.325) (2.396) (2.199) (2.059) Sub-Saharan Africa 0.178 0.881* −14.41*** −11.90*** −20.93*** −11.50*** (0.516) (0.530) (1.425) (1.464) (1.348) (1.258) Constant 2.238** 1.368 54.82*** 51.91*** 47.44*** 37.05*** Year dummies Yes Yes Yes Yes Yes Yes Observations 4,062 3,979 4,154 4,075 4,154 4,075 R-squared 0.034 0.043 0.055 0.069 0.090 0.260
Source: World Bank calculations based on World Development Indicators database. Note: Standard errors in parentheses. PPP = purchasing power parity. *** p < 0.01, ** p < 0.05, * p < 0.1.
TABLE 1A.2 Changes in FDI, imports to GDP, and exports to GDP by region, 1990–2014
ΔImports/ ΔImports/ ΔExports/ ΔExports/ Variables ΔFDI/GDP ΔFDI/GDP GDP GDP GDP GDP GDP per capita (PPP, 5.29e-05*** 0.000216*** 0.000833*** constant 2005) (1.21e-05) (4.29e-05) (4.04e-05) Europe and Central 1.888*** 1.428** −3.495* −4.859** −2.665 −7.906*** Asia (0.586) (0.599) (2.093) (2.105) (2.083) (1.984) Latin America and the 0.684 0.860 −12.61*** −10.71*** −11.97*** −6.366*** Caribbean (0.618) (0.630) (2.238) (2.256) (2.227) (2.126) (continues next page) THE REGION’S COMPETITIVENESS POTENTIAL REMAINS LARGELY UNREALIZED 27
TABLE 1A.2 FDI, imports to GDP, and exports to GDP by region, 1990–2014 (continued)
ΔImports/ ΔImports/ ΔExports/ ΔExports/ Variables ΔFDI/GDP ΔFDI/GDP GDP GDP GDP GDP Middle East and North −0.397 −0.906 −5.583** −5.603** −0.998 −7.491*** Africa (0.731) (0.771) (2.641) (2.714) (2.629) (2.557) North America 0.973 −0.450 −5.025 −10.26* −3.048 −23.23*** (1.644) (1.685) (5.925) (5.999) (5.896) (5.653) South Asia −0.669 −0.366 −4.277 −4.890 −16.20*** −10.54*** (0.998) (1.041) (3.475) (3.598) (3.458) (3.391) Sub-Saharan Africa 1.660*** 2.196*** −0.633 2.479 −10.02*** 0.495 (0.580) (0.599) (2.102) (2.175) (2.092) (2.050) Constant −1.592 −2.283* 4.477 1.022 6.143 −6.049* Year dummies Yes Yes Yes Yes Yes Yes Observations 3,892 3,811 3,896 3,830 3,896 3,830 R-squared 0.007 0.012 0.012 0.018 0.016 0.119
Source: World Bank calculations based on World Development Indicators database. Note: Standard errors in parentheses. PPP = purchasing power parity *** p < 0.01, ** p < 0.05, * p < 0.1.
FIGURE 1A.1 Export Orientation Index—Rankings for South Asia region countries and comparators VNM 60 MDV
40 L H P DN I 20 ND I N H C rientation Index O BGD K 0 PA xport E NPL N T G B F A –20
–40
Source: World Bank calculations based on World Development Indicators database. 28 SOUTH ASIA’S TURN
FIGURE 1A.2 Import Orientation Index—Rankings for South Asia region countries and comparators
50 VNM MDV L ND H I P DN I N G H F C A
0 N T BGD rientation Index B K NPL O PA Import
–50
Source: World Bank calculations based on World Development Indicators database.
Notes manner across 103 countries, five sectors (telecommunications, finance, transporta- 1. A slightly different picture emerges if, when tion, retail, and professional services) and measuring the export and import orientation the key modes of service supply from the per- of countries in South Asia, we control for spective of a foreign supplier who wishes to some nonpolicy determinants of openness. provide services to consumers in a particular For example, larger countries tend to trade country, with a focus on policy measures less with the rest of the world because their that discriminate against foreign services or domestic trade opportunities are larger than service providers. See Borchert, Gootiiz, and those of small countries. Similarly, land- Mattoo (2012) for more details. locked or island states face higher transpor- 3. A similar result was previously reported by tation costs and therefore tend to trade less. Amiti and Konings (2007) for Indonesia. Once size and whether a country is land- 4. This is consistent with cross-country evi- locked or an island state are taken into dence presented by Bolaky and Freund account, most countries in South Asia remain (2004) that the growth effect of trade depends less integrated into the global marketplace, on a country’s business regulations. both in terms of export and import orienta- 5. In the EXPY indicator, each export good is tion, than the average. However, India, the assigned the value of the average per capita largest economy in the region, appears to be income of other countries exporting that more integrated than the average (see plots of good; the country’s EXPY is the average of these Export and Import Orientation Indexes these values, weighted by the good’s share in figure 1A.1 and figure 1A.2). of total exports (Hausmann, Hwang, and 2. The Services Trade Restrictions Database Rodrik 2006). The EXPY indicator has to (used to calculate the indexes) collects and be interpreted with caution, because it makes publicly available information on ser- reflects the final product exported rather vices trade policy assembled in a comparable than the element of production carried out THE REGION’S COMPETITIVENESS POTENTIAL REMAINS LARGELY UNREALIZED 29
locally. In today’s increasingly fragmented Borchert, I., B. Gootiiz, and A. Mattoo. 2012. production processes, a country may export “Guide to the Services Trade Restrictions a sophisticated product, such as a computer, Database.” Policy Research Working Paper but contribute only low-skilled assembly 6108, World Bank, Washington, DC. activities using high-tech parts manufac- Goldberg, P., A. Khandelwal, N. Pavcnik, and tured elsewhere. P. Topalova. 2010. “Imported Intermediate 6. PRODY is an income level associated with a Inputs and Domestic Product Growth: Evidence given good, calculated as a weighted average from India.” Quarterly Journal of Economics of the per capita GDP of countries producing 125 (4): 1727–67. the good, with weights derived from revealed Gomez-Mera, L., T. Kenyon, Y. Margalit, J. Reis. comparative advantage calculations. and G. Varela. 2015. “New Voices in Invest- 7. The unit values secured by exporters are used ment: A Survey of Investors from Emerging here as a proxy for product quality because Countries.” Washington DC: World Bank. true product quality is unobserved. It is Hausmann, R., J. Hwang, and D. Rodrik. 2006. worth mentioning, however, that differences “What You Export Matters.” NBER Working in unit value may also reflect differences in Paper 11905, National Bureau of Economic manufacturing costs observed across firms or Research, Cambridge, MA. even across countries (see the discussion in Helpman, E., and P. R. Krugman. 1985. Market Khandelwal 2010). Structure and Foreign Trade: Increasing Returns, Imperfect Competition, and the International References Economy. Cambridge, MA: MIT Press. Hicks, J. R. 1935. “Annual Survey of Economic Aghion, P., N. Bloom, R. Blundell, R. Griffith, and Theory: The Theory of Monopoly.” Economet- P. Howitt. 2005. “Competition and Innovation: rica 3 (1): 1–20. An Inverted-U Relationship.” Quarterly IMF (International Monetary Fund). 2015. “Make Journal of Economics 120 (2): 701–28. in India: Which Exports Can Drive the Next Amiti, M., and J. Konings. 2007. “Trade Wave of Growth?” Working Paper 15119, Liberalization, Intermediate Inputs, and IMF, Washington, DC. Productivity: Evidence from Indonesia.” Kathuria, S., and M. Malouche. 2016. Strengthening American Economic Review 97 (5): 1611–38. Competitiveness in Bangladesh—Thematic Arnold, J., B. Javorcik, M. Lipscoomb, and Assessment: A Diagnostic Trade Integration A. Mattoo. 2015. “Services Reform and Study. Washington, DC: World Bank. Manufacturing Performance: Evidence from Kathuria, S., A. Portugal, and S. Shahid. 2015. India.” Economic Journal 126 (590): 1–39. “Barriers to Regional Integration in South Asia: Atkin, D., A. Khandelwal, and A. Osman. 2016. The Case of Para-tariffs.” World Bank, “Exporting and Firm Performance: Evidence Washington, DC (unpublished manuscript). from a Randomized Experiment.” NBER Khandelwal, Amit. 2010. “The Long and Short Working Paper 20690, National Bureau of (of) Quality Ladders.” Review of Economic Economic Research, Cambridge, MA. Studies 77 (4): 1450–76. Bernard, A., S. Redding, and P. Schott. 2011. Melitz, M. J. 2003. “The Impact of Trade on “Multiproduct Firms and Trade Liberalization.” Intra-industry Reallocations and Aggregate Quarterly Journal of Economics 126 (3): Industry Productivity.” Econometrica 71 (6): 1271–1318. 1695–1725. Bolaky, B., and C. Freund. 2004. “Trade, Topalova, P. and A. Khandelwal. 2011. “Trade Regulations, and Growth.” Policy Research Liberalization and Firm Productivity: The Case Working Paper 3255, World Bank, of India.” Review of Economics and Statistics Washington, DC. 93 (3): 995–1009.
Improving Competitiveness Requires Raising Productivity 2 Rather Than Keeping Costs Low
ompetitiveness occupies a central The comparison of South Asia’s and position in government and industry China’s apparel industries illustrates the diffi- Cagendas in South Asia.1 However, culty in defining national competitiveness, current performance—whether measured as because different countries have become participation in global markets or using “competitive” with different mixes of endow- common competitiveness benchmarks—has ments, factor prices, and policies (Porter been subdued. For example, the region pos- 1990). If competitiveness is defined as purely sesses several key advantages in the apparel gains in global market share, countries can sector: it has an abundant supply of work- remain competitive in the short term by keep- ers, labor costs are one-half to one-quarter ing production costs low through controlled those of China, and it is a top cotton pro- exchange rates, rigid factor markets, and sim- ducer in an industry in which textiles make ilar policies. However, such policies and the up close to 70 percent of production costs. larger focus on gaining a bigger slice of Some leading apparel firms in South Asia the pie are quite likely to be a zero-sum game have achieved world-class operational per- (Krugman 1994). A better strategy for formance by investing in training and tech- improved competitiveness is to reduce the nology, reaping economies of scale, and in transaction costs for firms to compete domes- India and Pakistan by integrating vertically tically and globally by providing efficient to avoid barriers to sourcing high-quality infrastructure services, a smoother business inputs on the global market. Nevertheless, environment, and more effective public ser- although South Asia increased its share vices. Still, reducing these costs has its limits. of the global apparel market from 7.5 to On the other hand, investing in productivity- 12.3 percent from 2000 to 2012, it continues enhancing measures can pay continuous divi- to lag well behind China, which accounts for dends over the long term. Porter (1990, 76) 41 percent of the market (table 2.1). Despite ties competitiveness to the efficiency with higher labor costs, China is able to attract which firms combine factors of production buyers by offering a wide range of apparel at (total factor productivity or TFP) and argues short lead times, while high productivity lim- that the “only meaningful concept of its total costs. No country in South Asia has competitiveness at the national level is thus far succeeded in offering a comparable productivity.” This report adopts a similar package of goods and services. perspective: productivity is what drives
31 32 SOUTH ASIA’S TURN
TABLE 2.1 South Asia lags well behind China in apparel exports despite much lower labor costs
Rank in top 15 Apparel exports as a share of Apparel exports as a share of Average apparel monthly Country apparel exporters world apparel exports (percent) country exports (percent) earnings (US$/per hour) Bangladesh 2 6.4 82.8 0.51 India 7 3.5 5.2 1.06 Pakistan 13 1.2 19.0 0.58 Sri Lanka 14 1.2 44.8 0.55 China 1 41 7.1 2.60
Source: World Bank calculations using UN COMTRADE data and household surveys. Note: Data are for 2012.
competitiveness in the long run, and boosting At the firm level, large productivity differ- productivity leads to rising living standards ences exist among South Asian firms, and through higher wages and returns on much of the region’s resources are locked investment. Productivity in South Asia has away in small, low-productivity firms that been less studied than various cost factors, so neither grow nor exit, indicating the existence the following discussion focuses on three of barriers to market entry and exit (Cabral major challenges to productivity growth in 2007; Li and Rama 2015; Tybout 1996). The the region. consequent “misallocation” of resources At the macro level, the contribution of TFP accounts for a large share of the difference in to growth in South Asia has declined in recent productivity between South Asia and high- years, and factors subject to diminishing income economies (Hsieh and Klenow 2009; returns—quantity rather than quality of labor Hsieh and Olken 2014; Pagés 2010).2 The and non–information and communication following discussion develops these observa- technology (ICT) investment—have been the tions in more detail. main drivers of growth. This calls for greater focus on improving productivity to sustain and accelerate growth, create jobs, reduce Macro challenge: Contribution of poverty, and boost shared prosperity. The TFP to growth is low and declining main forces that can increase productivity Gross domestic product (GDP) in South Asia growth are increased integration with the more than quadrupled from 1990 to 2014, global economy, the movement of resources and most countries enjoyed rapid growth in from agriculture to higher-productivity manu- output and per capita income (figure 2.1). facturing and services, and the movement of However, the contribution of TFP to GDP capital and labor from less-productive to growth has been mixed, as indicated by the more-productive firms within narrowly four countries for which data are available to defined economic activities. divide GDP growth into its components: At the sectoral level, the movement of changes in the quantity of labor, the quality labor from agriculture to industry and ser- of labor, ICT capital, non-ICT capital, and vices (structural transformation) has not TFP (figure 2.2).3 been rapid enough in South Asia to mark- Two main messages emerge from this edly reduce the large differences in produc- analysis: tivity across sectors. While countries in the region are at different points along this UÊ The contribution of TFP gains to eco- transformation, South Asia is overall in its nomic growth in the region has declined early stage, leaving significant untapped across the four countries for which data potential to reap productivity gains by are available. In India and Pakistan, the further reallocation of labor from lower- to contribution of TFP to GDP growth has higher-productivity activities. declined dramatically since 2011 and IMPROVING COMPETITIVENESS REQ UIRES RAISING PRODUCTIVITY RATHER THAN KEEPING COSTS LOW 33
FIGURE 2.1 Growth in real GDP and real GDP per capita in South Asia has accelerated
a. Real GDP growth 9 8 7 6 5
ercent 4 P 3 2 1 0 Afghanistan Bangladesh Bhutan India Maldives Nepal Pakistan Sri Lanka
b. Real GDP per capita growth 9 8 7 6 5
ercent 4 P 3 2 1 0 Afghanistan Bangladesh Bhutan India Maldives Nepal Pakistan Sri Lanka 1990–2000 2001–14 1990–2014
Source: World Bank calculations based on World Development Indicators database. Note: Afghanistan’s growth for both GDP and GDP per capita is calculated over 2002–14 based on data availability.
FIGURE 2.2 South Asia’s growth has been largely driven by factor accumulation
a. Bangladesh b. India 8 12
10 6 8
4 6
4 2 ercent ercent P P 2
0 0
–2 –2 –4
–4 –6
1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 2012 2014 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 2012 2014
(continues next page) 34 SOUTH ASIA’S TURN
FIGURE 2.2 South Asia’s growth has been largely driven by factor accumulation (continued)
c. Pakistan d. Sri Lanka 10 12 10 8 8 6 6 4 4 ercent ercent 2 P 2 P 0 0 –2 –2 –4 –4 –6
1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 2012 2014 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 2012 2014
TFP growth Contribution non-ICT capital Contribution ICT capital Contribution labor quantity Contribution labor quality GDP growth
Source: World Bank calculations based on Conference Board Total Economy database. Note: ICT = information and communication technology; TFP = total factor productivity.
FIGURE 2.3 Agriculture has given way to industry and services in 2006, respectively. In Sri Lanka, although South Asia the contribution of TFP picked up in 2014, it has declined from its high levels before 2009. In Bangladesh, TFP has played a negligible role in GDP growth 4 50 during the entire period of analysis. UÊ Non-ICT investment and increases in the number of workers have been the leading ) forces behind growth in all four countries.
GDP Although investment in ICT has increased 40 its contribution to growth in India, and substantially so in Sri Lanka, most of the growth is still accounted for by more labor (rather than higher-quality labor) and non-ICT investment—both factors 30 subject to diminishing returns. hare of value added (% hare S
Sectoral challenge: Slow pace of structural transformation 20 As in other countries at similar stages of devel- opment, resources in South Asia are moving 1960 1970 1980 1990 2000 2010 from agriculture to manufacturing and ser- Agriculture Services Industry vices (figure 2.3). This shift in economic activ- Source: World Development Indicators database. ity from lower-productivity, traditional IMPROVING COMPETITIVENESS REQ UIRES RAISING PRODUCTIVITY RATHER THAN KEEPING COSTS LOW 35
sectors to more modern and productive ones region remains high at over 50 percent, is known as structural transformation. despite the fact that labor productivity in Between 1960 and 2013, the average share of industry and services is several times that of agriculture in GDP in South Asia fell from agriculture. For example, in India average 44 percent to 19 percent, while the share of labor productivity in industry from 2004 to industry increased from 18 to 29 percent. 2013 was approximately 5 times, and ser- Over the same period, real GDP per capita of vices 6.5 times, the level of productivity in the region nearly quintupled. agriculture, illustrating the potential for Individual countries are at different major productivity gains in the region from points along this transformation, but the accelerating the process of structural overall trend is quite consistent (figures 2.4 transformation. and 2.5). An increase in real GDP per capita Differences in labor productivity play an has been associated with a decline in the important role in pulling labor to more pro- share of agriculture in employment and ductive sectors, above and beyond the value added and concomitant increases in “natural” rate of structural transformation the shares of services and industry.5 (table 2.2).7 In other words, movement Maldives, a small island economy, has tradi- from agriculture to industry and services is tionally had a large services sector owing to significantly faster in periods when the dif- tourism and has not experienced much ference in productivity between the two sec- structural change in the last decade. The tors is larger. In this way, the process of share of value added of industry, however, structural transformation in South Asia is appears to plateau at certain levels of income similar to the experience of the coun- (at least in the larger countries), consistent tries of the Organisation for Economic with the views of structural transformation Co-operation and Development (OECD), in the literature (Herrendorf, Rogerson, and although there are differences due either to Valentinyi 2013). The share of employment the two groups of countries being at a dif- in agriculture has been consistently higher ferent stage of the same path, or to the paths than the share of value added at all levels of being somewhat different. In either case, the GDP per capita for the four countries with response of employment to differences in adequate data. This suggests that agricul- sectoral productivity levels is much higher tural productivity has not improved sub- in South Asia than in OECD economies, stantially over the decades. The share of indicating that boosting productivity services in employment and value added in growth in manufacturing and services in this sector is high in South Asia, particularly South Asia carries a much greater potential considering the relatively low income levels. for accelerating structural transformation, The share of services in GDP appears to rise increasing nonfarm employment, and rais- relatively early in the development process, ing income growth. at below $700 per person in 2005 prices. This pattern differs from the experience of Firm challenge: Firm growth is mature industrialized countries such as the low and resources are trapped in United Kingdom, France, and the United small firms States, where the share of the services sector reached high levels only at high levels of In addition to structural transformation GDP per capita (Verma 2012). (resources moving from less- to more- Workers in South Asia are moving from productive sectors), productivity growth can agriculture to the higher-productivity manu- be driven by movement in resources from less- facturing and services sectors, a transition productive to more-productive firms within that has been associated with increases in narrowly defined economic activities. When aggregate productivity.6 However, the share this mechanism does not function as effec- of agriculture in total employment in the tively as it could—due, for example, to 36 SOUTH ASIA’S TURN
FIGURE 2.4 As incomes grow, resources shift from agriculture to industry and services Employment and share of value added by sector for Bangladesh, India, Pakistan, and Sri Lanka, 1975–2013
a. Share of employment, agriculture b. Share of value added, agriculture 70 40
35 60 30
50 25 ercent ercent P P 20 40 15
30 10
5.5 6.0 6.5 7.0 7.5 5.5 6.0 6.5 7.0 7.5 Log of GDP per capita (2005 constant $) Log of GDP per capita (2005 constant $) Bangladesh India Pakistan Sri Lanka Bangladesh India Pakistan Sri Lanka
c. Share of employment, industry d. Share of value added, industry
25 30
25 20 ercent ercent P P 20
15 15
5.5 6.0 6.5 7.0 7.5 5.5 6.0 6.5 7.0 7.5 Log of GDP per capita (2005 constant $) Log of GDP per capita (2005 constant $) Bangladesh India Pakistan Sri Lanka Bangladesh India Pakistan Sri Lanka
e. Share of employment, services f. Share of value added, services
45 60 40 55 35 50
ercent 30 ercent P P 45 25 40 20 35 5.5 6.0 6.5 7.0 7.5 5.5 6.0 6.5 7.0 7.5 Log of GDP per capita (2005 constant $) Log of GDP per capita (2005 constant $) Bangladesh India Pakistan Sri Lanka Bangladesh India Pakistan Sri Lanka
Source: World Development Indicators database. IMPROVING COMPETITIVENESS REQ UIRES RAISING PRODUCTIVITY RATHER THAN KEEPING COSTS LOW 37
FIGURE 2.5 As incomes grow, resources shift from agriculture to industry and services Share of value added by sector in Afghanistan, Bhutan, Maldives, and Nepal
a. Agriculture b. Industry
60 50
50 40 40
30 ercent 30 ercent P P
20 20
10 10 0 5.5 6.0 6.5 7.0 7.5 8.0 8.5 5.5 6.0 6.5 7.0 7.5 8.0 8.5 Log of GDP per capita (2005 constant $) Log of GDP per capita (2005 constant $) Afghanistan Bhutan Maldives Napal Afghanistan Bhutan Maldives Napal c. Services
80
70
60 ercent
P 50
40
30
5.5 6.0 6.5 7.0 7.5 8.0 8.5 Log of GDP per capita (2005 constant $)
Afghanistan Bhutan Maldives Napal
Source: World Development Indicators database. barriers to competition—the economy suffers different levels of productivity, with a range from misallocation of resources (box 2.1). that is much wider than that in the United Evidence shows that the misallocation can States. Reducing these productivity gaps to substantially reduce aggregate productivity. In the level of efficiency observed in the United a seminal study, Hsieh and Klenow (2009) States would increase TFP by 40–60 percent measure resource misallocation in China and and 30–50 percent in India and China, respec- India by comparing productivity dispersion tively, and output would increase by twice as among firms in these countries with the U.S. much if investment increased in response to market.8 They find that firms in China and higher productivity. Conversely, a more rapid India produce the same products with vastly expansion of less-efficient firms in the early 38 SOUTH ASIA’S TURN
TABLE 2.2 Labor productivity drives structural transformation From agriculture to industry From agriculture to services OECD countries Intercept 0.14 0.08 (0.09) (0.08) Lag employment share 0.97*** 0.97*** (0.00) (0.01) Labor productivity differential 0.18* 0.17* (0.08) (0.07) Observations 730 730 India, Pakistan, and Sri Lanka Intercept 15.68*** 16.53*** (4.15) (4.77) Lag employment share 0.84*** 0.83*** (0.05) (0.05) Labor productivity differential 4.43*** 3.70** (1.25) (1.21) Observations 70 70
Source: World Bank calculations based on World Development Indicators database. Note: Standard errors in parentheses, country fixed effects included but not shown. OECD = Organisation for Economic Co-operation and Development. * p < 0.05, ** p < 0.01, *** p < 0.001.
BOX 2.1 Barriers to competition and productivity dispersion
Consider a fictional example of an economy Barriers to competition in the real world also with two firms.a One operates at low produc- can create and perpetuate substantial misalloca- tivity, but manages to survive in the market tion of resources. For example, informal, low- because it has political connections through productivity retailers in Brazil hold a large share which it secures subsidized credit. The other of the market because they are subject to less firm has no political connections, borrows at stringent labor market regulations, and thus the market rate, and therefore faces higher lower labor costs, than higher-productivity costs. However, the second firm has higher pro- supermarkets (McKinsey Global Institute 1998). ductivity, which enables it to compete with the Subsidized loans and differential tax-code treat- first firm. If labor and capital were to move ment in Japan are used “to keep mom-and-pop from the firm with low productivity to the firm retailers from going out of business” (Lewis with high productivity, aggregate output would 2004, 14–15). And, prior to reforms, severe be higher, and the difference between produc- restrictions on foreign direct investment (FDI) tivity levels in the two firms would be lower. in retail prevented investment by global best- Thus, the misallocation of capital results in practice retailers in India. low output per worker, on average, across the two firms. a. This example is taken from Hsieh and Klenow (2009). IMPROVING COMPETITIVENESS REQ UIRES RAISING PRODUCTIVITY RATHER THAN KEEPING COSTS LOW 39
1990s reduced TFP growth in Indian manu- apparel sector, which is significantly exposed facturing by 2 percent over the 1987–94 to competition through exports, show less period (Hsieh and Klenow 2009). productivity dispersion than firms in India. Although the detailed data required to Some authors explain high productivity replicate the Hsieh and Klenow (2009) analy- dispersion—and, consequently, lower overall sis for other countries in the region are not productivity—as the result of a disproportion- available, indicators of productivity differ- ately high number of small, unproductive ences among firms compared to India can be firms that neither grow nor exit, releasing calculated for a few sectors in Bangladesh, resources into the economy (Li and Rama Nepal, and Sri Lanka. The results show sub- 2015). For example, the share of manufactur- stantial scope for improving productivity in ing firms with fewer than 10 employees in Nepal by shifting labor and capital to higher- India is almost visually indistinguishable from productivity firms in food and beverages 100 percent and the most common observation (firms in the lowest 10 percent of the TFP in the sample is a firm with a single employee distribution are more than five times less pro- (Hsieh and Olken 2014). By contrast, in the ductive than those in the highest 10 percent) United States, the most common observation and other manufacturing, but less scope in the sample is a firm with 45 employees. (relative to India) in Sri Lanka and Moreover, the dominance of small firms in Bangladesh (table 2.3).9 The results also India appears to be the same as, or perhaps hint at the importance of competition: in even greater than, it was more than twenty Bangladesh and Sri Lanka, firms in the years ago (figure 2.6). Evidence from the
TABLE 2.3 Productivity dispersion across South Asia’s firms tends to be large Coefficients of variation by sector and country
Sector Bangladesh Sri Lanka Food and beverages 0.64 1.56 Textiles 1.79 — Apparel 0.65 0.98 Basic metals 2.75 — Other manufacturing 1.49 2.95
Source: World Bank calculations based on World Bank Enterprise Surveys (Bangladesh 2013, India 2014, Sri Lanka 2011). Note: Dispersion in each sector is normalized to India (i.e., India’s dispersion is normalized to 1.00). — = not available.
TABLE 2.4 Small firms dominate the distribution across South Asia Distribution of firms by number of employees, South Asia and comparator countries
Country Year Small (5–19) Medium (20–99) Large (100 or more) Afghanistan 2014 69 26 5 Bangladesh 2013 37 36 27 Bhutan 2015 71 25 4 China 2012 55 32 13 India 2014 43 44 13 Indonesia 2009 88 10 2 Nepal 2013 82 16 2 Pakistan 2013 44 40 16 Philippines 2009 52 35 13 Sri Lanka 2011 76 18 6 Vietnam 2009 45 36 18
Source: Calculations based on World Bank Enterprise Surveys. Note: Distributions may not total 100 percent due to rounding. 40 SOUTH ASIA’S TURN
FIGURE 2.6 Size distribution of firms in India is heavily biased toward small firms
a. 1990–91 100 100 100
80 80 80
60 60 60 ercent ercent ercent P 40 P 40 P 40
20 20 20
0 0 0 0 0 2 0 20 40 60 80 4 6 8 10 100 20,000 40,000 60,000 80,000 100,000 Workers per firm Workers per firm Workers per firm
b. 2000–01 100 100 100
80 80 80
60 60 60 ercent ercent ercent P 40 P 40 P 40
20 20 20
0 0 0 0 0 0 2 4 6 8 20 40 60 80 100 10 10,000 20,000 30,000 40,000 Workers per firm Workers per firm Workers per firm
c. 2009–10 100 100 100
80 80 80
60 60 60 ercent ercent ercent P 40 P 40 P 40
20 20 20
0 0 0 0 0 0 2 4 6 8 20 40 60 80 100 10 50,000 100,000 Workers per firm Workers per firm Workers per firm
Source: World Bank calculations based on combined Annual Survey of Industries (ASI) and National Sample Survey (NSS) data. IMPROVING COMPETITIVENESS REQ UIRES RAISING PRODUCTIVITY RATHER THAN KEEPING COSTS LOW 41
World Bank Enterprise Surveys indicates that firms (figure 2.7).12 This pattern is also docu- the importance of small firms in South Asia is mented by Hsieh and Klenow (2009), who greater than in comparator East Asian argue that the relationship between productiv- countries (table 2.4).10 Countries with higher ity and size is stronger in China and India than levels of GDP per capita tend to have a smaller in the United States due to distortions that pre- share of firms with only a few employees (see vent firms from achieving optimal size (which figures 2A.1, 2A.2, and 2A.3). is also consistent with Banerjee and Duflo’s There are many reasons why small firms (2006) contention that Indian policies con- may be less productive than larger ones: econ- strain its most efficient producers and coddle omies of scale, access to finance, better its least efficient ones). Moreover, India’s pro- employees, and stronger business practices. ductivity growth between 1993 and 2007 was Larger firms tend to innovate more, particu- associated with productivity gains within large larly in process and organization, because they manufacturing plants (those with 200 or more can more easily secure financing for risky proj- workers) rather than with gains within small ects and because of the potential for econo- firms or with reallocation between plants mies of scale in research and development (Klenow, Sharma, and Bollard 2011). investments (see Del Mel, McKenzie, and Another symptom of resource misalloca- Woodruff 2008; Cohen and Klepper 1996, tion in the region is the difficulty firms face and Ayyagari, Demirgüç-Kunt, and in growing. In India, manufacturing plants Maksimovic 2007). Larger firms also tend to that are 40 years old are only 40 percent invest more in administration and adopt better larger than manufacturing plants that are management and overall business practices, less than 5 years old, while in the United which are highly correlated with firm perfor- States older plants are more than seven times mance (Bloom and Van Reenen 2007; Bloom larger than younger ones (Hsieh and Klenow et al. 2013).11 Regardless of the channel, pro- 2014). The same conclusion holds more ductivity does appear to be lower in small broadly across the region: Firms aged than in large firms in Asia. For example, in 25 years or more in Bangladesh, Bhutan, both India and China, value added per worker India, and Sri Lanka are only 50–90 percent in small firms is much lower than in large larger than firms aged less than 5 years.
FIGURE 2.7 Labor productivity is lower in small firms in India and China
Small (<20)
Medium (20–99)
Large (100 and over)
All
0 5,000 10,000 15,000 20,000 25,000 Value added per worker ($) China (2012) India (2014)
Source: World Bank calculations based on World Bank Enterprise Surveys. Note: In India, value added per worker for all firms is calculated from a sample of 4,774 firms; for large firms from a sample of 1,144; for medium, of 2,378 firms; and for small firms of 1,252. For China, the sample sizes are 1,349, 579, 587, and 183 firms, respectively. 42 SOUTH ASIA’S TURN
By contrast, older firms in Vietnam, (Hsieh and Klenow 2009). A higher marginal Indonesia, and China are on average 4.5, product of labor in large firms likely indi- 4.8, and 2.4 times the size of younger cates distortions that prevent firm growth; firms, respectively (figure 2.8). One explana- in a world without distortions, firms would tion for this is that in India and China, continue to expand until the marginal prod- within narrowly defined industries, larger uct of labor or capital equalizes across firms. plants have higher marginal products of Barriers to the growth of firms can likely labor and capital, while in the United be found in economic policy. Across the States the difference is much smaller region, licensing and size restrictions (which have declined in importance but still exist), labor regulations that increase the costs of hiring and firing, financial sector regulations FIGURE 2.8 Firms in South Asia grow less rapidly than those in that favor small enterprises, and inadequate comparator countries bankruptcy laws may limit the ability of effi- a. Average employment cient plants to grow and enable inefficient 5 plants to survive. Problems in enforcing con- tracts in India, for example, make it costly to hire the right managers, which is crucial for 4 firms’ growth (Bloom et al. 2013). Taxes or labor costs that affect larger firms more 3 than smaller firms may reduce the return on investment in large firms. Impediments to
umber of workers 2 reaching foreign markets, both from trade N policy and the high cost of logistics, can also impede expansion. 1 <5 5–14 15–24 >25 Four case studies of important Age of firm industries in South Asia b. Average export share over total sales Four industry case studies (agribusiness, elec- 6 tronics, apparel, and automotive) are an essential part of this report because they enable a better understanding of the driv- 4 ers and constraints of South Asia’s competitiveness. One of the case studies is fea- tured at the end of each chapter of part 2. ercent P 2 Extended versions of the industry case studies can be found online at www .worldbank .org / SouthAsiaCompetes. The main conclusions and recommendations from these case studies 0 are also featured in the overview and part 3 of <5 5–14 15–24 >25 the report. Age of firm The industry case studies show the links between the external environment of firms India (2014) China (2012) Bhutan (2015) and their behavior within a well-defined Bangladesh (2013) Indonesia (2009) Nepal (2013) Pakistan (2013) SriLanka (2011) Philippines (2009) industry in which industry dynamics (such as Vietnam (2009) competition) can be analyzed and perfor- mance benchmarked. They promote an Source: World Bank calculations based on World Bank Enterprise Surveys. Note: Age cohorts of firms have been normalized with respect to the first age cohort, firms fewer understanding of the relative importance of than 5 years old. the external factors that drive or constrain a IMPROVING COMPETITIVENESS REQ UIRES RAISING PRODUCTIVITY RATHER THAN KEEPING COSTS LOW 43
firm’s performance and productivity. Also, sector despite very competitive labor costs and crucially, the industry case studies assess and the fact that leading firms are achiev- the impact of industry-specific factors and ing world-class productivity in the region. policies that traditional cross-cutting analysis The case study is featured at the end is ill-equipped to identify. of chapter 4, “Productivity-Boosting And indeed, the main finding from the Agglomeration Economies,” because it four industry case studies is that industry- shows that South Asia is missing urban specific policies (also called product market ecosystems that provide thick markets for regulations), are the main constraint to skilled labor, large tracts of industrial land South Asia’s realizing its great untapped for clusters to thrive, and world-class competitiveness potential. These policies logistics to import and export seamlessly. include restrictions on trade, prices, prod- UÊ Apparel is the largest globally traded labor- ucts (through standards), and markets that intensive industry in the world. With rising have protected firms from exposure to labor cost in East Asia, South Asia has an global good practices (in automotive and historic opportunity to capture its fair share agribusiness) or have limited firms’ capac- of the global apparel market (having only ity to adopt these practices (in apparel 12 percent compared to 41 percent for and electronics). China alone), in the process pulling millions Manufacturing case studies were selected out of poverty, especially women. The case because the region’s performance has been is featured at the end of chapter 5, “Limited lagging in manufacturing relative to services. Success in Linking to Global Value Chains,” The following manufacturing sectors were because it shows how, despite the reforms chosen because they are important and repre- of the 1990s, trade barriers continue to sentative of different types of manufacturing stand in the way of South Asia’s realizing industries: its great potential in apparel. UÊ Automotive is one of the most important UÊ Agribusiness (including agriculture) industries globally and in South Asia, con- accounts for one-third of South Asia’s GDP tributing 19 million direct and indirect and is crucial for all countries. Increasing jobs in India alone. The potential for South income and urbanization create the pros- Asia to become globally competitive in this pect of significant growth. Livelihoods in sector is shown by the experience of Indian rural areas can be improved by linking auto-parts manufacturers who became farmers to processors and traders who are world leaders by having first acquired tech- willing to pay a premium for higher-quality nical and managerial skills from leading products as well as by providing off-farm original equipment manufacturers estab- job opportunities. The case study is fea- lished in India, followed by a process of tured at the end of chapter 3, “Business serving increasingly discerning customers Environment Challenges,” because it in competitive export markets. The case is shows how, despite much attention and featured at the end of chapter 6, “Firm reform, business environment issues, espe- Capabilities,” because it shows how firm cially industry-specific product market reg- capabilities are acquired and spread ulations, which have generally been through competitive exposure to global overlooked by policy makers, continue to good practices. be a big challenge in South Asia. UÊ Electronics is one of the largest and fastest The country coverage of each industry case growing industries in the world and has study (as shown in table 2.5) was determined played an important role in the develop- on the basis of its importance and data avail- ment trajectories of several newly industri- ability. For each industry case study, we also alized economies. Surprisingly, South Asia included relevant good practice benchmarks is not currently a significant player in the from outside South Asia, mostly from East 44 SOUTH ASIA’S TURN
TABLE 2.5 Country coverage of the four industry case studies
Afghanistan Bangladesh Bhutan India Maldives Nepal Pakistan Sri Lanka China Vietnam Thailand Agribusiness X X X X X X X X X Electronics X X X X X Apparel X X X X X X Automotive X X X
Asia, which shares many characteristics with UÊ The third step addresses constraints in South Asia and has been performing better. the external environment of firms that The analytical approach for these case limit their capacity or incentives to take studies relied on both quantitative and quali- advantage of these drivers. This step com- tative analyses and was carried out using the bines in-depth firm interviews with an following framework (see the online extended analysis of the impact of external factors versions for more details): (such as infrastructure constraints and trade regimes) on firms’ behavior and UÊ The first step assesses the competitiveness performance. performance of the industry in each coun- UÊ The fourth and final step develops try by comparing its performance in out- policy recommendations to remove the put, trade, productivity, and cost with the constraints and exploit the drivers of other selected South Asian countries, as competitiveness. This final step is inspired well as with the good practice comparator by the policy choices taken by the more countries from outside the region (see successful countries in the region and else- table 2.5). This first step relies primarily where such as, for example, the bonded on quantitative analysis using national sta- warehouse regime that facilitated access to tistics for output, World Integrated Trade imported textiles for Bangladesh’s apparel Solutions (World Bank and UNCTAD exporters. It also includes, in the case of 2015) for trade data, and enterprise sur- apparel, an estimate (based on a gravity veys (national surveys and standardized model) of how improved competitiveness, World Bank Enterprise Surveys) for pro- fostered by new policies, would affect out- ductivity and cost. put and jobs. UÊ The second step analyzes the drivers of pro- ductivity and cost at the firm and industry levels, including scale, skills, technology Annex 2A and innovation, agglomeration economies within clusters, and links along local and The following model describes a simple pro- global value chains. This step combines cess of reallocation of labor across different results from enterprise surveys (including sectors, assuming a two-sector economy the innovation and labor force modules of consisting of agriculture and industry. the World Bank Enterprise Surveys) and in- Employment in each sector can change depth firm interviews. It also includes, in because of a net addition of new workers as the case of apparel, a survey of global buy- the labor force grows or an intersectoral ers to understand what drives their deci- migration of labor. This can be represented sions apart from cost and quality—for as follows: example, short lead times or compliance AAA with social and environmental standards. LgLMt =+(1 ) t−1 + (1) IMPROVING COMPETITIVENESS REQ UIRES RAISING PRODUCTIVITY RATHER THAN KEEPING COSTS LOW 45
where LA is the number of people employed FIGURE 2A.1 Scatterplot of firms at the 25th percentile of in agriculture, gA is the constant net rate of employment distribution and income per capita employment growth in agriculture, and M 3.0 is the migration of laborers from industry to agriculture. Dividing equation (1) by BGD13 2.5 BGD07 total employment L, and assuming total PAK13 IND14 employment grows at a constant net rate of 2.0 g yields: AFG08 BTN09 AFG14 PAK07 LKA11 NPL13 A + A A NPL09 Lt = (1gL ) t −1 + M (2) 1.5
+ og 25th percentile Lt (1gL ) tt−1 L L 1.0 Define m as the percentage of population that migrates from one sector to another: 0.5 m = M/Lt. Assuming there are no frictions in 4681012 the movement of labor from agriculture to Log GDP per capita, 2005 constant $ industry (or, equivalently, that these frictions remain constant over time), intersectoral Source: Calculations based on World Bank Enterprise Surveys and World Development Indicators database. migration should be driven by the wage gap between the two sectors. For example, if the industrial wage rate is higher than the agricul- FIGURE 2A.2 Scatterplot of firms at the 50th percentile of tural wage rate, workers should migrate from employment distribution and income per capita agriculture to industry. Further, the wage rate in a sector is equal to the marginal product of 4.0 labor in that sector, which in turn is a func- tion of labor productivity. Thus m can be 3.5 BGD07 written as: BGD13 IND14 3.0 PAK13 m = f(wA,wI) = f(h(w A,w I)) = g(w A,w I) (3)
2.5 AFG08 AFG14 BTN09 where w is the wage rate in a sector, and PAK07 og 50th percentile L w is the labor productivity. Equation (2) can NPL09 LKA11 2.0 therefore be expressed as follows: NPL13
1.5 A A A L (1+ gL ) − t = t 1 + g(,)ωωAI (4) 4 6 8 10 12 + Lt (1gL ) t−1 Log GDP per capita, 2005 constant $ This equation can be estimated using the Source: Calculations based on World Bank Enterprise Surveys and World Development following reduced-form model for OECD Indicators database. countries and the South Asian economies for which we have sufficient data on where i represents agriculture and j repre- employment (India, Pakistan, and Sri sents industry or services, k represents the Lanka). Proxying labor productivity in each country dummy, and t represents time. sector with value added per worker yields V APW is the value added per worker calcu- the following: lated as the value added at constant prices
divided by the number of workers. Li/L is the Likt =α+γLikt−1 +β − ()(log()VAPWikt−1 employment share calculated as the number L L − (5) kt kt 1 of employees in a sector divided by total δ∈ log(V APWjkt−1 )) ++ k Countrykikt employment. 46 SOUTH ASIA’S TURN
FIGURE 2A.3 Scatterplot of firms at the 75th percentile of errors). Second, the calculations are made by employment distribution and income per capita assuming a degree of competition in factor markets sufficient to ensure that factor earn- 5 BGD07 ings are proportional to factor productivity.
BGD13 Third, growth accounting cannot measure the fundamental causes of growth (policies, insti- tutions, and history), but simply examines the 4 PAK13 proximate causes. Although these critiques IND14 have merit, the framework provides a simple and internally consistent way to organize data
AFG08AFG14 and is useful in generating insights into the BTN09 3 PAK07 process of economic growth. LKA11
Log 75th percentile Log 4. Note that the analysis does not distinguish NPL09 NPL13 between a structural decline in TFP growth and a decline from idle or poorly allocated 2 factors of production during cyclical periods of economic slowdown. 4 6 8 10 12 5. Sectoral employment shares data for Log GDP per capita, 2005 constant $ Afghanistan, Bhutan, Maldives, and Nepal Source: Calculations based on World Bank Enterprise Surveys and World Development are available only for very few years and have Indicators database. not been used in this analysis. 6. The economic development literature has long recognized the role of structural trans- Notes formation in boosting aggregate productivity. Baumol (1967), Dekle and Vandenbroucke 1. The Indian government’s Make in India (2012), Ngai and Pissarides (2007), Acemoglu initiative has industry competitiveness at its and Guerrieri (2008), and others have shown core. The Sri Lankan government has set up a that differential productivity growth across National Productivity Secretariat to help sectors will attract resources to more produc- enhance Sri Lankan productivity by energizing tive parts of the economy. the sector to face international competition. 7. A simple reduced-form model is used to 2. Hsieh and Klenow (2009), Pagés (2010), and quantify the role of productivity in driving others document that productivity dispersion the process of structural transformation. The between top and bottom firms is particularly movement of labor from one sector to another large in developing countries. The positive is a function of productivity differentials correlation between firm size and productivity across sectors (see annex 2A for a description (although weaker in developing countries), of the model). The natural rate of structural and the fact that medium and large firms are transformation is measured by the coefficient underrepresented in South Asia, suggests that on the share of employment lagged one the preponderance of small firms drags down period. The coefficient on the lagged employ- aggregate productivity, presenting a major ment share is always positive, significant, opportunity for improving productivity. and significantly less than one, indicating a 3. It is worth mentioning that TFP is measured “natural” downward trend in the employ- as the portion of output that is not explained ment share of agriculture. by increases in physical capital stocks or in 8. In theory, with no distortions firms produc- labor (both measured in quality and quantity). ing similar products would have the same In fact, three important critiques have been level of productivity. If that were not leveled at this growth accounting framework. the case, then resources would move from First, because TFP is measured as a residual, it the low-productivity firm to the higher provides an imperfect measure of shifts in the returns of the high-productivity firm, driving production function, which includes many the productivity of the former upward and determinants (particularly technical change, of the latter downward. Given that factors but also sustained political turmoil, external omitted in the model may be responsible shocks, institutional changes, or measurement for productivity differences, the benchmark IMPROVING COMPETITIVENESS REQ UIRES RAISING PRODUCTIVITY RATHER THAN KEEPING COSTS LOW 47
for comparison is not zero productivity dis- Crisis.” American Economic Review 57 (3): persion, but that of the United States—a 415–26. relatively undistorted market. Bloom, N., B. Eifert, A. Mahajan, D. McKenzie, 9. As argued by Li and Rama (2015), because and J. Roberts. 2013. “Does Management micro and small firms are generally underrep- Matter? Evidence from India.” Quarterly resented in datasets like the World Bank Journal of Economics 128 (1): 1–51. Enterprise Surveys, these indicators of dis- Bloom, N., and J. Van Reenen. 2007. “Measuring persion of productivity likely underestimate and Explaining Management Practices Across true dispersion. Firms and Countries.” Quarterly Journal of 10. The World Bank Enterprise Surveys collect Economics 122 (4): 1341–1408. information about firms in different coun- Cabral, L. 2007. “Small Firms in Portugal: A tries around the world in a harmonized Selective Survey of Stylized Facts, Economic way providing comparable cross-country Analysis, and Policy Implications.” Portuguese information. The dataset includes only for- Economic Journal 6 (1): 65–88. mal firms with at least five employees. The Cohen, W., and S. Klepper. 1996. “Firm Size and reader should keep in mind this truncation the Nature of Innovation Within Industries: when interpreting results. The Case of Product and Process R&D.” 11. This conclusion holds even when restricting Review of Economics and Statistics 78 (2): the sample to small and micro firms. Among 232–43. these, the larger did better at business Dekle, R., and G. Vandenbroucke. 2012. practices within a sample of small firms in “A Quantitative Analysis of China’s Structural Bangladesh, Chile, Ghana, Kenya, Mexico, Transformation.” Journal of Economic Nigeria and Sri Lanka (McKenzie and Dynamics and Control 36 (1): 119–35. Woodruff 2015). Del Mel, S., D. McKenzie, and C. Woodruff. 2008. 12. Note that this calculation is based on the “Returns to Capital in Microenterprises: World Bank Enterprise Surveys, which do Evidence from a Field Experiment.” Quarterly not include most micro firms, so it most Journal of Economics 123 (4): 1329–72. likely underestimates productivity differences Herrendorf, B., R. Rogerson, and Á. Valentinyi. that would be observed if firms of all sizes 2013. “Growth and Structural Transformation.” were considered. Another important caveat NBER Working Paper 18996, National Bureau is that differences in value added per worker of Economic Research, Cambridge, MA. between firms within different size classes Hsieh, C., and P. J. Klenow. 2009. “Misallocation may to some extent be explained by differ- and Manufacturing TFP in China and India.” ences in average worker quality, as workers Quarterly Journal of Economics 124 (4): with higher abilities or skills may self-select 1403–48. into larger firms. ———. 2014. “The Life Cycle of Plants in India and Mexico.” Quarterly Journal of Economics References 129 (3): 1035–83. Hsieh, C., and B. A. Olken. 2014. “The Missing Acemoglu, D., and V. Guerrieri. 2008. “Capital ‘Missing Middle.’” Journal of Economic Deepening and Nonbalanced Economic Perspectives 28 (3): 89–108. Growth.” Journal of Political Economy Klenow, P., Sharma, G., and A. Bollard. 2011. 116 (3): 467–98. “India’s Mysterious Manufacturing Miracle.” Ayyagari, M., A. Demirgüç-Kunt, and 2011 Meeting Paper No. 1176, Society for V. Maksimovic. 2007. “Firm Innovation in Economic Dynamics, Minneapolis, MN. Emerging Markets.” Policy Research Working Krugman, P. 1994. Peddling Prosperity: Economic Paper 4157, World Bank, Washington DC. Sense and Nonsense in the Age of Expectations. Banerjee, A., and E. Duflo. 2006. “Growth Theory New York: W. W. Norton & Co. through the Lens of Development Economics.” Lewis, W. 2004. The Power of Productivity: In Handbook of Economic Growth, edited by Wealth, Poverty, and the Threat to Global P. Aghion and S. Durlauf, vol. 1A, 473–552. Stability. Chicago: University of Chicago Press. Amsterdam: North-Holland Publishing Co. Li, Y., and M. Rama. 2015. “Firm Dynamics, Baumol, W. J. 1967. “Macroeconomics of Productivity Growth, and Job Creation in Unbalanced Growth: The Anatomy of Urban Developing Countries: The Role of Micro- and 48 SOUTH ASIA’S TURN
Small Enterprises.” World Bank Research Porter, M. 1990. “The Competitive Advantage of Observer 30 (1): 3–38. Nations.” Harvard Business Review 68 (2): McKenzie, D., and C. Woodruff. 2015. “Business 73–93. Practices in Small Firms in Developing Tybout, J. R. 1996. “Heterogeneity and Countries.” NBER Working Paper 21505, Productivity Growth: Assessing the Evidence.” National Bureau of Economic Research, In Industrial Evolution in Developing Cambridge, MA. Countries: Micro Patterns of Turnover, McKinsey Global Institute. 1998. Productivity: Productivity, and Market Structure, edited by The Key to an Accelerated Development Path M. J. Roberts and J. R. Tybout, 43–68. Oxford: for Brazil. Washington, DC: McKinsey Global Oxford University Press. Institute. Verma, R. 2012. “Can Total Factor Productivity Ngai, R., and C. A. Pissarides. 2007. “Structural Explain Value Added Growth in Services?” Change in a Multi-sector Model of Growth.” Journal of Development Economics 99 (1): American Economic Review 97 (1): 429–43. 163–77. Pagés, C., ed. 2010. The Age of Productivity: World Bank and UNCTAD (United Nations Transforming Economies from the Bottom Up. Conference on Trade and Development). 2015. New York: Palgrave Macmillan. World Integrated Trade Solutions Trade Data. PART 2
Productivity Performance: Firms and Linkages
lthough productivity can be measured (ICT), and improving the input mix. This is at different levels—macro, sectoral, the within-firm component of productivity Aand geographic, for example—the growth. Second, competition also induces most robust and intuitive representation is inefficient firms to transfer resources to more at the level of the firm. The focus on the efficient firms or exit altogether, boosting firm as the unit of analysis and on firm economy-wide productivity—the between- dynamics as the driver of productivity firm component of productivity growth growth goes back at least as far Schumpeter (Cabral 2007). Resources can flow from less- (1942), with competition playing a key role to more- productive uses through improve- in forcing inefficient, unproductive, or ments in standard factors such as unprofitable firms to either improve or exit infrastructure and business environment, but and transfer their resources to more efficient, also as a result of participation in global value productive, or profitable firms, thus boost- chains (GVCs) (Saia, Andrews, and Albrizio ing economy-wide productivity. 2015) and agglomeration economies (Desmet Formally, there are two mutually reinforc- and Rossi-Hansberg 2009; Michaels, Rauch, ing mechanisms, spurred on by competition in and Redding 2012). product and factor markets, that increase pro- The decomposition of changes in produc- ductivity. First, greater competition, from tivity into between- and within-firm compo- either domestic or international sources, nents (with the former further broken down pushes firms to become more efficient through into contributions from firm entry and exit) learning from international exposure, invest- has become a standard approach to thinking ing in innovation, improving business prac- about productivity dynamics (Olley and tices, adopting better technology, including Pakes 1996; Melitz and Polanec 2012). information and communication technology Unfortunately, none of the countries in the 50 SOUTH ASIA’S TURN
region carry out large, longitudinal firm sur- Participation in GVCs can raise productivity veys that would directly allow for this type of through exposure to competition and knowl- analysis.1 Therefore, this report approximates edge spillovers from connections with lead the spirit of the decomposition by using cross- firms; however, South Asian participation in sectional firm data to consider elements that GVCs is largely confined to apparel. are likely to affect productivity across and Reducing trade barriers, increasing skills, and within firms, in turn. This part begins with improving logistics would facilitate greater the business environment, agglomeration participation in GVCs. Finally, access to economies, and participation in GVCs as technology varies greatly across South Asian determinants of performance across firms, economies, ranging from extensive technol- and then considers how input use, technology ogy use in India to limited ICT adoption in adoption, and innovation affect within-firm Bangladesh and Nepal. Even among lead productivity. countries, however, the use of e-commerce Several broad conclusions emerge. South and other productivity-enhancing online Asia scores poorly on many indicators of the business tools is relatively low. Innovation quality of the business environment, which tends to be concentrated in few mature firms greatly constrains firms’ productivity in gen- and is likely to consist of imitating existing eral and particularly limits the growth of products rather than developing new ones. firms with high levels of productivity. Greater investment in technology diffusion, Leveraging agglomeration and urbanization improved resource management, and the economies requires a reduction in the distor- development of skills that are complementary tions in product and factor markets, particu- to technology could play a critical role in larly in the barriers that limit the flow increasing innovation and thus boosting of resources between districts and states. productivity. Business Environment Challenges Continue to 3 Weigh on Firm Performance
uch of the macro and sectoral 55th position, but lagging well behind China challenges to productivity in South at 28. India is followed by Sri Lanka at 68, MAsia can be traced to a difficult Nepal at 100, Bhutan at 105, Bangladesh at operating environment for the region’s firms. 107, and Pakistan at 126. Although all South The business environment, also called the Asian economies (with the exception of investment climate, has received a great deal Bhutan) improved since 2014, many have yet of attention in the policy and empirical liter- to regain the ground they lost since 2007. ature as a major constraint on firm produc- Pakistan, for example, has lost 34 places and tivity in the region. Most studies define the India, despite making major advances, still investment climate as the environment that ranks 7 positions lower than it did in 2007. affects entrepreneurs’ ability to work effi- When it comes to the components of the ciently, such as the degree of difficulty in overall ranking, the most challenging areas in accessing production inputs and dealing with Bangladesh, Bhutan, Nepal, and Pakistan regulatory and legal requirements and the include inadequate supply of infrastruc- level of security for running operations and ture and the effects of corruption, while in obtaining payments. As argued by Hallward- Sri Lanka the most problematic factors are Driemeier (2007), an inefficient business inefficient government bureaucracy and lim- environment will lead to low and uncertain ited access to finance. Although India has returns on investment, dragging down over- recently achieved better GCI scores on the all productivity and possibly more than off- macroeconomic environment, institutions, setting technical improvements on the and infrastructure, its score remains ham- factory floor. pered by inadequate electricity supply and On average, countries in South Asia score poor technology readiness of its businesses. poorly on two major indexes used globally to In the World Bank’s 2016 Doing Business capture key aspects of the business environ- report, all the South Asian economies, with the ment: the Global Competitiveness Index exception of Bhutan, are ranked in the bottom (GCI) (World Economic Forum 2015) and half of the “Ease of Doing Business,” scale Doing Business (World Bank 2016). In the with an average ranking of 128 (figure 3.1). most recent GCI rankings (for 2015–16), By contrast, many of South Asia’s competitors India is the only South Asian country in are in the top half of the ranking, such as the top half of nearly 140 countries, in the Thailand (49), China (84), and Vietnam (90).
51 52 SOUTH ASIA’S TURN
FIGURE 3.1 South Asia countries lag behind comparators in business environment rankings
100
90
80
70
60
50 ercent P 40
30
20
10
0
Regional sia hina (rank 84) epal (rank 99) A hutan (rank 71)C N B ietnam (rank 90) Lanka (rank 107)India (rank 130) Thailand (rank 49) V akistan (rank 138) Sri outh P S verage (rank 128) angladesh (rankfghanistan 174) (rank 177) A B A
Distance to frontier 2016 Distance to frontier 2015
Source: World Bank 2016. Note: The distance to frontier score benchmarks economies with respect to regulatory practice, showing the absolute distance to the best performance in each Doing Business indicator. An economy’s distance to frontier score is indicated on a scale from 0 to 100, where 0 represents the worst performance and 100 the frontier. Tan bars represent countries in the South Asia region; blue bars represent countries outside the region.
Bhutan has the region’s highest rank, at 71, in Thailand and 4.9 percent in Vietnam. The followed by Nepal (99), Sri Lanka (107), areas with the most opportunity for improve- India (130), Pakistan (138), Bangladesh ment are Enforcing Contracts (where the (174), and Afghanistan (177). Compared to region’s average ranking is 143), Registering their 2015 rankings, only India and Sri Lanka Property (136), and Resolving Insolvency improved—moving from 134 to 130 and 113 (129). On average, resolving a commercial to 107, respectively—while all the other coun- dispute through the courts takes 1,077 days tries experienced a setback. in South Asia—almost twice the global aver- As measured by the Doing Business indi- age of 630 days. cators, on average, the South Asian econo- The region also performs poorly in logis- mies rank highest in Protecting Minority tics rankings. According to the World Bank’s Investors, with India and Pakistan respec- Logistics Performance Index (LPI), in 2014 tively ranked 8th and 25th globally. The South Asia had the lowest logistics perfor- region’s next-best performing category is mance among all developing regions because Starting a Business, but rankings here are of its poor quality of trade and transport- well below those of comparators: India related infrastructure, time-consuming clear- ranked 155 out of 189 economies, and ance processes, low quality of logistics although Sri Lanka ranked better (98), it nev- services, and lack of timeliness of shipments ertheless cost 18.7 percent of income per cap- compared to economies such as China, ita to set up a firm, compared to 6.4 percent Vietnam, and Thailand. As with the GCI BUSINESS ENVIRONMENT CHALLENGES CONTINUE TO WEIGH ON FIRM PERFORMANCE 53
rankings, none of the South Asian economies uncertainty in delivery schedules (Jordan and placed in the top 50 of the LPI, while China Kamphuis 2014). is ranked at 30, Thailand at 31, and Vietnam Moving from expert surveys and de jure at 48. Between 2007 and 2014, only Nepal requirements (such as the Doing Business and Sri Lanka were able to improve their report) to perceptions by firms (as shown in logistics performance by gaining 25 places the World Bank’s Enterprise Surveys) further and 3 places, respectively, in the overall LPI underscores the pervasiveness of the chal- rankings—but much ground still remains to lenges. In a 2006 Investment Climate be covered. Assessment, the World Bank argued that Poor logistics can sharply reduce efficiency. South Asian countries underperform compar- Lengthy and unpredictable delays in customs ators on many investment climate dimen- clearance can force firms to hold high inven- sions, including infrastructure and electricity tories (regional firms in auto parts, textiles, supply, access to finance, employee skills, and electronics, and heavy engineering report corruption (World Bank 2006a). Similar maintaining on average 27 percent higher results emerge from the most recent round of than necessary inventories to deal with uncer- World Bank Enterprise Surveys in which tain delivery times) and can impose delays in an average firm in South Asia consistently production and increased turn-around times. ranks each investment climate constraint as Delays caused by poor road infrastructure more binding than does an average firm in and lengthy interstate clearance processes China or Vietnam (table 3.1). Although per- have similar effects. For example, in India, formance varies substantially across countries crossing two state borders between origin and and indicators—pointing to significant poten- destination can add as much as a week to the tial for improvement by leveraging best
TABLE 3.1 Firms in South Asia are at a disadvantage with respect to investment climate Percent of firms that view an obstacle as a major or severe constraint
South Asia South Afghanistan Bangladesh Bhutan India Nepal Pakistan Sri Lanka (average) China Africa Turkey Vietnam
(2014) (2013) (2015) (2014) (2013) (2013) (2011) (2012) (2007) (2013) (2015) Access to finance 49 23 19 15 40 22 33 18 5 16 10 14 Political environment 76 76 12 16 85 34 13 28 1 3 13 3 Crime 58 8 1 5 14 35 7 10 1 38 8 5 Taxes 56 20 24 31 23 55 41 36 7 8 25 8 Corruption 62 49 4 36 42 64 15 42 1 17 12 5 Informality 33 9 10 17 29 12 28 14 7 11 14 11 Infrastructure 81 55 29 26 79 79 36 42 6 24 25 18 Electricity 66 52 14 21 69 75 26 35 3 21 18 4 Telecom 59 3 15 4 3 14 6 7 4 4 9 8 Transport 43 15 14 10 32 27 12 15 3 4 10 10 Labor regulations 11 3 15 11 3 12 13 12 1 6 6 4 Workforce education 53 16 14 9 9 23 16 13 2 9 10 8 Trade & customs 47 8 9 12 29 30 31 18 4 2 11 24
Source: World Bank calculations based on World Bank Enterprise Surveys. 54 SOUTH ASIA’S TURN
practices from within the region—the overall Bangladesh, Bhutan, India, Nepal, Pakistan, gap puts South Asia’s firms at a clear disad- and Sri Lanka, both output and value added vantage compared to select comparators in per worker are systematically lower when other parts of the world. firms face greater business environment con- Lessons from the case studies in this report straints (table 3.2). echo the findings from the surveys. Difficulties A restrictive business environment can be in importing goods, poor trade logistics, and particularly damaging to firms that have the high protection rates have made participation most to contribute to productivity growth in global markets more costly for firms across and job creation. Many investment climate the region, while outdated standards and constraints can be particularly burdensome restrictive regulations have limited competi- for small firms (Word Bank 2006a), limiting tion in the automotive and agribusiness sec- their ability to grow and create employment. tors. Difficulties in accessing well-located and Some evidence also shows that higher- well-serviced industrial land and poor avail- productivity firms in the region may actually ability of skilled workers have also emerged face greater constraints in accessing public as important bottlenecks to firm growth. services, suggesting that investment climate Although firms may have different capa- deficiencies are particularly binding on firms bilities to overcome various investment cli- that would grow more rapidly and create mate constraints, studies show that an more jobs in the absence of distortions (Carlin average firm in South Asia experiences a and Schaffer 2012). sizeable productivity loss from the poor The severity of investment climate obsta- investment climate. For example, Hallward- cles in the region and their adverse impact on Driemeier (2007) finds a significant negative productivity have given rise to a series of effect on total factor productivity (TFP) and wide-ranging reforms to address constraints investment rates of garment firms across the in each aspect of the business environment: region from customs delays, power outages, for example, a range of policy actions has poor access to finance, and limited connec- been proposed to various regional authori- tivity. In particular, the author shows that, if ties by the Asian Development Bank (2006), the business environment for firms in India the Asian Development Bank and World were the same as that in China, firm produc- Bank (2004), Afram and Salvi Del Pero tivity could be one percentage point higher. (2012), Ferrari and Dhingra (2009), OECD Analysis that approximates the approach of (2009), and the World Bank (2006a, 2006b, Hallward-Driemeier (2007) using the most 2008a, 2008b, 2009, 2010). Improving recent round of World Bank Enterprise the investment climate is also high on the Surveys shows that, by and large, invest- regional authorities’ own policy agendas: ment climate challenges continue to affect nearly every country in the region is taking firm performance in the region. Across a concrete steps to strengthen the business wide sample of manufacturing firms in environment (box 3.1).
TABLE 3.2 South Asia’s investment climate deficiencies constrain firm performance
Output per worker (log) Value added per worker (log) Losses from power outages (log) −0.024** −0.038*** Losses in transit (log) −0.079*** −0.036*** Improved access to finance 0.342*** 0.403*** Observations 4,566 4,498 R-squared 0.039 0.027 Sector dummies (number) 20 19
Source: World Bank calculations based on World Bank Enterprise Surveys. Note: ***p < 0.01, **p < 0.05, *p < 0.1. BUSINESS ENVIRONMENT CHALLENGES CONTINUE TO WEIGH ON FIRM PERFORMANCE 55
BOX 3.1 Efforts to improve the investment climate in South Asia
In Bangladesh, the authorities initiated a num- company and can start business five days earlier ber of reforms to address binding constraints on than under previous regulations. Utilities in private sector growth in the recent years. New Delhi and Mumbai undertook significant busi- business-friendly legislation includes: (1) the ness process reengineering, combining inspec- Economic Zones Act of 2010, modernizing tions and procedures to reduce the time required the country’s economic zones agenda, includ- for companies to get connected to the electrical ing the institutional setup, and allowing for grid and get on with their business. In addition, more efficient incentives and private participa- the central government called for all states to tion; (2) the Competition Law, which is meant automate registration processes, move toward to uphold a level playing field for businesses; effective single window systems, and implement and (3) a new Value Added Tax Law that eases risk-based inspection regimes that introduce the compliance mechanisms for businesses and self-certification and third-party audit schemes reduces discretionary exemptions. The authori- to lessen the burden of inspections on low- and ties also introduced regulatory reforms that medium-risk businesses. streamlined business registration and trade- In Sri Lanka, the authorities have recently mark and patent registration and simplified taken steps to eliminate obstacles to foreign direct trade licenses and construction permits. A total investment (FDI), including (1) up-front payment of 56 regulatory processes have been reformed of the land-lease tax for foreign companies; (2) in recent years including company, investment, elimination of minimum investment requirements tax, and trademark registrations; trade licenses in ICT, R&D, and vocational training; and (3) at local government levels; subordinate rules implementation of online processing of business under three different tax laws for better con- visas. Regulatory barriers to trade are being tract enforcement; and dispute resolution. To reduced through an agreement to ratify the World foster trade competitiveness, the authorities have Trade Organization (WTO) Trade Facilitation launched a trade information portal and intro- Agreement (which is the basis for a medium-term duced risk management in the clearance process, trade reform agenda) and creation of the National as well as taking preparatory steps to a new Trade Facilitation Committee, which will be the Customs Act, a national single window for trade, body in charge of leading trade facilitation and making multimodal transport effective for reform. Finally, a new Secured Transactions Act trade logistics. Responding to concerns on frag- will enable the use of movable assets as collateral mented policy coordination, these reforms are for bank loans, improving access to finance for being carried out in the context of a formal, small and medium enterprises (SMEs). structured public-private dialogue. In Pakistan, the authorities have recently In India, the authorities recently launched a embarked on a two-year plan to improve the new, ambitious program of regulatory reform. country’s Doing Business ranking to the top 100 In 2015, the authorities eliminated the minimum by 2018, by preparing a Doing Business Reform capital requirement and ended the requirement Strategy. The strategy provides reform recom- of obtaining a certificate to commence business mendations for all the Doing Business indicators operations. Indian entrepreneurs no longer need and provides institutions at the provincial and to deposit $1,629 (100,000 Indian rupees)— federal level with the mandate to carry out the equivalent to 111 percent of annual income per reforms. The strategy is currently being imple- capita—in order to start a local limited liability mented both at the federal and provincial levels,
(continues next page) 56 SOUTH ASIA’S TURN
BOX 3.1 Efforts to improve the investment climate in South Asia (continued)
having been endorsed by an Ease of Doing implemented regulations to improve access to Business Committee formed by the authorities. credit, payment of taxes, and financial interme- In addition, and complementing the big push on diation (capital markets and housing finance), as Doing Business, the Government of Pakistan well as reforms in financial transparency and also took a series of legislative actions and oversight of state-owned enterprises.
Although the importance of addressing critical constraint to productivity growth investment climate constraints in the region (the industry case studies, especially the is beyond question, the issues are well- study of agribusiness value chains in this known and policy pathways to address them chapter, show that these regulations restrict have been mapped out by various institu- investment and competition). Therefore, the tions, including the region’s own govern- discussion in the following chapters focuses ments. There is little that the current study on newer, less-researched determinants and can add to the vast body of knowledge on correlates of firm productivity in South Asia: the issue, save for emphasizing that industry- agglomeration economies, value chains, and specific business environment issues in the firm capabilities, including technology and form of product market regulations remain a innovation. BUSINESS ENVIRONMENT CHALLENGES CONTINUE TO WEIGH ON FIRM PERFORMANCE 57
INDUSTRY CASE STUDY A
Industry-specific business South Asia’s great agribusiness environment issues in agribusiness opportunity Agribusiness (including agriculture) accounts Agribusiness (agriculture, food process- for one-third of South Asia’s GDP, and its ing, food retail, and restaurants) accounts share is expected to double by 2030 as a result for one-third of South Asia’s GDP and is of income growth and urbanization, creating estimated to double in size over the next 15 millions of productive jobs outside agricul- years, reaching $1.5 trillion by 2030. The ture and positive backward links to farmers, increase will be driven by income growth and most of whom are small and vulnerable. urbanization, shifting the demand toward Numerous examples show how leading higher-value products (such as horticulture, private firms and market forces are needed to livestock, and packaged and processed foods) develop and diffuse the new higher-value and food-related services. Figure 3.2 shows products and services as well as to facilitate that spending on food starts to grow signifi- access by small-holders to knowledge, finance, cantly at South Asia’s current GDP per capita and markets. Government-led arrangements, level of $1,500. put in place to support farmers and achieve Investments in agro-food processing result food security, are no longer relevant and are in input and income multipliers higher than in fact counterproductive. Some of these in any other industry, and the employment arrangements (such as trade barriers, price effect is about 2.5 times that of other sectors caps on higher-value goods, restrictions on (World Bank 2014). The increased demand private agricultural markets, storage, and FDI for higher-value agricultural products and in retail) discourage private investment and interactions with increasingly sophisticated limit competition in the new high-value mar- buyers increase the productivity and income kets. Minimum support prices and govern- of poor farmers. Furthermore, some of the ment procurements encourage excessive agricultural products with the highest growth production of low-value commodities, and potential, such as dairy products, can dispro- large blanket subsidies lead to overuse of portionately benefit women. water and other unsustainable agricultural South Asia has also the opportunity practices. to develop its exports from a low base of
FIGURE 3.2 Food spending in relation to GDP per capita
6,000
5,000 3,643 4,000
3,000 1,402
2,000 930
1,000 258
Food spending per capita (dollars) 0 Low income 1,000Lower middle 4,000Upper middle 12,000 High income income income
GDP per capita ($, thousands)
Source: International Comparison Project (ICP) Database, World Bank, Washington, DC (accessed October 2015). 58 SOUTH ASIA’S TURN
FIGURE 3.3 Horticulture yield
700
600
500 (thousands)
400 hectare
300 per
200 grams
100 ector H 0 1991–93 2011–13 1991–93 2011–13 1991–93 2011–13
Apples Onions Tomatoes
South Asia Brazil China EU
Source: FAO Database, United Nations, New York (accessed August 2015), http://faostat.fao.org/.
FIGURE 3.4 Productivity in agro-food processing firms The small-holder challenge Agricultural production in South Asia is pre- $26,317 dominantly in the hands of small-scale farm- ers who are among the population’s most vulnerable. Lack of market information and $6,874 $7,482 $3,799 $2,499 logistical difficulties prevent small-scale pro- ducers from accessing markets efficiently. If India Sri Lanka Bangladesh China Vietnam the product is to be exported, the logistics and financing requirements are usually Output value per employee beyond their capabilities. Source: World Bank Enterprise Survey 2014. The natural resources challenge 3 percent of world agro-food trade compared Agriculture’s share of total fresh water use is to 14 percent for East Asia. Export bright very high (at, for example, 99 percent in spots already include basmati rice (India and Afghanistan and 90 percent in India), and Pakistan), dried fruit and nuts (Afghanistan), competition for water is increasing with the and branded tea (Sri Lanka). growth in manufacturing and household con- Seizing these opportunities will require sumption (FAO 2014). Much of the region’s addressing three related challenges. irrigation depends on groundwater. In India, for example, groundwater supplies 60 percent The productivity challenge of the water used for irrigation, of which 15 Agriculture yields remain low, especially in percent is overexploited. (World Bank 2013). high-potential high-value products such as Climate change is exacerbating the need to horticulture (figure 3.3). Postharvest losses improve water management and use. are high due to a lack of storage capacity (India’s Planning Commission estimated the How leading agribusiness firms can help warehousing shortage at 35 million tons), and address these challenges productivity in agro-food processing firms is much lower in South Asia than in East Asia Thirty-six leading agribusiness firms in (figure 3.4). South Asia were interviewed for this report. BUSINESS ENVIRONMENT CHALLENGES CONTINUE TO WEIGH ON FIRM PERFORMANCE 59
Their experience sheds light on how, together Links with downstream producers can also with market forces and government support, improve farmers’ access to finance. Access they can help develop and bring to market can be direct, through a variety of contract superior products as well as improve farmers’ farming arrangements, with inputs provided access to knowledge, finance, and markets. on the basis of agreements to sell the output at a later date. There are indirect benefits as Developing and marketing new high-value well, as banks are more willing to lend to products farmers that have a contractual arrangement Dilmah Tea is Sri Lanka’s largest exporter of with a processor. Godrej Agrovet has helped tea and the sixth largest tea company in the 50,000 small-holders in eight Indian states world, exporting premium tea to more than 80 access bank financing by guaranteeing prices countries. It started in 1974 with 18 employees and standardizing financing agreements and as of 2016 has 35,000. It has remained between the banks and the farmers. The gov- competitive because of its continuous invest- ernment and development partners can also ment in R&D and innovative marketing. help promote such arrangements as shown in KRBL Ltd., one of the world’s largest exporters Afghanistan (box 3.2). of basmati rice, successfully marketed the Pusa- 1121 basmati rice developed by the Indian How industry-specific business Agricultural Research Institute. This variety environment issues stand in the way became a best-seller in the Persian Gulf and India. Higher prices led to rapid adoption by Agricultural policies put in place in the 1960s farmers—by 2013 the new variety accounted to support farmers and achieve food security for 84 percent and 68 percent of basmati plant- are now limiting private investment and com- ings in Punjab and Haryana, respectively. petition in the new high-value markets and encouraging excessive production of low- Facilitating small-holder access to value commodities through unsustainable knowledge, finance, and markets agricultural practices. In Bangladesh, Aftab Bahumuki Farms Ltd. introduced contract farming for commercial Barriers to trade broiler chickens on an experimental basis in Average most-favored-nation rates applied 1991, working with a select group of 20 farm- on food products remain high in the region ers. The number of farmers involved had (at 33 percent in India and 26 percent in increased to 650 by 2003. Begum (2008) found Sri Lanka compared to 16 percent in China that contracted farmers achieved 30 percent and Vietnam). There are also instances of higher net returns than noncontracted farmers. inverted tariff structures that discourage In Bhutan, Mountain Hazelnut Ventures was domestic production, such as in Afghanistan established in 2010 to plant and process hazel- where tariffs on intermediate goods nuts. The company distributed tissue-cultured (10.2 percent) are higher than those on final hazelnut plantlets and planting material to goods (7 percent). Nontariff barriers (para- farmers. In three years of operation 2,000 hect- tariffs, ad hoc quantitative restrictions, and ares had been planted and 5,000 farmers cumbersome standards and custom proce- trained. In India, Desai F&V, by taking control dures) have long been cited as major reasons of the logistics and managing the process from for low intraregional trade. For example, fruit formation onward, succeeded in supply- revenue collection from supplementary ing remote urban centers with quality bananas duties exceeded revenue collection from cus- and in exporting grade “A” fruit to distant toms duties in fiscal 2012/13 in Bangladesh. markets in the Middle East. Pepsico provided potato varieties suitable for the processing of Restrictions on agricultural markets potato chips to thousands of Indian small- Outdated regulations, such as the 1939 holders supplying its processing facilities. Agriculture Produce Market Ordinance Act 60 SOUTH ASIA’S TURN
BOX 3.2 Promoting access to finance and backward links in Afghanistan
In Afghanistan, the Agriculture Development condition that they lend a portion of their loans Fund (ADF) began in 2010 as a $100 million to their suppliers-farmers. This approach lever- project of the U.S. Agency for International ages both the agro-food processors’ knowledge Development (USAID) to provide much-needed as well as the business leverage they have over long-term financing along agribusiness value their suppliers to ensure proper use of the funds chains. The ADF was rated the most success- and repayment. By contrast, commercial banks ful USAID project in Afghanistan, with a more in Afghanistan mostly cater to urban areas and than 95 percent reimbursement rate and 60,000 lack such access and knowledge. The ADF has farmers benefitting. The ADF is facing excess been operated by professionals with extensive demand, and discussions are underway to scale it experience in commercial banking and agribusi- up. The key innovation of the ADF is to provide ness and incorporates financial products that are long-term loans to agro-food processors on the fully compliant with Sharia.
in Pakistan and the 1956 Agriculture Produce facilities to resume the export of Alphonso Marketing Act in India, have hindered private mangoes to the European Union. investment in market and storage infrastruc- ture as well as contract farming. The central Minimum support prices and subsidies on government of India has promoted reforms fertilizers and water (along the lines of a Model Agriculture The drive to food security in the 1960s and Produce Marketing Act) since 2003, and 1970s led to multiple public interventions to while several states have introduced new leg- support the production of low-value com- islation to modify the 1956 Act since, the modities. By providing high returns and low changes have been partial and uneven; fur- risks, these measures discourage farmers from ther, in many cases changes have not been moving into higher-value crops. Minimum fully implemented, largely preserving the sta- support prices and fertilizer subsidies are also tus quo. Modern food retailers, which foster backed by large government procurement and innovation and competition along the value distribution systems plagued with inefficien- chain, are also restricted by regulations (such cies. India’s large subsidies get the most pub- as the FDI restrictions in India). Early reforms licity because of their size ($38 billion dollars are promising. For example, the 2012 reforms in the 2016 budget, excluding power and irri- by the Sindh province in Pakistan, which gation subsidies) and the fact that they abolished notified market areas and market are poorly targeted, but the subsidy bills committees and allowed private markets and of Bangladesh and Sri Lanka are also direct buying, led to the creation of the pri- enormous—nearly as large as all public vate chili trading platform in Kunri. expenditures on agriculture. Pakistan subsi- dizes fertilizer through low urea prices. Large Restrictions on prices and inadequate water subsidies (such as irrigation charges product standards that cover only 10 percent of the cost of water Investment in higher-value food products is in Punjab, Pakistan, and free power for water discouraged by price caps on items such as pumping in Punjab, India) lead to an unsus- milk and meat in Punjab, Pakistan. tainable overuse of water. Inadequate food safety certification systems The extended version of this case study is can have a serious negative effect on exports. available online at www.worldbank.org For example, India had to upgrade its testing /SouthAsiaCompetes. BUSINESS ENVIRONMENT CHALLENGES CONTINUE TO WEIGH ON FIRM PERFORMANCE 61
Note Jordan, L. S., and B. Kamphuis. 2014. Supply Chain Delays and Uncertainty in India: The 1. There are two exceptions in the case of India, Hidden Constraint on Manufacturing Growth. but neither is fully satisfactory. India’s Annual Washington, DC: World Bank. Survey of Industries (ASI) data are available Melitz, M. J., and S. Polanec. 2012. “Dynamic with panel identifiers starting from 1998–99; Olley-Pakes Productivity Decomposition with however, firms with fewer than 100 employees Entry and Exit.” NBER Working Paper 18182, are sampled only once every 4–5 years, mak- National Bureau of Economic Research, ing it difficult to determine whether a missing Cambridge, MA. firm exited or was not sampled. The Prowess Michaels, G., F. Rauch, and S. J. Redding. 2012. database by the Centre for Monitoring Indian “Urbanization and Structural Transformation,” Economy has data for more than 10,000 man- Quarterly Journal of Economics 127 (2): ufacturing firms dating back to 1990, but these 535–86. are mostly large, publicly listed companies. OECD (Organisation for Economic Co-operation and Development). 2009. Investment Policy Review: India. Paris: OECD. References Olley, G. S., and A. Pakes. 1996. “The Dynamics Afram, G., and A. Salvi Del Pero. 2012. Nepal’s of Productivity in the Telecommunications Investment Climate: Leveraging the Private Equipment Industry.” Econometrica 64 (6): Sector for Job Creation and Growth. 1263–97. Washington, DC: World Bank. Saia, A., D. Andrews, and S. Albrizio. 2015. Asian Development Bank. 2006. South Asia “Public Policy and Spillovers from the Global Economic Report. Manila: ADB. Productivity Frontier: Industry Level Evidence.” Asian Development Bank and World Bank. Economics Department Working Paper No. 2004. Sri Lanka: Improving the Rural and 1238, OECD, Paris. Urban Investment Climate. Washington, DC: Schumpeter, J. 1942. Capitalism, Socialism, and World Bank. Democracy. New York: Harper & Bros. Begum, I. A. 2008. “Prospects and Potentialities World Bank. 2006a. The Investment Climate in of Vertically Integrated Contract Farming South Asia. Washington, DC: World Bank. in Bangladesh,” PhD thesis, Department ———. 2006b. The Maldives: Sustaining Growth of Agricultural Development Economics, and Improving the Investment Climate. Hokkaido University, Japan. Washington, DC: World Bank. Cabral, L. 2007. “Small Firms in Portugal: A ———. 2008a. Afghanistan Investment Climate Selective Survey of Stylized Facts, Economic Assessment. Washington, DC: World Bank. Analysis, and Policy Implications.” Portuguese ———. 2008b. Harnessing Competitiveness Economic Journal 6 (1): 65–88. for Stronger Inclusive Growth: Bangladesh Carlin, W., and M. Schaffer. 2012. “Understanding Investment Climate Assessment. Washington, the Business Environment in South Asia.” DC: World Bank. Policy Research Working Paper 6160, World ———. 2009. Pakistan Investment Climate Bank, Washington, DC. Assessment. Washington, DC: World Bank. Desmet, K., and E. Rossi-Hansberg. 2009. ———. 2010. Bhutan Investment Climate “Spatial Growth and Industry Age.” Journal of Assessment Report: Vitalizing the Private Economic Theory 144 (6): 2477–2502. sector, Creating Jobs. Washington, DC: FAO (U.N. Food and Agricultural Organization). World Bank. 2014. Country Profiles, United Nations, ———. 2013. Turn Down the Heat. Washington, New York. DC: World Bank. Ferrari, A., and I. S. Dhingra. 2009. India’s ———. 2014. India—Accelerating Agricultural Investment Climate: Voices of Indian Business. Productivity Growth. Washington, DC: World Washington, DC: World Bank. Bank. Hallward-Driemeier, M. C. 2007. “Improving the ———. 2016. Doing Business 2016. Washington, Climate for Investment and Business in South DC: World Bank. Asia,” in South Asia Growth and Regional World Economic Forum. 2015. The Global Integration, ed. S. Ahmed and E. Ghani. Delhi: Competitiveness Report 2015–2016. Geneva: Macmillan India Ltd. World Economic Forum.
Productivity-Boosting Agglomeration Economies 4 Are Underleveraged
gglomeration economies arise when highly dependent on particular raw materials firms and people locate near one (such as wood in the furniture industry), but Aanother (for example, in cities and extends to sectors in which distance or loca- industrial clusters). In South Asia, where the tion is less important, such as the software concentration of economic activity is rela- industry. Michaels, Rauch, and Redding tively high, agglomeration economies have a (2012) show that the transformation of the major role to play in increasing productivity. American and Brazilian economies over the However, distortions in goods and factor last 100 years increased the concentration of markets prevent resources from flowing to resources in a few locations. Studies that more productive firms, particularly across control for geographical scale and borders, state and district borders. Therefore, despite such as Duranton and Overman (2005), also the statistically significant effect of agglom- show that economic activity is heavily eration on productivity—which today in concentrated. South Asia operates mainly through urban- South Asia is no exception. Measures of ization (cities) rather than through localiza- firm concentration (using the locational Gini, tion (clusters)—the full productivity benefits which is calculated in the same way as of agglomeration economies are yet to be the Gini coefficient is used to measure income realized. The following discussion develops inequality) in manufacturing are quite high in these observations in more detail. Bangladesh (0.53), India (0.67), and Sri Lanka (0.48).1 In India, the 5 largest districts account Economic activity in South Asia is for 18 percent of total employment, a share highly concentrated that has not changed appreciably over time, although which districts were in the top 5 has Economic activity tends to be geographically changed (table 4.1).2 In Bangladesh, the share concentrated. This is true in every country of the 5 largest districts (which is much regardless of industry considered or greater than in India in part because concentration measure used. For example, Bangladesh has only about 60 districts com- Rosenthal and Strange (2004) show that in pared to India’s nearly 400) increased more the United States, heavy geographic concen- than 10 percentage points between 1995 and tration of industries is not limited to sectors 2012, with most of the increase coming from
63 64 SOUTH ASIA’S TURN
TABLE 4.1 Employment across South Asia is concentrated in a few districts Share of top five districts as a percentage of total employment
Rank District Name District Name India 1991 2009 Rank in 1991 1 Greater Bombay 5.0 Madras 4.7 3 2 Nizamabad 4.1 Bangalore 4.4 4 3 Madras 3.3 Coimbatore 3.7 8 4 Bangalore 2.6 Mahendragarh/Gurgaon 2.9 77 5 24 Parganas (North) 2.5 Rupnagar/Patiala 2.8 6 Total (top five) 17.4 Total (top five) 18.5 Bangladesh 1995 2012 Rank in 1995 1 Dhaka 36.6 Dhaka 35.4 1 2 Chittagong 15.9 Gazipur 16.6 6 3 Narayanganj 8.6 Chittagong 16.2 2 4 Sirajganj 5.0 Narayanganj 9.3 3 5 Khulna 3.8 Sirajganj 2.8 4 Total (top five) 69.8 Total (top five) 80.4 Sri Lanka 1995 2009 Rank in 1995 1 Colombo 37.4 Gampaha 27.7 2 2 Gampaha 29.7 Colombo 24.1 1 3 Galle 3.9 Kurunegala 9.4 5 4 Kandy 3.8 Kalutara 7.2 9 5 Kurunegala 3.1 Kandy 5.9 4 Total (top five) 77.9 Total (top five) 74.2
Source: World Bank calculations based on Annual Survey of Industries (ASI) in India and Sri Lanka and Survey of Manufacturing Industries (SMI) in Bangladesh
areas outside Dhaka.3 In Sri Lanka, the 5 departs from the geographic pattern of man- largest districts account for three-quarters of ufacturing employment in the country as a total employment, a share that has declined whole, with larger values indicating greater somewhat since the mid-1990s.4 concentration of activity.5 The Ellison- In most countries, modern manufacturing Glaeser (Ellison and Glaeser 1997) Index has tended to develop initially in very concen- corrects for a potential bias in the raw index trated locations, often on a coast with access (the concentration value is larger in indus- to international markets. As development pro- tries with a small number of very large ceeds, large coastal cities become congested plants) and allows for comparisons across and expensive, while at the same time the industries. In any event, neither measure of scale externalities they offer dissipate as agglomeration has changed significantly over manufacturing processes become more time (with the exception of one year in India standardized. Manufacturing plants then and two years in Sri Lanka), suggesting that move first to suburban or nearby satellite more productive locations have generally not locations and thereafter to secondary cities in been successful in attracting additional the hinterland. Yet the degree of geographic resources at the expense of less productive concentration of manufacturing activities in locations, therefore inhibiting overall pro- South Asia has not changed substantially in ductivity growth. Ghani, Goswami, and Kerr the last two decades (table 4.2). (2012a) provide one explanation for this: The Raw Concentration Index in table 4.2 the authors suggest that a compositional measures the degree to which the geographic change may be under way, with larger plants pattern of employment in the industry moving out of South Asia’s cities and PRODUCTIVITY-BOOSTING AGGLOMERATION ECONOMIES ARE UNDERLEVERAGED 65
TABLE 4.2 Spatial concentration of manufacturing in South Asia has not changed much over time
India: State India: District Bangladesh: District Sri Lanka: District Raw Index EG Index Raw Index EG Index Raw Index EG Index Raw Index EG Index 1994 0.001 −0.000 0.006 0.005 — — — — 1996 −0.013 0.001 −0.012 0.001 — — 0.001 −0.009 1997 — — — — 0.029 0.056 0.007 −0.011 1998 — — — — — — 0.021 −0.001 1999 — — — — 0.008 −0.025 0.030 −0.010 2000 0.009 0.018* 0.012 0.021** — — 0.026 −0.011 2001 — — — — −0.006 0.005 0.044 0.056 2002 — — — — — — 0.044 0.056 2003 — — — — — — 0.019 −0.023 2005 — — — — 0.009 −0.036 — — 2006 −0.009 −0.001 −0.013 −0.005 — — 0.071** −0.504 2007 — — — — — — 0.073** −1.660*** 2008 — — — — — — −0.000 −0.330 2009 0.019 0.012 0.016 0.003 — — −0.004 −0.167 2012 — — — — 0.047 −0.084 — — N 351 348 351 348 123 118 263 260
Source: World Bank calculations based on Annual Survey of Industries (ASI) in India and Sri Lanka and Survey of Manufacturing Industries (SMI) in Bangladesh; Ellison-Glaeser (EG) 1997. Note: Table cells show the coefficients from a regression of concentration indexes on time dummies. A statistically significant positive (negative) coefficient indicates that spatial concentration in that year was significantly above (below) the mean. — = not available. *p < 0.1; **p < 0.05; ***p < 0.01.
an increasing number of smaller service from lower-productivity agriculture to higher- establishments moving in. productivity manufacturing and services. Evidence from the industry studies con- Agglomeration economies raise firms the importance of agglomeration. The firm productivity biggest benefits from agglomeration econo- mies were found in the automotive industry, Evidence shows that agglomeration is posi- where geographic proximity to the customer tively associated with firm performance in has supported efforts to upgrade product, South Asia (see, for example, Glaeser and process, and function. There is a high, robust Kerr 2009; Ghani, Kerr, and O’Connell 2011; correlation between productivity and the pro- and Mukim 2011). Proximity to cities had a pensity of automotive firms to be located next positive impact on nonfarm employment in to other automotive firms. Although this Nepal (Fafchamps and Shilpi 2005). Similarly, correlation may arise in part from high- households in Bangladesh with better access productivity firms electing to locate next to to major urban centers achieved higher each other, interviews suggest that firms returns to nonfarm activities (Deichmann, derive substantial benefits from clusters. The Shilpi, and Vakis 2008). The concentration of location of leading firms close to suppliers high-return economic activities around and clients also has been important in apparel Bangladesh’s growth poles (Dhaka and and agribusiness; it is, however, too early to Chittagong) led to higher productivity in the see agglomeration effects in the small elec- eastern part of the country (World Bank tronics sector in the region. 2008). Therefore, the ongoing urbanization Agglomeration is usually discussed in terms process in South Asia (as described in World of localization (firms in the same industry Bank 2015) is likely to increase productivity locating close to one another) or urbanization due to agglomeration, which may reinforce (firms in diverse industries locating in the same an increase in productivity as workers move area), as described in box 4.1. Other indicators 66 SOUTH ASIA’S TURN
have also been used to measure the productiv- logistical difficulties has largely determined ity benefits of agglomeration: for example, two the physical distribution of business activity. alternative measures of agglomeration at the For example, in the early 2000s, Maruti plant level—market access and proximity to Suzuki, India’s largest carmaker, relied on transport hubs—were correlated with produc- some 400 major suppliers located across the tivity in four of the nine sectors in India inves- country, some almost 2,500 kilometers away tigated by Lall, Shalizi, and Deichmann from its main plant in Haryana. Its total logis- (2004). In contrast, the authors find that mea- tics costs were up to four times as high as its sures of localization are correlated with pro- wage bill, and it had to carry large buffer ductivity in only two sectors and measures of stocks. By 2013, buffer stocks were brought urbanization in none. These findings, however, down to zero and logistics costs slashed after rely on estimating the effects of agglomeration almost all suppliers were required to build, economies jointly with the estimation of the warehouse, or locate within a few hours’ production function, potentially widening radius of the plant. Today, approximately uncertainty around the estimates (see, for 80 percent of Maruti Suzuki’s suppliers are example, Van Beveren 2012; Combes, located within a 100-kilometer radius. In Duranton, and Gobillon 2011). Moreover, the Pakistan’s apparel industry, leading firms and correlation between measures of urbanization their suppliers in leather apparel—already and market access can be high, as both indica- clustered in Sialkot and benefitting from labor tors are calculated using the size of urban pop- pooling, knowledge diffusion, and a critical ulation in the same district. mass of offerings to encourage international In South Asia, much of the economic activ- buyers to travel to this remote place— ity tends to cluster (localize) either naturally privately financed the construction of an or in response to policy distortions. In India’s international airport and exhibition center to automotive industry, the need to overcome further develop the cluster.
BOX 4.1 Agglomeration and productivity
Economic growth and geographical concentra- Recently, other sources have been suggested, tion of economic activities reinforce each other such as home-market effects (in which the con- (Baldwin and Martin 2004; Martin and centration of demand encourages agglomeration) Ottaviano 2001) through the forces of localiza- and economies of consumption (because consum- tion and urbanization. Localization economies ers enjoy variety). are the gains in productivity that result when In a survey of the literature, Rosenthal and firms in the same industry locate close to one Strange (2004) report that the elasticity of pro- another and benefit from sharing inputs, labor ductivity with respect to the size of the city or to market pooling, and knowledge spillovers the size of the industry generally lies between 3 (Marshall 1920). Urbanization economies are the and 8 percent. In a recent paper, Martin, Mayer, gains in productivity that result when firms in and Mayneris (2011) argue that, in the case of different industries locate in the same area France, localization economies dominate urban- (Jacobs 1969). The diverse range of industries in ization effects—at least in the short term—and a particular location enables firms to access sup- raise TFP by 5 to 10 percent. Much like this pliers from different industries, benefit from study, however, most of the empirical studies research and development (R&D) spillovers, or are based on developed countries, and there is access a generally higher-quality labor market. little evidence for developing economies. PRODUCTIVITY-BOOSTING AGGLOMERATION ECONOMIES ARE UNDERLEVERAGED 67
On the other hand, empirical evidence 0.2 percent.7 In contrast, in Bangladesh the shows that urbanization economies appear to effect has ranged from 0.2 to 0.4 percent. have had a larger impact on plant productivity In Sri Lanka, the data allow for even more in South Asia than localization economies stringent controls by using plant fixed effects (table 4.3). The estimation approach follows (table 4.4). With this specification, the impact the two-step strategy of Martin, Mayer, and of urbanization economies on plant produc- Mayneris (2011): first deriving plant-level esti- tivity is positive, particularly in the earlier mates of total factor productivity (TFP) and part of the sample. Looking at the overall pic- then assessing the impact of various aspects of ture from 1995 to 2009, a 10 percent increase agglomeration economies while also control- in the number of employees in other sec- ling for geographical location (measured at the tors leads to an increase in productivity of state level) and industrial diversity and the 0.86 percent—higher than the cross-sectional degree of competition (measured at the district estimates for India and Bangladesh. This sug- or industry level).6 Although statistically sig- gests that our results using cross-sectional nificant, the effects are relatively small and, in data could be underestimated, and the effects the case of India, seem to decline over time. In of urbanization economies in India and 1991, an increase of 10 percent in the number Bangladesh might be higher than the esti- of employees in sectors other than that in mates show. which a firm operates was associated with a Further disaggregation of the results reveals 0.5 percent increase in plant productivity in that the impact of urbanization on firm pro- India; by 2009, the productivity impact fell to ductivity is greater for more-productive firms
TABLE 4.3 Agglomeration economies are associated with higher firm productivity in India and Bangladesh
India Bangladesh Variables 1991 1994 1996 2000 2006 2009 1995 1997 1999 2001 2005 2012 Localization 0.01 0.00 −0.01 0.01 0.00 0.01 0.06*** 0.04* 0.01 0.03** 0.08*** 0.02 Urbanization 0.05*** 0.05*** 0.00 0.03** 0.00 0.02** 0.03** 0.03** 0.03** 0.02** 0.04*** 0.03*** Diversity 0.06*** 0.03** 0.11*** 0.05** 0.10*** 0.03 −0.02 −0.12 0.01 0.08** −0.18*** 0.03 Competition 0.02** 0.02*** 0.02* −0.01 0.02* 0.01 −0.12*** −0.02 −0.02 −0.05** −0.05** −0.02 Observations 41,539 42,565 40,876 25,435 39,462 36,020 3,417 3,178 2,931 3,940 3,155 7,119 R-squared 0.05 0.05 0.02 0.03 0.04 0.03 0.03 0.02 0.01 0.02 0.03 0.01
Source: World Bank calculations based on ASI in India and SMI in Bangladesh. Note: Constant and state dummies included but not shown. ***p < 0.01, **p < 0.05, *p < 0.1.
TABLE 4.4 Agglomeration economies boost firm performance in Sri Lanka
Variables 1995–2003 2006–09 1995–2009 Localization 0.0315 −0.0300 −0.0002 Urbanization 0.1916*** −0.0410 0.0855*** Diversity 0.1207** 0.0217 0.0560 Competition 0.0159 0.0420 0.0141 Observations 17,125 4,873 21,998 R-squared 0.0044 0.0027 0.0014 Number of firms 4,877 3,528 8,405
Source: World Bank calculations based on ASI. Note: Constant term included but not shown. ***p < 0.01, **p < 0.05, *p < 0.1. 68 SOUTH ASIA’S TURN
and is not significantly different from zero for from addressing endogeneity when estimating less-productive firms.8 For example, in the impact of agglomeration economies on Bangladesh, the impact of urbanization at the productivity. Furthermore, the analysis did 75th percentile of the firm productivity distri- not take into account the potentially negative bution is more than twice as large as that at effects of agglomeration, such as congestion. the 50th percentile (table 4.5). In contrast to Addressing negative externalities might earlier results, this approach also identifies indicate that agglomeration promotes produc- significant positive effects of localization tivity, regardless of whether the effects are economies on firms’ productivity, although through localization or urbanization. results differ qualitatively for the two coun- tries. Less-productive firms in Bangladesh Resources do not flow easily to benefit from localization effects, but the most- more-productive firms productive firms benefit in India. Results using other years show similar results. Despite the productivity-enhancing benefits of Overall, these findings show that agglom- agglomeration, there appear to be significant eration economies matter in South Asia, and barriers to resources moving freely across the magnitude of the estimated impact of internal geographical borders in South Asian agglomeration on firm productivity is similar countries. Duranton et al. (2015) propose to that found in previous research that an empirically motivated counterpart to the focused on developed countries. But unlike misallocation measure of Hsieh and Klenow the evidence for high-income countries, (2009), defining misallocation as the (negative urbanization economies in South Asia seem to of) correlation between firm productivity and matter more than localization economies, some measure of firm size, whether output, although the two are not mutually exclusive. employment, the use of capital, or land and Evidence from the case studies documents the other resources. With this definition, Duranton emergence of clusters in or around cities—for et al. (2015) are able to decompose overall example, the apparel cluster in Lahore and misallocation at the country level into contri- the automotive clusters in Pune and butions from different factors of production, Aurangabad. It is important to emphasize as well as to distinguish between misallocation that these correlations do not indicate causal- of resources within entities (for example, states ity, because data limitations have prevented us or districts) and between them.
TABLE 4.5 More-productive firms derive greater benefits from agglomeration economies Estimation results at different percentiles of firm distribution
India (2009) Bangladesh (2012) Variables 25th percentile 50th percentile 75th percentile 25th percentile 50th percentile 75th percentile Localization −0.0060 −0.0010 0.0109* 0.0179*** 0.0049 −0.0079 Urbanization 0.0107 0.0207** 0.0119* 0.0022 0.0280*** 0.0674*** Diversity 0.0385*** 0.0505*** 0.0293* 0.0930*** 0.0308 0.0220 Competition 0.0299*** 0.0124* 0.0023 0.0231* 0.0063 −0.0150 State dummiesa Yes Yes Yes No No No Testing (p-value) 25 = 75 25 = 50 50 = 75 25 = 75 25 = 50 50 = 75 Localization (%) 0.7 25.5 2.3 4.7 17.7 22.6 Urbanization (%) 85.0 7.0 10.3 0.0 0.0 0.0 Observations 36,020 36,020 36,020 7,119 7,119 7,119
Source: World Bank calculations based on ASI in India and SMI in Bangladesh. Note: Constant term included but not shown. a. Model did not converge when using Bangladesh regional dummies. ***p < 0.01, **p < 0.05, *p < 0.1. PRODUCTIVITY-BOOSTING AGGLOMERATION ECONOMIES ARE UNDERLEVERAGED 69
Calculating misallocation in this way for FIGURE 4.1 “Thick” district borders prevent efficient allocation of three South Asian countries for which requi- resources site data are available reveals that, in most Decomposition of total misallocation into between- and within-district cases, district or state borders in the region components are “thick”—that is, impediments to efficient allocation of resources between districts are Output stronger than distortions within districts. Labor
This holds true for all countries in markets angladesh for goods, and in India in markets for labor B Capital and capital as well (as evidenced by the rela- Output tively small contribution of the between- district component to overall efficiency Labor plotted in figure 4.1). Moreover, the results Capital in figure 4.1 show that across South Asia, factor (labor and capital) markets are more Output
distorted than goods markets (as evidenced anka India
L Labor ri by more negative values for misallocation of S output than for misallocation of labor and Capital capital— a more negative number means less –1.2 –0.9 –0.6 –0.3 0 0.3 0.6 misallocation). Within districts Between districts Although the empirically motivated mea- Source: World Bank calculations based on ASI in India and SMI in Bangladesh. sure of misallocation used here does not read- Note: A more negative number means more efficient allocation of resources (less misallocation); ily lend itself to calculating productivity gains zero means no correlation between productivity and output or employment; a positive number means less-productive firms attract more labor or capital than more-productive firms. Numbers in from reduced misallocation (as done by Hsieh the figure are averages across misallocation indexes calculated for each of the years for which data and Klenow 2009), regression analysis by were available (Bangladesh: 1995, 1997, 1999, 2000, 2005, 2012; India: 1991, 1994, 1996, 2000, 2005, 2010; Sri Lanka: 1995–2003, 2006–09). Duranton et al. (2015) for India suggests that a 1 percent decrease in the index of employment than in manufacturing). Therefore, reducing misallocation could raise output per worker in factor market distortions to improve the abil- manufacturing by about 0.3 percent and in ity of more-productive firms to access inputs services by about 0.9 percent (because labor is could have important consequences for overall a relatively more important input in services productivity. 70 SOUTH ASIA’S TURN
INDUSTRY CASE STUDY B contribution to global growth. Global trade in electronic products, including communica- Better cities and trade logistics tions and information communication tech- are needed for the electronics nology (ICT) equipment and electronics-based sector to thrive consumer products, was estimated at $1.4 trillion in 2012, having grown 5.9 percent per Electronics is one of the largest and fastest year between 2008 and 2012.9 Electronics growing industries in the world and has production is an important source of employ- played an important role in the development ment with 18 million people worldwide in trajectories of several newly industrialized 2010 (ILO 2014a). economies. Surprisingly, South Asia is not The sector presents growth opportunities currently a significant player in the sector for developing countries. An important fea- despite very competitive labor costs and the ture of the sector is that production is highly fact that leading firms are achieving world- fragmented, with value often added in a vari- class productivity in the region. ety of countries before goods and services Missing are urban ecosystems that would make their way to end consumers. The ability provide thick markets for skilled labor, the ame- to shift parts of the value chain to low-cost nities that make them attractive to engineers locations has created opportunities for devel- together with large tracts of industrial land for oping countries to participate. Electronics clusters to thrive, and world-class logistics to companies from the developed world first enable the import and export of hundreds of started relocating to Malaysia, Singapore, components and products seamlessly. Some Taiwan, and Thailand during the 1970s and locations in South Asia are on the edge of being early 1980s, followed by China, Indonesia, able to provide such conditions. Electronics and the Philippines, primarily to take clusters have emerged in Bangalore, Chennai, advantage of lower labor costs. In recent Colombo, and Delhi-Noida, with growing years, Vietnam has become an important pro- interest from leading global investors for ducer for similar reasons. In 2013, informa- Chittagong and Mumbai-Pune. Much progress tion technology and electronics accounted for has been achieved on the regulatory and policy 7 percent of Vietnam’s exports. Asia has been fronts, but some issues related to “inverted a major beneficiary and has become an impor- tariffs” remain to be addressed in India. tant manufacturing hub, mainly due to its low Progress has also been made to make customs labor costs, established supply base, and prox- and ports more efficient, but more efforts are imity to key final markets (ILO 2014b). needed to bring them to East Asia’s level of effi- However, South Asia has yet to benefit ciency. Similarly, efforts are being made to from the global shift of electronics manufac- improve internal trade logistics. Governments turing. Other countries that started from a are investing in skills in partnership with the much weaker position have forged ahead and private sector. The livability of large cities is a established themselves as new global players well-known challenge and steps are being taken in electronics manufacturing exports. For to address it. In Bangladesh, a key constraint is example, electronics exports from South Asia the limited supply of large tracts of well-located are almost invisible compared to those from and readily available industrial land for large East Asian countries (figure 4.2). India is only investors and their suppliers. the 14th-largest electronics producer globally, The opportunity: South Asia on the behind countries such as Mexico (8th), Brazil edge of becoming globally competitive (10th), and Thailand (12th). in one of the world’s most important This is surprising both because of South and fastest-growing industries Asia’s growing labor cost advantage over East Asia and because leading firms in South Asia The electronics sector, one of the world’s larg- achieve productivity performance comparable est industrial sectors, has made a substantial to that of Chinese firms. Samsung reported to PRODUCTIVITY-BOOSTING AGGLOMERATION ECONOMIES ARE UNDERLEVERAGED 71
the authors that their Noida, India, plant ranks facilities next to Samsung following its invest- second in efficiency out of 30 comparable ment in Vietnam, and similar dynamics Samsung plants around the world. Similarly, can be observed at Samsung’s main plants in the World Bank Enterprise Surveys (2011, Noida and Chennai, India. Clustering next to 2012) show that productivity in the electronics large cities provides access to a large pool of sector is higher in Sri Lanka ($24,701) than in skilled workers and engineers as well as to the China ($22,382). Cost comparisons, obtained ancillary service providers required by this from a global manufacturer with facilities fast-moving, constantly innovating industry. around the world, show that manufacturing Thus by its nature, the electronics industry costs in India are 80 percent of those in the requires urban ecosystems that combine United States and, importantly, slightly lower than those in China and Vietnam (figure 4.3). FIGURE 4.2 Electronics production is low in South Asia
What will enable South Asia to succeed 300 in electronics? 250 Achieving low production costs at the plant level is not enough to be competitive in the 200 electronics industry. An electronic product is the result of the assembly of hundreds of 150
components, produced by skilled workers world ($) and engineers, from multiple suppliers in vari- 100 er capita exports er capita the to ous locations and countries. P Industrial clustering is a key feature of the 50 electronics industry because it helps reduce 0 transaction and logistic costs among dozens hina India anka C of related firms. It also helps facilitate quality akistan L ietnam P Sri V angladesh control and reduce transport-related uncer- B tainties in the supply chain. For example, 76 Source: World Bank calculations based on UN COMTRADE database and World Bank suppliers from the Republic of Korea located Enterprise Surveys.
FIGURE 4.3 India has a cost advantage in labor-intensive processes
1.0
0.5 . total cost S . indexed to U ost breakdown C 0 United States India Vietnam Shanghai, China Sales and administration Utilities and floor space Machinery, equipment, and tooling Semiskilled labor Unskilled labor Purchased parts Raw material
Note: Data drawn from auto parts manufacturer but representative of costs in electronics sector. 72 SOUTH ASIA’S TURN
a deep pool of skilled workers, amenities that cluster, especially to link it with a port and to make their locations attractive to engineers, major markets. Small and medium enterprises and large tracts of industrial land that can (SMEs) in particular require common facili- host clusters with very good internal and ties for R&D and testing, waste dumping, external connections, in particular to world- and recycling. Provisions for worker housing class ports. Because the industry is so globally within or close to the cluster is important. competitive, it also requires a low cost of Manufacturing facilities tend to be located doing business and an enabling trade policy outside urban areas to take advantage of environment. lower land costs. However, ensuring an ade- South Asia has made significant progress quate supply of labor requires companies to in regulation and trade policies, although pay to transport workers to the sites or to cre- some inverted tariff issues are left to be ate facilities for them to stay nearby, both of addressed in India (Subramanian and Modi which raise costs. 2015). The main challenge is to provide the Clustering in South Asia is difficult because urban ecosystems (including seamless internal buying land in suitable locations is often an and external trade logistics) for world-class arduous and expensive task. Verification of electronics clusters to emerge and thrive. title is complex, and procedures for purchas- ing land take time. It is difficult for large com- Barriers to clustering panies to assemble enough small plots of Although there is clustering around a few land. The scarcity of adequate land is also locations, access to land remains a challenge. reflected in very high prices. Companies pre- The electronics industry in India initially fer to locate close to major markets or ports, grew in the 1990s around three major cen- but some Indian and Sri Lankan periurban ters, Bangalore, Chennai, and Delhi-Noida. areas are among the most expensive in the Bangalore emerged as a hub early on, with world (Saleman and Jordan 2013). The prac- major public sector plants in defense tical solution to these issues has been indus- and telecommunication. In recent times, trial zones, which played a central role in the Bangalore also has attracted private sector development of manufacturing, and electron- firms in computer and industrial products. ics in particular, in East Asia. The lack of The Delhi region, in particular, has a large readily available and well-located industrial concentration of small-scale factories mak- land is probably the main constraint on the ing consumer electronic products and com- development of the sector in Bangladesh, puters. More recently, Hyderabad, Chennai, which came very close to breaking into the and the industrial corridor between Mumbai global electronics industry, as Vietnam has and Pune have become important manufac- done (box 4.2). turing locations. Few industrial areas have been able to pro- External and internal trade logistics issues vide what investors require in the right loca- Countries that are able to achieve faster turn- tions. Interviews suggest that manufacturers around times gain a significant competitive seeking to link to global supply chains prefer advantage, especially for more innovative, to locate in areas that are close to ports, to cutting-edge products. This is where South speed up the movement of goods along sup- Asia is at a disadvantage. Processing time is ply chains and reduce dependence on local higher in India than in China, Singapore, and infrastructure. Setting up clusters in such Taiwan (figure 4.4). areas—either as special enterprise zones Although much progress has been made, (SEZs) or industrial parks—that are large South Asian firms continue to report long and enough to house lead firms and suppliers unpredictable delays in customs clearance. would help attract electronics manufacturers. In India, the average time reported to Investors would also like to see world-class clear customs varied from 2 to 10 days for infrastructure developed around and within a large firms, and 14 to 21 days for SMEs. PRODUCTIVITY-BOOSTING AGGLOMERATION ECONOMIES ARE UNDERLEVERAGED 73
BOX 4.2 How Bangladesh missed the opportunity to break into the global electronics industry
In 2011, Samsung requested 250 acres in an empty 2,500-acre Korean Export Processing export processing zone (EPZ) to develop an Zone in Chittagong has been in dispute for electronics hub in Chittagong, Bangladesh more than 15 years. By contrast, Vietnam has (a $1.25 billion investment, with jobs for up to been able to provide large, readily available 50,000 workers). The investment did not mate- tracts of land to large investors and their rialize because no tract of land that large was suppliers. Samsung was able to locate there available in the area controlled by the with 76 of its Korean suppliers and now directly Bangladesh Export Processing Zones Authority employs 100,000 workers. (BEPZA), and the use of the land in the mostly Source: Ahsan 2014.
Remaining issues include ambiguities in pro- FIGURE 4.4 Process times for information technology hardware duct classification and difficulties in obtaining and electronics exemptions from import tariffs on raw mate- 40 rials and parts and components. Grievances 35 also can take considerable time, and compa- 30 nies fear reprisals, such as losing their trusted- 25 trader credentials. One firm stated that “[t]he 20 customs bureaucracy is very difficult to han- 15 umber of days dle when we import various equipment and N 10 that creates a big disincentive for anyone ven- 5 turing into this market.” Another interviewee 0 said that “[t]o gain one rupee in customs Total cycle time Time to market Time to market duties the country is losing thousands.” (domestic) (international) Issues also remain with internal logistics. India Singapore Taiwan China Indian firms reported that it takes 11 days for Source: D&B Analysis 2012. a container to travel from Shanghai to Mumbai but 20 days for it to travel from Mumbai to Delhi. A survey shows that a quarter of the journey is spent at checkposts, Access to skilled workers state borders, city entrances, and other regu- The availability of cheap and adaptable labor latory stops (figure 4.5). In order to deal with is one factor that makes South Asia attractive the resulting uncertainty, firms in four indus- as a manufacturing destination. However, tries (auto components, textiles, electronics, capitalizing on South Asia’s advantage and heavy engineering) report maintaining in workers requires significant investment in inventories 27 percent higher, on average, training and improving productivity. The than necessary. Total logistics costs, including returns can be large—international evidence inventory costs and lost sales, account for shows that a 1 percent increase in training is 14 percent of total costs for electronics firms, associated with 0.6 percent increase in value high by international standards (Jordan and added per hour (Dearden, Reed, and Van Kamphuis 2014). The recently passed reform Reenen 2006). The question is who makes this of the unified gross sales tax in India should investment? In South Asia, public investment improve matters considerably. in training has been low and of poor quality, 74 SOUTH ASIA’S TURN
FIGURE 4.5 Logistics issues in India
a. Percentage of firms identifying b. Average distance travelled by various logistics issues delivery trucks 1100 India’s road infrastructure not adequate 1000 Customs department not 900 efficient for imports State border checkpost 800 clearances 700 Nonfreight related (supplier or planning related) 600 Transporters or logistics provider 500
not efficient or professional Km per day 400 Rain or climatic conditions 300 Supplier’s lack of control over the logistics 200 infrastructure (doesn’t own trucks) 100 Use of older vehicle by transporters 0 0 5 10 15 20 25 30 35 40 45 50 55 60 65 United States Australia Brazil World India Percent average
Source: Jordan and Kamphuis 2014.
comparing unfavorably with other competitor Further, companies can reduce annual taxable nations. For example, vocational education income up to 10 percent through spending on programs in India can accommodate only R&D. Sharing the costs of skills education 5 percent of secondary school graduates, while between the state and private companies China has the infrastructure to train half of all can significantly encourage investment, espe- secondary school graduates. The quality of cially for companies with long-term plans. training is also an issue. Across the region, the development of new programs and curricula is South Asia is on the edge of breaking difficult in public institutions. Quality is fur- into the global electronics industry ther hampered by the lack of industry partici- pation in training (BCG 2013). Samsung As shown by the missed Samsung opportu- reported that it takes a full year of training to nity in Bangladesh (box 4.2), South Asia is bring its workers in India to global standards. very close to making it big in the global Tos Lanka, Sri Lanka’s leading electronics electronics industry. This will require con- company has also had to make large invest- tinued improvements in urban ecosystems ments in training (box 4.3). with respect to access to industrial land, In China, the government supports voca- internal and external trade logistics, and tional education with extensive industry skilled labor. participation. Curricula are flexible and, to Global investors are anticipating the shift make sure that faculty always keep abreast of and are on the move. Several multinational the latest industry practices, the Chinese gov- electronics firms (such as Samsung, HP, IBM, ernment has made it compulsory for voca- Motorola, Lenovo, Flextronics, and Foxconn) tional trainers to spend at least one month are present, or have announced plans to invest every year working in manufacturing compa- in the region (box 4.4), and many large firms nies. To facilitate access to labor, the govern- have set up R&D centers with world-class ment also supports employee housing next to capabilities in South Asian countries to sup- or within well-located industrial parks port global operations. This activity has been that also have technical and vocational col- encouraged by South Asia’s large, fast- leges and secondary schools. Vietnam growing markets (the electronics market in provides subsidized training for employees. India, for example, is expected to grow at Programs include soft skills, technical English, 24 percent per year to reach a market size of technical skills, and on-demand training. $400 billion by 2020) (Ernst 2014) and the PRODUCTIVITY-BOOSTING AGGLOMERATION ECONOMIES ARE UNDERLEVERAGED 75
BOX 4.3 How Tos Lanka is developing the skills of its workforce
Tos Lanka is Sri Lanka’s largest electronic assem- in training and development of staff to manufac- bly company. It commenced operations at the ture electronic products and components to Biyagama EPZ in 1998, with an initial investment international standards. Workers at Tos Lanka of SL Rs 220 million. Tos specializes in the surface- undergo training in Japan for a period of between mount technology assembly of printed circuit three and twelve months. The majority of the boards, electronic guitar tuners and effectors, and workforce has been trained in Japan in quality- coils and electronic components for the automo- oriented manufacturing processes. tive industry. The products are exported to Japan, The company has also established its own the United States, and the European Union. R&D section and looks forward to accessing the The factory has manual and automated elec- huge Indian market through the Comprehensive tronic assembly lines, supported by chip mounting Economic Partnership Agreement between India and wave soldering plants, together with extensive and Sri Lanka. testing facilities. Female workers dominate the 240-strong workforce. The company has invested Source: Tos Lanka.
BOX 4.4 Foxconn enters India
Foxconn, the world’s largest contract electronics has the bulk of its manufacturing capacity. manufacturer, has announced plans to open more Foxconn has also announced a joint venture with than 10 plants and employ a million workers in Xiaomi of China to produce devices in Andhra India by 2020. This will include R&D and hi- Pradesh. tech manufacturing facilities and would represent a major diversification away from China, where it Source: Rai 2015.
potential of the South Asian diaspora, which technologies. Indian companies such as has deep knowledge and extensive networks Micromaxx, Deltron, TVS Electronics, and in the global electronics industry. Sahasra are increasing their presence. Indian exports of components and pro- The extended version of this case study ducts have started to grow in areas such as is available online at www.worldbank.org mobile telephones, audio players, and display /SouthAsiaCompetes. 76 SOUTH ASIA’S TURN
sz Notes Diversityt =
−1 ⎛ ′ 2 ⎞ 1. These are simple averages of locational ⎛ employeessz ⎞ ln⎜ t ⎟ Gini coefficients calculated at the two-digit ⎜∑⎜ z − sz ⎟ ⎟ ⎝ ⎝ employeest employeest ⎠ ⎠ International Standard Industrial Classification ss′≠ (ISIC) industry level. −1 ⎛ 2 ⎞ 2. Only two of the top five districts in 1991 ⎛ sz ⎞ sz = ⎜ employeesjt ⎟ remained in the top five in 2009, and employ- Competitiont ln ∑ ⎜ ⎟ ⎜ ⎝ employeessz ⎠ ⎟ ⎝ ∈ sz t ⎠ ment in the five largest districts in 1991, which jSt constituted 17.4 percent of total employment, amounted to a much lower 14.0 percent in 7. Note that, due to data limitations, this report 2009. For the purposes of spatial analysis, the cannot adequately address the issue of input report consistently uses India’s 1989 districts, endogeneity (especially of capital) in the mapping new districts in later years to their production function following the traditional 1989 “parent” districts. approaches in the literature, such as Levinsohn 3. Dhaka continues to dominate in levels, and Petrin (2003) and Olley and Pakes (1996). accounting for more than 35 percent of total However, Van Beveren (2012) showed that employment. differences between most parametric or semi- 4. Sri Lanka has 25 districts, but 7 of these parametric methods to estimate TFP (including had too few firms to carry out the analysis. Levinsohn and Petrin and Olley and Pakes) and Therefore data for these districts (Jaffna, ordinary least squares (OLS) are minimal. Mannar, Vavuniya, Mullativu, Batticaloa, Therefore, the approach taken here is to Ampara, and Trincomalee) have been combined estimate the production function by OLS, into a single “residual” district, resulting in a allowing the parameters for capital (α) and total of 18 districts. labor (β) to vary for each two-digit ISIC sector. N 2 ⎛ L L ⎞ 8. To assess this, the report estimates quantile =−is, s , 5. This index is calculated as Gi ∑⎝⎜ ⎠⎟ regressions at the 25th, 50th, and 75th per- Li L s=1 centiles of the firm productivity distribution where i refers to a two-digit ISIC industry, for the last available year. s is location (district or state), and L is 9. World Bank calculations based on United employment. Nations COMTRADE database. The classifi- 6. Localization economies are defined as the log cation of electronic products is from Sturgeon of 1 plus the sum of all employees from sec- and Memedovic (2011). tors in region z other than the employees in plant i, while urbanization economies are defined as the log of 1 plus the sum of all References employees in region z other than employees Ahsan, Badrul. 2014. “Dearth of Land in BD in sector s (1 is added to ensure the inclu- Redirects FDI,” Financial Express, November 16. sion of all plants in the estimation; other- Baldwin, R. E., and P. Martin. 2004. “Agglomeration wise the existence of only one plant in a and Regional Growth.” In Vol. 4 of Handbook particular region would be discarded of Regional and Urban Economics, edited by because log of 0 is not defined). More spe- J. Vernon Henderson and Jacques-François cifically, the indexes are defined as follows Thisse, 2671–2711. Amsterdam: Elsevier B.V. for each plant i, sector s, region z, and time BCG (Boston Consulting Group). 2013. “People period t: Productivity. Key to Indian Manufacturing Competitiveness.” Boston Consulting Group. ⎛ sz ⎞ employeest − http://www.bcg.com/en-in/perspectives Localizationsz = ln⎜ ⎟ it sz /28825. ⎝ employeesit + 1⎠ Combes, P. P., G. Duranton, and L. Gobillon. 2011. “The Identification of Agglomeration ⎛ z ⎞ employeest − Urbanizationsz = ln⎜ ⎟ Economies.” Journal of Economic Geography t sz ⎝ employeest + 1⎠ 11 (2): 253–66. PRODUCTIVITY-BOOSTING AGGLOMERATION ECONOMIES ARE UNDERLEVERAGED 77
Dearden, L., H. Reed, and J. Van Reenen. 2006. of the Spatial Distribution Can We Explain?” “The Impact of Training on Productivity and Journal of Economics and Management Wages: Evidence from British Panel Data.” Strategy 18 (3): 623–63. Oxford Bulletin of Economics and Statistics Hsieh, C., and P. J. Klenow. 2009. “Misallocation 68 (4): 397–540. and Manufacturing TFP in China and India.” Deichmann, U., F. Shilpi, and R. Vakis. 2008. Quarterly Journal of Economics 124 (4): “Spatial Specialization and Farm-Nonfarm 1403–48. Linkages.” Policy Research Working Paper ILO (International Labour Organization). 2014a. 4611, World Bank, Washington, DC. “Better Work: Electronics Feasibility Study.” D&B Analysis. 2012. “IT Hardware & Electronics: ILO. http://betterwork.com/global/wp-conten Productivity & Efficiency Benchmarking.” /uploads/Better-work-Electronics -Feasibility Government of India, Ministry of Science and -Study-Executive-Summary.pdf. Technology, Department of Scientific and ———. 2014b. Ups and Downs in the Electronics Industrial Research, http://www.dsir.gov.in Industry: Fluctuating Production and the Use /reports/isr1/IT%20Hardware%20 and%20 of Temporary and Other Forms of Employment. Electronics/4_9.pdf. Geneva: ILO. Duranton, G., E. Ghani, A. Grover Goswami, and Jacobs, J. 1969. The Economy of Cities. New York: W. Kerr. 2015. “The Misallocation of Land Vintage Books. and Other Factors of Production in India.” Jordan L. S., and B. Kamphuis. 2014. “Supply Policy Research Working Paper 7221, World Chain Delays and Uncertainty in India: The Bank, Washington, DC. Hidden Constraint on Manufacturing Growth.” Duranton, G., and H. G. Overman. 2005. Working Paper ACS 14223, World Bank, “Testing for Localization Using Micro- Washington DC. Geographic Data.” Review of Economic Lall, S.V., Z. Shalizi, and U. Deichmann. 2004. Studies 72 (4): 1077–1106. “Agglomeration Economies and Productivity Ellison, G., and E. L. Glaeser. 1997. “Geographic in Indian Industry.” Journal of Development Concentration in U.S. Manufacturing Economics 73 (2): 643–73. Industries: A Dartboard Approach.” Journal of Levinsohn, J., and A. Petrin 2003. “Estimating Political Economy 105 (5): 889–927. Production Functions Using Inputs to Control Ernst, Dieter. 2014. “Republic of India for Unobservables.” Review of Economic Manufacturing Plan Implementation: Fast Studies 70 (2): 317–41. Tracking India’s Electronics Manufacturing Marshall, A. 1920. Principles of Economics. Industry—Business Environment and Industrial London: Macmillan. Policy.” World Bank, Washington, DC. Martin, P., and G. I. Ottaviano. 2001. “Growth Fafchamps, M. and F. Shilpi. 2005. “Cities and and Agglomeration.” International Economic Specialisation: Evidence from South Asia.” Review 42 (4): 947–68. Economic Journal 115 (503): 477–504. Martin, P., T. Mayer, and F. Mayneris. 2011. Ghani, E., A. Grover Goswami, and W. R. Kerr. “Spatial Concentration and Plant-Level 2012a. “Is India’s Manufacturing Sector Moving Productivity in France.” Journal of Urban Away From Cities?” NBER Working Paper Economics 69 (2): 182–95. 17992, National Bureau of Economic Research, Michaels G., F. Rauch, and S. J. Redding. 2012. Cambridge, MA. “Urbanization and Structural Transformation.” ———. 2012b. “Highway to Success: The Impact Quarterly Journal of Economics 127 (2): of the Golden Quadrilateral Project for the 535–86. Location and Performance of Indian Mukim, M. 2011. “Industry and the Urge to Manufacturing.” NBER Working Paper Cluster: A Study of the Informal Sector in 18524, National Bureau of Economic India.” Discussion Paper 72, Spatial Economics Research, Cambridge, MA. Research Centre, London. Ghani, E., W. Kerr, and S. D. O’Connell. 2011. Olley, G. S., and A. Pakes. 1996. “The Dynamics “Spatial Determinants of Entrepreneurship in of Productivity in the Telecommunications India.” NBER Working Paper 17514, National Equipment Industry.” Econometrica 64 (6): Bureau of Economic Research, Cambridge, MA. 1263–97. Glaeser, E., and W. Kerr. 2009. “Local Industrial Rai, S. 2015. “Foxconn Could Make India its Conditions and Entrepreneurship: How Much Next Manufacturing Base, after China, 78 SOUTH ASIA’S TURN
Investments Suggest.” Forbes, August 10, 2015. Working Paper 05/2010, United Nations http://www.forbes.com/sites/saritharai Industrial Development Organization, /2015/08/10/foxconn-could-make-india-its Vienna, Austria. -next-manufacturing-base-after-china Subramanian, A., and A. Modi. 2015. “ ‘Make in -investments-suggest/#1b3a49753bb9. India’ Not by Protecting but by Eliminating Rosenthal, S. S., and W. C. Strange. 2004. Negative Protectionism.” Business Standard. “Evidence on the Nature and Sources of http://www.business-standard.com/article Agglomeration Economies.” In Vol. 4 of / opinion/make-in-india-not-by-protecting-but Handbook of Regional and Urban -by-eliminating-negative-protectionism Economics, edited by J. Vernon Henderson -115050100026_1.html. and Jacques-François Thisse, 2119–71. Van Beveren, I. 2012. “Total Factor Productivity Amsterdam: Elsevier B.V. Estimation: A Practical Review.” Journal of Saleman, Y., and L. Jordan. 2013. “The Economic Surveys 26 (1): 98–128. Implementation of Industrial Parks: Some World Bank. 2008. Harnessing Competitiveness Lessons Learned from India.” World Bank, for Stronger Inclusive Growth: Bangladesh Washington, DC. Investment Climate Assessment. Washington, Sturgeon, T., and O. Memedovic. 2011. DC: World Bank. “Mapping Global Value Chains: Intermediate ———. 2015. Leveraging Urbanization in South Goods Trade and Structural Change in the Asia: Managing Spatial Transformation for World Economy.” UNIDO Development Prosperity and Livability. Washington, DC: Policy and Strategic Research Branch World Bank. Limited Success in Linking to Global Value Chains 5
art 1 of this report argued that global removal of trade barriers. The following dis- trade and investment integration are cussion develops these observations in Pimportant for productivity growth. In the more detail. 21st century, this integration takes place pri- marily by means of global value chains (GVCs), GVC participation supports which divide the production process into stages productivity and distribute these stages across different countries. This process, variously known as GVCs make it possible for firms in developing “production fragmentation” (Arndt and countries to participate in producing the Kierzkowski 2001), “processing trade” (Görg world’s most complex and sophisticated 2000), “vertical specialization” (Hummels, products by specializing in a piece of the pro- Rapoport, and Yi 1998), “slicing up the value duction process for which those firms have a chain” (Krugman, Cooper, and Srinivasan comparative advantage and produce at the 1995), or “the second unbundling” (Baldwin necessary large scale to be competitive glob- 2006), has been made possible by major ally (figure 5.1). Lead firms, typically located changes in logistics and managerial organiza- in advanced economies, perform the higher tion in the last third of the 20th century. value-added activities (such as design, brand- At the firm level, various aspects of GVC ing, and retail), but outsource most or all of participation, such as imports of parts and the manufacturing to a global network of components, entry into export markets, and producers. Beyond their direct contributions, knowledge spillovers from tie-ins with lead lead firms also have major positive effects firms, have been associated with higher pro- through the knowledge and support they pro- ductivity. However, South Asia’s participa- vide to suppliers and the competitive pressure tion in GVCs largely remains confined to they put on all firms in the industry. Foreign apparel. And even in this sector, the sophisti- firms’ subsidiaries in South Asia play a par- cation of the region’s products declined ticularly important role in complex, capital- between 2000 and 2010. To take better and knowledge-intensive activities such as car advantage of the productivity-enhancing assembly (such as Maruti Suzuki in India and opportunities offered by participation in Hyundai in Pakistan) and electronics (such as GVCs, South Asia needs to further develop Samsung in India and Tos Lanka, a subsid- the capabilities that matter for GVCs, includ- iary of Toslec from Japan, in Sri Lanka). The ing human capital, institutions, logistics, and relationships between foreign and domestic 79 80 SOUTH ASIA’S TURN
FIGURE 5.1 Structure of the global value chain for apparel
Increasing economic Design and branding value added Logistics and sourcing Production
Inputs Components (textiles) Final products Distribution, sourcing, and sales
Natural and Yarn Fabric Apparel Lead synthetic Intermediaries production production production firms fibers (cut and sew) Knit Brand manufacturers Trim (buttons, Woven Brand marketers zippers, Retailers elastic)
Equipment and machinery
Tangible Increasing value added Intangible activities activities
Source: World Bank 2016. Note: Blue indicates highest-value-added activities and control or power over the supply chain.
firms provide a vital first step toward increas- Mattoo, and Ruta 2015). Causality is difficult ing productivity and producing goods that to establish; more-productive firms are more meet world-market specifications with regard likely to enter export markets than less-pro- to technological content, quality, and design ductive firms (Melitz 2003; Clerides, Lack, (Helpman 1984). Nowhere is this more evi- and Tybout 1998; Abraham et al. 2010; Amiti dent than in East Asia, where the transfer of and Konings 2007; Goldberg et al. 2010; technology and knowledge facilitated through Atkin, Khandelwal, and Osman 2016). GVCs made it possible for economies at Empirical studies have used a variety of initially low levels of income—such as China; statistical techniques to control for this selec- Hong Kong SAR, China; Republic of Korea; tion bias. In a sample of Indonesian firms, Singapore; and Taiwan, China—to move up Arnold and Javorcik (2009) find evidence of the ladder of productivity, capital intensity, increased labor productivity resulting from and quality (Kimura and Ando 2005; Amiti capital investment and organizational and and Konings 2007; Kee and Tang 2015). management restructuring following acquisi- An extensive literature documents firm- tion by a foreign affiliate. In South Asia, evi- level productivity benefits from various aspects dence suggests that firms in Bangladesh that of GVC participation, such as access to larger receive FDI are more productive than firms markets, learning-by-exporting, and knowl- that do not (Kee 2005). In a review of these edge spillovers from foreign direct investment studies, Havranek and Irsova (2011) conclude (FDI), although recent evidence suggests that that a 10 percent increase in foreign presence the contribution of these forces to productivity is associated with a 9 percent increase in the growth declined in the period 2004–11 com- productivity of local suppliers through their pared to 1995–2003, primarily on account of exposure to foreign firms. This evidence sug- the global trade slowdown (Constantinescu, gests that firms often participate in GVCs first LIMITED SUCCESS IN LINKING TO GLOBAL VALUE CHAINS 81
and then become more productive, rather than 2007). Evidence from a randomized trial in the the other way around. Arab Republic of Egypt documents productiv- Classifying firms whose share of imported ity gains from exporting through knowledge raw materials is greater than 10 percent of the transfers (Atkin, Khandelwal, and Osman total raw materials used as GVC participants, 2016). In South Asia, studies have found that we find that the total factor productivity learning-through-exporting has boosted the (TFP) of GVC participants in Bangladesh and productivity of firms in India that enter export India tends to exceed that of nonparticipants markets (Mukim 2011), in part through scale (figure 5.2). The picture remains approxi- effects.2 Similarly, firms in Bangladesh with mately the same when GVC participants are more export experience exhibit higher produc- defined as firms whose share of imported raw tivity (Fernandes 2008). Among domestic pro- materials is greater than 20 percent of the ducers of apparel inputs in Bangladesh, even total raw materials used. The same pattern those firms that neither export nor supply to holds for participation in international mar- exporters can experience productivity spill- kets more generally: firms whose trade share overs by learning from the experience of firms (exports plus imports) is greater than 10 that are part of a shared supplier network that percent of value added tend to have, on aver- supports exporters (Kee 2015). On the import age, higher levels of TFP in Bangladesh, India, side, access to imported intermediate inputs and Sri Lanka. And this picture remains can also boost firm productivity; for example, robust to the use of different trade shares to when India liberalized its tariff regime, access define GVC participant firms. It should there- to a greater range of intermediate goods at fore come as no surprise that GVC partici- lower overall prices made manufacturing firms pants, as reflected in evidence from India and more productive (Goldberg et al. 2010). Case Bangladesh, are associated with a higher studies document the important productivity share of exports in value added.1 benefits from interactions with foreign firms in Various efforts have been made to pinpoint the context of GVCs. For example, the Desh- the source of productivity benefits from export- Daewoo joint venture, which included the ing. Increases in productivity following entry intense technical and managerial training of into foreign markets are attributed to greater 130 Bangladeshis in Daewoo’s Pusan plant in scale economies in Sub-Saharan Africa (Van 1979, established the foundation for the next Biesebroeck 2005) and Slovenia (De Loecker generation of Bangladeshi entrepreneurs.
FIGURE 5.2 Firms that participate in GVCs also have higher productivity
a. Bangladesh b. India
0.6 0.6
0.4 0.4 Firm density Firm density 0.2 0.2
0 0 –10 –5 0 5 –10 –5 0 5 10 Productivity (log) Productivity (log)
Non-GVC participants GVC participants Non-GVC participants GVC participants
Source: World Bank calculations based on the 2012–13 Annual Survey of Industries (ASI) in India and the 2012 Survey of Manufacturing Industries (SMI) in Bangladesh. Note: Productivity (TFP) estimates normalized to zero mean. GVC = global value chain. 82 SOUTH ASIA’S TURN
South Asia’s success in global and GVCs (box 5.1). In 2013, the share of total regional GVCs is limited to apparel merchandise exports in major GVC products (apparel, autos, electronics, and footwear) South Asia has the second-highest level of was approximately 80 percent for Bangladesh, GVC exports out of total exports among 45 percent for Sri Lanka, 40 percent for developing regions, almost entirely on the Pakistan, and 15 percent for India, while the strength of its large share of final and interme- participation of the remaining countries is diate apparel products (reinforcing the find- negligible (table 5.1). By this metric, ings on global market shares in part 1 of this Bangladesh has one of the highest GVC par- report). Approximately half of GVC exports ticipation rates in the world, although it from South Asia are in final apparel, whereas reflects the fact that Bangladesh exports little East Asia specializes in electronics, and besides final apparel (figure 5.4). India’s par- Europe, North America, and Latin America in ticipation in GVCs, by the same token, is low autos (figure 5.3). In imports, South Asia is precisely because it has a more diversified relatively less integrated in GVCs: firms mak- export basket, some of which may also have ing final products tend to obtain inputs from some of the characteristics of GVC produc- domestic sources (or themselves), indicating a tion but are not included in this analysis.4 lower level of GVC integration than implied With regard to products, each of the four 3 in the export data. In the case of apparel, South Asian countries has a significant however, local sourcing of intermediates com- export position in final apparel, covering bined with the predominance of final apparel nearly the full range of garment products, exports is consistent with FDI-led GVC activ- while intermediate apparel—dominated by ity in South Asia; that is, global lead firms set cotton textiles—is particularly important in up factories and use local materials to manu- India and Pakistan. Bangladesh’s exports of facture and export final apparel. final apparel in 2013, which have nearly Within the region countries vary greatly in tripled since 2007, amounted to over the extent to which they are integrated into $26 billion, making it the second-largest
FIGURE 5.3 South Asia’s strength in apparel explains the region’s high GVC integration
a. GVC exports by region, 2011 b. GVC imports by region, 2011
East Asia and Pacific East Asia and Pacific Europe and Central Asia Europe and Central Asia Latin America and Caribbean Latin America and Caribbean Middle East and North Africa Middle East and North Africa North America North America South Asia South Asia Sub-Saharan Africa Sub-Saharan Africa
0 20 40 60 80 100 0 20406080100 Percent Percent
Final apparel Intermediate apparel Intermediate apparel Final autos Intermediate autos Intermediate autos Final electronics Intermediate electronics Intermediate electronics Final footwear Intermediate footwear Intermediate footwear Non-GVC trade Others
Source: World Bank calculations based on UN COMTRADE database. LIMITED SUCCESS IN LINKING TO GLOBAL VALUE CHAINS 83
BOX 5.1 Measuring GVC participation
There are a number of ways to measure partici- (electronics), or buyers of final products working pation in GVCs. Some analyses, especially those with global middlemen (apparel and footwear). that focus on tracking global flows of value added This report uses a modified version of the through input-output methods, essentially view definition of the three classic GVCs in Sturgeon all trade as part of a GVC (Mattoo, Wang, and and Memedovic (2011). Products are classified Wei 2013). A country that exports only crude oil as belonging to one of the three GVCs based on or metallic ores may have a high degree of GVC a combination of expert opinion and their posi- participation under this definition because these tion in the U.N. Statistical Division’s Broad crude materials are eventually transformed into Economic Categories, which help to distinguish sophisticated goods or parts of other goods in between intermediate and final goods. This leads another country.a However, links with lead firms to a list of over 400 traded goods, identified in that lead to technology transfer and deeper inter- the United Nations Standard International Trade actions with final markets are more likely to take Classification (revision 3) classification at the place when countries are engaged in the middle four-digit or five-digit level. Each of the GVCs is or later stages of the production process. then divided into two subsectors to reflect inter- GVCs in vehicles, electronics, and apparel and mediate and final goods (such as intermediate footwear are characterized by a lead-firm net- electronics and final electronics), making six work structure and have been much studied. The GVC sectors in all. For the purposes of this anal- share of total global merchandise exports ysis of South Asia, the Sturgeon and Memedovic accounted for by these three GVCs has fluctuated (2011) categories are modified in two ways. between approximately 14 percent and 28 percent First, the footwear sector, both intermediate and since 1990. Studying the similarities and differ- final, is separated from apparel, making eight ences in the organization of these three GVCs can categories instead of six. Second, the definition improve our understanding of GVCs or, as they of the “autos” sector, which originally included are sometimes called, global supply chains (USITC only passenger motor vehicles and motorcycles, 2011). These three sectors differ in the methods is broadened to encompass other road vehicles, used to coordinate activity over long distances such as trucks, buses, and trailers. and the extent to which they are coordinated by a. Following the terminology of the input-output approach to measuring GVC traditional manufacturers (autos), owners of participation, exporters of primary products participate in GVCs through “forward linkages” while countries that export final goods requiring large amounts of brand names with strong research capabilities imported intermediate goods participate in GVCs through “backward linkages.”
TABLE 5.1 Outside of apparel, GVC exports from South Asia lag well behind East Asia US$ per capita
Afghanistan Bangladesh Bhutan India Nepal Pakistan Sri Lanka China Vietnam 2013 2013 2012 2013 2013 2013 2013 2013 2013 Final apparel 0.0 170.1 0.0 13.1 2.7 24.7 210.3 125.4 189.1 Final autos 0.0 0.0 0.0 6.5 0.0 0.1 1.9 23.6 4.7 Final electronics 2.0 0.0 0.5 3.7 0.2 0.6 6.9 252.8 274.1 Final footwear 0.0 3.2 0.0 1.8 0.0 0.6 1.1 35.5 93.6 Intermediate apparel 0.0 0.5 0.9 6.1 2.3 30.4 6.2 42.3 29.9 Intermediate autos 0.0 0.1 0.1 5.0 0.0 0.3 2.6 39.5 40.3 Intermediate electronics 0.0 0.1 0.0 1.0 0.0 0.0 1.1 67.5 16.2 Intermediate footwear 0.0 0.1 0.0 0.3 0.0 0.0 0.4 1.9 3.6 Total 2.0 174.1 1.5 37.5 5.2 56.1 230.5 588.5 651.5
Note: Data for Bangladesh 2013 are mirror data. 84 SOUTH ASIA’S TURN
FIGURE 5.4 Apparel accounts for the lion’s share of South Asia’s doubled in the last ten years to $6.4 billion in GVC exports 2013, reaching sophisticated markets such as Percent of total exports, 2013 the United States, the United Kingdom, Italy, 5 100 and Germany. India already exports more auto parts than Indonesia, Morocco, or 90 Vietnam, but only about one-tenth as much 80 as China. India is also one of the largest and 70 most rapidly growing developing country 60 exporters of final autos: it exports to middle- income countries (such as Algeria, Mexico, 50
ercent and South Africa) as well as developed coun- P 40 tries (such as Australia and the United 30 Kingdom), and, at current growth rates, its 20 exports of final autos may exceed those of China by 2020. In final electronics, India’s 10 exports have quadrupled in the six years 0 Bangladesh India Pakistan Sri Lanka ended 2013, but growth has come from a small base, and China’s exports dwarf those Intermediate footwear Intermediate electronics of India by a factor of 20. Intermediate autos Intermediate apparel India’s rapid export growth has come Final footwear Final electronics largely from using lower-priced products to Final autos Final apparel penetrate developing country markets Source: World Bank calculations based on UN COMTRADE database. (figure 5.5).6 Seventy-one percent of India’s exports of passenger motor vehicles go to the exporter of final apparel in the world after Middle East and North Africa, Latin America China. Sri Lanka and Bangladesh, in partic- and the Caribbean, the rest of South Asia, ular because of effective import facilities for and Sub-Saharan Africa. Similarly, the largest exporters, perform at East Asia’s level in destinations for Indian cell phone exports are terms of exports per capita ($216 and $147, Argentina, the Russian Federation, South respectively), while India and Pakistan are Africa, and the United Arab Emirates. On at an order of magnitude lower ($10 and average, therefore, India’s exports of GVC $23, respectively). Annual growth rates of products are skewed more toward more mid- exports over 2003–13 in India (9.6 percent) dle-income countries than are the exports of and Sri Lanka (5.3 percent) were modest, Germany or Japan, which are geared more and in Pakistan were sluggish (2.7 percent). toward developed country markets. These rates of growth, however, may not Trade data do not provide a full picture of necessarily reflect productivity improve- GVC integration, because many GVC-related ments because the region benefitted from sales take place within national borders. the 2005 elimination of the Multi-Fiber Measures of foreign value added in exports Agreement (MFA), which had restricted tex- that draw on multiregional input-output tables tile imports from developing countries to are useful indicators of the extent to which developed countries. countries are engaging fully in the international Most South Asian countries have negligi- division of labor that GVCs make possible.7 ble exports of GVC products other than The Organisation for Economic Co-operation apparel, with India accounting for almost all and Development (OECD)–World Trade of the region’s exports of autos and Organization (WTO) Trade in Value Added electronics. India has a large auto parts indus- (TiVA) data, for example, show that the share try, whose growth has been supported by of foreign value added in India’s exports rose increased exposure to international competi- from 9.4 percent in 1995 to 24.1 percent in tion since the lowering of trade barriers in the 2011, exceeding Indonesia’s share but continu- early 2000s. India’s auto parts exports ing to lag behind the Republic of Korea, China, LIMITED SUCCESS IN LINKING TO GLOBAL VALUE CHAINS 85
FIGURE 5.5 India’s auto and electronics exports have focused on the lower end of the market
a. Unit values for cellular telephones b. Unit values for small passenger vehicles 200 14,000 12,000 150 10,000 100 8,000
ollars ollars 6,000 D 50 D 4,000 2,000 0 0 China India Mexico Singapore United Belgium Czech India Japan Spain States Republic
Source: World Bank calculations based on UN COMTRADE database. Note: Product identification code for small passenger vehicles = HS870322. Note: Product identification code for telephones = HS851712.
FIGURE 5.6 India relies less on foreign inputs than its comparators
a. Foreign value added in total exports b. Foreign value added in exports of textiles, textile products, leather, and footwear 60 50 50 40 40 30 30 ercent ercent
20 P
P 20 10 10 0 0 Korea, China India Indonesia Malaysia Vietnam Korea, China India Indonesia Malaysia Vietnam Rep. Rep. 1995 2011
Source: World Bank calculations based on UN COMTRADE database.
Malaysia and Vietnam (figure 5.6). This sug- Bangladesh is even further downstream (0.4 gests that India is less integrated in GVCs than percent). These characterizations have been comparator countries in East Asia when rather stable over time and may be related to domestic value added is taken into account. the market structure of GVC sectors. In Focusing on textiles and apparel as a unified apparel, Bangladesh and Sri Lanka are domi- sector, the pattern is similar: India relies less on nated by large, formal firms geared toward the foreign inputs than its comparators, but that global market. India and Pakistan, however, reliance has increased in recent years. By con- have a sizeable informal apparel sector with trast, China’s foreign value added in textiles many firms employing fewer than 10 workers. and apparel has declined, perhaps indicating Yet none of the four countries is fully special- the country’s growing position in upstream ized. For example, Bangladesh imports large activities. amounts of cotton yarn from India but does Countries in South Asia vary in the extent not export very much fabric because, to a sig- to which they are upstream (specializing in nificant extent, the fabric made from the intermediate goods) or downstream (special- imported yarn is absorbed by the domestic izing in final goods) GVC participants. apparel industry. Furthermore, both Pakistan Pakistan is the furthest upstream (with 54 and India have significant exports of both percent of its GVC exports coming from inter- intermediate and final goods, making them mediate goods), followed by India (37 upstream in some product lines and down- percent). By contrast, Sri Lanka is much fur- stream in others. ther downstream (with 4.6 percent of its GVC The sophistication of South Asia’s exports coming from intermediate goods), and exports has improved in some respects. 86 SOUTH ASIA’S TURN
One way to assess sophistication is by mea- destination market, are more varied across suring the typical income level associated countries in South Asia than are the indica- with countries that export a basket of goods tors of product sophistication. Although similar to that of the country in question final apparel in Sri Lanka and Pakistan is (analogous to PRODY).8 In the decade directed at higher-income markets than is between 1999 and 2011, product sophisti- apparel from Bangladesh and India, market cation of final apparel increased in sophistication declined between 2000 and Bangladesh, India, and Sri Lanka but 2010 in all four countries (figure 5.7). This declined in Pakistan (figure 5.7). However, may reflect either increased sales to middle- product sophistication for cloth, yarn, and income markets or more intense competi- other apparel inputs converged over the tion in high-income markets, or both. In same decade, with sophistication increasing the case of intermediate apparel market in India and Pakistan and declining in sophistication, Pakistan and Bangladesh Bangladesh and Sri Lanka. showed gains between 1999 and 2001 and Indicators of market sophistication, between 2009 and 2011, indicating an measured by the average income level of the increasing ability to penetrate high-income
FIGURE 5.7 Although product sophistication has increased in some instances, market sophistication of South Asia’s textile and apparel exports has gone down
a. Average final apparel product sophistication b. Average intermediate apparel product sophistication 0.35 0.50 0.30 0.45 0.40 0.25 0.35 0.20 0.30 0.25 0.15 0.20 0.10 0.15 0.10 0.05 0.05 0 0
epal anka epal anka hutan India N L hutan India N akistan L B aldives akistan B P ri M P Sri S angladesh angladesh B B
c. Average final apparel market sophistication d. Average intermediate apparel market sophistication 1.0 0.6
0.8 0.5 0.4 0.6 0.3 0.4 0.2 0.2 0.1 0 0
epal epal hutan India N anka hutan India N anka B aldives akistan L B akistan L M P Sri P Sri angladesh angladesh B B 1999–2001 2009–11
Source: World Bank calculations based on UN COMTRADE database and World Development Indicators data. Note: Both product and market sophistication measures are normalized with the United States = 1. The relative income levels for South Asian countries are as follows: Bangladesh = 0.04, Bhutan = 0.11, India = 0.07, Maldives = 0.16, Nepal = 0.03, Pakistan = 0.06, and Sri Lanka = 0.11. LIMITED SUCCESS IN LINKING TO GLOBAL VALUE CHAINS 87
markets. India and Sri Lanka, on the other 2003, and in 2013 24 percent of imported hand, have shifted exports toward lower- intermediate apparel inputs came from income markets. within the region, up from 18 percent in Even though it is more feasible than ever 2003. In the region, Bangladesh and Sri to produce complex goods on a “made-in- Lanka have the highest share of final apparel the-world” basis, considerations of transport goods (at 86 percent and 44 percent of total and other transaction costs, as well as timely exports, respectively) and source many delivery, often cause value chains to cluster apparel inputs from Pakistan and India, on a regional basis. The best-known of these which focus relatively less on final products clusters are the East Asian regional value (at 18 percent and 6 percent of total exports, chains in electronics and the U.S.-Germany- respectively). In 2013, two-thirds of India’s Japan automotive regional value chains. In exports of knit and crochet fabric were des- South Asia, a regional value chain is emerg- tined for Sri Lanka and Bangladesh, while ing in intermediate apparel: intra–South nearly half of Pakistan’s exports of woven Asian apparel trade amounted to $2.5 billion cotton denim were destined for Bangladesh in 2013, up sharply from $400 million in and Sri Lanka (figure 5.8).
FIGURE 5.8 South Asia is developing a regional GVC in apparel
a. Exports of knit and crochet b. Exports of woven cotton denim, fabric (n.e.s.), India, 2013 Pakistan, 2013
Other Other 13% 27% Mauritius 3% Bangladesh Egypt, Sri Lanka 42% Arab Rep. 46% 3% United Cambodia States 3% 17% Total Sri Lanka Total exports: 5% exports: Bangladesh $210 million Egypt, Turkey $496 million Arab Rep. 18% 14% 9% Note: SITC 65529. Note: SITC 65243.
c. Exports of dyed woven d. Imports of dyed woven cotton, Sri Lanka , 2013 cotton, Sri Lanka , 2013
Indonesia Taiwan, China Other 1% Other India 1% 2% Hong Kong 2% 1% SAR, China Egypt, 8% Arab Rep. China 2% Pakistan 45% Turkey 18% 2% Total Total exports: imports: $29.8 million $163.6 million
Bangladesh India 92% 26%
Note: SITC 65242. Note: SITC 65242. Source: World Bank calculations based on UN COMTRADE database. 88 SOUTH ASIA’S TURN
There also is evidence of an East Asia– firm, local businesses need reasonably high South Asia regional value chain. Seventy productivity, a capacity to absorb new tech- percent of South Asia’s imported apparel nologies (skill and capital intensity), and, ide- inputs come from East Asia, with a growth ally, experience with trading across borders rate about the same as that of intra–South to be qualified suppliers (Corcos et al. 2013; Asian trade. Overall, South Asia sends Defever and Toubal 2013; and Jabbour 26 percent of its exports of GVC intermedi- 2012). The lack of these firm-level capabili- ates to East Asia and purchases 68 percent of ties can inhibit the extension of GVCs to cer- its extraregional inputs from East Asia tain countries (see Farole and Winkler (2014) (figure 5.9). This level of orientation of for the case of Africa). Other studies have South Asia’s exports of intermediates to East tested the importance of specific drivers of Asia is well above global averages in apparel, GVCs, including trade policy (Orefice and autos, and electronics. Rocha 2014), transport (Hummels and Schaur 2013), trade logistics (Saslavsky and Most policy determinants of GVC Shepherd 2012), and time zones (Dettmer participation are lacking 2014). Although these and other studies give us a sense of what factors are likely to be Although a number of studies have looked at important in determining GVC dynamics, the the determinants of production fragmenta- question of which specific drivers matter tion (for example, Hillberry 2011) and sup- most for country-level participation in GVCs ply chain trade (for example, Rahman and remains open. Zhao 2013), the literature is yet to give a In a recent study, Pathikonda and Farole clear picture of the drivers of GVC participa- (2015) find a greater intensity of GVC tion and competitiveness. At the level of the products compared to non-GVC products
FIGURE 5.9 South Asia is highly integrated with East Asia in GVC products
Sub-Saharan Africa
South Asia
North America
Middle East and North Africa
Latin America and Caribbean
Europe and Central Asia
East Asia and Pacific
0 1020304050607080 Percent of total intermediate exports/imports Intermediate imports sourced from East Asia Intermediate exports destined for East Asia
Source: World Bank calculations using UN COMTRADE database. LIMITED SUCCESS IN LINKING TO GLOBAL VALUE CHAINS 89
(as defined by combining lists generated by and connectivity, wage competitiveness, mar- Athukorala [2010] and Sturgeon and ket access, and access to inputs. Memedovic [2011]) across a sample of over South Asian countries vary on these capa- 100 countries for a range of capabilities that bilities (table 5.2). Sri Lanka scores highest on are most common in the theoretical, policy, the level of human capital as measured by and empirical literature on GVC trade. (The average years of schooling. India’s total natu- sample comprised 102 countries, which ral capital far outpaces that of other countries, together represented 81 percent of world with Pakistan a distant second. Bhutan’s insti- trade in 2012.) They divide capabilities into tutions are rated ahead of those of the other three categories: (1) fixed capabilities; (2) countries, followed by India, with Pakistan long-term policy variables; and (3) short-term and Bangladesh lagging behind considerably. policy variables. Fixed capabilities include Geographically, Afghanistan and Pakistan in proximity to markets and natural (resource) the region’s northwest appear the most disad- capital. Long-term policy variables, which vantaged, whereas Sri Lanka is closest to mar- can be changed gradually over a relatively kets. India is a much more sophisticated long period, include human capital, physical logistical hub, whereas some countries—such capital, and institutional capital. Short-term as Afghanistan, Bhutan, and Nepal—are in policy variables, which can be changed some of the most challenging logistics environ- directly through a policy shift or negotiations ments in the world. The minimum wage is in the short to medium term, include logistics not very different across Bangladesh, India,
TABLE 5.2 Many GVC capabilities and endowments in South Asia are limited
Category Capability Indicator Afghanistan Bangladesh Bhutan India Sri Lanka Maldives Nepal Pakistan Fixed Proximity to Proximity to markets 0.59 0.43 0.47 0.51 0.33 0.32 0.48 0.57 markets (GDP-weighted distance index; 0 to 1) Natural capital Total value of natural — 197.74 8.92 2,959.7 40.71 0.33 66.83 522.55 capital Long-term Human capital Average years of 3.85 5.91 — 6.24 10.06 6.02 4.23 5.02 schooling (16 years and older) Physical capital Capital stock ($ per — — — 2,764.3 — — — — capita) Institutional Rule of law (rating −1.90 −0.79 0.12 −0.04 −0.08 −0.33 −1.01 −0.74 capital from −2.5 to 2.5) Short-term Logistics and Logistics Performance 2.24 2.74 2.38 3.12 2.29 2.40 2.20 2.53 connectivity Index (rating; 1 to 5) Wage Minimum wage for a — 38.57 — 28.37 38.55 — 75.90 41.59 competitiveness 19-year-old worker or an apprentice ($ per month) Market access Overall Trade 23.49 16.87 1.68 8.37 18.47 34.23 1.68 14.73 Restrictiveness Index of trading partners (MAOTRI) Access to inputs Overall Trade — — — 14.90 7.42 20.20 12.63 7.37 Restrictiveness Index (OTRI)
Source: World Bank calculations. Note: The values for natural capital, wage competitiveness, and the two access indexes correspond to 2005, 2014, and 2009, respectively. In the indexes measuring market and input access, a higher number indicates lower access. — = not available. 90 SOUTH ASIA’S TURN
Pakistan, and Sri Lanka. For access to foreign physical capital, human capital, institutions, markets, India faces fewer trade barriers than and logistics. India erects barriers to imported Bangladesh, Pakistan, and Sri Lanka. For trade inputs that are higher than those of both barriers on imported inputs, however, India ASEAN and SACU. At the same time, on the appears to be more restrictive. To put these export side, India faces barriers lower than figures in context, figure 5.10 shows the stan- those of ASEAN but higher than those of dardized capability levels in South Asia com- SACU. Similarly, Bangladesh, Bhutan, pared to two regional blocs: the Association of Maldives, Nepal, Pakistan, and Sri Lanka are Southeast Asian Nations (ASEAN), which is on average more wage competitive than already of major importance in GVCs, and the ASEAN and SACU and are closer to markets. Southern African Customs Union (SACU), Their level of natural capital is higher than which is a potential competitor of South Asia. SACU’s, but lower than ASEAN’s. This group Data on India are shown separately from the of South Asian countries (excluding India), on rest of South Asia because of its size and level average, fares worse than SACU and ASEAN of industrialization. countries on other capabilities, including logis- India has several advantages relative to tics, institutions, human capital, and access to ASEAN and SACU: natural capital, wage markets overseas. Last, on average, they have competitiveness, and proximity to markets. better access to imports of intermediate inputs Yet it lags behind the two other blocs in than ASEAN, but worse access than SACU.9
FIGURE 5.10 GVC capability gaps compared to ASEAN and SACU are concentrated in areas amenable to policy intervention
a. India vs. ASEAN b. India vs. SACU
Access to inputs Market access Market access Access to inputs Wage competitiveness Wage competitiveness Logistics and connectivity Logistics and connectivity Institutional capital Institutional capital Physical capital Physical capital Human capital Human capital Natural capital Natural capital Proximity to markets Proximity to markets
–2.0 –1.5 –1.0–0.5 0 0.5 1.0 1.5 2.0 2.53.0 3.5 –2–101234
c. South Asia (excluding India) vs. ASEAN d. South Asia (excluding India) vs. SACU
Access to inputs Market access Market access Access to inputs Wage competitiveness Wage competitiveness Logistics and connectivity Logistics and connectivity Institutional capital Institutional capital Physical capital Physical capital Human capital Human capital Natural capital Natural capital Proximity to markets Proximity to markets
–1.5 –1.0 –0.5 0 0.5 1.0 1.5 –1.5 –1.0 –0.5 0 0.5 1.0 1.5
Source: World Bank calculations. Note: A positive number indicates an advantage for South Asia compared to the other regions. ASEAN = Association of Southeast Asian Nations; SACU = Southern African Customs Union. LIMITED SUCCESS IN LINKING TO GLOBAL VALUE CHAINS 91
INDUSTRY CASE STUDY C
The effect of trade barriers on the boost female employment in the formal sector apparel sector and provide workers with wages higher than those they can earn in agriculture or other This case study is based on Stitches to Riches?: informal sectors (Frederick and Staritz 2012). Apparel Employment, Trade, and Economic Women employed in the formal sector tend to Development in South Asia (World Bank 2016). have fewer children, which reduces population With rising labor costs in East Asia, South growth and improves children’s health, and Asia has an historic opportunity to capture its women are more likely than men to dedicate fair share of the global apparel market (cur- their income to the health and education of rently 12 percent compared to 41 percent for children (World Bank 2012). China alone), in the process pulling millions Rising wages in China may improve South out of poverty, especially women. Asia’s competitive position in the global However, taking advantage of lower labor apparel market. China accounted for 41 costs will not be sufficient as global buyers percent of global apparel exports in 2012 (up have ever more stringent conditions for qual- from 25 percent in 2000), compared to only ity, lead time, reliability, and social and envi- 12 percent for South Asia. A 2013 survey of ronmental compliance. Lead firms from leading global buyers in the sector found that across South Asia show that it can be done. A 72 percent of respondents planned to decrease key reason for their success was their ability to their share of sourcing from China over the connect to GVCs—both to source world-class next five years. fabrics and to serve demanding customers, which pushed them to ever greater heights. The challenge: Meeting ever more stringent Trade barriers are the main constraints to demands from global buyers on factors South Asia’s realizing its great potential in other than cost apparel. In particular, problematic duty To seize this opportunity, South Asia will drawback schemes in India and Pakistan need to compete not only on cost but also on make it difficult for exporters to import tex- quality and lead time as well as on social and tiles, imposing delays that are unacceptable to environmental compliance, which are increas- global buyers and cutting them off from the ingly important to buyers. increasingly important manmade fiber seg- Surveys of global buyers show that East ment. These issues have been resolved in Sri Asian apparel manufacturers rank well above Lanka (which has no import duties on tex- South Asian firms along these increasingly tiles) and Bangladesh (which has a very effec- important dimensions (table 5.3). tive system of bonded warehouses to facilitate Quality. Besides being cost competitive, duty-free import of textiles). As a result, Sri suppliers must also be able to consistently Lanka and Bangladesh export ten times more offer quality products. Quality is influenced apparel than India and Pakistan (adjusted for by the raw materials used, the skill level of the population size). sewing machine operator, and the thorough- ness of the quality control team. On the basis South Asia’s historic apparel of combined results from buyer surveys and opportunity interviews, countries can be placed in three groups according to the quality of apparel The apparel sector is one of the most impor- production, in order of strongest to weakest: tant employers in developing countries. Export- (1) China, Vietnam, and Sri Lanka; (2) oriented apparel production has the potential Indonesia, Cambodia, and Bangladesh; and to generate “good” jobs that contribute to ris- (3) India and Pakistan. ing living standards and poverty reduction. Lead time and reliability. Lead time and In particular, increased apparel exports tend to reliability are greatly affected by the efficiency 92 SOUTH ASIA’S TURN
TABLE 5.3 South Asia is less competitive than productivity, but also on their capacity to Southeast Asia in noncost areas source high-quality fabrics and to learn from Country comparison: non-cost-related factors impacting suppliers and demanding global customers. performance High labor productivity was achieved Buyers’ perceptions of through training and performance incentives, Lead Social such as incentives to reduce absenteeism. time and compliance and Computerized cutting machines are essential Country Quality reliability sustainability to reducing material waste and also are used China ● 1 ● 1 ▲ 3 for grading and marking by, for example, Bangladesh ● 5 ● 5 ● 6 Pacific Jeans of Bangladesh. Technology is India ● 6 ● 6 ● 5 also used to increase compliance with envi- Vietnam ● 2 ● 2 ● 2 ronmental standards (a must for selling to Cambodia ▲ 4 ▲ 4 ▲ 4 leading brands). For example, Pacific Jeans Indonesia ▲ 3 ▲ 3 ● 1 recycles its waste water through a very effi- cient effluent treatment plant. Orient Craft of Note: Based on data from (Birnbaum (2013) and stakeholder surveys conducted for this study. Countries were ranked from 1 to 6 on each India, like most Chinese apparel manufactur- factor, with 1 being the best and 6 being the worst. Blue indicates the ers, helped reduce costs and motivate work- factor is not a constraint; Yellow indicates some problems; Red shows an important constraint. ers by setting up housing close to the factory, which reduced the time and cost of commut- and availability of transportation networks ing and facilitated the employment of migrant and customs procedures. On the basis of sur- workers, including female workers. vey and interview results (Birnbaum 2013), Companies relied on foreign experts the countries can be placed in three groups, in (Bangladesh, for example, benefitted from order of strongest to weakest: (1) China, Sri Lankan expertise) to develop technical Vietnam, and Indonesia; (2) Sri Lanka and and managerial skills or benefitted from Cambodia; and (3) Bangladesh, India, and training abroad (a generation of Bangladeshi Pakistan. China has consistently had the entrepreneurs was trained in Korea by shortest lead times throughout the last decade Daewoo in the 1980s). Skills were also (Muzzini and Aparicio 2013; World Bank acquired through the machine suppliers, for 2005; World Bank 2013). example, by Orient Craft. Social compliance and sustainability. Innovation often arose as a result of interac- These criteria have become central to buyers’ tion with leading global buyers and suppliers. sourcing decisions in response to pressure For example, Pacific Jeans and MAS devel- from corporate social responsibility (CSR) oped new products in partnership with leading campaigns by nongovernmental organiza- brands from the United States, the European tions (NGOs) and compliance-conscious Union, and Japan. MAS of Sri Lanka has been consumers following safety incidents in developing innovative, high-performance apparel factories. Noncompliant countries sports apparel by investing heavily in research risk damaging their country’s brand. To that and development (R&D), as well as by import- end, Bangladesh, in partnership with global ing world-class textiles from around the world. buyers, is taking steps to boost compliance. On the other hand, leading firms from India and Pakistan (such as US Apparel and Orient The key to the success of lead firms: Craft), which had difficulty in importing high- Connectivity to GVCs performing textiles, either were limited to Lead apparel firms in South Asia (such as sourcing fabric locally (mostly cotton-based Pacific Jeans in Bangladesh, MAS in Sri fabrics) or had to develop their own textile Lanka, Orient Craft in India, and US Apparel production. When asked to name his main in Pakistan) show that world-class perfor- constraint, the senior executive from Orient mance can be achieved in the region. Their Craft said, “The difficulty to import manmade success was based not only on high labor fiber in India.” LIMITED SUCCESS IN LINKING TO GLOBAL VALUE CHAINS 93
Trade barriers hold back the growth of the weave and polyester filament, reach nearly sector, especially in India and Pakistan10 30 percent (Birnbaum 2013). Exporters can be Raw materials and other inputs to produc- competitive in global markets only if they are tion, over which each apparel supplier has exempt from these taxes on inputs. However, limited influence, make up two-thirds of total providing exemptions is fraught with diffi- cost. Fabrics are the most expensive input in culty. The categorization of different inputs is apparel production, and the quality of tex- subject to interpretation and negotiation, tiles is directly related to the quality of the creating risks for firms importing critical final product. Furthermore, the global inputs for the production of garments with apparel industry is quickly diversifying across tight production schedules. For example, a broad range of textiles (manmade fibers in duties are paid up-front and exporters apply particular) for which the most efficient pro- for a drawback, which is calculated on the cost ducers are located overseas. Thus, efficient of materials less the amount of duty paid—but import regimes, characterized by rapid clear- no drawback on trim items is permitted. ance through customs and low duties (or Administrative procedures are quite rigid. For effective duty drawback systems) are critical example, one firm reported that, if it obtained for export competitiveness. Bangladesh and preclearance to import synthetic fabric listed at Sri Lanka have achieved considerable prog- a certain weight but—because fabric produc- ress in improving their import regimes, while tion is inherently unpredictable—the actual India and Pakistan have yet to do so. consignment contained a few items at a slightly High import tariffs on cotton and man- different weight, customs officers, rather than made fibers (table 5.4) combined with ineffec- accepting minor differences from the original tive duty drawback mechanisms, have been application, might hold up the consignment to the main constraint to the growth of the apply a different tariff rate or on suspicion of a apparel sector in India and Pakistan. tariff violation (which carries very heavy fines). In India, manmade fiber imports are subject In the meantime, the firm would be unable to to a customs duty of 10 percent, which complete production, even if these fabrics were increases to the mid-teens for imports from only a small share of inputs (Jordan and Korea, China, and other principal produc- Kamphuis 2014). Similarly, in the advance ers because of antidumping measures. license scheme no duty is paid on imports used Furthermore, excise duties on the production in export products (Birnbaum 2013), but of manmade fibers are 12 percent, while natu- stakeholders contend that compliance with ral fibers are exempt. Total duty and tax rates procedures is extremely difficult and any error for some fabrics, such as polyester staple plain results in heavy fines.
TABLE 5.4 Import tariffs in South Asia and East Asia, 2014 Percent
Product category Bangladesh India Pakistan Sri Lanka Cambodia China Indonesia Vietnam Yarn Cotton (5203–5207) 5–10 10 5–25 0 0 5–6 5 5 MMF (5401–5406/ 5501–5511) 5–25 10 0–10 0 0 5 0–5 0–5
Woven fabric Cotton (5208–5212) 25 10 15–25 0 7 10–14 10–15 12 MMF (5407–5408/5512–5516) 25 10–12.5 15 0–15 7 10–18 10–15 12 Knit fabric (60) 25 10 20–25 0 7 10–12 10 12 MFN average Applied Duties 19.4 12.2 16.6 3.5 5.5 9.6 9.2 9.6 Textiles 12.9 8.5
Source: ITA-OTEXA 2014; WTO, UNCTAD and ITC 2014. Note: MMF = manmade fiber. MFN = most favored nation. 94 SOUTH ASIA’S TURN
Import barriers also affect the textiles and woven) from fabrics which are not industry. In India, purified terephthalic acid, widely available in India—Issuance of which is essential to the production of polyes- duty credit scrip (offsetting custom ter or synthetic fibers, can be purchased only duties) on import of specialty fabrics at from two Indian firms, one of which owns the rate of 5 percent for the export per- 79 percent of the production capacity (Jordan, formance in the year 2012–13 and in Kamphuis, and Setia 2014). the entire 12th five year plan. The duty and tax remission for export pro- In contrast, Sri Lanka eliminated all import gram (DTRE) in Pakistan is also problematic tariffs on textiles, and Bangladesh established (box 5.2). Remission can take two to four 2,000 bonded warehouses that enable rapid, months for textile imports, which is not duty-free import of textiles for exporters, acceptable to global buyers (Nabi and Hamid including small and medium enterprises 2013). As a result, the Pakistani apparel (SMEs). When asked to name the keys to its industry is dominated by the production of success, the founder of Pacific Jeans in low-value, cotton-based garments, using Bangladesh answered: “The system of bonded poor-quality textiles sourced domestically. warehouses put in place by the government The apparel export associations of India and together with back to back letters of credit.”11 Pakistan have put the reform of the import As a result of these constraints on the regime for textiles at the top of their “wish import of manmade fiber in India and lists” to their governments. The first proposal Pakistan, the industry in the region is exces- submitted by India’s Apparel Export Promotion sively concentrated in cotton fiber Council (AEPC) during an interministerial (figure 5.11), which is problematic because workshop held in April 2013 reads as follows: cotton-based apparel has been losing global Enlargement of the garment export bas- market share to manmade fiber–based ket by manufacturing garments (knitted apparel. It also limits the capacity of firms to
BOX 5.2 Expert evaluation of the duty drawback system in Pakistan
“The present system of suspension of duties and import and export—an archaic procedure; sus- taxes is governed by several Statutory Rule pension or cancellation of DTRE by the Orders (SROs) issued under Customs Rules 2001 Regulatory Collector as he may deem fit; exten- for ‘DTRE and Manufacturing Bond Licensing’ sive documentation requirements, and centraliza- and ‘DTRE Approval.’ The approval process tion of authority and approvals. involves multiple, parallel and overlapping The system complexities have led to extensive regimes; plethora of steps at each stage of which delays in processing of applications for DTRE, concerned officials have wide discretions; the sys- Manufacturing Bonds Licensing, and payment of tem itself is not clear and with no standard oper- DTRE claims resulting in the firms being unable ating procedures; intermediaries falsify the to timely import the quality inputs and meet supporting documentation which maximizes the their export orders; receive their blocked funds economic rent. The system involves physical veri- as well as pay the economic rent to the concerned fications of business premises; calculation of officials rendering these enterprises (and the ‘input wastages’ through physical checking by the export sector as a whole) non-competitive— Input Output Co-efficient Organization (IOCO); hence, there is need for fundamental changes in drawing of samples of imported input goods and the present DTRE system.”
output goods meant for exports at the time of Source: Ahmad Khan 2014. LIMITED SUCCESS IN LINKING TO GLOBAL VALUE CHAINS 95
FIGURE 5.11 SAR, unlike its competitors, focuses heavily on cotton Composition of apparel exports by region and fiber type, 2012
100
90
80
70
60
50
40
30
20 aterial share of apparel exports, 2012 (%) of apparel share aterial M 10
0 World SAR SEAB China N/A Other textile materials MMF Cotton
Source: UNSD 2014. Note: Exports represented by world imports; classifications created by author. MMF = Manmade fiber; SAR = South Asian sample countries (Bangladesh, India, Pakistan, and Sri Lanka); SEAB = Southeast Asian benchmark countries (Cambodia, Indonesia, and Vietnam). N/A indicates the material is not available in trade data classification definition.
FIGURE 5.12 India and Pakistan are far behind in apparel exports, 2005–12
CAGR, 2005–12
Sri Lanka 4%
Vietnam 18%
Bangladesh 16%
China 7%
Pakistan 7%
India 4%
0 50 100 150 200 250 Apparel exports per capita ($)
Source: Export data from UN COMTRADE database. Population figures from World Bank 2014. Note: Total Apparel Exports are a sum of HS 61 and 62. Exports are based on world imports from partner countries; data retrieved March 6, 2014. CAGR = compound annual growth rate. 96 SOUTH ASIA’S TURN
diversify their offerings and innovate. Finally, Thus, the much greater ease at which firms cotton-based apparel is sold mostly in the in Bangladesh and Sri Lanka can import fab- global spring and summer seasons while man- rics goes a long way in explaining why they made fiber–based apparel is sold mostly dur- export so much more apparel than do firms in ing the fall and winter seasons. Thus, absence India and Pakistan and are at levels compa- from the manmade fiber market reduces rable to East Asian countries on a per capita capacity use—apparel factories in India oper- basis (figure 5.12). ate only 6.5 months annually, while the global The extended version of this case study is average is 9 months (Jordan, Kamphuis, and available online at www.worldbank.org Setia 2014). /SouthAsiaCompetes. LIMITED SUCCESS IN LINKING TO GLOBAL VALUE CHAINS 97
Notes lowest for primary-product exporters such as Brazil, Russia, and Saudi Arabia (Koopman 1. In India, the share of exports in value added is, et al. 2010). on average, 12.9 percent for GVC participant 8. PRODY (as well as EXPY, described in an firms and 2.4 percent for nonparticipant firms. earlier section) is an index that measures the The corresponding figures for Bangladesh are quality of export baskets. The index, proposed 47.6 percent and 17.4 percent, respectively. by Hausmann, Hwang, and Rodrik (2007), is 2. A recent study based on a panel of 10,685 an average of per capita incomes of countries Indian manufacturing firms between 1990 producing a given product, weighted by global and 2011 found, however, that firms experi- export share. A higher PRODY means that a ence productivity growth a year prior to product is associated with a higher level of per entering export markets rather than after capita income (that is, the product is more entering—pointing to potential reverse cau- likely to be exported by a richer country). For sality (Gupta, Patnaik, and Shah 2013). South Asia, this means that the product 3. Bangladesh, India, Pakistan, and Sri Lanka sophistication of each South Asian country’s (the South Asian 4) run a substantial trade bundle corresponds to a level of income sig- surplus in final GVC goods ($68.0 billion of nificantly higher than today’s levels. exports versus $23.8 billion of imports) and 9. These comparisons remain unchanged have approximately balanced trade in inter- when South Asia is defined by the country mediate goods ($24.3 billion of exports ver- group formed by Bangladesh, Pakistan, and sus $25.1 billion of imports). They are Sri Lanka. significant net importers of electronics inter- 10. Other, less important, barriers also hold back mediates, modest net importers of automo- the sector. They are discussed in the extended tive intermediates, and net exporters of version of the case study available at www apparel and footwear intermediates. .worldbank.org/SouthAsiaCompetitiveness. 4. India’s largest exports are refined petroleum, 11. The system of bonded warehouses in diamonds, jewelry (including gold jewelry), Bangladesh caters mostly to the apparel pharmaceuticals for retail sale, and processed industry. Other high-potential industries rice. All of these lie outside the scope of this would greatly benefit from it, such as foot- analysis. wear, which has great difficulty importing 5. India exports chassis and engines, including leather (World Bank 2013). spark-ignition auto engines, diesel engines, and aircraft engines, as well as a variety of smaller parts. References 6. It is a stylized fact of international trade that rich countries import goods with higher Abraham, F., J. Konings, and V. Slootmaekers. 2010. unit values than do poorer countries, pre- “FDI Spillovers in the Chinese Manufacturing sumably because their consumers can Sector: Evidence of Firm Heterogeneity.” afford higher-quality varieties of products Economies of Transition 18 (1): 143–82. (Ferrantino, Feinberg, and Deason 2012; Amiti, M., and J. Konings. 2007. “Trade Manova and Zhang 2009; Bastos and Liberalization, Intermediate Inputs, and Silva 2010). Productivity: Evidence from Indonesia.” 7. In the archetypal case of China, foreign value American Economic Review 97 (5): 1611–38. added is highest in those sectors with the Arndt, S.W., and H. Kierzkowski, eds. 2001. highest degree of foreign investment and in Fragmentation: New Production Patterns in the more technologically progressive sectors such World Economy. Oxford: Oxford as computers and telecommunication equip- University Press. ment (Koopman, Wang, and Wei 2012). In Arnold, J. M., and B. S. Javorcik. 2009. “Gifted contrast, products like steel and ceramics Kids or Pushy Parents? Foreign Direct tend to have higher domestic value added in Investment and Plant Productivity in exports. Similarly, in international compari- Indonesia.” Journal of International Economics sons, the share of foreign value added is 79 (1): 42–53. higher for East Asian countries and Mexico, Athukorala, P. C. 2010. “Production Networks which are deeply imbedded in GVCs, and and Trade Patterns in East Asia: Regionalization 98 SOUTH ASIA’S TURN
or Globalization?” ADB Working Paper Series Sub-Saharan Africa: Local Spillovers and on Regional Economic Integration, No. 56. Competitiveness in Global Value Chains. Atkin, D., A. Khandelwal, and A. Osman. 2016. Washington, DC: World Bank. “Exporting and Firm Performance: Evidence from Ferrantino, M. J., R. M. Feinberg, and L. Deason. a Randomized Experiment,” NBER Working 2012. “Quality Competition and Pricing-to- Paper 20690, National Bureau of Economic Market: A Unified Framework for the Analysis Research, Cambridge, MA. of Bilateral Unit Values.” Southern Economic Bastos, P., and J. Silva. 2010. “The Quality of a Journal 78 (3): 860–77. Firm’s Exports: Where You Export to Matters.” Frederick, S., and C. Staritz. 2012. “Developments Journal of International Economics 82 (2): in the Global Apparel Industry after the MFA 99–111. Phaseout.” In Sewing Success? Employment, Baldwin, Richard. 2006. “Managing the Noodle Wages and Poverty Following the End of the Bowl: The Fragility of East Asian Regionalism,” Multi-fibre Arrangement, edited by G. Lopez- CEPR Discussion Paper 5561, CEPR, London. Acevedo and R. Robertson, 41–86. Birnbaum, D. 2013. “Competitiveness of India’s Washington, DC: World Bank. Apparel Export.” Apparel Export Promotion Goldberg, P., A. Khandelwal, N. Pavcnik, and Council (AEPC). P. Topalova. 2010. “Imported Intermediate Birnbaum, D. 2014. Comment: Bangladesh Inputs and Domestic Product Growth: Evidence Industry Development Moving Backwards. Just- from India.” Quarterly Journal of Economics style.com, April 16. http://www.just-style .com 125 (4): 1727–67. /comment/bangladesh-industry-development Görg, H. 2000. “Fragmentation and Trade: U.S. -moving-backwards_id121251.aspx. Inward Processing Trade in the EU.” Review of Clerides, S. K., S. Lack, and J. R. Tybout. 1998. World Economics 136 (3): 403–22. “Is Learning by Exporting Important? Gupta, A., I. Patnaik, and A. Shah. 2013. Microdynamic Evidence from Colombia, “Learning by Exporting: Evidence from India.” Mexico, and Morroco,” Quarterly Journal of Working Paper Series on Regional Economic Economics 113: 903–47. Integration No. 119, Asian Development Bank, Constantinescu, C., A. Mattoo, and M. Ruta. Manila. 2015. “The Global Trade Slowdown: Cyclical Hausmann, R., J. Hwang, and D. Rodrik. 2007. or Structural?” Policy Research Working Paper “What You Export Matters.” Journal of 7158, World Bank, Washington, DC. Economic Growth 12 (1): 1–25. Corcos, G., D.M. Irac, G. Mion, and T. Verdier. Havranek, T., and Z. Irsova. 2011. “Estimating 2013. “The Determinants of Intra-Firm Trade.” Vertical Spillovers from FDI: Why Results Vary Review of Economics and Statistics 95 (3): and What the True Effect Is.” Journal of 825–838. International Economics 85 (2): 234–44. Defever, F., and F. Toubal. 2013. “Productivity, Helpman, E. 1984. “A Simple Theory of Relationship-Specific Inputs and the Sourcing International Trade with Multinational Modes of Multinationals.” Journal of Corporations.” Journal of Political Economy Economic Behavior and Organization 94 92 (3): 451–71. (October): 345–57. Hillberry, R. 2011. “Causes of International De Loecker, J. 2007. “Do Exports Generate Production Fragmentation: Some Evidence.” In Higher Productivity? Evidence from Slovenia.” Global Value Chains: Impacts and Implications. Journal of International Economics 73 (1): Canadian Department of Foreign Affairs and 69–98. International Trade. Dettmer, B. 2014. “International Service Hummels, D. L., D. Rapoport, and K.-M. Yi. 1998. Transactions: Is Time a Trade Barrier in a “Vertical Specialization and the Changing Nature Connected World?” International Economic of World Trade.” Federal Reserve Bank of New Journal 28 (2): 225–54. York Economic Policy Review 4 (2): 79–99. Fernandes, A. M. 2008. “Firm Productivity in Hummels, D.L., and G. Schaur. 2013. “Time as a Bangladesh Manufacturing Industries.” World Trade Barrier.” American Economic Review Development 36 (10): 1725–44. 103 (7): 2935–59. Farole, T., and D. Winkler, eds. 2014. Making ITA-OTEXA (U.S. International Trade Foreign Direct Investment Work for Administration, Office of Textiles and Apparel). LIMITED SUCCESS IN LINKING TO GLOBAL VALUE CHAINS 99
2014. Market Reports/Tariffs: Textiles, Melitz, M. J. 2003. “The Impact of Trade on Apparel, Footwear and Travel Goods. Intra-industry Reallocations and Aggregate Washington, DC: ITA-OTEXA. Industry Productivity.” Econometrica 71: Jabbour, L. 2012. “Slicing the Value Chain 1695–1725. Internationally: Empirical Evidence on the Off- Mukim, M. 2011. “Industry and the Urge to Shoring Strategy by French Firms. The World Cluster: A Study of the Informal Sector in Economy 35 (11): 1417–47. India.” Discussion Paper 72, Spatial Economics Jordan, L.S., and B. Kamphuis. 2014. Supply Research Centre, London. Chain Delays and Uncertainty in India: The Muzzini, E., and G. Aparicio. 2013. Bangladesh: Hidden Constraint on Manufacturing Growth. The Path to Middle-Income Status from an Washington DC: World Bank. Urban Perspective. http://elibrary.worldbank Jordan, L.S., B. Kamphuis, and S.P. Setia. 2014. .org/doi/book/10.1596/978-0-8213-9859-3. A New Agenda: Improving the Competitiveness Nabi, I., and N. Hamid. 2013. Garments as a of the Textiles and Apparel Value Chain in Driver of Economic Growth: Insights from India. Washington DC: World Bank. Pakistan Case Studies. Lahore, Pakistan: Kee, H. L. 2005. “Foreign Ownership and Firm International Growth Centre. Productivity in Bangladesh Garment Sector.” Orefice, G., and N. Rocha. 2014. “Deep World Bank, Washington, DC. Integration and Production Networks: An ———. 2015. “Local Intermediate Inputs and the Empirical Analysis.” The World Economy Shared Supplier Spillovers of Foreign Direct 37 (1): 106–36. Investment.” Journal of Development Pathikonda, V.G., and T. Farole. 2015. A Economics 112 (January): 56–71. Capability-Based Assessment of GVC Kee, H. L., and H. Tang. 2015. “Domestic Value Competitiveness for the SACU Region. Added in Exports: Theory and Firm Evidence Washington, DC: World Bank. from China.” Policy Research Working Paper Rahman, J., and T. Zhao. 2013. “Export 7491, World Bank, Washington DC. Performance in Europe; What Do We Know Kimura, Fukunari, and Mitsuyo Ando. 2005. “The from Supply Links?” IMF Working Papers Formation of International Production and 13/62, Washington DC, International Distribution Networks in East Asia.” In Monetary Fund. International Trade in East Asia, edited by Saslavsky, D., and B. Shepherd. 2012. T. Ito and A. K. Rose 177–216. Chicago: “Facilitating International Production University of Chicago Press. Networks : The Role of Trade Logistics.” Koopman, R., W. Powers, Z. Wang, and S.-J. Wei. Policy Research Working Paper 6224, World 2010. “Give Credit Where Credit Is Due: Tracing Bank, Washington, DC. Value Added in Global Production Chains.” Staritz, C., and S. Frederick. 2014. “Sector Case NBER Working Paper No. 16426, National Study—Apparel.” In Making Foreign Direct Bureau of Economic Research, Cambridge, MA. Investment Work for Sub-Saharan Africa: Koopman, R., Z. Wang, and S.-J. Wei. 2012. Local Spillovers and Competitiveness in Global “Estimating Domestic Content in Exports Value Chains, edited by T. Farole and When Processing Trade is Pervasive.” Journal D. Winkler, 209–44. Washington, DC: of Development Economics 99 (1): 178–89. World Bank. Krugman, P., R. N. Cooper, and T. N. Srinivasan. Sturgeon, T., and O. Memedovic. 2011. Mapping 1995. “Growing World Trade: Causes and Global Value Chains: Intermediate Goods Consequences.” Brookings Papers on Trade and Structural Change in the World Economic Activity 26 (1): 327–77. Economy. Vienna: UNIDO. Manova, K., and Z. Zhang. 2009. “Export Prices USITC (U.S. International Trade Commission). Across Firms and Destinations.” NBER 2011. The Economic Effects of Significant Working Paper No. 15342, National Bureau of U.S. Import Restraints: Seventh Update Economic Research, Cambridge, MA. 2011. Special Topic: Global Supply Chains. Mattoo, A., Z. Wang, and S. J. Wei. 2013. Trade USITC Publication 4253. Washington, DC: in Value-Added: Developing New Measures of USITC. Cross-Border Trade. Washington, DC: Van Biesebroeck, J. 2005. “Exporting Raises World Bank. Productivity in Sub-Saharan African 100 SOUTH ASIA’S TURN
Manufacturing Firms.” Journal of International ———. 2014. World Development Indicators Economics 67 (2): 373–91. 2013. Washington DC: World Bank. UNSD (United Nations Statistics Division). 2014. ———. 2015. Leveraging Urbanization in South World Apparel Imports (2000, 2005, 2009, 2012). Asia: Managing Spatial Transformation for World Bank. 2005. “End of MFA Quotas: Key Prosperity and Livability. Washington, DC: Issues and Strategic Options for Bangladesh World Bank. Readymade Garment Industry.” Bangladesh ———. 2016. Stitches to Riches?: Apparel Development Series Paper 2, World Bank, Employment, Trade, and Economic Washington, DC. Development in South Asia. Washington, DC: ———. 2012. More and Better Jobs in South Asia. World Bank. Washington, DC: World Bank. WTO (World Trade Organization), UNCTAD ———. 2013. “Value Chain Analysis for Polo (United Nations Conference on Trade and Shirts.” In Bangladesh Diagnostic Trade Development), and ITC (International Trade Integration Study, Vol. 3, 107–59. Washington, Centre). 2014. World Tariff Profiles 2014 DC: World Bank. Geneva: WTO, UNCTAD and ITC. Firm Capabilities Are Constrained 6
n average firm in South Asia does not Firms lack managerial quality and operate at optimum efficiency. do not use resources efficiently Counterintuitively, the region’s firms A Differences in management practices can overemploy relatively scarce capital and underemploy labor, South Asia’s abundant account for as much as 30 percent of cross- resource. Performance varies widely with country differences in total factor productivity regard to knowledge inputs—ranging from (TFP) (Bloom, Sadun, and Van Reenen 2016). extensive technology use in India to limited In South Asia, organizational factors such as information and communication technology skills and management practices hold back (ICT) adoption in Bangladesh and Nepal— firm productivity (Bloom et al. 2010). For but even among leaders, the use of e-commerce example, management practices in India are and other productivity-enhancing online weaker than those in the United States, Brazil, business tools is relatively low. With respect and China (Bloom and Van Reenen 2010), as to innovation, the region again has both well as in many other developing countries leaders and laggards, but innovation is gen- (figure 6.1), because Indian firms generally do erally concentrated in a few, mature firms not collect and analyze data, set and monitor and, even then, is more likely to be of the clear performance targets, or explicitly link pay imitation, catching-up-to-the-frontier or promotion with performance. Conversely, variety. However, innovation of all types—to improved management practices in South Asia a large extent enabled through ICT have led to increased profitability and produc- adoption—is an important driver of produc- tivity. Bloom et al. (2013) performed an exper- tivity at the firm level. This suggests that fur- iment that provided management consulting to ther investment in firm capabilities, including large textile firms in Maharashtra, India, and better resource management, improving found that the intervention led to an 11 percent skills, deepening technology adoption, and improvement in firm productivity. McKenzie nurturing innovation, could raise productiv- and Woodruff (2015) show that business prac- ity in South Asia. The following discussion tices matter as much for microenterprises as develops these observations in more detail. for larger firms—boosting productivity,
101 102 SOUTH ASIA’S TURN
FIGURE 6.1 Management capabilities in South Asia are relatively weak
United States Canada Japan Singapore China India Myanmar Germany Sweden United Kingdom France Italy Poland Portugal Ireland Spain Greece Turkey Australia New Zealand Mexico Chile Argentina Brazil Colombia Nicaragua Kenya Nigeria Zambia Tanzania Ghana Ethiopia Mozambique 2.02.53.0 3.5 Average management scores, manufacturing
North America Asia Europe Australasia Latin America Africa
Source: World Management Surveys, 2013–14. Note: Management scores range from 1 (worst practice) to 5 (best practice).
profits, survival rates, and sales growth—and until the additional contribution of these fac- the positive effects of improved practices are tors to firm revenue (the marginal revenue robust to numerous measures of owners’ product) equals the going wage or rental rate. human capital. Of course, firms often do not observe mar- Managerial capabilities can be assessed ginal product directly; however, in the absence indirectly by looking at how firms use the of major distortions (physical, regulatory, or resources available to them. For example, information asymmetries), an average firm profit maximization requires that firms should employ something close to an optimal employ resources such as labor and capital factor mix—otherwise, a firm could gain by FIRM CAPABILITIES ARE CONSTRAINED 103
hiring relatively cheaper resources, and factor use, leverage economies of scale, and be closer owners could gain by commanding returns to the knowledge frontier. Instead, most large above their productivity levels, until marginal firms in South Asia do not operate close to products and costs were equalized. what would be considered optimum efficiency In South Asia, many firms do not use the levels given the prevailing factor prices, cost- available resources efficiently. Case studies ing themselves lost profits and bringing down reveal low capacity utilization in India and aggregate productivity. Pakistan among apparel makers (which oper- ate 6.5 months annually versus the global aver- Adoption of knowledge and age of 9 months) and automakers (which technology is low operate at 66 percent and 44 percent capacity, respectively, versus more than 75 percent in Firms rely on technology to enhance the effi- China). Only 4 of the 18 original equipment ciency of production processes and to connect manufacturers (OEMs) in the auto sector in more effectively with customers and suppliers. India and Pakistan operate at the industry stan- In particular, ICT is a major potential driver of dard for efficiency of 100,000 units per model. productivity growth, especially in countries and Previous research on India and Sri Lanka locations further from the technological has attributed weak performance of the man- frontier. In this report, ICT is defined as the use ufacturing sector to firms’ consistently under- of computers and other electronic equipment employing relatively abundant labor and and systems to collect, store, use, and send data overemploying relatively scarce capital electronically. Therefore, ICT includes any (Fernandes and Pakes 2008; Dougherty, communication device or application, such as Herd, and Chalaux 2009; Hasan, Mitra, and cellular telephones, computer and network Sundaram 2013).1 In India, these results sug- hardware and software, as well as the various gest that the optimal level of employment was services and applications associated with them, 6 times the actual level in 2001 and 3 times such as Internet and videoconferencing. the actual level in 2004, whereas in Sri Lanka, In developed countries, penetration of ICT the optimal level of labor was 1.1 times the is by now nearly universal: 97.9 percent of manufacturing employment in 2003.2 businesses with ten or more employees in the Repeating this analysis with the most recent countries of the Organisation for Economic data available for Bangladesh, India, Nepal, Co-operation and Development (OECD) Pakistan, and Sri Lanka reveals that underuti- have an Internet connection (OECD 2012). lization of labor remains a persistent feature Many developing countries also have high of the operating environment for firms in levels of ICT penetration: for example, over most of South Asia.3 The optimal level of 90 percent of Turkish and Mexican firms use employment of firms in India and Sri Lanka is the Internet (OECD 2012). However, based 1.7 and 2.2 times current employment levels, on data collected by the World Bank in the respectively, while estimates for Nepal and latest round of Enterprise Surveys, adoption Pakistan suggest underutilization on the order of ICT in South Asia is uneven (box 6.1). of 14 to 16 times the existing workforce. Only half of Nepalese and Bangladeshi firms Bangladeshi firms, on the other hand, appear use computers in their businesses, lower than to overutilize labor: firms hire approximately the average in Africa (figure 6.2).5 On the 18 percent more workers than would be opti- other hand, nearly all Indian firms use com- mal at the prevailing wage rate.4 puters, at a level similar to that in the Although the data include businesses as European Union. Twice as many Indian firms small as five employees, larger firms make up use computers and software as firms in the majority of the sample. However, this Bangladesh and Nepal, and 30 percent more makes the findings of even greater concern, firms use computers and software in India because larger firms could be expected to than in Pakistan. Pakistani firms lie between have a greater capacity to manage resource India, on the one hand, and Bangladesh and 104 SOUTH ASIA’S TURN
FIGURE 6.2 ICT adoption varies substantially across countries in South Asia
Share of firms that
Have Internet connection
Regularly use computer
Purchase or develop software
0102030 40 50 60 70 80 90 100 Percent India Pakistan Bangladesh Nepal South Asia (average) Africa Represents the range between minimum and maximum for countries in the Africa sample
Source: World Bank calculations based on 2013 Enterprise Surveys for Bangladesh, Nepal, and Pakistan, and 2014 Enterprise Survey for India. Note: Calculations for Africa based on data from the Democratic Republic of Congo, Ghana, Kenya, Namibia, Nigeria, South Sudan, Sudan, Tanzania, Uganda, and Zambia.
BOX 6.1 Data for analysis of ICT adoption in South Asia
The analysis of ICT and innovation practices is firms invest in ICT. Specifically, the ICT section based on the data collected by the World Bank of the innovation module provides information Enterprise Surveys, which were implemented in on two aggregate dimensions of ICT use: (1) com- Bangladesh, Nepal, and Pakistan in 2013 and in puter and software use and (2) different types of India in 2014. Overall, the dataset has approxi- Internet use. Computer and software use is a criti- mately 5,500 observations unevenly distributed cal channel for the improvement of the produc- among Bangladesh, India, Nepal, and Pakistan tion process, and Internet use can be a critical tool (with the number of firms surveyed in India for improving performance by reducing informa- exceeding the total number surveyed in the other tion costs, enabling e-commerce, and facilitating three countries). More than 4,000 manufacturing communication. Table B6.1.1 provides an over- firms and 1,266 service firms were surveyed. The view of the questions in the module. survey is representative of mostly medium, and to In order to obtain a meaningful measure of a lesser extent larger, firms, although a few firms ICT use, and given the lack of information on in the survey would have been characterized as what specific subdimensions of ICT use are more micro when surveyed because they had fewer important for performance, this report calcu- than five employees. lates a synthetic index for each ICT use using the The innovation module from the World Bank average of the normalized subcomponents in Enterprise Survey includes questions on ICT use, each country. For computer use, the two con- although the survey does not report how much tinuous variables (percentage of workers using a
(continues next page) FIRM CAPABILITIES ARE CONSTRAINED 105
BOX 6.1 Data for analysis of ICT adoption in South Asia (continued)
TABLE B6.1.1 ICT questions from the Innovation Module
Computer and Percentage of workers using a computer regularly software use Whether a firm has purchased or developed any software in-house Whether a firm has information technology (IT) staff Total cost of hiring an external computer or software consultant Internet use Communication: whether a firms uses the Internet for internal communication among its employees or for communication with clients and suppliers E-commerce: whether a firm uses the Internet for online purchases or sales Information: whether a firm uses the Internet for managing inventory, marketing products, or researching or developing ideas for new products and services
computer regularly and total cost of ICT consul- FIGURE B6.1.1 Components of the ICT Index tants) are normalized by subtracting the sample mean and dividing by the standard deviation. Computers Seven indicators are available for Internet Computers and use: two for communication (whether a firm software Software uses Internet for internal or external communi- cation), two for e-commerce (buying or selling Overall ICT index online), and three for information (online man- E-commerce agement of inventory, marketing, and research), which are then normalized and averaged to form Internet use Online marketing an Internet use index. Finally, an aggregate ICT total index (figure B6.1.1) is calculated as the Online inventory management average of the computer and Internet indexes.
Nepal, on the other, with approximately around 6.5 percentage points (at 99.6 percent 71 percent of firms having at least one worker and 93.1 percent, respectively), it increases to using a computer. ICT adoption rates also 23.4 percentage points in Nepal and 18.1 per- vary considerably by sector within and across centage points in Bangladesh. In Nepal, countries. In most South Asian countries, the Internet use appears to increase with age— use of ICT is much higher in electronics than only half of firms less than 5 years old use the in apparel, and ICT use is higher in sectors Internet, while 88 percent of those older than more exposed to international competition, 20 years do. The picture in Pakistan is, how- such as apparel and some automotive sectors, ever, the opposite—younger firms are more than in sectors that are less exposed to likely to use the Internet. In all countries, ser- competition, such as agribusiness. vice sector firms are more connected to the As expected, almost every firm using a Internet than are firms in manufacturing, computer also uses the Internet (figure 6.2). although the difference is not large. Similar to computer use, Internet use is higher Firms appear to use the Internet more often in larger and more foreign-exposed firms. to reach customers than to connect with suppli- Although the difference in the share of ers (figure 6.3). Given the high rates of connec- Internet use in large and small firms in India is tivity in South Asia (and particularly in India), 106 SOUTH ASIA’S TURN
FIGURE 6.3 Firms use the Internet to reach customers more than to connect with suppliers
Share of Internet-connected firms that:
Market online
Manage inventory online
Purchase online
Sell online
0 10 20 30 40 50 60 70 80 90 100 Percent India Pakistan Bangladesh Nepal South Asia (average) Africa Represents the range between minimum and maximum for countries in the Africa sample
Source: World Bank calculations based on 2013 Enterprise Surveys for Bangladesh, Nepal, and Pakistan, and 2014 Enterprise Survey for India. Note: Calculations for Africa based on data from the Democratic Republic of Congo, Ghana, Kenya, Namibia, Nigeria, South Sudan, Sudan, Tanzania, Uganda, and Zambia.
it is striking that South Asian firms are doing adoption varies significantly across countries: less e-commerce than firms in a number of India scores highly on multiple dimensions of African countries. In all four countries, the technology use and Pakistan is in line with share of firms selling goods on the Internet is global peers, but ICT adoption in Bangladesh higher than the share of firms purchasing and Nepal is very low and lower than that in online, a pattern different from that observed African countries. Second, India is a regional in Africa. South Asian firms generally do less leader in computer and software use among marketing on the Internet than do African firms, suggesting that there are potential spill- countries; 58 percent of firms in African coun- overs from the country’s strong software tries market online, but only Indian firms (the industry. Third, despite prevalent Internet use, highest share in the region) advertise on the the adoption of Internet commercialization web at the same rate. Although Internet use for practices for marketing products (e-commerce) marketing increases with size, large firms in is relatively low, with the difference particu- Bangladesh, India, and Pakistan do less mar- larly stark in India. keting online than large firms in African What determines these different rates of countries. Overall, and despite greater Internet ICT adoption? The literature identifies four access in South Asia than in Africa, the use of sets of factors: firm characteristics, market important Internet management strategies, structure, demand-side variables, and com- such as e-commerce and marketing, is still plementary factors such as skills, other tech- much lower than it has the potential to be. nologies, and agglomeration economies Overall, the data reveal three important that may facilitate diffusion (box 6.2). patterns for ICT use in South Asia. First, ICT Econometric analysis shows that all are FIRM CAPABILITIES ARE CONSTRAINED 107
BOX 6.2 Determinants of ICT adoption
The literature has identified a number of firm The environment in which the firm operates characteristics that are important for adoption of also matters for ICT adoption. Firms facing ICT at the firm level. (Some of these parallel the stronger competition are more inclined to inno- observations from the World Bank Enterprise vate and adopt new technologies, such as ICT, Surveys discussed in the preceding section.) in order to improve their performance and A number of studies find a positive correlation chances of survival. Some studies show that between firm size and the adoption of ICT (Teo competitive pressure is positively associated and Tan 1998; Thong 1999; Fabiani et al. 2005; with ICT adoption (Dasgupta et al. 1999; Giunta and Triveri 2007; and Haller and Kowtha and Choon 2001; Hollenstein 2004; Siedschlag 2011). Walczuch, Van Braven, and and Kretschmer, Miravete, and Pernías 2012). Lundgren (2000) point out that small firms in the Firms exposed to international competition in Netherlands are not adopting Internet use at export markets may be more inclined to adopt the same speed as larger firms.a Beyond size, some new technologies (Hollenstein 2004; Lucchetti studies suggest that adoption and use of ICT is and Sterlacchini 2004; Bayo-Moriones and higher in younger firms (Commander et al. 2006; Lera-López 2007; Giunta and Trivieri 2007; and Haller and Siedschlag 2011). Some studies have Haller and Siedschlag 2011). Similarly, foreign- examined the role of education or skills (human owned firms are more likely to be early adopters capital) in the adoption of new technologies of new technology as well as potentially impor- (Bartel and Lichtenberg 1987; Chun 2003), while tant channels of new technology diffusion others have shown that the demand for educated (Keller 2004; Narula and Zanfei 2005). workers rises with the use of the new technology a. Other studies find a weak or insignificant relationship (Teo, Lim, and Lai (Berman, Bound, and Machin 1994; Doms, 1998; Lefebvre et al. 2005; and Love et al. 2005). Hollenstein (2004) argues Dunne, and Troske 1997; Haskel and Heden that the relationship between ICT adoption and firm size might be nonlin- 1999; and Bugamelli and Pagano 2004). ear, which would partially explain the weak or insignificant relationship.
important in South Asia. Larger firms and Machin 1994; Doms, Dunne, and Troske 1997; exporters are more likely to adopt ICT prac- Haskel and Heden 1999; and Bugamelli and tices, with the size of each coefficient inversely Pagano 2004), skills matter critically for tech- related to the country’s economic develop- nology adoption in South Asia. The share of ment, suggesting larger effects in poorer high school graduates among firm employees is countries in the region (see table 6A.1 for positively and significantly associated with ICT detailed regression results).6 Younger firms adoption in the region pooled sample and in all are more likely to adopt ICT practices only in the country estimations except Pakistan. In India, which contrasts with the results of Bangladesh and Nepal, access to finance also Commander et al. (2011), and foreign-owned matters, while in India, access to foreign tech- firms are not more ICT intensive. The link nology through licensing is an important chan- between importing and ICT adoption, sug- nel for ITC adoption. Agglomeration matters gested in Hollenstein (2004) and Haller and for ICT diffusion only partially: in India (but Siedschlag (2011), is statistically significant not in other countries) firms in the main busi- only for India and Bangladesh. ness cities are more likely to adopt ICT, while in Consistent with the complementarity Nepal city size matters a great deal. Results for between technology and skills observed individual components of ICT adoption largely in OECD countries (Berman, Bound, and parallel those for the aggregate index. 108 SOUTH ASIA’S TURN
These findings suggest different policy pri- for the commercialization of products, facili- orities for countries within the region. Nepal tated by improved access to finance. Given and Bangladesh could concentrate on sup- the large number of firms in software devel- porting the adoption of the Internet and the opment and the availability of IT engineers, it use of computers in the private sector. The is likely that improving access to finance and public sector can play a role by investing in the establishment of broad-based online infrastructure (especially in Nepal where dif- financial transactions platforms could help fusion is higher in larger cities), by helping to broaden e-commerce use. train skilled workers, and by supporting the diffusion of technology. Once basic ICT Innovation is widespread but adoption is mainstreamed across all firms in novelty is limited these countries, the focus should shift toward greater integration of ICT practices that As with ICT use, adoption of innovation improve management and performance, rep- practices (box 6.3) differs significantly across resented in our indexes by the use of software South Asian countries. Bangladesh and India and the use of the Internet for the commer- have a larger percentage of firms conducting cialization of products. In India, where the research and development (R&D) than the use of ICT is pervasive in the private sector, average in Eastern Europe and Central Asia the focus could be on the use of the Internet and in Africa, while Nepal and Pakistan
BOX 6.3 Innovation activities and outputs in the Enterprise Surveys
Innovation can be measured by looking at inno- both knowledge-capital inputs and innovation vation inputs, innovation outputs, or both. outputs. The World Bank Enterprise Surveys pro- However, the subjective nature of many of the vide the following information on various sources questions used in innovation surveys presents a of knowledge capital and innovation outputs.a challenge. The Oslo manual, which is the main Innovation or knowledge inputs include: reference for this type of survey, defines innova- tion as “the implementation of a new or signifi- UÊ Research and development. Source and expen- cantly improved product (good or service), or ditures on R&D (internal versus external) process, a new marketing method, or a new orga- UÊ Capacity building. Training, including nizational method in business practices, work- expenditures, provided as a result of new place organization or external relations” (OECD/ innovations Eurostat 2005). Most surveys use this definition UÊ Purchase or licensing of inventions or other to identify innovations by directly asking firm knowledge forms. Expenditures on inventions or managers and owners whether they have imple- intellectual property that helped firms innovate mented “new” or “significant” changes or UÊ Acquisition of business intelligence. Key improvements in the last three years. This is sources of information and ideas for their problematic because “significant” is a highly sub- innovative activities jective term and any implementation is self- UÊ Intellectual property. Patents, utility models, reported. In general, any sound analysis of trademarks, or copyright designs, or regis- innovation activity should combine a focus on tered industrial designs
(continues next page) FIRM CAPABILITIES ARE CONSTRAINED 109
BOX 6.3 Innovation activities and outputs in the Enterprise Surveys (continued)
With respect to innovation outputs, the UÊ Organizational innovations. These are new Enterprise Survey innovation module differenti- organizational methods in business practices, ates between two types of technological innova- workplace organization, or external relations. tions (product and process) and two types of Organizational innovations are grouped into nontechnological innovations (organization and one of two categories. Structural innovations marketing): affect responsibilities, accountability, com- mand lines, and information flows, as well as UÊ Product innovations. New, redesigned, or the number of hierarchical levels, the divi- substantially improved goods or services: sional structure of functions (such as R&D, ° Products new to the firm production, human resources, and financing), ° Significantly improved products or the separation between line and support ° Products new to the market functions. Procedural innovations consist of UÊ Process innovations. Implementation of new changes to the routines, processes, and opera- or significantly improved production or deliv- tions of a company. Procedural innovations ery methods (including significant changes in change or implement new procedures and techniques, equipment, or software): processes within the company, such as simul- ° Innovation methods for manufacturing taneous engineering or zero buffer rules. products or offering services UÊ Marketing innovations. These include changes ° Innovative logistics, delivery, or distribu- to incorporate advances in marketing science, tion methods for inputs, products, or technology, or engineering to increase the services effectiveness and efficiency of marketing for a ° Innovative supporting activity for pro- competitive advantage. cesses, such as maintenance systems or
operations for purchasing, accounting, or a. The distinction between types of innovations, although clear in theory, computing can be a matter of some confusion for survey respondents. For example, new marketing processes (such as discounts), new packaging, and new The following are not considered to be inno- client segments are sometimes confused with process or product innova- vations: minor changes or improvements; an tions. Interviewees provide a recorded description of product and pro- cess innovations, which allows the user to verify the identified increase in production or service capabilities innovations, reclassify wrongly attributed cases to the proper category, through the addition of manufacturing or logis- and invalidate cases that do not constitute an innovation. This exercise tical systems that are very similar to those has been conducted by Cirera et al. (2015), who kindly provided us with already in use; ceasing to use a process; simple “clean innovation data.” For the overall sample of South Asia firms, the capital replacement or extension; changes result- “cleaning” exercise decreased the rate of product innovation from 53 percent to 51 percent and the rate of process innovation from 64 percent ing purely from changes in factor prices, cus- to 58 percent. Although the cleaning exercise reduced innovation rates tomization, regular seasonal, and other cyclical for most countries, in Nepal the rate of product innovation increased from changes; and the trading of new or significantly 10 to 12 percent as the result of the reclassification of some innovations improved products. originally designated as process innovations.
display a lower percentage (table 6.1). With intensity of R&D is low. There are some respect to R&D expenditures per employee, exceptions, of course. Bangalore, India, is however, South Asia’s performance across the home to one of Bosch’s global R&D centers region is below both Eastern Europe and (where it employs 15,000 personnel) and to a Central Asia and Africa. Thus, even though recently opened IBM enterprise mobility plat- many firms invest in R&D, the average form for developing iOS applications. 110 SOUTH ASIA’S TURN
TABLE 6.1 Knowledge capital intensity in South Asia is lower than in other regions
Europe and Type Indicator Bangladesh Pakistan India Nepal South Asia Central Asia Africa R&D Percent of firms 19 6 56 4 21 9 19 $ per worker (median) 8 — 14 6.5 14 498 18 Equipment Percent of firms 75 17 68 23 46 — 29 $ per worker (median) 92 197 227 130 179 — 180 Licensing Percent of firms 5 3 4 1 3 — 8 $ per worker (median) 6 82 25 234 27 — 21 Training Percent of firms 19 6 56 4 21 9 19 $ per worker (median) 12 107 21 73 21 — 47
Source: World Bank calculations based on 2013 Enterprise Surveys for Bangladesh, Nepal, and Pakistan, and 2014 Enterprise Survey for India. Note: Intensity is calculated only for firms engaged in research and development (R&D). Only 23 firms reported this information in Pakistan. — = not available.
Investment in innovation differs signifi- innovation rates of 15 percent and 21 percent, cantly across firms. Small, nonexporting, respectively (figure 6.4). Process innovation is national, and very young firms are more more prevalent in Bangladesh and India, R&D–intensive in India, while in Bangladesh, while product innovation is more common in large, exporting, foreign, and old firms are Nepal and Pakistan. Consistent with the ear- significantly more R&D–intensive. In lier discussion of firm capabilities, firms in all Pakistan, there is a very large concentration countries are more likely to innovate in mar- of R&D activity in a very small number of keting than in organization. firms. Investment in innovation in the global As suggested by the literature (Hall and value chain (GVC) sectors discussed earlier in Lerner 2009; and Kerr and Nanda 2014), the report is greater in large firms. In the technological and nontechnological innova- automotive sector, for example, field inter- tion rates are significantly higher for larger views show that large firms spend more on firms, consistent with evidence from Europe innovation and R&D than do smaller firms. and Central Asia and Africa.8 On the other And in agribusiness, 65 percent of large firms hand, the evidence for the relationship reported expenditures on innovation, com- between innovation and firm age is mixed: pared to only 49 percent of small and in India, younger firms display significantly medium enterprises (SMEs). Innovation higher rates of organizational innovation expenditures in electronics are relatively low and marketing while the opposite is true in at only 1.1 percent of sales in firms employ- the rest of the region. Trader firms (export- ing more than 100 workers and 4.7 percent ers, importers, or both) have higher rates of in smaller firms. technological and nontechnological innova- Studies show that training and R&D are tion, again consistent with the literature complementary in supporting productivity (Lileeva and Trefler 2010), although the dif- outcomes.7 Firms across South Asia, however, ferences are statistically significant in few spend relatively little on training. Lead firms cases. In all countries except Pakistan, do substantially more training than laggards, exporters are more innovative than import- although it is challenging to compare training ers in creating new products; but importers levels because so few firms in the laggard and two-way traders are more innovative group extend training to their employees. than exporters in process and organizational Turning to innovation outputs, Bangladesh innovation (except in Nepal). With the and India exhibit overall innovation rates of exception of Pakistan, foreign-owned firms approximately 80 percent, well above the are also more innovative (Brambilla 2009; average of Europe and Central Asia and Aghion et al. 2013), although the differences Africa, while Pakistan and Nepal have are relatively minor.9 FIRM CAPABILITIES ARE CONSTRAINED 111
FIGURE 6.4 Innovation rates vary significantly across South Asia Share of firms engaged in innovation activities
Bangladesh
India
Nepal
Pakistan
Europe and Central Asia
Russian Federation
Turkey
Africa
0204060 80 100 Percent Technological: Product and process Process Product Marketing Organizational
Source: World Bank calculations based on 2013 Enterprise Surveys for Bangladesh, Nepal, and Pakistan, and 2014 Enterprise Survey for India. Note: Calculations for Africa based on data from the Democratic Republic of Congo, Ghana, Kenya, Namibia, Nigeria, South Sudan, Sudan, Tanzania, Uganda, and Zambia.
Few firms engage in disruptive innovative innovation (table 6.2). Even in Bangladesh activities, such as introducing new products and India, the region’s innovation leaders, to the country or to the world; the majority of most innovation involves the imitation of firms conduct incremental innovations or existing products or processes. In these coun- pure imitation—either by upgrading the qual- tries, firms that introduce radical innovations ity of existing goods or introducing products are young, middle- or larger-sized, exporters, new to the firm (Cirera et al. 2015). This pat- and domestically owned. At the opposite end, tern, observed in Europe and Central Asia Nepal and Pakistan show very low innova- and in Africa, also holds in South Asia: tion rates in general, including imitation despite high average innovation rates in the activities. In these countries, innovating firms region, there is only a low degree of novelty in tend to be older, larger, nonexporters, and 112 SOUTH ASIA’S TURN
TABLE 6.2 Imitation is the most common form of innovation in South Asia Percent of firms that have introduced an innovation
South Asia Europe and Bangladesh Pakistan India Nepal (average) Central Asia Africa New to firm 44 8 54 12.3 30 18 25 Of which, new product 20 38 56 0 29 74 68 Of which, product upgrade 80 62 44 100 72 26 32 New to firm or local market 37 6 47 12 26 10 20 New to national market 4 2 4 0.3 3 6 3 New to international market 3 0.5 2 0.03 1 2 2
Source: World Bank calculations based on 2013 Enterprise Surveys for Bangladesh, Nepal, and Pakistan, and 2014 Enterprise Survey for India. Note: Calculations for Africa based on data from the Democratic Republic of Congo, Ghana, Kenya, Namibia, Nigeria, South Sudan, Sudan, Tanzania, Uganda, and Zambia.
FIGURE 6.5 The majority of innovation takes place in-house rather than collaboratively
a. Firms engaged in product innovation b. Firms engaged in process innovation
Nepal Nepal
India India
Pakistan Pakistan
Bangladesh Bangladesh
0 20406080100 0 20406080100 Percent Percent All in-house Cooperation with other firms University or government Don’t know
Source: World Bank calculations based on 2013 Enterprise Surveys for Bangladesh, Nepal, and Pakistan, and 2014 Enterprise Survey for India.
more likely foreign. In general, firms that underpins the large imitation rates in India have a higher level of R&D intensity intro- and Bangladesh. duce more radical innovations. Innovation activity is also spatially con- Most of the product and process innova- centrated, but only with respect to the nov- tions in the region are developed in-house elty of the innovation. Although in general, (figure 6.5). External cooperation is, in most R&D investments and innovation are more cases, linked to other firms, although cooper- concentrated than employment (Carlino and ation with the private sector plays a more Kerr 2014), with patents originating mostly important role for process innovation than in few and large cities (Fornahl and Brenner for product innovation. Successful firms 2009; Bairoch 1988), innovation activities in report that in-house R&D capability has been South Asia are less concentrated than an important driver of competitiveness, employment except in Pakistan (table 6.3). enabling them to compete on quality as well However, higher degrees of novelty, such as as cost. Nevertheless, this high reliance on in- products that are new to the national or house innovation development—larger than international market, are more concentrated in Africa and much larger than in Europe and than employment. Thus, agglomeration in Central Asia—implies a limited scope to South Asia may matter more for radical introduce more novel products and likely innovation than for imitation. FIRM CAPABILITIES ARE CONSTRAINED 113
TABLE 6.3 Agglomeration in South Asia matters most for radical innovation Concentration of economic and innovative activities per Hirschman-Herfindahl Index
Employment Innovators (product Innovators Radical (national- share Firm share or process) (product) international) R&D Bangladesh 0.390 0.280 0.280 0.280 0.400 0.240 India 0.017 0.006 0.006 0.009 0.040 0.007 Nepal 0.510 0.310 0.270 0.220 0.630 0.700 Pakistan 0.260 0.170 0.440 0.550 0.420 0.290
Source: World Bank calculations based on 2013 Enterprise Surveys for Bangladesh, Nepal, and Pakistan, and 2014 Enterprise Survey for India. Note: R&D = research and development.
Returns to innovation are high sources and effects of innovation in South Asia (box 6.4). With respect to knowledge inputs, Innovation is a key determinant of firm-level the most important determinant of R&D adop- productivity. Process innovation increases tion for all countries in South Asia is firm size, firm productivity through more efficient use of with larger firms more likely to engage in intermediate inputs and factors of production, R&D. Having a license to use foreign technol- while organizational innovation encourages ogy increases R&D in all countries except the reallocation of inputs and factors of pro- Bangladesh. Exporters in India and older firms duction across activities within firms. Product in Pakistan are also more likely to engage in innovation facilitates learning-by-doing and R&D than nonexporters and young firms, helps firms offer new and upgraded products, respectively. Further, financial constraints are while marketing innovation and innovative associated with lower investment in R&D for branding strategies allow firms to differentiate all countries except Bangladesh. Market struc- their products from those of their competitors ture appears to affect R&D only through infor- and gain market share. Precise estimation of mal sector competition in India. Other variables the impact is clouded by complementarities related to market structure are not significant, across different innovation concepts and perhaps because less than 9 percent of the issues of causality and endogeneity. However, sample firms compete in an oligopolistic or once these are addressed convincingly, the monopolistic market. estimated impact of innovation on productiv- Moving on to the determinants of innova- ity can be substantial. In a survey of evidence, tion outputs, R&D drives the intensity of inno- Mohnen and Hall (2013) show that the most vation (that is, the share of a company’s sales common value of the elasticity of firm-level that can be attributed to the introduction of productivity with respect to the intensity of product or process innovation), but does not product innovation—measured as the contri- affect the probability of adopting a technologi- bution of new products developed in the last cal innovation (table 6A.2). ICT is significantly three years to total sales—is 0.25, suggesting related to innovation intensity only for India that a 10 percent increase in intensity raises and the adoption of technological innovations productivity by 2.5 percent. This relationship in Nepal, but not in Pakistan or Bangladesh. is stronger in manufacturing than in services Lack of complementary factors, such as (Criscuolo 2009). skilled labor, reduces innovation intensity, The approach pioneered by Crepon, although marginally, in all countries except Duguet, and Mairesse (1998), which links Bangladesh. Other constraints, however, such knowledge inputs (such as R&D and ICT as access to external sources of funding, do adoption) to innovation outputs (such as better not appear to play a significant role. machines or more efficient managerial prac- Knowledge spillovers have a positive effect on tices) and innovation outputs to productivity, innovation-induced turnover for leading generates several insights concerning the firms, but they are insignificant for laggards. 114 SOUTH ASIA’S TURN
BOX 6.4 The Crepon-Duguet-Mairesse model
The Crepon-Duguet-Mairesse (CDM) model conditions and structure, and technology push explores the relationships between the basic factors. The first set includes variables captur- determinants of firms’ investment in knowledge ing firm size, age, and financial constraints and productivity. Firms invest in knowledge (measured by the share of internal sources used inputs that can be transformed into innovation to finance working capital). With regard to outputs. At a later stage, these outputs have an market structure, early empirical evidence pro- effect on firm-level productivity, depending on vided by Porter (1990), Geroski (1990), Baily the capacity of firms to transform innovation and Gersbach (1995), Nickell (1996), and outputs into improvements in product quality Blundell, Griffith, and Van Reenen (1995) sup- and efficiency (Crepon, Duguet, and Mairesse ports the view that competitive pressures 1998). The CDM model requires the estimation encourage innovation, while more recent evi- of three main components: (1) the knowledge dence by Aghion et al. (2005) shows that the function, which involves estimating the determi- relation is shaped like an inverted U. With nants of R&D and ICT adoption; (2) the innova- regard to composition, Cusolito (2009) shows tion equation; and (3) the productivity equation. that competition induces firms to specialize ver- The model is a recursive system of four blocks of tically by upgrading the quality of existing equations, in which each endogenous variable is goods. To account for these effects, the model determined sequentially. Firms first decide the includes variables measuring whether competi- intensity of two input choices—R&D and ICT. tion from informal firms is an obstacle for the These input choices and other factors feed into firm, if the sector in which the firm operates has different types of innovation outcomes (product a duopoly structure, and the extent of integra- or process, or innovation sales). Finally, innova- tion into international markets through trade. tion drives productivity (measured as output per With regard to technology, the model considers worker) at the firm level through an augmented whether the firm recently upgraded some of its Cobb-Douglas production function, which working capital and whether the firm has a includes innovation outcomes as inputs. license to use foreign technology because these The determinants of the adoption of knowl- variables can make investments in knowledge edge inputs include firm characteristics, market capital more attractive.
Further, agglomeration or urbanization effects than that in OECD countries (table 6.4). In do not appear to be an important determinant India, the large number of observations of innovation, while in Nepal most of the allows for separate estimation of product and innovation activity occurs outside main busi- process innovation, with both coefficients ness cities. Demand-pull factors, which reflect positive and statistically significant. The consumers’ willingness to pay a higher price degree of novelty does not introduce any for a given quantity, are important in explain- additional effect on productivity, and the ing innovation-induced sales gains in returns are the same as for imitation. Thus, Bangladesh, India, and Pakistan. the evidence suggests that there are positive Results from the final stage of the analysis, returns to imitation in South Asia, mostly which links innovation outputs to labor pro- coming from very incremental innovations in ductivity, show that the impact of innovation Bangladesh and India, but radical innovations on productivity in Nepal and Bangladesh is do not increase firm performance above and positive, statistically significant, and larger beyond the gains from imitation.10 FIRM CAPABILITIES ARE CONSTRAINED 115
TABLE 6.4 Innovation helps increase firm productivity
Bangladesh India Nepal Log(L) 0.1429*** 0.1416*** 0.0886*** 0.0901*** 0.3769*** 0.3586*** Log(K/L) 0.2827*** 0.3006*** 0.1567*** 0.1567*** 0.2369*** 0.2421*** Product or process 0.5544* 0.6902** 1.3959*** 1.5707*** Product innovation 1.2050*** 1.2146*** Process innovation 0.9759*** 0.9739*** Product or process (national) 0.0094 0.2742 Product or process (international) −0.0103 0.4718 Product (national) 0.0233 Product (international) −0.0972 Process (national) −0.0273 Process (international) −0.1172 Observations 990 990 3,481 3,481 470 470
Source: World Bank calculations based on 2013 Enterprise Surveys for Bangladesh, Nepal, and Pakistan, and 2014 Enterprise Survey for India. Note: Constant and sector dummies included but not shown. ***p < 0.01 **p < 0.05 *p < 0.1. Annex 6A TABLE 6A.1 Determinants of overall ICT adoption
South Asia (average) Nepal Bangladesh India Pakistan Firm size (log) 0.1807*** 0.3598*** 0.1654*** 0.1198*** 0.1908*** (0.0072) (0.0529) (0.0244) (0.0170) (0.0412) Firm age (log) −0.0033 −0.0055 0.0021 −0.0759*** −0.0106 (0.0108) (0.0602) (0.0308) (0.0276) (0.0812) Exporter 0.1603*** 0.2770** 0.4836*** 0.1230*** 0.2857** (0.0223) (0.1368) (0.1275) (0.0426) (0.1132) Importer 0.2058*** −0.1158 0.1757* 0.3120*** 0.0431 (0.0273) (0.1121) (0.0937) (0.0655) (0.1278) Foreign 0.1074 −0.0688 −0.0127 −0.1115 −0.5897* (0.0799) (0.3687) (0.1162) (0.1373) (0.3173) New capital t–1 0.0174 0.2311** 0.1194** −0.0138 −0.1777 (0.0166) (0.0910) (0.0607) (0.0342) (0.1101) Informal sector obstacle −0.0146 0.0013 −0.1220** 0.0417 0.2723** (0.0240) (0.0885) (0.0537) (0.0575) (0.1088) License for foreign technology 0.0528* −0.0136 0.0291 0.1753** 0.1072 (0.0288) (0.1703) (0.0775) (0.0880) (0.1210) Share used for working capital −0.0004* −0.0012 −0.0004 −0.0001 −0.0041** (0.0002) (0.0011) (0.0006) (0.0006) (0.0021) Duopoly or monopoly 0.0499* 0.4256** 0.2233** 0.0540 −0.1430 (0.0274) (0.1867) (0.1117) (0.0508) (0.1164) Business city 0.0163 0.1664 0.0562 0.0990** 0.0633 (0.0209) (0.1018) (0.0689) (0.0498) (0.1344) City over 1 million 0.1332*** 0.4064*** −0.1652* 0.1399 −0.4271 (0.0510) (0.1483) (0.0943) (0.1051) (0.3941) City 250,000 to 1 million 0.1050** 0.4660*** −0.0674 0.0888 −0.5418 (0.0503) (0.1548) (0.0939) (0.1083) (0.4103) City 50,000 to 250,000 0.0583 0.3629*** −0.1528 0.1335 −0.1968 (0.0506) (0.1234) (0.1034) (0.1092) (0.4049) High school workers 0.0017*** 0.0045*** 0.0058*** 0.0009* 0.0013 (0.0003) (0.0013) (0.0013) (0.0006) (0.0014) Observations 5116 470 967 3318 361
Source: World Bank calculations based on 2013 Enterprise Surveys for Bangladesh, Nepal, and Pakistan, and 2014 Enterprise Survey for India. Note: Robust standard errors in parentheses. Estimates use sampling weights, and country dummies are included in regional pooled estimates. Constant term not shown. ***p < 0.01 **p < 0.05 *p < 0.1. 116 SOUTH ASIA’S TURN
TABLE 6A.2 Determinants of innovation
Nepal Bangladesh India Pakistan Technological Innovation Technological Innovation Technological Innovation Technological Innovation innovation sales innovation sales innovation sales innovation sales Firm size (log) −0.5037*** −0.0190 0.2318 0.0055 0.2800*** −0.0504*** 0.3496*** 0.0124 (0.118) (0.035) (0.228) (0.054) (0.028) (0.016) (0.048) (0.010) Invests in R&D 0.2946 0.1817*** −1.1062*** 0.4595*** −1.8674*** 0.1978** 0.1914 0.0924*** (0.375) (0.025) (0.339) (0.033) (0.066) (0.096) (0.451) (0.035) ICT Index 1.9345*** 0.0497 −0.5478 −0.1465 0.2354 0.1908*** −1.5988*** −0.0468 (0.197) (0.096) (0.911) (0.203) (0.239) (0.062) (0.122) (0.043) Firm age (log) −0.1699* −0.0126 0.0246 −0.0085 −0.0697 −0.0088 0.2090** 0.0152 (0.089) (0.010) (0.063) (0.015) (0.043) (0.006) (0.085) (0.011) Education as obstacle −0.0404 −0.0280* 0.2207* −0.0045 −0.0412 −0.0218* −0.1052 −0.0310** (0.117) (0.016) (0.124) (0.022) (0.041) (0.013) (0.109) (0.015) Exporter −0.4398*** −0.0475* 0.3432 −0.0309 0.1827*** 0.0290* 0.4155*** −0.0076 (0.151) (0.027) (0.288) (0.076) (0.053) (0.016) (0.159) (0.021) Demand-pull effect −0.0611 −0.0173 0.1592* −0.0526*** 0.0376 0.0220** 0.0413 0.0264** (0.098) (0.015) (0.095) (0.019) (0.037) (0.010) (0.061) (0.013) Share used for working −0.0005 −0.0001 −0.0017 0.0008** −0.0005 −0.0001 −0.0059* −0.0001 capital (0.002) (0.000) (0.001) (0.000) (0.001) (0.000) (0.003) (0.000) Duopoly or monopoly −0.1417 0.0355 −0.2531 0.0467 0.0796 0.0133 −0.0351 0.0061 (0.226) (0.030) (0.350) (0.069) (0.065) (0.015) (0.140) (0.017) Business city −0.2419** −0.0294* −0.0306 0.0253 0.0177 0.0020 0.1533 0.0204 (0.118) (0.017) (0.097) (0.020) (0.023) (0.010) (0.154) (0.022) Spillover −2.8679 −0.1570 2.4450* −0.7407*** 8.4422 3.1811** 1.5389 0.3170 (1.877) (0.274) (1.482) (0.284) (7.763) (1.536) (1.258) (0.205) Constant 2.2236*** 0.1590 −0.3917 −0.0933 −0.8571*** 0.2784*** −1.7833*** −0.0998 (0.433) (0.189) (1.426) (0.296) (0.241) (0.063) (0.569) (0.087) Observations 470 470 990 990 3,481 3,480 499 502 Sector dummies ISIC-1digit ISIC-1digit ISIC-2digit ISIC-2digit ISIC-2digit ISIC-2digit ISIC-1digit ISIC-1digit
Source: World Bank calculations based on 2013 Enterprise Surveys for Bangladesh, Nepal, and Pakistan, and 2014 Enterprise Survey for India. Note: Technological innovation is a dummy variable with the value 1 if any new or significantly improved product, service, or processhas been introduced by a firm in the previ- ous three years. Innovation sales is the share of sales that can be attributed to the introduction of a new or upgraded innovation process. ICT = information and communication technology; R&D = research and development. Standard errors in parentheses. ***p < 0.01, **p < 0.05, *p < 0.1. FIRM CAPABILITIES ARE CONSTRAINED 117
INDUSTRY CASE STUDY D
Protection from global good Pakistan than in India because only a few practices limits the spread of OEMs are “competing” behind even higher world-class firm capabilities in the import tariffs on both CBUs and auto parts. automotive industry The experience of the Indian auto-parts and commercial vehicle sectors shows that a grad- The automotive sector is one of the most ual reduction of import tariffs, far from lead- important industries globally and in South ing to the debilitation of an industry, could be Asia, contributing 19 million direct and indi- a powerful catalyst to its global success. rect jobs in India alone. The potential for Converging toward international environ- South Asia to become globally competitive in mental and safety standards would further this sector is shown by the experience of encourage automotive firms in South Asia to Indian auto-parts manufacturers, who became adopt and contribute to international good world leaders by first having acquired techni- practices. cal and managerial skills from leading OEMs in India, followed by a process of serving South Asia’s opportunity: Become increasingly discerning customers in competi- globally competitive in a major industry tive export markets. Although the level of investment in R&D remains low, a few lead- The automotive sector (including the auto- ing global manufacturers are moving their parts industry) is a key contributor to jobs R&D centers to India, pointing to the region’s and economic growth in India and Pakistan potential to be at the heart of the technologi- with significant domestic growth potential. cal revolution taking hold in a critical indus- Approximately 19 million and 2.5 million try that has important ramifications for people are employed (directly or indirectly) in others, such as electronics, machining, and the Indian and Pakistani auto industries, tooling. respectively. There is much scope for further The challenge for the region is twofold. growth with rising income. At approximately First is to spread these world-class firm capa- 20 cars per 1,000 inhabitants, the car pene- bilities throughout the industry, from OEMs tration rate in India is one-sixth the Chinese and tier 1 companies to tier 2 and tier 3 sup- level, which is one-sixth the level in the United pliers. In effect, large productivity gaps persist States. Car penetration is 30 percent lower in in the sector, with most OEMs and their Pakistan than in India because of higher suppliers having subscale or fragmented prices that result from less competition and operations with low capacity use, quality lev- lower productivity. els, and investments in skills below interna- The sector has also a lot of room to grow tional benchmarks. A second—and in export markets. Even though India is the connected—challenge is that of moving up world’s sixth-largest auto producer by the GVC through greater innovation, invest- volume, it holds less than 1 percent of the ment in R&D, and commercialization of new global export market compared with more products, all of which remain below global than 3 percent for China, 4.5 percent for averages, with local suppliers relying primar- Korea, and 7 percent for Mexico (figure 6.6). ily on build-to-print models. In India, exports accounted for only 5 percent Policies such as high import tariffs on com- of the total sales of the average auto pletely built units (CBUs) of passenger cars, firm, compared to 16 percent in China. which helped attract market-seeking OEMs Intraregional trade is not significant— in the 1990s and 2000s, are now slowing Pakistan, the closest largest automaker, down the spread of world-class managerial remains a relatively closed market and did not good practices. The situation is worse in feature in the top 40 exporters in 2014. 118 SOUTH ASIA’S TURN
FIGURE 6.6 Share of exports in production in the automotive sector, 2014
25
20
15 ercent P 10
5
0 . ep zil tates exico R pain hina Italy ra India Japan anada France S C oland urkey ustria B ermany S M C ingdom elgiumepublic P ungary epublicSwedenT A omania orea, B H hailand R G K K R T R nited etherlands U ech N nited lovak U Cz S
Source: World Integrated Trade Solutions database 2013–14, http://wits.worldbank.org/.
The export performance of the auto-parts Bangalore. There is a similar trend in the elec- industry in India shows the potential of the tronics industry, which is increasingly impor- region. Forty-four percent of auto parts pro- tant to the automotive industry; for example, duced in India are exported. Exports of auto Samsung has one of its three global R&D components increased by 15 percent per year centers in Noida, India. The region’s global from 2009 to 2014 (ACMA and McKinsey preeminence in the ICT industry is no doubt a and Co. 2012).11 OEMs and tier 1 firms key explanation for the arrival of these global accounted for 80 percent of end customers, manufacturing R&D centers—R&D in man- with the share of sophisticated end markets ufacturing increasingly relies on ICT skills. rising, indicating high-quality exports. Less This is all very good news for the region, than a decade ago, only 35 percent of parts which could position itself as offering a com- were sold to OEMs, with the rest going to the bination of strong R&D and manufacturing much less demanding aftermarket. An capabilities with competitive labor costs increasing number of Indian auto-parts man- together with a very large and fast-growing ufacturers have become first-class global com- market. panies, such as Bharat Forge and Motherson Sumi, which supply most global car compa- South Asia’s challenges nies and have developed world-class capabilities. Spreading strong capabilities evenly to the In fact, India is showing that leading-edge rest of the industry R&D can be done in South Asia at a fraction Unfortunately, most of the automotive indus- of the price. Leading foreign companies such try in South Asia does not achieve these levels as BMW, Mercedes, Renault-Nissan, Volvo, of manufacturing excellence. The average GM, and Honda are establishing R&D cen- labor productivity of the 500 automotive ters in India, emulating Bosch, which is firms surveyed in India by the World Bank already conducting most of its global R&D in was less than one-third the level in China, India with 15,000 workers based in with Pakistan further behind (figure 6.7). FIRM CAPABILITIES ARE CONSTRAINED 119
FIGURE 6.7 Value added per worker in the automotive sector
30,000
25,000
20,000
15,000 ollars D 10,000
5,000
0 All automotive Auto exporters Auto nonexporters India China Pakistan Vietnam
Source: World Bank Enterprise Surveys for India (2014), Pakistan (2013), China (2013), and Vietnam (2009).
Firm size, export orientation, and share of 3 suppliers. External rejection rates are one foreign ownership are positively associated good measure of quality. External rejection with productivity in both India and Pakistan. rates are product specific; therefore we use The first explanation for low productivity the example of seat makers in India (Sutton is low scale at the plant level. In India, only 4 2004). The international best-practice stan- of the 18 OEMs operate at the industry stan- dard for seat makers is between 100 and 500 dard for efficiency of 100,000 units per defects per million produced (ppm). In India, model. In contrast, at least three Maruti- some leading suppliers were nearing 120 Suzuki models achieve more than 200,000 ppm, but one-fifth were experiencing rates as units annually and are profitable.12 Hyundai, high as 2,000 ppm (Sutton 2004).13 As 40 Honda, and Mahindra & Mahindra also percent of the value added of a car lies in the have managed to cross the 100,000-unit effi- tier 2 and tier 3 segments, the competitiveness ciency mark. Pakistan also suffers from sub- of the auto industry depends on its ability to scale production in all segments except improve quality, deliveries, and efficiency in tractors. In China, however, 25 out of 27 these segments. OEMs have reached this level. In 2014, 47 Many of these quality issues can be related models were produced at annual volumes to large skill gaps, which are prevalent among higher than 100,000 units, including 22 mod- both production and nonproduction workers, els at more than 200,000 units. particularly managers. As final products Low scale is compounded by low capacity become more complex, manufacturing pro- utilization in many OEMs. India produced cesses require an increasingly diverse range of 4 million cars in 2013 despite a production problem-solving skills, instead of traditional, capacity of 6 million—66 percent capacity linear ones. In China, 90 percent of auto firms utilization compared to 90 percent in China. provide training to their employees, as Capacity utilization among OEMs in Pakistan opposed to 37 percent of firms in India. Only is below 50 percent. Furthermore, low capac- 43 percent of nonproduction workers in auto ity utilization among OEMs in a vertically firms are formally trained in India, compared integrated industry, such as the automotive to nearly 70 percent in China (figure 6.8). sector, is often reflected in low capacity utili- When asked about its main challenge to zation among suppliers. growth, the chairman of Bharat Forge said, The next explanation for low productivity “Talent.” Even though there are numerous is low quality, especially among tier 2 and tier publicly subsidized training programs, leading 120 SOUTH ASIA’S TURN
FIGURE 6.8 Overview of skills in Indian and Chinese automotive firms
100 80 60
ercent 40 P 20 0 Share of employees Share of firms Share of Share of Share of firms completed secondary providing formal production nonproduction with quality education training programs workers trained workers trained certifications India China
Source: World Bank Enterprise Surveys for India (2014) and China (2013).
FIGURE 6.9 Expenditures on R&D in India and globally in India and Pakistan are not demonstrating sufficient and quick use of design capabilities. 5 In most cases, specifications are provided by 4 the customer and the execution is done by the sales 3 local firm on the basis of build-to-print plans of provided by the OEM or tier 1 company. 2 Design abilities are closely linked to a firm’s ercent P 1 R&D capabilities. The average expenditure on R&D in Indian auto firms ranges between 0 0 percent and 2 percent of sales, much orp uto zuki otors C A u lower than the global average of 4.7 percent oto ajaj -S average M B TVS M aruti (figure 6.9). ero lobal H M G 2010–11 2013–14 2013–14 Leading South Asian firms built their Source: Odgers Berndtson India 2014. capabilities through exposure to global good practices firms invest in their own training programs to The world-class Indian auto-parts firms have ensure a constant supply of talented line man- acquired their capabilities by linking with agers, business managers, and floor-level tech- leading global firms, either as suppliers or nical workers. through technology agreements. Many started as suppliers to Maruti-Suzuki, which trans- Moving up the GVC in design and R&D ferred its technical and operational know- capabilities how, and above all, according to its past Although increasingly important, R&D capa- chairman, its management ethic. Co-location bilities are spread thin across the automotive in clusters also greatly facilitated these industry in South Asia. Global OEMs now transfers. Subsequent exposure to demanding expect design capabilities from firms at all customers in competitive export markets levels of the value chain, because subcontract- increased economies of scale and induced ing makes sense only when the supplier can quicker adoption of modern, international be held responsible for entire modules of standards, leading to faster and deeper knowl- tasks. During interviews, global tier 1 compa- edge transfers and technological spillovers nies mentioned that design capabilities are than available in less-advanced and less- becoming critical factors in selecting tier 2 competitive South Asian automotive markets subcontractors. With a few exceptions, firms (box 6.5). FIRM CAPABILITIES ARE CONSTRAINED 121
BOX 6.5 How leading Indian automotive firms acquired their world-class skills
Alliances with OEMs for early capability were making interior locks and needed zinc parts, acquisition Sandhar improved its ability to work with zinc. “If we do a good job with locks, we would sug- Leading suppliers in India first acquired their gest we could handle the client’s plastic needs,” technical and managerial capabilities from lead- said the senior executive at Sandhar. Once MSSL ing domestic OEMs like Maruti-Suzuki and Hero had acquired new technology and fulfilled its Honda (now Hero MotoCorp). According to a contract with a customer, management would senior executive at Motherson Sumi Systems Ltd. ask their engineers, “What more could we do (MSSL), “We did not have a background in auto- with it?” MSSL expanded from basic plastic motive pre-Maruti. Through a technical collabo- components to building tooling and injection- ration . . . with Sumitomo, we set up a wiring molding machines to deliver complex plastic harness in India. Within 1–2 years, the [technical products. MSSL initially imported wires for their collaboration] became a [joint venture] leading to wire harness products, but then bought copper to MSSL.” At that time, more than 85 percent of manufacture wires. This allowed them to increase MSSL’s sales were to Maruti. Similarly, in the sales to existing customers and enter new product 1980s, Maruti was the most important customer markets. Bharat Forge has managed to break into of Bharat Forge, which became India’s largest design, engineering, testing, and other higher- auto-part exporter. The story is similar for firms value-added services through such approaches involved in motorcycles. Hi-Tech Gear Ltd. and in-house R&D. (HTGL) started as a preferred supplier of gear- cutting tools to Hero Honda. After receiving their first order from Hero Honda, HTGL moved into Co-location to facilitate learning and aluminum- and steel-based tubular parts. HTGL business development chose this product segment because steel of the standard required by Hero Honda was not avail- Physical proximity to the customer helped able in India. HTGL worked with Hero Honda upgrade products and processes, co-location to locate two steel suppliers. They climbed a steep allowed MSSL to hold frequent meetings with learning curve to meet the high standards Hero Honda, and new requirements would demanded by Hero Honda. At the same time, sometimes emerge during the course of these Honda was convincing some of its own suppliers discussions. “We make the decision to co-locate to locate in India and, according to HTGL, the based on several factors. Is the job big enough in “interactive working relationship between size to justify co-location? Is the OEM reputed HTGL, Honda’s suppliers, and [Hero Honda] enough to learn from? Is there potential to helped HTGL learn rapidly.” increase share of wallet? Is there potential to learn something new completely?” asked the Rapid absorption and adaptation of MSSL senior executive. Similarly, proximity to technologies the customer helped Sandhar become a designer as well as a supplier of locks and mirrors. As Leading firms became expert at absorbing the reported by the Sandhar senior executive, “A technology acquired through technical collabora- leading OEM was having trouble with one of its tion and joint ventures and adapting it to expand Indian suppliers which wasn’t meeting delivery into production of related products—both for- or quality standards. During a lunch with the cli- ward and backward. For instance, when they ent I proposed myself even though we had never
(continues next page) 122 SOUTH ASIA’S TURN
BOX 6.5 How leading Indian automotive firms acquired their world-class skills (continued)
made locks before.” One thing led to another, plague,” said the MSSL senior executive. and, pleased with Sandhar’s performance in metal Working for a demanding customer meant that sheets, Hero Honda helped it set up a technical the firm was forced to be efficient, adopt interna- collaboration with one of their lock suppliers in tional standards, and keep costs down. For Japan. Sandhar became their single-source sup- Bharat Forge, exports started as early as 1995. plier of locks. Bharat Forge bought plants in “Exports challenged us to design, develop, manu- Germany to be physically closer to its leading facture, and supply products to discerning customers. customers in global markets. This in turn moti- vated us to scale up the value chain and adopt Diversifying to more sophisticated and new technologies,” said the Bharat Forge senior demanding export markets executive. An HTGL senior executive noted that, “I wanted to find the most discerning customers, Many auto-parts suppliers increased production whether in India or abroad. I would bend over and productivity by serving sophisticated and backwards to work with them because I found competitive export markets. “Many players at we learnt the most when we worked with OEMs that time went into aftermarkets because barriers who held very high standards.” to entry were low and there was promise of high margins, but we avoided this route like the Source: World Bank interviews in India.
The spread of world-class capabilities, auto-parts sector to global good practices and including in design and R&D, is limited pushed it to improve productivity and build by protection from global good design skills to compete in world markets. practices Exports now constitute more than 40 percent of production. Millions of local jobs were High levels of import tariffs for final cars in created. Similarly, the decline in import duties India and Pakistan contributed to attracting on commercial vehicles led to increases in pro- foreign OEMs, but now these protections are duction and employment, and the subsector reducing their incentives to export and slow- currently shows a trade surplus (figure 6.11). ing down the diffusion of good practices. The International experience shows the positive situation is worse in Pakistan than in India, impact of trade liberalization. A large and where only three OEMs “compete” domesti- globally competitive automotive industry cally behind higher import tariffs (76 percent developed rapidly in Mexico following the and 60 percent, respectively). The experience implementation of the North American Free of the Indian auto-parts and commercial vehi- Trade Agreement (NAFTA). More recently, cle sectors shows that a gradual reduction of between 2002 and 2005, China reduced import tariffs, far from debilitating an indus- import tariffs from 90 percent to 25 percent. try, could be a powerful catalyst for its global Subsequent competitive pressure was felt success and support the adoption and spread throughout the automotive value chains. of world-class capabilities. Tariffs on auto Net trade in cars rose from $672 million to parts and commercial vehicles in India have $5.3 billion, and output among suppliers gradually fallen since the 1990s with a con- increased by 25 percent per year. comitant sharp rise in production and exports Obsolete and unpredictable safety and (figure 6.10). Competition exposed the environmental standards have also shielded FIRM CAPABILITIES ARE CONSTRAINED 123
FIGURE 6.10 Domestic production of auto parts and nominal tariff reduction in India
60 15
40 10 ercent
P 20 5 ollars (billions) D 0 0
1996–97 1997–98 1998–99 2000–01 2001–02 2002–03 2003–04 2004–05 2005–06 1999–2000 Nominal tariff (%) Domestic production
Source: World Integrated Trade Solutions database, http://wits.worldbank.org/; ACMA and McKinsey and Co. 2012; Narayan and Vashisht 2008. Note: Domestic production is stated in current prices.
FIGURE 6.11 Exports of commercial vehicles and nominal tariff reduction in India
90 500 80 70 400 60 50 300
ercent 40
P 200 30 ollars (millions)
20 100 D 10 0 0
1993–94 1994–95 1995–96 1996–97 1997–98 1998–99 2000–01 2001–02 2002–03 2003–04 2004–05 2005–06 2006–07 1999–2000 Nominal tariff (in %) Nominal tariff (%) Exports
Source: World Integrated Trade Solutions database, http://wits.worldbank.org/; ACMA and McKinsey and Co. 2012; Narayan and Vashisht 2008.
the South Asian automotive industries from Euro III. To complicate matters, India’s auto- global good practices and have reduced the motive sector suffered from frequent changes incentives for firms to invest and export. The and backtracking in regulations for fuel taxes Euro I emissions standards, introduced in and emission standards. The situation is worse the European Union in 1983, came into force in Pakistan, which has not yet adopted Euro II in India in 1996. Although the Euro II stan- norms. dards were introduced in the European Union The extended version of this case study is in 1997, they were not applied throughout available online at www.worldbank.org India until 2005. Europe has now moved to /SouthAsiaCompetes. 124 SOUTH ASIA’S TURN
Notes innovation module from the Enterprise Survey was used: the Democratic Republic of 1. The analysis in Fernandes and Pakes (2008) Congo, Ghana, Kenya, Namibia, Nigeria, is based on the 2002 and 2005 rounds of South Sudan, Sudan, Tanzania, Uganda, and the Investment Climate Survey (now the Zambia. Although South Asia and Africa are Investment Enterprise Survey) in India, and very different in many respects, levels of the 2005 round of the Investment Climate gross domestic product (GDP) per capita are Survey in Sri Lanka. The authors measure similar among the sampled countries. factor under- and overutilization by estimat- 6. All equations control for sector effects using ing a Cobb-Douglas production function to 2-digit International Standard Industrial calculate the marginal products of labor and Classification (ISIC) dummies. Country mod- capital and comparing these marginal prod- els are estimated individually, and a pooled ucts with factor costs (wages for labor and sample with country fixed effects is estimated interest and depreciation rates for capital). for the South Asia region. The regional The estimation approach followed Olley and pooled estimates are dominated by India Pakes (1996) to correct for entry and exit because of its large number of observations. (selection) and the endogeneity of input use All individual country estimates use sampling (that inputs may be selected at the same time weights. as output). 7. In a study for Mexico, Lopez-Acevedo and 2. The reported underutilization rates do not Tan (2003) found that training produced imply that employment would rise by the large and statistically significant wage and indicated factor if firms were able to use fac- productivity outcomes, that joint training tors optimally—as firms hired more labor, and R&D yielded larger returns than invest- wages would rise, putting downward pres- ments in one or the other, and that both sure on optimal employment levels. training and technology investments enabled 3. These estimates are based on the World Bank firms to improve their relative position in the Enterprise Survey for each country for the wage and productivity distribution. year indicated: Pakistan (2010), Sri Lanka 8. These differences are significantly different (2011), Bangladesh and Nepal (2013), and for almost all types of innovation in India India (2014). and Nepal, for product and marketing inno- 4. To obtain these estimates, using an approach vation in Bangladesh, and for organizational similar to that of Fernandes and Pakes innovation in Pakistan. (2008), marginal products of labor and capi- 9. Foreign-owned firms are defined as those tal are estimated using country-specific with more than 25 percent ownership by pri- Cobb-Douglas production functions with vate foreign individuals, companies, or orga- sector dummies and then compared with the nizations. The differences are statistically prevailing wage and rental rates for each significant for product and process innova- country. However, because of the absence of tion in Bangladesh and India, marketing in panel data, the estimation approach does not Nepal, and process innovation and market- explicitly account for firm selection and ing in Pakistan. endogeneity of input choices. Although the 10. The estimates are robust to alternative meth- extent of the implied basis is difficult to odologies. Because the generalized structural judge, results reported by Fernandes and equation model (GSEM) methodology is Pakes (2008) for India show that coefficient more robust with large samples and well- estimates obtained by ordinary least squares specified models, we also estimate the same (OLS) lie within two standard errors of models using three-stage least squares (3sls). results obtained with the Olley-Pakes estima- The disadvantage of this methodology is tor when firms are allowed to optimize labor that it uses the sample of the stage with volumes in every period, even when control- lower number of estimates and does not ling for firm exit. allow for a mixed process, because all 5. In the discussion that follows, South Asia’s the stages have to be estimated linearly. On performance is benchmarked against the the other hand, it is computationally less average in the Africa region, which includes demanding than full information maximum the following 10 countries in which the likelihood and still addresses the issue of FIRM CAPABILITIES ARE CONSTRAINED 125
endogeneity instrumenting at each stage. Countries Have Low Productivity?” American The results for the returns to innovation, Economic Review: Papers & Proceedings 2010 although larger than those under GSEM, are 100 (2): 619–23. identical in statistical significance. Bloom, N., R. Sadun, and J. Van Reenen. 2016. 11. World Bank and UNCTAD (2015) estimates “Management as Technology?,” Harvard auto-parts exports at $2 billion in 2009 and Business School Working Paper 16-133, $6 billion in 2014. Harvard University: Cambridge, MA. 12. Maruti-Suzuki Annual Reports. Bloom, N., and J. Van Reenen. 2007. “Measuring 13. These data were confirmed by World Bank and Explaining Management Practices Across site visits to Indian auto clusters in December Firms and Countries.” Quarterly Journal of 2014. Economics 122 (4): 1341–1408. ———. 2010. “Why Do Management Practices References Differ across Firms and Countries?” Journal of Economic Perspectives, American Economic Aghion, P., U. Akcigit, and P. Howitt. 2013. Association 24 (1): 203–24. “What Do We Learn from Schumpeterian Blundell, R., R. Griffith, and J. Van Reenen. 1995. Growth Theory?” NBER Working Paper “Dynamic Count Data Models of Technological 18824. National Bureau of Economic Research, Innovation.” The Economic Journal 105 (429): Cambridge, MA. 333–44. Aghion, P., N. Bloom, R. Blundell, R. Griffith, and Brambilla, I. 2009. “Multinationals, Technology, P. Howitt. 2005. “Competition and Innovation: and the Introduction of Varieties of Goods.” An Inverted-U Relationship.” Quarterly Journal of International Economics 79 (1): Journal of Economics 120 (2): 701–28. 89–101. ACMA (Automotive Component Manufacturers Bugamelli, M., and P. Pagano. 2004. “Barriers to Association of India) and McKinsey and Co. Investment in ICT.” Applied Economics 36 2012. Auto Component Industry—Ready for (20): 2275–86. the Transition. New Delhi: ACMA. Carlino, G., and W. R. Kerr. 2014. “Agglomeration Baily, M. and Gersbach, H. 1995. “Efficiency in and Innovation.” NBER Working Paper 20367, Manufacturing and the Need for Global National Bureau of Economic Research, Competition.” In Brookings Papers on Cambridge, MA. Economic Activity: Microeconomics. Chun, H. 2003. “Information Technology and Washington, DC: The Brookings Institution. the Demand for Educated Workers: Bairoch, P. 1988. “Cities and Economic Disentangling the Impacts of Adoption versus Development from the Dawn of History to the Use.” Review of Economics and Statistics Present.” University of Chicago Press, Chicago. 85 (1): 1–8. Bartel, A. P., and F. R. Lichtenberg. 1987. “The Cirera, X., D. Lederman, J. A. Máñez, M. E. Comparative Advantage of Educated Workers Rochina, and J. A. Sanchis. 2015. “The Export in Implementing New Technology.” The Review Productivity Link for Brazilian Manufacturing of Economics and Statistics 69 (1): 1–11. Firms.” Economics Discussion Papers 2015-26. Bayo-Moriones, A., and F. Lera-López. 2007. “A Kiel Institute for the World Economy. http:// Firm Level Analysis of Determinants of ICT www.economics-ejournal.org / economics/ Adoption in Spain.” Technovation 27 (6/7): discussionpapers/2015-26. 352–66. Commander, S., R. Harrison, and N. Menezes- Berman, E., J. Bound, and S. Machin. 1994. Filho. 2011. “ICT and Productivity in “Changes in the Demand for Skilled Labor Developing Countries: New Firm-Level within U.S. Manufacturing: Evidence from the Evidence from Brazil and India.” Review of Annual Survey of Manufactures.” Quarterly Economics and Statistics 93 (2): 528–41. Journal of Economics 109 (2): 367–97. Crepon, B., E. Duguet, and J. Mairesse. 1998. Bloom, N., B. Eifert, A. Mahajan, D. McKenzie, “Research, Innovation and Productivity: An and J. Roberts. 2013. “Does Management Econometric Analysis at the Firm Level.” Matter? Evidence from India.” Quarterly Economics of Innovation and New Technology Journal of Economics 128 (1): 1–51. 7 (2): 115–58. Bloom, N., A. Mahajan, D. McKenzie, and Criscuolo, P. 2009. “Inter-firm Reverse J. Roberts. 2010. “Why Do Firms in Developing Technology Transfer: The Home Country 126 SOUTH ASIA’S TURN
Effect of R&D Internationalization.” Industrial Hasan, R., D. Mitra, and A. Sundaram. 2013. and Corporate Change 18 (5): 869–99. “What Explains the High Capital Intensity of Cusolito, Ana. 2009. “Competition, Imitation, and Indian Manufacturing?” Indian Growth and Technical Change: Quality vs. Variety.” Policy Development Review 6 (2): 212–41. Research Working Paper 4997, World Bank, Haskel, J., and Y. Heden. 1999. “Computers and Washington, DC. the Demand for Skilled Labour: Industry- and Dasgupta, S., D. Agarwal, A. Ioannidis, and Establishment-Level Panel Evidence for the S. Gopalakrishnan. 1999. “Determinants of UK.” Economic Journal 109 (454): 68–79. Information Technology Adoption: An Hollenstein, H. 2004. “Determinants of the Extension of Existing Models to Firms in a Adoption of Information and Communication Developing Country.” Journal of Global Technologies (ICT): An Empirical Analysis Information Management 7 (3): 30–40. Based on Firm-Level Data for the Swiss Business Doms, M., T. Dunne, and K. R. Troske. 1997. Sector.” Structural Change and Economic “Workers, Wages, and Technology.” Quarterly Dynamics 15 (3): 315–42. Journal of Economics 112 (1): 253–90. Keller, W. 2004. “International Technology Dougherty, S., R. Herd, and T. Chalaux. 2009. Diffusion.” Journal of Economic Literature “What Is Holding Back Productivity Growth in 42 (September): 752–82. India? Recent Microevidence.” OECD Journal: Kerr, W. R., and R. Nanda. 2014. “Financing Economic Studies 2009 (1): 1–22. Innovation.” NBER Working Paper 20676, Fabiani, S., F. Schivardi, and S. Trento. 2005. “ICT National Bureau of Economic Research, Adoption in Italian Manufacturing: Firm-Level Cambridge, MA. Evidence.” Industrial and Corporate Change Kowtha, N. R., and T. W. I. Choon. 2001. 14 (2): 225–49. “Determinants of Website Development: A Fabiani, S., F. Schivardi, and S. Trento. 2005. “ICT Study of Electronic Commerce in Singapore.” Adoption in Italian Manufacturing: Firm-Level Information and Management 39 (3): 227–42. Evidence.” Industrial and Corporate Change Kretschmer, T., E. J. Miravete, and J. C. Pernías. 14 (2): 225–49. 2012. “Competitive Pressure and the Adoption Fernandes, A., and A. Pakes. 2008. “Factor of Complementary Innovations.” American Utilization in Indian Manufacturing: A Look at Economic Review 102 (4): 1540–70. the World Bank Investment Climate Surveys Lefebvre, L. A., É. Lefebvre, E. Elia, and H. Boeck. Data.” NBER Working Paper 14178, National 2005. “Exploring B-to-B E-Commerce Bureau of Economic Research, Cambridge, MA. Adoption Trajectories in Manufacturing Forman, C., A. Goldfarb, and S. Greenstein. 2005. SMEs.” Technovation 25 (12): 1443–56. “Geographic Location and the Diffusion of Lileeva, A., and D. Trefler. 2010. “Improved Access Internet Technology.” Electronic Commerce to Foreign Markets Raises Plant-Level Research and Applications 4 (1): 1–13. Productivity... for Some Plants.” Quarterly Fornahl, D., and T. Brenner. 2009. “Geographic Journal of Economics 125 (3): 1051–99. http:// Concentration of Innovative Activities in ideas.repec.org/a/tpr/qjecon/v125y2010i3p1051 Germany.” Structural Change and Economic -1099.html. Dynamics 20 (3): 163–82. Lopez-Acevedo, G., and H. Tan. 2003. “Mexico: Geroski, P. A. 1990. “Innovation, Technological In-Firm Training for the Knowledge Economy.” Opportunity, and Market Structure.” Oxford Policy Research Working Paper 2957, World Economic Papers 42 (3): 586–602. Bank, Washington, DC. Giunta, A., and F. Trivieri. 2007. “Understanding Love, P. E., Z. Irani, C. Standing, C. Lin, and the Determinants of Information Technology J. M. Burn. 2005. “The Enigma of Evaluation: Adoption: Evidence from Italian Manufacturing Benefits, Costs and Risks of IT in Australian Firms.” Applied Economics 39 (10): 1325–34. Small–Medium-Sized Enterprises.” Information Hall, B. H., J. Lerner. 2009. “The Financing of & Management 42 (7): 947–64. R&D and Innovation.” NBER Working Paper Lucchetti, R., and A. Sterlacchini. 2004. “The w15325. National Bureau of Economic Adoption of ICT among SMEs: Evidence from Research, Cambridge, MA. an Italian Survey.” Small Business Economics Haller, S., and I. Siedschlag. 2011. “Determinants 23 (2): 151–68. of ICT Adoption: Evidence from Firm-Level McKenzie, D., and C. Woodruff. 2015. “Business Data.” Applied Economics 43 (26): 3775–88. Practices in Small Firms in Developing FIRM CAPABILITIES ARE CONSTRAINED 127
Countries.” NBER Working Paper 21505, Equipment Industry.” Econometrica 64 (6): National Bureau of Economic Research, 1263–97. Cambridge, MA. Porter, M. 1990. “The Competitive Advantage of Mohnen, P., and B. H. Hall. 2013. “Innovation Nations.” Harvard Business Review 68 (2): 73–93. and Productivity: An Update.” Eurasian Sutton, J. 2004. “The Auto-Component Supply Business Review 3 (1): 47–65. Chain in China and India—A Benchmark Narayan, B. G., and P. Vashisht. 2008. Study.” Research Paper No. EI34, London “Determinants of Competitiveness of the Indian School of Economics, London, England. Auto Industry.” Working Paper 201, Indian Teo, T. S. H., V. K. G. Lim, and R. Y. C. Lai. 1998. Council for Research on International Intrinsic and Extrinsic Motivation in Internet Economic Relations, New Delhi. Usage. Omega 27 (1): 25–37. Narula, R., and A. Zanfei. 2005. “Globalisation of Teo, T. S. H., and M. Tan. 1998. “An Empirical Innovation: The Role of Multinational Study of Adoptors and Non-adopters of the Enterprises.” In Oxford Handbook of Internet in Singapore.” Information and Innovation, edited by J. Fagerberg, D. Mowery, Management 34 (6): 339–45. and R. Nelson 318–45. New York: Oxford Thong, J. Y. L. 1999. “An Integrated Model of University Press. Information Systems Adoption in Small Nickell, S. 1996. “Competition and Corporate Business.” Journal of Management Information Performance.” Journal of Political Economy Systems 15 (4): 187–214. 104 (4): 724–46. Walczuch, R., G. Van Braven, and H. Lundgren. Odgers Berndtson India. 2014. “Indian 2000. “Internet Adoption Barriers for Small Automotive Sector Gearing Up to Accelerate.” Firms in the Netherlands.” European Search Intelligence Automotive Business Management Journal 18 (5): 561–72. Leaders Survey. World Bank. 2013. World Bank Enterprise Survey, OECD (Organisation for Economic Co-operation China. Washington, DC: World Bank. http:// and Development). 2012. “Community Survey www.enterprisesurveys.org/. on ICT Usage in Enterprises.” OECD. ———. 2014. World Bank Enterprise Survey, OECD/Eurostat. 2005. Oslo Manual: Guidelines India. Washington, DC: World Bank. http:// for Collecting and Interpreting Innovation www.enterprisesurveys.org/. Data. 3rd Edition. Paris: OECD Publishing. World Bank and UNCTAD (United Nations Olley, G. S., and A. Pakes. 1996. “The Dynamics Conference on Trade and Development). 2015. of Productivity in the Telecommunications World Integrated Trade Solutions.
PART 3
The Way Forward
Potential for Increased Growth through Policy Reforms 7
he potential benefits of becoming recent decade of growth in the 2000s, but more productive and, therefore, more well above the experience thus far in the Tcompetitive are discussed below at current decade (about 0.9 percentage points the macro, sectoral, and firm levels. The pol- per year). icy actions underlying these scenarios are In the baseline scenario, South Asia’s rela- discussed in the next chapter. tively young population and strong produc- tivity growth underpin a substantial increase Macro benefits: Faster export in per capita income through 2030. Unlike growth through higher productivity China, where total population is projected to decline after 2025, South Asia’s population South Asia has tremendous potential to is expected to continue growing. In particu- raise incomes through policies that enhance lar, India will become the world’s most popu- productivity and increase exports. The lous country shortly after 2020. The number economy-wide implications of these policies of skilled workers—those with a secondary can be assessed using a global computable school degree or higher—is expected to rise general equilibrium model, which is used here by an average of 84 percent by 2030, rang- to consider how a reduction in both interna- ing from 42 percent to 125 percent across tional and domestic trade costs could increase the countries in the region. The shift from income growth in South Asia through 2030. agriculture to industry and services is The model uses the latest Global Trade expected to accelerate, driven by rising Analysis Project (GTAP) dataset with a 2011 incomes and much lower wages in agricul- base year, and its key assumptions are cali- tural versus nonagricultural activities. The brated to the Middle of the Road scenario of agricultural workforce is projected to the Integrated Assessment Modeling increase only 10 percent, but the labor force Consortium (IAMC).1 (Annex 7A provides a in nonagricultural activities is projected to brief introduction to the model.) Most impor- rise by approximately 60 percent (adding up tant, productivity in South Asia is calibrated to an average increase in the overall labor to contribute an average of 2 percentage force of 30 percent). Consequently, the share points to total gross domestic product (GDP) of the region’s workers employed in nonagri- growth through 2030, consistent with what cultural activities is projected to increase the region was able to achieve during its best from 40 percent to 50 percent.
131 132 SOUTH ASIA’S TURN
Real GDP in the region is projected to rise much more moderate 27 percent. This reflects by 6 percent per year (tripling by 2030), giv- a limited increase in the agricultural labor ing South Asia one of the highest GDP growth force and availability of land, some decelera- rates in the world and mirroring the growth tion in yield growth, and relatively high in East Asia. Per capita income would rise income growth, which shifts demand toward from about $1,400 in 2011 to about $3,400 more highly valued agricultural goods (such in 2030 (in 2011 prices and market exchange as fruits, vegetables, and dairy products) and rates), but remain well below the average of nonagricultural goods. On the other hand, $55,000 in high-income countries. Despite services exports would more than triple rapid gains in productivity, increases in the and manufacturing exports rise by nearly number of workers and especially in the vol- 300 percent, as South Asia’s rapid labor force ume of capital represent the most important growth, rising skill endowments, and rapid source of growth, as has been true in the past. productivity growth (implicit in the baseline Under these assumptions, South Asia growth scenario) capture a growing share of becomes the world’s fastest-growing region the global market in higher-value-added prod- for exports. Merchandise exports (in constant ucts (figure 7.1). Overall, the region would U.S. dollars) rise by 264 percent between more than double its global export share in 2011 and 2030, compared with a 138 percent manufacturing and increase its global share of increase for all developing countries and 83 services exports by 75 percent. percent for the world. The majority of this Within manufacturing, more skill-intensive growth comes from manufacturing and ser- sectors account for a larger portion of the vice exports, whereas exports of agricultural overall growth: export growth rates range goods (excluding processed food) rise by a from 193 percent for textiles to 220 percent
FIGURE 7.1 Continued productivity growth could lead to substantial gains in South Asia’s global market share
Services
Manufacturing
Agriculture
Electronics
Motor vehicles
Wearing apparel
051015 20 25 Percent 2011 Baseline, 2030 Further improvements Further reductions Further improvements in international LPI, in domestic trade costs, in domestic LPI, 2030 2030 2030
Source: World Bank calculations. Note: See annex 7A for a detailed description of the scenarios. LPI is the Logistics Performance Index. POTENTIAL FOR INCREASED GROWTH THROUGH POLICY REFORMS 133
for wearing apparel to 400 percent for motor performance on the domestic component of vehicles to 435 percent for electronics. the Logistics Performance Index (LPI)), could Starting from a relatively low base, by 2030 raise total export growth over the forecast South Asia more than triples its share of period from 256 percent in the baseline to global exports of electronics and motor vehi- nearly 340 percent and increase the trade-to- cles, and comes close to doubling its already GDP elasticity from 1.1 to 1.5. An alternative significant market share of wearing apparel scenario that targets the international compo- (excluding textiles and leather). In electronics nent of the LPI through more rapid imple- and motor vehicles, nearly all the growth mentation of ongoing improvements in comes from India while other countries— port-to-port trade and transportation costs despite rapidly increasing their exports— would generate smaller effects, because, remain small players in global markets. Even according to the GTAP database, the margins though India’s exports in these sectors between free on board (FOB) and cost, insur- increase more rapidly than China’s, by 2030 ance, and freight (CIF) prices are already low India’s exports of motor vehicles only just (4 percent to 8 percent on average). Exports approach China’s levels in 2011, and its 2030 would increase by 11 percent in 2030 relative electronics exports remain an order of magni- to the baseline, or roughly half the increase tude below what China exported in 2011. In generated by the first scenario. wearing apparel, performance is more equal Improved economic performance under across the region. these scenarios gives rise to increased labor If productivity growth is instead closer to demand in manufacturing. Employment rises what the region has achieved in the current relative to the baseline in agribusiness and decade, the growth of exports and incomes falls in sectors such as agricultural crops; fos- will be significantly slower. Exports would sil fuels; and trade, transport, and business rise by 5.7 percent per year, compared to 6.9 services—the last because of increased effi- percent in the baseline, and gains in global ciency in providing trade and transport ser- market share of manufacturing and services vices. As the aggregate supply of labor is the would be significantly lower. However, even same across simulations, rising demand for with lower productivity growth, real GDP in labor because of lower trade costs leads to the region would still expand by 5 percent per higher wages.2 Reductions in logistics costs year, on a par with the developing country and in domestic trade costs raise the average average and well above the 2 percent GDP wage by approximately 12 percent, while the growth in high-income countries. The region drop in international trade costs increases the still increases its share of global export mar- average wage by only 1 percent. The average kets, although its performance suffers com- wage of unskilled and skilled workers rises by pared to China and the rest of East Asia. 17 percent and 8 percent, respectively, com- Slower productivity growth reduces South pared to the baseline. Asia’s share of the global apparel market by The efforts required to achieve the cost nearly 4 percentage points, and its shares of reductions differ across the three scenarios. automobiles and electronics by 0.5 percent- Thus, while the scenarios indicate the magni- age points each, compared to the baseline tude of the effect of policy improvements, scenario. decisions on what policies to undertake On the other hand, productivity-enhancing would also require identifying the needed improvements in trade facilitation and the measures and quantifying their costs. functioning of domestic markets could gener- ate additional gains in exports. Improvements Sectoral benefits: More jobs, higher in port infrastructure and warehousing and a earnings, greater inclusion reduction in burdensome customs regulations that would lower the region’s high logistics An additional perspective on how productiv- costs (comparable, for example, to improved ity improvements could affect the domestic 134 SOUTH ASIA’S TURN
economy can be gained by estimating an elas- markets would have little effect on exports ticity of substitution for South Asia’s from Bangladesh and Pakistan, but would exports—that is, an expected increase in the raise exports from India and Sri Lanka by region’s exports given a change in relative 19.0 percent and 22.5 percent, respectively, prices compared to South Asia’s competitors. consistent with the current production rela- (See annex 7A for the methodology used.) tionships between these countries and the The advantage of this partial equilibrium European Union. These results suggest that approach, compared to the general equilib- competition in the apparel markets is intense, rium approach in the previous section, is that with buyers highly sensitive to price changes. estimates can be obtained at a detailed prod- Thus, policies that increase productivity in uct level and with much higher precision. To apparel may be effective in generating large keep the exercise manageable, the analysis export gains. focuses on apparel, which accounts for Increases in exports in the apparel and 12 percent of the region’s merchandise textile sector would be particularly beneficial exports and employs 3 percent of the region’s to employment in South Asia, because the workers, and only on exports to the United textile and apparel sectors are relatively labor States and the 15 countries of the European intensive. (The procedure used to estimate Union (EU-15).3 labor demand is explained in annex 7A.) South Asian exports could increase sharply These jobs are also more likely to be formal. if prices were to rise more rapidly in China Evidence from Bangladesh suggests that than in South Asia. (Table 7.1 combines the permanent employment is more sensitive coefficients estimated by the procedure out- to increases in output than is informal lined in annex 7A with the import shares of employment—formal workers are more each country to generate the elasticity of sub- expensive, but they are also more productive stitution.) A 10 percent increase in Chinese (Diaz-Mayans and Sanchez 2004).5 prices would reduce U.S. imports from China Increased employment and wages also by 7.9 percent (almost $700 million), while would particularly benefit women (box 7.1). exports from Bangladesh, India, and Pakistan The demand for female labor in Bangladesh’s would increase by 13.6 percent ($519 mil- garment sector is more elastic than the lion), 14.6 percent ($414 million), and 25.3 demand for male labor—a 1 percent percent ($336 million), respectively.4 South increase in foreign sales is associated with a Asia’s emerging competitors could benefit 0.04 percent increase in female labor demand even more: Vietnam’s exports could increase and 0.02 percent increase in male labor by 37.7 percent ($2.2 billion) and Cambodia’s demand. In 2012, a 1 percent increase in by 51.3 percent ($1.1 billion). The same rela- expected wages was associated with an tive price increase in the European Union increase in the probability of female labor
TABLE 7.1 Southeast Asia benefits more than South Asia from a rise in Chinese apparel prices Elasticity of Substitution, U.S. and EU Imports
Bangladesh Cambodia India Pakistan Sri Lanka Vietnam U.S. 1.358a 5.125a 1.462a 2.531a 0.024 3.770a (0.039) (0.093) (0.027) (0.086) (0.058) (0.029) EU −0.238 2.525 1.895a −0.060 2.249a 1.644a (0.534) (2.031) (0.455) (1.068) (0.745) (0.960)
Source: World Bank calculations using data from the U.S. Department of Commerce’s Office of Textiles and Apparel (OTEXA). Note: Seemingly unrelated regression (SUR) estimates with homogeneity and symmetry and fixed effects and weights. The numbers in this table are elasticities for a 1 percent increase in prices of Chinese apparel. Standard errors are in parentheses. The change in the exports of a given country heading the columns would be the elasticity times a given percent change in China. A negative value means a decline. a = statistically significant at 1 percent. POTENTIAL FOR INCREASED GROWTH THROUGH POLICY REFORMS 135
force participation of 89 percent in Sri Lanka, percent, compared to 58 percent in Latin 31 percent in Bangladesh, 19 in India, and America and the Caribbean, 62 percent in 16 in Pakistan, although the importance of Europe and Central Asia, and 67 percent in this channel seems to have declined over East Asia (World Bank 2014). Because time.6 (See annex 7A for the estimation pro- apparel is a relatively low-skilled industry, cedure and table 7.2 for results.) This is par- employment opportunities in apparel that ticularly important for South Asia, which pay more than agriculture could potentially has a large pool of potential female workers: draw South Asia’s nonparticipating women its female labor participation rate is only 32 into the labor force.
BOX 7.1 Why focus on women?
There is growing evidence that gender gaps in the along with better nutrition and health invest- labor market and low female labor force partici- ments for school-aged girls (Jensen 2012). pation rates have a major impact on incomes. UÊ Also in India, a recent study on women Lower employment and wage rates for women employed in the textile industry finds that are estimated to reduce GDP per capita by as those with a longer history of employment much as 27 percent in some regions (Cuberes and tended to delay marriage and had a lower Teignier 2012). Another study estimates that rais- desired fertility rate. Moreover, these effects ing the female labor force participation rate to had spillovers within the family—the younger equal the male rate would raise the GDP in the sisters of women who worked in textiles United States by 5 percent, in Japan by 9 percent, also married later, and their younger brothers in the United Arab Emirates by 12 percent, and were less likely to drop out of school in the Arab Republic of Egypt by 34 percent (Sivasankaran 2014). (Aguirre et al. 2012). A third study finds that UÊ In Bangladesh, a study shows that the growth countries with a comparative advantage in of the garment sector was associated with a female-labor-intensive goods are characterized by 0.27 percentage point increase in girls’ school lower fertility, likely indicating that women delay enrollment over 1983–2000—a more sizeable marriage and childbearing, which can result in effect than a simultaneous supply-side inter- better pregnancy outcomes and better health vention of providing a subsidy for girls to (Do, Levchenko, and Raddatz 2014). remain in school (Heath and Mobarak 2012). At the microeconomic level, some studies Girls who live near a garment factory are show that female labor force participation and 28 percent less likely to be married and employment are beneficial for a number of 29 percent less likely to have given birth than household indicators, including children’s health those living in villages farther away from a and education and decision making about fertil- factory. ity and marriage. UÊ Also in Bangladesh, a recent study finds that formally employed women had fewer children UÊ In India, a randomized experiment finds that and possessed greater decision-making power an increase in labor market opportunities for over their own health expenses and formal women raised their labor force participation savings (either through insurance or a bank and their probability of going to school account) (Kabeer et al. 2013). instead of getting married or having children,
(continues next page) 136 SOUTH ASIA’S TURN
BOX 7.1 Why focus on women? (continued)
FIGURE B7.1.1 Working in garments with fewer children Number of children ages 5 years and younger in a household by female sector of employment
1.40 1.20 1.00 0.80 0.60 0.40 younger in household younger umber of children ages 5 and umber of children
N 0.20 0 Bangladesh India Pakistan Sri Lanka Female sector of employment Agriculture Textile and apparel Other industries Unemployed or not in the labor force
Source: World Bank estimation from household data.
The estimates in this report confirm that South (figure B7.1.1). Estimates from the Sri Lanka Asian households with working women, espe- 2008 household survey show that households cially in the textile and apparel sectors in India with women working in textiles or apparel and Pakistan, have fewer young children on spend almost twice as much per month on average than those with women working education per student (SL Rs 1,112) than in agriculture or those with women who households with women working in agriculture are not in the labor force or unemployed (SL Rs 657).
TABLE 7.2 Higher wages increase women’s labor force participation Combining the empirical evidence presented Marginal effects of female labor participation with respect to log so far in this section—the price sensitivity of expected wage South Asian apparel exports to high-income 1995 2000 2005 2012 markets and the responsiveness of employment to apparel output—provides an estimate of the Bangladesh — 1.646*** 0.141*** 0.306*** potential number of jobs that South Asia could India 0.551*** 0.426*** 0.410*** 0.189*** Pakistan 0.085*** 0.194*** 0.188*** 0.163*** generate through greater apparel exports. For Sri Lanka 1.011*** 0.939*** 0.696*** 0.892*** the U.S. market, a 10 percent increase in Chinese apparel prices would increase apparel Source: World Bank calculations using household and labor force surveys of various years. Note: Bangladesh—the last column is 2010; India—the first column is 1994, second is 2001, last is employment in Pakistan for males by 2010; Pakistan—the first column is 1996, second is 2001; Sri Lanka—first column is 1996, third is 8.93 percent—by far the largest increase 2006. — = not available. among South Asian economies—followed by Bangladesh and India. The gains for POTENTIAL FOR INCREASED GROWTH THROUGH POLICY REFORMS 137
Sri Lanka are less than 1 percent, but it is Firm benefits: Greater density of important to keep in mind that these esti- successful firms mates are for exports to the United States only. For the European Union market, a 10 At the firm level, the region’s potential is percent increase in Chinese apparel prices shown by the achievements of its leading would increase apparel employment for males firms, which have risen to standards of global by 8.55 percent in Sri Lanka and 4.30 percent excellence. These firms have flourished in India, but Bangladesh and Pakistan would because they operate in countries and sectors experience small decreases because their (for example, apparel in Bangladesh and Sri products are not close substitutes for Chinese Lanka) or subsectors (for example, auto parts apparel in the European Union. All of the in India) in which the policy environment is results are qualitatively similar for females conducive to success or because they were (table 7.3). able to internalize some of the constraints in
TABLE 7.3 Increased apparel price competitiveness vis-à-vis China can create many jobs in South Asia Panel a: Male employment responses for exports to United States
Elasticity of jobs to prices Elasticity of exports Elasticity of jobs ⎛ %ΔEmployment ⎞ ⎝⎜ ⎠⎟ Country to prices (εxp ) to exports (εEx ) %ΔPrices Bangladesh 1.358* 0.311*** 0.422 India 1.462* 0.176*** 0.332 Pakistan 2.531* 0.353*** 0.893 Sri Lanka 0.024 0.380*** 0.009 Panel b: Female employment responses for exports to United States
%ΔEmployment %ΔPrices Country εxp εEx Bangladesh 1.358* 0.323*** 0.439 India 1.462* 0.172*** 0.251 Pakistan 2.531* 0.336*** 0.850 Sri Lanka 0.024 0.350*** 0.008 Panel c: Male employment responses for exports to the EU
%ΔEmployment %ΔPrices Country εxp εEx Bangladesh −0.238 0.311*** −0.074 India 1.895* 0.176*** 0.430 Pakistan −0.060 0.353*** −0.021 Sri Lanka 2.249* 0.380*** 0.855 Panel d: Female employment responses for exports to the EU
%ΔEmployment %ΔPrices Country εxp εEx Bangladesh −0.238 0.323*** −0.077 India 1.895* 0.172*** 0.326 Pakistan −0.060 0.336*** −0.020 Sri Lanka 2.249* 0.350*** 0.787
Source: Lopez-Acevedo and Robertson 2016. Note: The elasticities reported here are for a 1 percent increase in prices of Chinese apparel. The ratios denoted in bold highlight high values of the elasticity of jobs to prices. ***p < .01, **p < .05, *p < .1. 138 SOUTH ASIA’S TURN
their external environment (for example, efficiency out of 30 comparable Samsung through vertical integration in agribusiness plants around the world, Dilmah and KRBL and apparel in both Pakistan and India). are recognized as premium tea and rice These examples show the potential of more brands globally, and MAS has developed conducive and supportive policies to increase a range of high-performance sportswear the number of successful firms. based on its innovative synthetic fabric. Some Leading firms play a critical development of these firms are extending their global reach role by providing major sources of productive by acquiring leading firms abroad. Bharat employment, exports, and innovation; help- Forge—a company that has broken into ing improve the performance of suppliers by design, engineering, R&D, testing, calibra- providing them with access to high-value tion, and other higher-value-added services markets, technology, skills, and financing; and integrated these with their existing manu- increasing competitive pressure on other facturing product lines—has acquired auto- firms; and, through example, providing a motive companies in Germany. source of inspiration for local businesses and In order to acquire these capabilities, South sending a strong message to potential interna- Asia’s leading firms pursued international tional investors. This section summarizes the integration, reaped the productivity gains experience of more than 100 leading firms in generated by locating close to suppliers and South Asia interviewed for the industry case clients, invested in skills and improved man- studies—apparel, automotive, electronics, agement practices, and benefitted from public and agribusiness. investment in trade logistics and innovation. The case studies reveal that many of South For example: Asia’s top firms are indigenous. These include most of the leading firms in apparel (such as UÊ Global links. Many of South Asia’s leading US Apparel in Pakistan, Orient Craft in India, firms actively sought to connect with global Pacific Jeans in Bangladesh, and MAS in Sri leaders through links to suppliers. Examples Lanka) and a growing number of auto-parts include Bharat Forge and Motherson Sumi suppliers (such as Bharat Forge and Hi-Tech Systems Ltd. (MSSL) with Maruti-Suzuki, Gears Ltd. [HTGL] in India). Notable leading HTGL with Hero Honda (now Hero South Asian firms in agribusiness include MotoCorp), and MAS with Victoria’s Fauji Foundation (a food conglomerate in Secret. Over time, these companies Pakistan), Dilmah (producing high-value tea challenged themselves further through in Sri Lanka), and KRBL Ltd. (processing exposure to export markets and very com- basmati rice in India). Even in the electronics petitive domestic markets (in auto parts, for sector, which is relatively new to the region, example, following the reduction in import there are examples of emerging world-class tariffs and in electronics). South Asian firms, such as Dixon UÊ Agglomeration economies. Geographic Technologies and Micromax in India. Most proximity to customers appears to have of these firms started from modest beginnings aided efforts to upgrade products, process, but expanded substantially over time; Dilmah, and functions for these firms. Their close for example, started with 18 staff in 1974 and location enabled MSSL to hold frequent has grown to 35,000 employees. meetings with original equipment manu- The experience of these leading firms dem- facturers (OEMs) on existing products. At onstrates that world-class levels of opera- times during the course of these discus- tional excellence, efficiency, and innovation sions, a new need would reveal itself, lead- can be achieved with the right management, ing to subsequent meetings to identify scale and technology, and worker training. OEMs’ requirements. For example, the Samsung plant in Noida, UÊ Skills. To compete in global markets, these India (outside New Delhi) ranks second in firms made significant investments in POTENTIAL FOR INCREASED GROWTH THROUGH POLICY REFORMS 139
acquiring skilled manpower at all levels Annex 7A and meeting international standards. For example, workers at Tos Lanka Estimating changes in market share: undergo training in Japan for a period the global computable general ranging from three months to one year. equilibrium model The chairman of Bharat Forge said, “We The forward-looking analysis in this report is have leveraged our tie-ups with leading based on a global recursive dynamic comput- academic institutions to create a strong able general equilibrium (CGE) model. The talent pipeline. Our efforts have resulted CGE approach has several key advantages in creation of an over 7,000-strong global that can be used to complement other forms pool of skilled engineers and technicians.” of analysis. First, it is deeply structural with According to a senior executive from multiple sectors and globally integrated mar- HTGL, “We train workers and lose them kets that determine bilateral flows of goods to OEMs. But we still train because the and services. Second, it is both country- and ones who stay are crucial for our produc- globally consistent. Third, its richness allows tivity. Unskilled workers are cheaper but productivity growth to be introduced in sev- costs match up when their mistakes are eral ways. financially accounted for.” The model is calibrated in the base year UÊ Innovation. When MSSL successfully (2011) to Version 9 of the Global Trade acquired a new technology and delivered Analysis Project (GTAP) database—the latest the product to the customer, management available version. The database has been con- would ask its engineers, “How can we figured for 15 regions—including all of those leverage this technology for adjacent available for South Asia: Bangladesh, India, products? What more could we do with Nepal, Pakistan, Sri Lanka, and rest of South it? What would that take?” In several Asia aggregate region—and 32 economic instances MSSL upgraded products or activities: 12 in agriculture and food, 4 in fos- began to produce the inputs for their sil fuels and other mining, 10 manufacturing existing products. It expanded from pro- sectors (including wearing apparel, electronic ducing basic plastic components to build- equipment, and motor vehicles, which are the ing tooling and injection-molding subject of in-depth case studies carried out for machines to deliver a range of complex this report), and 6 service sectors plastic products. As a result, MSSL was The model is constructed as a time able to deepen its relationships with and sequence of comparative static equilibria with increase sales to existing customers, enter dynamic equations linking the periods. In new product categories, and expand its each static equilibrium, the model is a rela- participation in global value chains tively standard CGE model. Production is (GVCs). modeled as a nested series of constant elastic- UÊ Public investment. Public investment in ity of substitution (CES) functions, with a dif- trade logistics and innovation capacity has ferent nesting for crops, livestock, and other also been important to these firms’ suc- sectors. Energy is treated as a special input—a cesses. Pacific Jeans in Bangladesh said complement to capital in the short run, but a that the system of bonded warehouses and substitute in the long run. One additional fea- back-to-back letters of credit provided by ture of the production structure is its vintage the government in the 1970s supported structure. Substitutability is assumed to be the industry by providing access to critical lower with installed capital, and there is a imported inputs. In the case of KRBL, the degree of capital mobility friction with Indian government played a critical role in installed capital. Thus a negative shock to a the development of the Pusa-1121 variety sector leads to only a partial adjustment to of basmati rice. the capital stock in that sector. 140 SOUTH ASIA’S TURN
Household demand is based on the Dynamics is composed of three broad ele- constant-differences-in-elasticity (CDE) utility ments. The aggregate supply of labor is function that is nonhomothetic.7 The govern- assumed to grow at the rate of growth of the ment sector purchases goods and services, and working-age population—defined as the pop- collects taxes on sales, imports, exports, fac- ulation aged between 15 and 64. The growth tors, and household income. The government of skilled labor is aligned with the growth deficit is assumed to be held constant—and rate of higher education (those with a second- the direct tax on households adjusts to meet ary degree or higher) as projected by the the fiscal target. Investment is savings driven International Institute for Applied Systems and is equal to the sum of household, public, Analysis (IIASA) for the shared socioeco- and foreign savings. The baseline assumes a nomic pathways, with unskilled labor mea- targeted path for investment as a share of sured as the difference between aggregate and GDP; given that government and foreign sav- skilled labor. Capital is equated to the previ- ings are assumed to be fixed at base year lev- ous period’s capital stock, less depreciation, els, household savings are allowed to adjust to plus the previous period’s investment—the meet the investment target. standard motion equation for capital. Trade is modeled using the ubiquitous Productivity is labor-augmenting: it is uni- Armington assumption (Armington 1969), form between skilled and unskilled labor, but whereby goods are differentiated by region the model allows it to be differentiated across of origin. In the current version of the base- sectors. In the current baseline, it is assumed line, preferences between domestic and for- to be uniform across all sectors and is cali- eign consumption are held constant, and brated to target a given growth path for real thus trade shares, in the absence of changes income per capita. in relative prices, move in rough proportion In the baseline scenario, real GDP in South to GDP (with adjustments largely due to Asia rises by 6 percent per year (tripling by compositional effects). With a fixed capital 2030), supported by rapid population growth, account balance, the real exchange rate an increasing share of skilled workers, and the adjusts to ex ante movements in the trade relatively optimistic productivity growth balance. Bilateral trade is identified with assumptions embedded in the OECD GDP four different prices and three wedges. The projections for South Asia. The sources of first wedge is between the producer price growth are projected to change somewhat and the exporter’s border (or FOB) price— from recent experience. The share of GDP this is an export tax (or subsidy). The second growth accounted for by increases in labor and wedge reflects the costs of international trade capital falls from 70 percent in 2011–15 to a and transport and converts the FOB price to little over 60 percent in 2026–30 (figure 7A.1). the CIF price at the border of the importer. Increases in labor productivity make a greater The importer adds to this a tariff, the third contribution to GDP growth in the later and final wedge. period, averaging a contribution of 2 percent- Factor markets are assumed to clear. The age points to total GDP growth through 2030 model also allows for a segmented labor mar- (figure 7A.1). Still, as in the past, increases in ket between agriculture and nonagriculture the number of workers and in the volume of using a Harris-Todaro specification (Harris capital (particularly the latter) represent the and Todaro 1970). In the case of a segmented most important source of growth. market, migration reflects changes in the Because the baseline scenario holds the relative return to labor across the two labor policy environment constant, three alternative markets. Capital markets clear at the national scenarios model improvements in trade facili- level. However, installed capital is only tation and the functioning of domestic mar- partially mobile in sectors with deficient kets in order to quantify the possible effects of growth. Finally, land is only partially mobile policy interventions aimed at enhancing across agricultural activities. productivity: POTENTIAL FOR INCREASED GROWTH THROUGH POLICY REFORMS 141
UÊ The first scenario lowers the “iceberg” FIGURE 7A.1 South Asia’s sources of growth, baseline, 2011–30 cost of trade. Iceberg costs refer to logistic 100 difficulties, for example weak port infra- structure, burdensome customs regula- 90 tions, or inefficient behind-the-border services (such as warehousing and trans- 80 portation) that raise the costs of trade. 70 In the model, iceberg costs are measured by a lower quantity of imports arriving at 60 a destination than what was exported at origin. In this simulation, the iceberg 50 ercent parameter is modified to produce a P 40 10 percent increase in the quantity of exports arriving at destination (compared 30 to the baseline), but only for bilateral trade between South Asian countries and 20 the rest of the world. The magnitude of this change is consistent with the potential 10 for improving logistics in the region, given 0 its poor performance on the domestic 2011–15 2015–20 2021–25 2026–30 component of the LPI. Productivity land Productivity capital Productivity skilled UÊ The second scenario assumes a more rapid Productivity unskilled Volume land Volume natural resources implementation of ongoing improvements Volume capital Volume skilled Volume unskilled in the port-to-port trade and transporta- Source: tion costs in the region, as measured by a World Bank calculations. decline in the trading margin between the FOB and CIF prices. In the baseline sce- improved functioning of domestic markets nario, the margin is assumed to fall by 1 through better product market regulation percent per year across all trade relation- and reduced distortions in the movement ships; in this simulation the margin of goods within countries. declines by 2 percent per year for all trade relationships involving the South Asian countries—for both their exports and their Estimating the elasticity of substitution imports. This would be consistent, for for exports example, with South Asia taking advan- tage of improved international transporta- The empirical approach employed here is sim- tion networks (for example, a denser ilar to Feenstra (1994), adapted to be used network), enhanced competition, and with panel data rather than his cross-sectional some technological upgrading of ports to approach. The methodology therefore com- handle larger, more sophisticated vessels— bines elements from a standard gravity model, consistent with improvements in the inter- direct estimation of elasticities, and Feenstra’s national component of the region’s LPI. model. As in any typical demand equation, UÊ In the third scenario, a similar assumption the quantity of apparel that buyers want to is used to lower the domestic cost of trade. purchase from each country depends on the
We assume that the cost of moving goods price Pi that country offers, the prices Pj that from the farm or factory gate to local mar- other countries offer, and the total income Y kets or the border for export is reduced of the buyer. For tractability, the analysis from an average of 10 percent of the total focuses on two major destinations (the United value of goods to as little as 5 percent.8 States and the EU-15, which are the two top This scenario would be consistent with import markets for apparel with 63 percent 142 SOUTH ASIA’S TURN
of global imports in 2012), two market lead- apparel in response to an increase in apparel ers in apparel exports (China and Latin exports—would attract more women into the America), and two emerging competitors of labor force. Here, female labor supply is a South Asia (Cambodia and Vietnam). The function of the expected wage as well as mar- system is estimated with three equations: U.S. ital status, education, household size, educa- or EU-15 imports from focus countries tion of household head, number of children (Bangladesh, Cambodia, India, Pakistan, Sri between birth and age 5 and between the Lanka, and Vietnam), U.S. imports from ages of 6 and 18, and the rural-urban loca- China, and U.S. imports from Latin America. tion dummy. Although Klasen and Pieters (2012) use India data to estimate the female labor supply, this is the first study to do this Estimating labor demand exercise for the region. This approach modifies the classic labor demand model (Hamermesh 1993) by con- Notes trolling for structural differences in labor pro- 1. The IAMC community has developed five ductivity related to the size of firms, distinct scenarios (referred to as social macro-global changes over time, and impos- economic pathways) with different story- ing symmetrical cross-wage elasticities.9 lines, such as equitable and environmen- Labor is heterogeneous across males and tally sustainable growth or a fragmented females, all factor prices including wages and world with poor global governance, low rental rates are exogenous to the firm, and growth, and persistent high poverty levels. All projections are available at the website output serves as a proxy for exports because of the International Institute for Applied most apparel firms in the region are export- Systems Analysis, http://www.iiasa.ac.at/. oriented. The model is similar in spirit to The Middle of the Road scenario used here Grossman (1986), who proposes that inter- is based on Organisation for Economic sectoral labor mobility is responsible for how Co-operation and Development (OECD) import competition affects jobs.10 Along the projections for gross domestic product same lines, Revenga (1997) studies the impact (GDP) and the medium variant of the 2010 of trade liberalization on wages and employ- revision of the United Nations population ment in Mexico’s manufacturing sector, and projections. The growth projections have Currie and Harrison (1997) conduct a similar been modified to gap fill for the missing study for Morocco. countries, re-base to a different base year (holding the growth rates constant), and Estimates of labor demand are based on a annualize the projections that were made two-step procedure. First, standard Mincer- available at five-year intervals. type equations are used to establish whether 2. By design, there is no change in the aggregate working in apparel carries a wage premium level of employment across scenarios. More over agriculture (a labor-intensive, low- plausibly, the changes induced by each sce- skilled alternative), especially for women. nario would lead to changes in labor force Individual wages are estimated as a function participation rates, in the aggregate as real of age, education, and a set of industry and wages increase, and perhaps across gender occupation dummies, controlling for self- lines if sectors such as wearing apparel, with selection as in Heckman (1978). Second, a higher concentration of female employ- building on Becker (1965, 1973, 1974), ment, expand. 3. The European Union countries used in the whose seminal work developed a framework analysis are Austria, Belgium, Denmark, for the analysis and the classic labor supply Finland, France, Germany, Greece, Ireland, model (Hausman 1980; Blundell and Italy, Luxembourg, the Netherlands, Portugal, MaCurdy 1999), we explore whether Spain, Sweden, and the United Kingdom. expected higher wages—which could be 4. Sri Lanka would experience an increase in induced by a greater availability of jobs in exports of less than 1 percent in this scenario. POTENTIAL FOR INCREASED GROWTH THROUGH POLICY REFORMS 143
5. Results for India—the only other country for 9. Cross-wage elasticity is modeled as elasticity which the data distinguish between permanent of male (female) employment with respect to and temporary employees—suggest that the change in female (male) wage. elasticities for the two types of labor are about 10. Seddon and Wacziarg (2001) and Levinsohn the same. (1999) provide further reading on the inter- 6. There are a number of potential explanations sectoral reallocation of labor. for the decline, most involving a U-shaped relationship between female labor force participation and economic development References (Goldin 1995; Verik 2014). For example, Aguirre, D., L. Hoteit, C. Rupp, and K. Sabbagh. female participation rates may be high in the 2012. Empowering the Third Billion: Women poorest countries, where many women are and the World of Work in 2012. New York: engaged in subsistence activities, and lower Booz & Co. in countries with somewhat higher incomes Armington, Paul. 1969. “A Theory of Demand for because of the rising importance of industrial Products Distinguished by Place of Production.” jobs. At some point in the development pro- Staff Paper 16, International Monetary Fund, cess, however, higher female education levels, Washington DC. lower fertility, and a larger share of services Becker, G. S. 1965. “A Theory of the Allocation of in output, which opens up opportunities for Time.” Economic Journal 75 (299): 493–517. women, result in higher female participation ———. 1973. “A Theory of Marriage: Part I.” rates. There is also evidence in India that Journal of Political Economy 81 (4): 813–46. labor market outcomes depend in part on ———. 1974. “A Theory of Social Interactions.” differences in the level of urbanization, with NBER Working Paper 74, National Bureau of relatively few employment opportunities for Economic Research, New York. women in growing areas that are more Blundell, R., and T. MaCurdy. 1999. “Labor urbanized than villages but less urbanized Supply: A Review of Alternative Approaches.” than large cities (Chatterjee, Murgai, and In Handbook of Labor Economics vol. 3, edited Rama 2015). Other factors might be lim- by O. Ashenfelter and D. Card, 1559–695. ited availability of transportation to work, Amsterdam: Elsevier. bad working conditions, and a lack of Chatterjee, U., R. Murgai, and M. Rama. 2015. institutions for early childhood education. “Job Opportunities along the Rural-Urban 7. The CDE utility function allows for much Gradation and Female Labor Force Participation greater richness in cross-price substitutabil- in India.” Policy Research Working Paper, 7412, ity than the ubiquitous linear expenditure World Bank, Washington, DC. system (LES). But nonetheless, both systems Cuberes, D., and M. Teignier. 2012. “Gender Gaps suffer from relatively poor dynamic behav- in the Labor Market and Aggregate ior. In the case of the CDE, income elastici- Productivity.” SEPR Working Paper 2012017, ties are relatively stable relative to income University of Sheffield, Department of growth. Economics, Sheffield, U.K. 8. The current version of the GTAP database Currie, J., and A. Harrison. 1997. “Trade Reform incorporates the domestic trade margins in and Labor Market Adjustment in Morocco.” the input-output table. The decline in the Journal of Labor Economics 15 (3): S44–71. domestic trade margin is achieved by reduc- Diaz-Mayans, M., and R. Sanchez. 2004. ing the input-output coefficient of selected “Temporary Employment and Technical service sectors (if there are not enough ser- Efficiency in Spain.” International Journal of vices to justify a domestic margin of 10 Manpower 25 (2): 181–94. percent, service requirements are halved). Do, Q. T., A. A. Levchenko, and C. E. Raddatz. Note that this scenario has no direct impact 2014. “Comparative Advantage, International on the cost of imports because it does not Trade, and Fertility.” Policy Research Working directly affect the end-user price of imports. Paper 6930, World Bank, Washington, DC. This is somewhat contrary to reality, as one Feenstra, R. C. 1994. “New Product Varieties would anticipate that improvements in and the Measurement of International domestic margins would also lead to a reduc- Prices.” American Economic Review 84 (1): tion in the end-user price of imports. 157–77. 144 SOUTH ASIA’S TURN
Goldin, C. 1995. “The U-Shaped Female Labor Kabeer, N., R. Assaad, A. Darkwah, S. Mahmud, Force Function in Economic Development and H. Sholkamy, S. Tasneem, D. Tsikata, and Economic History.” In Investment in Women’s M. Sulaiman. 2013. Paid Work, Women’s Human Capital and Economic Development, Empowerment and Inclusive Growth: edited by T. P. Schultz, 61–90. Chicago: Transforming the Structures of Constraint. University of Chicago Press. New York: UN Entity for Gender Equality and Grossman, G. M. 1986. “Imports as a Cause of the Empowerment of Women. Injury: The Case of the U.S. Steel Industry.” Klasen, S., and J. Pieters. 2012. “Push or Pull? Journal of International Economics 20 (3-4): Drivers of Female Labor Force Participation 201–23. during India’s Economic Boom.” IZA Paper Hamermesh, D. 1993. Labor Demand. Princeton: 6395, Institute for the Study of Labor, Bonn, Princeton University Press. Germany. Harris, J., and M. Todaro. 1970. “Migration, Levinsohn, J. 1999. “Employment Responses to Unemployment, and Development: A Two- International Liberalization in Chile.” Journal Sector Analysis.” American Economic Review of International Economics 47: 321–44. 60 (1): 126–42. Lopez-Acevedo, G., and R. Robertson. 2016. Hausman, J. A. 1980. “The Effect of Wages, Taxes, Stitches to Riches? Apparel Employment, and Fixed Costs on Women’s Labor Force Trade, and Economic Development in South Participation.” Journal of Public Economics Asia. Washington, DC: World Bank. 14 (2): 161–94. Revenga, A. 1997. “Employment and Wage Effects Heath, R., and A. M. Mobarak. 2012. “Does of Trade Liberalization: The Case of Mexican Demand or Supply Constrain Investments in Manufacturing.” Journal of Labor Economics Education? Evidence from Garment Sector Jobs 15 (53): S20–43. in Bangladesh.” Report 81292, World Bank, Seddon, J., and R. Wacziarg. 2001. “Trade Washington DC. Liberalization and Intersectoral Labor Heckman, J. J. 1978. “A Partial Survey of Movements.” Stanford GSB Working Paper Recent Research on the Labor Supply of 1652, Stanford Graduate School of Business, Women.” American Economic Review 68 (2): Stanford, CA. 200–07. Sivasankaran, Anitha. 2014. “Work and Women’s ITA-OTEXA (U.S. International Trade Marriage, Fertility and Empowerment: Administration, Office of Textiles and Apparel). Evidence from Textile Mill Employment in 2014. Market Reports/Tariffs: Textiles, India.” Harvard University Job Market Paper, Apparel, Footwear and Travel Goods. Harvard University, Cambridge, MA. Washington, DC: ITA-OTEXA. Verick, Sher. 2014. “Female Labor Force Jensen, R. 2012. “Do Labor Market Opportunities Participation in Developing Countries.” IZA Affect Young Women’s Work and Family World of Labor, September 2014. doi: Decisions? Experimental Evidence from India.” 10.15185/izawol.87. Quarterly Journal of Economics 27 (2): World Bank. 2014. World Development Indicators. 753–92. Washington, DC: World Bank. Need for Greater Emphasis on Trade Policies, Spatial Policies, 8 and Firm Capabilities
This report has focused on four main policy With respect to trade and trade-related pol- levers to boost South Asia’s competitiveness. icies, South Asia does not compare favorably The first is policies to improve the business with the Association of Southeast Asian environment. This lever is relatively well Nations (ASEAN) and Southern African known and multiple efforts are underway Customs Union (SACU) countries. In 2010, within the region to address some of the trade costs within South Asia were almost major shortcomings—although, as argued by double those in East Asia, and trade costs the agribusiness industry case study at the end between the two regions were almost as high of chapter 3, more emphasis could be given to as those within South Asia.1 Part 1 of this industry-specific business environment issues report described how tariffs, paratariffs,2 non- (product market regulations). The policy rec- tariff measures (NTMs), and logistics ineffi- ommendations address the remaining three ciencies boost trade costs. In GVCs—which by critical, and underused, drivers of competi- definition require that parts and components tiveness in the region: (1) better connection to move back and forth across international global value chains (GVCs); (2) maximizing borders—the effect of a marginal increase in agglomeration benefits; and (3) support for trade costs is much larger than in “regular” innovation and productivity. We discuss each trade flows. For example, most-favored nation policy below. (MFN) tariffs on intermediate apparel goods average from 15 percent to 21 percent in 3 Policies to better connect to Bangladesh, Maldives, and Pakistan. Duties global value chains exceed 30 percent on auto parts in Pakistan and MFN tariffs are high on final autos Trade-related issues have been found to be the (from 23 percent to 100 percent) in all regional most important constraints on competitive- countries except Sri Lanka. The electronics ness and productivity in the four industry case sector is treated most favorably among the studies—for example, the difficulties export- GVC sectors, with single-digit tariffs in most ers face in importing inputs in a timely man- cases. NTMs in the region are also pervasive, ner at world-market prices (apparel and including in GVC sectors: motorcycles and electronics), poor trade logistics (electronics), vehicles in Pakistan face the largest number of and high protection rates (automotive assem- NTMs. Although both tariffs and NTMs mat- bly and agribusiness). ter for GVC participation (Ferrantino 2012a;
145 146 SOUTH ASIA’S TURN
Ferrantino 2012b), Kee and Nicita (2013) to firms’ ability to sell to external markets and finds that tariffs are the main policy-induced source inputs efficiently. In India, the average obstacle to regional trade. time reported to clear customs varied from Hence, gradually reducing import tariffs, 2 to 10 days for large firms and 14 to 21 days paratariffs, and NTMs toward a common low for small and medium enterprises (SMEs). baseline to increase exposure and access to Firms are often forced to hold higher-than- global good practices is a policy priority that necessary inventories to compensate for could reap substantial productivity gains. lengthy and unpredictable delays in customs, Tariffs should be gradually reduced when high and nevertheless may be forced to delay tariffs shield industries from international shipments, both of which can severely erode good practices. High tariffs have been found competitiveness. and discussed in the automotive assembly However, external challenges are often industry as well as in the agribusiness industry dwarfed by internal ones. Indian firms (7 to 30 percent applied, 16 to 190 percent reported that, although it takes 11 days for a bound tariffs). Reducing tariffs on final goods container to travel from Shanghai to Mumbai, would improve incentives for innovation and it takes 20 days to travel from Mumbai to shift labor and capital from low-productivity Delhi. Firms in auto components, textiles, firms that cannot survive in a more competi- electronics, and heavy engineering report tive environment to high-productivity firms. maintaining 27 percent higher inventories Tariffs on intermediate goods that are higher than necessary to cope with these internal than those on final goods should also be obstacles. Poor infrastructure is one reason reduced as this protection structure discour- for these delays, but a survey shows that a ages domestic manufacturing.4 The auto-parts quarter of the journey is spent at checkposts, industry in India shows that a gradual reduc- state borders, city entrances, and other regu- tion in tariffs can lead to increased growth latory stoppages. In India, differences in tax and competitiveness (figure 8.1). regimes between states are important reasons There is also substantial room for improve- for the need for, and time consumed by, inter- ment with regard to trade facilitation. The nal clearances. poor efficiency of customs and other clearance An improvement in all trade facilitation procedures for traded goods, as well as inade- measures could raise South Asian exports by quate logistics services, are major impediments 40.3 percent—the largest increase among
FIGURE 8.1 Tariff reduction and faster growth of auto parts production in India went hand in hand
60 12
50 10
40 8
30 6 ercent P
20 4 ollars (billions) D 10 2
0 0
1996–97 1997–98 1998–99 2000–01 2001–02 2002–03 2003–04 2004–05 2005–06 1999–2000 Nominal tariff (%) Domestic production
Source: World Integrated Trade Solutions database, http://wits.worldbank.org/; ACMA and McKinsey and Co. 2012; Narayan and Vashisht 2008. Note: Domestic production is stated in current prices. NEED FOR GREATER EMPHASIS ON TRADE POLICIES, SPATIAL POLICIES, AND FIRM CAPABILITIES 147
global regions, followed by Europe and agencies; and establishing an effective and Central Asia at 30 percent (Wilson, Mann, quick grievance redress mechanism (the cur- and Otsuki 2005). Within South Asia, India rent administrative mechanisms are lengthy, would have the highest increase in dollar and firms fear reprisals). Improvements in port terms ($10.4 billion), and Bangladesh the efficiency and service-sector infrastructure will highest percentage increase (68.3 percent). also be critical, in particular the capacity of These high export gains in the region would ports to handle larger, more-sophisticated ves- result more from improvements in port sels. When asked why he was not investing in efficiency and service sector infrastructure Bihar, where most of his labor comes from, a than from improvements in the regulatory leading apparel manufacturer from Rajasthan environment and the customs environment. answered, “Fix the Calcutta port.” Regional differences in trade gains (in per- Improvements in product market regula- centage terms) are much smaller for imports, tion are also critical. For example, in apparel, but South Asia remains one of the biggest South Asia’s GVC success story, the input beneficiaries. mix is not fully consistent with global Trade facilitation could be improved by demand patterns. Although 32 percent of reforming the duty and tax remission for apparel globally is made from synthetic export (DTRE) schemes to facilitate access to materials, the share of synthetics in South imported inputs for exporters. These schemes Asian exports ranges from 5 percent in are supposed to enable exporters to import key Sri Lanka to 18 percent in India (figure 8.2). inputs free of duties and taxes, but in practice Synthetic fibers are increasingly in demand they seldom work. Exporters are limited to for high-performance garments, such as exporting products made with locally sourced sports uniforms and protective gear, and inputs, which greatly constrains their capacity require a greater degree of technological to expand or improve quality. The apparel sophistication than products made from case study is the best illustration of the impor- traditional fibers. Anecdotal evidence sug- tance of such schemes, as shown by the supe- gests that either high tariffs or NTMs may be rior performance of the Bangladesh apparel reducing the supply of synthetics in South industry (the only industry that enjoys exten- Asia. Such protection, to the extent it exists, sive access to bonded warehouses and accounts could be driven by the interests of the cotton for almost 90 percent of Bangladesh’s exports) industry or of the small domestic synthetics and the Sri Lanka apparel industry (which industry in India that competes with imports. does not require a DTRE scheme because it For example, in 2011–12 India produced 6.1 has no tariffs on textiles) compared to the million tons of raw cotton fiber, compared to apparel industries in India and Pakistan, where just 1.2 million tons of synthetic staple fibers such schemes are plagued with red tape.5 It is (Saheed 2012).6 thus no surprise that India’s and Pakistan’s Firms also cite problems in domestic prod- apparel export associations have put liberaliz- uct markets, including controls on prices, ing the import regimes for inputs at the top of inappropriate product standards, other their list of policy recommendations. constraints on markets, and administrative A longer-term solution to trade facilitation requirements for the transport of goods, as issues must include strengthening the soft and important constraints on production. hard infrastructure for domestic and external Agribusiness firms state that restrictions on trade. Important steps to speed clearance at markets limit their operations. Outdated reg- the border include providing for fully ulatory barriers hinder the development of electronic submission of documents (the single storage and processing infrastructure. In par- electronic window); a risk-based inspection ticular, violations of stock limits and price system for imported containers (reducing the caps carry penalties that can include jail sen- need to physically inspect all of them); improv- tences of up to seven years, which severely ing coordination of border management limits private sector interest in these markets. 148 SOUTH ASIA’S TURN
FIGURE 8.2 South Asia’s apparel input mix is not fully consistent with global demand patterns
100
90
80
70
60
50 ercent P 40
30
20
10
0 World (average) India Bangladesh Pakistan Sri Lanka Silk Wool Other materials Other textiles MMF Cotton
Source: World Bank 2016. Note: MMF = manmade fiber.
Market committees impose strict controls on emission norms in India, force firms to the marketing of agricultural produce. change their technology, imposing heavy Produce can be traded only through the mar- losses. The announcement in 2016 by the ket; for some crops (such as sugar cane) direct Indian government to adopt Euro 6 norms purchases are allowed subject to a fee. As a by 2020 (leaping over Euro 5 norms) is a consequence, there is no competition from welcome development. private markets, services are poor, and the set- Firms also call for greater representation ting of fees is opaque. Price caps combined of the private sector in the formulation and with minimum support prices on commodity implementation of standards. In the agribusi- products discourage investment in higher- ness industry, processors and traders con- quality products, and subsidies on fertilizer sulted for this report believed that food safety and water tend to benefit larger farmers and regulations often are rigid, not in accord with sustain low productivity and environmentally scientific advancements, and not in line with damaging practices. the World Trade Organization (WTO) Product market standards that are unnec- Agreement on Sanitary and Phyto-Sanitary essarily restrictive, do not reflect the latest measures. Overlapping responsibilities technology, or are seriously out of line with among government bodies responsible for international standards particularly limit food safety, coupled with a lack of productivity in sophisticated industries. For coordination, impair transparency and the example, automotive firms in India are not ability of firms to comply with regulations required to use the latest international tech- in Bangladesh, India, and Pakistan. nical standards required by the European Enforcement of regulations is reportedly Union and the United States, reducing their inefficient or lacking. Food safety laborato- incentives and capacity to compete in global ries are not recognized by international markets. Moreover, frequent changes in bodies and lack the capacity for certain tests, regulations, such as those pertaining to such as for pesticides, mycotoxins, and NEED FOR GREATER EMPHASIS ON TRADE POLICIES, SPATIAL POLICIES, AND FIRM CAPABILITIES 149
antibiotic residues. As a result, the system items on their shelves in very short cycles fails to effectively protect consumers and (Abernathy, Volpe, and Weil 2006). Local, impedes firms’ access to foreign markets. or at least regional, access to and avail- The development of capabilities for GVC ability of fabric inputs is also closely participation could help raise South Asian related. However, fabric production needs participation in GVCs beyond apparel. to be competitive in price, quality, lead Simulations carried out in part 3 of this report time, and variety. In this context, the show that, while South Asia is projected to ability to import inputs duty-free is more than triple its share in global exports of advantageous. electronics and motor vehicles by 2030, coun- UÊ Full-package services. Full-package capa- tries other than India will remain small play- bility refers to the ability of a firm to offer ers in these industries. Even India in 2030 will accompanying services that increase the only just approach China’s current levels of value added of manufacturers. The buyer exports of autos and electronics. To achieve surveys show that the most important more rapid growth in GVC products, coun- services include input and material sourc- tries in South Asia will need to substantially ing and financing, and apparel product enhance GVC-relevant capabilities, for which development services. Buyers’ desire to they currently lag behind potential competi- reduce the complexity of their supply tors in ASEAN and SACU. Improvements in chains has spurred this shift from working logistics alone are unlikely to offset the with assembly suppliers (cut, make, and challenges to competitiveness faced by the trim assembly) to full-package suppliers. region; deeper institutional reform and human UÊ Social and, to a lesser extent, environmen- capital development will be equally crucial. tal compliance and political stability and A pragmatic approach to increasing predictability. Because buyers play a lim- participation in GVCs is the development of ited role in the actual production process, the specific characteristics that lead firms look country-specific factors are generally less for when selecting suppliers. Surveys of global important than firm-specific factors in the apparel buyers (Birnbaum 2014; Daher and supplier selection process. However, social Chmielewski 2013; Kurt Salmon Associates compliance has increased in importance in and Apparel Magazine 2007–13; Nathan response to pressure from corporate social Associates 2005) and interviews conducted for responsibility (CSR) campaigns by non- this report identify the following key factors: governmental organizations (NGOs) and compliance-conscious consumers. UÊ Cost and quality. These two firm-specific criteria ranked the highest in all buyer Policies to maximize agglomeration surveys reviewed over the last decade. benefits Local firms can enhance their ability to meet the cost, quality, and timeliness stan- Agglomeration benefits, or the benefits that dards of multinational corporations accrue to firms and workers from locating through investments in firm capabilities close together in cities or clusters, are impor- and innovation—whether of processes, tant for productivity. Although measures of products, organization or managerial concentration are high in South Asia, concen- capacity, and skills. tration has not increased substantially over UÊ Lead time and reliability, including access the past two decades, suggesting that more- to inputs. The increasing importance of productive locations have not generally been lead time is related to the shift toward successful in attracting additional resources lean retailing and just-in-time delivery, in away from less-productive locations. This which buyers reduce the inventory risks reflects significant barriers to the move- associated with supplying apparel to fast- ment of goods, labor, and capital across changing, volatile markets by replenishing internal borders in South Asian countries. 150 SOUTH ASIA’S TURN
Indeed, impediments to efficient allocation of importance of public intervention in improv- resources between districts are stronger ing access to land. Historically, public sup- than distortions within districts, creating sig- port has been provided in most countries nificant barriers to firms’ ability to reap the through industrial zone developments. These benefits of agglomeration. have had a mixed record of success—many The industry case studies confirm that, of the public zones were not in appropriate when it is allowed to happen, agglomeration locations or have been poorly managed (for is positively associated with firm perfor- example, the Punjab Small Industries mance in South Asia. For example, interviews Corporation zones in Pakistan), while not suggest that automotive firms gain substan- enough quality industrial land was provided tial benefits from being located in clusters in the most suitable areas. because of the importance of frequent techni- The lack of well-located and well-serviced cal interactions, and location next to other industrial land limits export-oriented foreign automotive firms is highly correlated with direct investment (FDI) in electronics and productivity levels. Another interesting case apparel; investors can choose instead to is the light manufacturing cluster in Sialkot, invest in Vietnam, for example, which has Pakistan, where agglomeration benefits more readily available industrial land in prime than compensate for a challenging invest- locations. A notorious example is Samsung’s ment climate 1,000 kilometers from the decision to withdraw a planned $1.25 billion Karachi port. The Sialkot cluster derives its investment in Bangladesh, which would have competitive advantage from the ability of employed 50,000 workers, because the com- firms to hire workers from a large pool of pany could not obtain 250 acres in an export skilled labor as well as from the ability to processing zone. Samsung is now a major offer a one-stop solution to global buyers— investor in Vietnam, where it contributed to the private enterprises in the cluster financed the launch of the electronics industry. the development of an international airport, Conversely and unfortunately, Indian states which provides direct connections to Dubai, have competed fiercely to attract major origi- and the development of new industrial zones nal equipment manufacturers (OEMs) with to accommodate their growth and help com- tax incentives and land deals, risking a “fiscal ply with ever more stringent social and envi- race to the bottom,” suboptimal investment ronmental norms. locations, and industry fragmentation. Restrictions on land markets in South Ironically, governments are not providing Asia discourage domestic and foreign invest- sufficiently good land for export-oriented ment and limit the benefits firms can gain FDI (which can be made elsewhere), while through agglomeration and clustering. Firms providing too much land and incentives to mention the difficulty in accessing industrial market-oriented FDI, which would have been land in Bangladesh as an important con- made in any case. straint on development. Inadequate space in Cooperation with the private sector can well-serviced clusters also impairs productiv- play an important role in improving the effi- ity in apparel SMEs that are stranded in ciency and availability of clusters for indus- congested city centers across South Asia. trial development. Industrial zones can enable Difficulties in the land market that take a SMEs to cluster around their main customers long time to resolve (such as a lack of secure (for example in automotive and electronics) land titles), the need to provide infrastruc- as well as to have access to common facilities ture and overcome coordination issues, as for research and development (R&D) and well as the need to overcome negative exter- testing, waste disposal, and recycling. The nalities (such as pollution) and foster positive Combined Effluent Treatment Plants in the externalities (such as attracting leading inves- upcoming leather and apparel parks in tors and generating agglomeration econo- Punjab, Pakistan, are an example. An impor- mies and cluster effects) underline the tant lesson for South Asia can be drawn from NEED FOR GREATER EMPHASIS ON TRADE POLICIES, SPATIAL POLICIES, AND FIRM CAPABILITIES 151
China’s cooperation with private firms to markets (including over property taxes) to develop “plug and play” industrial zones, elected local governments to provide them which provide SMEs with ready-to-use stan- with the authority, resources, and incen- dardized industrial buildings and provide tives they need to promote industrial devel- decent worker housing close to the factories. opment by facilitating private sector–led India has developed an interesting public- industrial zones. This is the path followed private partnership solution to address the by China, which started with five Special coordination and financing issues associated Economic Zones launched by the central with moving an urban SME cluster to government followed by thousands of an industrial estate outside the city. In the industrial zones launched by the private Scheme for Integrated Textile Parks, sector with the support of local govern- Infrastructure Leasing and Financial Services ments, which financed the infrastructure Ltd. (ILFS) (a company of mixed public and and facilitated access to land. private ownership) helps SME clusters set up In addition to land market reforms, special purpose vehicles, find appropriate improving the flexibility of markets for labor land, and secure the required financing. ILFS and capital is likely to facilitate further gains also provides managerial and technical train- from agglomeration. In particular, policies to ing to the members of the cluster. increase the flexibility of labor markets, espe- It is interesting to note that many of the cially for women, who face particularly high modern industrial clusters in South Asia discrimination in South Asia’s labor markets remain located within or near large urban (World Bank 2012), are likely to substan- centers—for example, Dhaka and Chittagong tially reduce misallocation of labor and for apparel in Bangladesh, Delhi/Noida and improve productivity. Additional flexibility Chennai for electronics in India, and Karachi could also improve labor mobility, which is and Lahore for automotive in Pakistan. This relatively low in the region. Chauvin, Glaeser, helps explain the result at the aggregate level and Tobio (2011) found that only 0.4 percent that agglomeration benefits in South Asia of the population in South Asia had moved come primarily from urbanization rather to a different state within the previous five than specialization effects; the industry case years, compared with 9 percent in the United studies show that most specialization hap- States. Labor market policies in the region pens in large cities. The case studies also remain an important constraint, especially as show, however, the emergence of specializa- per capita incomes rise (figure 8.3). Although tion within smaller or specialized cities—for hiring rules in the region are rather flexible, example, the light manufacturing cluster in dismissal procedures in South Asia are Sialkot, Pakistan, and the automotive clus- among the most onerous in the world ters in Pune and Aurangabad, India. This (World Bank 2011). may signal the next wave of economies of Minimizing the misallocation of labor and agglomeration, which should be driven by capital and maximizing the benefits of smaller or specialized cities, as has happened agglomeration economies, therefore, go hand in China and more developed regions as pri- in hand. Policies directed at improving urban mary cities become too congested and governance and bridging the region’s infra- expensive. structure gap will ensure that firms and work- To enable this natural and desirable ers will be matched more easily. Achieving evolution, and as discussed in Ellis and this will require tackling congestion issues Roberts (2016), South Asian governments head-on (box 8.1). In particular, investments should continue to invest in infrastructure in improved urban connectivity (going to better connect and equip secondary cit- beyond roads to invest in public transit),7 ies and pursue decentralization. One criti- provision of quality affordable housing cal aspect of this decentralization will be and other basic infrastructure services, and the delegation of authority over land reducing the negative social impact of 152 SOUTH ASIA’S TURN
FIGURE 8.3 Labor regulations are a more important constraint in South Asia than in other regions
4
3
2 everity of constrainteverity S 1
06 7 8 91011 Log of per capita GDP in purchasing power parity dollars Nepal Bangladesh Afghanistan Pakistan India Sri Lanka Maldives Bhutan
Source: World Bank 2011. Note: The cross-country regression lion shows the relationship between the reported severity of the constraint for a benchmark firm and the log of per capita GDP. The shaded area is the 95 percent confidence interval band around the regression line. Vertical bars show confidenceintervals of 95 percent around the reported severity of the constraint for countries in South Asia. The lack of overlap between the South Asian country confidence interval and the regression line confidence interval is a conservative test of the statistically significant difference between the reported severity of a constraint for the South Asian country and the average reported severity of constraint for countries at the same level of per capita GDP. The reported severity could still be signifi- cantly different even with overlap. Analysis is based on pooled sample of enterprise survey conducted between 2000 and 2010. The severity of constraint is rated by firms on a 5-point scale, with 0 being no obstacle, 1 being a minor obstacle, 2 being a moderate obstacle, 3 being a major obstacle, and 4 being a very severe obstacle.
BOX 8.1 Leveraging urbanization in South Asia
Ellis and Roberts (2016) urge policy action to clarifying intergovernmental fiscal legal frame- reduce the high costs of urban congestion in the works—amending existing laws, enforcing region. To address key congestion constraints, them, and, in some cases, enacting new and they call the attention of policy makers to three simpler laws. Significant effort will also be fundamental urban governance deficits: empow- required to establish and align incentives for erment, resources, and accountability. urban management, governance, and finance. UÊ Resources. Revenue mobilization and man- UÊ Empowerment. Most urban local govern- agement are difficult for most urban local ments in South Asia suffer from unclear insti- governments in South Asia. Revenue mobili- tutional roles, limited functional and revenue zation is constrained by established fees and assignments, and limited control over human tax rates, narrow tax bases, and weak admin- resources. Empowering urban local govern- istrative capacity to fully use existing revenue ments will require a dedicated commitment to opportunities. Budgetary transfers, while
(continues next page) NEED FOR GREATER EMPHASIS ON TRADE POLICIES, SPATIAL POLICIES, AND FIRM CAPABILITIES 153
BOX 8.1 Leveraging urbanization in South Asia (continued)
officially unconditional, often come with UÊ Land and housing. City and suburban gov- higher-level rules and “guidance” on use. ernments need to go beyond the upgrading of Improved design, implementation, and effec- slums and embrace measures to stimulate the tiveness of intergovernmental fiscal transfers supply of affordable housing and offer more are required to close the resource gap. options to both low- and middle-income UÊ Accountability. Many of the formal adminis- households. The supply of affordable housing trative accountability systems in the region can be increased over time through more per- are weak or little used. Even though audits missive land-use and development regula- are legally mandated, poorly performing local tions. Also needed is infrastructure investment governments continue to receive transfers to open land for residential development, without penalty. Bridging the accountability easy-to-use land titling and registration sys- deficit requires the development of better sys- tems, and greater access to construction and tems and practices and the building of capac- mortgage finance. In addition, government ity of both government (at all levels) and regulations need to be revised to foster the citizens, including nurturing the social con- provision of more affordable rental housing. tract between local governments and citizens UÊ Resilience to disasters and climate change and clarifying fiscal relations between local effects. Cities in South Asia are particularly and higher tiers of government. exposed to disaster shocks. The first step in developing a resilience strategy is to accu- Ellis and Roberts (2016) also raise three addi- rately identify and quantify the national, sub- tional, and interrelated, areas for policy action: national, and city risks, and build national UÊ Connectivity and planning. Decision makers geo- referenced hazard exposure databases. should focus on strengthening transport links With the help of urban planners, engineers, to improve connectivity between urban areas and academics, cities should revisit the design (for example, between large and secondary cit- and enforcement of building codes and land- ies and between secondary cities and towns), use plans to avoid further building in risk- adopt forward-looking planning approaches prone areas and to reinforce structures so they to guide expansion on city peripheries (where are resilient to various hazards. In addition, it is most rapid), revitalize city cores by invest- national disaster risk-financing frameworks ing in better-quality public urban spaces need to be developed on the basis of risk lay- to enhance pedestrian walkability and livabil- ering to match risks with appropriate financ- ity, and adopt granular spatial planning ing instruments. approaches that permit greater variation in land use and intensity of development. Source: Ellis and Roberts 2016.
agglomeration (such as crime),8 should be capabilities, technology adoption, and worker high on the policy makers’ agendas. skills, even though evidence shows that relax- ing these constraints can lead to substantial Policies to strengthen firm improvements in productivity. The findings of capabilities this report suggest different approaches to innovation policy for both countries that are Many firms in South Asia are at a disadvan- innovation leaders and countries that are lag- tage compared to competitors in other coun- gards. For leaders, the critical challenge is tries with respect to innovation, managerial how to generate novel and, if possible, radical 154 SOUTH ASIA’S TURN
innovations. Enhancement of complementary Van Reenen 2013) and may be even higher in factors—skills and finance—is warranted, but developing countries further away from the more important is breaking the pattern of technological frontier (Griffith, Redding, and inward innovation development by support- Van Reenen 2004). However, as shown by ing cooperation with other firms and institu- Goni and Maloney (2014), returns to formal tions. For laggards, policies need to R&D are likely to be extremely low in the concentrate on increasing the number of firms absence of complementary factors such as engaged in incremental innovation. education and the quality of the private Public support for innovation can take var- sector, including managerial capabilities. ious forms. Investment in R&D is an impor- Therefore, authorities should focus on tant determinant of innovation and enhancing the inputs that are complementary productivity. There are many examples of to R&D investments—technology, skills, and public R&D interventions catalyzing firm finance. Given the varying rates of technology growth in the region, especially in agribusi- adoption in the region, increasing the limited ness, where research needs to be localized and adoption of the Internet and the use of com- spread widely among a large number of farm- puters among private firms, and then turning ers (box 8.2). Yet, overall, public and private to increasing the use of information and com- investment in R&D in South Asia is low munication technology (ICT) to improve (figure 8.4) and has remained relatively management and performance is particularly unchanged over the past decade, while Latin important in Nepal and Bangladesh. In con- America and East Asia in particular have trast, the use of ICT is common in Indian increased their investment in R&D. The firms, so efforts should be devoted to increas- growing gap is particularly worrisome in light ing e-commerce, the use of the Internet for the of empirical evidence that shows that social commercialization of products. returns from R&D are at least twice as high Public support for improving firm capabil- as private returns (Bloom, Schankerman, and ities through technology extension, access to
BOX 8.2 Public support for the development of Pusa-1121 basmati rice in India
At the turn of the twenty-first century, managers test marketing. KRBL had already established a at KRBL Ltd. (KRBL) attended a demonstration network of farmers through their attempts at by the Indian Agricultural Research Institute contract production. The knowledge that KRBL (IARI) at which a new “evolved” variety of bas- would buy Pusa-1121 in the local wholesale mar- mati rice, numbered 1121, was presented. KRBL kets eliminated the marketing risk for farmers staff were shown its extraordinary cooking char- growing the new variety. The results of testing acteristics, which resulted in the longest cooked were overwhelmingly positive, both from grow- grain of any basmati type. Subsequently, KRBL ers, who recognized higher returns from higher acquired a 3.5-kilogram sample of variety 1121 yields on a shorter growing cycle with a lower from IARI, and in 2001 began growing it for water requirement, and from the consumers in eventual seed multiplication even before the line the Persian Gulf markets who found that a cup of had entered national trials. Three seasons later, milled rice gave 4.5 cups of boiled rice instead of when the variety was officially released as Pusa- the more typical 4 cups. From there, adoption of 1121, KRBL had 20,000 tons ready. Over the the new variety spread rapidly to cover 84 percent next three seasons a portion of the crop was of basmati plantings in Punjab and 68 percent in saved for multiplication and a portion milled for Haryana by 2013. NEED FOR GREATER EMPHASIS ON TRADE POLICIES, SPATIAL POLICIES, AND FIRM CAPABILITIES 155
FIGURE 8.4 South Asia invests relatively less in R&D
Sri Lanka
Nepal
Pakistan
India
South Asia
Europe and Central Asia
Latin America and the Caribbean All low- and middle-income economies
East Asia and Pacific
0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6
Total public and private expenditure on R&D (% GDP)
Source: World Bank calculations based on World Development Indicators for 2004–13.
consulting services, networking, and informa- learn from ongoing efforts and implement tion has a long history in high-income coun- their own pilot initiatives. tries, in which returns on investment in these Skills matter critically for technology services to SMEs have been estimated to be adoption in South Asia, and worker skills are as high as 10–30 times (Ezell and Atkinson an important complement to firm investment 2011). Although most of these programs have in technology, research, and management not been evaluated using randomized con- capabilities. The share of high school gradu- trolled trials, a recent rigorous evaluation in ates among firm employees is positively and Indian textiles found an 11 percent increase in significantly associated with ICT adoption in productivity over one year in response to the region as a whole, and in every country management consulting (Bloom et al. 2013). except Pakistan. Many of the lead firms inter- Recognizing the importance of these invest- viewed for the industry case studies cited low ments, governments in Latin America and skill levels as a major constraint on Africa are piloting interventions that provide productivity. For example, skills were viewed SMEs with access to individualized consulting as a key factor in the success of firms in the services, as well as the more novel approach automotive industry. The lack of adequate of group consulting services, which can be managerial skills was seen as a serious delivered at lower cost and leverage group- problem. For example, only 43 percent of learning dynamics (similar to agricultural nonproduction workers in the automotive extension services). Much of the original sector in India are formally trained, compared research and experimentation into the impor- to nearly 70 percent in China. tance of managerial capabilities for firm Governments can and should play a performance originates in East and South leading role in the development of techni- Asia, affording authorities opportunities to cal, managerial, and vocational skills. 156 SOUTH ASIA’S TURN
BOX 8.3 China’s approach to workforce skills
China has taken effective steps to deal with diploma, ensuring faster absorption into the the demand-side challenges associated with job market. training and providing skills to its industrial Similarly, to make sure that the faculty keeps workforce. Over the years, the Chinese gov- abreast of the latest industry practices, the Chinese ernment has invested extensively in vocational government has made it compulsory for voca- education. As a result, nearly 50 percent of tional trainers to spend at least a month every the secondary school–level students in China year in manufacturing companies. Additionally, have access to vocational education. The China has made it very easy for vocational students quality of training in Chinese vocational to move back into general academic programs by institutions is good, mainly because of exten- sufficiently covering general academic skills in sive industry participation, favorable govern- vocational curricula. Chinese firms take employee ment policies, and a flexible curriculum. training seriously, spending twice as much on train- The key stakeholders in the system work ing and development as their Indian counterparts. hand in hand. Chinese courses require that students undergo one year of training to get a Source: BCG 2013.
Establishing educational partnerships, as China, where vocational training benefits well as upgrading university and vocational from extensive industry participation curricula for procurement, supply chain, (box 8.3). On-the-job training (including and marketing competencies, including apprenticeships) is also a very effective e-marketing and e-commerce, will be impor- way to acquire skills (often superior to tant for creating a generation of business government-led programs or a worker’s managers who can successfully communi- own investment) although it has some bias cate with global firms. Firms should forge toward existing skill needs. Company-led more robust links with local and technical training programs by Samsung, LG, and universities. Training, however, cannot be Intel in Vietnam and by Daewoo in limited to preservice training from public Bangladesh have addressed important skills technical and vocational education and gaps. Large Pakistani apparel firms report training institutes, which have shown mixed that they conduct in-house training for most success in India and Pakistan compared to of their workers (Nabi and Hamid 2013). NEED FOR GREATER EMPHASIS ON TRADE POLICIES, SPATIAL POLICIES, AND FIRM CAPABILITIES 157
LESSONS FROM THE FOUR INDUSTRY CASE STUDIES
Drivers and constraints of their experience with respect to the drivers of competitiveness as seen by and constraints on competitiveness. the managers of leading firms This section summarizes the experiences of Technology adoption and the more than 80 leading firms in South Asia innovation interviewed for the four industry case studies in this report. Many of them are from South Leading firms are demonstrating that world- Asia and include major apparel exporters class and innovative products can be (such as US Apparel in Pakistan, Orient Craft developed in South Asia by South Asian in India, Pacific Jeans in Bangladesh, and firms—Dilmah and KRBL are recognized, MAS in Sri Lanka), leading auto-parts manu- respectively, as premium tea and rice brands facturers (such as Bharat Forge, Hi-Tech globally, and MAS has developed a range of Gear, Motherson Sumi Systems Ltd. [MSSL] high-performance sportswear using its inno- and Hi-Tech Gear Ltd. in India), notable agri- vative synthetic fabric. Some of these firms business firms (such as Fauji Foundation in are becoming global through the acquisition Pakistan, Dilmah Tea in Sri Lanka, and KRBL of other leading firms abroad—Bharat Forge Ltd. in India), and emerging world-class firms has acquired automotive companies in in the relatively new electronics sector (such Germany and now engages in design, engi- as Dixon Technologies and Micromax in neering, R&D, testing, and calibration. In India). Most of these firms started from very electronics, Dixon is leading a new generation modest beginnings—Dilmah, for example, of productive and innovative Indian firms in started in 1974 with 18 staff but now has home appliances, with two R&D centers 35,000 employees, while US Apparel started located in Delhi. with four sewing machines in the 1930s. In some cases, the government played an Foreign firms continue to play an important facilitating role. For example, important role in complex, capital- and Pacific Jeans in Bangladesh said that the sys- knowledge-intensive activities such as car tem of bonded warehouses and back-to-back assembly (Maruti-Suzuki in India and letters of credit provided by the government Hyundai in Pakistan), electronics (Samsung in the 1970s promoted the industry by pro- in India and Tos Lanka, a subsidiary of viding it access to critical imported inputs. In Toslec of Japan, in Sri Lanka), and agribusi- the case of KRBL, the Indian government ness (Hindustan Lever, Nestlé, and Pepsico, played a critical role in the development of which have transmitted leading-edge knowl- the Pusa-1121 basmati rice. edge to tens of thousands of farmers). Beyond their direct contributions, leading Agglomeration economies and firms have important positive effects through diffusion of knowledge through the knowledge and support they provide to clustering suppliers and the competitive pressure they put on all firms in the industry. Their exam- Agglomeration economies are the benefits ple and competition compel other firms to that arise when firms and people locate near improve and signal to the international one another, either in cities or industrial investor community what can be achieved in clusters. There are five sources of agglomer- their country. ation economies: access to and sharing These leading firms have demonstrated of inputs and services (increasing returns that world-class levels of efficiency and qual- to scale); labor market pooling (better ity could be achieved in South Asia across all matches between employers and employees); the studied industries. This section presents knowledge spillovers (exchanges of ideas); 158 SOUTH ASIA’S TURN
market effects (the concentration of demand established in 2010 to plant and process encourages agglomeration); and economies hazelnuts. The company imports hazel of consumption (because consumers enjoy tissue–cultured plantlets and seeds from a variety). related operation in China, which are distrib- The biggest benefits from agglomeration uted among farmers. After three years of economies were found in the electronics and operation, 2,000 hectares had been planted automotive industries, in which geographic and 5,000 farmers trained. In India, Nestlé proximity to the customer has supported has helped 190,000 farmers increase the efforts to upgrade product, process, and func- quality of their milk and access and develop tion. Quoting a senior executive from MSSL, formal dairy markets in urban areas. a lead automotive firm in India: “We make the decision to co-locate based on several Learning from the best and factors. Is the job big enough in size to justify improving continuously by co-location? Is the OEM reputed enough to linking to GVCs learn from? Is there potential to increase share of wallet? Is there potential to learn Some of the leading South Asian entrepreneurs, something new completely?” In fact, the such as the founder of Dilmah, acquired their Indian and Pakistan automotive industries knowledge by studying and working abroad. are concentrated in major clusters in Karachi, The Desh-Daewoo joint venture, which Lahore, Chennai, and Pune. It is already pos- included the intense technical and managerial sible to see agglomeration effects in South training of 130 Bangladeshis in Daewoo’s Asia’s nascent electronics industry in clusters Pusan plant in 1979, established the founda- in Bangalore, Chennai, and Delhi, India. But tion for the next generation of Bangladeshi these effects can also be seen in the rapidly entrepreneurs. Similarly, many of the leading growing electronics industry in Vietnam. Indian auto-part companies acquired their Leading firms (such as the Mahr Group knowledge as suppliers to foreign companies, and Wol Plus Incorp.) and their suppliers in for example, Bharat Forge and MSSL from the leather apparel cluster in Sialkot benefit Maruti-Suzuki and Hi-Tech Gear from Hero from close proximity, which facilitates labor MotoCorp. (formerly Hero-Honda). The same pooling and knowledge diffusion and pro- is true of MAS through its close partnership vides international buyers a critical mass of with its main customer, Victoria’s Secret. offerings to encourage travel to this remote These companies developed their capabili- location in Pakistan. Buyers’ access was fur- ties over time by participating in export mar- ther facilitated by the development of an kets (such as apparel) or very competitive international airport and exhibition center domestic markets (such as auto parts and elec- privately financed by the cluster. tronics following the reduction in import Leading firms’ links with local suppliers tariffs). Quoting a senior executive at Hi-Tech also provide a positive impact on firms in the Gear: “From an operational perspective, agribusiness industry. For the Pusa-1121 exports challenge companies to design, develop, variety of basmati rice, KRBL transferred manufacture and supply products to discerning crucial market information to farmers by customers in global markets. This in turn moti- ensuring that they produced the “right” vates companies to scale up the value chain, product for the overseas markets. In I wanted to find the most discerning customers, Bangladesh, Aftab Bahumuki Farms Ltd. whether in India or abroad. I would bend over introduced contract farming with poultry backwards to work with them because I found farmers; contracted farmers recorded a sig- I learnt the most when I worked with OEMs nificantly higher level of output (at 11,783 who held very high standards.” kilograms per year) than noncontract farmers Developing trusted relationships with lead- (at 6,763 kilograms per year) (Begum 2008). ing international customers also provided a In Bhutan, Mountain Hazelnut Ventures was platform for further expansion. MSSL, for NEED FOR GREATER EMPHASIS ON TRADE POLICIES, SPATIAL POLICIES, AND FIRM CAPABILITIES 159
example, expanded from making basic plastic one constraint, echoing the apparel export components to building tooling and injection- associations in these two countries. Instead molding machines. they focused on cotton-based textiles and To compete in global markets, these firms integrated vertically to ensure quality made significant investments in acquiring textiles—a costly solution not available to skilled manpower at all levels and meeting SMEs in the sector. Conversely, Pacific international standards. Workers at Tos Jeans (Bangladesh) and MAS (Sri Lanka) Lanka undergo training in Japan for a period reported that the ability to import fabric ranging from three months to one year. Just duty-free was critical to their success. as many of these leading firms acquired their ° Poor trade logistics and inverted tariffs: capabilities as suppliers to leading OEMs, In electronics, Samsung mentioned the they in turn are having a major positive inverted tariff structure in India, and impact on their own suppliers. For example, very high effective protection rates were MSSL’s chief technology officer visits suppli- identified in the automotive and agri- ers and works closely with them to define and business sectors. Trade-related issues guide product specifications. also include major barriers to regional trade as well as barriers to internal trade The main constraints reported within India, which affects all industries. by leading firms UÊ Industry-specific product market regula- tions are important constraints in the Industry-specific policies, in the form of prod- automotive industry (through standards) uct market regulations, are the main con- and agribusiness (through standards, sub- straints found across the four industry case sidies, and restrictive regulations on prices studies. These policies include restrictions on and markets). trade, prices, products (through standards), UÊ The lack of managerial and technical skills and markets that have protected firms from is a constraint in automotive, electronics, exposure to global good practices (automo- and agribusiness. When asked to identify tive and agribusiness) or have limited firms’ the main challenge to growth, the chair- capacity to adopt these practices (apparel and man of Bharat Forge said, “Talent,” echo- electronics). These constraints are summa- ing other leading auto-part manufacturers. rized as follows and in table 8.1: UÊ Difficulties in accessing well-located and UÊ Trade-related issues were the most impor- well-serviced industrial land is a very seri- tant constraints mentioned by the leading ous issue for FDI (especially in electronics firms: and apparel in Bangladesh) and for clus- ° Difficulties apparel exporters face in ters of SMEs in apparel stranded in city importing inputs at world market prices: centers in all countries. Samsung could not Orient Craft (India) and US Apparel invest in Bangladesh because it could not (Pakistan) identified the difficulty of find 250 acres for itself and its suppliers importing synthetic fiber as their number around Chittagong.
TABLE 8.1 Major constraints to competitiveness identified in the industry case studies
Apparel Electronics Automotive Agribusiness Trade barriers Very important Very important Very important Important Product standards and Less important Less important Important Very important market restrictions Lack of technical and Less important Important Very important Important managerial skills Difficulties in accessing Important Important Important Less important well-located industrial land 160 SOUTH ASIA’S TURN
Notes congestion only in the short run, which means that expansion of roads is unlikely to 1. World Bank calculations using data from the relieve congestion in the long run. Public U.N. Economic and Social Commission for transport improvements appear to be the Asia and the Pacific–World Bank International most powerful tool for alleviating the incon- Trade Costs database. South Asia is repre- venience of commuting in urban areas. sented by India, Pakistan, and Sri Lanka. 8. Using Brazilian city-level data, Lage de Sousa East Asia is represented by China, Japan, (2014) showed that migration is negatively Malaysia, Thailand, and Vietnam. affected not only by local crime rates but also 2. For information on paratariffs, see Sattar by those in neighboring areas. (2014). 3. In some cases, these may be offset by duty References drawbacks. 4. In Nepal, an inverted tariff structure was Abernathy, F., A. Volpe, and D. Weil. 2006. “The found in yarn. Bangladesh, Maldives, and Future of the Apparel and Textile Industries: Pakistan impose 15 to 21 percent tariffs on Prospects and Choices for Public and Private intermediate apparel goods. Tariffs of more Actors.” Environment and Planning 38 (12): than 30 percent were imposed on auto parts 2207–32. in Pakistan. In India, a 7.5 percent tariff was ACMA (Automotive Component Manufacturers imposed on materials for medical equipment Association of India) and McKinsey & Co. while final goods faced a tariff of 5 percent. 2012. Auto Component Industry—Ready for In Pakistan, finished poultry products are the Transition. New Delhi: ACMA. imported at zero duty from Malaysia and at BCG (Boston Consulting Group). 2013. “People 16 percent duty from China, yet duties on the Productivity. Key to Indian Manufacturing inputs for local poultry processors are 15 to Competitiveness.” Boston Consulting Group. 30 percent, in addition to the goods and ser- http://www.bcgindia.com/documents vices tax of 17 percent. In the Indian elec- /file130179.pdf. tronics sector, producers face high tariffs on Begum, I. A. 2008. “Prospects and Potentialities of materials designated as dual-use under the Vertically Integrated Contract Farming in WTO Information Technology Agreement, Bangladesh.” Department of Agricultural and the process for obtaining exemption Development Economics, Hokkaido University, from duty is cumbersome. And in the Indian Japan. automotive sector, final goods are subjected Birnbaum, D. 2014. “Comment: Bangladesh’s to no tariffs under bilateral trade agreements Garment Trend Lines Look Pretty Poor.” Just- (such as that with Thailand) while intermedi- style.com, June 8. http://www.just-style.com ate inputs still face tariffs. /comment/bangladeshs-garment-trend-lines 5. Difficulties with duty drawback schemes -look-pretty-poor_id121926.aspx. were also found to affect the electronics Bloom, N., B. Eifert, A. Mahajan, D. McKenzie, and industry in India (the extremely cumbersome J. Roberts. 2013. “Does Management Matter? procedures around notification 25/99 dis- Evidence from India.” Quarterly Journal of courage firms from using it) and the auto- Economics 128 (1): 1–51. parts industry in Pakistan. Bloom, N., M. Schankerman, and J. Van Reenen. 6. Globally, apparel manufactured from syn- 2013. “Identifying Technology Spillovers and thetic fibers grew at a rate of 6.7 percent over Product Market Rivalry.” Econometrica 81: the 2005–12 period, and its share of the 1347–93. world apparel trade increased from 26 percent Chauvin, J. P., E. Glaeser, and K. Tobio. 2011. to 32 percent, while cotton’s share of the “Urban Economics in the U.S. and India.” global market decreased from 51 percent to Paper presented at the Economic Geography 46 percent over the same time period. United Conference, Seoul, Republic of Korea, June 29. Nations Statistics Division (database), United Daher, M., and J. Chmielewski. 2013. Private Nations, New York (accessed June 20, 2014), Label Sourcing Strategies to Differentiate and http://wits.worldbank.org. Defend: Insights from the 2012–2013 Private 7. Duranton and Turner (2011) find that in U.S. Label Sourcing Survey. New York: Deloitte cities, improved road provision eases traffic Consulting LLP. NEED FOR GREATER EMPHASIS ON TRADE POLICIES, SPATIAL POLICIES, AND FIRM CAPABILITIES 161
Duranton, G., and M. A. Turner. 2011. “The Survey: Annual Apparel Research Study & Fundamental Law of Road Congestion: Analysis. Evidence from U.S. Cities.” American Economic Lage de Sousa, F. 2014. “Does Crime Affect Review 101 (6): 2616–52. Migration Flows?” Papers in Regional Science Ellis, P., and M. Roberts. 2016. Leveraging 93 (S1): S99–S111. Urbanization in South Asia: Managing Spatial Nabi, I., and N. Hamid. 2013. Garments as a Transformation for Prosperity and Livability. Driver of Economic Growth: Insights from World Bank: Washington, DC. Pakistan Case Studies. Lahore, Pakistan: Ezell, S. and R. D. Atkinson. 2011. International International Growth Centre. Benchmarking of Countries’ Policies and Narayan, B. G., and P. Vashisht. 2008. Programs Supporting SME Manufacturers. “Determinants of Competitiveness of the Indian The Information Technology & Innovation Auto Industry.” Working Paper 201, Indian Foundation. Council for Research on International Ferrantino, M. 2012a. “Using Supply Chain Economic Relations, New Delhi. Analysis to Examine the Costs of Non-Tariff Nathan Associates. 2005. “Survey of U.S. Apparel Measures (NTMs) and the Benefits of Trade Buyers: Sourcing from Sub-Saharan Africa in Facilitation.” Staff Working Paper ERSD-2012- the Post-Quota Era.” University of Sussex. 02, World Trade Organization, Geneva. Pursell, G., and F. M. Ziaul Ahsan. 2011. “Sri ———. 2012b. “Supply Chains and Behind-the- Lanka’s Trade Policies: Back to Protectionism.” Border Trade Barriers: Implications for ASARC Working Paper 2011/03, Australian Developing Nations.” Centre for Economic National University, Canberra. Policy Research. http://voxeu.org/article /why Reis, J.G., and D. Taglioni. 2013. “Pakistan: -non-tariff-measures-matter-more-world-sliced Reinvigorating the Trade Agenda.” Policy -supply-chains. Paper Series on Pakistan PK 15/12, World Goni, E., and W. F. Maloney. 2014. “Why Don’t Bank, Washington, DC. Poor Countries Do R&D?” Policy Research Saheed, H. 2012. “Prospects for the Textile and Working Paper 6811, World Bank, Washington, Clothing Industry in India.” Textile Outlook DC. International 156: 86–127. Griffith, R., E. Huergo, J. Mairesse, and B. Peters. Sattar, Z. 2014. “Challenges of Export-Led 2006. “Innovation and Productivity Across Growth: Breaking Into New Markets And Four European Countries.” Oxford Review of Products.” Presentation for Bangladesh Economic Policy 22 (4): 483–98. Economists’ Forum, Dhaka, June 21–22. Griffith, R., S. Redding, and J. Van Reenen. 2004. Wilson, J., C. Mann, and T. Otsuki. 2005. “Mapping the Two Faces of R&D: Productivity “Assessing the Benefits of Trade Facilitation: Growth in a Panel of OECD Industries.” Review A Global Perspective.” World Economy 28 (6): of Economics and Statistics 86 (4):883–95. 841–71. Kee, H. L., and A. Nicita. 2016. “Trade Frauds, World Bank. 2011. More and Better Jobs in South Trade Elasticities, and Non-Tariff Measures.” Asia. Washington, DC: World Bank. Paper prepared for 5th IMF-World Bank-WTO ———. 2012. World Development Report: Jobs. Trade Research Workshop, Washington, DC, Washington, DC: World Bank. November 30. ———. 2016. Stitches to Riches?: Apparel Kurt Salmon Associates and Apparel Magazine. Employment, Trade, and Economic Development 2007–13. Excellence in Global Sourcing in South Asia. Washington, DC: World Bank. 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 chlorine–free (EECF) processes. More information about the Bank’s environmental philosophy can be found at http://www.worldbank.org/corporateresponsibility.