Leveraging Urbanization in South Asia

Leveraging Urbanization in South Asia Managing Spatial Transformation for Prosperity and Livability

Some rights reserved 1 2 3 4 18 17 16 15

This work is a product of the staff of The World Bank with external contributions. The fi ndings, interpretations, and conclusions expressed in this work do not necessarily refl ect 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 specifi cally 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: Ellis, Peter, and Mark Roberts. 2016. Leveraging Urban- ization in South Asia: Managing Spatial Transformation for Prosperity and Livability. South Asia Devel- opment Matters. Washington, DC: World Bank. doi: 10.1596/978-1-4648-0662-9. License: Creative Com- mons 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 ofﬁ cial 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, ﬁ gures, or images. All queries on rights and licenses should be addressed to the Publishing and Knowledge Division, The World Bank, 1818 H Street NW, Washington, DC 20433, USA; fax: 202-522-2625; e-mail: pubrights@ worldbank.org.

ISBN (paper): 978-1-4648-0662-9 ISBN (electronic): 978-1-4648-0663-6 DOI: 10.1596/978-1-4648-0662-9

Cover photo: © Barjor Mehta. Location is Kanpur, India. Used with the permission of Barjor Mehta. Further permission required for reuse. Cover design: Bill Pragluski, Critical Stages, LLC.

Library of Congress Cataloging-in-Publication Data has been requested. Contents

Foreword ...... xi

Acknowledgments ...... xiii

Abbreviations ...... xv

Overview ...... 1 Introduction ...... 1 Urban prosperity and livability below potential ...... 2 Urbanization—Messy and hidden ...... 3 Congestion constraining the benefits of agglomeration ...... 4 Addressing deficits in urban governance and finance ...... 5 Three additional areas for policy action ...... 7 Moving forward: Leveraging urbanization for greater prosperity and livability ...... 9 Notes...... 10 Introduction ...... 11 Summary ...... 11 The growing recognition of urbanization’s importance ...... 12 A framework for assessing urbanization and city performance ...... 12 Sources of agglomeration economies and congestion forces ...... 14 Notes...... 17 References ...... 18

Part 1 PATTERNS OF URBANIZATION AND STRUCTURAL TRANSFORMATION 1 Leveraging Urbanization for Prosperity and Livability ...... 23 Introduction ...... 23 South Asia’s recent urbanization in global and historical perspective ...... 24 Livability of South Asia’s cities ...... 31 Looking ahead—Leveraging urbanization for prosperity and livability ...... 37 Notes ...... 39 References ...... 40 v vi CONTENTS

2 Spatial Patterns of Subnational Performance and Urban Growth ...... 43 Introduction ...... 43 Measuring differences in subnational performance ...... 44 Variations in the level and pace of urbanization across South Asian countries ...... 51 Urban population growth and its drivers ...... 53 Population growth across urban hierarchies and geographic space ...... 59 Rapid relative expansion of urban footprints and the rise of the multicity agglomeration ...... 61 Spatial patterns of economic growth ...... 65 Growth of major agglomerations ...... 66 Spatial deconcentration of manufacturing ...... 68 Summary ...... 71 Notes...... 72 References ...... 74

Part 2 POLICIES FOR IMPROVED LEVERAGING 3 Revitalizing Urban Governance and Finance ...... 79 Introduction ...... 79 Basic government and intergovernmental structures and decentralization policy ...... 81 The empowerment deficit ...... 84 The resource deficit ...... 87 The accountability deficit ...... 93 Where empowerment, resources, and accountability meet: Implications for urban service delivery ...... 97 Bridging gaps in urban empowerment, resources, and accountability ...... 102 Notes...... 106 References ...... 106 4 Planning and Managing Spatial Structure and Connectivity ...... 109 Introduction ...... 110 Developing a system of well-connected and productive cities ...... 112 Managing urban expansion and rejuvenating city cores ...... 123 Urban governance in spatial and transportation planning in cities ...... 134 Notes...... 146 References ...... 146 5 Providing Affordable Land and Housing ...... 149 Introduction ...... 149 Impact of urbanization on housing ...... 150 The consequences of failing to provide affordable housing: At least one in four urban dwellers lives in slums ...... 152 Overcoming housing supply constraints: A two-pronged approach ...... 155 Key recommendations ...... 163 Notes ...... 165 References ...... 165 6 Sustaining Prosperity by Building Disaster-Resilient Cities ...... 169 Introduction ...... 169 Urban risks and exposure ...... 171 CONTENTS vii

Urban vulnerability to natural hazards ...... 172 Impacts of climate change and global warming ...... 173 Classification of cities based on risk and vulnerability profiles ...... 175 Key recommendations ...... 177 Notes...... 185 References ...... 186

Boxes 1 Types of localization economies ...... 15 2 Sources of congestion forces ...... 16 1.1 Investment in data for South Asian countries is urgently required ...... 25 1.2 Defining and comparing urban areas—The agglomeration index and night-lights ...... 26 1.3 Different trajectories—Urbanization, economic growth, and manufacturing exports in China and South Asia ...... 32 1.4 Is early deindustrialization inevitable? Not if the advantages of cities can be brought into play ...... 33 1.5 More than dust in Delhi ...... 35 2.1 Using nighttime lights data to infer patterns of economic activity and urban expansion ...... 45 2.2 Constructing the prosperity index ...... 47 2.3 India’s rapidly proliferating census towns ...... 53 2.4 Urbanization and health in South Asia—A missed opportunity? ...... 55 2.5 The rise of the South Asian megacity and the Malthusian trap ...... 58 4.1 Urbanization pressures in Kabul, Dhaka, and Thimphu ...... 111 4.2 Interconnecting cities in India and Sri Lanka ...... 116 4.3 Rejuvenating Puerto Madero, Buenos Aires ...... 119 4.4 Ahmedabad Bus Rapid Transit System: The Janmarg ...... 121 4.5 Equitable planning: Planning, when done right, benefits the poor in multiple ways ...... 125 4.6 Transforming Ahmedabad’s commercial center through granular planning ...... 129 4.7 Santiago’s repopulation and housing rehabilitation programs ...... 132 4.8 Land use in Kanpur and Amritsar from satellite imagery analysis and ground surveys ...... 135 4.9 Balancing trade-offs in urban development, housing land supply, traffic, tourism, cultural heritage, and environmental assets in Kandy ...... 137 4.10 Singapore’s urban planning framework and approach to granularity and implementation ...... 140 4.11 Land-value capture in Hong Kong SAR, China ...... 141 4.12 Land pooling and its application in Bhutan and Gujarat ...... 142 4.13 Inclusive urban planning and city revitalization in Medellín ...... 144 4.14 Preserving heritage and improving livelihoods in Lahore Walled City through social mobilization ...... 145 5.1 Estimating India’s slum population ...... 152 5.2 Impact of regulations on housing supply elasticity ...... 160 6.1 Investing in urban resilience: Colombo metropolitan area ...... 178 6.2 Integrating risk assessment in development planning: Aqaba, Jordan ...... 181 6.3 International experiences in risk financing ...... 184 viii CONTENTS

Figures 1 Report structure and framework for assessing urbanization and performance ...... 13 1.1 South Asia is among the least urbanized regions ...... 25 1.2 South Asia’s low agglomeration index is consistent with its level of development . . . . 27 1.3 Annual growth rate of South Asia’s urban share compared with other regions, 2000–11 ...... 28 1.4 Annual growth rate of South Asia’s urban population share compared with annual growth rates historically experienced by today’s developed countries ...... 29 1.5 Absolute increase in South Asia’s urban population compared with other regions, 2000–11 ...... 29 1.6 Global GDP share minus global urban population share for major world regions, 2011 ...... 30 1.7 Structural change and urbanization: Relation between agglomeration index and value added by nonagricultural activities, 2010 ...... 31 1.8 Annual growth rates of shares of GDP generated by nonagricultural and manufacturing activities, 2000–10 ...... 31 B1.3.1 Economic growth, urbanization, and manufacturing exports in South Asia versus China ...... 32 1.9 Prevalence of urban slums and poverty across South Asian countries ...... 34

B1.5.1 Relationship between annual mean concentration of PM2.5 and city population density for 381 developing-country cities ...... 36 1.10 Livability in four major South Asian cities and comparator cities ...... 38 1.11 Projected growth of urban population, 2011–30 ...... 38 B2.1.1 Distribution of nighttime light intensity between agricultural and urban areas in South Asia, 2009 ...... 46 B2.1.2 Relationship between GDP growth and growth of nighttime light intensity across low- and middle-income countries, 1999–2010 ...... 46 2.1 Performance distribution on the prosperity index across all districts ...... 48 2.2 Prosperity tends to be higher in more populous districts, but with significant variation in performance ...... 49 2.3 Relationship between district population and components of the prosperity index ...... 50 2.4 Very strong performing districts: A strong negative relationship between productivity and dynamism ...... 51 2.5 Prosperity score and dynamism by population quartile ...... 51 2.6 Share of the population classified as urban: Official definitions and the agglomeration index, 2010 ...... 52 B2.3.1 Number of Indian statutory and census towns, 1981–2011 ...... 53 2.7 Annual growth rate of urban share of population, by country, 2000–10 ...... 54 2.8 Annual growth rate of urban population, by country, 2000–10 ...... 54 B2.4.1 Under-five mortality rates across socioeconomic index quintiles ...... 56 B2.4.2 Percentage of births in health facilities across socioeconomic index quintiles ...... 56 2.9 Convergence of poverty rates across Sri Lankan districts, 2002–09 ...... 57 2.10 City population growth rate and initial population ...... 59 2.11 Patterns of population growth across districts, years including 2001–10 ...... 60 2.12 Urban nighttime light footprints ...... 62 2.13 Built-up urban area located outside official boundaries ...... 63 2.14 The birth of the Coimbatore agglomeration, India ...... 64 2.15 The Delhi-Lahore agglomeration ...... 65 CONTENTS ix

2.16 GDP growth and population growth across South Asian districts, 1999–2010 ...... 66 2.17 Patterns of annual nighttime lights growth around selected major cities, 1999–2010 ...... 67 2.18 Employment growth in Indian metropolitan cores and their peripheries, by sector, 1998–2005 ...... 68 2.19 Increased urbanization of unorganized manufacturing in India, 1989–2005 ...... 69 2.20 Employment in Dhaka ...... 70 3.1 Three key urban governance deficits in South Asia ...... 81 4.1 National, city, and institutional priority areas ...... 112 4.2 The Republic of Korea’s highway network and GRDP ...... 113 4.3 Road network characteristics ...... 114 4.4 Connectivity of South Asian cities in infrastructure networks: Airlines, information technology, roads, and rail ...... 115 B4.2.1 The national physical structure plan showing major planned road and rail network in Sri Lanka connecting to major urban centers by 2030 ...... 117 B4.3.1 Illustrative plan of Puerto Madero, 2014 ...... 119 B4.4.1 Bird’s-eye view of the BRTS station at Jhansi ki Rani junction, Ahmedabad, India ...... 121 B4.4.2 The Bus Rapid Transit System network ...... 122 4.5 Expansion of built-up area ...... 124 4.6 Karachi’s 2020 Development Plan ...... 126 4.7 Global urban population density and quality of living, 2012 ...... 127 4.8 Floor space index patterns in Mumbai ...... 128 B4.6.1 The Ahmedabad central business district ...... 130 B4.7.1 Cost of developing a new residential unit in city center versus periphery ...... 133 4.9 Redevelopment in major cities ...... 134 B4.8.1 Kanpur land use, 2011 ...... 136 B4.8.2 Amritsar land use, 2010–11 ...... 136 B4.9.1 Zoning plan for Kandy ...... 138 B4.9.2 Hazard zonation map for Kandy and surrounding areas ...... 138 B4.10.1 Singapore’s urban planning framework ...... 140 B4.12.1 Map of town planning schemes in Ahmedabad, 1915–76 ...... 143 B4.12.2 Number of town planning schemes in Ahmedabad ...... 144 5.1 Slum share exceeds poverty rates, 2005–11 ...... 153 5.2 Large and fragmented public holdings of land in Karachi ...... 158 6.1 Urban population growth by hazard risk, 2000 and 2050 ...... 171 6.2 Extent of existing urbanization in risk-prone regions, 2010 ...... 172 6.3 Estimated urban economic loss and mortality loss rates in South Asia ...... 173 6.4 Map of disaster typologies ...... 176 6.5 Financial instruments to address different layers of risk ...... 183

Tables 1.1 Global shares of urban population, GDP, and manufacturing value added ...... 30 1.2 Ranking of South Asian cities in the Economist Intelligence Unit’s livability index . . . . . 37 2.1 The prosperity index captures three dimensions of performance ...... 44 2.2 District performance on the prosperity index, by performance category ...... 48 2.3 Growth of multicity agglomerations, South Asia ...... 63 3.1 Levels or categories of subnational government and administration ...... 82 x CONTENTS

3.2 Decentralization policies and frameworks ...... 83 3.3 Local government functions, revenues, and expenditure shares ...... 85 3.4 Interjurisdictional cooperation and coordination mechanisms ...... 86 3.5 Intermediate tier and local government autonomy and discretion ...... 88 3.6 Own-source revenues, shared taxes, and local revenue significance ...... 89 3.7 Intergovernmental transfers ...... 90 3.8 Subnational borrowing frameworks ...... 92 3.9 Administrative accountability and managerial mechanisms ...... 94 3.10 Subnational elections and assemblies ...... 95 3.11 Transparency and nonelectoral accountability mechanisms ...... 96 3.12 Organization of services delivery in Bangladesh, India, Nepal, Pakistan, and Sri Lanka ...... 99 3.13 Functional responsibilities in selected global metropolitan areas ...... 102 4.1 World Economic Forum Global Competitiveness Index and infrastructure rankings ...... 114 5.1 Estimated slum population and number of slum households in South Asian cities, 2010 ...... 151 5.2 Proportion of urban population living in slums for South Asian countries ...... 154 5.3 Urban land requirement scenarios, 2010–50 ...... 156 5.4 Land registration and titling performance, by three measures, 2014 ...... 158 5.5 Urban infrastructure (new and replacement) investment requirements, 2010–50, South Asia ...... 160 6.1 Number of cities in South Asia affected by each of the four major hazards ...... 175 6.2 Characteristics of hazards and assessment data requirements ...... 182 Foreword

outh Asia’s urban population grew by least 14 percent of its urban population. 130 million people between 2001 and According to statistics for 2011, it is also S2011, and it is forecast to rise by almost home to the largest concentration of people in 250 million more in the next 15 years. The the world living on less than $1.25 per day region has already started to see the eco- (the World Bank’s global poverty line).1 It nomic growth and poverty-reduction bene- boasts six of the world’s mega-cities— fits associated with urbanization; if Bangalore, Delhi, Dhaka, Karachi, Kolkata, managed properly, further urbanization and Mumbai—with more on the way offers the potential for more prosperous and as populations grow in Ahmedabad, livable cities. To confirm this prospect, one Chennai, Hyderabad, and Lahore. At need look no further than the transforma- the same time, new settlements with urban- tive effect that urbanization has had, within like characteristics have been proliferating living memory, on many East Asian coun- since the beginning of the century, even tries and, looking further back, on the now- though many of them continue to be gov- advanced economies of Western Europe and erned as rural entities. North America. This report assesses South Asia’s progress Throughout South Asia, an important pol- in realizing the immense potential of its cities icy debate is under way on the role of cities for promoting prosperity and livability. Using and, more generally, the role of urbanization innovative nighttime lights data, as well as a in promoting economic development. Policy host of other data sets, it examines how the makers are thus concerned with how the region’s cities and subnational areas are per- region’s cities can be transformed to better forming and provides new evidence on the drive economic growth and poverty reduc- nature of urbanization in the region. In tion, as well as to become better places in particular, the report documents a process of which to live. The aim of this report is to contribute to and further stimulate the debate 1 Until the year 2015, extreme poverty was defined about the role of cities and urbanization in using the $1.25 poverty line, based on the 2005 promoting the region’s development. PPP exchange rates. Since the publication of the The opportunities are tremendous. Global Monitoring Report 2015/2016, the pov- South Asia is currently home to more than erty line is defined by a consumption threshold 23 percent of the world’s population and at at $1.90, using the 2011 PPP exchange rates.

xi xii FOREWORD

“messy” and “hidden” urbanization. Messy identifies urban governance and finance as urbanization is reflected in the widespread keys to the successful realization of the poten- existence of slums and sprawl, and hidden tial of South Asia’s cities. Urban local govern- urbanization is seen in the large share of the ments throughout the region need to be region’s population that lives in settlements properly empowered and adequately that possess urban characteristics but do not resourced. In addition, reforms need to be put satisfy the criteria required to be officially in place to improve urban local government classified as urban. Striking new evidence is accountability, both to upper tiers of govern- also provided on the emergence of an increas- ment and to the local populations they serve. ingly interconnected network of cities arising Such reforms are a precondition for meaning- from the growth of continuous belts of urban ful progress, but they will not, by themselves, development. These conurbations straddle suffice in the addressing of congestion con- subnational administrative boundaries and, in straints. In this context, the report identifies one case, even the border between India and three additional areas of critical policy action: Pakistan. improvements in connectivity and spatial The report describes the often severe planning, the efficiency of land and housing stresses brought about by growing urban markets, and the resilience of cities to the populations on infrastructure, basic city ser- effects of natural disasters and climate change. vices, land use, housing, and the environment. These three areas are intimately interrelated, It is the inability to adequately address these and the ability to progress on each will be stresses that provides the root cause of messy facilitated by improvements in urban gover- and hidden urbanization. Equally, it is these nance and finance. same congestion forces that are constraining As the debate about how best to cultivate the region’s ability to realize the vision of the benefits of urbanization continues, the prosperous and livable cities. Although the report offers the contours of a road map for existence of congestion forces is a fact of life the region’s policy makers—a road map for any city, the report argues that in the case toward a future of more prosperous and liv- of South Asia, these forces are exacerbated by able cities. failures not only of the market, but also of policy. Annette Dixon Finally, the report discusses how best to Vice President for South Asia ameliorate and manage those stresses. It The World Bank Acknowledgments

his report was prepared by a team led Maria Florencia Pinto, Prem Sangraula, Marc by Peter Ellis and Mark Roberts Schramm, Linda Shi, Forhad Shilpi, and Tunder the guidance of the South Asia Naijun Zhou. The National Geophysical Regional Chief Economist Martin Rama Data Center at the National Oceanic and and the South Asia Urban Practice Manager Atmospheric Administration (NOAA) also Ming Zhang. The work took place under provided support in the processing and inter- the South Asia Regional Vice President, pretation of the nighttime lights data used in Annette Dixon, and her predecessor, the report, and the team is particularly grate- Philippe Le Houerou. Core members of the ful to Christopher Elvidge and Kimberly team included Parul Agarwala, Rana Bough for the time that they generously Amirtahmasebi, David Dowall, Chyi-Yun donated in this area. A University of Huang, Jon Kher Kaw, Sangmoo Kim, Abdu Peradeniya team led by Jagath Gunatilake Muwonge, Fatima Shah, Paul Smoke, and assisted with ground-truthing fieldwork for Charles Undeland. the high-resolution satellite mapping of Additional important contributions were Kandy, Sri Lanka, reported on in chapter 4, made to the preparation of the report by while Shubham Mishra assisted with similar Robert Chase, Janna El-Horr, Marc Forni, work for Amritsar and Kanpur, India. Stephen Hammer, Ejaz Ghani, Charles The team gratefully acknowledges the Kunaka, Yue Li, Katie McWilliams, Patrick valuable comments provided by the peer Mullen, Urvashi Narain, Manjusha Rai, reviewers at the Project Concept Note (PCN), Shigeyuki Sakaki, Jessica Schmidt, and Quality Enhancement Review (QER), and Benjamin Stewart. Syed Usman Javaid, Eshrat Decision Review stages, which have contrib- Waris, and Mihir Prakash provided research uted greatly to both shaping and improving assistance at various points during the course the report. Abha Joshi-Ghani and Victor of the preparation of the report. Michelle Lisa Vergara acted as peer reviewers at the PCN Chen and Sumbo Adeyemo were the program stage, while Dean Cira and Ellen Hamilton assistants. were peer reviewers at the QER stage. The Work on the report was underpinned by a peer reviewers at the Decision Review stage series of background papers. Authors of these were Takuya Kamata, Mary Hallward- background papers who have not already Driemier, Jaafar Sadok Friaa, and Jonas been named include CIESIN (Columbia Frank. Uwe Deichmann was a peer reviewer University), the RAND Corporation, Jennifer at all three review stages, and Javier Sanchez- Day, Ben Derudder, Klaus Hubacek, Blane Reaza at both the QER and Decision Review Lewis, Xingjian Liu, O. P. Mathur, Bimal Patel, stages. In preparing the PCN, the team also xiii xiv ACKNOWLEDGMENTS

benefited from the feedback provided by an Kandy (Sri Lanka), and Thimphu (Bhutan), Advisory Panel consisting of Songsu Choi, organized as part of the Flagship’s Network of Sonia Hammam, Indermit Gill, and Sameh Practitioners, as well as from presentations Wahba. Vernon Henderson also provided use- made in New Delhi (India), Frascati (Italy), ful feedback on a draft of the PCN, while New York, and at various events at the World Kalpana Kochhar, who was the Chief Bank. Economist for South Asia when the work was Bruce Ross-Larson was the principal editor initially launched, provided important guid- and Joe Caponio the production coordinator, ance in the early stages of the work. while the World Bank’s Publishing and In preparing the report, the team has bene- Knowledge Unit, led by Mary Fisk, Patricia fited from discussions with senior manage- Katayama, and Nora Ridolfi, was in charge ment of the World Bank’s Social, Urban, Rural of the design, typesetting, printing, and dis- and Resilience Global Practice—notably semination of both the hard- and soft-copy Senior Director Ede Jorge Ijjasz-Vasquez, versions of the report. The team also worked Director Maninder Gill, and Director closely with the World Bank’s South Asia Marisela Munoz. The team is thankful for External Communications team, led by Alex the support provided by John Henry Stein Ferguson and Robert Doherty, in the launch (former Director, Sustainable Development and dissemination of the report. Network) and both current and former coun- Finally, the team acknowledges the gener- try office colleagues—notably, Barjor Mehta, ous financial support of Australian Aid and Deepali Tewari, Shahnaz Arshad, Augustin the Korean government, and the Norwegian Maria, Christopher Pablo, and Zhiyu Jerry contribution to the Multi-Donor Trust Fund Chen. Invaluable feedback on various parts of (MDTF) for Sustainable Urban Development, the report was also received at workshops in which helped to support much of the Daegu and Seoul (the Republic of Korea), analytical work. Abbreviations

AFG Afghanistan AI agglomeration index AUDA Ahmedabad Urban Redevelopment Authority BGD Bangladesh BRTS Bus Rapid Transit System BTN Bhutan CBD central business district CG central government DDC District Development Committee DMSP-OLS Defense Meteorological Satellite Program–Operational Linescan System DN digital number DRM disaster risk management EAP East Asia and Pacifi c ECA Europe and Central Asia EIU Economist Intelligence Unit FAR fl oor area ratio FSI fl oor space index GDP gross domestic product GQ Golden Quadrilateral GRDP gross regional domestic product IND India LAC Latin America and the Caribbean LG local government LKA Sri Lanka MDV The Maldives MENA Middle East and North Africa NAC North America NPL Nepal PAK Pakistan PFM public fi nancial management

PM2.5 particulate matter less than 2.5 micrometers in diameter xv xvi ABBREVIATIONS

PPP purchasing power parity; public-private partnership SAR South Asia; or Special Administrative Region (Hong Kong) SDC Santiago Development Corporation SSA Sub-Saharan Africa UNISDR United Nations International Strategy for Disaster Reduction VDC Village Development Committee WHO World Health Organization

Note Until the year 2015, extreme poverty was deﬁ ned using the $1.25 poverty line, based on the 2005 PPP exchange rates. Since the publication of the Global Monitoring Report 2015/2016, the poverty line is deﬁ ned by a consumption threshold at $1.90, using the 2011 PPP exchange rates. Overview

Introduction has been the inadequate provision of infrastructure and basic urban services. Two South Asia’s urban population is poised to others are insufficient housing and a failure grow by almost 250 million people by 2030. to deal with pollution. A key characteristic of If recent history is any guide, this increase urbanization is that agglomeration econo- could propel the region toward greater eco- mies, which are the unintended benefits that nomic growth and prosperity. Between 2000 firms and workers experience from one and 2011 the region’s urban population another as they cluster together,2 improve expanded by 130 million—more than the productivity and spur job creation, specifi- entire population of Japan. The growth ben- cally in manufacturing and services. In efits associated with urbanization also South Asia, this process is clearly visible in increased. South Asia made good strides in the structural transformation of the region’s achieving greater prosperity, with the increase economy, with manufacturing and services in productivity linked to the growing number now accounting for more than 80 percent of people living in the region’s towns and of GDP. cities. Average GDP per capita in the region In the long term, successful urbanization grew by almost 56 percent during 2000– is accompanied by the convergence of living 2012, from $2,560 to $4,000, for average standards between urban and rural areas as 1 annual growth of more than 3.8 percent. At economic and social benefits spill beyond the same time, absolute poverty declined urban boundaries. But these positive trends from one in two people living on less than can be undermined by the pressures of urban $1.25 a day in 1999 to less than one in three population growth on infrastructure, basic in 2010. Urbanization thus presents South services, land, housing, and the environment. Asian countries with an opportunity to trans- Estimates suggest that at least 130 million of form their economies and join the ranks of South Asia’s urban residents live in slums and richer countries in both prosperity and are disproportionately deprived of basic livability. infrastructure and access to basic services. South Asia can gain from urbanization by South Asia is not fully realizing the poten- fostering productivity through the agglomer- tial of its cities for prosperity and livability. ation of both people and enterprises in its One significant reason is that its urbanization towns and cities. One shortcoming, however,

1 2 LEVERAGING URBANIZATION IN SOUTH ASIA

has been messy and hidden. Messy urbaniza- 32 percent (a ratio of 0.91); South Asia pro- tion is reflected in the widespread existence duced 8 percent of global GDP with a share of slums and sprawl. Sprawl, in turn, helps of the global urban population of 14 percent give rise to hidden urbanization, particularly (a ratio of 0.57). This comparison suggests on the peripheries of major cities, which is that South Asia has been much less successful not captured by official statistics. Messy and than East Asia in leveraging its urbanization hidden urbanization is symptomatic of the for gains in productivity and prosperity. failure to adequately address congestion con- Indeed, South Asia looks more like Sub- straints that arise from the pressure of urban Saharan Africa, where the ratio of the population. For South Asian cities and for the region’s share of global GDP (3 percent) to region, these congestion constraints are its share of the global urban population undermining both livability and the agglom- (9 percent) was 0.34 in 2011. eration economies that can produce greater Although progress since 2000 has been prosperity. impressive, the majority of South Asia’s cities Policy makers and urban practitioners in remain characterized by high levels of pov- South Asia face common challenges for effec- erty, bad housing conditions, and generally tive urban management. The traditional res- poor livability for many of their inhabitants. ervations of South Asia’s politicians and According to the most recent estimates, policy makers about the benefits of urbaniza- extreme urban poverty has been largely tion have been replaced by more optimistic eradicated in both Sri Lanka and Bhutan. discourse on leveraging the benefits of urban- However, for the five most populous ization for growth and prosperity. The countries in the region—India, Pakistan, changing attitudes of governments are visible Bangladesh, Afghanistan, and Nepal, in that as national and local policy makers ask: order—the number of urban dwellers below What do cities need to do to meet the the national poverty line ranges from about demands of their growing populations and to one in eight in Pakistan to more than one in manage transformation? How can we create four in Afghanistan.3 an effective and functioning system of cities? In absolute numbers, at least 130 million National and local policy makers are starting South Asians—equivalent to more than the to recognize the need to address these chal- entire population of Mexico—live in infor- lenges in a timely and systematic manner if mal settlements characterized by poor con- they are to alleviate congestion pressures for struction, insecure tenure, and underserviced better performance of cities and, in so doing, housing plots. The lack of decent, affordable create an environment conducive to the stron- housing not only impairs the welfare of mil- ger leveraging of agglomeration economies. lions of South Asians, it also has potentially Improvements in urban governance and adverse implications for health outcomes and finance—in empowerment, in resources, and for female labor force participation. The in accountability systems—hold the keys to prevalence of urban slums in South Asian cit- both challenges. ies reflects a failure to adequately manage the congestion forces—in this case, in land and Urban prosperity and livability housing markets—associated with urban below potential population growth. Meanwhile, for the very poorest in Bangladesh, India, Nepal, and Despite strong growth since the beginning of Pakistan, under-five mortality is higher in the century, South Asia’s share of the global urban than in rural settings. South Asia’s cit- economy remains strikingly low relative to its ies are also notable for their polluted air. In share of the world’s urban population. In Delhi, recorded air pollution is almost three 2011, the East Asia and the Pacific region times higher than in Beijing, giving it the generated 29 percent of the world’s GDP with dubious distinction of being the “world’s a share of the global urban population of most polluted city.”4 OVERVIEW 3

The failure of South Asia’s urban areas to In a process of messy urbanization, how- adequately cope with the pressure of rising ever, a sizable proportion of the region’s populations is also reflected in the poor per- urban population lives in slums, and cities formance of its largest cities—those with the have been growing outward, spilling over most well-developed infrastructure networks, their administrative boundaries, rather than best access to basic urban services, and the upward through the construction of taller highest standards of living in the region—in buildings. And with growth occurring international rankings of cities for their beyond city limits, much urbanization has livability. One of the most respected rankings been hidden—a growing number of people in is the livability index published by the the region live in places that possess strong Economist Intelligence Unit (EIU), which urban characteristics but that are not offi- assesses cities on their performance across cially recognized as urban. five dimensions of a “livable city”—stability, health care, culture and the environment, Messy education, and infrastructure. According to the EIU’s 2015 rankings, the highest ranking Messy urbanization is refl ected in the esti- of the six South Asian cities in the index is mated 130 million people who live in slums New Delhi, which, out of 140 cities globally, typifi ed by poor quality housing in hazard- ranks 110, ahead of Mumbai (115), ous areas and a lack of access to basic Kathmandu (125), Colombo (127), and services. It is also refl ected in faster popula- Karachi (135). Dhaka, meanwhile, comes in tion growth on the peripheries of major cities at 139, better only than Damascus, Syria.5 in areas beyond municipal boundaries. For More generally, the average ranking of the the 12 largest Indian cities, satellite imagery six South Asian countries in the index falls shows that, for many of these, the propor- below the averages for both all non–South tion of built-up area outside a city’s offi cial Asian developing-country cities in the index boundaries exceeds that within its boundar- and all developing-country cities in the East ies. For all 12 cities, the proportion of built- Asia and Pacific region. up area outside city boundaries exceeds the proportion of population, implying that the Urbanization—Messy and hidden spillover is associated with relatively low- density sprawl. Urbanization in the region remains under- The spillover of cities across their bound- leveraged. The share of the region’s popula- aries creates challenges for metropolitan tion officially classified as living in urban coordination in the delivery of basic services settlements increased only marginally, from and the provision of infrastructure. And the 27.4 percent in 2000 to 30.9 percent in scale of the challenge has grown, evident in 2011, for annual growth of 1.1 percent a the rapid spread of urban footprints. Analysis year. By contrast, when it was at a level of based on night-lights data shows that the urbanization similar to that of South Asia region’s urban areas expanded at slightly today, China experienced growth in its more than 5 percent a year between 1999 urban share of population of 3.1 percent a and 2010. But the region’s urban population year, moving from 26.4 percent in 1990 to grew a little less than 2.5 percent a year. So 35.9 percent in 2000. Likewise, Brazil’s cities grew about twice as fast in area as they urban share grew at 2.5 percent a year grew in population, which suggests declining between 1950 and 1960, while moving from average city population densities and increas- 36.2 percent to 46.1 percent. Going back ing sprawl. Furthermore, as the footprints even further, for the United States, the urban of neighboring cities have expanded, the share rose from 25 percent to 35.9 percent number of multicity agglomerations— between 1880 and 1900, for growth of continuously lit belts of urbanization 1.8 percent a year. that contain two or more cities, each with 4 LEVERAGING URBANIZATION IN SOUTH ASIA

a population of at least 100,000—in the Thimbirigasyaya—saw their populations region has also grown, from 37 in 1999 to 45 decline, but the immediately surrounding in 2010. These urbanization belts present an localities had comparatively rapid popula- opportunity for greater agglomeration econo- tion growth. mies, but realizing these economies will again require better coordination between different Congestion constraining the urban local governments. benefi ts of agglomeration

Hidden South Asia’s urbanization has been messy and hidden in part because its towns and Hidden urbanization stems from official cities have been struggling to deal with the national statistics that understate the share of pressures of population on their infrastruc- South Asia’s population living in areas with ture, basic services, land, and housing, not urban characteristics. An alternative measure to mention the environment. These conges- of urbanization, the agglomeration index— tion pressures undermine the region’s exploi- which, unlike offi cial measures, is compara- tation of the full range of agglomeration ble across countries and regions—shows that economies and its ability to compete in offi cial statistics may substantially understate international export markets. In doing so, the number of South Asians living in areas they also constrain the ability of the urban- that look and feel urban, even if they are not ization process overall and of cities individu- counted as such in national population and ally to deliver improvements in both housing censuses. This undercounting is in prosperity and livability. This struggle to addition to the population in India’s census deal with congestion pressures is due not towns, which are towns that the country’s only to failures of the market but also to census classifi es as urban even though they failures of policy. continue to be governed as rural entities. The The strength of congestion forces can be reclassifi cation of rural settlements into cen- mitigated, to an important extent, if invest- sus towns was responsible for 30 percent of ments in infrastructure and basic services India’s urban population growth between keep pace with demand as more people and 2001 and 2011, reflecting a more general firms congregate in urban areas. Without process of in situ urbanization across much sufficient investment, urban infrastructure of the region.6 and services become stretched, reducing For major cities in India, population quality and access. The effects of congestion growth has been fastest on their peripheries forces also depend on the ability of land and in areas beyond their official administrative housing markets to respond to rising demand boundaries. This type of urban spread is for urban residential, industrial, and com- reflected in a large growth differential mercial property—and the ability to address between the districts in which the cities are the environmental concerns associated with located and some of the immediately neigh- pollution. boring districts. For example, the district of Not only does the interplay of agglomera- Delhi experienced population growth of tion economies with congestion forces deter- 1.9 percent a year between 2001 and 2011, mine the pace and character of urbanization, while population growth in Gautam Budh it also determines the prosperity and livabil- Nagar, just to the east, was 4.1 percent a ity outcomes delivered. This is true both for year. The picture is similar for major cities the urban system overall and for individual in other countries in the region. For exam- towns and cities. ple, several divisional secretariat (DS) A key insight is that successful cities inevi- divisions within the district of Colombo— tably are also congested cities, at least for Colombo, Dehiwala–Mount Lavinia, their transport infrastructure and their land Moratuwa, Sri Jayewardenepura Kotte, and and housing markets. Geneva; Hong Kong OVERVIEW 5

SAR, China; London; New York; Paris; and Empowerment Singapore are among the most prosperous Most urban local governments in South Asia and livable cities in the world. But they also suffer from unclear institutional roles and have property markets that rank among the limited functional and revenue assignments. most expensive globally. Similarly, the speed That leaves local governments with uncer- of traffic in downtown London today is little tain authority and limited power to make changed from the speed when horse-drawn decisions for most service delivery obligations. carriages rather than cars dominated its Despite generally having the authority to pre- streets. What separates world-class cities like pare their own budgets, local governments New York and London from South Asian have limited capacity and few incentives cities is not that they have conquered to do so. They depend greatly on transfers congestion—it is that they have much higher from upper tiers of government, and the prosperity and livability at comparable or reporting requirements for budget approvals lower levels of congestion. are heavy. The focus in this report is more on conges- Most urban local governments across the tion costs than on agglomeration economies. region have limited control over hiring, per- The forces that generate agglomeration formance appraisal, and firing. In Bhutan, economies—for example, spillovers of ideas Maldives, and Nepal, senior local govern- between firms and workers—provide rela- ment staff appointments tend to be made by tively difficult targets for policy to address. the national government. Exceptions include The forces of congestion, by contrast, are India and Pakistan, where some local govern- directly influenced by policy decisions ments have control over the hiring and firing regarding the supply of both infrastructure of lower cadre staff, but even their powers and basic services and the way cities are are subject to concurrence and clearance by planned. Congestion forces, moreover, act as the states or provinces. a direct constraint on the exploitation of Empowering urban local governments in agglomeration economies. Why? Because South Asia will require a dedicated commit- high congestion costs constrain both urban ment to clarifying intergovernmental fiscal growth and agglomeration by making cities legal frameworks by amending existing laws, less attractive places to migrate to and enforcing them, and in some cases, establish- encouraging cities to grow outward rather ing new and simple laws. Amending the legal than upward. framework will clarify the institutional roles and the revenue and expenditure obligations Addressing defi cits in urban of various tiers of government. It will also governance and fi nance increase the decision-making authority and fiscal autonomy of local governments. To address key congestion constraints, policy Significant effort will be required to estab- makers need to address three fundamental lish and align incentives for urban manage- urban governance defi cits—an empowerment ment, governance, and finance. Higher tiers defi cit, a resource defi cit, and an accountabil- of government should provide guidance and ity deficit. Addressing these deficits will oversight to ensure attention to national pol- require improving intergovernmental fi scal icy goals. The central government should relations to empower urban local govern- encourage and, where possible, facilitate ments. It will also require identifying practi- greater interjurisdictional cooperation and cal ways to increase the resources available to coordination among tiers of local govern- urban local governments to allow them to ment and agencies. Improving this enabling perform their mandated functions. And it environment will allow urban local govern- will require strengthening the mechanisms ments to better deliver on their compacts that hold local governments accountable for with the local populations they serve. their actions. 6 LEVERAGING URBANIZATION IN SOUTH ASIA

Resources region, the key challenge is to design, implement, and increase the effectiveness of inter- Revenue mobilization and management are governmental fiscal transfers. diffi cult for most urban local governments. Revenue mobilization is constrained by Accountability established fees and tax rates, as well as by narrow tax bases. In Afghanistan, Maldives, Several accountability mechanisms are in and Nepal, the central government sets all place for many urban local governments. local revenue rates. In Bangladesh and They range from fi nancial and performance Bhutan, local governments must follow audits to citizen report cards and social nationally defi ned guidelines or secure cen- audits. These mechanisms could be an impor- tral government approval in setting tax rates. tant source of information to be used by In India and Pakistan, local governments higher-level governments and citizens for have some formal discretion over setting monitoring local government performance, local tax rates but are generally subject to particularly governance and service delivery. strong state and provincial revenue regula- However, in practice, their effectiveness var- tions and oversight. ies markedly across the region. Local governments generally do not tap Formal administrative accountability sys- into all of their tax bases as prescribed tems generally exist in the region, but many by law. The reasons include weak adminis- are fairly weak or little used. The main causes trative capacity, service delivery break- for their infrequent implementation are the downs, outdated registers, rigid and fragmentation and lack of clarity in institu- ambiguous laws, limited private sector tional roles and the lack of interjurisdictional involvement, and complex governance set- cooperation. Nor are the links strong ups and political economies. Very few local between development plans, public invest- governments have robust revenue-enhancement programs, and annual budgets. ment plans and programs, while the ones All countries in the region have some type that do have very weak political and bureau- of audit mechanism led by a national audit cratic support for implementing them. institution, in many cases with a mandate Urban local governments, regardless of their for auditing both local and higher-tier size and economic significance, must governments. In Bangladesh, private auditors develop and exploit practical options for support the center in auditing local own-revenue mobilization, for more effec- governments. However, even though the tive use of transfers, and where appropriate, audits are legally mandated, poorly perform- for borrowing to finance infrastructure. ing local governments continue to receive Most South Asian countries have some transfers without penalty. Poor enforcement type of formula-allocated, unconditional mechanisms, weak capacity, and political transfers from central to urban local govern- interference impair the responses to audit ments, ranging from large allocations in findings. India and Pakistan to much more modest Bridging the accountability deficit will transfers in Bangladesh, Bhutan, Maldives, require the development of better systems and Nepal. However, although the transfers and practices and building the capacity of are officially unconditional, they often come both government (at all levels) and citizens. with higher-level rules and “guidance” on Accomplishing these tasks means nurturing use. In Afghanistan, municipalities do not the social contract between citizens and local receive transfers from the central govern- governments and clarifying fiscal relations ment; instead, block grants are channeled between local governments and higher tiers directly to community development commit- of government. In addition, local elections tees that deliver local services, bypassing the need to be transparent and sufficiently com- municipalities in the process. Across the petitive to give voters meaningful choices. OVERVIEW 7

Nonelectoral mechanisms—input-oriented to become more specialized in what they pro- processes, such as participatory planning and duce. Better transport links will lead to the budgeting, and feedback mechanisms, such development of more efficient systems of cit- as complaint bureaus and report cards—can ies. Complementing this, investments in be highly productive if well designed and improved intracity connectivity and traffic appropriately implemented. But they require management can enhance mobility within building the capacity of citizens to use them. urban areas and ease problems of traffic congestion. Three additional areas for The second strategy is to adopt forward- policy action looking planning approaches to guide expansion where it is most rapid—on city peripheries. Three additional, and interrelated, areas for Urban growth projections will have to be policy action are instrumental to addressing continually updated to better respond to congestion constraints and facilitating longer-term needs and to safeguard space for the further leveraging of urbanization future development. This strategy will reduce for improved prosperity and livability— the messiness of urbanization, prevent the connectivity and planning, land and housing, locking in of undesirable spatial forms, and and resilience to disaster and the effects of facilitate the future provision of infrastruc- climate change. ture and basic urban services. As a complement, the third strategy is to Connectivity and planning unlock the potential of city cores, rejuvenating those in decline. Revitalizing city cores For South Asian cities to realize their poten- will require investing in better-quality public tial and transform themselves into prosper- urban spaces to enhance pedestrian walkabil- ous and livable centers, they must not only ity and livability. It will also require promot- manage the frequently rapid expansion on ing better management of developable land in their peripheries; they must also address city cores through effective land-assembly existing and future challenges at their cores, mechanisms, freeing up publicly owned land, which often have largely locked-in spatial and repurposing structures for appropriate structures, contributing to heavy trafﬁ c con- uses. gestion and congestion in land and housing Fourth, to facilitate the formation of more markets. At the national level, how cities are vibrant neighborhoods, granular spatial connected as a system through flows of planning approaches can permit greater goods, labor, and ideas is important. Market variation in land uses and development forces interact with public policy decisions intensities. Such planning should be dynamic regarding both the placement of infrastruc- and flexible, allowing land uses to adapt to ture and the ways cities are planned to inﬂ u- changes in market demand in a framework ence both congestion forces and the evolution that takes a long-term view of a city’s of cities. development. This effort should be sup- To bolster opportunities for prosperous ported by strengthening the capacity of city and livable cities, planners and government planners and local governments to plan, decision makers can focus on four strategies. implement, and enforce development to The first is to invest in strengthening trans- deliver integrated, coordinated, and smarter port links that improve connectivity between planning policies. urban areas—between large and secondary cities, and secondary cities and towns. Land and housing Strengthening these links can facilitate the efficient deconcentration of land and capital- Highly congested land and housing markets intensive industries from more to less con- are exacerbating South Asia’s affordable gested cities and allow different urban areas housing crisis and undermining the livability 8 LEVERAGING URBANIZATION IN SOUTH ASIA

of the region’s cities. A shortage of affordable Faced with this situation, private developers housing already contributes to messy urban- are reluctant to acquire and assemble multi- ization, with at least 130 million people liv- ple parcels of land for residential developing in slums. Furthermore, it is not just the ment because they are unsure of the security poor, but also many middle-income house- of title. Financial institutions, likewise, are holds, that lack access to affordable housing. reluctant to finance land development or to Without fundamental reform, the crisis will accept land as collateral. South Asia can only get worse—a further 203 million hous- learn, based on the experience of other ing units will be required between 2010 and regions, how to make land tenure and titling 2050 to meet the projected growth in urban procedures more efficient, accurate, and population. transparent, and thus easier to navigate. To turn back the tide of proliferating Improving access to affordable land and slums, South Asian cities must embark on housing will require significant investments land and housing reforms and foster innova- in sustainable and affordable models of tive housing finance. City and suburban gov- finance to provide sufficient funds for building, ernments need to go beyond slum upgrading operating, and maintaining infrastructure. and embrace measures to stimulate the sup- Housing markets fail when developers ply of affordable housing and offer more ( private and public) cannot recover costs options to both low- and middle-income because of the lack of financial instruments. households. The supply of affordable housing Innovative approaches to address infrastruc- can be increased over time through more per- ture provision should therefore include the missive land-use and development regula- assessment and overhaul of existing policies tions. Also needed are infrastructure to open and tools to enable infrastructure financing up land for residential development, easy-to- through, for example, strategic disposition of use land titling and registration systems, and publicly held land, betterment levies, devel- greater access to construction and mortgage oper exactions, impact fees, and public- finance. In addition, government regulations private partnerships. need to be revised to foster the provision of more affordable rental housing. Resilience to disasters and the eff ects of Making land management more effective climate change will require cities to enhance their capacity to guide urban development and provide a By concentrating people and property in risk- framework for planning infrastructure prone areas, such as deltas, floodplains, investments that can make them livable and coasts, and the Himalayan belt, urban popu- inclusive. One aspect of land management is lation growth and economic development making land available for development. have increased the exposure of people, prop- Across South Asian cities, government agen- erty, and structures to natural hazards. Some cies own vast tracks of prime land, which are 80 percent of major South Asian cities are typically mismanaged and inadvertently con- exposed to fl oods, with about 45 percent of strain the urban land supply. urbanized locales lying in fl ood-prone areas South Asia’s cities desperately need effi- and 14 percent in extremely flood-prone cient land tenure and ownership record areas. Cities in Afghanistan, Pakistan, Nepal, systems. The lack of such systems is prevent- and northern India that lie along the ing private residential development in South Himalayan range are at risk of earthquakes; Asian countries from reaching sufficient scale many cities in Pakistan and northern India to accommodate urbanization. Except in are also at risk of heavy inland fl ooding. Nepal, land titling in South Asia is inefficient The first step in developing a resilience and expensive, encouraging many households strategy is to accurately identify and quantify to avoid formal processes, particularly when the national, subnational, and city risks. the risk of eviction or sanctions is limited. Governments at all levels should conduct risk OVERVIEW 9

assessments to define the potential risks and reduce the financial risk of disasters even the characteristics (such as frequency and more, cities can develop risk-insurance frame- severity) of potential hazards and to identify works independently and under a national the vulnerabilities of communities and the umbrella. The most extensive current risk potential exposures to given hazards. A risk- financing in the region is in Sri Lanka, which assessment framework can also guide govern- has a contingent credit line that provides ments in prioritizing measures for risk immediate liquidity to the country should it management, giving due consideration to the declare a state of emergency after a natural probability and impact of potential events, the disaster strikes. cost-effectiveness of preventive measures, and the availability of resources. The next step Moving forward: Leveraging would be to build a national georeferenced urbanization for greater prosperity hazard exposure database, which would and livability include public and private assets. Such information is critical to facilitating the insurance South Asia has so far struggled to make the industry in its offering of affordable property most of its urbanization. Difﬁ culties in deal- catastrophe insurance products. ing with the congestion forces brought about With the help of urban planners, engi- by the pressure of population on land, hous- neers, and academics, cities should revisit the ing, infrastructure, basic services, and the design and enforcement of building codes environment lie at the heart of the relative and land-use plans to avoid further building lack of its cities’ livability. By fostering messy in risk-prone areas and to reinforce structures and hidden urbanization, those forces are so they are resilient to various hazards. As also constraining the potential of powerful South Asia further develops its construction agglomeration economies to bring about industry, its cities can prevent substandard faster improvements in prosperity. structures from being built and substandard Looking ahead, South Asia’s policy mak- construction practices from being used. City ers face a choice between two paths. The first leaders can use various policy tools and is to continue with the same policies that have incentives to enforce building codes, which allowed congestion pressures in urban areas may mean extra costs for the government but to mount faster than might have otherwise will be an investment that can mitigate the been the case, thus undermining the exploita- costs associated with the aftermath of tion of agglomeration economies. This path disasters. would leave South Asia on its current trajec- As South Asia addresses its infrastructure tory of underleveraged urbanization, struc- deficit, and in so doing relieves congestion tural change, and development—locking pressures, city leaders must be cognizant of in the worst of the region’s urban problems future risks and hazards and ensure that new and risks. infrastructure is not built in hazard-prone The second path is to undertake difficult areas and does not expose communities to and appropriate policy reforms to alleviate additional risks. And in building infrastruc- both current and future congestion pressures ture, they need to identify and plan for and to facilitate the exploitation of agglom- critical infrastructure that is subject to eration economies, thereby enabling the tre- higher-than-usual “margins of safety” (the mendous untapped potential of its cities to be extra strength that engineers build into realized. This path will significantly improve designs). the likelihood that South Asia’s development National disaster risk-financing frame- trajectory will follow that of other countries works need to be developed based on risk lay- that have successfully transitioned to upper- ering to match risks with appropriate middle and high incomes. It will not be easy. financing instruments. Such frameworks will But it is essential to making the region’s cities help urban areas cope with disasters. But to prosperous and livable. 10 LEVERAGING URBANIZATION IN SOUTH ASIA

Notes matter with a diameter of less than 2.5 microns (that is, PM2.5) from the World Health 1. GDP per capita is measured in 2011 con- Organization’s “Ambient (outdoor) air stant international dollars using purchasing- pollution in cities database 2014” (http:// power-parity exchange rates. www.who.int/phe/health_topics/outdoorair 2. These include benefi ts arising from, for /databases/cities/en/). example, the spillover of ideas and knowl- 5. Data reused by permission of The Economist edge between fi rms and workers, the better Intelligence Unit. Further permission required matching of fi rms and workers that tends to for reuse. occur in dense urban labor markets, and the 6. In situ urbanization is urbanization that existence of dense networks of local suppliers is driven by natural population growth, of intermediate inputs. by the reclassifi cation of a settlement from 3. Data on urban poverty are not available for rural to urban, or both. Such urbanization Maldives. differs from that driven by net rural-urban 4. This designation is according to data on migration. annual mean concentrations of particulate Introduction

Summary disaster and the effects of climate change. The four areas are interlinked. Governance This report describes the progress of urban- and finance (chapter 3) are fundamental to ization and structural transformation in the supply of urban infrastructure and basic South Asia, the market and policy failures services and thus also undergird the three that have shaped the region’s towns and cit- other policy areas. Indeed, improvements in ies, and the decisive actions needed to better the way South Asia’s cities are governed leverage urbanization for South Asian pros- and financed may be a precondition for perity and livability. Part 1 (chapters 1–2) sets achieving meaningful progress in these other the context by analyzing key urbanization areas and therefore in prosperity and developments and related trends of structural livability. Of course, the ways in which cities transformation and economic growth. Part 2 are planned and connected (chapter 4) also (chapters 3–6) discusses and diagnoses the influence the supply of affordable housing main policy problems, identifies areas for (chapter 5). Likewise, urban planning poli- policy reform, and provides examples of best cies, and their enforcement, help determine a practices from both within and outside the city’s resilience to natural disaster and the region to help inform the decisions of policy effects of climate change (chapter 6). makers and practitioners. This report considers how these four pol- Chapter 1 focuses on recent urbanization- icy areas influence both prosperity and liva- related outcomes and trends at the regional bility through their impacts on the congestion level. It provides a broad comparison of the forces that afflict cities. Congestion forces region’s current state with other world emanate not just from the pressure of a city’s regions and with historical trends in devel- population on its transportation infrastruc- oped countries. Chapter 2 then provides a ture, but also on basic services, land, hous- spatially detailed analysis of trends and out- ing, and the environment. In particular, comes within the region and at the subna- policies in the four areas interact with the tional level. decisions of individual firms and households The chapters of part 2 identify four funda- to determine the strength of these forces, mental areas for policy action: governance which limit a city’s ability to benefit from and finance; connectivity and planning; land agglomeration economies—the benefits that and housing; and resilience to natural firms and workers derive from locating close

11 12 LEVERAGING URBANIZATION IN SOUTH ASIA

to one another. At the same time, excessive elected in 2014 to view urbanization as an congestion directly undermines a city’s liva- opportunity rather than a threat, able to bility and makes urbanization messier by transform the country’s cities so that they encouraging both slums and sprawl. At the “no longer remain a reflection of poverty and aggregate level, congestion also tends to act bottlenecks” but rather “become symbols of as a brake on the overall pace of urbanization. efficiency, speed and scale” (BJP 2014, 18). It By relaxing congestion constraints, policies is also seen in Pakistan’s Vision 2025 pro- can positively affect the key outcomes of gram, which places cities at the center of prosperity and livability at the local, national, national policy for sustained and inclusive and regional levels. economic growth.2 And in Bangladesh the government has recently revived discussions The growing recognition of about a national urban policy. Finally, urbanization’s importance Nepal’s 2012 creation of a Ministry of Urban Development (and the associated Kathmandu Urbanization is a cross-cutting and iterative Valley Development Authority) reveals that process involving demographic, economic, government’s recognition of the significance and spatial transformations. As urbanization of urban policy. concentrates people geographically, it also concentrates economic activity in those geo- A framework for assessing graphical areas. The founding fathers of urbanization and city performance modern economics—notably Adam Smith and Alfred Marshall—were quick to recog- Edward Glaeser, the eminent Harvard nize the benefits of such concentration as University urban economist, has defi ned cit- they watched industrialization and urbaniza- ies as “the [relative] absence of physical tion unfold in today’s developed countries. space between people and companies” But only in the last quarter century have (Glaeser 2011, 6). This close proximity of economists fully understood the benefits, people and production inevitably leads to particularly increased productivity and various kinds of market failures, which pol- higher standards of living, accruing from the icy should aim to address;3 it is these failures concentration of people and production in that make urban development—and by urban areas.1 Within their formal models, extension, urban policy making—such a economists have also embraced urbanization complicated fi eld. To help make sense of the as intimately related to a country’s structural complexities of urbanization, this report transformation, in particular to the transi- relies on the analytical framework set out in tion from agriculture to manufacturing and fi gure 1, which also provides the structure services, without which no country can hope for the report. to rise to middle-income status (see, for As a country urbanizes, its urban settle- example, Henderson and Wang 2005; ments grow in size and number. This growth Henderson, Roberts, and Storeygard 2013). gives rise to two competing sets of forces: Even more important than this elevated agglomeration economies and congestion academic appreciation is the growing recog- forces. nition among South Asia’s policy makers of By boosting productivity and job creation, the critical need to address the problems of agglomeration economies—the benefits that urbanization to cultivate its benefits. Their firms and workers derive from locating close traditional hostility toward urbanization, to one another—create further impetus for though lingering in places, has been replaced city growth and urbanization.4 These econo- by a more positive attitude that comprehends mies can take one of two basic forms: local- the tremendous opportunities it offers for ization or urbanization economies. prosperity. This new attitude can be seen in Localization economies are the productivity the commitment of the Indian government and cost advantages that firms in the same INTRODUCTION 13

FIGURE 1 Report structure and framework for assessing urbanization and performance

Chapter 1. Leveraging Urbanization for Prosperity and Livability

Chapter 2. Spatial Patterns of Subnational Performance and Urban Growth and Structural Transformation and Structural Part 1 Patterns of Urbanization 1 Patterns Part

Chapter 3 Chapter 4 Governance Connectivity and and Finance Planning

Agglomeration economies Congestion forces Productivity, skills, jobs, and Pressures on infrastructure innovation and markets

Chapter 5 Chapter 6 Land and Disaster Housing Resilience

Outcomes Part 2 Policies for Leveraging Urbanization Leveraging for 2 Policies Part Prosperity Livability

Sources: Based on Fujita, Krugman, and Venables 1999; Duranton 2014. industry enjoy from locating close to one infrastructure, basic urban services, land and another (Marshall 1890). Urbanization econ- housing markets, and the environment. omies are the productivity advantages that The interaction of agglomeration economies firms in different industries derive from with congestion forces determines the charac- locating in the same urban area (Jacobs 1969; ter of urbanization. As a city’s population Glaeser and others 1992). Agglomeration grows and congestion forces mount, urban- economies also arise because urban popula- ization tends to become messier. Mounting tion growth allows the fixed costs of new congestion pressures (especially in land and infrastructure—including infrastructure for housing markets) encourage both slums (as basic urban services and amenities (transpor- affordable housing becomes scarce) and sprawl tation, utilities, solid waste management, (as a city comes under increasing pressure to health and education facilities, and others)— grow outward). Similarly, mounting conges- to be spread over a larger number of people, tion forces make it more likely that population lowering the average cost of provision growth will be “displaced” to smaller com- (Armstrong and Taylor 2000). munities beyond municipal boundaries that Working against agglomeration economies are not officially recognized as urban, contrib- are congestion forces, which act as a brake on uting to hidden urbanization that fails to get both city growth and urbanization, hence on captured by official statistics. the prosperity that agglomeration economies A country’s pace of urbanization is also produce. These congestion forces arise from influenced by the interaction of agglomeration the pressure of increased population on urban economies with congestion forces. Pace will 14 LEVERAGING URBANIZATION IN SOUTH ASIA

tend to pick up as agglomeration economies system overall and for individual towns and increase their strength relative to congestion cities. Successful cities are also congested forces and to stabilize when these forces are cities, at least so far as their land and housing more closely aligned.5 markets, transport infrastructure, and labor The strength of agglomeration economies markets go. Thus Geneva; Hong Kong SAR, is, in turn, largely determined by the struc- China; London; New York; Paris; and ture of a country’s economy—agglomeration Singapore are among the most prosperous economies are mainly confined to the manu- and livable cities in the world, but they also facturing and services sectors, which, unlike have some of the most expensive property agriculture, can experience productivity ben- globally.6 Similarly, the speed of traffic in cen- efits from locating in urban areas. Empirical tral London in the first quarter of 2011/12 evidence suggests that agglomeration econo- was a mere 8.8 miles per hour (Transport for mies tend to be stronger at earlier stages of a London 2012), little changed from the era of manufacturing industry’s life cycle, when horse-drawn carriages.7 What separates innovation is rapid, and in higher-value- world-class cities like New York and London added tradable services such as banking and from South Asian cities is not that the former financial services, which rely intensely on have conquered congestion, but that they have human capital (see, for example, World Bank achieved much higher prosperity and livability 2008, chapter 4). The pace at which a coun- at comparable or lower levels of congestion. try urbanizes is thus closely related to the pace at which it transitions from an agrarian- based to a manufacturing- and services-based Sources of agglomeration economy. economies and congestion forces By contrast, the strength of congestion Agglomeration economies forces depends largely on the ability of investment in infrastructure and basic urban ser- Agglomeration economies fall into two basic vices to keep pace with increasing demand as categories as discussed previously—localization the numbers of people and firms in cities rise. economies (which beneﬁ t ﬁ rms in the same Without sufficient investment to expand industry) and urbanization economies (which capacity, the quality of urban infrastructure benefit firms in different industries). and basic services deteriorates. The strength Recognition of localization economies dates of congestion forces also depends on the abil- back to 19th-century England and the found- ity of land and housing markets to respond to ing father of modern microeconomics, Alfred increasing demand for residential, industrial, Marshall (Marshall 1890). He identified and commercial property, along with the three basic types of localization economies— ability of labor markets to respond to rising labor-market pooling, intermediate-input demand for workers. sharing, and within-industry knowledge The same logic that applies to a country’s spillovers (box 1). These economies helped overall urbanization also applies to individ- drive the localization of Bangladesh’s gar- ual urban areas: the population of a town or ment industry in Dhaka and Chittagong city tends to grow when the agglomeration (Muzzini and Aparicio 2013), India’s infor- economies it enjoys are strong relative to the mation and communication technology congestion forces it faces, making it a more industry in Bangalore (World Bank 2013b), attractive destination for migrants. Cities and Pakistan’s finance industry in Karachi with greater concentrations of industries that (World Bank 2014a).8 benefit from agglomeration economies tend Urbanization economies are similar to the to be larger in the long term. third type of localization economies, insofar The interplay of agglomeration economies as they derive from knowledge spillovers. with congestion forces also determines pros- These knowledge spillovers, however, are the perity and livability—both for the urban ideas that a firm acquires from observing INTRODUCTION 15

BOX 1 Types of localization economies

• Labor-market pooling. Both fi rms and workers close to one another, they create a large enjoy advantages from the dense labor market local market for intermediate inputs. This their colocation creates. Both benefi t from bet- large local market then encourages the ter matching. An information technology (IT) growth of a diversity of local special- fi rm looking for a programmer is more likely ized suppliers, which creates cost and pro- to fi nd one if it is located in Bangalore rather ductivity advantages for the downstream than in Chittagong. Similarly, a programmer is industry. more likely to fi nd work in IT if she lives in • Within-industry knowledge spillovers. Firms Bangalore than if she lives in Chittagong. in the same industry glean ideas and infor- Another benefi t for a worker is that if she is mation regarding best practices from one unfortunate enough to lose one job, she is another as a result of locating in the same more likely to fi nd another job if other fi rms in city. Knowledge spills over as one firm the same industry are located nearby. The observes how another fi rm in the same indus- worker benefi ts from reduced risk and, there- try operates, as workers move between dif- fore, a more certain income; this course of ferent firms in the same industry, and as events also benefits firms in the industry, friends who work for different firms talk because the worker is more likely to be willing shop in the street or at social occasions. to accept a lower average wage in Bangalore These spillovers also benefi t workers when given the lower chance of reemployment in they learn the latest developments in their Chittagong. industry, giving them a competitive edge in • Sharing of intermediate inputs. When a the labor market that can help them enjoy number of fi rms in the same industry locate faster wage growth over time.

practices in a different industry rather than more urbanization, creating a second round the same industry (Jacobs 1969; Glaeser and of increasing productivity and competitive- others 1992). It follows that urbanization ness, which would lead to a third round, and economies tend to be stronger in cities with so on—until urbanization is complete and greater diversity in the goods and services high-income status attained.9 they produce. And as with localization econ- Agglomeration economies do not operate omies derived from within-industry knowl- unopposed, however. They encounter resis- edge spillovers, urbanization economies tend tance from congestion forces, which can to be dynamic—their presence has a positive slow or, if strong enough, even halt the effect on a city’s productivity growth rate, whole process of circular and cumulative not just its productivity level. causation. Congestion forces make cities less attractive places to live and work and erode Congestion forces the competitiveness of firms, either by pushing up their costs or undermining their In theory, unopposed agglomeration econo- productivity. mies will increase urbanization and raise Congestion can arise from many differ- prosperity in a circular and cumulative pro- ent potential sources (box 2). Although all cess. The increasing concentration of people of these sources are present to some degree and production in urban areas would stimu- in South Asian cities, this report mainly late improvements in productivity and greater focuses on congestion in infrastructure, competitiveness, which would, in turn, spur basic urban services, and land and housing 16 LEVERAGING URBANIZATION IN SOUTH ASIA

BOX 2 Sources of congestion forces

• Land and housing markets. Even in the most 2011—also suffer from debilitating traffic efficient of settings, land markets respond congestion (World Bank 2014b). According only sluggishly to the pressures of urban pop- to basic economic principles, building more ulation growth, not least because the acquisi- roads can, in the absence of congestion pric- tion and provision of services to land that ing, ease traffi c congestion only in the short render it fi t for development can be costly and term since more roads tend to induce more time-consuming processes. As a consequence, traffi c, and in the long term there are physical as cities grow, land prices rise. This pressure constraints on road expansion (Duranton and increases both the costs of production for Turner 2011; Newbury 2000). Traffi c conges- firms seeking to acquire or rent land for tion imposes costs on both workers, who suf- industrial or commercial purposes and the fer from longer commutes, and businesses, cost of living for households that face higher which need to transport goods and compen- rents and property prices. In South Asia, high sate workers for their longer commutes. rents and property prices in the formal sector • Other infrastructure and basic urban services. often cause households to choose informal Urban growth also places pressure on other housing. The severity of this problem is evi- infrastructure and basic urban services, dent in the region’s large slum population (see including solid waste management and such chapter 5). utilities as water and electricity. Unless invest- • Labor markets. Urban growth will cause ment is forthcoming, this pressure can lead to wages to be bid up, eating into the profi ts of deteriorating levels of quality and service pro- firms, unless urban labor markets respond vision, which reduce firms’ and workers’ with an increased supply of suitably qualifi ed incentives to locate in a particular city—or labor. When the industries fueling growth are even in urban areas at all. During the next low skilled, rural-urban migration can help four decades, South Asia’s urban areas will release the pressure in urban labor markets. require more than $800 billion (in constant Labor mobility in South Asia is, however, rel- 2004 U.S. dollars) of investment in new and atively low—in India, according to 2001 cen- replacement road, water, and sanitation infra- sus results, only 0.4 percent of the population structure to keep pace with urbanization and lived in a different state five years earlier avoid further deterioration in service quality compared with 9 percent in the United States (see chapter 5). (Glaeser, Chauvin, and Tobio 2011; see • Pollution. Because of increased traffi c conges- chapter 2). As urbanization and development tion, not to mention building activity, bigger proceed to a more advanced stage, the empha- cities tend to suffer from poorer air quality. In sis shifts to more human capital–intensive India, bigger cities have higher concentrations industries. This shift places ever-increasing of nitrogen oxide, both an important indica- pressure on the education sector to deliver tor of air quality and a serious cause of respi- suitably qualifi ed workers. ratory illness (Lamsal and others 2013). More • Roads. As cities grow, their roads tend to generally, the relationship between population become clogged with traffic. Workers in density and particulate matter concentrations New Delhi and Bangalore, for instance, is stronger for South Asian cities than for suffer among the worst commutes in the cities in the rest of the developing world world, according to IBM’s Commuter Pain (chapter 1). Linked also to the forces of con- Index.10 Some secondary cities such as Kandy gestion on infrastructure and basic urban ser- in Sri Lanka—whose traffi c volume grew at vices, urban density can cause both streets 5 percent annually between 1998 and and water supplies to become polluted with

(continues next page) INTRODUCTION 17

BOX 2 Sources of congestion forces (continued)

garbage and human waste, with serious pub- larger markets for drug dealers, fueling the lic health consequences. potential for drug-related crime. Karachi, for • Disease. In addition to diseases that can arise example, is well known for the presence of from pollution, infectious diseases can spread powerful criminal gangs that engage in extor- faster in dense urban settings than in rural tion and land-grabbing in addition to the drug ones. trade (World Bank 2013a, 2014a). Larger cit- • Crime. Glaeser has noted, “If I’m close ies also provide more attractive targets for ter- enough to sell a newspaper, you’re close rorist organizations intent on causing enough to rob me” (Glaeser 2011). Cities, by destruction and loss of life—a problem sadly virtue of their density, can create the condi- too evident in South Asia in recent decades tions for crime. Larger cities also provide (World Bank 2013a).

markets, as well as on the pressures that 3. Both the negative and positive externali- congestion imposes on the environment. ties that are associated with urban size and These are the most generally important in density are sources of market failure. The explaining the character of urbanization in negative externalities include, for example, South Asia. This is in addition to the pollution and congestion of infrastructure and land markets, while the positive exter- report’s focus on the risks from natural nalities include the different sources of disasters and the effects of climate change, agglomeration economies. Both negative and risks that can be exacerbated by a failure to positive externalities lead to socially subopti- adequately deal with the pressure of urban mal results and provide potential justiﬁ cation population growth, where this contributes for public policy intervention. The existence to, for example, poorly maintained infra- of the need for public goods, such as street structure and the growth of slum popula- lighting, that would be underprovided if the tions who live in poorly constructed market were left to itself is another source of housing. More generally, natural disasters market failure that afﬂ icts urban areas. can threaten the hard-won gains in prosper- 4. Urban bias, which can arise, for example, ity and livability associated with the man- from the distribution of agricultural sur- pluses to cities, can provide an additional agement of congestion forces and cultivation impetus to the growth of cities and urbaniza- of agglomeration economies. tion (Renaud 1981; Henderson 1988; Ades and Glaeser 1995; Davis and Henderson Notes 2003). But there is little evidence of such a bias for South Asia, policy having favored 1. This increased appreciation and understand- rural areas historically. In recent years, policy ing is due, in part, to the emergence of the has become less biased toward rural areas, so-called new economic geography—a body as policy makers realize the opportunities of theoretical literature, most notably asso- for economic development that urbanization ciated with economics Nobel laureate Paul provides. Krugman, that aims to better understand the 5. More generally, the pace of urbanization micro-foundations of large-scale processes of within a country depends on the differential agglomeration (see Krugman 1991a, 1991b; between real wages and living conditions in Fujita, Krugman, and Venables 1999). urban versus rural areas. The pace of urban- 2. “Pakistan 2025: Together for Better Tomorrow” ization therefore tends to accelerate as these (http://www.pc.gov.pk/?page_id=73). differentials increase. It follows that any 18 LEVERAGING URBANIZATION IN SOUTH ASIA

deterioration in rural conditions as a conse- International Trade. Cambridge, MA: MIT quence of, for example, crop failure or con- Press. ﬂ ict focused on rural areas can also lead to an Glaeser, E. 2011. Triumph of the City: How Our accelerated pace of urbanization. Greatest Invention Makes Us Richer, Smarter, 6. See, for example, data from the Global Greener, Healthier, and Happier. New York: Property Market Guide (http://www Penguin Press. .globalpropertyguide.com/most-expensive Glaeser, E., J.-P. Chauvin, and K. Tobio. 2011. -cities). “Urban Economics in the U.S. and India.” 7. “London’s Smart Move” (http://connectedcities Presented at the Economic Geography . eu/downloads/magazines/nt_2008_april_tdm Conference, Seoul, Republic of Korea, June 29. . pdf). http://www.scribd .com/doc/59978593/Prof-Ed 8. Localization of industry can also be driven -Glaeser-Urban-Economics-in-the-US-and-India. by spatial competition as in the classic Glaeser, E. L., H. D. Kallal, J. A. Scheinkman, and Hotelling (1929) model. This spatial com- A. Shleifer. 1992. “Growth in Cities.” Journal petition, rather than localization economies, of Political Economy 100 (6): 1126–52. probably explains the clustering of retailers Henderson, J. V. 1988. Urban Development: of similar products observed, for example, in Theory, Fact, and Illusion. New York: Oxford Old Delhi. University Press. 9. This process follows from both early theo- Henderson, J. V., M. Roberts, and A. Storeygard. ries of circular and cumulative causation 2013. “Is Urbanization in Sub-Saharan Africa (Myrdal 1957; Kaldor 1970) and more Different?” Policy Research Working Paper recent new economic geography models 6481, World Bank, Washington, DC. (Krugman 1991a, 1991b; Fujita, Krugman, Henderson, J. V., and H. G. Wang. 2005. “Aspects and Venables 1999). of the Rural-Urban Transformation of 10. “Frustration Rising” IBM 2011 Commuter Countries.” Journal of Economic Geography Pain Survey, http://www-03.ibm.com/press 5 (1): 23–42. / us/en/pressrelease/35359.wss. Hotelling, H. 1929. “Stability in Competition.” Economic Journal 39: 41–57. References Jacobs, J. 1969. The Economy of Cities. New York: Random House. Ades, A., and E. L. Glaeser. 1995. “Trade Kaldor, N. 1970. “The Case for Regional and Circuses: Explaining Urban Giants.” Policies.” Scottish Journal of Political Quarterly Journal of Economics 110 (1): Economy 17 (3): 337–48. 195–227. Krugman, P. 1991a. Geography and Trade. Armstrong, J., and H. Taylor. 2000. Regional Cambridge, MA: MIT Press. Economics and Policy. 3rd ed. Oxford: ———. 1991b. “Increasing Returns and Economic Blackwell. Geography.” Journal of Political Economy 99 BJP (Bharatiya Janata Party). 2014. Election (3): 483–99. Manifesto 2014. New Delhi: Bharatiya Janata Lamsal, L. N., R. V. Martin, D. D. Parrish, and Party. http://www.bjp.org/manifesto2014. N. A. Krotkov. 2013. “Scaling Relationship for

Davis, J., and J. V. Henderson. 2003. “Evidence NO2 Pollution and Urban Population Size: A on the Political Economy of the Urbanization Satellite Perspective.” Environmental Science Process.” Journal of Urban Economics 53 (1): & Technology 47 (14): 7855–61. 98–125. Marshall, A. 1890. Principles of Economics. Duranton, G. 2014. “Growing through London: Macmillan. Cities in Developing Countries.” Policy Muzzini, E., and G. Aparicio. 2013. Bangladesh: Research Working Paper 6818, World Bank, The Path to Middle-Income Status from an Washington, DC. Urban Perspective. Washington, DC: World Duranton, G., and M. Turner. 2011. “The Bank. Fundamental Law of Road Congestion: Myrdal, G. 1957. Economic Theory and Evidence from U.S. Cities.” American Underdeveloped Regions. London: Economic Review 101 (6): 2616–52. Duckworth. Fujita, M., P. Krugman, and A. J. Venables. 1999. Newbury, D. 2000. “Pricing and Congestion: The Spatial Economy: Cities, Regions and Economic Principles Relevant to Pricing INTRODUCTION 19

Roads.” In Readings in Microeconomics, 2nd ———. 2013a. “Note on Urbanization and ed., edited by T. Jenkinson, 141–54. Oxford: Violence in the South Asia Region.” Oxford University Press. Background paper for this report, World Bank, Renaud, B. 1981. National Urbanization Policy Washington, DC. in Developing Countries. New York: Oxford ———. 2013b. Urbanization beyond Municipal University Press. Boundaries: Nurturing Metropolitan Transport for London. 2012. “London Streets: Economies and Connecting Peri-Urban Areas Performance Report, Quarter 1, 2011/12.” in India. Washington, DC: World Bank Greater London Authority, London. http:// ———. 2014a. “Draft Pakistan Urban Sector www.tfl.gov.uk/assets/downloads/streets Assessment.” World Bank, Washington, DC. -quarter-one-2012.pdf. ———. 2014b. “Economic Analysis of Traffic World Bank. 2008. World Development Report Improvements, Kandy City-Region.” 2009: Reshaping Economic Geography. Background paper for the Sri Lanka Strategic Washington, DC: World Bank. Cities Project, World Bank, Washington, DC.

PART 1

Patterns of Urbanization and Structural Transformation

s South Asia’s urban population and livability—delivered by that process. grows, the resulting concentration of Part 1 of this report analyzes South Asia’s Apeople and production, and the urbanization process and the associated eco- structural transformation that accompanies nomic and structural transformations the this concentration, provide opportunities for region is undergoing. It also analyzes the enhanced prosperity and livability through outcomes delivered by the process, with a the exploitation of agglomeration econo- focus on the two key outcomes of prosperity mies. But constraining the realization of this and livability. Chapter 1 compares South potential are congestion forces arising from Asia’s urbanization experience with that of the pressure of urban population on infra- other regions and with the historical experi- structure, basic services, land, housing, and ences of today’s developed countries. the environment. The interaction of agglom- Chapter 2 carries out a more spatially eration economies with congestion forces detailed analysis, largely at the subnational determines the character and speed of a level, of trends and outcomes within the country’s urbanization process, as well as region. This analysis informs the policy the outcomes—in terms of both prosperity discussion in part 2 of the report.

Leveraging Urbanization for Prosperity and Livability 1

Key messages

South Asia’s urban population has increased the East Asia and Pacific region, as well as steeply since the turn of the century, and the that historically experienced in today’s region has made impressive progress toward developed countries. greater prosperity. Urbanization now presents • South Asia’s share of global GDP, though South Asian countries with a chance to transform rising since the turn of this century, is much their economies and reach development’s upper lower than its share of the global urban tiers of prosperity and livability. To take full population. advantage, however, these countries must under- • Urbanization and structural transformation in take difﬁ cult policy actions. many South Asian countries, driven in part by global trends, are being led by nontradable • Urbanization in South Asia has been messy services rather than by manufacturing, and hidden. South Asian cities, with some making it very hard to exploit the full range exceptions, still have poor livability as of agglomeration economies. evidenced by the prevalence of slums and • South Asia’s congestion pressures are a funda- sprawl, not to mention poverty and pollution. mental problem for its cities, undermining the And much of this urbanization has not been potential economic leverage that should be captured by official statistics. South Asia’s derived from urbanization. urbanization has also been slower than that in

Introduction populous country, Japan—from 382 million to 511 million.1 The productivity benefits Between 2000 and 2011, the number of peo- linked to this large urban population have ple ofﬁ cially living in South Asia’s towns and been important in the region’s progress cities swelled by slightly more than toward higher prosperity. Between 2000 and 130 million—equivalent to more than the 2012, average real gross domestic product entire population of the world’s 10th most (GDP) per capita increased by almost 23 24 LEVERAGING URBANIZATION IN SOUTH ASIA

56 percent, from $2,556 to $3,999, with offices in South Asia do not recognize them as annual GDP per capita growth rates of more such. And third, although the absolute than 4.5 percent a year in all countries except increase in the official urban population has Nepal and Pakistan. And while half of South been large—South Asia’s population is huge Asians were living on less than $1.25 a day in after all—the rate of gain, when compared 1999, fewer than a third were by 2010.2 with that in the East Asia and Pacific region The gains in prosperity associated with and the historical experiences of today’s South Asia’s rising urban population bode developed countries, has been relatively slow, well for the region’s development, particularly and the share of its population living in such with a forecasted increase of almost settlements was only slightly higher in 2011 302 million people living in recognized urban than it was at the beginning of the century. settlements between 2011 and 2030—almost Messy and hidden patterns of urbanization equivalent to the entire population of the are symptoms of market and policy failure: United States. Urbanization and the associ- the region’s towns and cities fail to adequately ated structural transformation present South ease the pressures of their growing popula- Asia with a tremendous opportunity to reach tions on land, housing, infrastructure, ser- the upper echelons of economic development. vices, and the environment. These congestion As theory teaches and history demonstrates, pressures undermine the region’s ability both successful urbanization is accompanied by to exploit the full range of agglomeration successful economic development as workers economies and to compete in international move from low-productivity agricultural export markets, especially for manufactured activities to high-productivity urban jobs in goods. They are also constraining the capac- manufacturing and services (Lewis 1954). ity of urbanization overall and of cities indi- The higher productivity of urban jobs, in vidually to improve prosperity and livability. turn, stems partly from the agglomeration To better leverage urbanization, policy benefits that cities offer (Glaeser 2011). makers will need to push through difficult Moreover, the benefits of successful urbaniza- policy actions. If they do not, the region’s tion for South Asia will not be confined to its path of urbanization, structural change, and cities but will spill over to its rural areas. development is unlikely to shift to a higher Successful urbanization is thus accompanied trajectory. Some of the worst urban problems by a long-term convergence in living stan- are also likely to become more tightly locked dards between urban and rural areas (World in, making future policy actions even more Bank 2008). Empirical evidence already difficult. The historical experience of today’s shows strong beneficial spillover effects from advanced countries and East Asia’s more urban to rural areas for India and Nepal (Cali recent experience show that policies to lever- and Menon 2013; Fafchamps and Shilpi age urbanization offer faster progress toward 2005). prosperity and livability. South Asia must fol- Still, urbanization in the region remains low a similar path. underleveraged and has, since the turn of the century, been messy and hidden. First, urban- South Asia’s recent urbanization in ization has been messy in that a majority of global and historical perspective cities in the region still exhibit poor livability, as seen in the widespread prevalence of slums If we were to believe offi cial national estimates and sprawl. Second, because of tight official of the share of people living in urban areas definitions of urban areas, significant hidden from the United Nations’ World Urbanization urbanization is not captured in official statis- Prospects database, South Asia would be the tics. A nonnegligible share of the region’s least urbanized region in the world, behind population is living in areas with traits that even Sub-Saharan Africa (fi gure 1.1).3 As is would be considered urban in many other well known, however, such comparisons are regions, even though national statistical hampered by defi nitions of urban areas that LEVERAGING URBANIZATION FOR PROSPERITY AND LIVABILITY 25

vary widely across countries. The difﬁ culties FIGURE 1.1 South Asia is among the least urbanized regions are symptomatic of wider data problems, limiting descriptive analysis of urbanization 90 and related economic trends (box 1.1). 80 The agglomeration index (AI)—originally 70 developed for World Development Report 60 2009: Reshaping Economic Geography 50 (World Bank 2008)—provides a more consistent basis for comparing urbanization across 40

countries and regions by adopting a uniform Urban share (%) 30 definition of urban areas (box 1.2).4 Based 20 on the AI, South Asia is not the world’s least 10 urbanized region: Whereas official estimates 0 for 2010 suggest that slightly more than one OHIE OECD MENA LAC ECA SAR EAP SSA in three South Asians live in towns and cities, AI UN the AI suggests that a little more than one in two do (figure 1.1).5 This AI estimate places Source: Calculations based on UN World Urbanization Prospects: 2011 Revision data. South Asia ahead of Sub-Saharan Africa and Note: AI indicates urban shares based on the agglomeration index; UN indicates offi cial national East Asia and Pacific. It also suggests that the estimates of urban shares (which, for South Asia, include the share of the population living in India’s census towns) as taken from World Urbanization Prospects. EAP = East Asia and Pacifi c; ECA = Europe region has significant “hidden” urbaniza- and Central Asia; LAC = Latin America and the Caribbean; MENA = Middle East and North Africa; tion, not picked up by official statistics. This OECD = Organisation for Economic Co-operation and Development; OHIE = other high-income economies; SAR = South Asia; SSA = Sub-Saharan Africa. hidden urbanization is particularly prevalent on the peripheries of the region’s major

BOX 1.1 Investment in data for South Asian countries is urgently required

Three basic data shortcomings limit rigorous database provides a set of economic, environ- descriptive analysis of urbanization and related mental, and demographic indicators for all economic trends in South Asia. functional urban areas with populations of more than 500,000 in member countries.a • Administrative-based definitions of urban • Lack of comparability across countries and over areas. Offi cial statistics for cities produced by time. Beyond the differing defi nitions of urban national statistical offi ces and other govern- that complicate cross-country comparisons (see ment agencies tend to be based on administra- box 1.2), defi nitions of “urban” within a coun- tive defi nitions. The administrative boundaries try can change over time. Take Sri Lanka: in of a city, however, often fail to accurately 1987, it tightened its defi nition by reclassifying delineate a city’s true built-up extent. In South town councils as rural areas (pradeshiya sabha). Asia, many cities are “under-bounded” This move contributed to a fall in the country’s (Colombo) and some are “over-bounded” official urban share from 21.5 percent in the (Mumbai). Nor do administrative defi nitions 1981 census to 14.6 percent in 2001. consider commuting patterns. Ideally, for a • Lack of data. Much progress has been made consistent analysis of urbanization trends, cit- in recent decades in putting developing coun- ies should be defined using functional, not tries across the globe on a common time cycle administrative, criteria. An example of best for national population and housing censuses practice is the Metropolitan Areas Database and in improving the quality of census opera- of the Organisation for Economic tions. Bangladesh, India, Nepal, and Sri Lanka Co-operation and Development (OECD). This (continues next page) 26 LEVERAGING URBANIZATION IN SOUTH ASIA

BOX 1.1 Investment in data for South Asian countries is urgently required (continued)

each conducted a census in 2011, and the This report attempts to overcome some of the Maldives in 2014. Bhutan has not conducted region’s data deﬁ ciencies by drawing on uncon- a census since 2005, and Afghanistan and ventional data sources, such as nighttime lights Pakistan completed their last censuses in 1979 and other forms of remotely sensed earth obser- (partial) and 1988, respectively. Sri Lanka’s vation data. It also attempts to tell an internally 2001 census was also partial, covering only consistent story of South Asia’s urbanization 18 of the country’s 25 districts because of civil based on the data available. A general conclu- war. Comprehensive subnational data on sion, however, is that South Asian countries GDP are also absent for the region, although require urgent investment in their capacity to GDP data with limited temporal coverage are generate higher-quality and more comprehen- available for Indian districts. By contrast, sive data. Until that happens, any analysis will Eurostat—the European Union’s (EU’s) statis- have important caveats. tical agency—publishes data on GDP per inhabitant dating back to 2000 and down to a. For the Metropolitan Areas Database, see http://www.oecd.org/gov/regional -policy/regionalstatisticsandindicators.htm. the third administrative level for all EU b. See http://epp.eurostat.ec.europa.eu/portal/page/portal/region_cities member countries.b /regional_statistics/data/database.

BOX 1.2 Defi ning and comparing urban areas —The agglomeration index and night-lights

Urban demographers and economists face a criterion to defi ne urban settlements (more than persistent and onerous problem when com- 5,000) but has a qualifying statement that the paring urbanization trends across countries or settlement’s population must live in a “continu- regions: no single standard defi nition of urban ous collection of houses where the community areas is used by national census bureaus. This sense is well developed.” Furthermore, an urban lack of standardization complicates compari- settlement must have (among other things) a sons of urbanization trends across South Asian “community [that] maintains public utilities, countries, just as it complicates comparisons of such as roads, street lighting, water supply, South Asia with other world regions. In particu- sanitary arrangements, etc.” and be a center “of lar, comparisons of trends that rely on the UN trade and commerce where the labor force is World Urbanization Prospects—the most glob- mostly non-agricultural.”6 ally comprehensive source of urbanization data— Seven criteria usually feature in the defi nitions have to be treated with caution because this data of “urban” in South Asian countries: local gov- set relies on urban population data published by ernment, population, population density, area these bureaus. of settlement, access to services, structure of the As an illustration of the differing defi ni- local economy, and literacy rate. Bangladesh tions of urban settlements across South Asian uses fi ve of these criteria. Other South Asian countries, consider Nepal and Bangladesh. countries vary on the combination of the seven Nepal uses a single basic criterion to defi ne an and in their defi nitions of them. urban settlement—a population of more than To overcome these problems of comparabil- 9,000. Bangladesh also uses a single population ity, this report uses two methods.

(continues next page) LEVERAGING URBANIZATION FOR PROSPERITY AND LIVABILITY 27

BOX 1.2 Defi ning and comparing urban areas —The agglomeration index and night-lights (continued)

First, for comparing urban population shares Second, for comparing spatially detailed pat- across countries and regions, it uses the agglomer- terns of urbanization, particularly focusing on ation index (AI) developed by Chomitz, Buys, and urban footprint expansion patterns and pat- Thomas (2005) and Uchida and Nelson (2008). terns of economic growth within South Asia, The AI is based on three parameters: population the report uses night-light earth observation density (at least 150 people per square kilometer), data remotely collected by satellites that are a threshold population of a “large” urban cen- part of the U.S. Defense Meteorological Satellite ter (50,000), and a maximum travel time to that Program. These innovative data allow the center (60 minutes). Using this method, national extents of urban areas to be consistently deﬁ ned urban shares that allow consistent comparisons across the eight countries based on their aver- of South Asia with other world regions and of age nighttime luminosity, revealing patterns of individual South Asian countries with each other urbanization that would be hard to obtain with were estimated. The AI uses a population den- conventional sources (see chapter 2). sity threshold that, although suitable for global comparisons is low relative to average population a. In 2010, six of the region’s eight countries (not Afghanistan and Bhutan) had average population densities in excess of the AI threshold, implying that if this densities in South Asia; it thus provides upper- were the sole criterion for defi ning urban, they would each be classifi ed as 100 bound estimates of urban shares.a This index also percent urbanized. b. These updated estimates were fi rst developed for Global Monitoring Report featured in World Development Report 2009 2013: Rural-Urban Dynamics and the Millennium Development Goals (World Bank (with estimates for 2000), updated here to 2010 and IMF 2013). Unfortunately, issues with the data mean that the new updated b estimates cannot be compared with the earlier estimates for 2000 to draw using the same parameters as in that report. conclusions about trends over time.

cities, where congestion pressures have FIGURE 1.2 South Asia’s low agglomeration index is consistent encouraged sprawl and growth beyond with its level of development municipal boundaries (see chapter 2). Nevertheless, even given this hidden urban- 100 ization and especially compared with 90 Organisation for Economic Co-operation 80 and Development (OECD) and other high- income economies, South Asia remains at a 70 60 PAK relatively early stage of urbanization, consis- IND MDV tent with the relatively low levels of develop- 50 LKA ment that characterize the region’s eight 40 BGD AFG countries (figure 1.2). 30 NPL South Asia’s urban share might lag behind Agglomeration index (%) 20 those of high-income and developing regions 10 BTN like Latin America and the Caribbean, but it 0 has been catching up. However, to assess the 0 20,000 40,000 60,000 80,000 speed with which it is catching up the analy- GDP per capita, 2010 sis must fall back on official national estimates of urban population shares owing to Source: Calculations based on World Bank World Development Indicators data. Note: GDP per capita is expressed in constant 2011 international dollars at purchasing power parity the lack of comparable AI estimates over exchange rates. AFG = Afghanistan; BGD = Bangladesh; BTN = Bhutan; IND = India; LKA = Sri Lanka; time. Use of national estimates is far from MDV = Maldives; NPL = Nepal; PAK = Pakistan. 28 LEVERAGING URBANIZATION IN SOUTH ASIA

ideal, but the results here are robust to grew by almost 4.0 percent a year, from reasonable assumptions about potential 27.7 percent to 40.7 percent. biases in the official data. The results are also Even if South Asia’s urbanization speed is robust when the analysis is confined to coun- relatively slow, the absolute increase in its tries that use similar definitions of “urban” urban population has been huge (figure 1.5). (see web-based annex 1A and Roberts During 2000–11, the number of people offi- [2015]).7 cially classified as living in urban areas At 1.1 percent a year, the growth rate of increased by slightly more than 130 million— South Asia’s urban population share during more than the entire population of Japan, the 2000–11 was on par with Sub-Saharan world’s 10th most populous country. South Africa’s ( figure 1.3). Rapid by some measures, Asia’s share of the global urban population the pace looks slow when set against East increased by 0.6 percentage points in that Asia and Pacific, whose urban share grew period, at the expense of more urbanized and more than twice as fast as South Asia’s.8 developed regions (table 1.1). Its share of The impression of slow urbanization con- global GDP rose by 2.2 percentage points, tinues in a comparison of South Asia with the reflecting South Asia’s strong economic historical experiences of today’s developed growth relative to other regions since the start countries when they were at urbanization lev- of the century, growth that has facilitated els similar to South Asia’s today (figure 1.4). some catch-up in prosperity. Europe’s urban share grew 1.3 percent a year Yet it is striking how low South Asia’s during 1880–1900, taking it from share of global GDP remains relative to its 23.5 percent to 30.4 percent. In Australia, share of the global urban population, even Canada, New Zealand, and the United States compared with other regions containing (other developed countries), the urban share developing countries (figure 1.6). For North climbed from 24.4 percent to 35.6 percent, America and Europe and Central Asia, the for annual growth of 1.9 percent. During difference between each region’s share of 1960–70, the Republic of Korea’s urban share global GDP and its share of global urban population is positive. The difference is also slightly positive for the Middle East and North Africa. By contrast, for the remaining, FIGURE 1.3 Annual growth rate of South Asia’s urban share more developing regions, the difference is compared with other regions, 2000–11 negative, most so for South Asia—consistent with the notion that other regions have more 3.0 successfully leveraged their urbanization for productivity and prosperity.9 2.5 As South Asia has increased its share of global urban population and global GDP, it 2.0 has also expanded its share of global manufacturing value added (table 1.1). This change 1.5 in manufacturing value added suggests that urbanization and economic growth in the (% per year) 1.0 region have been accompanied by structural

Urban share growth rate transformation, consistent with the historical 0.5 evidence that as countries urbanize, they also become increasingly less dependent on agri- 0 culture (figure 1.7). Indeed, structural trans- EAPSSA SAR MENA LAC ECA formation is thought to provide the key link between urbanization and economic growth Source: Calculations based on UN World Urbanization Prospects: 2011 Revision data. Note: EAP = East Asia and Pacifi c; ECA = Europe and Central Asia; LAC = Latin America and the (Henderson and Wang 2005; Michaels, Caribbean; MENA = Middle East and North Africa; SAR = South Asia; SSA = Sub-Saharan Africa. Rauch, and Redding 2012). The reason? LEVERAGING URBANIZATION FOR PROSPERITY AND LIVABILITY 29

Agriculture is characterized by constant FIGURE 1.4 Annual growth rate of South Asia’s urban population returns to scale, whereby output increases share compared with annual growth rates historically experienced only proportionally with inputs; but nonagri- by today’s developed countries cultural activities tend to benefit more from increasing returns and agglomeration econo- 4.0 mies. This effect is particularly true of manu- 3.5 facturing and higher-value-added tradable services, notably information and communi- 3.0 cations technology, banking and finance, and 2.5 other knowledge-based services. But the increase in South Asia’s share of 2.0 global manufacturing activity in table 1.1 is deceptive in that it is due more to deindustri- 1.5 alization in North America and Europe and Urban share growth rate (%) rate growth Urban share 1.0 Central Asia than to rapid manufacturing growth in South Asia. In Afghanistan, Nepal, 0.5 Maldives, and Pakistan, the shift out of agri- 0.0 culture has been accompanied by a large South Asia Europe Other Republic decline in the proportion of GDP from manu- (2000–11) (1880–1900) developed of Korea facturing (figure 1.8). India, although avoid- countries (1960–70) ing a decline, has managed only to keep its (1880–1900) manufacturing contribution to GDP level. Sources: Calculations based on UN World Urbanization Prospects: 2011 Revision data and historical This finding implies that—apart from urban share data from Bairoch and Goertz (1986, 288). Bangladesh and Sri Lanka, with their export Note: Europe comprises Austria, Hungary, Belgium, Bulgaria, Denmark, Finland, France, Germany, Greece, Italy, the Netherlands, Norway, Portugal, Romania, the Russian Federation, Serbia, Spain, success primarily in garments and textiles— Sweden, Switzerland, and the United Kingdom. Other developed countries comprise Australia, urbanization since the turn of the century has Canada, New Zealand, and the United States. been associated more with the growth of services.10 Services-led urbanization, accompanied by FIGURE 1.5 Absolute increase in South Asia’s urban population either stagnation or decline in the relative compared with other regions, 2000–11 importance of manufacturing, is something of a departure from expected trends based on 350 the historical experiences of today’s developed countries (also see Rodrik 2015). It is 300 also in stark contrast to China, where both urbanization and development have been 250 driven by manufacturing, export-led growth (box 1.3). 200 South Asia’s atypical pattern of urbanization and structural transformation need not 150 be a cause for concern if the services industries that workers are moving into have higher 100 productivity and are more dynamic than agri- (million) increase Absolute culture and manufacturing. The burgeoning 50 of India’s information technology sector seems particularly promising, having grown 0 extremely fast in recent decades and generat- EAP SAR SSA LAC MENA ECA ing about 6–7 percent of GDP and about Source: Calculations based on UN World Urbanization Prospects: 2011 Revision data. 18 percent of exports (D’Costa 2011). The Note: EAP = East Asia and Pacifi c; ECA = Europe and Central Asia; LAC = Latin America and the industry has benefited from powerful Caribbean; MENA = Middle East and North Africa; SAR = South Asia; SSA = Sub-Saharan Africa. 30 LEVERAGING URBANIZATION IN SOUTH ASIA

TABLE 1.1 Global shares of urban population, GDP, and manufacturing value added

Urban population GDP Manufacturing value added 2000 2011 Change 2000 2011 Change 2000 2011 Change SAR 13.3 14.0 0.6 5.7 7.9 2.2 1.7 3.1 1.4 EAP 29.4 32.0 2.6 23.0 29.1 6.0 30.3 33.9 3.7 SSA 7.5 8.9 1.5 2.5 3.0 0.6 0.9 1.1 0.1 MENA 6.4 6.8 0.3 6.4 7.1 0.6 1.7 2.6 0.9 LAC 14.0 13.2 −0.8 9.6 9.2 −0.4 6.5 6.7 0.2 NAC 8.7 7.9 −0.9 22.9 18.3 −4.6 24.6 22.0 −2.5 ECA 20.7 17.3 −3.4 29.9 25.6 −4.4 33.9 31.5 −2.4

Sources: Calculations based on UN World Urbanization Prospects: 2011 Revision and World Bank World Development Indicators data. Note: GDP is measured at 2011 constant (purchasing power parity) international dollars. EAP = East Asia and Pacifi c; ECA = Europe and Central Asia; LAC = Latin America and the Caribbean; MENA = Middle East and North Africa; NAC = North America; SAR = South Asia; SSA = Sub-Saharan Africa.

FIGURE 1.6 Global GDP share minus global urban population Sub-Saharan Africa. Between 1990 and 2010, share for major world regions, 2011 agriculture’s share in the aggregate employment of 11 African countries fell from 15 61.6 percent to 49.8 percent and manufacturing’s from 8.9 percent to 8.3 percent, as ser- 10 vices’ share expanded from 24.1 percent to 36.8 percent. The expansion was, however, led 5 by services such as retail trade and distribution, 0 which saw declining average labor productivity relative to the economy as these sectors –5 absorbed more workers. Africa’s pattern of Share of global GDP minus

share of global urban population structural change has yielded increases in pro- –10 ductivity levels (that is, static productivity SAR SSA LAC EAP MENA ECA NAC gains), while the overall contribution of structural change to productivity growth has been Sources: Calculations based on UN World Urbanization Prospects: 2011 Revision and World Bank World Development Indicators data. limited (de Vries, Timmer, and de Vries 2013). Note: GDP is measured at 2011 constant (purchasing power parity) international dollars. EAP = East Asia The challenge for South Asia is to avoid and Pacifi c; ECA = Europe and Central Asia; LAC = Latin America and the Caribbean; MENA = Middle East and North Africa; NAC = North America; SAR = South Asia; SSA = Sub-Saharan Africa. following a path similar to Africa’s—it needs to create urban environments that revitalize agglomeration economies associated with the manufacturing and encourage the emergence geographic concentration in Bangalore. Other and growth of higher-value-added tradable countries in the region have also experienced services. As export-led theories of economic rapid growth of their information technology growth show, nontradable services are impor- sectors, albeit from an extremely low base tant employment generators, but they lack the (Muzzini and Aparicio 2013a, 2013b). dynamism to drive long-term development. Much of the region’s expansion in services Given that the demand for such services jobs has not, however, been in high-value- depends on local incomes, the overall health added, dynamic—not to mention tradable— of the nontradable services sector ultimately services sectors like information technology, depends on that of tradables (McCombie and but in small-scale, nontradable services less Thirlwall 1994). The same is as true for sub- likely to benefit from dynamic agglomeration national urban economies as it is for national economies (Muzzini and Aparicio 2013a, economies (North 1955; Rowthorn 2000). 2013b). The declining importance of manufactur- In this services-led urbanization, ing in many South Asian countries has a host South Asia bears some resemblance to of causes. Emerging research suggests that LEVERAGING URBANIZATION FOR PROSPERITY AND LIVABILITY 31

developing countries outside of East Asia FIGURE 1.7 Structural change and urbanization: Relation between more generally are suffering from a process agglomeration index and value added by nonagricultural activities, that Rodrik has dubbed “premature deindus- 2010 trialization” due to globalization and labor- saving technological progress in 100 manufacturing (Rodrik 2015) (box 1.4). But beyond that, the failure of policy to ade- 80 quately respond to the pressure exerted by rising urban populations on land, housing, infrastructure, and basic services may be an 60 important contributor. These congestion PAK IND MDV forces are therefore likely acting both as a BGD LKA brake on the exploitation of agglomeration 40 economies and as a drag on the region’s abil- AFG NPL ity to compete in world manufacturing mar- index (%) Agglomeration 20 kets, thereby contributing to slower progress in prosperity gains than otherwise would BTN occur. 0

40 60 80 100 Livability of South Asia’s cities Value added by nonagricultural activities (% of GDP)

Despite the strong prosperity gains that the Source: Calculations based on World Bank World Development Indicators data. region has enjoyed since the turn of the cen- Note: AFG = Afghanistan; BGD = Bangladesh; BTN = Bhutan; IND = India; LKA = Sri Lanka; MDV = Maldives; NPL = Nepal; PAK = Pakistan. tury, the majority of South Asia’s cities endure high poverty, extremely poor housing conditions, and poor livability for many of their inhabitants. For the ﬁ ve most populous coun- FIGURE 1.8 Annual growth rates of shares of GDP generated by tries, the number of urban dwellers below the nonagricultural and manufacturing activities, 2000–10 national poverty line ranges from about one in eight in Pakistan to more than one in four 3 in Afghanistan (ﬁ gure 1.9). In Sri Lanka and Bhutan, extreme urban 2 poverty has been largely eradicated. Sri Lanka cut urban poverty particularly quickly, from 1 7.9 percent in 2002 to about 2 percent in 2013. It also fares well on the relatively small 0 share of its urban population living in slums. So, by regional standards, on these two indi- –1 cators at least, Sri Lanka has cities that are –2 relatively prosperous and livable. More generally, however, the share of the –3 urban population living in slums across the five other South Asian countries for which Annual growth rate of shares GDP (%) –4 data are available is high, ranging from 17.1 percent in India to 88.6 percent in –5 Afghanistan.11,12 The estimate for Afghanistan AFG BTN LKA BGD IND NPL MDV PAK is rather old, dating to 2005. Even so, analy- Nonagricultural Manufacturing sis of very high resolution satellite imagery for Source: Calculations based on World Bank World Development Indicators data. Kabul—where 54 percent of the country’s Note: AFG = Afghanistan; BGD = Bangladesh; BTN = Bhutan; IND = India; LKA = Sri Lanka; urban population lives—suggests that the vast MDV = Maldives; NPL = Nepal; PAK = Pakistan. 32 LEVERAGING URBANIZATION IN SOUTH ASIA

BOX 1.3 Diff erent trajectories—Urbanization, economic growth, and manufacturing exports in China and South Asiaa

In 1990, average GDP per capita in fi ve South of the fi ve, while the gap in urban shares widened Asian countries—Bangladesh, India, Nepal, to 25 percentage points. Pakistan, and Sri Lanka—was slightly higher The key to China’s dynamic urbanization and than China’s, while their average urban share (per growth during 1990–2011 was the strengthening UN World Urbanization Prospects data) lagged of its position as the “workshop of the world,” by 6 percentage points (panels a and b of fi gure in a process promoted by economic reform and B1.3.1). Since then, China’s urbanization and by the central government deliberately allowing economic growth have both accelerated beyond economic activity to concentrate in urban areas those of these fi ve economies. By 2011, China’s along the eastern seaboard. GDP per capita reached $10,041 (in 2011 con- China’s success is refl ected in the dramatic stant international prices)—2.2 times the average increase in its manufacturing exports (as a share

FIGURE B1.3.1 Economic growth, urbanization, and manufacturing exports in South Asia versus China

a. Real GDP per capita (2011 constant international dollars) b. Urban population share 12,060 60

10,060 50 8,060 40 6,060

Percent 30 4,060 GDP per capita GDP per 2,060 20

60 10

1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 2012 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 2012

c. Manufacturing share of total merchandise exports 100 95 90 85 80

Percent 75 70 65 60

1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 2012 China South Asia average

Sources: Calculations based on UN World Urbanization Prospects: 2011 Revision and World Bank World Development Indicators data.

(continues next page) LEVERAGING URBANIZATION FOR PROSPERITY AND LIVABILITY 33

BOX 1.3 Diff erent trajectories—Urbanization, economic growth, and manufacturing exports in China and South Asiaa (continued)

of its total merchandise exports) from 72 percent The share stabilized in 2008, but it was 69 percent in 1990 to 94 percent in 2011 (ﬁ gure B1.3.1, in 2011, below its starting point in 1990.

panel c). By contrast, the average share of manu- Sources: Calculations based on UN World Urbanization Prospects: 2011 Revision facturing exports for the ﬁ ve, although closely and World Bank World Development Indicators data. a. Afghanistan, Bhutan, and Maldives lack suffi cient historical data on GDP and following China’s path for the ﬁ rst few years of manufacturing exports to fi gure in the comparison with China undertaken in the 1990s, stagnated and then rapidly declined. this box.

BOX 1.4 Is early deindustrialization inevitable? Not if the advantages of cities can be brought into play

In 1966, the Cambridge economist Nicholas economies. In recent research, Dani Rodrik diag- Kaldor delivered a lecture on the causes of the nosed developing economies as being affl icted by slow rate of economic growth in the United “premature deindustrialization” (Rodrik 2015). Kingdom (Kaldor 1966). He sought to explain Whereas the employment share of manufactur- why the United Kingdom’s post–World War II ing in the United Kingdom peaked in 1961 at growth had been lagging behind that of its main somewhat more than 30 percent, with income rivals. per capita of about $14,000 (in 1990 dollars), He identifi ed the proximate cause as the today’s developing countries are seeing manu- economy’s relatively slow rate of manufactur- facturing peak at much lower shares of employing growth. Rapid manufacturing growth, he ment and levels of income. argued, was pivotal to rapid economic growth Rodrik suggests that premature deindustri- at the macro level, with manufacturing growth alization is largely a post-1990 phenomenon particularly contributing to economywide pro- attributable to the twin processes of global- ductivity growth both through static gains ization and labor-saving technological prog- associated with the reallocation of labor from ress. Increased automation of manufacturing lower-productivity agriculture and dynamic in developed countries has led to the sector’s gains associated with faster productivity growth. shrinkage in employment (but crucially not in Kaldor diagnosed the country as suffering from real output) in these countries, while driving “premature maturity” in which, because all down the relative price of manufacturing goods surplus labor from agriculture had long been globally. Developing countries would previously absorbed, manufacturing was losing strength have been isolated from this relative price trend, against other economies at similar incomes, stat- but globalization and lower barriers to inter- ing that the country had “exhausted its growth national trade have removed this protection. potential before attaining particularly high Manufacturing in developing countries has thus levels … of average per capita income” (Kaldor become less profi table, leading to deindustrial- 1966, 102). ization being “imported” by developing coun- Fifty years later, related concerns are expressed tries from the United States and other advanced about manufacturing in today’s developing economies.

(continues next page) 34 LEVERAGING URBANIZATION IN SOUTH ASIA

BOX 1.4 Is early deindustrialization inevitable? Not if the advantages of cities can be brought into play (continued)

Rodrik’s concerns about the repercussions of share of GDP during 2000–10 grew in Bangladesh premature deindustrialization for today’s devel- and in Sri Lanka, and in Sri Lanka’s case, despite oping countries are similar to those that Kaldor the country’s relatively high GDP per capita. voiced for the United Kingdom in the mid-1960s Yet avoiding the peril of premature deindus- and for the same reasons—that anemic manu- trialization is no easy feat, and requires even facturing growth will translate into slow over- faster productivity growth in manufacturing all macroeconomic growth. But his concerns relative to services than in a world without are all the more serious given that the United globalization. This condition suggests that the Kingdom’s deindustrialization, in hindsight, was need to address the congestion forces restrain- anything but premature. A far greater anxiety ing urban agglomeration economies (which can from a developing-country perspective is that enhance manufacturing productivity) is even the traditional route to rapid economic develop- more pressing. Higher-value-added tradable ser- ment through industrialization may have been vices also offer a potential route to faster eco- permanently closed. nomic growth, although their growth depends Does this mean that South Asia’s economies on high levels of human capital. Urban areas are inevitably conﬁ ned to a future of slow growth tend to provide greater returns and incentives to and economic development? Not necessarily. As human capital accumulation, which again sug- Rodrik shows, East Asian economies have largely gests a potentially crucial role for cities. escaped premature deindustrialization. Likewise, in South Asia, manufacturing value added as a Sources: Kaldor 1966; Rodrik 2015.

FIGURE 1.9 Prevalence of urban slums and poverty across South Asian countries

a. Estimated urban poverty rates b. Estimated shares of urban population living in slums 30 100

25 80

20 60

15 40 10 Urban poverty rate (%) Urban population in slums (%) 20 5

0 0 AFG BGD NPL IND PAK LKA BTN AFG BGD NPL PAK IND LKA

Sources: For urban poverty, calculations based on World Bank World Development Indicators data for the following years: Pakistan (2006); Bangladesh and Nepal (2010); Afghanistan (2011); Bhutan and India (2012); and Sri Lanka (2013). Data on Maldives’ urban poverty rate are missing. For the share in slums, calculations based on data for the following years: Afghanistan and Sri Lanka (2005); Bangladesh, Nepal, and Pakistan (2009); and India (2011). Data are from the following sources: Afghanistan and Sri Lanka (UNESCAP 2012, 126); Bangladesh, Nepal, and Pakistan (UN-HABITAT 2013, 126–28); and India (Census of India Offi ce of the Registrar General and Census Commissioner 2013). Note: AFG = Afghanistan; BGD = Bangladesh; BTN = Bhutan; IND = India; LKA = Sri Lanka; NPL = Nepal; PAK = Pakistan. LEVERAGING URBANIZATION FOR PROSPERITY AND LIVABILITY 35

majority of residential areas in the city con- with urban population growth (see tinue to exhibit slum-like characteristics chapter 5). (Zhou 2014). The failure of South Asia’s urban areas to The lack of decent, affordable housing for cope with the pressure of population is also large numbers of South Asia’s urban dwellers evident in the severe air pollution that afflicts directly curtails their welfare and has poten- the region’s cities (box 1.5) and, more gener- tially adverse implications for health out- ally, in the poor performance of the region’s comes and for female labor force largest cities—those where access to basic participation.13 And the prevalence of slums urban services tends to be best and standards in South Asian cities reflects a failure, once of living highest—in international rankings of again, to manage congestion forces—in land cities’ livability. One of the most respected and housing markets especially—associated rankings is the livability index of the

BOX 1.5 More than dust in Delhi

Of all the sources of congestion forces associated Air pollution in cities is inﬂ uenced by many with the growth of cities, one of the most serious factors beyond the control of policy makers, for health and human welfare is ambient air pol- including climate (such as levels and variabil- lution from vehicle emissions and the burning of ity of rainfall and temperature) and geography fossil fuels by industry. Particularly harmful are (such as distance to the coast). Controlling for high concentrations of fine particulate matter, these factors, analysis of the data shows that, for especially that of 2.5 microns or less in diameter developing-country cities globally, annual mean

(PM2.5), which can penetrate deep into the lungs, concentrations of PM2.5 are positively and signif- increasing the likelihood of asthma, lung can- icantly correlated with city size and population cer, severe respiratory illness, and heart disease. density, where population density is measured Data released by the World Health Organization within a 20 kilometer radius of the city center.c (WHO) in May 2014 shows Delhi to have the Although these relationships are expected, they most polluted air of any city in the world, with appear to be stronger in South Asian cities for

an annual mean concentration of PM2.5 of population density than in other developing- 152.6 µg/m3, or more than 15 times greater than country cities (ﬁ gure B1.5.1). In developing- the WHO’s guideline value and high enough to country cities outside South Asia, a doubling of make Beijing’s air—known for its bad quality— population density is associated with a 24.2 per- a look comparatively clean. cent increase in PM2.5, but in South Asian cities, Although Delhi provides the most extreme the increase is 34.8 percent. example of foul air, detailed analysis of WHO What accounts for this uniquely strong data for ambient (outdoor) air pollution in cities impact in South Asian cities? It seems plausi- reveals that Delhi is far from unique in South ble that the answer lies somewhere in the rela- Asia in having dangerously high concentra- tionship between city population density, the

tions of PM2.5. Among a global sample of 381 number and spatial conﬁ guration of potential developing-country cities, 19 of the 20 with pollution sources within a city, and the volume the highest annual mean concentrations are in of pollution emitted by each source (although South Asia.b And the issue is not just in India— more research is needed). For example, given Karachi, Dhaka, and Kabul all feature in the the lack of availability of and access to clean top 20. public transport in India, one can speculate that,

(continues next page) 36 LEVERAGING URBANIZATION IN SOUTH ASIA

BOX 1.5 More than dust in Delhi (continued)

for any given increase in demand for mobility at particular times. Even Delhi implemented associated with a given increase in population a wide range of pollution control measures density, more air pollution will occur there than in the early 2000s to reverse, if for a short elsewhere. Similarly, given failures in planning, time, increasing air pollution. Under the direc- one can speculate that a given increase in popu- tion of the Indian Supreme Court, which was lation density is associated with a greater prob- responding to a public interest lawsuit, Delhi ability of trafﬁ c gridlock in South Asian versus converted its entire ﬂ eet of public buses from non–South Asian cities, again contributing to a diesel to compressed natural gas. Public taxis higher relative increase in air pollution. and three-wheeler rickshaws were also con- Tackling excessive air pollution in South verted from petrol to compressed natural gas. Asia’s cities requires policy responses—quickly. Delhi also moved polluting industries out In India, about 660 million people (more than of its city limits, improved fuel quality and half the national population) live in areas where vehicle technology, banned highly polluting

annual mean concentrations of PM2.5 exceed vehicles, and improved its vehicle inspection national guidelines—with important adverse and maintenance program (Bell and others consequences for health.d 2004). Other cities followed and implemented Indian cities have been known to battle air measures to curb pollution. While beneﬁ ting pollution quite successfully in some places and from nationally implemented improvements in fuel quality and vehicle technology, some FIGURE B1.5.1 Relationship between annual mean cities brought in bus rapid transit systems as

concentration of PM2.5 and city population density for a cleaner mode of public transport, and oth- 381 developing-country cities ers introduced liquid petroleum gas vehicles to replace more-polluting ones. 5 Despite these measures, most cities are losing the war. A new wave of pollution control initia- 4 tives is needed to stem the current crippling lev- )

2.5 els of air pollution. These measures will have to range from further improvements in fuel qual-

Ln(PM 3 ity and vehicle technology to greater access to public transport (curbing the dramatic increase in private passenger vehicles) and changes in 2 patterns of urban development that reduce the –5 0 5 10 need for transport. Indian cities will also need Ln(population density) to improve air quality monitoring to get a bet- South Asia Other ter handle on the extent of the problem and to invest in source apportionment studies to better Sources: Calculations based on analysis of World Health Organization ambient (outdoor) air pollution in cities 2014 data (http://www.who.int/phe understand pollution sources. / health_topics/outdoorair/databases/cities/en) and other sources of data on city population levels, densities, and climate; and geographical-related 3 determinants of air pollution levels. Full regression results available on request. a. Beijing’s annual mean PM2.5 rate in the WHO’s 2014 database is 55.6 μg/m . Note: City population density is measured as population density within a 20 b. The only non–South Asian city in the top 20 is Khoramabad in the Islamic 3 kilometer radius of a city’s center. Sample covers 381 developing-country (that Republic of Iran, with PM2.5 of 101.9 μg/m . is, non-OECD and non–other high income) cities, with 139 of the cities in South c. Measures of population density were constructed using LandScan-gridded Asia (121 are Indian cities). (Similar relationships hold when restricting attention population data. to a population density range that is the same for both South Asian and non– d. http://www.economist.com/news/asia/21642224-air-indians-breathe South Asian developing-country cities.) -dangerously-toxic-breathe-uneasy. LEVERAGING URBANIZATION FOR PROSPERITY AND LIVABILITY 37

Economist Intelligence Unit (EIU), which uses in four areas—education, health, safety, and a range of indicators to assess cities’ perfor- the environment. In each pairwise compari- mance along five dimensions: stability, health son, the South Asian city ranks as less livable care, culture and environment, education, and on the index and on each of the four index infrastructure. According to the EIU’s 2015 components (except safety, based on reported rankings, the highest placed of the six South homicide rates). Asian cities in the index is New Delhi, India, which ranks 110 (out of 140 cities in all). This Looking ahead—Leveraging means that New Delhi ranks not only behind urbanization for prosperity the developed-country cities of Australia, and livability North America, and Western Europe that inevitably dominate the rankings, but also South Asia has generally struggled to make behind such cities as Baku (Azerbaijan), the most of its urbanization, particularly with the Manila (the Philippines), and Tunis (Tunisia). forces of congestion, and it is precisely these Mumbai (India; 115), Kathmandu (Nepal; forces that undermine the livability of the 125), Colombo (Sri Lanka; 127), and Karachi region’s cities and make its urbanization (Pakistan; 135) follow New Delhi in the rank- messy and hidden. These same forces seem to ings, while Dhaka’s (Bangladesh) livability is be holding back the region’s urbanization ranked 139, surpassing only that of Damascus pace and its potential exploitation of power- (Syrian Arab Republic). The average ranking ful agglomeration economies for faster pros- of the six South Asian countries in the index is perity gains as they whittle away the region’s 125, compared with an average ranking of international manufacturing competitiveness, 103 for all non–South Asian developing- contributing to more services-led urbaniza- country cities in the index and 93 for all cities tion. If South Asia had dealt better with these in the developing countries of East Asia and congestion forces, economic theory suggests the Pacific (table 1.2). that its urbanization speed would have been Further evidence of the poor livability of closer to East Asia’s experience or that of South Asia’s cities comes from assessing them now-advanced countries more than a century against appropriate comparators from other ago. Such congestion forces put a brake on regions (figure 1.10). Each comparator city the concentration of people and of economic was selected to match its South Asian coun- activity. terpart as closely as possible across popula- The UN’s World Urbanization Prospects tion, city area, and population density. The project that South Asia will continue its recent livability index is constructed from indicators pace of urbanization for the foreseeable future.14 South Asia will be the second-fastest urbanizing region during 2011–30, but due TABLE 1.2 Ranking of South Asian cities in the Economist Intelligence Unit’s livability index more to projected slowing in East Asia, a natural consequence of that region’s transition to City Ranking higher incomes. In absolute terms, South New Delhi 110 Asia’s urbanization pace will add almost 302 Mumbai 115 million people to its towns and cities, inexo- Kathmandu 125 rably raising already severe congestion pres- Colombo 127 sures (figure 1.11). Karachi 135 Dhaka 139 South Asia’s policy makers face a choice between two paths. The first is to continue Developing-country averages with the same policies that have allowed con- South Asia 125 All developing countries excluding South Asia 103 gestion pressures in urban areas to mount East Asia and Pacifi c 93 rapidly, thereby undermining proper exploita-

Source: EIU 2015. Data reused by permission of The Economist Intelligence tion of agglomeration economies and leaving Unit. Permission required for further reuse. the region on the current trajectory of 38 LEVERAGING URBANIZATION IN SOUTH ASIA

FIGURE 1.10 Livability in four major South Asian cities and comparator cities

Livability Livability Index Index 100 100 80 80 60 60 40 40 Education Health Education Health 20 20 0 0

Safety Environment Safety Environment Delhi, India Istanbul, Turkey Karachi, Pakistan São Paulo, Brazil

Livability Livability Index Index 100 100 80 80 60 60 40 40 Education Health Education Health 20 20 0 0

Safety Environment Safety Environment Kathmandu, Nepal Medellín, Colombia Dhaka, Bangladesh Bogotá, Colombia

Sources: See web-based annex 1B and Amirtahmasebi and Kim (2014) for full details.

FIGURE 1.11 Projected growth of urban population, 2011–30

a. Growth of urban share (%) b. Growth of absolute urban population 1.4 450

1.2 400 350 1.0 300 0.8 250 0.6 200 150 0.4 100 Growth of urban share (%) Absolute increase (millions) 0.2 50 0 0 EAP SAR SSA MENA ECA LAC EAP SAR SSA LAC MENA ECA

Source: Calculations based on UN World Urbanization Prospects: 2011 Revision data. Note: EAP = East Asia and Pacifi c; ECA = Europe and Central Asia; LAC = Latin America and the Caribbean; MENA = Middle East and North Africa; SAR = South Asia; SSA = Sub-Saharan Africa. LEVERAGING URBANIZATION FOR PROSPERITY AND LIVABILITY 39

underleveraged urbanization, structural OECD [2012] for a description of methods change, and development. This path also risks underlying this database). Both of these initia- locking in the worst of the region’s urban tives are based, in part, on the identifi cation problems. of built-up urban area from satellite imagery; The second is to undertake difficult policy however, they lack the comprehensive global coverage of the AI. reforms that help alleviate current and future 5. The exact estimate of the urban share pro- congestion pressures and that allow countries vided by the AI depends on the source of to better exploit agglomeration economies. global gridded population data used in its This second path—based on enhanced lever- construction. The use of Global Rural-Urban aging of urbanization—offers the potential to Mapping Project (GRUMP) data tends to improve South Asia’s development trajectory provide lower estimates than the use of so that its countries can follow, better and LandScan data. Therefore, the AI estimates faster, other countries that have transitioned obtained from these two different data to upper-middle and high incomes. sources are averaged. 6. Defi nitions are from United Nations, Department of Economic and Social Affairs, Notes Population Division, World Urbanization Prospects, the 2014 revision, Excel spread- 1. This number includes people living in India’s sheet titled “Data sources and statistical con- census towns, which are settlements that cepts for estimating the urban population” the Indian census recognizes as urban even (http://esa.un.org/unpd/wup/DataSources/). though they are governed as rural areas. To 7. In general, estimates of growth rates of qualify as a census town, an administratively urban shares and populations based on offi - rural settlement must meet the following cial national statistics will be less biased than three criteria: population in excess of 5,000 estimates of levels of urban shares and popu- persons, population density greater than lations, provided the bias associated with 400 people per square kilometer, and at least estimating levels has remained reasonably 75 percent of male main workers involved constant over time (World Bank 2008). in nonagricultural pursuits. (See chapter 2— 8. As chapter 2 discusses, there are, of course, particularly box 2.3—for more discussion important variations in the pace of urban- of the rapid proliferation of census towns in ization across countries in the region. Thus, India since 2001.) the pace of urbanization in both Bangladesh 2. The statistics relating to GDP per capita and and Nepal has been more on par with today’s poverty in this paragraph are based on World developed countries in the late 19th century, Bank World Development Indicators data. although they still fall short of the pace in GDP per capita fi gures are expressed in 2011 the East Asia and Pacifi c region since 2000. constant international dollars using purchas- Only Bhutan and Maldives have matched ing power parity (PPP) exchange rates. The this region’s pace. $1.25 a day poverty line is based on 2005 9. South Asia’s performance based on table 1.1 international prices. Figures on the abso- and fi gure 1.6 may be overstated if one con- lute increase in urban population are from siders that the shares of global urban popu- the United Nations’ World Urbanization lation on which these are based are derived Prospects Database: 2011 Revision (http:// from offi cial national estimates of urban esa.un.org/unpd/wup/). population, which tend to understate urban 3. This is, again, despite the inclusion of census shares in South Asian countries relative to towns in India’s offi cial estimates of the share non–South Asian countries. of its population living in urban areas (see 10. One caveat to this conclusion is that the offi - also note 1 and chapter 2, box 2.3). cial data on which fi gure 1.8 are based are 4. Alternative initiatives aimed at providing a unlikely to fully capture either the informal consistent defi nition of urban areas across sector or illegal activities. Although national countries include the e-Geopolis project (http:// statistical offi ces in the region do use methods www.e-geopolis.eu/spip.php?rubrique67) and to help estimate the contribution of the infor- the OECD’s metropolitan areas database (see mal sector to GDP, these methods suffer from 40 LEVERAGING URBANIZATION IN SOUTH ASIA

a number of shortcomings. In India, the ser- guage=ENG&source=ulg&format=PDF&docu vices sector is most prone to underestimation ment_id=1019689371&serialid=mdiZYWg% of its contribution as a result of informality 2FdX1ecMlIQLLr4eMqzR3oMhwWG2gYnB (Credit Suisse 2013). This suggests that, for PsBuA%3D. India at least, the conclusion of services-led D’Costa, A. P. 2011. “Geography, Uneven urbanization is unlikely to be an artifact of Development and Distributive Justice: The the data. Political Economy of IT Growth in India.” 11. The most recently available slum popula- Cambridge Journal of Regions, Economy and tion data available for Bhutan are for 2000, Society 4 (2): 237–51. and no data are available for Maldives. de Vries, G., M. Timmer, and K. de Vries. 2013. Both countries are therefore excluded from “Structural Transformation in Africa: Static ﬁ gure 1.9. Gains, Dynamic Losses.” GGDC Research 12. As discussed in chapter 5, there are con- Memorandum 136, Groningen Growth cerns that India’s slum population is seri- and Development Center, Groningen, The ously underestimated by the Census of India. Netherlands. UN-HABITAT (2013) estimates India’s slum EIU (Economist Intelligence Unit). 2015. population in 2009 was 29.4 percent of the “Livability Ranking and Overview: May urban population. 2015.” https://store.eiu.com/product 13. For quasi-experimental evidence of these .aspx?pid=475217632. adverse effects, see Field (2007). Fafchamps, M., and F. Shilpi. 2005. “Cities and 14. Swerts, Pumain, and Denis (2014) provide Specialisation: Evidence from South Asia.” a similar projection for India based on the Economic Journal 115 (503): 477–504. extrapolation of growth trends of urban Field, E. 2007. “Entitled to Work: Urban Property areas observed over 1961–2011 in e- Geopolis Rights and the Labor Supply in Peru.” data. Quarterly Journal of Economics 122 (4): 1561–602. References Glaeser, E. 2011. Triumph of the City: How Our Greatest Invention Makes Us Richer, Smarter, Amirtahmasebi, R., and S. Kim. 2014. “Beyond Greener, Healthier, and Happier. New York: Economic Growth: Livability Performance in Penguin Press. SAR Cities.” Background paper for this report, Henderson, J. V., and H. G. Wang. 2005. “Aspects World Bank, Washington, DC. of the Rural-Urban Transformation of Bairoch, P., and G. Goertz. 1986. “Factors of Countries.” Journal of Economic Geography 5 Urbanization in the Nineteenth Century (1): 23–42. Developed Countries: A Descriptive and HUPA (Ministry of Housing and Urban Poverty Econometric Analysis.” Urban Studies 23 (4): Alleviation). 2008. Report of the Committee on 285–305. Slum Statistics/Census. New Delhi: Government Bell, R. G., K. Mathur, U. Narain, and D. Simpson. of India. http://mhupa.gov.in/w_new/slum 2004. “Clearing the Air: How Delhi Broke the _report_nbo.pdf. Logjam on Air Quality Reform.” Environment Kaldor, N. 1966. Causes of the Slow Rate of 46 (3): 22–39. Economic Growth of the United Kingdom: Cali, M., and C. Menon. 2013. “Does An Inaugural Lecture. London: Cambridge Urbanization Affect Rural Poverty? Evidence University Press. from Indian Districts.” World Bank Economic Lewis, W. A. 1954. “Economic Development with Review 27 (2): 171–201. Unlimited Supplies of Labour.” The Manchester Chomitz, K., P. Buys, and T. S. Thomas. 2005. School 22 (2): 139–91. “Quantifying the Rural-Urban Gradient in McCombie, J. S. L., and A. P. Thirlwall. 1994. Latin America and the Caribbean.” World Economic Growth and the Balance-of- Bank Policy Research Working Paper 3634, Payments Constraint. London: Macmillan. World Bank, Washington, DC. Michaels, G., F. Rauch, and S. J. Redding. 2012. Credit Suisse. 2013. “India Market Strategy.” “Urbanization and Structural Transformation.” Equity Research report, July 9. https://doc Quarterly Journal of Economics 127 (2): .research-and-analytics.csfb.com/docView?lan 535–86. LEVERAGING URBANIZATION FOR PROSPERITY AND LIVABILITY 41

Muzzini, E., and G. Aparicio. 2013a. Bangladesh: Swerts, E., D. Pumain, and E. Denis. 2014. “The The Path to Middle-Income Status from an Urban Future of India’s Urbanization.” Futures 56: Perspective. Washington, DC: World Bank. 43–52. ———. Urban Growth and Spatial Transition in Uchida, H., and A. Nelson. 2008. “Agglomeration Nepal: An Initial Assessment. Washington, DC: Index: Towards a New Measure of Urban World Bank. Concentration.” 2009 World Development North, D. C. 1955. “Location Theory and Report Background Paper, World Bank, Regional Economic Growth.” Journal of Washington, DC. Political Economy 63 (3): 243–58. UN (United Nations). 2012. World OECD (Organisation for Economic Co-operation Urbanization Prospects, the 2011 Revision. and Development). 2012. Redeﬁ ning “Urban”: New York: UN Department of Economic A New Way to Measure Metropolitan Areas. and Social Affairs. Paris: OECD Publishing. UNESCAP (United Nations Economic and Social Office of the Registrar General and Census Commission for Asia and the Pacific). 2012. Commissioner. 2013. Primary Census Abstract The State of Asian Cities 2010/2011. Bangkok: for Slum. New Delhi: Government of India. UNESCAP. http://www.censusindia.gov.in/2011-Documents UN-HABITAT (United Nations Human /Slum-26-09-13.pdf. Settlements Programme). 2013. Streets as Roberts, M. 2015. “Assessing India’s Pace Public Spaces and Drivers of Urban Prosperity. of Urbanization: A Robustness Analysis.” Nairobi: UN-HABITAT. Background paper for this report, World World Bank. 2008. World Development Report Bank, Washington, DC. 2009: Reshaping Economic Geography. Rodrik, D. 2015. “Premature Deindustrialization.” Washington, DC: World Bank. IAS School of Social Science Economics World Bank and IMF. 2013. Global Monitoring Working Paper 107, Princeton University, Report 2013: Rural-Urban Dynamics and the Princeton, NJ. Millennium Development Goals. Washington, Rowthorn, R. E. 2000. “Kalecki Centenary DC: World Bank. Lecture: The Political Economy of Full Zhou, N. 2014. “Task Report: Land Use Employment in Modern Britain.” Oxford Mapping and Analysis with VHR Images.” Bulletin of Economics and Statistics 62 (2): Background paper for this report, World 139–73. Bank, Washington, DC.

Spatial Patterns of Subnational Performance and Urban Growth 2

Key messages

South Asia’s cities exhibit a wide range of out- reclassification of rural settlements, not by comes, and patterns of urban growth vary greatly rural-urban migration. At the same time, across countries and subnational areas. To many settlements with urban characteristics inform policy for better outcomes, this chapter are governed as rural areas. introduces a new metric of subnational perfor- • Many cities have been rapidly expanding mance—the prosperity index—and provides a beyond their administrative boundaries as the more in-depth analysis of national and subna- pace of urban expansion has outstripped tional patterns of urbanization, urban expansion, urban population growth. Related to this, and local economic growth. multicity agglomerations have become • Driven mainly by superior productivity, the increasingly important. most populous cities exhibit the strongest per- • Manufacturing, particularly in the formal sec- formance in generating prosperity. But size by tor, has been moving away from the cores of itself accounts for only a fraction of the varia- the region’s major agglomerations to their tion in performance, especially for poverty peripheries. Many city cores have stagnated, and economic growth. but some have retained their vibrancy by • Urban population growth has been engaging in higher-value-added tradable driven largely by natural increase and the services.

Introduction has been relatively slow in Afghanistan, India, Pakistan, and, for somewhat different South Asia is not realizing urbanization’s full reasons, Sri Lanka; it has progressed much potential to improve prosperity and livability faster elsewhere, especially in the smaller (see chapter 1). This overall picture, however, countries of Bhutan and Maldives. masks tremendous variation in productivity, Policy makers and others need a deeper poverty, and economic growth. Urbanization understanding of these variations to develop 43 44 LEVERAGING URBANIZATION IN SOUTH ASIA

differentiated policies for areas of varying areas is inherently difficult given the data patterns and outcomes. By providing impor- challenges that plague the region (see tant clues to the constraints on South Asia’s chapter 1, box 1.1). In particular, up-to-date cities, and the region’s urbanization more gen- subnational gross domestic product (GDP) erally, accurate analysis can help diagnose the data that would allow the construction of dimensions of the problem and devise appro- measures of both levels of productivity and priate responses. rates of economic growth (which provide an This chapter builds on the broad regional indication of subnational dynamism) at a and historical comparisons of chapter 1 reﬁ ned spatial scale are absent for much of to construct a more detailed analysis, the region.1 Furthermore, measures of pro- introducing a new metric of subnational ductivity and economic growth do not, by performance—the prosperity index. As its themselves, provide a full picture of an area’s name implies, this index captures dimensions success in generating prosperity for its inhab- of an area’s success in generating prosperity itants. Prosperity also depends, for example, for its inhabitants. It allows identification and on success in reducing poverty. quantification of differences in performance The prosperity index is a new measure of among subnational urban areas across coun- subnational performance that addresses the tries. The chapter also analyzes variations in above issues. It captures three dimensions of patterns of urbanization and urban population South Asian urban areas’ widely varying suc- growth across the eight South Asian countries, cess in generating prosperity: poverty, pro- as well as variations in population growth ductivity, and dynamism (table 2.1). To across both cities and subnational areas. To overcome the limited availability of subna- overcome data limitations (see chapter 1, box tional GDP data, it makes use of data on the 1.1), innovative “nighttime lights” data were intensity of nighttime lights to construct used to help construct the prosperity index, as proxy measures of both productivity and well as to document subnational patterns of dynamism (box 2.1). These data have the both urban expansion (Zhang and Seto 2011) advantage of being both consistently mea- and economic growth (Henderson, Storeygard, sured and available at a refined spatial scale and Weil 2011, 2012). for the entire region. The data also provide a good proxy measure of subnational GDP. Measuring diff erences in Meanwhile, by combining measures of pro- subnational performance ductivity and dynamism with a measure of poverty, the prosperity index also provides a Consistently measuring differences in subna- more robust and multidimensional measure tional performance across South Asian urban of performance than does a single indicator.

TABLE 2.1 The prosperity index captures three dimensions of performance

Dimension Rationale Indicator Poverty Inversely measures an area’s ability to generate widespread Percentage of population living on less than $1.25 prosperity and avoid extreme income deprivation. a day, 2010a (the World Bank’s global poverty line). Productivity A key determinant of prosperity. In accepted economic Intensity of nighttime light per square kilometer growth models, productivity determines an area’s of land, 2010, as a measure of the density of standard of living in the long term (Solow 1956; economic activity. Empirical research shows that Romer 1990). It also determines its long-term market measures of economic density correlate strongly competitiveness and thus its ability to achieve export-led with productivity (Ciccone and Hall 1996; Ciccone growth through trade (McCombie and Thirlwall 1994). 2002; Roberts and Goh 2011). Dynamism More rapid economic growth achieves greater progress Real GDP growth, 1999–2010, as estimated from in increasing prosperity over time. nighttime lights data during that period.

a. The poverty line is measured in 2005 constant international prices using purchasing power parity exchange rates. Poverty rates enter negatively into the construction of the prosperity index, so that larger values are associated with better poverty outcomes. SPATIAL PATTERNS OF SUBNATIONAL PERFORMANCE AND URBAN GROWTH 45

Given data limitations that prevent the data for areas at the third administrative building of the prosperity index at the city (Admin-3) level—the divisional secretariat level, it is, with the exception of Sri Lanka, (DS-division) level—were available. For brev- instead constructed for subnational areas at ity, these administrative units are referred to the second administrative (Admin-2) level. (In as “districts” regardless of local terminology. Bangladesh, India, and Pakistan, these units One advantage of using such district correspond to districts, while in Bhutan and data is that they allow the index to capture Nepal they correspond to gewogs and admin- the wider influence of urban areas on the dis- istrative zones, respectively.) For Sri Lanka, tricts in which they are located. This influence

BOX 2.1 Using nighttime lights data to infer patterns of economic activity and urban expansion

The nighttime lights data that underpin the pro- rates) at the subnational level (see, for exam- ductivity and dynamism indicators are derived ple, Alder 2015; Baum-Snow and Turner 2012; from “raw” lights data collected every night, typ- Pinkovskiy 2013). ically between 8:30 pm and 10 pm, local time, by The lights data are also used later in this satellites orbiting the poles. The satellites belong chapter to provide an overview of built-up to a program that dates back to the mid-1960s urban areas’ expansion patterns across South and were designed to collect meteorological data, Asia, extending a long line of previous research. particularly data on clouds illuminated by the This research has demonstrated that the lights moon, for making short-term cloud-cover fore- data can provide a suitable basis for monitor- casts (Doll 2008). ing how urban footprints expand at national It was realized in the late 1970s that the satel- or regional scales (see, for example, Zhang and lite sensors could also detect artiﬁ cial sources of Seto 2011). One concern about such use of lights light emanating from the earth’s surface, includ- data in South Asia is the increasing susceptibility ing city lights (Croft 1978). Since this discovery, of many of the region’s cities to power outages. a wealth of research has looked into using the But analysis suggests that the results are robust lights data to track patterns of human activity. to this concern.b Furthermore, the urban expan- Most notably, in the 21st century, economists sion picked up in the lights data is also seen in Vernon Henderson, Adam Storeygard, and David other higher-resolution satellite imagery for cit- Weil published work in the American Economic ies in the region (see “Rapid relative expansion Review showing that growth in intensity of a of urban footprints and the rise of the multicity country’s nighttime lights correlates strongly agglomeration” later in this chapter). with its GDP growth (Henderson, Storeygard, The nighttime lights data used in this chapter are and Weil 2011, 2012). This relationship occurs the Global Radiance Calibrated Nighttime Lights because consumption of nearly all goods in the product.c This differs from the product forming evening requires light, and as incomes increase, the basis of Henderson, Storeygard, and Weil’s so does light usage per person, in both consump- and most others’ research in that it does not suffer tion and investment activities.a Given this ﬁ nd- from “sensor saturation.” Whereas the standard ing, the economists argue that lights data can be product measures the average annual intensity of a proxy for GDP growth where GDP data are nighttime light over all cloud-free nights on a digi- either poor or missing. Their work is inspiring a tal number (DN) scale of 0–63, with higher values quickly expanding use of lights data to serve as indicating greater intensity, the product underly- a proxy for economic activity (levels and growth ing this chapter’s analysis measures light on a DN (continues next page) 46 LEVERAGING URBANIZATION IN SOUTH ASIA

BOX 2.1 Using nighttime lights data to infer patterns of economic activity and urban expansion (continued)

range of 0–1,500. This wider range allows varia- also helps mitigate the well-known “over-glow” tions in the luminosity of bright urban cores of or “blooming” problem, which would otherwise major cities such as Karachi, which are missed by cause urban footprints derived from the lights the standard product, to be detected. data to be signiﬁ cantly overstated (Doll 2008). To identify an urban area’s footprint, a DN Figure B2.1.2 shows the strong correlation threshold of 13 is used, selected by calibrating between the growth in the intensity of lights and the lights data against higher-resolution land- GDP for low- and middle-income countries in use cover maps, in particular, the European 1999–2010. This relationship underlies the deri- Space Agency’s GlobCover 2009 map and the vation of the subnational GDP growth estimates MODIS Collection 5 Land Cover Type mapd for South Asia used in the dynamism dimension for the region. A comparison of the lights data of the prosperity index.e with these maps reveals that nighttime light of less than DN 13 tends to be only rarely seen in FIGURE B2.1.2 Relationship between GDP growth the urban areas of South Asia but is typical for and growth of nighttime light intensity across agricultural areas (ﬁ gure B2.1.1). The use of a low- and middle-income countries, 1999–2010 calibrated threshold to demarcate urban areas 2.0 y = 0.2426x + 0.3587 1.5 R2 = 0.1728 FIGURE B2.1.1 Distribution of nighttime light 1.0 intensity between agricultural and urban areas in South Asia, 2009 0.5

0.0 0.35 Change in ln(GDP) 0.30 –0.5

0.25 –1 –2 –1 0 1 2 3 0.20 Change in ln(nighttime lights) 0.15

0.10 Sources: Calculations based on analysis of Defense Meteorological Satellite Program–Operational Linescan System nighttime lights data and GDP data from

Proportion of observations 0.05 the World Bank World Development Indicators. Note: GDP growth is measured in local currency units following Henderson, 0.00 Storeygard, and Weil (2011, 2012). 0 a. Measured nighttime light intensity refl ects outdoor and some indoor uses 0.6 1.2 1.8 2.4 3.0 3.6 4.2 4.8 5.4 6.0 6.6 More (Henderson, Storeygard, and Weil 2012, 999). Nighttime light intensity ln(DN value) b. In particular, in work undertaken in collaboration with the National Geophysical Data Center of the National Oceanic and Atmospheric Association, Agriculture Urban results were found to be robust to the “trimming out” of abnormally dimly lit nights, which could be driven by power outages, from the nighttime lights data. Sources: Calculations based on analysis of Defense Meteorological Satellite See web-based annex 2A for further details of this robustness test. Program–Operational Linescan System nighttime lights data and European c. The latest version of this product is available for download from http:// Space Agency GlobCover 2009 map land-use class data. ngdc.noaa.gov/eog/dmsp/download_radcal.html. The results reported in this Note: The fi gure shows frequency of diff erent average annual digital number chapter are based on an earlier (pre-general release) version of this product that (DN) values (on a natural logarithmic scale) in areas that may be considered to was supplied by the National Oceanic and Atmospheric Association’s National be urban and agricultural based on the European Space Agency’s GlobCover Centers for Environmental Information Earth Observation Group. 2009 map. Urban areas are taken to correspond to land-use class 190 (“Artifi cial d. http://due.esrin.esa.int/globcover/ and https://lpdaac.usgs.gov/products surfaces and associated areas [urban areas > 50 %]”) and agricultural areas /modis_products_table/mcd12q1. For more details on calibrating the lights to land use classes 11 (“Post-fl ooding or irrigated croplands [or aquatic]”), 14 data, see Roberts (2014). (“Rainfed croplands”), 20 (“Mosaic cropland [50–70%]/vegetation [grassland/ e. Based on a limited sample and excluding outliers, an even stronger correlation shrubland/forest; 20–50%]”), and 30 (“Mosaic vegetation [grassland/shrubland/ comes through between growth in the intensity of nighttime lights and GDP for forest; 50–70%]/cropland [20–50%]”) in the GlobCover map. The dotted line Indian districts during 1999–2010. Web-based annex 2B provides a more formal indicates DN = 13. analysis of the empirical relationship between nighttime light intensity and GDP. SPATIAL PATTERNS OF SUBNATIONAL PERFORMANCE AND URBAN GROWTH 47

depends, in part, on the strength of their links performance may differ systematically across with rural areas, particularly in less urbanized countries—due, for example, to differing mac- districts. In total, the index covers 699 dis- roeconomic conditions beyond the influence of tricts across six of South Asia’s eight countries local policy makers—the index is designed so (Afghanistan and Maldives are excluded that a score greater (or less) than 50 indicates because of a lack of poverty data). Districts that an area’s performance exceeds (or falls that lacked any evidence of urbanization (that below) the country average. Therefore, the is, no light intensity measured above a thresh- index provides a relative, not an absolute, old DN value of 13) were also excluded from measure of performance. Box 2.2 lists the steps the index. in constructing the prosperity index. To achieve the maximum score of 100 on Figure 2.1 illustrates the variation in per- the index, an area needs to perform the best in formance across districts. Weak performers South Asia on all three dimensions. Because were districts whose overall prosperity index

BOX 2.2 Constructing the prosperity index

The prosperity index is designed to make the standard deviations above (below) the aver- best use of limited subnational data to provide age. These scores represent extreme out- or a robust performance metric comparable across underperformance. South Asia. Its construction involves four steps: • Step 3—combine the transformed scores across the three components. A simple aver- • Step 1—measure “raw” performance on each age of the z-scores across the three associated of the three index components. These compo- indicators is calculated. The resulting score nents are the share of the population living on shows a district’s overall performance across less than $1.25 per day (poverty); intensity of the three components relative to the average nighttime light per square kilometer of land in for all districts in the country. 2010 (productivity);a and estimated real GDP • Step 4—rescale the combined scores to arrive growth rate during 1999–2010 (dynamism). at the final prosperity index. To make the • Step 2—transform the raw performance index more intuitive, it is rescaled such that a measures into comparable units across the district will score 50 if it performs exactly at three components by converting them into the country average on each of the three com- standardized z-scores. For each indicator, ponents; to achieve a score of 100 a district this step is achieved by subtracting the aver- must be the best performer on all three com- age performance across districts and dividing ponents (which in reality no district is). Scores through by the standard deviation. Because above (below) 50 imply performance above performance is likely to differ systematically (below) the country average. across countries in ways beyond the control of local policy makers in each country, the All districts lacking evidence of urbanization—in average performance and the standard devia- particular, lacking nighttime light intensity above tion are allowed to vary across countries. a DN of 13 (see box 2.1)—were excluded from Hence, for each indicator, the transformed the index. score takes on a value greater than (less than) zero if a district outperforms (underper- a. Performance on this indicator is measured in natural logs because the distribution of the intensity of nighttime light per square kilometer of land is forms) the average district performance in heavily right-skewed, meaning that there are a small number of districts with the country. A score greater than +2 (less extremely large values. Taking natural logs gives a more normal distribution, thereby facilitating comparability with the poverty and dynamism indicators than –2) on an indicator shows that a district and preventing results at the top end of the overall index from being unduly has performance more than (less than) two driven by the productivity indicator. 48 LEVERAGING URBANIZATION IN SOUTH ASIA

scores were less than or equal to 38, a perfor- distribution, within this group a further sub- mance that falls short of the average for the group of very strong performers can be identi- country by one standard deviation or more. fied.2 These districts have prosperity index Average performers have a score greater than scores exceeding their country averages by two 38 but less than 62. These districts have standard deviations or more, a score equal to scores that fall within one standard deviation or greater than 74. They represent extreme of their country averages. cases of positive relative performance. Strong performers have a score equal to or Table 2.2 provides examples of districts in greater than 62, and outperform their country each category.3 The very strong performers averages by one standard deviation or more. are the districts that are home to the region’s Given the overall shape of the performance largest cities, except for Gautam Budh Nagar, which is, nevertheless, part of India’s National Capital Region. Districts containing large cit- FIGURE 2.1 Performance distribution on the prosperity index ies are also evident in the strong category. By across all districts contrast, less urbanized districts containing smaller cities or towns appear more often in 50 the average and weak categories. Weak Average performers Strong Very The tendency of districts containing more 45 performers performers strong populous cities to perform better on the pros- 40 performers perity index is also seen in figure 2.2, which 35 shows a positive relationship between district 30 population (relative to the country average) 25 and performance.

Frequency 20 As can be seen from figure 2.3 (panel a), 15 this positive relationship is driven mainly by the productivity component. There is, there- 10 fore, a strong and significant positive relation- 5 ship between district population and 0 productivity. A similar, but much less strong, 26 29 32 35 38 41 44 47 50 53 56 59 62 65 68 71 74 77 80 More relationship exists between population and dynamism (figure 2.3, panel b). By contrast, Note: All 699 districts are included. there is no relationship between a district’s

TABLE 2.2 District performance on the prosperity index, by performance category

Performance Number of Category index range districts Example districts Very strong ≥ 74 9 Bangalore Urban (Karnataka, IND), Chennai (Tamil Nadu, IND), Delhi (Delhi, IND), Gautam Budh Nagar (UP, IND), Greater Bombay (Maharashtra, IND), Hyderabad (Telangana), Islamabad (ICT, PAK), Kolkata (W Bengal, IND), Sri Jayewardenepura Kotte (Colombo, LKA) Strong ≥ 62 47 Dhaka (Dhaka, BGD), Puducherry (Puducherry, IND), Thimbirigasyaya (Colombo, LKA), Bagmati (Central, NPL), Karachi (Sindh, PAK) Average > 38 and < 62 587 Karandeniya (Galle, LKA), Kushtia (Khulna, BGD), Kullu (Himachal Pradesh, IND), Quetta (Balochistan, PAK), Bheri (Mid-Western, NPL), Phuentsholing (Chhukha, BTN) Weak £ 38 56 Barwani (Madhya Pradesh, IND), Koralai Pattu North (Batticaloa, LKA), Bannu (KP, PAK), Sagarmatha (East, NPL)

Note: The Admin-1 (or, for Sri Lanka, Admin-2) unit to which a district belongs is shown in parentheses, along with country code. The Admin-1 (Admin-2 for Sri Lanka) level units are as follows: Bangladesh—divisions; Bhutan—dzongkhags; India—states; Nepal—development regions; Pakistan—provinces; and Sri Lanka—districts. BGD = Bangladesh; BTN = Bhutan; ICT = Islamabad Capital Territory; IND = India; KP = Khyber Pakhtunkhwa; LKA = Sri Lanka; NPL = Nepal; PAK = Pakistan; UP = Uttar Pradesh. Greater Bombay includes both Mumbai City and Mumbai Suburban districts of India, per the census of India’s 2011 administrative divisions. SPATIAL PATTERNS OF SUBNATIONAL PERFORMANCE AND URBAN GROWTH 49

FIGURE 2.2 Prosperity tends to be higher in more populous districts, but with signifi cant variation in performance

5 y = 0.1863x+0.0001 R2 = 0.0645 4

3 Bangalore, Islamabad, PAK Gautam Budh Kolkata, IND IND Nagar, IND Greater Colombo, Bombay, IND 2 LKA Puducherry, IND Delhi, Daman, IND IND Yanam, IND Karaikal, IND 1 Lahore PAK Thane, IND 0 Kinnuar, IND

Prosperity index (z-score) Prosperity Nuwara Eliya, LKA –1 Junagadh, IND Hardoi, IND Katharagama, LKA W. Nimar, IND –2

–3 –5.0 –4.5 –4.0–3.5 –3.0 –2.5–2.0 –1.5 –1.0–0.50.0 0.5 1.0 1.5 2.0 2.5 3.0 Population (z-score)

Note: Both the prosperity index and population are expressed as z-scores, where a value greater than (less than) zero indicates that a district has a prosperity index score (population) greater than (less than) the average for all districts in the same country. IND = India; LKA = Sri Lanka; PAK = Pakistan. population and its poverty rate (figure 2.3, a positive, albeit less strong, relationship panel c). between district size and dynamism suggests The positive relationship between the pro- the presence of dynamic agglomeration econ- ductivity component and district population omies arising from knowledge spillover is consistent with the theory that South Asian effects, which allow more heavily populated cities are characterized by potentially power- places to grow faster. ful agglomeration economies arising from However, although performance on the localization and urbanization economies (see prosperity index may be positively corre- “A framework for assessing urbanization and lated with population, wide variation can be city performance” in the Introduction). For seen around the fitted line (figure 2.2). Places developed countries, evidence of such econo- like Bangalore Urban and Puducherry mies is to be found in that a doubling of city (India), Islamabad (Pakistan), and Colombo size, for a wide range of city sizes, is associ- (Sri Lanka) strongly outperform expecta- ated with an increase in productivity of 3–8 tions based on population size alone, while percent (Rosenthal and Strange 2004). The districts like Nuwara Eliya (Sri Lanka) and empirical evidence for South Asia suggests West Nimar (India) show heavy underper- that agglomeration economies are stronger formance. These variations stem, in part, there than in developed countries (Chauvin, from the apparent lack of relationship Glaeser, and Tobio 2013). The fact that more between a district’s population and its per- populous districts exhibit higher levels of pro- formance on the poverty component. District ductivity is in keeping with the existence of poverty rates appear to be more or less inde- static agglomeration economies stemming pendent of district size. In short, while popu- from, for example, pooled labor markets and lation seems to matter for productivity and the availability of a wider variety of locally growth, it has no bearing on district poverty produced intermediate inputs. Similarly, outcomes.4 50 LEVERAGING URBANIZATION IN SOUTH ASIA

FIGURE 2.3 Relationship between district population and components of the prosperity index

a. Relationship between population and productivity b. Relationship between population and dynamism 6 6 y = 0.3998x–1E-07 y = 0.1205x–2E-06 5 R2 = 0.1596 4 R2 = 0.0145 4 3 2 2 0 1 0 –2

Productivity (z-score) –1 –4 –2 GDP growth, 1999–2010 (z-score) –3 –6 –6 –4 –2 0 2 4 –6 –4 –2 0 2 4 Population (z-score) Population (z-score)

c. Relationship between population and poverty 4 y = –0.0385x–2E-07 3 R2 = 0.0015

–1 Poverty rate (z-score) –2

–3 –6 –4 –2 0 2 4 Population (z-score)

Note: Poverty, productivity, GDP growth, and population are all expressed as z-scores, where a value greater than (less than) zero indicates that a district has an indicator value greater than (less than) the average for all districts in the same country.

Delving deeper into the relationships Gautam Budh Nagar despite its compara- between district performance on each of the tively much higher level of productivity. This three components reveals that they tend to be finding accords with the notion that, as a positively related. Districts that perform well group, the region’s major agglomerations are on the productivity indicator also tend to per- encountering diminishing returns due to, form well on the poverty and dynamism indi- most notably, congestion of infrastructure, cators.5 This tendency breaks down, however, basic services, and land and housing markets, when one looks only at the group of very which are overwhelming these cities’ agglom- strong performers, where the correlation eration economies. between productivity and dynamism becomes Figure 2.5 similarly shows a tendency for strongly negative (figure 2.4). For example, the districts containing some of the region’s Hyderabad, India, lacks dynamism relative to most populous cities to perform worse than SPATIAL PATTERNS OF SUBNATIONAL PERFORMANCE AND URBAN GROWTH 51

expected on dynamism given their overall FIGURE 2.4 Very strong performing districts: A strong negative prosperity index scores, notably Chennai, relationship between productivity and dynamism Delhi, Greater Bombay, Hyderabad, and Kolkata (all below the fitted line). But both 4.0 Y= –0.9159x + 5.0578 Islamabad 2 Colombo and Bangalore Urban perform as 3.5 R = 0.7338 expected on dynamism given their overall 3.0 Gautam Budh 2.5 Nagar scores. Gautam Budh Nagar, on the periph- Bangalore Urban 2.0 ery of Delhi, also exhibits more dynamism Sri Jayewardenepura Kotte than predicted by its overall score.6 As will 1.5 Kolkata 1.0 Delhi

be seen later in this chapter (see “Spatial Dynamism (z-score) Hyderabad Greater Bombay Chennai deconcentration of manufacturing”), these 0.5 differences in dynamism reflect differences in 0 the ability to retain vibrancy in core city 2.0 2.53.0 3.5 4.0 4.5 5.0 5.5 areas in the face of a process of manufactur- Productivity (z-score) ing suburbanization. Note: Both dynamism (as measured by the rate of real GDP growth, 1999–2010) and productivity The remainder of this chapter analyzes (as measured by the intensity of nighttime lights per square kilometer of land) are expressed as patterns of urbanization, urban expansion z-scores, where a positive value indicates that a district’s performance exceeds the average for all (including the expansion of urban footprints), districts in the same country. and local economic growth within South Asia, starting with cross-country differences in the level and pace of urbanization. These FIGURE 2.5 Prosperity score and dynamism by population patterns provide the backdrop to the differ- quartile

ences in performance on the prosperity index –5.5 described in this section. –4.5

–3.5 Gautum Budh Nagar Variations in the level and pace –2.5 Bangalore of urbanization across South Colombo –1.5 Kolkata Asian countries Delhi Hyderabad 0.5 Chennai South Asia remains at an incipient stage of Greater Bombay –0.5 urbanization, and its pace of urbanization during 2000–11 was slow relative to both the –1.5 GDP growth, 1999–2010 (z-score) GDP growth, East Asia and Paciﬁ c region and the historical –2.5

experiences of today’s developed countries. –3.5 y = 0.9162x–1E-06 R2 = 0.451 Even though ofﬁ cial statistics tend to under- –4.5 state the region’s level of urbanization— –2.5 –2.0 –1.5 –1.0 –0.5–0.0 0.5 1.0 1.5 2.0 2.5 3.0 indicating the existence of significant Prosperity index (z-score)

“hidden” urbanization—the share of its pop- Population ulation living in settlements with urban-like Below 25th percentile 25th–50th percentile characteristics remains low compared with 50th–75th percentile Top 75th percentile more developed regions of the world. Thus, Note: Both the prosperity index and rate of real GDP growth, 1999–2010, are expressed as z-scores, according to the agglomeration index (AI), where a value greater than (less than) zero indicates that a district has a prosperity index score (real the share of South Asia’s population living GDP growth rate) greater than (less than) the average for all districts in the same country. in urban areas in 2010 was 52.5 percent (see chapter 1). Because India accounts for 75 percent of the region’s population, this shares of Maldives and Pakistan (ﬁ gure 2.6). figure closely reflects conditions in that The other ﬁ ve countries all have urban shares country. India’s urban population share, as (according to the AI) of less than 50 percent, estimated using the AI, is 55.3 percent, which with the urban shares of Afghanistan, Nepal, is also very similar to the AI-estimated urban and especially Bhutan particularly low. 52 LEVERAGING URBANIZATION IN SOUTH ASIA

FIGURE 2.6 Share of the population classifi ed as were officially classified as urban alongside urban: Offi cial defi nitions and the agglomeration the country’s municipal and urban councils index, 2010 (World Bank 2012). The downgrading of these town councils to rural areas (pradeshiya 60 sabha) contributed to a fall in the country’s 50 official urban share from 21.5 percent in the 40 1981 census to 14.6 percent in the 2001 cen- 30 sus (see chapter 1, box 1.1).

Percent 20 In India, the extra population implied to be 10 living in unofficial urban settlements is in 0 addition to the growing share of the country’s MDV PAK IND LKA BGD AFG NPL BTN urban population living in “census towns.” AI UN These towns are settlements that the Census of India already recognizes as urban and that Note: AI indicates urban shares based on the agglomeration index are included in the official urban share esti- methodology; UN indicates estimates based on offi cial national defi nitions of urban areas as taken from UN World Urbanization Prospects: 2011 mates, even though they are administratively Revision. For India, the offi cial defi nition of “urban areas” includes census rural. Some of these census towns are large. towns. AFG = Afghanistan; BGD = Bangladesh; BTN = Bhutan; IND = India; LKA = Sri Lanka; MDV = Maldives; NPL = Nepal; PAK = Pakistan. According to the 2011 census, 20 census towns have populations greater than 100,000, the largest of which, Noida, has a population 9 Discrepancies between the offi cial and of 642,381, or larger than Sheffield, the fifth- AI fi gures for urban population shares largest “primary” urban area in the United Kingdom. One of the most striking results of All eight countries also show large urban- India’s 2011 census was the dramatic growth share discrepancies between the AI and that in the number of census towns since the previ- based on a country’s offi cial defi nition that ous census (box 2.3). comes from the United Nations’ World Bhutan is the exception with regard to the Urbanization Prospects: 2011 Revision data- underestimation of its urban share by official base. Other than for Bhutan, AI urban shares figures, being the only South Asian country exceed the offi cial fi gures, pointing to hidden with an AI urban share less than its official urbanization, given South Asia’s relatively urban share. This outcome largely reflects the strict offi cial defi nitions compared with other country’s very low population density, which global regions.7 in 2010 was just 18.7 people per square kilo- In Bangladesh, India, Maldives, Pakistan, meter. (The average population density across and Sri Lanka, the implied official underesti- South Asia that year was 432 people per mations are particularly wide, suggesting sim- square kilometer, more than 23 times as ilarly large hidden urbanization—in other high.)10 words, sizable portions of their populations are living in settlements that, although they The relatively slow pace of South Asian may exhibit urban characteristics, are gov- urbanization erned as rural areas. In the most extreme case of Sri Lanka, the difference between the The relatively slow regional pace of urban- AI-estimated urban share and the estimate ization (see chapter 1) is refl ected in country based on the country’s official definition of data for Afghanistan, India, Pakistan, and urban areas suggests that as much as one- Sri Lanka (fi gure 2.7).11 For Sri Lanka, offi - third of its entire population may be living in cial estimates indicate that the share of the unrecognized urban settlements.8 These population living in towns and cities actu- unrecognized urban settlements are likely to ally fell slightly between 2000 and 2010. include former town councils that, before Bangladesh and Nepal experienced faster Sri Lanka’s 1987 tightening of its definition, urbanization, with rates more on par with SPATIAL PATTERNS OF SUBNATIONAL PERFORMANCE AND URBAN GROWTH 53

BOX 2.3 India’s rapidly proliferating census towns

Census towns are those that the Registrar General sector. These criteria are more demanding than and Census Commissioner of India, which is those used by the agglomeration index to defi ne responsible for the country’s census operations, areas as “functionally” urban; therefore, the classifi es as urban even though they lack statu- aggregate population of the census towns (about tory status and are therefore formally governed as 53 million) is only a lower-bound estimate of the rural areas. One of India’s 2011 census fi ndings number of Indians living in urban-like areas that that has attracted particular attention is the rapid are governed as though they were rural. growth in the number of census towns since the 2001 census. India in 2011 had almost as many FIGURE B2.3.1 Number of Indian statutory and census towns as statutory ones (each either side census towns, 1981–2011 of 4,000; fi gure B2.3.1), and their overall share of the offi cial urban population increased from 5,000 7.4 percent in 2001 to about 14 percent in 2011. 4,041 4,000 3,799 The locations of the new census towns are quite 3,894 dispersed, with only 13.1 percent of their popula- 2,996 tion within 25 kilometers of India’s million-plus 3,000 2,758 population cities; only about one-third of all new census towns are close to a Class I town (a town 2,000

with a population in excess of 100,000). Number of towns 1,693 1,362 The three criteria that the Registrar General 1,000 1,271 uses to identify census towns are rather strin- 0 gent: the town must have a population of at least 1981 1991 2001 2011 5,000, a population density of at least 400 peo- Statutory towns Census towns ple per square kilometer, and at least 75 percent of its male workforce engaged in the nonfarm Source: Pradhan 2013.

today’s developed countries in the late 19th two are related, the distinction is important, century (see chapter 1, ﬁ gure 1.3). Yet they because a country could have fast urban pop- still fall short of the pace in the East Asia ulation growth without becoming more and Paciﬁ c region since 2000. Only Bhutan urbanized if urban and rural areas were and Maldives have matched the pace of this growing at largely similar rates. In this case, region. towns and cities will still have to provide services to an expanding number of people, even Urban population growth and though nationally society is not becoming its drivers more urbanized. In all South Asian countries, growth rates Whereas the pace of urbanization just of urban population exceed the correspond- discussed (measured by the growth rate of the ing growth rates of urban shares (figure 2.8 urban population share) indicates the speed versus figure 2.7), meaning that population at which a country is transitioning from a growth continues in rural areas. Sri Lanka rural to an urban society, the growth rate of aside, all countries’ urban populations grew its urban population measures the speed at faster than 2 percent a year, showing depar- which the absolute number of people residing tures from their growth rates of urban in towns and cities is growing. Although the share. For instance, Maldives has been the 54 LEVERAGING URBANIZATION IN SOUTH ASIA

FIGURE 2.7 Annual growth rate of urban share of population, Urban population growth stems mainly by country, 2000–10 from natural increase or reclassifi cation In addition to natural increase, urban popula- 4 tion growth can occur through the reclassiﬁ - 3 cation of settlements from rural to urban and through net rural-urban migration. Though 2 the tendency often is to think of urban population growth as being driven by rural-urban

Percent 1 migration, the majority of such growth in South Asia has actually been due to either 0 natural increase or reclassiﬁ cation. For example, 44 percent of the urban population –1 growth that India experienced between 2001 MDVBTN NPL BGD AFG IND PAK LKA and 2011 was due to natural increase and 29.5 percent to the reclassification of rural Source: Calculations based on UN World Urbanization Prospects: 2011 Revision data. Note: AFG = Afghanistan; BGD = Bangladesh; BTN = Bhutan; IND = India; LKA = Sri Lanka; settlements into census towns (see box 2.3; MDV = Maldives; NPL = Nepal; PAK = Pakistan. Pradhan 2013). Similarly, in Nepal migration contributed just less than one-third of total urban population growth during the 1990s FIGURE 2.8 Annual growth rate of urban population, by (though this rises to as much as 40 percent country, 2000–10 when considering only Kathmandu; Muzzini and Aparicio 2013a). For Pakistan, although 6 recent data are lacking, the contribution of 5 rural-urban migration to total urban population growth during 1981–98 was just 26 per- 4 cent (Karim and Nasar 2003).13, 14 3 China stands in stark contrast: 56 percent of its urban population growth between 2000 and

Percent 2 2010 was attributable to net rural-urban migra- 1 tion (World Bank and DRC 2014), despite its 0 hukou (household registration system), which sets powerful disincentives for rural residents –1 BTN MDVAFG NPL BGD PAK IND LKA to move to urban areas by denying them rights of access to public education and health Source: Calculations based on UN World Urbanization Prospects: 2011 Revision data. benefits (Chan and Buckingham 2008; Bosker, Note: AFG = Afghanistan; BGD = Bangladesh; BTN = Bhutan; IND = India; LKA = Sri Lanka; Deichmann, and Roberts 2015). MDV = Maldives; NPL = Nepal; PAK = Pakistan. This small migration contribution to urban population growth in South Asia suggests that the region’s towns and cities are relatively most quickly urbanizing country since the unattractive to would-be migrants from rural beginning of the century, but the rate of areas. Evidence on the determinants of rural- urban population growth has been slightly urban migration, including in South Asian higher in Bhutan. Likewise, Pakistan’s countries like Nepal (Shilpi, Sangraula, and Li urban population growth rate has been 2014) and Sri Lanka (World Bank 2011), has marginally faster than that of India even established that a city’s attractiveness for rural though India has been the more rapidly migrants depends not only on income-earning urbanizing. These differences reflect differ- opportunities but also on whether it can offer ences in underlying rates of national popu- better amenities, including basic infrastruc- lation growth.12 ture and services. But although average health SPATIAL PATTERNS OF SUBNATIONAL PERFORMANCE AND URBAN GROWTH 55

outcomes tend to be higher in South Asia’s impressive progress in achieving spatial equity urban areas than in its rural areas, for the between rural and urban areas in the provi- poorest in each of these areas the opposite sion of basic public services and living stan- generally holds (box 2.4). dards (World Bank 2011). For example, in The picture for Sri Lanka is somewhat dif- 2012, the literacy rate among the population ferent. As elsewhere in South Asia, cities ages 15–24 years was 98.6 percent in Sri appear to be unattractive to migrants, but in Lanka’s rural sector, compared with 98.9 per- Sri Lanka not so much because of poor urban cent in its urban sector.15 The country’s living conditions as because of the country’s Household Income and Expenditure Surveys

BOX 2.4 Urbanization and health in South Asia—A missed opportunity?

The experience of the Industrial Revolution in Better socioeconomic status in urban areas the 18th and 19th centuries has shaped modern translates into better nutrition and improved perceptions of urbanization and health because access to safe water and adequate sanita- the populations of rapidly growing cities suf- tion. It also results in better access to health fered worse health conditions than those in rural care, and (by contrast with health outcomes) areas owing to poor and overcrowded housing surveys measure higher services utilization and sanitation conditions. But by the early 20th across socioeconomic groups. For example, in century, as a result of better sanitation, advances all quintiles across the four countries, moth- in medical technology, and overall reduced urban ers resident in urban areas are more likely to poverty, urban-rural health disparities reversed. give birth in a health facility. For Pakistan, a Household survey data for urban Bangladesh, large part of this difference is driven by higher India, Nepal, and Pakistan reflect both these utilization of private health facilities, par- negative and positive scenarios, in a mixed health ticularly by better-off quintiles (fi gure B2.4.2, picture for the region. panel a). And in all four countries and at all Household surveys indicate that the nega- socioeconomic levels, utilization of public ser- tive urban health scenario applies to the poorest vices for delivery care in urban areas exceeds, in urban areas. Compared with rural popula- or is at a similar level to, that in rural areas tions, under-fi ve mortality was higher among (fi gure B2.4.2, panel b). urban households in India (2005–06) and Nepal What conclusions can be drawn? That better (2011) in the bottom socioeconomic quintile, in average health outcomes in urban areas refl ect Pakistan (2006–07) in the bottom two quintiles, greater prosperity among urban populations and in Bangladesh (2011) in all but the highest seems to refl ect an overall benefi cial impact of quintile.a urbanization. A more specifi c agglomeration Overall averages for health and nutrition are, benefi t may be a concentration of private sector however, generally better for urban than for health services that helps translate better eco- rural populations (in India, Nepal, and Pakistan; nomic conditions into better health outcomes. in Bangladesh urban averages are similar to Policy implications for government include rural, though still slightly better). These better addressing the economic burden of out-of- outcomes occur because the majority of urban pocket health spending by households—includ- populations are in higher socioeconomic quin- ing reducing the risk of impoverishment due to tiles, whereas the reverse is true for rural popu- health care costs—and improving the function- lations (fi gure B2.4.1). ing and quality of private services.

(continues next page) 56 LEVERAGING URBANIZATION IN SOUTH ASIA

BOX 2.4 Urbanization and health in South Asia—A missed opportunity? (continued)

FIGURE B2.4.1 Under-fi ve mortality rates across socioeconomic index quintiles

a. India, 2005–06 b. Bangladesh, 2011 120 90

80 100 70

80 60

50 60 40

Per 1,000 live births 40 Per 1,000 live births 30

20 20 10

0 0 01234560 123456 Urban Rural

Note: Quintiles are defi ned using an index that refl ects the relative socioeconomic status of households (based on asset ownership and other data), and are calculated using the entire sample (that is, both urban and rural households together). Data points are circles proportionate to the population in each quintile.

FIGURE B2.4.2 Percentage of births in health facilities across socioeconomic index quintiles

a. Pakistan, percentage of births in private health facilities, 2006–07 b. Nepal, percentage of births in public health facilities, 2011 70 70

60 60

50 50

40 40

30 30

Percentage of births 20 Percentage of births 20

10 10

0 0 0123456 0 123456

Urban Rural

BOX 2.4 Urbanization and health in South Asia—A missed opportunity? (continued)

The data also hint at a large, missed oppor- of basic health services (such as immunization tunity. Child mortality and malnutrition indi- and preventive care for chronic conditions) cators in particular show that South Asia’s and of other basic public services with health urban poor have health outcomes as bad as impacts, notably improved water supply and or worse than the rural poor. More extensive sanitation. and better targeted government action would help address such inequalities, while contrib- Sources: For Bangladesh, Nepal, and Pakistan, analysis was based on data from Demographic and Health Surveys for the following years: Bangladesh, 2011; uting to a social safety net that would pro- Nepal, 2011; and Pakistan, 2006–07. For India, analysis was based on 2005–06 vide poor households with better access to the National Family Health Survey data. a. This pattern is even more evident for chronic malnutrition among under-fi ve beneﬁ ts of agglomeration. Such action would children in Bangladesh and India, the prevalence of which in both countries was take the form of ensuring adequate provision higher in urban areas for all but the highest socioeconomic quintile.

likewise reveal a dramatic convergence of FIGURE 2.9 Convergence of poverty rates across Sri Lankan poverty rates across its districts during 2002– districts, 2002–09 09 (figure 2.9). Thus, in 2002 a district’s urbanization was significantly negatively 35 related to its headcount poverty rate; by 2009 this relationship had disappeared. This find- 30 ing suggests that the impetus for “push” 25 migration—rural residents moving to a city more out of distress than for, say, higher 20 wages—is generally weak in Sri Lanka, which 15 may help explain the stability of the country’s urban population share. 10 Headcount poverty index Many towns and cities in South Asia 5 (Sri Lanka aside) are unattractive destinations for rural migrants because of the disappoint- ing conditions they offer. This view is consis- 0 10 20 30 40 50 tent with the more general picture of cities Urban population share (%) struggling with congestion forces emanating 2009 2002 from the pressure of population on infrastruc- Source: Based on analysis of data from the 2002 and 2009 Household Income and Expenditure ture, basic services, land, housing, and the Surveys for Sri Lanka. environment. Yet urban areas’ populations Note: Urban shares of the population for each district are calculated by aggregating the populations of municipal and urban councils. The poverty data for 2002 are matched with urban share data have still been growing, in part reflecting rela- for 2001, while the poverty data for 2009 are matched with urban share data for 2012. Analysis is tively high urban fertility (natural increase). restricted to the 15 (out of 25) districts for which data were available. Emerging academic research suggests that a self-reinforcing loop between such fertility and excessive congestion costs may mean contribution of rural-urban migration to that the region’s megacities are stuck in a urban population growth in South Asia. Such Malthusian trap from which they will find it costs include not only the monetary costs of hard to escape without decisive policy actions moving but also nonmonetary costs from, for (box 2.5). example, differences in culture between origin High costs of migration may be another and destination and leaving the family behind. potential factor in the relatively small Although countries in the region do not 58 LEVERAGING URBANIZATION IN SOUTH ASIA

BOX 2.5 The rise of the South Asian megacity and the Malthusian trap

South Asia is home to 6 of the world’s 29 mega- agglomeration economies, absolute increases cities (that is, those with populations of at least in population since World War II have been so 10 million): Bangalore, Delhi, Dhaka, Karachi, great that these agglomeration economies have Kolkata, and Mumbai. Only six megacities— been overwhelmed by urban congestion forces, London, Los Angeles, New York, Osaka, Paris, stifl ing growth of productivity and real wages and Tokyo—are now in developed countries. through a variety of mechanisms (for example, But even as recently as 1950 New York claimed lost productive time of workers due to excessive the title of the world’s largest city, with Tokyo, commute times, poor human capital outcomes London, and Paris all immediately behind. The owing to crowding of educational services, and rise of the South Asian and developing-country the Malthusian force of excessive labor supply megacity is a largely post–World War II phenom- growth) and contributing to continued high enon. What accounts for this rise? And what dif- rates of natural population increase by slowing ferentiates it from the rise of earlier megacities of the demographic transition. Developed-world the developed world? megacities, by contrast, because they increased In newly emerging research, academic econo- their populations over a longer period, avoided mists Remi Jedwab and Dietrich Vollrath (2015) being overwhelmed by congestion costs, allow- argue that—in keeping with the general fi nding ing for positive real wage growth, which helped of this chapter that urban population growth has create lower crude birth rates. By the time vac- been largely a matter of natural increase—the rapid cines and new medicines became available, the growth of developing-country megacities during megacities of today’s developed countries had the last 60–70 years has also been mainly driven already made the demographic transition (from by natural population increase. A post–World War high birth and death rates to low birth and death II acceleration in the natural rate of population rates) and avoided the Malthusian trap. growth in developing countries was attributable What are the policy implications of this hypoth- to a sharp drop in crude death rates, caused by the esis? One is that South Asian megacities could drive to make vaccines and new treatments (such escape the trap if policies produce a large enough as antibiotics) available globally after the 1940s. drop in congestion costs to allow an upward The sharp drop in death rates allowed develop- jump in wages and incomes, thereby slowing birth ing-country megacities to grow in absolute size at rates. This outcome could potentially be achieved rates that had not previously been possible. through, for example, large-scale new town devel- Much of the growth of the developed-world opment on greenfi eld sites aimed at slowing, or megacities took place over centuries rather even reversing, population growth in the megaci- than decades, and with a large portion of their ties. But the historical experience with new town growth before the early 20th century (and before development has been mixed, with many new the arrival of vaccines and modern medicines). towns failing to reach planned population targets Their growth occurred against a backdrop of (World Bank 2008). Alternatively, investments in high crude death rates and relatively slow rates congestion-reducing technologies like improved of natural population growth. In stark contrast inter- and intracity transport systems and with the average developing-country megacity, improved basic services, with reforms to alleviate their growth was driven primarily by rural-urban congestion in land, housing, and labor markets, migration and accompanying urbanization. could, on a suffi cient scale, succeed. Achieving Jedwab and Vollrath (2015) hypothesize these outcomes will require fundamental reforms that developing-county megacities are stuck aimed at improving governance and fi nancing of in a self-reinforcing Malthusian trap that rein- South Asian megacities (see chapter 3). forces their relative lack of prosperity and livability. Although such megacities benefi t from Source: Based on Jedwab and Vollrath 2015. SPATIAL PATTERNS OF SUBNATIONAL PERFORMANCE AND URBAN GROWTH 59

impose the sort of formal restrictions on and cities of different sizes for four of South migration that exist in China, geographic Asia’s eight countries—Bangladesh, India, labor mobility appears to be generally lack- Nepal, and Sri Lanka—in 2001–11.16 ing; for example, whereas 9 percent of people in the United States lived in a different state Relationship between an urban five years ago and 40 percent were born in a settlement’s initial population and its different state, the equivalent figures for subsequent growth rate India, according to 2001 census data, were just 0.4 percent and 3.6 percent (Glaeser, The main ﬁ nding that emerges from this anal- Chauvin, and Tobio 2011). ysis is that, for both Bangladesh and India, no A lack of labor mobility such as this can relationship can be discerned between an prevent the relaxation of congestion pres- urban settlement’s initial population and its sures in urban labor markets in which subsequent growth rate. Nepal and Sri Lanka demand is bidding up wages. It also hinders display more evidence of a relationship. In the widespread spillover of urban agglom- Nepal larger cities tend to grow faster than eration benefits to rural areas—by increas- smaller ones, while for Sri Lanka, medium- ing labor supply in urban areas and reducing sized cities, led by Vavuniya, Batticaloa, and it in rural areas, rural-urban migration can Kattankudy, show some tendency to grow stimulate rural-urban wage convergence. faster (ﬁ gure 2.10). Empirical evidence suggests that this was a The relationships observed for Nepal and historically important force in promoting Sri Lanka might be explained by their levels rural-urban integration during the urban- of economic development. Whereas Nepal ization of today’s developed countries remains a low-income country, with GDP per (World Bank 2008). capita of $2,173 in 2013, Sri Lanka qualifies as a lower-middle-income country, with GDP 17 Population growth across urban per capita more than four times as high. As hierarchies and geographic space the World Bank’s World Development Report 2009 notes, at low levels of GDP per capita, Drilling down to data on individual urban population tends to spatially concentrate, but settlements allows the analysis to consider as countries move to middle-income status patterns of population growth across towns population growth increasingly spreads to

FIGURE 2.10 City population growth rate and initial population

a. Nepal, 2001–11 b. Sri Lanka, 2001–12 y = 0.5545x–2.9018 y = –0.1554x2+ 3.374x–17.267 7.0 4.0 R2 = 0.0635 6.0 3.5 R2 = 0.0793 5.0 3.0 4.0 2.5 3.0 2.0 2.0 1.5 Growth rate (%) 1.0 Growth rate (%) 1.0 0.0 0.5 –1.0 0.0 91113 91113 Ln(2001 population) Ln(2001 population)

Source: Based on data extracted from City Population (http://www.citypopulation.de). Note: For Sri Lanka, 2012 population levels were derived by extrapolating the latest available estimates (generally from 2007) using district population growth rates in 2001–12; district population growth rates were calculated using census data. These district population growth rates were adjusted to account for a change in the method of enumerating the population between the 2001 and 2012 censuses—from the de facto method used in 2001 to the de jure method used in 2012 (see Sri Lanka, Department of Census and Statistics [2012] for further details). 60 LEVERAGING URBANIZATION IN SOUTH ASIA

secondary urban centers (World Bank 2008). a “law,” namely, Gibrat’s law (Gabaix and However, even for Nepal and Sri Lanka, dif- Ioannides 2004). ferences in city size account for only a small The emerging policy lesson is that South proportion of the overall observed differ- Asian countries need to learn to live with their ences in population growth rates across cities existing urban hierarchies. The megacities of (figure 2.10).18 today will remain the megacities of tomor- From a population perspective, towns row, and the same is likely to be true of large, and cities in South Asia show a strong ten- medium, and small urban settlements. All dency to grow in parallel, so that urban these cities will grow in population as their hierarchies exhibit great stability. For relative sizes remain, more or less, the same. example, the ranking of cities according to Policy makers need to accept this reality and size within Nepal’s urban hierarchy in 2011 focus on addressing the congestion con- was largely identical to that in 2001.19 straints that inhibit prosperity and livability India’s urban hierarchy has been stable for in cities of different sizes. more than a century (Swerts, Pumain, and Denis 2014). South Asia is by no means Variation in population growth rates unique in this respect, and the tendency of a across districts—Two features city’s population growth to be independent of its size is, in fact, a common feature of Adopting a more spatial perspective, urbanization processes—so much so that ﬁ gure 2.11 (panel a) shows the variation in the tendency has acquired the status of population growth rates across subnational

FIGURE 2.11 Patterns of population growth across districts, years including 2001–10

a. South Asia b. Western province, Sri Lanka

– – –

– –– – –– – –– – –– – – – – – – –

Source: Based on data for subnational administrative units from national housing and population censuses extracted from City Population (http://www.citypopulation.de). Note: For Bangladesh, India, and Nepal, annual population growth rates are for 2001–11; for Sri Lanka, 2001–12; for Pakistan, 1998–2010. The population fi gures for Pakistan 2010 are also estimates given the absence of a census since 1998. Afghanistan, Bhutan, and Maldives are not included because of the absence of data. SPATIAL PATTERNS OF SUBNATIONAL PERFORMANCE AND URBAN GROWTH 61

administrative units (“districts”) in South annual population growth rates were fairly Asia.20 Two features stand out. uniform spatially: for 65 percent of districts, First, major cities such as Delhi, the rate was 1–3 percent. This relative unifor- Hyderabad, Kolkata, and Mumbai in India mity aligns with the notion that the majority and Colombo in Sri Lanka show a large of urban population growth within South growth differential between the districts in Asia has taken place through natural increase which the cities are located and some of the and the reclassification of rural settlements immediately neighboring districts. Population rather than through heavy rural-urban growth has been faster in these neighboring migration. districts than in the districts containing these cities. The district of Delhi experienced popu- Rapid relative expansion of lation growth of 1.9 percent a year, but urban footprints and the rise of Gurgaon, just to the south, grew at 4.5 per- the multicity agglomeration cent a year. Similarly, whereas Greater Bombay and Hyderabad had population Because some of the fastest population growth rates of 0.4 percent and 0.5 percent, growth rates have occurred in districts that respectively, Thane (which borders Greater abut districts containing major cities, these Bombay to the north) and Rangareddy cities may have quickly expanding urban (which neighbors Hyderabad to the west) footprints, spilling over formal administrative grew at 3.1 percent and 3.9 percent a year, boundaries. Using nighttime lights data (see respectively. Kolkata’s population declined box 2.1), ﬁ gure 2.12 (panel a) shows for 1999 0.19 percent a year, whereas populations in and 2010 the urban footprints of all South the remainder of the districts of West Bengal Asian cities with a population circa 2010 of grew at 1–2 percent a year. Finally, outside more than 100,000.23 As ﬁ gure 2.12 (panel b) India, Colombo’s experience was similar to illustrates for Hyderabad, urban footprints Kolkata’s: while populations contracted in have a strong tendency, especially for the several DS-divisions in the district of largest cities, to spill over administrative Colombo—Colombo, Dehiwala–Mount boundaries. Lavinia, Moratuwa, Sri Jayewardenepura This trend is not only evident in the night- Kotte, and Thimbirigasyaya—the immedi- time lights data but also from other (higher- ately surrounding localities grew relatively resolution) sources of satellite imagery. quickly (figure 2.11, panel b).21,22 The finding is mirrored in results from the Dhaka provides an exception to this trend Indian Institute for Human Settlements of relatively slow population growth com- (IIHS 2011) for Kolkata (figure 2.13, panel pared with neighboring areas. Its growth rate a). Figure 2.13, panel b presents estimates was 3.4 percent a year, making it the fourth- from IIHS of the proportion of built-up area fastest growing district in Bangladesh. From and population located outside official the figure, Kathmandu also appears to be an boundaries for India’s 12 largest cities in exception, but care in interpretation is 2010. The proportion of built-up area out- required, because the figure hides significant side a city’s official boundary often exceeds spatial variation within Bagmati, that within its boundary, especially in Kathmandu’s district. In fact, although popu- Chennai and Kolkata. In all cases, the pro- lation growth in Kathmandu Metropolitan portion of built-up area outside a city’s City has been rapid, it was even faster in the boundary exceeds the corresponding peripheral municipalities of Kirtipur and population. Madhyapur Thimi and in peri-urban areas IIHS reports that expansion of built-up officially classified as rural (Muzzini and urban areas outside official boundaries Aparicio 2013a). has been faster than that for population for The second important feature of 8 out of 12 of India’s largest cities. The figure 2.11 is that, outside major cities, main exceptions are Hyderabad and Agra, 62 LEVERAGING URBANIZATION IN SOUTH ASIA

FIGURE 2.12 Urban nighttime light footprints

a. South Asia, 1999 and 2010 b. Hyderabad, 2010

Source: Based on analysis of Defense Meteorological Satellite Program–Operational Linescan System nighttime lights data. Note: In panel b, Hyderabad’s administrative boundary is shown in black. The city’s urban nighttime light footprint is defi ned using a DN = 13 threshold (see box 2.1) with warmer colors corresponding to higher levels of nighttime light intensity.

which appear to have experienced significant The foregoing overall regional picture of densification in peripheral areas outside offi- rapid relative expansion of urban footprints cial boundaries. again largely reflects trends in India, whose The spillover of cities across their bound- share of total urban night-lit area in South aries creates challenges for metropolitan Asia in 2010 was 57.8 percent. However, the coordination in delivering basic services and fastest rates of expansion in urban area providing infrastructure. The scale of this occurred in Afghanistan and Bhutan, which expansion has grown over time, as is evi- recorded annual growth rates higher than 13 dent from the rapid growth of urban foot- percent and which showed faster rates of prints. For example, the region’s urban expansion relative to urban population than night-lit area expanded at slightly more than did the region overall. Sri Lanka had the fast- 5 percent a year during 1999–2010, against est expansion of urban area relative to urban urban population growth for the region of a population, with a ratio of more than seven— little less than 2.5 percent a year.24 Cities, whereas the country’s total urban area grew at therefore, grew in area about twice as fast a rate close to that for the region overall, its as they grew in population. This finding is urban population growth rate was much similar to most international experience slower than for the region overall. The rapid (Angel and others 2011) and suggests an expansion of urban area relative to urban pop- overall tendency toward declining average ulation reflects the sprawl and ribbon develop- city population densities and increasing ment that are characteristic of Sri Lanka’s urban sprawl. urban development (World Bank 2012). SPATIAL PATTERNS OF SUBNATIONAL PERFORMANCE AND URBAN GROWTH 63

FIGURE 2.13 Built-up urban area located outside offi cial boundaries

a. Kolkata, 2010 b. Percentage of built-up urban area and population located outside ULB boundaries, major Indian cities, 2010

300

250

200

150

100

Outside ULB boundaries (percent) 50

Agra Delhi Pune Surat Jaipur Kolkata Chennai BangaloreHyderabad Ahmedabad Chandigarh Greater Mumbai Populaton Built-up area

Source: IIHS 2011. Note: In panel a, the black line indicates Kolkata’s administrative (urban local body) boundary and red shows built-up area as detected based on Landsat satellite imagery. ULB = urban local body.

TABLE 2.3 Growth of multicity agglomerations, South Asia

Number of agglomerations Number of cities in agglomeration Area (square kilometers) 1999 2010 Annual growth Country 1999 2010 Mean Maximum Mean Maximum 1999 2010 (percent) India 23 30 4.09 17 4.73 38 22,240 75,499 11.75 Pakistan 12 10 4.00 10 6.50 29 1,536 2,558 4.75 Bangladesh – 2 – – 2.00 2 – 1,340 – Sri Lanka 1 2 5.00 5 3.50 5 182 205 1.08 Nepal 1 1 2.00 2 2.00 2 12,969 12,495 –0.30 South Asia 37 45 3.92 17 4.89 38 36,927 92,097 8.66

Source: Based on analysis of Defense Meteorological Satellite Program–Operational Linescan System nighttime lights data. Note: Afghanistan, Bhutan, and Maldives are not included in the table because they had no agglomerations in either 1999 or 2010.

By contrast, Bangladesh, Nepal, and Pakistan least 100,000 living within its administrative all experienced slower rates of growth of boundaries in 2010. In 1999, there were 37 urban area, in absolute terms and relative to such agglomerations; in 2010, 45 (table 2.3). the growth of their urban populations. India added seven agglomerations during the South Asia is home to a growing number period, while Sri Lanka witnessed the emer- of multicity agglomerations, where a multicity gence of the Galle-Matara agglomeration to agglomeration is defined as a continuously lit add to its Colombo agglomeration. belt of urbanization containing two or more Bangladesh saw the emergence of its first two cities, each of which had a population of at agglomerations—the Dhaka agglomeration 64 LEVERAGING URBANIZATION IN SOUTH ASIA

(Dhaka and Rupganj) and the Brahmanbaria for one large multicity agglomeration of agglomeration (Brahmanbaria and Bhairab). five cities with a combined population of Pakistan experienced a net decline in 8.8 million.26 agglomerations from 12 to 10 as new agglom- The most striking example of two or more erations were outpaced by the merging of agglomerations merging is that of the Delhi existing agglomerations. The Lahore agglom- and Lahore agglomerations, which now form eration expanded to absorb those of Chiniot, one enormous continuously lit belt with an Gujranwala, Gujrat, Lalamusa, and Sialkot, estimated population of 73.4 million, or just which explains the large increase in the aver- less than Turkey’s population (figure 2.15, age number of cities per agglomeration in panel a). This mega-agglomeration stretches Pakistan from 4 in 1999 to 6.5 in 2010, mak- from Palwal, south of Delhi, all the way to ing Pakistani agglomerations the largest in the Kharian in the Pakistani province of Punjab. region on this criterion. Across the region the Figure 2.15, panel b shows this mega- average number of cities per agglomeration agglomeration in high-resolution nighttime climbed from 3.9 in 1999 to 4.9 in 2010. 25 lights satellite imagery as a brightly lit corri- Coimbatore provides a good example of dor of interlinked cities between Delhi and the birth of a new agglomeration (figure 2.14). Lahore.27 In 1999, Coimbatore existed as a “single Off the corridor, smaller, less brightly lit city”—that is, a city with its own separately urban and rural centers are also evident with identifiable urban night-lit footprint. By 2010, a general decay in brightness as distance Coimbatore’s footprint had become indistin- from the corridor increases, indicating an guishable from those of the nearby cities of urban-rural continuum or gradient. The for- Bhavani, Erode, Salem, and Tiruppur, making mation of the Delhi-Lahore agglomeration

FIGURE 2.14 The birth of the Coimbatore agglomeration, India

Mandya Urban extent

2010 agglomeration Mysore 1999 agglomerations malai 1999 single cities within 2010 agglomerations

2010 single cities

Cities

Salem

Bhavani Coonoor Erode Kozhikode (Calicut)

Tiruppur Malappuram Coimbatore Karur

Tiruchirappalli Palakkad Thanjavur

Pollachi Guruvayur Thrissur 0 12.5 25 50 Km

Source: Background paper prepared by the Center for International Earth Science Information Network 2013 for this report based on analysis of Defense Meteorological Satellite Program–Operational Linescan System nighttime lights data. SPATIAL PATTERNS OF SUBNATIONAL PERFORMANCE AND URBAN GROWTH 65

FIGURE 2.15 The Delhi-Lahore agglomeration

a. Standard nighttime lights data b. Higher-resolution imagery

Source: Background paper prepared by the Center for International Earth Source: Based on data extracted from Tile 3 (75N/060E) of the Visible Infrared Imaging Science Information Network 2013 for this report based on analysis of Radiometer Suite Day/Night Band Cloud Free Monthly Composite Version 1 data set for Defense Meteorological Satellite Program–Operational Linescan System December 2014, National Oceanic and Atmospheric Administration (http://ngdc.noaa.gov nighttime lights data. /eog/viirs.html).

resembles the emergence of cross-border Spatial patterns of economic mega-agglomerations of economic activity in growth more developed regions of the world—for example, the formation of Europe’s “Hot Panel a of ﬁ gure 2.16 illustrates patterns of Banana” (referring to the rough shape of the real GDP growth across South Asian dis- fruit), which stretches from London to tricts in 1999–2010; rates of real GDP Milan. However, the Hot Banana is in the growth were derived from nighttime lights 28 most highly integrated region in the world, data. Some of the fastest GDP growth rates whereas the Delhi-Lahore mega-agglomera- have been in areas surrounding the region’s tion has formed across one of the world’s major cities—Bangalore, Hyderabad, least permeable borders. Kolkata, and New Delhi in India; Karachi The birth of new agglomerations such as and Hyderabad in Pakistan; and Dhaka in 29 Coimbatore and the fusing of existing Bangladesh. Spatial patterns of economic agglomerations points to an increasingly con- growth appear to have mirrored those of nected network of cities across South Asia. If population growth (see “Population growth the challenges that they present for urban across urban hierarchies and geographic governance can be overcome, these agglom- space”). However, looking at districts erations carry great potential for the exploita- more widely, no evidence emerges of a tion of agglomeration economies and the clear relationship between spatial patterns building of economic prosperity. of economic and population growth 66 LEVERAGING URBANIZATION IN SOUTH ASIA

FIGURE 2.16 GDP growth and population growth across South Asian districts, 1999–2010

a. Patterns of real GDP growth b. Real GDP growth vs. population growth

5 R2 = 0.016 – – 4 – –– – – –– 3 – – 2 1 0 –1 –2 –3 –4 Adjusted GDP growth rate (z-score) –5 –5.50 –3.50 –1.50 0.50 2.50 4.50 6.50 Adjusted population growth rate (z-score)

Source: Estimates of real GDP growth derived from Defense Meteorological Satellite Program–Operational Linescan System nighttime lights data. Note: Both district rates of real GDP and population growth are calculated as z-scores to adjust for national diff erences in average growth rates with positive (negative) values refl ecting growth rates above (below) the district average for a country.

(fi gure 2.16, panel b). In other words, away attributable to fast growth on their peripher- from the major cities and their surrounding ies, including areas governed as rural. areas, economic growth has been occurring Consider, for example, the better-than-pre- in different places from population growth. dicted performance of the district of Gautam Once again, this result concurs with the ear- Budh Nagar on the periphery of Delhi (and lier fi nding that the majority of urban popu- home to Noida, India’s largest census town) lation growth has been driven more by and its impact on the dynamism component natural increase and the reclassifi cation of of the prosperity index (see fi gure 2.5). By rural settlements rather than by large-scale contrast, the cores of these cities tended to rural-urban migration. It is also consistent stagnate or even decline, as seen, for exam- with the geographically dispersed pattern of ple, in New Delhi, Dhaka, and Lahore India’s new census towns (see box 2.3). ( fi gure 2.17, panels a, b, and c). The failure of economic growth to occur in For all three of these cities, warmer colors the same places as population growth sug- depicting fast growth of lights are apparent in gests that the benefits of urbanization, which immediately surrounding areas, but the cores arise from the exploitation of agglomeration of the cities themselves are bluish-gray, indi- economies and which the prosperity index cating either minimal growth or even a dim- suggests are accruing to larger districts (see, in ming of lights. Colombo (figure 2.17, panel d) particular, figure 2.3, panel a), are not being is different in that the growth of lights at its widely shared across the region. core appears not too dissimilar from that in its peripheral areas. (Though not shown, Growth of major agglomerations Bangalore is also different in that, while the growth of lights has again been fastest in The relatively fast growth of South Asia’s areas on the periphery of the city, growth has major cities between 1999 and 2010 is also remained rapid at the center.) SPATIAL PATTERNS OF SUBNATIONAL PERFORMANCE AND URBAN GROWTH 67

FIGURE 2.17 Patterns of annual nighttime lights growth around selected major cities, 1999–2010

a. Annual growth around New Delhi b. Annual growth around Dhaka

– – – –– ––– – – – – – – – – – – – – – – – –

c. Annual growth around Lahore d. Annual growth around Colombo

– – – –– – –– – – –– – – – – – – – – – – – –

Source: Based on analysis of Defense Meteorological Satellite Program–Operational Linescan System nighttime lights data. 68 LEVERAGING URBANIZATION IN SOUTH ASIA

Spatial deconcentration of from their centers) increased by almost 12 manufacturing percent. The relocation of manufacturing employment from the cores of India’s seven The pattern of relative stagnation at the cores largest metropolitan areas to their peripheries of many of South Asia’s major cities and was particularly pronounced in high-tech rapid growth on their peripheries can be and fast-growing export manufacturing partly explained by the process of manufac- industries. In high-tech manufacturing, for turing deconcentration from the centers and example, the cores experienced a 60 percent toward the outskirts of these cities without decline in employment, while suburban the emergence of suitable replacement indus- towns and suburban villages located 10–50 tries. Take India’s seven largest metropolitan kilometers from these centers experienced areas: Mumbai, Delhi, Bangalore, Kolkata, growth of 17 and 180 percent, respectively. Chennai, Hyderabad, and Ahmedabad Stagnation has been particularly pro- (World Bank 2013). They dominate the nounced for Delhi and Mumbai. Metropolitan country’s economic landscape, but between Delhi, defined as the area within a 50 kilome- 1998 and 2005 manufacturing employment ter radius of the city’s center and which within 10 kilometers of their city centers includes both the metropolitan core and sub- declined by 16 percent (ﬁ gure 2.18). By con- urban towns and villages, suffered a 0.5 per- trast, manufacturing employment in their cent decline in its share of national immediate peripheries (that is, in suburban employment between 1998 and 2005. For towns and villages located 10–50 kilometers metropolitan Mumbai, the equivalent decline

FIGURE 2.18 Employment growth in Indian metropolitan cores and their peripheries, by sector, 1998–2005

200

150

100

50 % change

–50

–100 ManufacturingFast-growing Medium High-tech Real estate Transport, export low-tech Manufacturing storage, and manufacturing manufacturing communications Metropolitan core Suburban towns Suburban villages

Source: World Bank 2013 based on Economic Census data covering manufacturing establishments of all sizes (organized and unorganized). Note: Metropolitan core includes an area with a radius of 10 kilometers centered on the main metropolis. Suburban towns comprise urban areas 10–50 kilometers from the metropolitan core, and suburban villages comprise rural areas in the same vicinity. These fi gures are averages for the seven largest metropolitan areas (in descending order of population): Mumbai, Delhi, Bangalore, Kolkata, Chennai, Hyderabad, and Ahmedabad. SPATIAL PATTERNS OF SUBNATIONAL PERFORMANCE AND URBAN GROWTH 69

was even steeper, at 1.3 percent. By contrast, FIGURE 2.19 Increased urbanization of unorganized the Chennai-Hyderabad-Bangalore metropol- manufacturing in India, 1989–2005 itan areas taken together saw an increased share in national employment of 1.1 percent 80 during the period. Bangalore, matching its 70 pattern of nighttime lights growth, attracted many medium- and high-tech manufacturing 60 jobs (World Bank 2013). 50 In India as a whole, an important difference in trends can be observed between the 40 organized and unorganized manufacturing 30 sectors, where organized manufacturing consists of all manufacturing establishments that (percent) Urban shares 20 employ 10 or more workers or, if the estab- 10 lishments do not use electricity, that employ 20 or more workers.30 Thus, while organized 0 manufacturing became significantly less 1989 1994 2000 2005 urbanized between 1994 and 2005, unorga- Unorganized output Unorganized employment nized manufacturing became significantly Unorganized plants more urbanized (figure 2.19). The share of organized manufacturing employment located Source: Ghani, Goswami, and Kerr 2012 based on National Sample Survey Organization data. in urban areas fell from 62.0 percent to 51.4 percent, whereas the share of unorganized Corporation, but by 2009 the city’s share of manufacturing employment located in these formal garment industry employment had same areas increased from 30.2 to 34.7 per- fallen to 30 percent. By contrast, the share cent (Ghani, Goswami, and Kerr 2012). In located in Dhaka’s peri-urban areas increased line with the patterns in the lights data for from 20 percent to 38 percent during the Delhi, organized manufacturing has been same period. In addition to the emergence of deconcentrating from India’s major urban a garment cluster approximately 15 kilome- centers, only to be partly replaced by lower- ters from Dhaka’s center (figure 2.20, panel productivity, unorganized manufacturing. a), garment employment has sprawled out- Although evidence also shows a growing con- ward to the municipalities of Sreepur and, to centration of services in the highest-density a lesser extent, Kaliakair, both of which lie locations—that is, in the largest cities—in just outside the boundaries of metropolitan India (Desmet and others 2015), the dimming Dhaka (figure 2.20, panel b). Just as with of lights at the cores of cities such as Delhi many of India’s major cities, only limited evi- suggests that services have not been sufficient dence can be found of the emergence of indus- to plug the gap left by the exodus of orga- tries capable of replacing the lost nized manufacturing firms. Again, Bangalore manufacturing within Dhaka. Thus, although is a major exception in its ability to retain the information and communications technol- economic vibrancy at its core. ogy sector is an emerging cluster within the A story similar to that of Delhi and city, it remains small and has not yet filled the Mumbai can be seen in Dhaka, Bangladesh void left by the garment industry. (Muzzini and Aparicio 2013b). Although In Colombo, the share of the city’s metro- Dhaka still dominates Bangladesh’s economy, politan region in national manufacturing the country’s most important industry—its employment, which is also dominated by the garment industry—has been shifting out of garment industry, declined from 52 percent the city’s core and into its peri-urban areas. In in 2001 to 43 percent in 2009. However, like 2001, more than one-half of all formal jobs in Bangalore, Colombo has successfully the industry were located in Dhaka City retained its vitality, having avoided a 70 LEVERAGING URBANIZATION IN SOUTH ASIA

pronounced dimming of lights in core areas. post–World War II low of 3.4 million in 1983, This outcome occurred because the Colombo after which it began to recover (Sassen 2001). metropolitan region has been able to attract The difference between New York and replacement industries of sufficient size and London on the one hand and South Asia’s dynamism to supplant the lost manufactur- major cities on the other is that the former ing. Thus, the region’s transport, communi- went through this process at a much more cations, and knowledge services industries advanced stage of development. In 1961, the are relatively large and rapidly growing United States and the United Kingdom had (World Bank 2012). GDP per capita levels of $11,402 and $8,857, The deconcentration of manufacturing— respectively. By contrast, in 2010, the average particularly formal and organized manufac- GDP per capita level across the four South turing—and associated stagnation of urban Asian countries of Bangladesh, India, cores is by no means a process that is histori- Pakistan, and Sri Lanka was $3,126.31 In this cally unique to South Asia. Indeed, major sense, the process of manufacturing decon- North American and European cities under- centration and stagnation of urban cores is went similar processes within the last 50–75 occurring early within South Asia; this may years. New York and London, for example, also be linked to the phenomenon of services- experienced absolute declines at their cores led urbanization that is being witnessed across and significant losses in manufacturing much of the region (see chapter 1). employment in the 1960s and 1970s. In particular, the deconcentration of for- Manufacturing employment in London fell mal and organized manufacturing that is from 1.4 million in 1961 to 680,000 in 1981, occurring at such an early stage of develop- which was a relatively larger loss than for the ment without, with some exceptions, the United Kingdom overall. At the same time, emergence of replacement industries that overall employment in the city declined by push cities up the ladder to high-value-added 900,000 from 4.3 million in 1961 to reach a activities suggests that excessive congestion

FIGURE 2.20 Employment in Dhaka

b. Employment density of garment industry a. Evolution of density of nonfarm employment, Dhaka, 2001–09 in Dhaka metropolitan area, 2009 5

1 more employees (thousands)

0 Nonfarm employment in businesses with 10 or 0 5 101520253035404550 Distance from Dhaka city center (square kilometers) Employment 2009 Employment 2006 Employment 2001 – – – –

Source: Muzzini and Aparicio 2013b. Note: CC = city corporation; P = pourashava (municipality). The boundaries of Dhaka CC and Dhaka metropolitan area are marked in red and black, respectively. SPATIAL PATTERNS OF SUBNATIONAL PERFORMANCE AND URBAN GROWTH 71

forces associated, in part, with failures in pol- induced by regulations on urban development icy may be driving the process. As discussed densities imposes welfare losses equivalent to in the Introduction (“A framework for assess- 2–4 percent of household income due to ing urbanization and city performance”), higher commuting costs (Bertaud and such congestion forces extend beyond exces- Brueckner 2004). sive traffic congestion associated with a lack of internal city connectivity and can originate Summary from a variety of different sources, including the pressure of urban population on basic Performance across subnational districts urban services, on land and housing markets, varies tremendously within South Asian and on the environment. countries. This outcome is clear from the The exact mix of congestion forces push- prosperity index. And although these varia- ing out formal and organized manufacturing tions are related to population size, that alone differs from case to case. For Dhaka, traffic is a far-from-perfect predictor of overall congestion is a relatively stronger culprit performance. Furthermore, the relatively (Muzzini and Aparicio 2013b), whereas for strong performance of the districts containing Indian cities, congestion in land markets exac- South Asia’s largest cities is driven by their erbated by overly stringent regulations on comparatively high levels of productivity, but urban development densities are more of a they tend to exhibit (with exceptions like force (World Bank 2013). Colombo’s manu- Bangalore and Colombo) lower-than-expected facturing deconcentration process is, in con- dynamism. trast, relatively more “natural” in the sense These differences in performance between that Sri Lanka, with a GDP per capita of the largest cities reflect differences in success $9,426, is at a more advanced stage of devel- in retaining vibrancy in the face of manufac- opment than either Bangladesh or India, turing deconcentration, a process that has whose respective levels of GDP per capita are caused rapid economic growth on the periph- $2,476 and $5,238,32 and also given the evi- eries of the region’s major agglomerations. dence of emerging replacement industries in With the possible exception of Sri Lanka, this Colombo. The differences observed between process is taking place at an earlier stage of most of India’s major cities and Dhaka on economic development than might be the one hand and Bangalore and Colombo on expected based on the historical experiences the other hand are also reflected in the pros- of today’s developed countries, and may also perity index. Hence, while the districts of be linked to the region’s historically atypical Chennai, Delhi, Dhaka, Greater Bombay, and pattern of services-led urbanization. Kolkata all show strong overall performance No clear relationship between spatial pat- on the index, they perform worse than terns of economic and population growth can expected on the index’s dynamism compo- be discerned within South Asian countries. nent. By contrast, both Bangalore and Rather, urban population growth has been Colombo perform as expected on this compo- relatively spatially dispersed with the popula- nent (see figure 2.5). tions of towns and cities growing more or less Also, while sub- and peri-urban areas may in parallel, indicating that countries in the have been benefiting from the deconcentra- region need to accept the policy reality that tion process in countries such as Bangladesh their existing urban hierarchies are here to and India, the overall process is nevertheless stay. Consistent with the region’s relatively suboptimal. The sprawl resulting from policy- slow pace of urbanization (see chapter 1), induced congestion constraints inhibits the urban population growth has been driven pri- exploitation of agglomeration economies and marily by natural increase and the reclassifi- elevates market connection costs for firms. cation of rural settlements rather than by Sprawl also imposes welfare costs on house- large-scale rural-urban migration. And holds. In Bangalore, for instance, sprawl reflecting the existence of widespread hidden 72 LEVERAGING URBANIZATION IN SOUTH ASIA

urbanization, a large proportion of South (relative to the average for the country in Asia’s population lives in settlements that, which it is located) and its prosperity index even though they exhibit important urban performance. In particular, regressing a dis- characteristics, are governed as rural entities. trict’s prosperity index score on its popula- South Asian cities are also witnessing rapid tion density (with both calculated as z-scores) gives an R2 of 0.24: in other words, variation growth of their urban extents relative to their in population density is able to “explain” populations, resulting in sprawl and contrib- 24 percent of the observed variation in pros- uting to messy as well as hidden urbanization. perity index performance across districts. Linked to this trend is the emergence of a Furthermore, although no statistically signif- growing number of multicity agglomerations, icant relationship exists between a district’s which present huge opportunities for the population size and its poverty rate, a statis- exploitation of agglomeration economies— tically signifi cant negative relationship does provided the associated challenges for gover- exist between a district’s population density nance in providing basic services are and its poverty rate. overcome. 5. The coeffi cient of correlation between the The overall urbanization and spatial eco- productivity and poverty indicators is +0.44, while that for the correlation between the nomic dynamics described in this chapter and productivity and dynamism indicators is the differences in performance across areas +0.39. The correlation between the poverty can be explained by the differing interactions and dynamism indicators, although positive, of agglomeration economies with congestion is much weaker (+0.09). forces, where the latter arise from the pres- 6. From fi gure 2.5, Valikamam in the district of sure of growing city populations on basic Jaffna in Sri Lanka appears to be particularly urban infrastructure and services, land and dynamic. This dynamism is likely attribut- housing markets, and the environment. Part 2 able to rapid growth following the end of (chapters 3–6) explores the interrelationships Sri Lanka’s civil confl ict, which particularly between policy and outcomes in more detail. affected the northern and eastern regions of the country, in 2009. 7. See also box 1.2 in chapter 1 for a more Notes detailed discussion of offi cial defi nitions of urban areas in South Asia. 1. As noted in chapter 1 (box 1.1), GDP data 8. The difference between a country’s are available for Indian districts, but with AI-estimated urban share and its estimated only limited temporal coverage. The Central urban share based on offi cial defi nitions Bank of Sri Lanka publishes subnational of urban areas provides a very rough and GDP data, but only for the country’s prov- extreme upper-bound estimate of the share of inces, which is an insuffi ciently refi ned spatial its population living in unrecognized urban scale to allow detailed analysis of variations settlements. In particular, the difference pro- in subnational performance. vides an estimate of the share of the popu- 2. As fi gure 2.1 shows, the distribution of scores lation living in areas that, although they are on the prosperity index exhibits a slight posi- offi cially classifi ed as rural, satisfy the crite- tive skew. As a consequence, there are several ria that the AI uses to characterize them as districts whose performance exceeds their urban: namely, they have a population den- country averages by two standard deviations sity of at least 150 people per square kilome- or more, but there are no districts whose ter and are within 60 minutes’ travel time of score falls two standard deviations or more an urban settlement with a population of at below average. least 50,000. 3. Table 2.1 presents categories of performance 9. http://www.censusindia.gov.in/default.aspx. rather than precise rankings and exact pros- 10. Figures on population density are based on perity index scores to avoid giving a spurious World Bank World Development Indicators sense of precision. data. 4. A signifi cant positive relationship also exists 11. As in chapter 1, in assessing the pace of between a district’s population density urbanization, we are forced to fall back on SPATIAL PATTERNS OF SUBNATIONAL PERFORMANCE AND URBAN GROWTH 73

estimates of urban shares and population 17. GDP per capita fi gures are expressed in 2011 based on offi cial national defi nitions of urban constant international dollars at purchas- areas. However, this study’s fi ndings on the ing power parity exchange rates and were pace of urbanization appear reasonably taken from World Bank World Development robust. In particular, for India, large biases Indicators. Sri Lanka’s GDP per capita in in both levels and growth rates of the urban 2013 was $9,426. share would be required to overturn the 18. This is evident from the relatively poor fi ts conclusion of relatively slow urbanization. of the regression lines in fi gure 2.10, as indi- Furthermore, for limited samples allowed by cated by the low R2 statistics. data, urbanization continues to appear to be 19. The stability of a country’s urban hierarchy relatively slow in India when the country is can be assessed using Spearman’s rank cor- compared with other countries based on simi- relation coeffi cient. A rank correlation coeffi - lar defi nitions of urban. See web-based annex cient of zero indicates no correlation between 1A and Roberts (2015) for more details. city rankings in different years, whereas a 12. During 2000–10 the annual rate of popula- coeffi cient of +1 represents a perfect cor- tion growth was 2.4 percent in Bhutan, 1.9 relation with no changes in rank between percent in Pakistan, 1.8 percent in Maldives, years. For Bangladesh, India, Nepal, and Sri and 1.5 percent in India. Lanka, the values of Spearman’s rank corre- 13. Although its contribution relative to other lation coeffi cient for 2001–11 (2001–12 for sources remains small, there is evidence, at Sri Lanka) are 0.9564, 0.9603, 0.9702, and least for some South Asian countries, that 0.9954, respectively. rural-urban migration is increasing, albeit 20. In general, these subnational units are at the marginally, in importance as a source of Admin-2 level. The exceptions are Pakistan, urban population growth. In particular, in where they are at the Admin-1 (that is, India the share of urban population growth provincial) level, and Sri Lanka, where they attributable to net inward migration from are at the Admin-3 (that is, DS-division) rural areas increased from 21.2 percent to level. Regardless of country, and similar to 24.1 percent between the periods 1991–2001 the analysis of the prosperity index (see the and 2001–11 (IIHS 2011). section “Measuring differences in subna- 14. The fi gures on migration that have been tional performance”), the subnational units cited in this paragraph do not include in fi gure 2.11 are referred to as “districts” for seasonal or temporary rural-urban migra- brevity. tion, which, in general, is not well captured 21. Some care is required in interpreting the by either national population censuses or results for Sri Lanka because of the change standard survey instruments within the in the method of enumerating the popula- region. Estimates of the magnitude of sea- tion between the 2001 and 2012 censuses. In sonal migration vary dramatically. For particular, while enumeration for the 2012 India, Keshri and Bhagat (2013) report, census took place on a de jure basis (persons based on National Sample Survey data, were enumerated based on their usual place that 13.6 million people in 2007–08 were of residence), that for the 2001 census took seasonal migrants. This is equivalent to place on a de facto basis (persons were enu- 3.8 percent of India’s offi cial urban popula- merated based on where they were on the tion in 2008. By contrast, Deshingkar and night of the census). others (2009) cite results from village surveys 22. Besides Kolkata and Colombo, an additional indicating that 50–60 percent of households 41 districts (out of a total of 935) experi- in villages are involved in seasonal migration. enced population decline during the study 15. The literacy rates among 15–24-year-olds in period. For 24 of these districts the decline Sri Lanka’s estate sector in 2012 was only was near negligible, but it was higher for the slightly lower, at 95.6 percent. The estate remainder. sector consists of tea and rubber planta- 23. A small number of additional cities with circa tions and is separately reported on in offi - 2010 population less than 100,000 were also cial Sri Lanka, Department of Census and included in the analysis on the grounds of Statistics publications. their importance to a particular country’s 16. The analysis for Sri Lanka covers 2001–12. urbanization process. Thus, for example, 74 LEVERAGING URBANIZATION IN SOUTH ASIA

a number of secondary cities in Nepal and parity exchange rates and are taken from Sri Lanka were included despite not meeting the Maddison Project database (http://www the 100,000 population threshold. A full list .ggdc.net/maddison/maddison-project/home of the cities included in the sample is avail- .htm). able on request. 32. GDP per capita fi gures are expressed in 2011 24. One caveat when considering this result constant international dollars at purchasing is that the growth rate for urban popula- power parity exchange rates and were taken tion is calculated using World Urbanization from the World Bank World Development Prospects, 2011 Revision data, which are Indicators. based on offi cial national defi nitions of urban areas. To the extent that these defi nitions tend to underestimate the sizes of urban References areas, they will also tend to underestimate urban population. This underestimation will Alder, S. 2015. “Chinese Roads in India: lead to biased estimates of urban population The Effect of Transport Infrastructure on growth rates if the degree of underestimation Economic Development.” Working Paper, has changed over time. University of North Carolina at Chapel Hill. 25. Note, however, that in both 1999 and 2010, Angel, S., J. Parent, D. L. Civco, and A. M. Blei. a large number of agglomerations consisted 2011. Making Room for a Planet of Cities. of just two 100,000+ population cities. Thus, Cambridge, MA: Lincoln Institute of Land 17 out of 37 agglomerations in 1999 con- Policy. sisted of just two cities. In 2010, 22 out of 45 Baum-Snow, N., and M. A. Turner. 2012. consisted of two cities. “Transportation and the Decentralization 26. This population estimate was arrived at by of Chinese Cities.” Unpublished, Brown using geographic information system tech- University, Providence, RI. niques to layer Coimbatore’s urban foot- Bertaud, A., and J. K. Brueckner. 2004. print with gridded population data for 2011 “Analyzing Building-Height Restrictions: taken from LandScan (http://web.ornl.gov Predicted Impacts, Welfare Costs, and a Case / sci / landscan). The same techniques were Study of Bangalore, India.” World Bank Policy also used to estimate population for the Research Working Paper 3290, World Bank, Delhi-Lahore mega-agglomeration. Washington, DC. 27. This imagery was obtained from the Visible Bosker, M., U. Deichmann, and M. Roberts. 2015. Infrared Imaging Radiometer Suite (VIIRS). “Hukou and Highways: The Impact of China’s The instrument was launched aboard the Spatial Development Policies on Urbanization Suomi National Polar-Orbiting Partnership and Regional Inequality.” World Bank Policy satellite in October 2011, and the imagery has Research Working Paper 7350, World Bank, a resolution of 0.75 kilometer. This compares Washington, DC. to a native spatial resolution of approximately Chan, K. W., and W. Buckingham. 2008. 2.7 kilometers for the Defense Meteorological “Is China Abolishing the Hukou System?” Satellite Program–Operational Linescan China Quarterly 195: 582–606. System instrument. Chauvin, P., E. L. Glaeser, and K. Tobio. 28. Figure 2.16 is based on the same set of sub- 2013. “Urban Economics in the U.S. and national administrative units as fi gure 2.11. India.” Unpublished, Harvard University, As with fi gure 2.11, these units are referred Cambridge, MA. to as “districts” for brevity. Ciccone, A. 2002. “Agglomeration Effects in 29. After correcting for differences stemming Europe.” European Economic Review 46 (2): from national factors. 213–27. 30. (Un)organized manufacturing is usually Ciccone, A., and R. E. Hall. 1996. “Productivity associated with (in)formal manufacturing, and the Density of Economic Activity.” although, strictly speaking, the concepts are American Economic Review 86 (1): 54–70. not identical. Center for International Earth Science 31. The fi gures for GDP per capita in this Information Network. 2013. “Report for paragraph are expressed in 1990 constant Phase I: Mapping, Quantifi cation and Analysis international dollars at purchasing power of Evolution of Patterns of Urban Physical SPATIAL PATTERNS OF SUBNATIONAL PERFORMANCE AND URBAN GROWTH 75

Extent and Morphology in South Asian Cities, 16, Marron Institute of Urban Management, 1999-2010.” Background paper for this report, New York University. World Bank, Washington, DC. Karim, M. S., and A. Nasar. 2003. “Migration Croft, T. A. 1978. “Night-Time Images of the Patterns and Differentials in Pakistan: Based Earth from Space.” Scientiﬁ c American 239: on the Analysis of the 1998 Census Data.” In 86–98. Population of Pakistan: An Analysis of 1998 Deshingkar, P., M. Deshpande, S. Kumar, A. Population and Housing Census, edited by Paradkar, L. Rao, and P. Sharma. 2009. A. R. Kemal, M. Irfan, and N. Mahmood, “Governance for Local Development in 173–80. Pakistan Institute of Development Small Towns: Addressing the Challenges Economics, Islamabad. and Opportunities of Increasing Migration Keshri, K., and R. B. Bhagat. 2013. and Mobility in India.” Rural-Urban “Socioeconomic Determinants of Temporary Working Paper 18, International Institute Labour Migration in India: A Regional for Environment and Development, London. Analysis.” Asian Population Studies 9 (2): Desmet, K., E. Ghani, S. A. O’Connell, and 175–95. E. Rossi-Hansberg. 2015. “The Spatial McCombie, J. S. L., and A. P. Thirlwall. 1994. Development of India.” Journal of Regional Economic Growth and the Balance-of- Science 55 (1): 10–30. Payments Constraint. London: Macmillan. Doll, C. N. H. 2008. “CIESIN Thematic Guide Muzzini, E., and G. Aparicio. 2013a. Bangladesh: to Night-Time Light Remote Sensing and The Path to Middle-Income Status from Its Applications.” Center for International an Urban Perspective. Washington, DC: Earth Science Information Network World Bank. (CIESIN), Columbia University, Palisades, ———. 2013b. Urban Growth and Spatial New York. Transition in Nepal: An Initial Assessment. Gabaix, X., and Y. M. Ioannides. 2004. “The Washington, DC: World Bank. Evolution of City Size Distributions.” In Pinkovskiy, M. L. 2013. “Economic Handbook of Regional and Urban Economics, Discontinuities at Borders: Evidence Volume 4: Cities and Geography, edited by from Satellite Data on Lights at Night.” J. V. Henderson and J.-F. Thisse, 2341–78. Unpublished, Massachusetts Institute of Amsterdam: Elsevier. Technology, Cambridge, MA. Ghani E., A. G. Goswami, and W. R. Kerr. 2012. Pradhan, K. C. 2013. “Unacknowledged “Is India’s Manufacturing Sector Moving Urbanization: The New Census Towns of away from Cities?” World Bank Policy India.” Center for Policy Research, New Delhi. Research Working Paper 6271, World Bank, Roberts, M. 2014. “Urbanization in South Asia: Washington, DC. A View from Outer Space.” Background Glaeser, E. L., J.-P. Chauvin, and K. Tobio. paper for this report, World Bank, 2011. “Urban Economics in the U.S. and Washington, DC. India.” Presented at the Economic Geography Roberts, M., and C. Goh. 2011. “Density, Conference, Seoul, June 29. http://www Distance and Division: The Case of Chongqing .scribd.com/doc/59978593/Prof-Ed -Glaeser- Municipality, China.” Cambridge Journal of Urban-Economics-in-the-US-and -India. Regions, Economy and Society 4 (2): 189–204. Henderson, J. V., A. Storeygard, and D. N. Weil. ———. 2015. “Assessing India’s Pace of 2011. “A Bright Idea for Measuring Economic Urbanization: A Robustness Analysis.” Growth.” American Economic Review 101 Background paper for this report, World Bank, (3): 194–99. Washington, DC. ———. 2012. “Measuring Economic Growth Romer, P. M. 1990. “Endogenous Technological from Outer Space.” American Economic Progress.” Journal of Political Economy 98 Review 102 (2): 994–1028. (5): S71–S102. IIHS (Indian Institute for Human Settlements). Rosenthal, S. S., and W. C. Strange. 2004. 2011. Urban India 2011: Evidence. Bangalore: “Evidence on the Nature and Sources of IIHS. Agglomeration Economies.” In Handbook of Jedwab, R., and D. Vollrath. 2015. “The Mortality Regional and Urban Economics, Volume 4: Transition, Malthusian Dynamics, and the Cities and Geography, edited by J. V. Henderson Rise of Poor Mega-Cities.” Working Paper and J.-F. Thisse, 2119–71. Amsterdam: Elsevier. 76 LEVERAGING URBANIZATION IN SOUTH ASIA

Sassen, S. 2001. The Global City: New York, ———. 2011. Sri Lanka: Reshaping Economic London, Tokyo. 2nd ed. Princeton, NJ: Geography—Connecting People to Prosperity. Princeton University Press. Washington, DC: World Bank. Shilpi, F., P. Sangraula, and Y. Li. 2014. “Voting ———. 2012. Turning Sri Lanka’s Urban with Feet? Access to Infrastructure and Vision into Policy and Action. Colombo: Migration in Nepal.” Background paper for World Bank. this report, World Bank, Washington, DC. ———. 2013. Urbanization beyond Municipal Solow, R. M. 1956. “A Contribution to the Boundaries: Nurturing Metropolitan Theory of Economic Growth.” Quarterly Economies and Connecting Peri-Urban Areas Journal of Economics 70 (1): 65–94. in India. Washington, DC: World Bank. Sri Lanka, Department of Census and Statistics. doi:10.1596/978-0-8213-9840-1. 2012. “Census of Population and Housing World Bank and DRC (Development Research 2011: Population of Sri Lanka by District— Center of the State Council). 2014. Urban Preliminary Report.” Department of Census China: Toward Efficient, Inclusive, and and Statistics, Sri Lanka. Sustainable Urbanization. Washington, DC: Swerts, E., D. Pumain, and E. Denis. 2014. “The World Bank and DRC. Future of India’s Urbanization.” Futures 56: Zhang, Q., and K. C. Seto. 2011. “Mapping 43–52. Urbanization Dynamics at Regional and World Bank. 2008. World Development Report Global Scales Using Multi-temporal DMSP/ 2009: Reshaping Economic Geography. OLS Nighttime Light Data.” Remote Sensing Washington, DC: World Bank. of Environment 115 (9): 2320–29. PART 2

Policies for Improved Leveraging

outh Asia’s cities are not fulfilling their area—in empowerment, in resources, and in development potential as characterized accountability—that the regions’ policy mak- Sby either prosperity or livability because ers need to address to bring about these of the congestion pressures that growing improvements. But while addressing these populations are exerting on infrastructure, deficits is a necessary condition for meaning- basic services, land, housing, and the envi- ful progress, it will not, by itself, be suffi- ronment. How can policy makers address the cient. Chapters 4–6 therefore discuss three main congestion constraints on the region’s additional, interrelated areas for policy cities and better leverage the urbanization action that are instrumental to the further process for improvements in prosperity and leveraging of urbanization for improved livability? Part 2 of the report provides prosperity and livability—connectivity and answers. Improvements in urban governance planning (chapter 4), land and housing and finance hold the keys. Chapter 3 dis- (chapter 5), and resilience to disaster and the cusses three fundamental deficits in this effects of climate change (chapter 6).

Revitalizing Urban Governance and Finance 3

Key messages

If urban governments are to play their potentially discretion and national priorities while pro- vital role as agents of local and national moting responsible fiscal behavior. Urban growth, countries must reduce the deficits in governments should creatively strive to raise empowerment, resources, and accountability local revenues, including through borrowing, that hinder urban peformance. National, to enhance their credibility and effectiveness. intermediate, and local governments need to take • Strengthen urban governments’ accountability steps to do the following: systems through better and more transparent • Expand and enhance urban empowerment, fi nancial management, workable coordination capacity, and incentives to plan and deliver arrangements with key partners, enhanced services essential for economic and social citizen feedback channels, and stronger urban development. These include services that autonomy. Citizens and fi rms need informa- urban governments can deliver on their own tion to back up their efforts to pressure urban and those best shared with higher and peer governments to deliver. levels of government and private and commu- • Strategically and pragmatically implement tai- nity partners. lored reforms to ensure that new systems and • Reformulate fl awed, anemic intergovernmental processes will be feasible, productive, and revenue systems. Transfers can support urban sustainable.

Introduction planning of urbanization for improved connectivity, land use, and urban resilience. This Previous and subsequent chapters document chapter focuses on characterizing, comparing, the importance of urbanization in South Asia, and assessing essential governance and ﬁ scal assess how it is proceeding, and outline criti- mechanisms in South Asian countries’ inter- cal issues as well as policy options that could governmental systems. These mechanisms be pursued to improve the management and substantially shape the ability of urban 79 80 LEVERAGING URBANIZATION IN SOUTH ASIA

governments to meet the considerable and and not readily comparable, but a few growing range of demands and challenges impressions can be drawn with the caveat they face. that some of the information may not be The nature and characteristics of urban fully current.1 areas in the countries of South Asia vary For a start, local governments in most greatly. Similarly, the demands faced by South Asian countries account for small governments for local public services, as shares of total public spending, mostly in well as the viable options to meet them, the range of 3–10 percent. Selective data also vary substantially. Policies appropriate from around the world show that local for relatively small urban areas in Bhutan expenditure shares in developing countries and Maldives, for example, may not be are commonly less than 10 percent and usu- workable in more than a general sense for ally much less on the revenue side (United the large urban areas of Bangladesh, India, Cities and Local Governments 2010). In and Pakistan. In some countries, urban other regions, however, local governments areas are more self-contained (that is, they in multiple countries have substantially are under the primary management of a larger roles—including in Ethiopia, Kenya, single subnational entity); in others, inter- Nigeria, South Africa, and Uganda in jurisdictional encroachment and interde- Africa; China, Indonesia, the Philippines, pendency are extensive, complicating and Vietnam in East Asia; and Argentina, public service delivery, revenue collection, Brazil, Chile, Colombia, and Peru in Latin and territorial governance. Many other America. In South Asia, only India has a differences—including economic base, set- high local share of public spending, but the tlement density, development patterns, and autonomy of Indian local bodies is signifi- socioeconomic characteristics of the cantly constrained. Local governments in population—are also evident and are some of the countries with higher local relevant for the effective management of expenditure shares in other regions also face urban areas. constraints on spending autonomy, but they Despite these consequential differences, the tend to have more discretion than do their region’s countries generally share certain counterparts in India and other South Asian common challenges and limitations in their countries with lower local expenditure systems of local governance and finance. shares.2 Most urban, municipal, and local govern- As South Asian urban areas become ments face severe systemic constraints that larger, more diverse, and more economically limit their autonomy, fiscal role, and important, there is a pressing need for accountability. strong action. Urban governments can often This chapter documents a core triad of take independent steps to improve perfor- consequential deficits in South Asian urban mance, but they also need support from governance and finance: in empowerment, higher levels in the form of robust and inno- resources, and accountability (figure 3.1). vative policies that promote empowered, Although urban governance and decen- adequately resourced, and well-governed tralization efforts face challenges globally, local governments. Such measures hold con- South Asia appears to be among the weaker siderable potential to allow local govern- regions in the pace of its reform adoption ments to meet the needs of urban and implementation progress. Central (and populations more effectively and sustain- in some cases intermediate or state) govern- ably, as well as to play a more explicit and ments are typically very powerful or even vigorous role in contributing to the success- dominant, perhaps even to an extent compa- ful overall development of South Asian rable only to the Middle East. Definitive countries. assessments of local government roles are This chapter has six sections. The first com- difficult to make because data are limited pares basic subnational and intergovernmental REVITALIZING URBAN GOVERNANCE AND FINANCE 81

FIGURE 3.1 Three key urban governance defi cits in South Asia

Empowerment Limited, overlapping, and fragmented functional assignments with deficit inappropriate and excessive central, state, and provincial control

Resource Extremely weak financial bases: Limited revenue powers and deficit inappropriately conceived and targeted intergovernmental transfers

Accountability Weak and underutilized accountability mechanisms despite deficit elections and rights to information

structures of South Asian countries, including levels of government as well as a range of non- the key parameters and foundations of their governmental actors, each of which has some respective policies regarding decentralization.3 role to play in urban development, indepen- The next three sections consider the three defi- dently and in support of urban government cits mentioned,4 followed by a discussion illus- performance, is recognized. trating how these deficits interact and collectively affect urban service delivery. The Basic government and final section pulls together the various topics, intergovernmental structures draws some conclusions about the state of and decentralization policy urban governance and finance in South Asia, and makes a number of tentative broad rec- Intergovernmental systems vary widely in ommendations, with the understanding that South Asia. Countries have differing levels of additional work is required to develop a set of subnational government and types of local more specific policy options for each country. governments. These levels and types may be Given the wide range of issues, the diversity differentially empowered with regard to the of South Asian countries, and major con- nature of decentralization (deconcentration straints on securing relevant and comparable of higher-level functions versus devolution of data and information, the chapter is not meant autonomous powers) and its importance (the to be exhaustive or definitive; instead, it is scope and scale of local empowerment). intended to highlight basic information to Relationships among subnational actors facilitate a critical assessment of the gover- (more hierarchical versus more independent) nance and finance preparedness of subnational also vary among the countries, as do the governments to meet urbanization challenges. nature and extent of the mechanisms set up Selected system features are the subject of to facilitate interjurisdictional coordination deeper analysis in certain countries or groups (including within large metropolitan areas) of countries. The focus is on urban areas, and in meeting public functions. Many observed the primary concern is with appropriate characteristics of intergovernmental systems empowerment for more effective local govern- in South Asia have deep historical and politi- ments. At the same time, the need for an over- cal foundations that affect the prospects and arching national framework that includes all avenues for reform. 82 LEVERAGING URBANIZATION IN SOUTH ASIA

Levels of government and administration designation for cities, municipalities, and more urbanized local bodies, although The diversity of the intergovernmental land- these designations can be formal to varying scape in which South Asian cities operate is degrees and are more consequential in evident from table 3.1. The number of levels practice in some countries than in others. of subnational (intermediate and local) government or administration range from two in Decentralization and subnational Maldives to several in other countries, government policy and frameworks although whether some of the ofﬁ cial jurisdictional distinctions designate distinct levels of Six of the eight countries operate as unitary government or categories of government at states (India and Pakistan, which are federal, essentially the same level that play different are the exceptions). Devolution is relatively roles in public functions is subject to some modest. The two federal countries have pow- debate. For present purposes, the main point erful and fairly independent intermediate tiers, is that the nature and size of subnational enti- which have substantial control over further ties vary considerably across countries. decentralization to lower tiers and heavily These basic designations also fail to do influence local government behavior. India justice to the multiplicity of types of local enacted the 73rd and 74th Constitutional bodies in most South Asian countries and Amendments in 1992, which aim to may not recognize that they might not only strengthen rural and urban local governments be differentially empowered but also have by assigning them speciﬁ c civic functions sub- varying relationships with each other (for ject to state level legislation and regulation. example, hierarchical versus independent). However, devolution has been limited and Most countries have some type of specific uneven across and within states/provinces.

TABLE 3.1 Levels or categories of subnational government and administration

Afghanistana Provinces (34) Urban LGs: Municipalities (153), including Kabul with special legal status Rural LGs (not formal LGs): Districts (399), Villages (community development councils; 40,020) Bangladesh Zila parishads (districts; 64) Urban LGs: City corporations (11) and pourashavas (municipalities; 315) Rural LGs: Upazila parishads (subdistricts; 510), Union parishads (5,000) Bhutan Dzongkhags (districts; 20) Thromdes (municipalities; 4) (note: 16 small municipalities are under district administration) Gewogs (blocks; 205) India States (28) and Union Territories (7) including Federal District Urban LGs: Municipal corporations (138), Municipalities (1,595), Town councils (2,108) Rural LGs: (Panchayati Raj): Zilla (593), Samities (6,087), Gram (villages; 239,432) Maldivesb Cities (2) and Atolls (19) Islands (189) Nepal District: District development committee (75) Urban LGs: Municipalities (191, including recently amalgamated villages) Rural LGs: Village development committees (3,276) Pakistan Provinces (4) and Territories (4) LGs: Zillas (districts; 96); Tehsils (337); Unions (6,022) Sri Lanka Provinces (9) Districts (25) Urban LGs: Municipal councils (23), Urban councils (41) Rural LGs: Pradeshiya Sabhas (257)

Sources: Based on the literature (see references). Note: CG = central government; LG = local government. a. There is some disagreement about the exact numbers of local governments. b. Atolls and islands are administrative designations; atoll councils supervise (in all but one case) multiple island councils. REVITALIZING URBAN GOVERNANCE AND FINANCE 83

In most of the nonfederal countries, some efforts, however, are too undefined, too form of deconcentration prevails overall or at new to evaluate, or not fully operational. certain levels. Provinces, for example, are In most cases, cities, municipalities, or powerful in Afghanistan. Zilas in Bangladesh urban local bodies have somewhat more are administrative rather than political enti- responsibility than other local levels, either ties. In Sri Lanka, local governments are for- formally through legal provisions or in prac- mally recognized by constitutional amendment tice. At the same time, most of these urban as under the jurisdiction of provinces rather governments are greatly affected by higher- than independent entities. Complex hybrid level decisions and oversight. arrangements at the same level can also create A prominent feature of South Asia is that challenges: for example, the president subnational governments throughout the appoints provincial governors but the winning region have both a constitutional and a party or coalition in provincial council elec- legal basis (table 3.2, second column). The tions appoints the chief minister. situation is similar elsewhere (as in Brazil, To varying degrees some countries are Mexico, and South Africa), but decentraliza- beginning to strengthen the roles and auton- tion policies in many developing and omy of subnational bodies. In Nepal ongo- middle-income countries have their primary ing deliberations about a new constitution basis in law. This is a less robust and durable have focused on introducing a federal foundation than constitutional provisions system with local governments, and because laws can be more easily changed. other countries, including Bhutan and Although many countries constitutionally or Maldives, are making efforts to empower legally empower subnational levels, these pro- specific types of local bodies. Most of these visions are not particularly strong in that they

TABLE 3.2 Decentralization policies and frameworks

Decentralization policy Underlying framework (nature and extent of state decentralization) (constitutional and legal provisions) Afghanistan Centralized unitary state with little local autonomy, 2004 Constitution established a centralized state and except municipalities. Recent reforms have devolved elected provincial, district, and municipal bodies; minor administrative and fi scal authority; however, the delegation to LGs partly realized by Subnational national government retains most authority, and only Governance Policy (2010). Municipalities Law (2000) provincial elections have been held. provides some urban LG autonomy; other laws apply, but overall legal framework is inconsistent.

Bangladesh Unitary democratic republic with largely Articles 59 and 60 of the 1972 Constitution provide deconcentrated local entities. Urban LGs’ autonomy for elected LGs and require Parliament to determine is limited, but councils are elected, as are local specifi c LG functions. Subsequent amendments and governments at lower rural and subdistrict levels. laws have experimented with various levels of LG autonomy, roles, and responsibilities.

Bhutan Unitary constitutional parliamentary monarchy with 2008 Constitution established “decentralized gradually increasing decentralization (1981–2010). and devolved” governments with broad revenue Large urban areas are more independent, but at and service-delivery functions. Roles have been present LGs essentially remain deconcentrated tiers. clarifi ed, expanded, and empowered by the Local Governments Act (2009) and national policies.

India Federal system with strong states. Rural and urban The federal system is outlined in the 1949 local bodies are recognized under state oversight. Constitution; the 73rd and 74th amendments (1992) Larger urban areas tend to have more independence. aim to strengthen substate governments but subject National debate on encouraging states to further to state legislation and regulation; each state issues a empower local tiers has reemerged. municipal act that outlines the specifi c functions of local bodies in that state (many variations). (continues next page) 84 LEVERAGING URBANIZATION IN SOUTH ASIA

TABLE 3.2 Decentralization policies and frameworks (continued)

Decentralization policy Underlying framework (nature and extent of state decentralization) (constitutional and legal provisions) Maldives Unitary system with authority recently decentralized The Decentralization Act of 2010 outlines the to two levels. Large cities are independent from atolls decentralization framework and establishes (fi rst tier of subnational government), while atolls subnational governments. Amendments are being oversee most island governments (second tier). considered by the president and the Majlis (national legislature) to merge some levels of government.

Nepal Unitary system with strong center (constitutional Federal system under the Interim Constitution arrangements under discussion call for federalism); (2007); Local Self-Governance Act (1999) and Local lower levels of government have some autonomy Self-Governance Regulations (2000) establish LG but are currently run by centrally appointed civil functions regulated by the CG, while proposed state servants pending a new constitution. and provincial jurisdictions are to be determined in drafting the new constitution.

Pakistan Federal system with strong center and provinces; The 18th amendment to the Constitution devolved Constitution requires elected LGs; elections had not several shared functions fully to the provinces; recent been held since 2009, but have begun again in 2015; provincial LG acts reduce 2001 devolution; LGs are district is the most important LG tier; union councils, administered by provincially appointed civil servants. the lowest tier, are directly elected.

Sri Lanka Unitary system with strong center and provinces; LGs received constitutional recognition in the 13th the provinces oversee LGs; many powers assigned to amendment (1987) but are under provincial control; the provinces and LGs are not devolved in practice. separate laws establish service responsibilities for urban councils (1939) and municipal councils (1947).

Sources: Based on the literature (see references). Note: CG = central government; LG = local government.

are general and provide limited specificity on seriously hampered by relatively modest local government powers and functions. The empowerment. This challenge has three com- details of these general provisions in several mon dimensions: limited formal functional countries (discussed below) have not been and revenue assignment, lack of clarity in further developed or have not been substan- institutional roles, and inadequate autonomy tially implemented as defined, obviously lim- to make independent decisions on matters iting the ability of urban governments to play generally considered important to realizing the their intended and potential roles. beneﬁ ts of decentralization. The nature and Decentralization has been subject to cycli- extent of these issues vary across countries, cal (often politically driven) variations in but they are present throughout the region. some countries. Nepal, for example, used to have local government elections (prior to Limited subnational assignment of 2002) before a period of political turmoil, functions and resources and both Bangladesh and Pakistan have at times had stronger provisions for local gov- Many types of public services are provided at ernments. Thus, reformers need to be aware the local level in South Asia, but the functions of historical influences and carefully consider executed directly and independently by sub- when and where national political space national governments vary in scope and sig- might be available to create a stronger role niﬁ cance across countries. There is also often for urban governments. a wide difference between functions assigned de jure to local governments and the de facto The empowerment defi cit performance of local functions by deconcentrated entities, which can vary within coun- Although progress has been made in recent tries (table 3.3). Revenues are discussed more years, urban governments in South Asia are fully in the next section. REVITALIZING URBAN GOVERNANCE AND FINANCE 85

Local governments are generally assigned least one type of local government in each expenditure functions considered theoretically country in the region has a constitutional or appropriate for local provision. This simple legal right to use a range of revenue sources, statement, however, does not adequately cap- there are constraints on this authority, and it ture considerable variations in the actual pro- is rarely used adequately. vision of these functions within countries, The significance of local government spend- some lack of clarity/too much overlap in the ing in total public spending also varies a great provision of specific functions by different deal. Indian local governments have the great- levels and types of local government, and the est share, at about a third of public spending, sometimes highly consequential effects of and this fraction is dominated by urban areas. controls imposed and ad hoc interference by At the low end are Sri Lanka (0.5 percent), higher levels of government in local fiscal Afghanistan (2 percent), and Bangladesh matters. (3 percent). Pakistan is at 5 percent and Own-source revenues are one of the great- Bhutan 16 percent. These figures, however, est weaknesses of local government systems in mask considerable differences in local auton- many developing and middle-income coun- omy because many types of local functions are tries, particularly in South Asia. Although at largely deconcentrated or centrally controlled

TABLE 3.3 Local government functions, revenues, and expenditure shares

Local functions Local revenuesa Local spending share Afghanistan Despite some formal clarity on LG functions, Municipalities are allowed to raise 24% of public spending is subnational CG agencies and their subnational funds through a variety of tax and but only 8% subprovincial and 2% departments manage many key services. nontax revenues. municipal (2007).

Bangladesh Most LGs have no strong direct role in LGs have some formal revenue options 3% of expenditure by LGs (2011/12); providing services and largely function as in the form of taxes, fees, rates, and other local-level spending is centrally deconcentrated central government agents. rentals. controlled.

Bhutan LGs have defi ned roles;thromdes have 27 Thromdes and gewogs are allowed to 22% of public spending (2008) is functions, including water, sanitation, and levy and collect certain tax and nontax local; larger share for dzongkhags than solid waste, as per 2011 Thromde Rules. revenues. gewogs.

India Municipalities have 18 functions (Constitution, Municipal bodies may levy and collect About 66% of public spending is 12th Schedule), but each state determines taxes and other revenues from a subnational (2004), nearly evenly specifi cs (much variation), focus is on constitutional list (7th Schedule) if divided between states and lower tiers; core (urban) functions (water, streetlights, allowed by state governments. high share in urban areas. sanitation, roads).

Maldives Atolls have 20 functions (oversight, Atolls, islands, and cities can only About 5% of public expenditures are consultation, revenue raising); islands and charge fees for services delivered. made by LGs (2011). cities have 25 functions.

Nepal System evolving; some functions assigned LGs have access to diverse bases, more About 10% of public expenditures are to development committees (village or so in municipalities. made by LGs (2013). municipal and district levels).

Pakistan Functions vary by province and LG: districts Sources vary by LG type (districts, LG expenditures constitute about 5% of (mostly rural infrastructure, some basic social tehsils, unions); LG resources limited public spending (2011). Total provincial services); tehsils (urban services); unions help with most funding transferred. and local spending is about 33%. villages; neighborhoods propose projects.

Sri Lanka LGs are responsible for municipal solid waste, Municipal councils are allowed to levy Municipalities account independently utilities, and markets, but share responsibility taxes and user fees within limits set by for less than 0.5% of public spending with the center for other sectors. the central government. (2006).

Sources: Based on the literature (see references). Note: CG = central government; LG = local government. a. See table 3.6 for more detailed information on local own-source revenues. 86 LEVERAGING URBANIZATION IN SOUTH ASIA

through transfer policy or overt interference, about what higher levels of government and so they are not truly local. citizens should hold urban governments In some cases, services are provided by accountable for. alternative means. In Afghanistan, for exam- Equally important, particularly for urban ple, local service delivery entities fill gaps in areas, is that jurisdictional cooperation and delivery of services assigned to weak formal coordination mechanisms are largely under- local governments. These entities include developed or ineffectively used in South Asia community development councils and school (table 3.4). Such mechanisms can be vertical management committees. Use of such mecha- (among levels), horizontal (across the same nisms may be needed at certain stages of sys- level, including in metropolitan areas), gen- tem development, but they can also hinder eral purpose (for example, a broad-based the ability of formal local governments to metropolitan development authority), or spe- take up their legal functions. cific in purpose (for a particular service sector, such as roads or water, for instance). In some Lack of clarity in institutional roles and cases, of course, local governments are fairly in interjurisdictional coordination self- contained and have few responsibilities, limiting the need for coordination across Insufﬁ cient clarity in the powers and respon- jurisdictions. In the federal countries, deci- sibilities of urban governments—resulting sions and behaviors of the state and provin- from the legal framework or the practices of cial governments strongly affect avenues for government actors—is not uncommon. This cooperation. murkiness can lead to gaps and redundancies In more decentralized environments and in service delivery, complicate resource gener- where jurisdictional proximity is important, ation and allocation, and create confusion intergovernmental action requires careful

TABLE 3.4 Interjurisdictional cooperation and coordination mechanisms

Afghanistan There are no formal mechanisms for jurisdictional cooperation. International Provincial Reconstruction Teams work across provinces but somewhat outside formal government.

Bangladesh Given the highly centralized nature of Bangladesh’s system, cooperation is largely managed by the Local Government Division of the central government.

Bhutan Members of thromde governments attend dzongkhag sessions to coordinate decisions, and any issues are referred to the Ministry of Works and Human Settlement for adjudication.

India Each state determines its institutional framework for execution of functions and coordination within its jurisdiction, including among LGs and various types of state-level parastatal organizations. In general, however, coordination is often considered inadequate.

Maldives The country’s dispersed islands preclude robust cooperation in most areas; CG provides certain services, for example, education in regional hubs.

Nepal Extensive planning processes and local coordination bodies provide for coordination among jurisdictions and deconcentrated agencies, although practice is mixed. CG agencies coordinate delivery of most public services across jurisdictions.

Pakistan Provinces set mandatory LG policy; LG commissions and appointed civil servants may settle interjurisdictional disputes; jurisdictions responsible for municipal services in rural and urban areas can enter into service agreements with other LGs, higher or lower, to achieve economies of scale.

Sri Lanka Ministry of Local Government and Provincial Councils holds monthly meetings to coordinate with provincial commissioners and external organizations working with local governments. Municipal services are largely coordinated by the CG Urban Development Authority.

Sources: Based on the literature (see references). Note: CG = central government; LG = local government. REVITALIZING URBAN GOVERNANCE AND FINANCE 87

coordination. Because central and regional may exist at the lowest tiers in some coun- governments are strong, some coordination tries (communities, wards, village commit- may be handled by mandate from above, but tees), but only over very minor functions and this setup does not preclude the need for other resources. Recent reforms in some countries mechanisms, particularly for more indepen- appear to signal a willingness to increase dent urban governments and as local govern- local autonomy, but much is still to be done ments in general more fully assume their across the board. intended roles. Local governments have authority to prepare their own budgets in most countries, Inadequate subnational autonomy but this power is frequently offset by heavy and discretion dependence on transfers, requirements for budget approval by higher tiers, lack of Decentralization theory holds that effective capacity or incentives to prepare proper local governments need a degree of autonomy budgets, and so on. Local governments are to respond to the demands of constituents, even more constrained in controlling local meet their functional obligations, and support staff and human resource management. the development of their territories. Autonomy In some cases, some or all senior local is never absolute, however, and frameworks, government staff appointments are made by capacity, and accountability must be devel- the national government (Bhutan, Maldives, oped to avoid abuse. Local autonomy is mod- and Nepal) or require central approval est at best in South Asia, making it hard for (Bangladesh). Intermediate officials local governments to act decisively and inde- appointed by the center can also get involved pendently and impeding the development of in or even manage local hiring decisions governance and accountability links between (Afghanistan, Sri Lanka). Local governments local governments and citizens. have somewhat more staff control in India Table 3.5 assesses the degree of autonomy and Pakistan, subject to state and provincial and discretion that local governments are regulations that vary across each country. allowed along four dimensions. The first is an Perhaps the most heavily constrained aspect overall assessment of the general indepen- of local government autonomy is revenue gen- dence of local governments and how much eration. In some cases, all local revenue rates control they are subject to. The strongest are prescribed by the central government autonomy is at the intermediate level in the (Afghanistan, Maldives, and Nepal). In others two federal systems, particularly Indian (Bangladesh and Bhutan) local governments states. Indian states have considerable power have a bit more discretion but must follow over their own affairs as well as over devolu- nationally defined guidelines or secure central tion to lower tiers, and they can constrain the government approval. In India and Pakistan, powers and choices even of important cities local governments may have some formal dis- despite the constitutional amendments in the cretion over setting local tax rates but are gen- 1990s that were intended to create the basis erally subject to strong state or provincial for empowering local bodies. revenue regulations and oversight. Where the unitary countries use intermediate tiers, these tiers tend to have powers The resource defi cit (including formal powers over lower tiers, as in Sri Lanka), but they are substantially A stark urban resource deﬁ cit reinforces the accountable to the central government. The above urban empowerment deficit in two various types of local governments have ways: an overall lack of funds for local public some discretion over certain functions and services and a ﬁ scal gap between the cost of resources in most countries, but they are assigned functions and the proceeds of allow- often heavily influenced by higher tiers and able local revenues. Both problems are hard particular ministries. The greatest autonomy to quantify given the lack of precise functional 88 LEVERAGING URBANIZATION IN SOUTH ASIA

TABLE 3.5 Intermediate tier and local government autonomy and discretion

Hiring and human resource General Budgeting management Revenue Afghanistan Centralized system; CG develops budget; President appoints LG Municipalities have no autonomy community development provinces infl uence district executives; ministries largely to set offi cial tax rates and charge councils (not formal LGs) allocations; municipalities choose staff , but appointed levels, which are determined by have some discretion. have some autonomy, but provincial governors have some the CG. CG appoints mayors. infl uence.

Bangladesh Very limited LG autonomy in CG ministry funding Hiring approved by Local Municipalities set rates and most respects. dominates budgets; LGs Government Division; all LG staff charges based on CG guidelines, rely on opaque and erratic report to CG ministries. but CG can override. transfers.

Bhutan Increasing autonomy, and Lower tiers develop Chief administrative offi cers are Thromdes set rates and charges, greatest at the lowest level. budgets for approval and appointed by CG, which largely may devise new taxes; but all aggregation at higher tiers. controls staffi ng. require CG approval.

India States have strong autonomy LGs prepare their budgets State Public Service LGs have limited revenue and control substate roles; LG subject to state regulation; if Commissions regulate LG hires, autonomy and are constrained autonomy is relatively limited. not indebted they are fairly but specifi c practices can vary by other factors, for example, independent. across states. rent controls.

Maldives Roles defi ned, but some LG All subnational governments Local administrators are LGs only collect fees for services discretion in fulfi lling them. develop their own budgets. appointed by Civil Service (defi ned by CG law). Commission (consults councils).

Nepal Functions are established, Budget is under CG Local offi cials are hired by CG. All main LG revenues are subject but under review for the new medium-term plans; formal to rates prescribed by CG. constitution. autonomy for LG budgets.

Pakistan Much spending is devolved Districts form own budgets District and tehsil governments Limited; city districts and tehsils to the provinces, but per district government can hire personnel as per set property tax (a provincial tax) provinces have been slow to budgeting rules. provincial policy guidelines. rates subject to guidelines. empower districts.

Sri Lanka Low LG autonomy; CG retains Local bodies have some Provinces hire LG staff ; provincial LGs have very little revenue planning function, giving control over budget process chief secretary named by autonomy. wide scope for intervention but frequently lack technical national cabinet; governors in local aff airs. capacity. infl uence appointments and human resource management.

Sources: Based on the literature (see references). Note: CG = central government; LG = local government.

assignments and severe data constraints, but a in decentralized fiscal systems (table 3.6). sense of its severity is evident. This section A number of points can be drawn from avail- provides additional information on the able information. range of revenue sources available to local First, at least one type of local government governments—own-source, intergovernmen- in each country has constitutional or legal tal transfers, and borrowing. rights to use a range of revenue sources. But individual taxes are selectively shared by Own-source revenues higher levels of government only in India, Nepal, and Pakistan. (Practices for sharing Own-source revenues raised directly by local pooled [from multiple revenues] higher-level governments or shared by higher levels by resources with local governments are summa- law or local government discretion are critical rized in table 3.7.) REVITALIZING URBAN GOVERNANCE AND FINANCE 89

TABLE 3.6 Own-source revenues, shared taxes, and local revenue significance

Own-source revenue Own-source revenues Shared revenues importance Afghanistan LGs use property and business No major individually shared LG revenues vary greatly; some taxes, octroi (an entry tax on certain taxes.a (urban) raise half to most of their consumables), and service fees, income, others little. among others; some extralegal fees.

Bangladesh LGs have access to various taxes, No major individually shared Varies: cities raise about 20%, fees, and rentals, but limited use taxes.a union parishads about 14%. except in city corporations.

Bhutan Thromdes and gewogs can levy No major individually shared LGs raise only about 1% of their limited local taxes. taxes.a revenues.

India Urban LGs tax property, CG and states mainly use pooled LGs raise < 3% of public revenues nonmotorized vehicles, professions, transfers (table 3.7), but state (about 33% of spending) and and advertising; octroi formerly practices vary; shared CG tax on 10% of their income; 90% of LG widely used, now abolished in all goods and services has been revenue from urban LGs; variation states but one; rural LGs use minor proposed. high across and within states. fees and charges.

Maldives Atolls, islands, and cities can charge No major individually shared LGs raise very small amounts of fees for services delivered. taxes.a revenue.

Nepal Municipalities have access to CG shares with DDCs (for example, LG own-source revenue was 2% property, entertainment, vehicle, land and housing registration), of total public revenues (about advertising, commercial video, VDCs (natural resources), and 10% of public spending) and and business enterprise taxes; user municipalities (vehicle); a few about 13% of total LG revenues charges; building permit fees. municipal-DDC shares (2012).

Pakistan Vary: districts (education and 2.5% of general sales tax shared to Provincial revenues account for health taxes, licenses, charges); off set loss of octroi. only about 7% of total public tehsil (property tax, entertainment revenues (compared with 35% of fees, permits, cattle markets, user expenditures); LG revenues are charges); union (fees and charges). even smaller.

Sri Lanka Municipalities use taxes on property, No major individually shared Subnational taxes account for less vehicles, and animals; user charges; taxes.a than 7% of total public revenues. and license fees.

Sources: Based on the literature (see references). Note: CG = central government; DDC = district development committee; LG = local government; VDC = village development committee. a. See table 3.7 on intergovernmental transfers fi nanced by pooled revenues from multiple taxes and sources.

Second, virtually all allowable taxes and selected cities of some countries, such as nontax revenues—property taxes, license fees, Bangalore in India (Rao 2013). permits, user charges, and so on—would be Third, the local fiscal (revenue- expenditure) considered appropriate local revenues. Most gap appears to be substantial, demonstrably of them, however, are subject to a degree of so in countries for which data are available. higher-level control in all countries. Many Local bodies in India, for example, account also suffer from inappropriate policy provi- for about 33 percent of public spending but sions, and collection is typically weak. In raise only 3 percent of total public revenues short, even normatively desirable and allow- and 10 percent of total local revenues. In able local taxes are not well used, although Pakistan, local governments account for promising reform efforts are under way in 35 percent of spending but raise only 7 percent 90 LEVERAGING URBANIZATION IN SOUTH ASIA

of revenues. Local governments in Nepal and require intergovernmental transfers, even in Pakistan, respectively, raise 2 percent and the most advanced economies. Table 3.7 7 percent of total public sector revenue. summarizes fiscal transfers from general Fourth, in most countries in the region, revenue or pools of multiple individual rev- urban governments have more revenue enue sources, making a distinction between authority or raise a considerably greater unconditional and conditional transfers. share of their total revenues than do other Unconditional transfers are important in types of local governments. In India, for decentralized systems to augment the fis- example, there is general concern about cal power of autonomous local govern- inadequate exploitation of allowable local ments, while conditional transfers provide a revenue sources, but 90 percent of total local means for higher levels to encourage spend- government revenue is raised in urban areas. ing on basic services that advance national priorities. Conditional transfers target a Intergovernmental transfers particular sector (for instance, health or education) or a specific type of expendi- Local governments globally suffer from nat- ture (for instance, salaries or development ural ﬁ scal imbalances and almost invariably expenditure).

TABLE 3.7 Intergovernmental transfers

Unconditional Conditional Afghanistan No major unconditional transfers; minor funding fl ows to Transfers to provinces are essentially disbursements from line Community Development Committees (not offi cial local ministries to provincial departments; no formal municipal transfers. governments).

Bangladesh Annual Development Program Block Grant is allocated The vast majority of LG transfers are earmarked grants for salaries, by formula; accounts for about 5–20% of revenue, often ministry projects, and specifi c development projects (including aid depending on LG type. projects).

Bhutan Thromdes receive unconditional grants equal to the Various CG agencies provide earmarked grants for particular current or capital defi cit (subject to capital expenditure purposes to LGs. justifi cation). Other levels receive transfers based on a clearly defi ned formula.

India Indian transfers are complicated; a large pool-based, Growing conditional transfers (most from line ministries), formula-allocated federal transfer goes to states; State some allocated by criteria, others less transparent; a key urban Finance Commissions share revenues (from a pool and infrastructure program (linked to reforms) is the Jawaharlal Nehru in some cases specifi c taxes) with lower tiers; federal National Urban Renewal Mission; increasing use of performance- government transfers to lower tiers pass through states. based grants for specifi c purposes.

Maldives CG provides grants to LGs that rarely exceed expenditures No conditional transfers to LGs. on salaries of councilors and staff ; revenue from national facilities is transferred to subdivision of facility location.

Nepal Two unconditional block grants: minimum and a formula- Formal conditional grants to LGs are devoted primarily to education, based municipal grant; though offi cially unconditional, CG roads, and other specifi c local infrastructure. provides directives to target benefi ciaries of spending.

Pakistan Provinces rely heavily on CG unconditional transfers. Ad hoc federal and provincial grants to LGs earmarked for recurrent (salaries, O&M) and capital spending (for example, federal Khushhal Pakistan Program).

Sri Lanka Finance Commission can make ad hoc grants to local Earmarked central transfers go to provinces and LGs, largely for bodies passed through provinces. salaries.

Sources: Based on the literature (see references). Note: CG = central government; LG = local government; O&M = operations and maintenance. REVITALIZING URBAN GOVERNANCE AND FINANCE 91

Assessing the overall situation with trans- Fourth, most countries operate some type fers is complicated by the complexity of inter- of conditional transfer program for local gov- governmental systems, a murkiness in the ernments. Local governments in Bangladesh degree of autonomy of some levels and types and Sri Lanka, for example, receive earmarked of local governments, and the lack of clarity transfers for salaries; Bhutan and Nepal use a on their specific functions. But a few points wider range of conditional transfers for spe- can be made. First, in some countries, one or cific purposes. In India and Pakistan, local more subnational government levels are governments receive various conditional embedded in the budgets of the central or transfers (often through programs of individ- intermediate government (essentially decon- ual line ministries) from both federal and state centrated), for example, Afghan provinces or provincial governments. and Sri Lankan local governments. Other Fifth, some efforts have been made in types of differentiation are also observed; for recent years to promote reform through example, only the four largest thromdes in compliance- or performance-based grants in Bhutan are self-governing, with the others the region. Some of these grants are simply under the oversight of the dzongkhags. special cases of regular conditional transfers, Second, a majority of South Asian coun- while others are or were broader programs of tries have some type of formula-allocated various scales (for example, Bangladesh and unconditional transfer. These transfers range Nepal; in Nepal, they were absorbed into the from large allocations from the federal gov- unconditional transfer system). The ernment to intermediate tiers in India and Jawaharlal Nehru National Urban Renewal Pakistan to much more modest or minor Mission in India is an example of a very large transfers to local governments in Bangladesh, program intended to support critical urban Bhutan, Maldives, and Nepal. Even if trans- government expenditures while creating fers are officially unconditional, there are incentives for the adoption of much-needed often higher-level rules and “guidance” placed reforms. Critics maintain that these on use. In some cases, minor block grants performance-based grants, although poten- flow to levels or bodies below official local tially desirable in theory and intent, are often governments, such as Community in practice treated too separately from regular Development Committees in Afghanistan, but operations and suffer from weaknesses in the municipalities in this case receive no trans- design and implementation. fers at all. Potential concerns arise in all countries about the effects of unconditional Subnational borrowing transfers, for example, whether they undermine incentives for local revenue generation, Local governments, especially in urban whether they in fact fairly treat the neediest areas, have great needs for infrastructure areas, and so on. Limited evidence reinforces ﬁ nance. As local government systems mature these potential concerns, but better informa- and decentralization advances, subnational tion and analysis are needed. borrowing can become an important source Third, in cases where the national (federal) of funds for urban infrastructure develop- government focuses on transfers to intermedi- ment. International experience indicates a ate tiers (India and Pakistan), these tiers often well-defined borrowing framework is have discretion over what to share and how to needed to ensure fiscal responsibility. share it with lower tiers. Sharing is often done Subnational borrowing is generally under- using clear criteria or a formula. Empowering exploited and not very advanced in the intermediate tiers with this role can result in region (table 3.8), but a few basic observa- considerable differential treatment of local tions can be made. governments among states and provinces (or First, except for Afghanistan, some type of even within them), which may be justifiable local government borrowing is legally allowed but can also cause problems. throughout the region. The nature of 92 LEVERAGING URBANIZATION IN SOUTH ASIA

TABLE 3.8 Subnational borrowing frameworks

Afghanistan Municipal borrowing (domestic or foreign) through Ministry of Finance sources is allowed by the Subnational Governance Policy and the Municipalities Law (2000), but exact requirements are not defi ned. The Public Finance and Expenditure Management Law states that cities can only borrow from the CG; in practice, cities have been borrowing but from unknown sources.

Bangladesh LG borrowing from external sources is allowed with CG approval but rare. Urban LGs can and do borrow from the Bangladesh Municipal Development Fund.

Bhutan Thromdes may borrow funds through the Ministry of Finance or with its approval.

India LG borrowing is governed by the Local Authorities Loans Act (1914) and increasingly accessed from multiple sources, including bonds. Urban LGs can borrow from the market without federal or state guarantee if they meet specifi c criteria. A Pool Finance Development Fund is intended for smaller LGs, and some states have their own mechanisms. Urban LGs dominate local borrowing. The Income Tax Act was amended to allow urban LGs to issue tax-free bonds.

Maldives All levels of subnational government are technically empowered to borrow—from other LGs, banks, and international fi nancial institutions—and to issue fi nancial instruments, such as bonds and securities. To date there has been no borrowing.

Nepal Municipalities can, in principle, borrow using collateral or CG guarantees. Borrowing from foreign governments or entities requires CG approval. To date there has been no borrowing.

Pakistan Provinces are allowed to borrow with informal CG monitoring. Outstanding provincial debt is about 5% of GDP. Metropolitan corporations can borrow with CG approval.

Sri Lanka Municipal councils may secure bank loans and issue bonds with approval from the Minister of Local Governments and the relevant provincial council, subject to certain criteria. The CG runs the Local Loan and Development Fund, which off ers loans for local infrastructure investment at below market rates.

Sources: Based on the literature (see references). Note: CG = central government; LG = local government.

borrowing, however, varies. In a number of in most countries. These requirements relate cases, the scope is very limited (for example, to general financial practices as well as to spe- in Bhutan it occurs only through, or with cific indicators of overall local government approval of, the Ministry of Finance, and cen- debt levels and debt-service burdens. Most of tral government approvals or guarantees are the borrowing frameworks need further required in Bangladesh, Nepal, Pakistan, and development and require better mechanisms Sri Lanka). In some of these countries, only for monitoring and enforcement. certain types of local governments may bor- Third, several South Asian countries have row: thromdes in Bhutan, municipalities in special mechanisms for lending to local Nepal and Sri Lanka, and metropolitan cor- governments unable to directly access finan- porations in Pakistan. In practice little evi- cial markets; for example, a Municipal dence can be found of significant borrowing Development Fund in Bangladesh and a Local in most countries, although it is increasing in Loan and Development Fund in Sri Lanka. India in a few of the larger urban bodies. The Some states in India, including Tamil Nadu, lack of borrowing for critical infrastructure have established dedicated local government investment is a missed opportunity for South lending mechanisms. Problems often arise in Asian urban governments to play their managing such funds, and they can be heavily increasingly needed developmental role. influenced by politics, yet they can play a key Second, some type of framework must be role in countries where local governments in place and certain conditions must be fol- need to develop creditworthiness before lowed before local governments can borrow accessing capital markets. Other mechanisms, REVITALIZING URBAN GOVERNANCE AND FINANCE 93

such as the Pooled Finance Development systems are rarely strong, but various attempts Fund in India, can also open the borrowing are under way to improve them. In many cases door to smaller or fiscally weaker local gov- local government PFM systems have received ernments that could not otherwise borrow on little dedicated attention, but some newer their own. Also in India, the Income Tax Act reforms (Bhutan) or cases in which donors has been amended to allow urban govern- helped build certain elements of local systems ments to issue tax-free bonds. (for example, Bangladesh, Nepal) may have left PFM in a somewhat better position. The accountability defi cit Still, such mechanisms may not be fully implemented or institutionalized or are hindered by Beyond basic structures and decentralized ﬁ s- weak capacity. In India, state municipal acts cal functions, a range of governance and define procedures for local government PFM. accountability weaknesses require attention if Links between development plans, public urban governments are to function well and investment programs, and annual budgets sustainably. Of critical concern are local gov- have long been a weakness of public sector ernment administrative, managerial, and management in the region, and there are no oversight mechanisms that provide consis- exemplary cases of surmounting this conse- tency and transparency and allow for appro- quential challenge. Reform attempts occur, priate levels of the upward reporting and but are seldom well implemented or are accountability that are essential even in somewhat separated from formal govern- decentralized systems. Equally important are ment processes. And in many cases plans provisions for downward accountability remain more like wish lists or are heavily mechanisms—electoral and nonelectoral— influenced by higher-level agents. Institutional that create space for citizens to place demands fragmentation (including in metropolitan on elected local governments. areas) and the lack of interjurisdictional cooperation exacerbate the problem. In some Administrative and managerial countries, features of the institutional frame- mechanisms work reinforce the lack of integration and create a mindset that permeates government Table 3.9 provides basic information on operations—for example, the long-time sepa- accountability mechanisms in local govern- ration between the National Planning ment administrative and management sys- Commission and the National Finance tems, including the state of formal public Commission in India before the former was financial management (PFM) systems, the abolished in 2014. extent to which planning and budgeting Each country has some type of audit mech- mechanisms are adequately linked, whether anism led by a national audit institution, audit mechanisms are in place, and whether a often with a mandate for audit of both local framework for public-private partnerships and higher-level governments. In some coun- (PPPs) has been developed. Given the rela- tries, such as Bangladesh, audits must be com- tively weak or nascent state of decentraliza- pleted by private auditing companies for local tion in South Asia, the extent to which these governments to qualify for intergovernmental mechanisms exist and are used can vary transfers. Even legally mandated audits, how- extensively, and there is considerable room ever, are not always conducted as prescribed for improvement in virtually all cases. or with an adequate quality level or fre- Formal administrative accountability and quency. In particular, private audits are not local government managerial systems are gen- reviewed by the supreme audit institution. erally present in the region, but many are rela- Thus, the results of audits likely do not have tively weak or not well used, especially in any obvious consequences for poor perform- countries new to decentralization and in areas ers. However, because audits are a potentially with particularly limited capacity. PFM important source of information to be used 94 LEVERAGING URBANIZATION IN SOUTH ASIA

TABLE 3.9 Administrative accountability and managerial mechanisms

Plan-budget Formal PFM systems linkages Audit mechanisms PPP framework Afghanistan No clear LG PFM system and Provincial plans and budgets No formal LG audit PPPs legal but limited; some limited formal reporting not well linked; municipal mechanisms. mayors independently requirements. plans lacking. exploring options.

Bangladesh Central PFM weak but Budgets (especially cities) Annual external audits of Recent PPP policies but only at improving; local PFM capacity target accessible CG funds; municipal budgets mandated national level; PPPs limited in is very low. weak formal plans. but uneven. practice.

Bhutan LGs use a multiyear fi nance Planning and budgeting CG is supposed to audit LGs MoEA issued initial framework framework; well monitored; linked, but LGs take liberties in every two years. for private participation in capacity limited. execution. infrastructure.

India State municipal acts defi ne Formal linkages between LG Supreme Audit Institution Evolving national framework, procedures for LG PFM; some plans and budgets have been inspects public (including but only a few states have urban movement from cash to challenging to develop. LG) accounts; also lower-level well-developed PPP laws and accrual accounting. audit. frameworks.

Maldives LGs fall under CG PFM legal Limited planning capacity; Supreme Audit Institution and PPPs implemented on an ad framework, including Public budgets rarely cover more CG LG Authority have the right hoc basis. Finance Act. than salaries. to audit any LG accounts.

Nepal MoLG manages three-year Periodic fi ve-year plans and The Supreme Audit Institution National Planning Commission implementation of formal annual plans, which provide conducts audits of block issued a White Paper on Public LG public expenditure and the basis for expenditure grants to the district level; Private Partnership (2011); CG fi nancial accountability budgets. below the district, audits are has PPP policy committee; enhancement plans. conducted by registered or LG law mandates LG PPP private auditors. committees.

Pakistan Financial reporting and Annual budget plans; Auditor General has audit No overall LG framework, but audit reform under way; also execution and links to other authority over provincial and some provinces (for example, provincial PFM regulation. levels weak. local accounts. Punjab) have laws on PPP.

Sri Lanka Ministry of Finance and Medium-Term Expenditure Auditor General authorized to No PPP framework; Planning regulates PFM, Framework links plans audit subnational government procurement rules are part of management of assets, and and budget; center drives, accounts. overall fi nancial regulations. procurement. feedback mechanisms weak.

Sources: Based on the literature (see references). Note: CG = central government; LG = local government; MoEA = Ministry of Economic Aff airs; MoLG = Ministry of Local Government; PFM = public fi nancial management; PPP = public-private partnership.

by higher-level governments and citizens to productive initiatives in some countries, and monitor local government performance, they remain a potentially desirable means for extending their use and improving their qual- urban government service provision. ity would be valuable. PPPs are receiving increasing attention and Subnational elections appear to be legal in all countries. However, not many dedicated policies for local PPPs are Local government elections are a cornerstone evident. A few countries, such as Nepal, have of devolution. In South Asia, many countries specific mandates to promote them, and in are in varying stages of transition to at least India and Pakistan a number of states and partially devolved systems, so there are mul- provinces have developed PPP frameworks tiple types of subnational elections at different for lower tiers. PPPs remain relatively limited levels (table 3.10), although not at deconcen- in practice, but there are instances of trated tiers with purely administrative roles. REVITALIZING URBAN GOVERNANCE AND FINANCE 95

TABLE 3.10 Subnational elections and assemblies

Extent of political Level competition Direct election of mayor or head Afghanistan Only provincial, but law allows Many competing parties, but system Municipal mayors are appointed. municipal and district; community favors independent candidates from development councils (not LGs) specifi c districts. elected.

Bangladesh Upazila and union parishads, Two main parties, but additional All local government chairpersons and mayors pourashavas, city corporations. smaller ones exist. are directly elected.

Bhutan All levels since 2011, but some issues Candidates are not allowed to be Mayors are directly elected. (for example, low turnout of women). members of political parties; few candidates in fi rst local elections.

India State (some bicameral) level and High (many parties), but the situation The election of mayors is direct in some states, various substate levels, including in varies across areas; a few members are indirect in others. municipal bodies and in three-tier nominated to local councils. panchayati raj (rural local bodies) in each state with population > 2 million.

Maldives All atolls, cities, and islands elect local Multiparty system; three main parties; The elected councilors elect the mayor or councilors. competitive elections. president.

Nepal LG elections not held since 2002. National electoral competition is high Mayors are selected by local committees. with 122 registered parties.

Pakistan Provincial and local; three-tier LG Strong political competition often Councils select chief executive offi cer, mayors, system; union council directly elected; along clan or tribal lines. and deputies in tehsils and districts; some (reserved number of seats for women provinces can, after due process, dismiss LG and minorities). offi cials.

Sri Lanka Council elections at provincial, National parties and alliances control President appoints provincial governors; majority municipal, and rural levels. nomination; mobilization and voting party nominates urban LG chairs. often high.

Sources: Based on the literature (see references). Note: LG = local government.

All countries with state or provincial tiers The extent of political competition in the have elections at those levels, but the use of region also varies, with a few countries, such local elections is more mixed. In some coun- as Bangladesh and Maldives, dominated by a tries, such as Bhutan and Maldives, elections few political parties. Other countries, such as are held at all levels. In other countries, such Afghanistan, India, Nepal, and Pakistan, have as Bangladesh, Pakistan, and Sri Lanka, elec- many political parties, some with ethnic or tions are held only at selected (including religious associations with influence that var- urban) subnational levels. In a few countries, ies by area. A number of specific rules or con- there are no local elections. In Afghanistan, ditions are reported; for example, candidates for example, municipal and district elections in Bhutan may not be members of political are provided for in legislation, but they have parties, and local nominations in Sri Lanka yet to be held. Political turmoil has pre- are controlled by national political parties. vented local elections in Nepal since 2002, Another commonly accepted indicator of but they are expected to be called when local democracy is whether local government there is political settlement around a new mayors are directly elected. In Bangladesh constitution. and Bhutan mayors are elected, but it is more 96 LEVERAGING URBANIZATION IN SOUTH ASIA

common for South Asian mayors to be for citizens to discipline local government selected by elected councilors, as in Maldives, performance. Pakistan, and Sri Lanka. In India, the practice varies across states, and all municipal mayors Transparency and nonelectoral are appointed in Afghanistan. accountability mechanisms In short, the state of local electoral democracy in South Asia is uneven, but many coun- Although local elections are a core gover- tries are increasing efforts to promote it. nance priority in devolved systems, they are Local elections, of course, can be captured by considered a blunt instrument of accountabil- local elites or offset by other factors, so even a ity that should be complemented by other seemingly good electoral system is no guaran- means of promoting transparency and down- tee of the downward accountability needed ward accountability. Table 3.11 illustrates

TABLE 3.11 Transparency and nonelectoral accountability mechanisms

Right to or freedom of information law Input mechanisms Feedback mechanisms Afghanistan None. No formal LG civic input process; No formal feedback means; Subnational community development councils use Governance Policy has participatory participatory planning and budgeting provisions but does not require (role unclear). implementation.

Bangladesh LG Act 2009 and RTI Act 2009 require 2009 law created shava (ward) and local Shava meetings allow LG feedback but publishing of plans, budgets, and committees; shava decisions require may be vulnerable to elite capture, and the like, and protect citizen access to LG review; committees have one LG CG has fi nal authority over use. public documents. representative and four to six others.

Bhutan RTI law in process; LGs must publish No direct participation, but LG sessions No information available. agendas, development and work are public; plans and budgets are bottom plans, and budgets. up; lower tiers submit to higher tiers (unclear eff ect).

India Powerful 2005 national RTI Act and RTI Some push for participatory processes Some feedback mechanisms exist, but Acts in most states; several states have (for example, 11th Five-Year Plan) they vary widely in form and whether adopted disclosure acts for urban LGs. and some LG eff orts; but uneven, driven by LG or civil society. underdeveloped, or underused in many areas.

Maldives 2014 RTI Act; Commissioner of LG Act (2010) mandates atoll councils to Ward meetings are held, and council Information to be appointed to promote citizen involvement in planning meetings are open to the public. enforce act. and development programs.

Nepal 2007 RTI Act requires access to Various mechanisms at ward level and Good Governance Unit reviews information and documents of public to link levels, but participation not yet corruption charges; large number of social importance. strong, and subject to some political accountability initiatives. interference.

Pakistan Strong FOI and RTI laws in Punjab Citizen consultation is required before Punjab has instituted large-scale proactive and Khyber Pakhtunkhwa apply to budget letter is issued, but limited feedback mechanisms; smaller pilots by LGs; law in Sindh is weak; general interaction in practice. some donors and NGOs. implementation challenges.

Sri Lanka RTI Act drafted and expected to pass Citizens can submit project requests in CG issued a Citizens’ Charter to support in 2015. budget planning. local grievance mechanisms.

Sources: Based on the literature (see references). Note: CG = central government; FOI = freedom of information; LG = local government; NGO = nongovernmental organization; RTI = right to information. REVITALIZING URBAN GOVERNANCE AND FINANCE 97

selected mechanisms that provide information Asia, but several countries have some experi- to citizens and empower citizens to inﬂ uence ence with them. Citizen report cards, in par- and evaluate local governments. ticular, have been a focus of considerable Most South Asian countries have passed experimentation by both governments and right to information or freedom of informa- civil society groups, particularly in parts of tion acts, including Bangladesh, Maldives, India, and have also been piloted in Nepal, India (some Indian states also have local Pakistan, and Sri Lanka. government disclosure acts), Nepal, and Right to services laws have emerged in Pakistan. Bhutan and Sri Lanka have draft Indian states in the past few years as a poten- laws (with unclear traction in Sri Lanka). tially important mechanism for holding local Only Afghanistan has apparently not made governments accountable. This type of legis- efforts on this front. Some laws are, of course, lation sets out service standards and provides stronger than others: Pakistan, for example, legal recourse to citizens who have not been has many exemptions to public access. More receiving the legally mandated quality of ser- generally there is a long way to go before vice. Such legislation has helped prompt the these, mostly relatively recent, laws are development of citizen service centers with broadly implemented, understood, and have higher levels of responsiveness and account- the desired effects. ability for performance. Other potential enablers of local accountability include mechanisms that offer citizens Where empowerment, a way to engage in local government decision resources, and accountability making (for example, participatory planning meet: Implications for urban and budgeting). Attempts have been made to service delivery develop or encourage means for citizen input in all South Asian countries, but they vary The general intergovernmental institutional greatly by level (many are at the community structures, fiscal arrangements, and gover- or village level); degree of formality (for nance mechanisms reviewed above—and the example, organized deliberations as in three deficits—collectively determine how Bangladesh versus general access to commit- well urban governments can meet their core tee and council meetings as in Bhutan); stage services provision role, take charge of their (for example, proposing development projects territorial development, and develop produc- as in Sri Lanka versus commenting on plans tive relationships with other governmental and budgets prepared by local governments and nongovernmental actors. This section as in Nepal); and demonstrable impact on tentatively explores how these factors may (or local government decisions. do) affect urban governments, with a focus The lack of systematic assessments allows on selected services. no firm overall conclusions to be made, but Although some cities perform relatively many relevant mechanisms in the region well, South Asian urban governments overall appear to be relatively weak and pro forma. do not adequately provide key public services There is anecdotal evidence of positive needed for economic and social development. experiences—some mechanisms have been The region ranks ahead of only Sub-Saharan reported to be at least somewhat effective in Africa on the share of the urban population some localities—as well as weak, ineffective, with access to improved water sources, and it or token practices. Mechanisms that allow ranks last on access to improved sanitation. citizens to lodge petitions and provide feed- Moreover, service provision typically becomes back on the performance of their local gov- more challenging as cities face increases in ernments (for example, tax-liability appeal size, population, and the range of demands mechanisms, complaint bureaus, citizen resulting from higher population diversity. In report cards) appear to be less prevalent than much of South Asia the rapid influx of urban participatory mechanisms in much of South residents is overwhelming the ability of cities 98 LEVERAGING URBANIZATION IN SOUTH ASIA

to provide basic services. For instance, access provision. But the table shows that South to improved sanitation for South Asian urban Asian urban governments play a relatively residents has declined from 61 percent to only modest role (note the few gray shaded cells 10 percent during the past two decades. where urban governments have strong Improvements in urban road networks also authority). Evidence suggests they compare lag behind the pace of population growth.5 poorly with major cities in other regions No “best practice” approach to structur- (table 3.13).6 ing urban governance for effective services Most urban governments in South Asia delivery exists. Suitable structures vary have primary responsibility only for solid greatly, and comparisons are hindered by waste (but they lack full administrative con- idiosyncrasies across and within countries. trol even of that). India’s cities have financing The identification of institutional arrange- and administrative authority for local roads, ments that work well is further hindered by but the situation is mixed elsewhere in the the fact that such arrangements cumulatively region. Financing is included in local budgets and reactively evolve. Problems with the road in most countries, but mainly as earmarked network in Dhaka, for example, result from transfers. years of underinvestment and inadequate Private operators generally deliver local planning at least as much as from current public transport, with higher-level govern- institutional and fiscal arrangements. ments setting routes and licensing operators. Within the diversity of how services deliv- Given the local nature of the service and the ery is organized, the institutional framework modest technical expertise required, urban is critical: resources flow through and gover- governments could learn to perform these nance mechanisms are embedded in this functions. Of the four services, water and sew- framework, shaping the larger process of erage is technically the most complex. In large turning resource inputs into urban public ser- cities in Bangladesh, India, Nepal, and Sri vices that respond to citizens’ needs. The Lanka, centrally created and managed water framework provides incentives for services and sewerage boards run services with limited providers to perform, particularly through involvement from urban governments. defined lines of accountability to government The centralized control over staffing oversight bodies and to users of services. touched on above is a major issue for urban The institutional framework for public services delivery. In Bangladesh, Nepal, and services delivery in the South Asian urban Sri Lanka, the central government hires and landscape is largely one of functional fragmen- fires key staff members who oversee delivery tation, that is, there are commonly dedicated of these four services, and states and prov- single-purpose agencies for public services inces play a significant role in India and with substantial operational autonomy. Such Pakistan. In cases in which services are deliv- autonomy allows for greater technical special- ered by a special-purpose entity, such as water ization and capacity as well as a degree of and sewerage authorities in big city corpora- insulation from unwarranted political interfer- tions in Bangladesh, administrative control is ence, but the fragmentation impedes urban exercised entirely by the central government. autonomy and broader coordination. Functional fragmentation is not necessarily Table 3.12 illustrates how functional frag- undesirable. A dedicated entity controlled by mentation dominates the way five South the central government may allow pooling of Asian countries provide four basic urban ser- technical and human resources across juris- vices. These services are usually considered dictions and provide a measure of insulation legitimate local functions given their limited from potentially problematic local politics. externalities. In addition, benefits from these But an institutional framework in which ser- services accrue primarily to city residents, vices delivery is fragmented among agencies which is why urban governments would be with limited local government authority can expected to play a major role in their have negative consequences as well. The TABLE 3.12 Organization of services delivery in Bangladesh, India, Nepal, Pakistan, and Sri Lanka

Bangladesh India Nepal Pakistan Sri Lanka Municipal 11 city corporations, 315 138 municipal corporations, 58 municipalities 5 city district governments 23 municipal councils (MC) units pourashavas 1,595 municipalities, 2,108 town (CDGs), 6 municipal or councils metropolitan councils (MECs) Solid waste management Function M M M M M Municipalities de jure and de Municipalities de jure and de Municipalities de jure and de Municipalities de jure and de Municipalities de jure and de facto responsible; delivery mix facto responsible; delivery mix of facto responsible; delivery mix facto responsible; delivery mix of facto responsible; delivery mix of of public and private operators public and private operators of public and private operators public and private operators public and private operators

Financing M M M M CG, M Municipalities cover capital and Municipalities cover capital and Municipalities cover capital Municipalities cover capital and Municipalities cover capital and operating costs with own funds, operating costs with own funds; and operating costs with own operating costs with own funds; operating costs with own funds, transfers, contracts; set fees authorize fees funds, authorize fees PG authorizes taxes and fees but CG pays staff

Administrative CG, M SG, M CG PG CG, M control CG appoints chief executive SG appoints head of department; CG appoints all heads of local PG appoints senior staff and CG appoints staff , but MC decides offi cer and chief inspector, municipalities other staff ; administration and handles oversees operators on exact posting; MC handles who hire other staff ; municipalities oversee planning performance monitoring and planning and performance municipalities review plans and performance, with SG input; evaluation framework and performance, but lack limited CG oversight power to sanction Water and sewerage Function CG, M SG, M CG PG, M CG, M Dhaka and Chittagong have Municipalities de jure responsible CG Kathmandu Upatyaka Depending on PG, either MC is de jure responsible, but autonomous water and but signifi cant variety de facto; Khanepani Limited (KUKL) municipalities or PG de jure and National Water Supply and sewerage authorities (WASAs) delivery mix of municipalities delivers services in Kathmandu de facto responsible Drainage Board is de facto overseen by CG; smaller and M and SG utilities Valley responsible municipalities have own CG Nepal Water Supply departments for piped water; Corporation (NWSC) performs provision often nonpiped, PS functions elsewhere

Financing CG, M SG, M CG PG CG CG fi nances WASAs’ capital Municipalities and municipal KUKL and NWSC set and collect PG covers capital and operating CG covers capital and operating and operating costs, sets fees; utilities set and collect fees and water and sewerage fees costs with own funds; authorizes costs with own funds; sets fees municipalities fi nance capital cover capital and operating costs for O&M costs; capital costs fees and operating costs (often with own revenues and transfers paid by municipalities and from CG transfers) and set fees community contributions (continues next page) 99 100

TABLE 3.12 Organization of services delivery in Bangladesh, India, Nepal, Pakistan, and Sri Lanka (continued)

Bangladesh India Nepal Pakistan Sri Lanka Administrative CG, M SG, M CG PG, M CG, M control CG appoints WASA staff and SG appoints all senior staff ; low- CG appoints staff of both KUKL PGs fully oversee CDGs and CG appoints staff and decides head offi cers of municipal level staff hired by organization; and NWSC; appoint all senior staff for all performance framework; planning departments; low-level SG utility oversight is shared by municipalities on board of municipalities; MECs that deliver is handled by CG in consultation staff are appointed by SG and municipality; municipal KUKL and infl uence planning; services will have more low- with MCs on a project basis municipalities utility can be fully autonomous otherwise CG oversees; CG fully level staffi ng, planning, and CG, municipalities control oversees NWSC monitoring control WASAs, own departments

Local roads Function M M M PG, M M Municipalities de jure and de Municipalities de jure Municipalities de jure and de PG and municipalities share de Municipalities de jure and de facto responsible responsible; delivery mix of facto responsible jure and de facto responsibility; facto responsible municipal and independent PG delivers main and outer roads, agencies municipalities inner and small roads

Financing CG, M M M PG CG, M Municipalities cover capital Municipalities cover capital and Municipalities cover capital PG covers capital and operating Municipalities cover capital and and operating costs with operating costs with own funds, and operating costs with own costs with own funds operating costs with own funds, own funds, authorize fees; CG authorize fees funds, Nepal Roads Board but CG pays staff approves budget funding, transfers, loans from Town Development Fund, and cash contributions from residents

Administrative CG, M M CG, M PG PG, M control CG appoints head of Municipality handles staffi ng, CG appoints head of municipal PG appoints all staff PG appoints staff ; municipalities department and undertakes planning, and performance departments; municipal handle planning and performance most human resource monitoring when providing departments handle planning framework functions; municipalities service and performance monitoring handle planning and performance monitoring

TABLE 3.12 Organization of services delivery in Bangladesh, India, Nepal, Pakistan, and Sri Lanka (continued)

Bangladesh India Nepal Pakistan Sri Lanka

Local public transport Function CG CG, SG, M CG, M PG CG, PG CG is de jure responsible; PS Varied arrangements and Delivery by PS with oversight PG de jure responsible; PG and PS CG and PG are de jure responsible; delivers service fragmentation across public from CG and municipalities deliver services both public and PS deliver services and PS

Financing CG CG, SG, M CG, M, PG CG CG sets fee levels, but in Municipalities cover capital Financing comes from PS and PG covers capital and operating Operators cover capital and practice, PS charges at whim; and operating costs with own fees collected by operators; costs with own funds; PS operating costs with own funds; CG has access to international revenues and transfers; SG and municipalities cover costs (for provides for own funding when fees negotiated with and collected loans for capital investment CG set fees, collected by operator example, depots) with own contracted by PS, which provides own revenues fi nancing

Administrative CG CG, SG CG, M PG CG, PG control CG handles route permits SG handles route permits, CG is responsible for permits, PG appoints all staff , issues PG issues licenses and route and licenses, regulations and licenses, and staffi ng; SG and licenses, overall planning, and route permits and licenses, permits; operators hire and performance, and staffi ng; CG handle strategic planning, traffi c police; municipalities and undertakes planning and appoint their own staff ; CG and PG private companies hire staff ; regulation, and monitoring; plan and manage parking and monitoring functions; PS hires handle planning and performance municipalities on CG boards municipalities have input into bus facilities and nonmotorized own staff monitoring but little direct infl uence planning and performance transport; private operators hire monitoring own staff

Sources: Based on the literature (see references). Note: Gray shaded cells indicate that urban governments have strong authority. CDG (Pakistan) = city district government; CG = central government; KUKL = Kathmandu Upatyaka Khanepani Limited; M = municipality; MC (Sri Lanka) = municipal council; MEC (Pakistan) = municipal/metro council; NWSC = Nepal Water Supply Corporation; O&M = operations and maintenance; PG = provincial government; PS = private sector; SG = state government; WASA = water and sewerage authorities. 101 102 LEVERAGING URBANIZATION IN SOUTH ASIA

TABLE 3.13 Functional responsibilities in selected global metropolitan areas

Buenos Addis Dar es Function Mumbai Istanbul Jakarta Manila São Paulo Aires Bogotá Ababa Salaam Kampala Johannesburg Roads and streets C C M C M, C M, C M, C R, Z, C M, C M C Public transport N, P, C M N, M PS M, C M M M M M C Water and sewerage C, R, P M M PS M, C M, C M, C M M M C Solid waste P, C M, C M M, C M, C M, C M, C M N, M M C

Source: Sud and Yilmaz 2013. Note: C = city government; M = metropolitan government; N = national government; P = provincial or state government; PS = private sector; R = regional government; Z = zonal government.

region’s experience with services delivery and service entities rather than a unified, locally the inability of most South Asian cities to driven vision. Local initiative and ingenuity keep up with their growth suggest several are fairly limited and ad hoc because there are concerns. few avenues for their emergence. Higher-level First, the diffusion of functional responsi- governments may enthusiastically promote bility, financing power, and administrative development in cities, as in Sri Lanka, but control among levels of government compli- top-down initiative is not likely to foster suf- cates effective urban management. General ficient local dynamism to address urban prob- rules and standards for revenue generation lems effectively or in accordance with local and personnel management, for example, are preferences. certainly appropriate, but they need to allow Fourth, lines of accountability to citizens a workable measure of local discretion to deal of urban areas are generally weak, limiting with unique and evolving local conditions. incentives for local government performance. Second, overall urban development is hin- Service provision depends on relationships dered by the lack of strong coordinating enti- between the citizens who are users of services, ties at the city level that consider service the service providers themselves, and the gov- provision holistically. There are some cen- ernment entities that mediate between the trally managed urban development authori- two (responding to citizens and delegating ties, but these entities tend to focus on capital authority and financing to service providers). investments rather than the planning and This process is inherently challenging, but if delivery of core services. Although South the overseeing government is not in the local- Asian countries have planning systems and ity, citizens will be hard-pressed to exercise mechanisms that provide some notional their voice effectively and trigger robust means for coordination, these systems are accountability. often tied to individual service areas and specific funding mechanisms, weakening overall Bridging gaps in urban systematic coordination of services that must empowerment, resources, work synergistically. and accountability Third, a striking feature of South Asian cities is the common absence of powerful per- A number of South Asian countries and indi- sons or entities at the local level to help drive vidual cities have taken positive steps in recent development. Mayors and councils are usu- years to improve urban governance and ally not very prominent (and are not always finance. On balance, however, South Asian elected); even plans of relatively well-off or urban governments face considerable con- well-developed cities tend to be based on a straints that limit their ability to assume the superficial amalgamation of plans by various leading role they could play in promoting local REVITALIZING URBAN GOVERNANCE AND FINANCE 103

and national development. A conspicuous triad capacity concerns, but greater clarity is of consequential defi cits in urban governance generally needed to improve the targeting and fi nance—in empowerment, resources, and of public resources and to enhance local accountability—requires urgent attention. accountability. • Reduce unwarranted higher-level interference in legal and legitimate local functions. The empowerment defi cit Some guidance and oversight are always Every country in the region has adopted some valid in decentralized systems, but they are type of decentralization and intergovernmen- primarily supposed to ensure sufficient tal framework that appears to empower local attention to national needs and provide a governments, typically with at least some spe- supportive climate in which urban govern- cifi c provisions for urban areas. These provi- ments can make their own decisions more sions, however, are often weak or only partly effectively. They are not intended to facili- implemented, and local shares of public tate central dominance of local decisions, spending are relatively low or shaped by which tends to weaken incentives for pur- higher levels in all countries. In many cases, suing an integrated urban development service delivery is fragmented among multiple vision and to undermine the development actors, and higher-level agencies get involved of genuine local accountability. in services that are or should be primarily • Develop more robust mechanisms and urban government responsibilities. opportunities for interjurisdictional coop- Coordination mechanisms across actors, if eration in planning, fi nancing, and deliver- they even exist, are generally weak or not suf- ing urban services. Such coordination can fi ciently well used. be an integral part of effective service deliv- In all South Asian countries, urban govern- ery and can also help enhance opportunities ment autonomy, an essential foundation for for and the effi ciency of urban fi nancing. effective decentralized governance, is constrained by central governments (and in fed- These three steps are by necessity primarily in eral and some unitary countries by state and the realm of central (and in some cases inter- provincial governments). These restrictions mediate) governments: empowerment of local take many forms, from explicit control over governments inherently involves action on the urban plans and budgets, urban revenue gen- part of higher levels, whose authority will be eration, urban hiring decisions, and personnel reduced and whose role will remain critical management to more subtle and even infor- but will be modifi ed under decentralization mal interference by higher-level actors in the reforms. Yet in some cases, urban govern- operations of urban governments. Such ments may be able to take independent restrictions may be well intended, but empow- actions to improve performance. erment and autonomy are increasingly recognized to be critical for local accountability, The resource defi cit the foundation for realizing the potential benefits of decentralization. Although hard formal evidence is limited, The weak state of empowerment of urban powerful indications suggest that urban gov- governments is pervasive and consequential, ernments in South Asia do not have enough but there are steps that can be taken to help resources to meet their obligations and behave improve matters: like genuine local governments. Local own- source revenues are limited in most countries, • Promote greater clarity in functional although urban governments typically do bet- assignments among levels and types of ter than other types of local governments. government and other providers. Some Many potentially important revenue sources variations will always be necessary for for urban governments, such as property dealing with special conditions and taxes and user charges, are allowed, but they 104 LEVERAGING URBANIZATION IN SOUTH ASIA

are poorly administered, and collections are between unconditional and conditional weak (with limited instances of better perfor- funding, and reduce any disincentives for mance). Rarely do South Asian urban govern- urban government own-source revenue ments have much control over local revenue generation and borrowing. Transfers can bases or rates. also be used to create incentives for adopt- Some South Asian countries receive gener- ing needed reforms and improving ser- ous intergovernmental transfers, but often vice delivery, a potentially productive they are insufficient to supplement their weak approach that some South Asian countries revenue bases. In a few cases urban govern- have experimented with in the form of ments are embedded in national budgets, performance-based grants. making transfers more like sectoral budget • Develop a framework for local borrowing. allocations. There are issues in many South Well executed, such a framework would Asian countries with the structure of trans- encourage fi scal responsibility and provide fers, including the balance between condi- for an appropriate range of options— tional and unconditional, the criteria and public, private, and mixed—that meet the formulas used to allocate the transfer pool, diverse and growing needs of South Asian and disincentives for local revenue generation urban governments for development or borrowing created by transfers. fi nance. With few exceptions, opportunities for urban government borrowing are restricted. Again, many of the fundamental reforms on Access to private market sources is particu- the resource front require action by higher larly restricted by a combination of national levels of government. At the same time, reve- regulations and weak creditworthiness. Low nue generation is often an area in which moti- levels of borrowing cause significant prob- vated urban governments can work within lems in light of the considerable backlog of existing frameworks to improve collection (and growing need for) urban infrastructure and yields. Actions to improve yields will be investments. most effective, however, if revenue increases The role of own-source revenues in a are associated with service delivery improve- decentralized system cannot be overstated. ments and other measures supported by local Taxes and fees paid to urban governments are businesses and residents. the foundation of the social contract between these governments and their constituents. If The accountability defi cit businesses and residents are willing to contribute to the costs of local public services, Most countries in South Asia have developed urban governments are doing their job. frameworks for urban management func- To improve the financial status quo, tions and local and urban elections, adopted South Asian countries could take the follow- transparency legislation and other provisions ing steps: for open government, and created a variety of mechanisms for citizen input and feed- • Enhance opportunities for urban govern- back. Even so, substantial gaps in account- ments to raise their own revenues. ability remain across the region. Even where Necessary actions would include allowing good systems are in place or under develop- additional autonomy over sources that ment, adequate knowledge, incentives, and they are already allowed to collect and capacity to use them well may not be in facilitating better management of them, as place at any level. well as assessing the need for and develop- A key concern is the autonomy gap. If ing additional sources as required. urban governments have little genuine control • Rethink and reform intergovernmental fi s- over the raising and spending of resources, cal transfers. The aim is to reduce frag- they are unlikely to feel pressured to, or be mentation, achieve an appropriate balance able to, improve performance, which they REVITALIZING URBAN GOVERNANCE AND FINANCE 105

may perceive or frame as largely beyond their contribute to the costs of service provision control. If citizens believe urban governments and more generally to engage with elected are weak and ineffective, they will not respect representatives and urban ofﬁ cials. or trust them, and they may not participate in local elections, take advantage of other Summary of challenges and moving accountability mechanisms to influence urban forward government behavior, or be willing to pay local government taxes. Narrowing the empowerment, resource, and There are three critical dimensions to alle- accountability deficits is a challenging but viating the accountability deficit—developing critical requirement for substantially improv- better systems and practices, building suffi- ing urban government performance. cient capacity on the part of all parties con- Overcoming these deficits is challenging cerned to use them, and nurturing trust and because of the long history of strong central- productive working relationships all around: ization in South Asia, and many public sector between urban governments and higher-level systems and procedures have not been well governments, between urban government developed, broadly accepted, or effectively elected officials and staff, and between urban used. It is critical because urbanization is pro- governments and their citizens. ceeding so rapidly and altering or increasing A number of steps could help bridge the the variety and level of demands on urban accountability gap, some of which must nec- governments. In their present state, many of essarily be undertaken by central govern- these governments cannot meet these acceler- ments, but local measures are often within the ating pressures and challenges, thus hindering power of urban governments. urban, national, and regional development. The agenda outlined above is general: • Reformulate, as needed, key elements of Urban governments across and within coun- the overall formal framework and systems tries are widely diverse. Different weaknesses for urban governance. These revisions dominate or are more urgent in some coun- include enhancing public mechanisms and tries, and situations are evolving, so that spe- procedures for financial management, cific reforms must be adapted to both country planning, budgeting, reporting, and audit- and subnational conditions. Some essentially ing, as well as developing more robust universal needs stand out, such as the require- frameworks for PPPs. ment for greater urban government auton- • Extend or improve local electoral and non- omy, but additional work in each country is electoral accountability mechanisms. needed to prioritize critical needs and identify Elections need to be transparent and sufﬁ - specific remedial options. ciently competitive to give citizen vot- An overarching concern for conducting ers meaningful choice. Nonelectoral further empirical analysis is that limited, mechanisms—input-oriented processes, inconsistent, and unreliable fiscal data across such as participatory planning and budget- South Asia preclude robust identification and ing, and feedback mechanisms, such as analysis of national and regional fiscal trends complaint boards, report cards, and right and limit accessibility to the information and to services regimes—can be highly produc- transparency that are essential for good gov- tive if well designed and appropriately ernance. The development of regular collec- implemented, including by supporting the tion, accessibility, and analysis of urban data capacity of citizens to use them. is a pressing need throughout the region. • Strengthen the link between urban revenue In addition, most countries require many generation and urban service delivery. urban governance and finance reforms, which Making this link tighter and more evident cannot all be realized quickly. It will be neces- should improve citizens’ trust in local gov- sary to prioritize reforms and develop a prag- ernments and enhance their willingness to matic, strategic approach to implementing 106 LEVERAGING URBANIZATION IN SOUTH ASIA

them. Such strategies are also specific to each (2013); and Local Development International country, as well as to individual states or (2013). provinces and localities. Yet it is reasonable to 3. The data provided in the tables in this chapter state that strategic approaches will involve, come from many different sources and have among other possible elements, some combi- been cross-checked where possible. There are too many sources to cite them all in individ- nation of periodic situational assessments and ual tables, but all of the data have been drawn the development of criteria-based, asymmet- from one or more (usually country-specifi c) ric starting points; mechanisms to enhance sources cited in the chapter references. Given fiscal and political powers and build appro- the range of sources and incomplete explana- priate and usable capacity; and the mapping tions of how some data were derived, there of progressive trajectories that build toward are undoubtedly comparability issues, so the stronger systems and results over time. In all data should be seen as illustrative rather than cases, carefully targeted incentives would help defi nitive. encourage better urban government perfor- 4. Because these three defi cits are interrelated, mance. In many cases, motivated urban gov- some decisions were made about where to pro- ernments can independently take strategic vide the primary treatment of issues that are relevant for multiple defi cits; in such cases the steps to improve their performance even interdependencies are recognized in the appro- before broader national system reforms are priate sections. adopted. 5. The service delivery information reported here The longstanding and nontrivial political is based on World Development Indicators obstacles to strong urban government 2011 (World Bank; http://data.worldbank.org empowerment in South Asia are likely to per- /data-catalog/world-development- indicators sist to varying degrees. But as pressure grows /wdi-2011). for improved urban government performance, 6. A comparison of delivery of basic municipal political openings for reform will surely services (roads and streets, public transport, emerge. And as these openings arise and are water and sewerage, solid waste) in large cit- acted on, forward-thinking urban govern- ies around the world is provided in the table. Although there is much “devil in the detail,” ments will seize the space and opportunities the comparison confi rms that South Asian cit- to be more prominent and effective players on ies have relatively limited involvement in pro- the local and national development stages. viding major services; in most other large cities, Initial instances of successful change should either the city government or the central met- help create a demonstration effect for others ropolitan authorities play a major role in the and, it is hoped, stimulate greater demand provision of services, except for roads. Cities from citizens and businesses for more active, in other regions also commonly use metropoli- influential, and efficacious urban govern- tan governance structures that take responsi- ments across South Asia. bility for functions without extensive sharing of authority with other governmental entities. Notes There is comparatively more diffusion of functional responsibility among levels of govern- 1. Every effort was made to update the infor- ment in South Asian cities, including Mumbai. mation as of early 2015, but the situation in a number of the countries in the region is References evolving and the most recent developments may not be refl ected here. Akramov, K., S. Qureshi, R. Birner, and 2. This is an “informed impression” based on B. H. Khan. 2008. “Decentralization, Local careful consideration of extensive litera- Government Elections and Voter Turnout ture, including several recent publications in Pakistan.” Discussion Paper 00754, that attempt to provide comparative assess- International Food Policy Research Institute, ments of local government strength around Washington, DC. the world, including United Cities and Local Arshad, R. 2003. Pakistan’s Decentralization: Governments (2010); Bahl, Linn, and Wetzel Implications and Challenges for Delivering REVITALIZING URBAN GOVERNANCE AND FINANCE 107

Water and Sanitation. Water and Sanitation Fox, W. F., and B. Menon. 2008. Program for South Asia. Washington, DC: “Decentralization in Bangladesh: Change World Bank. Has Been Elusive.” Working Paper, Asian Development Bank, Department for International Center for Public Policy, International Development, and the World Andrew Young School of Policy Studies, Bank. 2004. Devolution in Pakistan. Overview Georgia State University, Atlanta, GA. of the ADB/DFID/World Bank Study. Kurosaki, T. 2006. “Community and Economic Washington, DC: World Bank. Development in Pakistan: The Case of Citizen Bahl, R. W. 1999. “Implementation Rules for Community Boards in Hafizabad and a Fiscal Decentralization.” Working Paper, Japanese Perspective.” Pakistan Development International Studies Program, Andrew Review 45 (4): 575–85. Young School of Policy Studies, Georgia State Leitan, G., and R. Tressie. 2010. “Overview of University, Atlanta, GA. Decentralization and Local Governance in ———, M. Cyan, and S. Wallace. 2008. “The Sri Lanka.” Swiss Agency for Development Challenge of Intergovernmental Fiscal Cooperation, Berne, Switzerland. Relations in Pakistan: The Property Tax Local Development International. 2013. “The Dimension.” Working Paper, International Role of Decentralization/Devolution in Studies Program, Andrew Young School Improving Development Outcomes at the of Policy Studies, Georgia State University, Local Level: A Review of the Literature and Atlanta, GA. Selected Cases.” Department for International Bahl, R. W., J. J. Linn, and D. Wetzel. 2013. Development, London. Financing Metropolitan Governments in Maldives, Ministry of Home Affairs. 2010. “Act Developing Countries. Cambridge, MA: on Decentralization of the Administrative Lincoln Institute of Land Policy. Divisions of the Maldives.” Ministry of Home Bahl, R. W., G. Sethi, and S. Wallace. 2011. Affairs, Maldives. “Fiscal Decentralization to Rural Local Martinez-Vazquez, J. 2013. “Local Nonproperty Governments in India: A Case Study of West Revenues.” In Financing Metropolitan Bengal State.” Publius 40 (2): 312–31. Governments in Developing Countries, edited Bhutan, National Assembly. 2009. The Local by Roy Bahl, Johannes Linn, and Deborah Government Act of Bhutan. National Wetzel, 183–211. Cambridge, MA: Lincoln Assembly, Bhutan. Institute of Land Policy. Bird, R. M., and E. Slack. 2013. “Metropolitan Mathur, O. P. 2008. “Decentralized Urban Data Public Finance.” In Financing Metropolitan System.” Preliminary Note, National Institute Governments in Developing Countries, edited of Public Finance and Policy, New Delhi. by Roy Bahl, Johannes Linn, and Deborah ———. 2013. “Finances of Municipalities: Issues Wetzel, 135–58. Cambridge, MA: Lincoln before the Fourteenth Finance Commission.” Institute of Land Policy. Economic and Political Weekly 2103 (June): Carter Center. 2014. “Local Governance in 23–27. Nepal: Public Participation and Perception.” McCluskey, W. J., and R. Franzsen. 2013. Carter Center, Atlanta, GA. “Property Taxes in Metropolitan Cities.” Cheema, A., A. I. Khawaja, and A. Qadir. 2006. In Financing Metropolitan Governments in “Local Government Reforms in Pakistan: Developing Countries, edited by Roy Bahl, Context, Content and Causes.” John F. Johannes Linn, and Deborah Wetzel, 159–81. Kennedy School of Government, Harvard Cambridge, MA: Lincoln Institute of Land University, Cambridge, MA. Policy. Chhoden. 2009. “An Evaluation of McLure, C. E., Jr. 2000. “Tax Assignment and Decentralization System in Bhutan.” Thesis, Fiscal Autonomy.” Bulletin for International KDI School of Public Policy and Management, Fiscal Documentation 2000 (December): Seoul, Republic of Korea. 626–35. Commonwealth Local Government Forum. 2013. Nepal, Ministry of Finance. 2012. “Economic The Local Government System in Maldives. Survey Fiscal Year 2011/12.” Ministry of http://www.clgf.org.uk/userfiles/1/file Finance, Nepal. /Maldives_Local_Government_Proﬁ le_2013 Nepal, Ministry of Local Development. 2012. _CLGF.pdf. “Public Expenditure & Financial Accountability 108 LEVERAGING URBANIZATION IN SOUTH ASIA

and Fiduciary Risk Reduction Action Plan.” In Financing Metropolitan Governments in Ministry of Local Development, Nepal. Developing Countries, edited by R. Bahl, Pasha, Aisha Ghaus. 2011. Fiscal Implications J. Linn, and D. Wetzel, 107–34. Cambridge, of the 18th Amendment: The Outlook for MA: Lincoln Institute of Land Policy. Provincial Finance. World Bank Policy Thomson, A. 2010. “Thematic Paper on Notes Series on Pakistan 1. Washington, DC: Decentralisation and Local Empowerment in World Bank. Bhutan.” Oxford Policy Management, Oxford, Pethe, A. 2013. “Metropolitan Public Finances: U.K. The Case of Mumbai.” In Financing United Cities and Local Governments. 2010. Metropolitan Governments in Developing Local Government Finances: The Challenges Countries, edited by Roy Bahl, Johannes Linn, of the 21st Century. Second Global Report and Deborah Wetzel, 243–71. Cambridge, on Decentralization and Local Democracy. MA: Lincoln Institute of Land Policy. Barcelona, Spain: United Cities and Local Rao, G. M. 2006. “Republic of India.” In The Governments. Practice of Fiscal Federalism: Comparative White, R., and J. Boex. 2008. “Intergovernmental Perspectives, volume 4, edited by Anwar Shah, Fiscal Relations and Subnational Expenditures 151–77. Montreal, Canada: McGill-Queens in Afghanistan.” Working Paper 45105, World University Press. Bank, Washington, DC. ———. 2013. “Property Tax System in India: World Bank. 2006. “Property Taxes in Large Problems and Prospects of Reform.” Working Cities of Punjab Province, Pakistan.” World Paper 2013-114, National Institute of Public Bank, Washington, DC. Finance and Policy, New Delhi. ———. 2006. “Service Delivery and ———, and M. R. Bird. 2010. “Urban Governance Decentralization in Sri Lanka: Assessment and and Finance in India.” Working Paper 2010- Options.” World Bank, Washington, DC. 68, National Institute of Public Finance and ———. 2007. “Service Delivery and Governance Policy, New Delhi. at the Subnational Level in Afghanistan.” Shah, A. 2012. “Grant Financing of Metropolitan World Bank, Washington, DC. Areas: A Review of Principles and Worldwide ———. 2009. Improving Municipal Management Practices.” World Bank Policy Research Paper for Cities to Succeed: An IEG Special Study. 6002, World Bank, Washington, DC. Washington, DC: World Bank. Sharma, Y., and A. Muwonge. 2010. “An ———. 2010. Bangladesh Public Expenditure Opportunity to Improve Service Delivery and Institutional Review: Towards a Better in Nepal through Local Governance.” Quality of Public Expenditure. Washington, Himalayan Research Papers Archive, Nepal DC: World Bank. Study Center, University of New Mexico, ———. 2012. “Fiscal and Municipal Finance Albuquerque, NM. Analysis of Five Large Cities in Punjab Shurkin, M. 2011. “Subnational Government Province.” World Bank, Washington, DC. in Afghanistan.” Occasional Paper, Rand ———. 2014a. Local Service Delivery in Nepal. Corporation, Santa Monica, CA. Washington, DC: World Bank. Smoke, P. 2001. Fiscal Decentralization in ———. 2014b. “Municipal Finance in India: Developing Countries: A Review of Current Opportunities, Issues, and Way Forward.” Concepts and Practice. Geneva, Switzerland: Unpublished, World Bank, Washington, DC. United Nations Research Institute for Social ———. 2014c. “SAR Urban Flagship Fiscal Development. Decentralization and Governance Component Sud, I., and S. Yilmaz. 2013. “Institutions and Country Information Template.” Internal Politics of Metropolitan Management.” memo, World Bank, Washington, DC. Planning and Managing Spatial Structure and Connectivity 4

Key messages

To enhance livability and improve opportunities declined by investing in improvements such as for prosperity, planners, government policy mak- better public urban spaces to enhance pedes- ers, and stakeholders need to better manage the trian walkability and livability. Promote spatial structure and intra- and interurban con- better management of developable land in city nectivity of South Asian cities at several levels: cores through effective land-assembly mecha- • National. Invest in the strengthening of trans- nisms; freeing up or making better use of pub- port links that improve connectivity between licly owned land; and reusing existing cities—between large and secondary cities, and structures in an adaptive, appropriate, and between secondary cities and towns—to create innovative manner. more efﬁ cient systems of cities. Also invest in • Institutional. Facilitate the formation of improved intracity connectivity and traffic more vibrant neighborhoods through granu- management to enhance mobility within urban lar and contextual spatial planning areas and ease trafﬁ c congestion. approaches that permit greater variation in • City peripheries. Adopt forward-looking land uses and intensity of development. Such approaches to planning and guiding expan- planning should be dynamic and flexible, sion where it is most rapid—on city peripher- allowing land uses to adapt to changes in ies. This approach will reduce the messiness market demand within a framework that of urbanization, prevent undesirable spatial takes a long-term view of a city’s develop- forms from being locked in, and facilitate ment. Strengthen city planners’ and local future provision of infrastructure and basic governments’ capacity to plan, coordinate, urban services. implement, and enforce development to • City cores. Unlock the potential of city cores deliver integrated, coordinated, inclusive, and carry out rejuvenation where cores have and smarter planning policies.

109 110 LEVERAGING URBANIZATION IN SOUTH ASIA

Introduction footprint exceeds the rate at which it can expand infrastructure and regulate develop- As discussed in chapter 2, many South Asian ment, spatial planning and services provision cities are faced with difﬁ cult challenges aris- typically suffer. ing from the rapid expansion of their built- In Kabul, the city’s population grew by a up areas and the low-density sprawl that, all staggering 4.5 percent a year between 2010 too often, has gone hand-in-hand with (3.72 million population) and 2015 expansion. Managing these challenges is (4.64 million population). Urbanization was made even tougher by the expansion of largely informal, with an estimated 73 percent built-up areas beyond cities’ administrative of the population living in unplanned areas. boundaries. Addressing the ability of South These unplanned areas not only make services Asian countries to manage their cities’ spa- provision hard, but have also started to tial development is critical for two key encroach on valuable agricultural land on the reasons: peripheries. Dhaka—with a population of First, good connectivity and efficient slightly more than 14.5 million,1 one of South spatial structure are essential to alleviating Asia’s megacities—faces similar challenges. congestion pressures that both undermine a Issues related to quickly expanding urban city’s livability and hamper the agglomeration areas are seen in many smaller cities as well, economies that hold the key to prosperity. such as Thimphu (box 4.1). The positive relationship between a country’s For South Asian cities to transform them- level of urbanization and its level of gross selves into productive and livable centers, domestic product (GDP) per capita is well they must not only manage rapid expansion established (see chapter 1)—a doubling at their peripheries but also address existing of city size is, in general, associated with and future challenges at their city cores. These a 3–5 percent increase in productivity areas often have largely locked-in spatial (Rosenthal and Strange 2004). However, structures that lead to congestion forces that although cities continue to be at the forefront result in clogged streets, polluted air, unaf- of South Asia’s economic growth, shaping cit- fordable land, and the prevalence of slums. At ies to accommodate population growth with the national level, how cities are connected as jobs, housing, livable environments, and ser- a system, and how they can become spatially vices in an equitable, inclusive, and sustain- differentiated, become important. All these able matter will become more challenging and issues are made more daunting by the need complex as cities grow. for appropriate and effective policies and reg- Second, managing connectivity and spatial ulations, given the current governance and structure will be critical in preventing South finance deficit in most South Asian countries. Asian cities from being further “locked in” to Although market forces underlie agglomera- a pattern of urban sprawl that is prohibitively tion economies and congestion forces, they do costly to reverse. Underutilized land, planning not address the key issues arising from exter- constraints, and the resulting rise in land and nalities and public infrastructure, which only rental costs also make it difficult for cities to government intervention can address. But support affordable housing or commercial policy failures can further hinder, rather than and industrial space (see also chapter 5 for facilitate, development. Therefore, critical discussion of this point in specific relation to enablers such as sound urban governance, the supply of affordable housing). Many cities capable institutions, and good leadership in South Asia also lack adequate transporta- must underpin the capacity for good policy tion systems, with insufficient road networks, making and effective spatial planning are (see poor pedestrian amenities, and poor or non- also chapter 3). existent public transit, all of which drive up South Asian cities can transform spatially private commuting costs and decrease mobil- through more granular, contextual, and inte- ity and job access. When the growth of a city’s grated approaches to spatial planning to shape PLANNING AND MANAGING SPATIAL STRUCTURE AND CONNECTIVITY 111

BOX 4.1 Urbanization pressures in Kabul, Dhaka, and Thimphu

Managing connectivity in Kabul. Rapid popula- government impedes spatial planning and infra- tion growth caused motorization between 2006 structure programming. Consecutive urban mas- and 2007 to increase by 26 percent; the transpor- ter plans have underestimated population growth tation network has been unable to keep pace with and urban development and failed to allocate traffi c or trip generation. A major bottleneck is enough land for urban expansion. Transportation the absence of a proper regional highway net- planning has not only failed to anticipate growth, work, such as a circular road connecting radial but has also failed to mobilize suffi cient invest- roads from the city center. Furthermore, localities ment for roads and transit networks, leaving within Kabul are poorly connected; this isolation Dhaka poorly connected and highly congested. urgently needs to be remedied along with better Spatial planning in Thimphu. As the political overall accessibility. and economic center of Bhutan and the high- A lack of adequate transportation planning est recipient of migrants from other parts of and investment programming impedes connec- the country, Thimphu accounts for 15 percent tivity within Kabul and makes shaping an effi - of the country’s population and 17.9 percent of cient spatial structure diffi cult. Kabul’s public its area. Agricultural land (including orchard transport setup is currently unable to satisfy area) and forestland, respectively, constitute demand and suffers strong competition from 23.4 percent and 18.7 percent of the total urban all sizes of vehicles on major routes. Higher- area. However, sprawling urbanization is rapidly capacity public transport along major roads, reducing the supply of agricultural land. rationalized bus routes, and an adequate supply Thimphu experienced unprecedented popula- of buses are required. In addition, bus termi- tion growth of 10 percent in 2010, and Bhutan’s nals and intermodal integration would further urban population is expected to increase by improve public transport. a factor of 1.7, from 253,000 to 434,000, Institutional coordination in Dhaka. Urban between 2010 and 2030. This growth presents structure and transportation planning have not huge challenges for spatial planning and manag- kept pace with population growth or urban ing urbanization. development in Dhaka, where the lack of horizontal and vertical coordination within municipal Source: World Bank 2013b.

urban forms that are characterized by a vari- The following sections discuss the dynam- ety of land-use patterns, articulated intensities ics of intercity and intracity connectivity, the of development, and vibrant neighborhoods. key challenges of spatial and transportation Such approaches must include attention to planning at the city level, and the supporting details such as local area development plans, institutional and implementation require- streetscape improvements, public space provi- ments for managing spatial planning. The dis- sion, urban design, good connectivity, and cussion frames the key challenges, offers place making. These details need to be imple- broad policy guidance, and showcases exam- mented in a way that will allow greater dyna- ples of what other cities inside and outside the mism and flexibility and that can better region have done. respond to short-term needs while taking a Experience in managing urbanization in long-term view of spatial development. South Asian cities of various sizes is mixed. Figure 4.1 suggests some key priority areas for Some smaller cities, despite limited financial South Asian cities. and human resources, have done a better job. 112 LEVERAGING URBANIZATION IN SOUTH ASIA

FIGURE 4.1 National, city, and institutional priority areas

Intercity connectivity • Facilitate formation of well-connected and efficient system of cities

Rapid periphery growth • Improve intracity connectivity • Plan ahead and facilitate new centers

City core with relatively locked- in spatial structure

• Unlock assets and land resources • Enhance public spaces • Implement innovative policies to rejuvenate core

Institutions • Facilitate granular planning • Take integrated, coordinated, and dynamic planning approaches • Strengthen urban governance and capacity • Promote participatory planning

Larger cities (with exceptions such as Developing a system of well- Ahmedabad and Colombo) have, in general, connected and productive cities been unable to keep pace with growth because of larger absolute population numbers, Intercity connectivity—how regions, cities, inadequate resources, and institutional frag- and ports are linked—is critical in fostering mentation. Much of this chapter’s focus, economic prosperity, reducing poverty, and therefore, is on the region’s megacities, such promoting livability. as Mumbai and Delhi in India and Karachi in Pakistan, with some analysis of medium-sized Promoting intercity connectivity cities such as Amritsar and Kanpur in India and Kandy in Sri Lanka. These cities offer a At the national and subnational regional glimpse into the early symptoms of the bigger scales, intercity connectivity facilitates labor challenges to come if urban growth is not mobility and access to both markets and sup- properly managed. pliers, which, in turn, stimulate domestic PLANNING AND MANAGING SPATIAL STRUCTURE AND CONNECTIVITY 113

trade in goods and services. Better connectiv- panel a). Some South Asian countries, such as ity of cities to airports and ports, which pro- India and Sri Lanka, have better connectivity vide the gateways to international markets, (figure 4.3, panel b), but most suffer from seri- also fosters trade with cities both regionally ous shortcomings such as lack of intraregional and globally. Expanded trade opportunities connectivity for national road networks, unre- then allow ﬁ rms to better exploit scale econo- alized potential for rail and inland water mies and generate urban-rural spillovers. freight transport, and inadequate road and rail Connectivity also fosters better specialization connectivity of ports with their hinterlands. across cities, which helps to alleviate system- An investment climate survey carried out wide congestion costs. Hence, countries with by the World Economic Forum ranked promi- good intercity connectivity often promote nent cities on competitiveness factors, includ- agglomeration economies in their cities and ing multisectoral infrastructure. Six out of the along transport corridors, whereas countries 148 countries surveyed are in South Asia with poor or underdeveloped national trans- (Bangladesh, Bhutan, India, Nepal, Pakistan, portation infrastructure tend to be less com- and Sri Lanka), but they rank relatively low petitive in domestic and international trade. on competitiveness (Global Competitiveness The Republic of Korea launched an aggres- Index) and their infrastructure rankings are sive road-building program to connect prov- even lower (table 4.1). Apart from Sri Lanka, inces and cities, and the result is a notable infrastructure is singled out as one of the top example of good connectivity. Between 1995 five problems in doing business with every and 2010, levels of gross regional domestic country in South Asia. Only India and Sri product (GRDP) increased dramatically as Lanka are in the top 100 countries for com- the country built its highway system. The petitiveness, and they do not rank well on greatest GRDP increases occurred in areas infrastructure. with better connectivity, especially those at An analysis of connectivity patterns across major intersections (figure 4.2). South Asia finds that the strongest infrastruc- The density of South Asia’s road network— ture linkages in the region are between the as measured by its length per 1,000 people— largest metropolitan cities, and connectivity lags behind the densities of the networks in and communities are strongly influenced by North America and Europe and Central Asia, national borders. Derudder and others (2014) and is much more on par with the densities of analyze connectivity patterns based on airline networks in the rest of the world (figure 4.3, flights, Internet Protocol links, roads, and rail

FIGURE 4.2 The Republic of Korea’s highway network and GRDP

Source: Maps prepared by Korea Research Institute for Human Settlements for this report. Note: Darker shades indicate higher gross regional domestic product (GRDP). 114 LEVERAGING URBANIZATION IN SOUTH ASIA

FIGURE 4.3 Road network characteristics

a. Total length of road network per 1,000 b. Access to roads in South Asian countries persons, major world regions

25 25 100

20 20 80

15 15 60

10 10 40 Kilometers per 1,000 people Kilometers per 1,000 people

5 5 20 Percent of roads that are paved

0 0 0

SSA EAP SAR ECA NAC India World MENA Nepal Bhutan Maldives Pakistan Sri Lanka Bangladesh Afghanistan Road network (left scale) Road quality (right scale)

Source: Andrés, Biller, and Dappe 2013. Note: EAP = East Asia and Pacifi c; ECA = Europe and Central Asia; MENA = Middle East and North Africa; NAC = North America; SAR = South Asia; SSA = Sub-Saharan Africa.

TABLE 4.1 World Economic Forum Global Competitiveness Index and infrastructure rankings

GCI Infrastructure GCI Infrastructure GCI Infrastructure Country rank rank Country rank rank Country rank rank Cyprus 58 44 Libya 108 103 Liberia 128 131 Philippines 59 96 Bhutan 109 87 Uganda 129 133 India 60 85 Bangladesh 110 132 Benin 130 129 Peru 61 91 Honduras 111 115 Zimbabwe 131 126 Slovenia 62 36 Gabon 112 114 Madagascar 132 136 Hungary 63 51 Senegal 113 117 Pakistan 133 121 Russian Federation 64 45 Ghana 114 109 Venezuela, RB 134 125 Sri Lanka 65 73 Cameroon 115 128 Mali 135 108 Rwanda 66 104 Gambia, The 116 95 Malawi 136 137 Montenegro 67 70 Nepal 117 144 Mozambique 137 130 Jordan 68 54 Egypt, Arab Rep. 118 98 Timor-Leste 138 138

Source: Schwab 2013. Note: GCI = Global Competitiveness Index. A rank of 1 is best out of 148 countries for both categories.

for 67 key cities in the region, or all cities in Five connective clusters are apparent in the region with a 2010 population of more South Asia’s infrastructure networks: a cluster than 750,000, as well as the capital cities of bringing together Pakistan’s cities plus Kabul; a Colombo, Thimphu, and Malé (to ensure that northern Indian cluster centered on Delhi that all countries in the region are represented). also extends to Kathmandu and Thimphu; PLANNING AND MANAGING SPATIAL STRUCTURE AND CONNECTIVITY 115

FIGURE 4.4 Connectivity of South Asian cities in infrastructure networks: Airlines, information technology, roads, and rail

Source: Derudder and others 2014. Note: The colors represent clusters. The thickness of the line connecting any pair of cities is proportional to the strength of connectivity in four networks: Airlines: Based on the number of direct weekly fl ights off ered during the last week of May 2013. Internet protocol links: Based on data from DIMES, a distributed scientifi c research project studying the structure and topology of the Internet (http://www .netdimes.org/new). Roads: Based on a network effi ciency measure, computed by dividing anticipated travel time between cities by the Euclidean distance separating them. Rail: Based on the number of direct weekly trains off ered during the last week of May 2013.

a southern Indian cluster including Mumbai, connectivity between Delhi and Lahore, which Kolkata, Hyderabad, Chennai, and Bangalore; are shown belonging to separate clusters and two smaller clusters comprising Dhaka and despite being in the same extended urban Chittagong in Bangladesh in one cluster and region. (Nighttime lights data in chapter 2 Malé and Colombo in the other (figure 4.4). reveal the growing physical connection Connectivity appears to be stronger—that between Delhi and Lahore, among other is, better at fostering trade and economic cities.) This current lack of connectivity sug- interaction—within than between clusters. gests that huge productivity gains could be A lack of regional integration impedes realized if frictions created by borders were 116 LEVERAGING URBANIZATION IN SOUTH ASIA

reduced (World Bank 2013c). Pakistani cities particularly secondary and tertiary cities. are more integrated on transport infrastruc- Correspondingly, the implications of planned ture networks than are cities in northern India. regional transport networks (such as India’s These findings support the efforts by the gov- dedicated freight corridors) should be explored ernment of India to invest in regional trans- and exploited to maximize economic benefits. portation networks, such as the Western and Some South Asian countries have begun to Eastern Dedicated Freight Corridors in the prioritize investment in interurban connectivity Golden Quadrilateral (the large network con- and infrastructure. In India, intercity connec- necting Chennai, Delhi, Kolkata, and tivity is generating growth in secondary cities Mumbai). through the roads and rail connectivity of the Improved connectivity often leads to devel- Golden Quadrilateral. The dedicated freight opment of secondary and tertiary cities and, corridors, when fully completed, will connect therefore, the creation of a system of cities. many of the major urban centers in India, such Systems of cities can help alleviate congestion as Delhi, Mumbai, Kolkata, and Chennai. forces in large cities by providing opportuni- These corridors were supported by institu- ties for the spatial deconcentration of land- tional reform in the logistics and rail industries. and capital-intensive operations such as Sri Lanka has also embarked on the develop- mature manufacturing industries. Therefore, ment of a well-functioning and well-connected transport network planning should consider system of cities through investments in inter- the impact of transport on urban systems, city highways and rail networks (box 4.2).

BOX 4.2 Interconnecting cities in India and Sri Lanka

India’s Golden Quadrilateral. The Golden that had India followed the example of China Quadrilateral (GQ) is a highway system launched by building its highway network to directly in 2001 comprising close to 6,000 kilometers of link intermediate cities instead of to link large roadway and connecting the four major Indian centers, more intermediate cities would have industrial and cultural centers: Delhi, Mumbai, benefi ted from better connectivity (Alder 2015). Kolkata, and Chennai. In empirical work con- This approach nonetheless can be taken up in ducted for this report, Ghani, Goswami, and the next phase of India’s transportation network Kerr (2013) seek to evaluate the impacts of this development, given that signifi cant gaps still network on levels of manufacturing activity in remain at the subnational level. For example, nonnodal districts (that is, districts outside the Uttar Pradesh, which is one of the least devel- four major cities) that fall along its route. To oped states in India, not only has high overall do so, they compare changes in levels of manu- poverty but also a high degree of spatial inequal- facturing activity in nonnodal districts within ity. Poor roads and congested rail links within 10 kilometers of the route before and after the the state are constraining the prosperity of the building of the GQ with changes in levels of man- National Capital Region from spilling over to ufacturing activity in more distant districts. They the state. fi nd that the GQ had signifi cant positive impacts on the entry of new manufacturing fi rms within India’s Dedicated Freight Corridors. The main 10 kilometers of the highway. rail corridors in India are also part of the GQ. However, although the GQ highways have Although they account for only 16 percent encouraged manufacturing activity along the of the railway network’s length, they carry corridors they have created, some critics suggest more than 60 percent of its freight load.

(continues next page) PLANNING AND MANAGING SPATIAL STRUCTURE AND CONNECTIVITY 117

BOX 4.2 Interconnecting cities in India and Sri Lanka (continued)

The government of India, recognizing that FIGURE B4.2.1 The national the rail sector urgently needs to add capacity physical structure plan showing to these routes, approved a long-term plan to major planned road and rail build dedicated freight-only lines, parallel to network in Sri Lanka connecting to the existing GQ passenger and freight mixed- major urban centers by 2030 trafﬁ c routes. Relief for the existing lines will allow improvements to be made in passenger service. On completion, the total corridor railway capacity will double. Construction of the lines has started with the Western and Eastern Dedicated Freight Corridors. The Western corridor, which will be almost 1,500 kilometers in length, will run along the Delhi-Mumbai axis of the GQ. Meanwhile, the Eastern corridor will be even longer—slightly more than 1,800 kilometers—and will run parallel to the existing trunk line on the Delhi-Kolkata axis of the GQ.a Sri Lanka’s vision for a system of cities. The government of Sri Lanka also has begun an ambitious plan to connect its cities via a comprehensive road and rail network to fulfill

its vision of a well-planned system of cities Source: Sri Lanka National Physical Planning (figure B4.2.1). This system uses a strategy Department (n.d.). of fostering economic growth in major urban centers outside Colombo to produce a more Expressway, travel time between Colombo and spatially balanced distribution of economic Matara has been cut to 2 hours from 4.5–5 opportunities and reduce congestion in the hours (Road Development Authority 2014). capital. The government has begun building an expressway network to improve connections a. For more information on the dedicated freight corridors project, see between cities. With the opening of the Southern http://www.dfccil.gov.in/dfccil_app/home.jsp.

City and regional access to ports and mul- Colombo, Dhaka, Karachi, Kolkata, and timodal transportation networks is a critical Mumbai, hence the importance of invest- part of intercity connectivity. As a result of ments in connectivity. Airports are also the National Trade Corridor project, a con- important for shipping high-value-to-weight cept approved in 2005, Pakistan today has outputs, as seen in Sialkot’s development of a one of the most developed networks of high- freight airport. To complement effective road ways in the region, linking its three Arabian and railway investment, trade facilitation, Sea ports to the rest of the country and far- railway stations, logistics hubs, and customs ther north to Afghanistan, Central Asia, and clearances must also be adopted (World Bank China. 2013c). Many of the region’s major ports are along Investing in such dense networks is often its southern edge, such as those in Chennai, expensive: rapidly developing countries have 118 LEVERAGING URBANIZATION IN SOUTH ASIA

devoted more than 15 percent of their GDP to Industrial or business clusters cannot form infrastructure investment. But improved without supportive land-use policies that intercity and port connectivity contributes to encourage formation of a critical mass of the development of a network of prosperous complementary activities. cities by reducing trade costs and facilitating Over time, such policies could lead to a more efficient allocation of resources across shift to higher-value clusters and the creation cities. Gradually, as cities and their hinter- of new subcenters. As land-use intensity lands become better connected, land- and increases, land prices increase, and many capital-intensive firms in maturing industries mature firms that no longer need to focus on relocate to lower-cost areas while retaining product or service innovation naturally seek good access to large cities. This process also locations in lower-cost areas, essentially “sub- frees up land in the cores of large cities for the urbanizing.” Improved connectivity helps this potential entry of higher-value-added, more process along by making less congested sub- human-capital-intensive, services sector urban locations relatively more attractive. In industries such as banking and finance, creat- this case, improved connectivity helps to ing agglomeration economies across systems “pull” firms out from more congested to less of cities. congested places. Firms, however, may also Schramm (2013), in modeling the impact suburbanize even in the absence of significant of India’s dedicated freight corridors on improvements in connectivity when, for regional wages using a “new economic geog- example, congestion forces are exacerbated raphy” approach, reinforces this finding. by failures in policy and thereby serve to Transportation connectivity, along with the “push” firms out from what would otherwise size of the city, are the principal determinants be attractive locations. In Mumbai, manufac- of market access: increasing connectivity turing activities have moved further inland helps boost trade and economic activity and because of high land and associated costs, has the added advantage of fostering rural- partly due to Mumbai’s restrictive urban urban migration and providing typically poor development plans and partly because its rural residents faster and cheaper access to transportation system is underdeveloped for a urban services such as health care and educa- city of its size. tion. Similarly, in a detailed study of transpor- In successful and well-planned cities, tation and urbanization in Sri Lanka, Lall and areas vacated by industries are usually revital- Claus (2009) find that proximity to Colombo ized by high-value services. For example, and better transportation infrastructure fos- Puerto Madero in Buenos Aires was once ters migration of the poor to employment a maritime-oriented warehouse district centers within the metropolitan area. but is now a vibrant entertainment area Intercity connectivity has profound effects (box 4.3). Unfortunately, in major South on the formation of urban areas by facilitat- Asian cities—with the exceptions of Colombo ing agglomeration economies, through either and Bangalore—this process has not localization or urbanization economies (see happened. Rather, as formal manufacturing chapter 1). Examples of localization econo- has suburbanized in response to excessive mies include Bangalore, Gurgaon, and Noida; congestion forces in cities such as Delhi specifically for banks and financial institu- and Dhaka, urban cores have stagnated (see tions, Colombo, Karachi, and Mumbai; and chapter 2, “Spatial deconcentration of for textiles, Chittagong, Dhaka, and Tiruppur. manufacturing”). Many of these initiatives were privately A city’s spatial structure can gradually driven. Once a large private firm locates or transform through land-use changes to grows, it attracts smaller firms that provide it respond to market needs and higher intensi- with services and inputs. Public policy plays a ties in central locations or new subcenters. critical role as well, exerting real influence by This transformation is often marked by a shift providing critical backbone infrastructure. from monocentric to polycentric urban forms, PLANNING AND MANAGING SPATIAL STRUCTURE AND CONNECTIVITY 119

BOX 4.3 Rejuvenating Puerto Madero, Buenos Aires

FIGURE B4.3.1 Illustrative plan of Puerto Madero, Puerto Madero’s vacant land was not new, dat- 2014 ing as far back as the 1940 Buenos Aires master plan, which called for combined recreational and cultural uses. Administrative and fi nancial conditions, as well as land-ownership issues, prevented the plan’s adoption. The economic crisis of 1989 stimulated the passage of the Administrative Emergency Law and the Economic Emergency Law, which allowed federal property to be privatized, particularly in Puerto Madero. The land was transferred from the General Administration of Ports Source: Based on Corporacion Antiguo Puerto Madero S.A., Plano Madero. to the Corporacion Antiguo Puerto Madero Sociedad, a quasi-private corporation created for running the project without the standard The autonomous city of Buenos Aires, the congressional approvals. Given its central loca- Argentinian capital, has a stable population of tion and perceived high value of the waterfront 3 million and an area of 200 square kilometers land, the project was fi nanced by the sale of some and is part of a greater metropolitan area with of the land without the need for the municipality about 13 million inhabitants (2010) covering to contribute. 4,500 square kilometers. The city accounts for The Corporacion used the 1940 master plan less than 8 percent of the country’s population as an instrument for negotiating with developers but generates 25 percent of GDP. and local professional bodies rather than as a The regeneration of Puerto Madero harbor strict regulatory framework. The plan also played a key role in revitalizing the central busi- underwent a national design competition in ness district and surrounding neighborhoods. In 1992, with a strong emphasis on green space, the 1980s and 1990s, businesses rapidly left the waterfront access, mixed use, and preservation downtown area in the north and northwest of of historic warehouses (fi gure B4.3.1). the city, seriously threatening Buenos Aires’s economic sustainability. The idea of redeveloping Source: Amirtahmasebi and others, forthcoming.

resulting in multiple subcenters specialized metrorail connectivity and government incen- according to services or land-use activities. tives were essential. Such forms often require master plans that Many South Asian cities need to approach allow for changes in land use and intensity. transportation planning at a broad subna- Mumbai could redevelop central districts tional scale to ease congestion constraints and such as the Mumbai Port area as Buenos leverage agglomeration economies. Planning Aires did, bringing vital benefits to the city. at this scale will require metropolitan-level Transformation typically requires infrastruc- plans and programs and the institutional ture investment—particularly in enhanced structures to implement them. South Asian connectivity—to motivate change and create countries such as India and Sri Lanka have efficient links between multiple centers, as in already made large investments in transporta- Canary Wharf in London, where improved tion networks across subnational regions, 120 LEVERAGING URBANIZATION IN SOUTH ASIA

but it is critical that this momentum be con- To manage this increase in vehicles, South tinued to lead to better overall development Asian cities can focus on developing public outcomes. transport infrastructure and policies as As many South Asian cities strengthen Ahmedabad has done with its Bus Rapid their intercity connectivity to facilitate effi- Transit System (box 4.4). Cities facing traffic cient systems of cities, they will also need to congestion pressures can look toward either improve intracity infrastructure and manage reducing overall demand or distributing it local congestion pressures. In Bangladesh, over time, as well as pairing investment in congestion and high land and housing prices public transport with policies to manage have encouraged garment manufacturers to automobile use. Singapore is one of the best- relocate to suburban areas outside Dhaka but known examples of managed motorization, vacated buildings have not been repurposed, having combined high vehicle taxes on auto most likely because transportation infrastruc- purchases, ownership quotas, congestion ture has not been upgraded. Therefore, pricing that charges cars to enter the central Dhaka is not developing activities with business district (CBD), and an extensive and higher economic productivity (Muzzini and efficient mass transit system to help keep Aparicio 2013a). automobile ownership rates relatively low (Han 2010). Although South Asian cities gen- Managing intracity connectivity and erally may not be ready to adopt congestion congestion pricing and tolling, implement parking controls, or levy taxes on luxury vehicles, they Managing intracity connectivity and conges- could start with basic traffic management tion forces arising from the pressure of popu- initiatives, such as charging for on-street lation on land, housing, and basic services is parking, and then move toward pay-to-use critical to leveraging agglomeration econo- parking lots and garages. mies. Congestion forces work against city Most South Asian cities are poorly planned growth and urbanization and hence the and managed, but the transformative initia- prosperity afforded by agglomeration econo- tives of cities like Ahmedabad suggest that mies (see chapter 1). When urbanization managing motorization and urbanization is increases faster than a city’s infrastructure possible. In fact, cities like Chennai and Delhi capacity, businesses may relocate to other cit- have also recently invested in new metrorail ies or countries with more reliable services. lines to improve intracity connectivity. Cities This trend is common to many South Asian that mobilize resources to provide infrastruc- cities, which suffer from unreliable power and ture to support growth and those that man- water supply (World Bank 2013c). age congestion and invest in enhanced The challenge of ensuring good intracity mobility fare better. Balancing intracity connectivity tends to increase with city size, mobility with physical expansion is a way and a common threat to large cities is the lack that cities can achieve a more efficient spatial of adequate transportation infrastructure, structure and provide better prosperity and roads, and mass transit. As per capita income livability outcomes. rises, demand for automobiles, motorbikes, To meet increasing travel demand from and small buses typically increases in a population growth, even cities with relatively motorization trend seen in many of the larger developed infrastructure and public transport South Asian cities. The number of powered must continue to optimize their services. At vehicles in Delhi, for instance, surged peak hours, road congestion can easily cause from 521,000 in 1981 to 7.44 million in travel times to double and metrorail systems 2012. In Pakistan, the number of motorized and buses to reach capacity, leading to a vehicles per 1,000 persons increased more decline in service quality. Governments and than 50 percent between 2002 and 2009 service providers need to improve their ability (World Bank 2013e). to anticipate demand and provide more road PLANNING AND MANAGING SPATIAL STRUCTURE AND CONNECTIVITY 121

BOX 4.4 Ahmedabad Bus Rapid Transit System: The Janmarg

Ahmedabad’s Bus Rapid Transit System (BRTS) facilities (ﬁ gure B4.4.1). The BRTS began opera- has emerged as an example of how a planned tion in 2009 and has grown from a route length commuting system can help reduce emissions of 12 kilometers serving 18,000 passengers a day and improve air quality as well as have a posi- to about 90 kilometers serving 175,000 passen- tive impact on urban development. Going by the gers a day in 2014. City residents are also served name “Janmarg” or “the people’s way,” the BRTS by private bus operators, buses run by Gujarat is a road-based public transport service operated State Transport Corporation, and railway. by Ahmedabad Municipal Transport Services, State and local governments have prepared featuring a closed system with bus stations an expansion plan of the mass transit sys- along road medians as well as revamped rights- tem comprising metrorail and BRTS (World of-way to include cycle tracks and pedestrian Bank 2013b). Ahmedabad has also used land

FIGURE B4.4.1 Bird’s-eye view of the BRTS station at Jhansi ki Rani junction, Ahmedabad, India

Source: Centre of Excellence in Urban Transport, CEPT University from Innovations in Design: Ahmedabad Bus Rapid Transit System, published by Mapin Publishing in association with the Centre of Excellence in Urban Transport, CEPT University.

(continues next page) 122 LEVERAGING URBANIZATION IN SOUTH ASIA

BOX 4.4 Ahmedabad Bus Rapid Transit System: The Janmarg (continued)

readjustment to upgrade informal areas and to FIGURE B4.4.2 The Bus Rapid Transit increase density in central-city areas (Annez and System network others 2010). The BRTS is considered a success for several reasons: • Environmental benefi ts. About 20–22 percent of commuters have moved from motorcycles to buses, saving almost 200,000 vehicle- kilometers per day. The Janmarg is expected to continue attracting passengers and is part of a larger regional plan for Ahmedabad with far wider coverage. • Social benefi ts and catalyst for development. The extensive network (figure B4.4.2) has benefi ted previously underserved groups. For example, there has been a rise in female trav- elers, and almost 40 percent of commuters in the afternoon off-peak hours are women. Source: Ahmedabad Municipal Corporation, Ahmedabad Bus Janmarg has also helped rejuvenate Rapid Transit System (http://www.ahmedabadbrts.org/web Ahmedabad, encouraging redevelopment of /operationplanmap.html). vacant, former mill lands and the building of new housing and shopping areas for the urban municipal council held regular press brief- poor along its corridors. ings, public exhibitions, and presentations, • Citizen engagement. A big factor in the suc- engaging citizens for suggestions and cess of Janmarg has been the positive role recommendations. played by citizens through consultation. Sources: World Bank 2013b; United Nations Framework Convention on Climate During planning and design, the Ahmedabad Change 2012.

network capacity and public transport ser- modes of travel if such function is enabled—is vices as well as enhance service levels. in operation in the Delhi Metro and experi- Mumbai has implemented a traffic control mentally in buses of the Bangladesh Road system for monitoring traffic in the central Transport Corporation in Dhaka. part of the city with about 700 cameras, capa- In the future, strategies to better integrate ble of controlling roughly 250 signal intersec- transport networks with land use, such as tions to achieve network-wide optimization transit-oriented development and transport of traffic flow. Intelligent transport systems demand management, can be further devel- are becoming increasingly popular in bus ser- oped in the larger cities. Integration of land vices in Indian cities. Buses are equipped with use and development densities into transport on-board units to transmit location, speed, networks at a granular level will be critical to and other operational information to control improving intracity connectivity, including centers, enabling more efficient dispatching of modal integration among rail, bus, road, and vehicles and provision of passenger informa- nonmotorized transport. Ensuring sufficient tion. Smart-card ticketing—a prepaid card sidewalk space at metro entrances and safe system that can also be used for different waiting areas for buses and taxis can also PLANNING AND MANAGING SPATIAL STRUCTURE AND CONNECTIVITY 123

tighten intermodal links. Metro stations can reverse, rendering service provision diffi cult add more pedestrian-friendly features and and upgrades, like road widening, more incorporate barrier-free access for the dis- challenging. abled. As new metro networks are introduced (Chennai) or current ones expanded (Delhi), Planning for effi cient spatial structures more planned features around metro sta- and integrating with transportation tions, such as bridges across main roads and pedestrian crossings, should integrate more Market forces play an important role in fluidly with the rest of the city’s pedestrian shaping the spatial structure of cities, but network. planning is required to provide infrastructure and other public goods; to coordinate and promote synergies between land uses; Managing urban expansion and and, more generally, to help manage the rejuvenating city cores trade-offs that cities face between agglomeration economies and congestion forces. Spatial structure broadly refers to the organi- Planning can also be important in promoting zation and patterns of land use (for instance, equity, especially for the poor (box 4.5 and businesses, housing, public amenities), densi- box 4.13). Good connectivity combined with ties (for instance, building height, fl oor area, effi cient spatial structures facilitates face-to- persons per unit of land area), and connecting face business transactions and goods infrastructure (for instance, roads, metrorail movement and minimizes the costs of com- systems, and the like) within cities and across muting. And cities that can integrate and subnational regions. Sarzynski and Levy coordinate land use and transportation tend (2010, 4) defi ne spatial effi ciency as to have effi cient and productive economies. Spatial planning also affects the organiza- the geographic arrangement of businesses tion of activities within cities, from location and residences, the physical infrastructure that connects the region (i.e., transportation, of businesses to housing provision. Because communication, green space), and the orien- employment density is partly shaped by land tation of each towards the other that mini- use and spatial structure—commercial dis- mizes the time, effort, or cost required to tricts and building heights—planners need to conduct economic activities for the entire consider how to promote dense employment metropolitan region. districts through effective planning and integrated infrastructure programming. Globally, urbanized land area is expanding Employment density, in turn, drives economic faster than population—in South Asia, at productivity; Abel, Dey, and Gabe (2013), for about twice the rate (see chapter 2). However, example, find that a doubling of employment urban expansion is often poorly planned and density increases economic productivity by not well integrated with land use and trans- 2–4 percent. Much of this work is based on portation networks. In cities such as Kabul, studies of developed economies, but research- Karachi, and Kathmandu, rapid urban popu- ers are now starting to examine developing lation growth has led to sprawl that is poorly regions, including India (Chauvin, Glaeser, connected to urban activity centers, exacer- and Tobio 2013). bating congestion forces (figure 4.5). This In managing the spatial structure of cities, rapid expansion of urban areas relative to the spatial and urban planning should consider urban population is also reflected in the how the various tools available to control sprawl and ribbon development that is char- density—such as floor area ratio (FAR) and acteristic of Sri Lanka (Lall and Claus 2009). building envelope, land-use controls, and Urban expansion, if not managed preemp- urban planning and design standards—can be tively, becomes locked into the city’s spatial used to facilitate development and to mitigate structure and becomes almost impossible to infrastructure bottlenecks, manage adverse 124 LEVERAGING URBANIZATION IN SOUTH ASIA

FIGURE 4.5 Expansion of built-up area

Source: Zhou 2014.

Source: JICA 2012. PLANNING AND MANAGING SPATIAL STRUCTURE AND CONNECTIVITY 125

BOX 4.5 Equitable planning: Planning, when done right, benefi ts the poor in multiple ways

First, if land-use patterns permit or facilitate because of high transit fares and lower opportu- affordable housing close to work, the poor have nity costs. This fi nding suggests that affordable better access and lower commuting costs. Second, intracity transit and pedestrian-friendly streets when the poor have close and easy access to are important for helping the poor access jobs urban services—water, health care, child care, and urban services. A second fi nding is that pub- welfare offi ces, and other community services— lic transit is important in the mobility of the poor they will be more likely to use them and spend and lower-middle class. Rail is the main mode less on traveling to them. for commuting to work for 23 percent of com- In a detailed study of travel patterns of the muters, while bus remains the main mode for poor in Mumbai, Baker and others (2005) fi nd 16 percent. The modal shares for bus are highest that the poor traveled shorter distances than the for the poor in areas with service, whereas rail lower-middle class and tended to travel by foot shares are highest for the poor in the suburbs.

impacts from incompatible uses, and provide commonly found in many South Asian cities amenities. Increasing density in urban areas undermine the exploitation of agglomeration and promoting compact urban development economies and fail to optimize land resources. can facilitate agglomeration economies, Spatial planning, land management, and improve access to services, and generate prop- addressing informality are thus important in erty tax revenues. If implemented well, these guiding the expansion of cities. density and spatial-structure-shaping actions Cities in South Asia can be better struc- can also lead to more efficient transportation tured as they expand through the creation of networks. The converse is also true: unbridled compact, polycentric, and articulated spatial increase in the density of a city could lead to structures. Multiple specialized subdistricts unmanageable congestion forces that mani- can drive localization economies (Henderson fest themselves in the form of, among other 2003). Examples of such clustering can be ways, streets always clogged with traffic, found around the world: financial districts in unaffordable commercial and residential Mumbai as well as New York, London, and space, and polluted air. Many successful cities Tokyo; entertainment districts in Hollywood have designed and used density controls cre- and Bollywood; and high-tech districts such atively with urban planning design guidelines as Silicon Valley, Silicon Alley in New York, to offer incentives to developers to provide and Bangalore. Multiple centers also gener- public amenities on their development in ally mean shorter commuting times for work- return for higher plot ratios. ers because they have a wider range of choices to locate near work. Managing sprawl and density in Karachi, for example, is a sprawling city of urban areas about 19.5 million people where low-income settlements and unplanned residential areas Overly dispersed land-use patterns with (mainly squatter settlements) make up homogeneous, low densities and urban 16.9 percent (152.8 square kilometers) of the sprawl usually require more land for develop- city’s urban land area, and more than half of ment, and infrastructure costs per person are its population lives more than 10 kilometers higher than where densities are higher. The from the center (Qureshi 2010). It also lacks spread effects and ribbon developments modern mass transit and is wholly dependent 126 LEVERAGING URBANIZATION IN SOUTH ASIA

on road-based public transport, primarily a staggering 3.8-fold spatial expansion in taxis and rickshaws. Karachi should place night-lit areas in 2001–12, indicating that cit- more emphasis on transit systems such as bus ies are expanding very rapidly on their periph- rapid transit, light rail, and subways. It could eries.2 The poor are highly concentrated in seek to emulate cities such as Delhi and, out- this space, with severe unmet demand for side of the region, Bogotá, where the urban services. Upgrading informal settle- TransMileno BRTS has succeeded in reduc- ments, including regularizing land tenure, is ing the average commute time experi- crucial to land management in Afghanistan. enced by workers by more than 10 minutes In Kabul, rapid population growth has been (Yepes 2008). partly driven by the return to the country But Karachi can do more to revise its since 2002 of more than 5.8 million refugees, development plans (figure 4.6) to prevent fur- representing 20 percent of the country’s ther low-density sprawl by considering population, as well as an inflow of internally increased land-use intensity and the forma- displaced persons from, in particular, tion of polycentric structures across the city rural areas.3 As a result, approximately rather than just focusing on the already con- 80 percent of households have some irregu- gested city center. But a more critical problem larity in land tenure, including limitations on is that Karachi needs to improve its inter- using land as collateral. agency coordination, given that the city is At the other extreme, very dense cities lead fragmented into more than 25 land-owning to congestion of roads, basic services, and agencies that need to come together to ensure land and housing markets, not to mention that any planned centers will have the neces- pollution. However, an increasing number of sary infrastructure in place and services to cities are successfully combining density with meet the needs of the residences and busi- high standards of living through innovative nesses located there. spatial planning policies. A comparison of In Afghan cities such as Kabul, unbridled city densities and Mercer’s livability survey expansion and sprawl on the city periphery (figure 4.7) suggests that high livability can be also undermine livability and lead to costly achieved at either high densities (although provision of services, inefficient use of land, fewer cities succeed) or lower densities. and loss of agricultural land. Afghanistan saw Singapore and New York are two examples that offer many lessons in spatial planning and urban management for the dense megaci- FIGURE 4.6 Karachi’s 2020 Development Plan ties of South Asia. After all, only 50 years ago it was almost unimaginable that the small city-state of Singapore—plagued with slums, poor infrastructure, high unemployment, and a growing population—could make the transition from a developing nation into a thriv- ing economy with one of the highest densities of people per land area along with very high living standards. This outcome can also be achieved in South Asian cities through innovative spatial planning policies.

Shaping vibrant neighborhoods through granular planning and smart policies Density can be balanced with livability at a Source: Based on Karachi City District Government 2007. granular level by using innovative urban PLANNING AND MANAGING SPATIAL STRUCTURE AND CONNECTIVITY 127

FIGURE 4.7 Global urban population density and quality of living, 2012

25,000 Chennai Kolkata

Paris 20,000 Mumbai ) 2 New Delhi Seoul 15,000 Tokyo Jakarta Osaka

10,000 New York Urban density (persons/km Yokohama Santiago Nagoya Milan Singapore Hong Kong SAR, China 5,000 MadridBoston Nairobi Busan Munich Berlin

Beijing Barcelona 0 200 150 100 50 0 Mercer quality of living survey ranking, 2012

South Asia Other

Sources: Ranking surveys by Mercer (https://www.imercer.com/products/2014/quality-of-living.aspx); population density derived from UNSD 2014. Note: Size of bubble is proportional to total population of city. Dashed lines in fi gure show median values. design and ﬂ exible policies. Recognizing the conserved buildings at the rear while preserv- need to manage density within a limited land ing the visible features of traditional shop- area, Singapore created market-based policies houses is some areas (Singapore, Urban and incentives in the form of additional ﬂ oor Redevelopment Authority n.d.). area (above master plan–approved plot ratios) South Asian cities can do more to unlock and relaxation of building height controls to development potential by promoting more shape urban design. These policies balance granular patterns of variation in land uses the need to conserve and adaptively reuse and land-use intensities. In Mumbai, which historic buildings and the provision of ample has roughly the same land area as Singapore, open space while encouraging denser devel- undifferentiated floor space indexes (FSIs; opment. Some of these innovative policies can similar to FARs) throughout large areas, eco- be explored in South Asian cities that have nomically incoherent use of transferable strong clusters of heritage buildings within development rights where higher FSIs are their city cores, such as Kandy, Kathmandu, allowed farther from the city center, and his- and Thimphu. Singapore’s Landscaping for torical withholding of large tracts of public Urban Spaces and High Rises program pro- land from the market are clearly responsible motes green and open spaces in dense built- for congestion forces. up areas (Singapore, Urban Redevelopment A comparison of Mumbai’s and New Authority 2014a), and, in some areas, conser- York’s or Singapore’s zoning ordinance vation guidelines allow the extension of maps shows this dramatically. The current 128 LEVERAGING URBANIZATION IN SOUTH ASIA

FSI zones in Mumbai (figure 4.8, panel a) granularity and variation in land uses and appear uniform over large tracts of land and land-use intensities in key areas. Mumbai is are not well connected with specific land revising its development plans to better uses. By contrast, Singapore’s zoning and accommodate development pressures and development control regulations exhibit the need for open space at a granular level high granularity and high levels of land-use by giving more flexibility for higher density variation, and the patterns that emerge in its around transport nodes (figure 4.8, panel b). core are highly livable because of the layer- Ahmedabad also has a draft of a detailed ing of various components. While new areas local area plan and guidelines at the parcel near transport nodes foster high-density level to facilitate the revitalization of its commercial activities and residential hous- CBD, with higher densities, more green ing, other areas preserve historic buildings cover, and a better street network that is for adaptive reuse. Where the CBD sees pedestrian-friendly (box 4.6). demand for taller office towers, care is If managing land use and density on taken to also create open spaces and parks. developable urban land is critical to promot- Within each parcel of land, urban design ing prosperity outcomes, enhancing the and planning guidelines are put in place to open spaces between buildings, such as manage density. streetscapes within the right-of-way and South Asian cities need to reform their public urban spaces, is equally crucial in fos- land-use practices and promote more tering greater livability. Many of the most

FIGURE 4.8 Floor space index patterns in Mumbai

a. Current b. Draft plans to revise index intensities and zones based on transport corridors

Sources: World Bank 2013d; Municipal Corporation of Greater Mumbai 2015. Note: Greater Mumbai land area = 603 square kilometers. CBD = central business district; FAR = fl oor area ratio. The numbers included in panel a denote the maximum allowed FAR. Areas that are shaded the same color have the same maximum allowed FAR. PLANNING AND MANAGING SPATIAL STRUCTURE AND CONNECTIVITY 129

livable cities in the world, such as London, development—are known to be key issues New York, and Vancouver, place heavy facing South Asian cities today. These struc- emphasis on management of public spaces, tural and policy failures impede development streetscape improvements, and pedestrian opportunities and result in artificially high walkability. land prices for businesses. They also restrict Some South Asian cities have also imple- the supply of serviced land for residential mented initiatives to improve public spaces. development and distort the housing market. Colombo has invested heavily in its public The results are poorly serviced informal set- spaces, streetscapes, and waterfront areas. tlements where sizable segments of the urban Ahmedabad has used rights-of-way to incor- population live and high congestion costs porate its BRTS and pedestrian pathways that harm both firms and households. (see box 4.4). Chennai has plans to improve The outcomes associated with poor pedestrian walkability and encourage planning are shared with much of the rest cycling at the commercial hub of Thyagaraya of the developing world. Yet other cities in Nagar through a comprehensive area plan developing countries—such as Surabaya, focused on public land: streetscape enhance- Indonesia—provide examples of good prac- ments, pedestrianized streets, and public tices and offer lessons for fostering sustain- parks (Tamil Nadu Urban Development able regeneration in the city center by Fund, Jones Lang Laselle, and Townland improving urban-suburban connectivity. 2011). This plan is complemented by inter- Surabaya has also made parks and open ventions to manage traffic and improve pub- spaces a priority. lic transport, such as bus priority lanes and Learning from the planning failures of the designated zones for on-street and multi- past means abandoning uniform blanket den- story car parks. sities, static plans, and underutilization of Policies that exacerbate congestion large tracts of public land and adopting new forces—overly restrictive density controls, tools. Granular planning and development the withholding of land from the market, control strategies, successfully adopted by lack of coordination of infrastructure cities such as New York, allow a city to vary

BOX 4.6 Transforming Ahmedabad’s commercial center through granular planning

Ahmedabad’s central business district (CBD) is Urban Redevelopment Authority (AUDA). The characterized by a fragmented stretch of mostly plan proposes to transform the CBD by leverag- small-scale buildings The current fl oor area ratio ing its connectivity to a proposed metrorail sys- in some areas is a mere 1.0 and the CBD is often tem and its location along the waterfront. It also choked with traffi c, impeding pedestrian move- proposes to triple the fl oor area ratio from 1.8 ment. During the past few decades, highly restric- to 5.4, quadruple the population from 85,000 to tive building bylaws have led to new commercial 357,000, and double the street network cover- and residential buildings being scattered across age and green cover (fi gure B4.6.1). the city. These new developments could have The plan includes key elements such as adding added luster to the skyline and helped to defi ne a new streets to improve connectivity and walk- vibrant focal point if they had been concentrated ability, appropriating street-side setbacks and in one area. reducing building footprints, relaxing building Change is happening, however, with an ambi- envelope and height restrictions, incorporating tious local area plan drafted by the Ahmedabad design elements such as arcades, and transferring 130 LEVERAGING URBANIZATION IN SOUTH ASIA

BOX 4.6 Transforming Ahmedabad’s commercial center through granular planning (continued)

FIGURE B4.6.1 The Ahmedabad central business district

b. Proposed plans for Ahmedabad a. Existing street network and density central business district

Existing street network Proposed street network

Existing density Proposed density

Source: Based on Patel 2014.

development rights up to the maximum building features of the area into account, includ- height. In driving the design and implementation ing plots, buildings, streets, trees, and of the local area plan, the AUDA focused on the infrastructure. following: • Supporting regulations. Next, progressive building regulations were drawn up. These • Detailed granular design. The AUDA first regulations allow street widening in parallel conducted a detailed survey of the area, which with redevelopment of existing buildings allowed it to draw up a detailed plan to and will allow more efﬁ cient use of precious improve the street network while taking all (continues next page) PLANNING AND MANAGING SPATIAL STRUCTURE AND CONNECTIVITY 131

BOX 4.6 Transforming Ahmedabad’s commercial center through granular planning (continued)

city-center land. They will also require devel- designed to work with, rather than against, opers to provide public arcades and other market forces to realize public objectives. urban design elements. Even though local planning agencies can only • Working with the market. The AUDA ensured provide incentives for development, build infra- that developers will pay for the additional structure, and regulate development, these ﬂ oor area ratio so as to provide resources for actions play a critical role in sculpting the spatial infrastructure development. Thus, the plan is structure of cities.

land-use types, densities, and built forms planning and implementing policies to (such as height) at the neighborhood level. increase dynamism in city cores. One of the main benefits of granular planning The potential of existing city centers can is that a city can increase the diversity and be unlocked with better land management texture of its neighborhoods by promot- and with infill development. Puerto Madero ing high densities in CBDs and strategic tran- (see box 4.3) and Santiago, Chile, are exam- sit nodes, as is planned for Ahmedabad ples of cities that have undertaken success- (box 4.6), while preserving the intimacy of ful and transformative redevelopment. historic buildings through adaptive reuse and Santiago found that it was more cost-effec- low densities. These plans must be accompa- tive to build infill developments in the city, nied by periodic reviews to help the city supported by existing infrastructure, than respond to changing market conditions and to build completely new developments on demographic shocks. In essence, urban plan- the city periphery (box 4.7), especially if ners need to balance the real estate market’s buildings on the periphery risked being demand for land against development that underserved if amenities and connecting is sustainable and a city’s prosperity and infrastructure were not built. Inner-city livability. projects are often more expensive and challenging than greenfield development Unlocking resources and innovating to because of the frequently significant demoli- rejuvenate city cores tion costs that can be involved and the need to engage in environmental cleanup of sites. As manufacturing moves out of cities like However, such projects can also greatly Dhaka, Kanpur, and Kolkata, the centers benefit surrounding neighborhoods and the need to upgrade economically by attracting rest of the city by acting as a catalyst for advanced services sector or other firms. regeneration. Physical rejuvenation is needed to take on Improved land-administration systems these new functions. Chapter 2 observes that would not just enable systematic land-use several South Asian cities exhibit patterns of planning and support urbanization, but stagnation and even decline in their core would also facilitate investment in industry areas, explained in part by their inability to and infrastructure development. According ﬁ ll the void caused by the suburbanization of to the Association of Builders and Developers formal manufacturing. These trends provide of Pakistan, the government of Pakistan good opportunities to plug the gap with owns 40 percent of land in the country, higher-value-added uses. Cities must actively which is far higher than the 5 percent or so respond to this deconcentration trend by owned by the government in other countries 132 LEVERAGING URBANIZATION IN SOUTH ASIA

with well- functioning land markets. These has plans to move some of its central minis- lands are often auctioned off by land-owning tries away, thereby freeing up land for more government agencies in Pakistan to the high- productive uses. Several historic buildings— est bidder, causing speculative secondary the Auditor General’s Department, an Urban trading and increased land values that are not Development Authority office, and an asylum linked to underlying market fundamentals. In established by the British—were renovated India, poor land records are a critical compo- and adaptively reused as Independence nent of larger land-administration bottle- Square, a commercial complex with public necks that have hindered growth across many open spaces. The waterfront in Colombo is sectors. already being developed for public uses and Colombo has recognized the value of con- hotel development to cater to tourism growth. verting central-city land to better uses and But such development must be accompanied

BOX 4.7 Santiago’s repopulation and housing rehabilitation programs

Santiago’s urban municipal district deterio- SDC calculated that providing services and rated for several decades as residents moved to infrastructure for low-density social housing on low-density housing on the city outskirts. By the outskirts with often underutilized infrastruc- the early 1990s, the city core had degenerated, ture costs about 17 times as much as allocating having lost almost 50 percent of its population the resources to municipalities with existing and 33 percent of its housing stock in just four infrastructure, services, and accessibility such decades. Housing was replaced by warehouses, as Santiago’s Municipal District (ﬁ gure B4.7.1). workshops, motels, and parking lots. This information helped convince the Ministry The mayor started a participatory planning of Housing and Urban Development to support process in 1990 for the renovation of Santiago. the program through a subsidy for urban reno- This process created two programs: Santiago’s vation with the aim of helping lower-middle- Repopulation Program (SRP), which used a and middle-income households buy property in public-private partnership structure to attract priority areas for urban development. new residents and galvanize the housing mar- These efforts were successful, and Santiago’s ket in the municipal district; and the Housing population grew by about 55 percent between Rehabilitation Program (REHA), which built or 2002 and 2012. Since 1990, about 650 building improved 500 apartment buildings comprising permits have been issued and 124,000 housing 7,500 housing units to serve the housing needs units were built, with total private sector invest- of the most vulnerable. ment of about $3 billion in the residential sector. SRP and REHA were part of the broader The main lessons are as follows: initiative to revitalize the inner city. The mayor • Adopt proactive leadership, holistic led this effort with the Santiago Development approaches, and appropriate institutional Corporation (SDC) as a vehicle to stimulate arrangements. The mayor was instrumental in land and housing supply and demand. The city articulating a vision for the city and obtaining sought to improve Santiago’s livability by build- support from the private sector as well as the ing accessible services and public spaces close Ministry of Housing and Urban Development. to workplaces and by persuading developers to Setting up the SDC also created the conditions invest, given renewed demand for housing in the for public-private partnerships with real estate city center. developers. (continues next page) PLANNING AND MANAGING SPATIAL STRUCTURE AND CONNECTIVITY 133

BOX 4.7 Santiago’s repopulation and housing rehabilitation programs (continued)

FIGURE B4.7.1 Cost of developing a new residential unit in city center versus periphery

7,000

6,000

5,000

4,000

Cost ($) 3,000

2,000

1,000

0 City center City outskirts Education Water pipes Sewerage Police surveillance Access and circulation roads Sports and recreation Health Electric power networks Rainwater drainage

Source: Verdugo 2003.

• Use master plans, urban design, and cost- was essential to the program and the commu- benefi t analysis. The master plan was used to nity was consulted at the planning and design attract real estate developers’ interest, which stages. Participation should have continued allowed great fl exibility; and the use of cost- into implementation with better public access benefi t analysis (city center versus periphery) to information, given that the program provided strong justifi cation for the program. revolved around the community. Continued However, with the high fl oor area ratios and community involvement could have prevented building heights, critics felt that urban design neighborhood associations’ lobbying efforts could have been more sensitive to the sur- to stop new development, which emerged in rounding context. the 2000s. • Ensure continuous stakeholder and community participation. Private sector participation Source: Amirtahmasebi and others, forthcoming.

by long-term transformative strategies that reuse existing structures, and formulate new avoid replicating the inefficiencies that policies to rejuvenate city centers. Singapore’s increase congestion forces. successful built heritage conservation pro- Cities should also develop innovative gram, as discussed, allows new uses that strategies to make the best use of their assets respond to market demand and creates and improve their productivity and livability. incentives for owners to maintain and renew Apart from tackling traffic congestion, they dilapidated structures. It even permits parts must redevelop freed-up space, adaptively of heritage buildings to be rebuilt to 134 LEVERAGING URBANIZATION IN SOUTH ASIA

accommodate higher-intensity uses without an important supporter, but it was the non- compromising the historic nature of the profit Friends of the High Line, formed in building. 1999 by two local residents, that advocated Partnerships with enlightened stakehold- for the structure’s preservation and reuse as ers who see value in city assets such as his- a public open space.4 toric buildings and public spaces for better livability can be an important driver of reju- Urban governance in spatial and venation. In Shanghai, China, a private transportation planning in cities developer saw value in preserving two dilapidated blocks of historic shikumen To support integrated management of cities (stone gatehouses) to enhance the urban and a more granular approach to spatial plan- street life and cultural setting as part of the ning, South Asian cities need to strengthen redevelopment of the Taipingqiao neighbor- urban governance and implementation capac- hood (figure 4.9, panel a). In New York, the ity and reform their approach to spatial plan- High Line is a 1.6 kilometer linear park ning and urban land management. A city’s built on a disused elevated railroad spur ability to plan is underpinned by the strength along the west side of Manhattan of its institutions and its implementation (figure 4.9, panel b). Based on the results of capacity. a design competition inspired by a similar project in Paris, the Promenade Plantée, the Better land-resource management railroad structure has been redesigned and reused. The park, which opened in 2009, In many South Asian cities, congestion has revitalized the Chelsea neighborhood pressures in land and property markets are and spurred real estate development exacerbated by large tracts of valuable land in neighborhoods along the out-of-use line. kept off the market and left undeveloped New York Mayor Michael Bloomberg was (see also chapter 5). Prime parcels of land

FIGURE 4.9 Redevelopment in major cities

a. Taipingqiao redevelopment master plan b. Winning design for the High Line

Taipingqiao Xintiandi Open/green development area space

Source: Based on Shui On Group (http://www.shuionland.com Source: James Corner Field Operations and Diller Scofi dio + Renfro, Courtesy of / en-us /property/project/detail/shanghai_xintiandi). the City of New York. Note: Historic Xintiandi blocks shown in gray on the western edge. PLANNING AND MANAGING SPATIAL STRUCTURE AND CONNECTIVITY 135

are often taken up by government buildings Many instances can be cited of poor policy or military cantonments or occupied by an implementation and distortions of land mar- activity that need not reside in the city kets in developing countries that have (box 4.8). Public landowners often do not resulted in strong incentives for squatting on follow local land-use regulations, resulting land in anticipation of windfall gains from in further issues ranging from inadequate rezoning and higher FARs. In addition, road networks to nonexistent urban ser- fraudulent practices are often used to obtain vices. But the main result of high rates of higher FARs. Fraud is more common when public landownership is high land prices the base FAR is kept low and homogeneous caused by supply constraints, which con- across large areas without accommodating tribute to the low affordability of housing localized pressures. The resulting delays in and disenfranchisement of lower-income judicial processes also often create further groups. Planners should ask: Do these pub- inefficiencies, poor accountability, and credi- licly held lands unduly constrain supply bility gaps with the public. and create connectivity bottlenecks between different areas within the city? And can Improving institutional and technical they be put to better use? capacity to manage cities Furthermore, weak institutional land- management structures have led to high A major success factor for urban renewal is transaction costs as well as rampant rent institutional and technical capacity that is seeking by government authorities and pri- sufﬁ cient to develop effective urban spatial vate sector players, often impeding the plans and policies. Achieving efﬁ cient spatial efficient functioning of land markets. It is form requires more than coordination ironic that many spatial planning and land- and management of density and land use. management policies aimed at addressing Shaping livable and prosperous cities spatial bottlenecks in South Asian cities have requires capable city planners to monitor only led to the informal system operating far and facilitate development at a local or gran- more efficiently than the formal system. ular level and to balance development

BOX 4.8 Land use in Kanpur and Amritsar from satellite imagery analysis and ground surveys

According to the United Nations, Kanpur (22 percent, red and magenta) dispersed has a population of about 2.6 million in the throughout the city. city proper and covers 267 square kilometers Amritsar (ﬁ gure B4.8.2) is slightly smaller, (9,565 people per square kilometer). Satellite with a city-proper population of about 967,000 imagery analysis and ground surveys of the on 136 square kilometers (7,109 people per core city area (measuring about 158 square square kilometer). Similar satellite analysis and kilometers) reveal that a large proportion of ground surveys covering about 107 square kilo- land is occupied by military cantonments meters of the core suggest that, here too, large (23 percent, light brown), which appear tracts of land in the core are reserved for mili- to divide the city (figure B4.8.1). Much of tary use (9 percent, light brown). Also similar to the land in the city core is also occupied by Kanpur, much of the city is occupied by informal informal and poorly planned neighborhoods neighborhoods (20 percent, red and magenta).

(continues next page) 136 LEVERAGING URBANIZATION IN SOUTH ASIA

BOX 4.8 Land use in Kanpur and Amritsar from satellite imagery analysis and ground surveys (continued)

FIGURE B4.8.1 Kanpur land use, 2011

Sources: Based on analysis of very high resolution satellite imagery (Zhou 2014). Population data and city size are from the United Nations 2012 Demographic Yearbook.

FIGURE B4.8.2 Amritsar land use, 2010–11

Sources: Based on analysis of very high resolution satellite imagery (Zhou 2014). Population data and city size are from the United Nations 2012 Demographic Yearbook. PLANNING AND MANAGING SPATIAL STRUCTURE AND CONNECTIVITY 137

options and their trade-offs: roads versus the numerous planning and design competi- walkability, density versus sprawl, built-up tions held in China, South Asian cities need to area versus open space, and development build their own technical capacity in the lon- versus heritage (box 4.9). ger term. Unfortunately, such capacity is commonly lacking in South Asia. In India, the Taking a long term view on urban number of planners registered with the development Institute of Town Planners, India (their quality aside) is estimated to be about Even as South Asian cities wrestle with urgent 3,000—or 1 planner per 100,000 urban short-term problems, administrators must residents, a far cry from, say, the United also take an integrated, coordinated, and pro- States and Canada with about 1 planner per active view toward their long-term transfor- 5,000 people (Ramanathan 2013). This mations. Long-term planning prevents new, capacity shortage is notable in small and foreseeable problems and ensures that future- medium cities as well as in large metropoli- focused projects, such as city resilience and tan areas fragmented into multiple cities disaster management, are begun. South Asian that show little coordination. Such adminis- cities must take proactive approaches to plan- trative fragmentation is widespread across ning and ensure that urban policies provide a South Asia, in cities such as Chennai, Delhi, supply of serviceable land for urban use in Kathmandu, and Lahore. both the short and long terms. A proactive Therefore, a new approach is needed to approach means coordinating land-use deci- address city planning and management in sions and transportation planning with an South Asia’s cities. Although international emphasis on accessibility; allocating FAR in a planners can help meet immediate needs, as in rational way that promotes intense land uses

BOX 4.9 Balancing trade-off s in urban development, housing land supply, traffi c, tourism, cultural heritage, and environmental assets in Kandy

The city of Kandy centers around an interna- and housing: much of Kandy is close to protected tionally acclaimed United Nations Educational, forest reserves and sacred areas. Its situation on Scientifi c, and Cultural Organization (UNESCO) hilly, landslide-prone terrain restricts access and world heritage site anchored by the Temple of the overcoming such hazards is costly. A comparison Sacred Tooth Relic. Kandy has an urban popula- of the zoning plan (fi gure B4.9.1) and the hazard tion of only about 166,000, but the city functions zonation map (fi gure B4.9.2; scale is roughly the as a regional transport and services hub in the same as the zoning plan map) shows that much Central Province, serving about 350,000 daily of the Kandy municipal council area zoned for commuters, more than 60 percent of whom arrive residential use is within landslide areas. Adding by public transport. Kandy also attracts many to these pressures are urban design restrictions: tourists. Sited in hilly terrain and an environmen- The main commercial areas (commercial zone 1) tally sensitive area, it faces critical challenges of are within the city center adjacent to the sacred traffi c congestion due to through traffi c, growing areas of the UNESCO site. Accordingly, the city demand for basic services such as drinking water, has imposed height controls on development and the need for environmental protection. and conservation requirements in this area. Kandy is also highly constrained spatially for Because Kandy continues to play a major the land and development intensity of its business transport and tourism role, the long-term (continues next page) 138 LEVERAGING URBANIZATION IN SOUTH ASIA

BOX 4.9 Balancing trade-off s in urban development, housing land supply, traffi c, tourism, cultural heritage, and environmental assets in Kandy (continued)

FIGURE B4.9.1 Zoning plan for Kandy FIGURE B4.9.2 Hazard zonation map for Kandy and surrounding areas

Source: Sri Lanka, Urban Development Authority, n.d. Source: Sri Lanka, Ministry of Disaster Management 2000.

challenges for its transformation are clear. It will trafﬁ c congestion pressures. As an example, be critical for Kandy to develop a spatial strat- the recent freeing up of a large disused former egy and innovative policies to balance economic prison in the city center offers an opportunity development and activities (such as tourism) with for transformative inﬁ ll interventions, while its the need to protect heritage and environmental numerous heritage buildings offer further unlev- assets by unlocking resources and mitigating eraged potential for adaptive reuse.

closer to city centers and facilitates private Any city development strategy, any spatial- sector development; identifying peripheral development master plan or policy, is only as lands where population expansion is likely; good as its content, inclusivity, and enforce- making preparations for future settlement ment. In South Asian cities, existing uses on demands, such as demarcating and protecting the ground are rarely consistent with their an arterial network of future transportation corresponding master plans. Master plans are rights-of-way; and protecting land resources also used as documents to convey ideas and from urban growth. as negotiation tools for arriving at consensus PLANNING AND MANAGING SPATIAL STRUCTURE AND CONNECTIVITY 139

with developers and various stakeholders, as Equally important is the ability to design in Puerto Madero (see box 4.3) and Santiago effective policies and tools to implement plans (see box 4.7). However, in South Asia, master and exercise development control. Singapore and land-use plans are often vague, outdated, has relied on a strong, proactive planning and lacking the type of review that would framework to guide its development in the ensure they meet long-term economic and long term. Since 1971, its Concept Plan has social needs. Most important, public partici- been reviewed four times (every 10 years) and pation is often inadequate, so that the visions incorporated revised growth projections. This these plans present are not truly shared by the progress is reflected in the statutory Master main stakeholders. Plan that guides land use, development inten- Many cities also lack targeted, proper sity, and land supply (box 4.10). The Master land-use and development controls; even Plan is reviewed every five years, and the when they do include regulations, such as release and allocation of land to the market zoning, they are infrequently enforced. by the government is reviewed every six Across South Asia, considerable land is months. owned by public agencies, and these entities Despite the need for long-term vision, typically do not follow municipal land-use plans and zoning designations cannot remain regulations. Land is also often developed static; they need to reflect immediate market without proper titling and registration (see realities while meeting long-term goals. If a also chapter 5). Informal settlements tend to city does not zone enough land or facilitate have roads that are neither of a width ade- land assembly for a particular use in a timely quate to accommodate vehicles nor inte- manner, the supply of land for that use will be grated with adjacent subdivisions, resulting constrained and could lead to higher land in poor accessibility. prices. It is therefore vital that plans and development control regulations be aligned Responding to current needs with market demands and be able to engage private sector developers. In many South Development master plans should be seen as Asian cities, land-market outcomes suffer living documents that are continuously from policies and plans founded on inade- updated and aligned with urbanization chal- quate land-market information or deficient lenges. They anticipate needs, function as interpretations of this information. Effective regulatory frameworks, and build consensus. planning requires up-to-date real estate A common problem in South Asia, however, information—in particular, spatially refined is that population increases frequently over- data on land and housing prices. These data shoot projections. For example, the city plan can be used to monitor land-market perfor- for Chandigarh assumed a population of mance and shortages in land supply and to 500,000, but as of 2011 the population had design effective land-value capture mecha- reached nearly 1.5 million, resulting in over- nisms to help finance infrastructure develop- whelmed transportation systems and infra- ment (box 4.11). structure. Because such underprojections are not uncommon, master plans need to be Communicating a strong vision, revised regularly to match population growth focusing on coordination and and to accommodate development—an activ- implementation ity not well practiced in South Asia. To serve urban population growth and To engage in integrated and coordinated land- economic growth, plans must provide land, use planning, city planners must be backed services, and infrastructure to meet current by strong urban governance and institutional commercial and housing needs. These plans coordination to prepare, implement, and should be responsive to market demand, peri- enforce plans. The key challenges are as odically reviewed, and updated frequently. follows: 140 LEVERAGING URBANIZATION IN SOUTH ASIA

BOX 4.10 Singapore’s urban planning framework and approach to granularity and implementation

Land is a scarce resource in Singapore. With existing and growth areas such as commerce, slightly more than 719 square kilometers support- industry, housing, amenities, open space and ing a growing population of about 5.5 million, greenery, transportation networks, and infra- the city-state has invested heavily in planning to structure. Beyond specifying land use and balance competing land uses and meet the needs density for developments, the Master Plan con- and aspirations of its people. It has put in place an tains detailed layers to further shape the spa- integrated planning and development framework tial structure of the city: plans for parks and that guides plans at the national and local scales, bodies of water; landed housing areas; build- including public-private partnerships, supply of ing heights; activity-generating uses; and street land, and development control (ﬁ gure B4.10.1). blocks, urban design, conservation areas, and The statutory Master Plan typically includes monuments (Singapore, Urban Redevelopment detailed land use to safeguard and accommodate Authority 2014b).

FIGURE B4.10.1 Singapore’s urban planning framework

Legal backing, CONCEPT PLAN Spatial structure with broad land Long-term strong urban 40–50 year time horizon allocation, factoring in long-term integrated governance, and (Reviewed every 10 years) population needs, economic growth planning institutional projections, and so forth capacity MASTER PLAN Detailed and granular land use, 10–15 year time horizon intensity, and accompanying layers (Reviewed every five years) such as urban design, conservation guidelines, and so forth

FLEXIBILITY and In-built flexibility in land-use zoning PLANNING INCENTIVES codes and planning incentives allows some variation to planning and design (Flexibility) parameters

LAND SUPPLY TO Land supply planned in tandem with MARKET market demand and cycles (Reviewed half-yearly)

Development PLANNING Developers and landowners can seek control and DEVIATIONS changes to planning parameters through various avenues based on implementation (Market responsiveness) Market merits of proposed development capacity responsiveness

• Preparation. Limited interpretation of heritage, transportation, and affordable plans; few detailed neighborhood plans; housing. limited integration with a regional frame- • Implementation. Issues of land acquisi- work; lack of integration and ownership tion for development projects; lack of across different agencies; and neglect of funding to implement state and central PLANNING AND MANAGING SPATIAL STRUCTURE AND CONNECTIVITY 141

BOX 4.11 Land-value capture in Hong Kong SAR, China

Transportation investments are largely publicly Hong Kong SAR, China’s metropolitan trans- fi nanced, yet owners of property adjacent to roads, portation authority, the MTR Corporation, transit stations, and other accessibility-enhancing provides a best practice for using land-value facilities typically see their property values increase. capture to bridge its infrastructure investment This outcome can be viewed as a social inequity funding gaps. MTR Corporation does not sell and a conflict of public and private interests— land-development rights to other private devel- although the public pays for the investment, the opers but instead partners with property devel- private owner benefi ts from it fi nancially. opers to construct both the transit infrastructure Many planning and infrastructure special- and property based on a market value that fac- ists argue that the value gained by landown- tors in the “new” transit line. It remains in full ers should be taxed or the total value increase control of the land and can subsequently sell recovered through land-value capture or benefi t the completed units. This mechanism is differ- levies. In many countries, transportation and ent from other models, which typically sell off infrastructure authorities do, in fact, capture the development rights of public land upfront to some or all of these gains. There are several private developers and thus risk losing control models, such as land readjustment, develop- over the land. MTR Corporation is thus able to ment impact fees, and tax increment fi nancing effectively capture the real estate income from (see Ingram and Hong [2012], for a detailed dis- the increased value of the land resulting from cussion of these models and their applications). improved connectivity and accessibility.

projects; lack of synchronization with in authority. In Pakistan, high levels of institu- master plan recommendations; ease of tional fragmentation in and around Karachi deviation from master plan land-use make it difﬁ cult to provide integrated basic recommendations; approval processes services. In India, one survey by the Indian burdened with multiple windows, delays, Institute for Human Settlements (IIHS 2011) and graft that affect financial viability; reveals fragmentation of responsibility for and poor ability to structure public- coverage and integration of spatial and land- private partnership incentives. use planning in large Indian cities. India’s key • Enforcement. Complicated development shortcomings in delivering effective planning controls and codes that are difﬁ cult to inter- have been documented by the Jana Urban pret, thereby creating uncertainty; poor Space Foundation (JUSF 2013), which esti- connection between planning and enforce- mates that only 30 percent of the 4,041 statu- ment bodies that have little ownership or tory towns in India have any spatial plans. input; nonparticipatory nature of planning Even a clear vision and a sound develop- document; ease of land-use manipulation ment plan will remain unrealized without and unauthorized boundary expansion appropriate land-management tools. Many beyond the plan by individuals or groups cities in South Asia face the problem of land with vested interests; and a general lack of assembly to regularize existing developments, enabling policies for enforcement. even before any new development can hap- pen. Even though land assembly is the first Key institutions typically have limited imple- step in allowing city planners to provide bet- mentation capacity, lack decentralization, pos- ter infrastructure and services, it is often a sess different information, and are fragmented contentious, complex, and politically fraught 142 LEVERAGING URBANIZATION IN SOUTH ASIA

process. Bhutan and Gujarat have had some urban planning and policy making is success in the use of land-pooling mechanisms needed. A focus on implementable solutions for land readjustment (box 4.12). Recognizing that work with markets is also required. In the need to manage urbanization, Bhutan has Santiago (see box 4.7), citizen engagement begun developing the necessary legal frame- was strong during the planning and design work to allow planners to better coordinate phases of the repopulation and urban regen- and guide urban development. eration program, but was not carried through to implementation, which resulted in resistance. Similarly, public consensus Engaging citizens and inclusive was needed for Puerto Madero before any planning master plans were implemented (see box 4.3). In Medellín, Colombia, the focus To complement these tools, continuous on public participation and inclusive urban public engagement and transparency in

BOX 4.12 Land pooling and its application in Bhutan and Gujarat

Land pooling and land readjustment refer to • Public participation. Typically, the agreement land assembly through a process by which land of a majority of individual landowners is parcels with different owners are combined into required to proceed. a larger, contiguous land area for more effi cient Bhutan. In 2007, land pooling was recognized subdivision and development. Landowners equi- by municipal act as a development tool and tably contribute a proportion of their plots for approved by the Cabinet in 2009, the same infrastructure rights-of-way and surplus land year the Local Government Act gave Thimphu parcels. These parcels may be sold for either com- powers to carry out land pooling. Bhutan land- mercial purposes or higher-income housing to pooling rules require more than two-thirds of recover part or all of the infrastructure construc- landowners to agree to land expropriation, with tion costs, or may be used for public amenities or a maximum land contribution of 30 percent. low-income housing. The economic rationale for These rules on land transactions in land-pooling land pooling is that the value of the redeveloped areas are complemented by processes and proce- land will be increased for the landowners. dures for community consultation and redress of Land pooling and readjustment are often grievances. favored for the following reasons: • Politically feasible. Because existing landown- Gujarat. Gujarat has widely and effectively used ers share in development gains and costs and its town planning scheme as a land assembly benefi ts are equitably distributed, land pool- tool for almost a century. It has been claimed ing and readjustment are less likely to be (Sharma, n.d.) that 95 percent of Ahmedabad contested. (outside downtown) was developed through the • Cost effective. The government does not have town planning scheme (fi gure B4.12.1). The fi rst to directly fi nance the purchase of rights-of- such scheme was implemented in 1920 and con- way or compensate resettled residents. Some sisted of 270 hectares, followed by a sharp rise infrastructure construction costs can be recov- in schemes after 1985. In 2012, town planning ered through the sale of surplus plots created schemes had increased to 1,200 hectares in urban by the redevelopment. areas (fi gure B4.12.2). The town planning scheme • Minimally disruptive. Owners retain a large has also evolved to allow land deductions of up part of their land. to 40 percent from the earlier 20 percent. (continues next page) PLANNING AND MANAGING SPATIAL STRUCTURE AND CONNECTIVITY 143

BOX 4.12 Land pooling and its application in Bhutan and Gujarat (continued)

FIGURE B4.12.1 Map of town planning schemes in Ahmedabad, 1915–76

Source: Ballaney 2008.

(continues next page) 144 LEVERAGING URBANIZATION IN SOUTH ASIA

BOX 4.12 Land pooling and its application in Bhutan and Gujarat (continued)

FIGURE B4.12.2 Number of town planning schemes in Ahmedabad

1,200 1,000 800 600 400 200 0 0 40 50 0 0 0 0 00 10 –3 – – –6 –7 –8 –9 – – 0 0 40 50 0 0 0 0 93 9 9 96 97 98 99 000 192 1 1 1 1 1 1 1 2

Source: Sharma, n.d.

BOX 4.13 Inclusive urban planning and city revitalization in Medellín

Medellín, the second-largest city in Colombia, has architecture. A metrorail system opened in 1995, experienced rapid growth since the middle of the linking the north and south of the Aburrá Valley. 20th century, from a population of about 358,000 To connect the poorest districts, the city later in 1951 to an estimated 2.44 million in 2013. built two aerial cable-car lines. It used the sys- Its metropolitan region consists of more than tem stations as anchors for “integrated urban 3.5 million people spread across the Aburrá Valley. plans”—a combination of new buildings (such In the 1990s, Medellín’s reputation was as as libraries, schools, and galleries), public spaces the world’s murder capital. This characterization (such as concert venues and parks), and social turned around when, in 2004, the city underwent programs. It has since built 120 schools and nine radical urban revitalization under the leadership signature library parks. A third of the city’s bud- of Mayor Sergio Fajardo. Medellín enhanced law get is allocated to education. enforcement and initiated a series of innovative The revitalization and social urbanism proj- public investments. Mayor Fajardo led a pro- ect was paid for by revenues from the city’s own- cess of community involvement in the planning ership of its utility company. Since its creation and design of these investments as well as public in the 1950s, Empresas Publicas de Medellín participation in municipal funds allocation. The has transferred approximately 30 percent of investments in public works focused on the poor- its annual proﬁ ts to the municipality for social est and most violent areas. Many of these projects investment projects. And while gang violence were designed to integrate the city’s low-income and homicides have not disappeared, they are residents and communities with its commercial remarkably lower than in the 1990s. Inequality center. In 2014, the municipal government spent and unemployment remain key challenges for 85 percent of its $2.2 billion budget on infrastruc- the city, but as a result of the vision of its lead- ture and services for the poorest parts of the city. ership and the buy-in of its citizens, Medellín That spending has included community was named the Urban Land Institute’s Most programs, public transportation, and modern Innovative City in 2013. PLANNING AND MANAGING SPATIAL STRUCTURE AND CONNECTIVITY 145

planning that included the informal settle- consensus with the business community and ment areas and investments in public spaces local population. They must win support and amenities, helped transform the city for transformation and be backed by capa- from a crime-ridden one into a dramatically ble and talented public employees who are more livable urban center (box 4.13). In accountable for their actions—just as the South Asia, social mobilization was funda- mayor of Santiago was instrumental in the mental to the success of the Lahore Walled success of its repopulation and urban regen- City historic residential district rehabilita- eration program, and the government of Sri tion (box 4.14). Lanka laid out a clear plan for a well- connected urban system. These champions Having a stake in shaping cities need not be politicians or administrators: they can include any stakeholders who help Last, transforming cities in South Asia shape the city, like the nonproﬁ t Friends of requires champions—leaders who are able the Highline, which advocated for preserva- to communicate a compelling vision for the tion of a disused elevated rail line as an city, to innovate, and to forge stakeholder urban park.

BOX 4.14 Preserving heritage and improving livelihoods in Lahore Walled City through social mobilization

Lahore’s Walled City was the capital of the electricity, gas, water, and sewerage. The project Mughal and Sikh empires and continues to be also regularized encroachments. a rich part of Pakistan’s cultural heritage. As Because the purpose of the project is to Lahore grew beyond the Walled City, much improve residents’ lives through the restora- of the grandeur of its historic structures, brick tion and rehabilitation of their heritage, social façades, carved wooden balconies, and overhang- mobilization was fundamental to its success. ing jharokas (windows with wooden shutters) fell Led by the local community, the project was into disrepair, in part due to unplanned and hap- supported by social mobilization teams that hazard construction, illegal encroachments, and convinced residents of its beneﬁ ts. About 1,500 neglect of municipal services. The resulting traf- households were engaged in this process, and ﬁ c congestion within the Walled City’s narrow this community involvement enabled the project streets also became a hazard to both homes and to succeed. Social mobilization and local activ- residents. ism also played an important role in engaging The Walled City of Lahore Authority, estab- with squatters outside the 380-year-old Shahi lished by the government of Punjab, has been Hamam (Royal Bath) in their voluntary resettle- working to restore these neighborhoods. The ment outside the Walled City and in negotiating authority has restored the Royal Trail, a 383 a compensation package. The area around the meter heritage trail leading from the Delhi Gate Shahi Hamam has since changed dramatically to the Chowk Kotwali market. It has completed as the building’s façade has been revealed and is restoration of heritage architecture, street paving, being restored. and street furniture and the replacement of infrastructure with underground telecommunications, Source: World Bank 2013a. 146 LEVERAGING URBANIZATION IN SOUTH ASIA

Notes Chauvin, P., E. L. Glaeser, and K. Tobio. 2013. “Urban Economics in the U.S. and India.” 1. The population of Dhaka Statistical Unpublished, Harvard University, Cambridge, Metropolitan Area according to Bangladesh’s MA. 2011 population and housing census was Derudder, B., X. Liu, C. Kunaka, and M. Roberts. 14,543,124. 2014. “The Connectivity of South Asian Cities 2. Night-lit areas were defi ned using a digi- in Infrastructure Networks.” Journal of Maps tal number threshold of 13 and provide a 10 (1): 47–52. proxy measure of built-up urban area (see Ghani, E., A. G. Goswami, and W. Kerr. 2013. chapter 2, box 2.1, for more discussion of “Highway to Success: The Impact of the the use of nighttime lights data to measure Golden Quadrilateral Project for the Location patterns of urban expansion). and Performance of Indian Manufacturing.” 3. See http://www.unhcr.org/pages/49e486eb6 World Bank, Washington, DC. http://elibrary .html. According to UN High Commission .world bank.org/doi/book/10.1596/1813 for Refugees estimates, 683,000 people were -9450-6320. internally displaced by confl ict in Afghanistan Han, S. S. 2010. “Managing Motorization in as of mid-2014, more than half of whom live Sustainable Transport Planning: The Singapore in urban areas. Experience.” Journal of Transport Geography 4. See http://www.thehighline.org for more 18: 314–21. information. Henderson, J. V. 2003. “Marshall’s Scale Economies.” Journal of Urban Economics References 53: 1–28. IIHS (Indian Institute for Human Settlements). Abel, J. R., I. Dey, and T. M. Gabe. 2013. 2011. Urban India 2011: Evidence. Bangalore: “Productivity and the Density of Human IIHS. https://www.citiesalliance.org/sites Capital.” Journal of Regional Science 52 (4): /citiesalliance.org/fi les/IUC%20Booklet%20 562–86. on%20Indian%20cities.pdf. Alder, S. 2015. “Chinese Roads in India: Ingram, G. K., and Y.-H. Hong, eds. 2012. Land The Effect of Transport Infrastructure on Value Capture and Land Policies. Cambridge, Economic Development.” Working Paper, MA: Lincoln Institute of Land Policy. University of North Carolina at Chapel Hill. JICA (Japan International Cooperation Agency). Amirtahmasebi, R., M. Orloff, S. Wahba, and 2012. Data Collection Survey on Traffic A. Altman, forthcoming. Regenerating urban Improvement in Kathmandu Valley: Final land, a practitioner’s guide to leveraging pri- Report. Tokyo: JICA. vate investment. Washington, DC: World Bank. JUSF (Jana Urban Space Foundation). 2013. Andrés, L., D. Biller, and M. H. Dappe. 2013. Urban Spatial Planning and Development Reducing Poverty by Closing South Asia’s Guidelines Report. Bangalore, India: JUSF. Infrastructure Gap. Washington, DC: http://portfolio.janausp.org/?page_id=986. World Bank. Karachi City District Government. 2007. Karachi Annez, P. C., A. Bertaud, B. Patel, and V. K. Strategic Development Plan 2020. Karachi: Phatak. 2010. Working with the Market: Master Plan Group of Offices. http://www A New Approach to Reducing Urban .shehri.org/2020.pdf. Slums in India. Washington, DC: World Lall, S. V., and A. Claus. 2009. Sri Lanka— Bank. http://elibrary.worldbank.org/doi Reshaping Economic Geography: Connecting /book/10.1596/1813-9450-5475. People to Prosperity. Washington, DC: World Baker, J., R. Basu, M. Cropper, S. Lall, and Bank. A. Takeuchi. 2005. “Urban Poverty and Municipal Corporation of Greater Mumbai. 2015. Transport: The Case of Mumbai.” Working “Report on Draft Development Plan—2034.” Paper 3693, World Bank, Washington, DC. Municipal Corporation of Greater Mumbai, Ballaney, S. 2008. The Town Planning Mumbai. http://mcgm.gov.in/irj/go/km/docs Mechanism in Gujarat, India. Washington, /documents/Draft%20Development%20 DC: World Bank. http://hcp.co.in/file Plan/5.%20Draft%20DP%20Report _ manager /publications/Town-Planning-of /DRAFT%20DP%20REPORT%20-%20 -Gujarat_Research-Paper.pdf. 2034%20Final.pdf. PLANNING AND MANAGING SPATIAL STRUCTURE AND CONNECTIVITY 147

Muzzini, E., and G. Aparicio. 2013a. Bangladesh: ———. n.d. “Conservation Guidelines.” http:// The Path to Middle-Income Status from an www.ura.gov.sg/uol/guidelines/conservation Urban Perspective. Washington, DC: World / Conservation-Guidelines.aspx. Bank. Sri Lanka, Ministry of Disaster Management. Patel, B. 2014. “Revitalizing the Center of 2013. “Preliminary Hazard Zonation Ahmedabad.” Background paper for this Map, Kandy District.” Ministry of Disaster report, World Bank, Washington, DC. Management, Sri Lanka. http://www.nbro Qureshi, S. 2010. “The Fast Growing Megacity .gov.lk/web/index.php?option=com_content& Karachi as a Frontier of Environmental view=article&id=168&Itemid=192&lang=en. Challenges: Urbanisation and Contemporary Sri Lanka, Urban Development Authority. n.d. Urbanism Issues.” Journal of Geography and “Kandy Municipal Council Area Zoning Plan Regional Planning 3 (11): 306–21. 2002–2016.” Urban Development Authority, Ramanathan, S. 2013. “Where Are the Sri Lanka. http://www.uda.lk/images Planners?” Jana Urban Space Foundation, /DevelopmentPlans/kandy_zoning.jpg. Bangalore, India. http://portfolio.janausp Tamil Nadu Urban Development Fund, Jones .org/?page_id=677. Lang Lasalle, and Townland. 2011. “Urban Road Development Authority. 2014. “Southern Redevelopment of Thyagaraya Nagar: Revised Expressway.” March 28. http://www.rda.gov Interim Report.” .lk/supported/expressways/stdp.htm. Taubenböck, H., T. Esch, A. Felbier, M. Wiesner, Rosenthal, S. S., and W. C. Strange. 2004. A. Roth, and S. Dech. 2012. “Monitoring “Evidence on the Nature and Sources of Urbanization in Mega Cities from Space.” Agglomeration Economies.” In Handbook Remote Sensing of Environment 117 of Regional and Urban Economics, Volume (February): 162–76. 4: Cities and Geography, edited by J. V. United Nations Framework Convention on Henderson and J.-F. Thisse, 2119–71. Climate Change. 2012. “Ahmedabad Bus Amsterdam: Elsevier. Rapid Transit System, Janmarg.” UNFCCC, Sarzynski, A., and A. Levy. 2010. “Spatial New York. http://unfccc.int/secretariat Efficiency and Regional Prosperity: A /momentum_for_change/items/7098.php. Literature Review and Policy Discussion.” UNSD (United Nations Statistics Division). 2014. Working paper, George Washington Institute “Population Density and Urbanization.” http:// of Public Policy, Washington, DC. http:// unstats.un.org/unsd/Demographic/sconcerns gwipp.gwu.edu/files/downloads/Working /densurb/default.htm. _Paper_041_RegionalProsperity.pdf. Verdugo, M. V. 2003. “Programa de Repoblamiento Schramm, M. 2013. “The Impact of Dedicated Comuna de Santiago: Un Programa de Gestion Freight Corridors on Regional Wages in India: Urbana.” Urbano 6 (8): 53–61. http://www A New Economic Geography Approach.” .redalyc.org/articulo .oa?id=19800811. Background paper for this report, World Bank, World Bank. 2013a. “In Lahore, Improving Washington, DC. Livelihoods and Preserving Heritage Go Hand Schwab, K. 2013. The Global Competitiveness in Hand.” World Bank, Washington, DC. http:// Report 2013–2014. Geneva: World Economic www.worldbank.org/en/news/feature/2013/10/28 Forum. http://www3.weforum.org/docs/WEF /in-lahore-improving- livelihoods-and-preserving _GlobalCompetitivenessReport_2013-14.pdf. -heritage-go-hand-in-hand. Sharma, P. L. n.d. “The Gujarat Land Pooling ———. 2013b. “South Asia Urban Transport Scheme.” Gujarat State: Gandhinagar, India. Analysis.” Background paper for this report, http://www .town planning.gujarat.gov.in World Bank, Washington, DC. / Documents/FinalLandPoolingMechanism.pdf. ———. 2013c. “Trading Across Borders.” World Singapore, Urban Redevelopment Authority. Bank, Washington, DC. http://www.doing 2014a. “LUSH 2.0: Extending the Greenery business .org/data/exploretopics/trading Journey Skywards.” June 12. http://www.ura -across-borders. .gov.sg/uol/media-room/news/2014/jun/pr14 ——— . 2 013 d . Urbanization Beyond Municipal -35.aspx. Boundaries: Nurturing Metropolitan ———. 2014b. “Master Plan.” http://www.ura Economies and Connecting Peri-Urban Areas .gov.sg/uol/master-plan.aspx?p1=View-Master in India. Washington, DC: World Bank. -Plan. doi:10.1596/978-0-8213-9840-1. 148 LEVERAGING URBANIZATION IN SOUTH ASIA

———. 2013e. World Development Indicators. dissertation, Universidad Nacional de World Bank, Washington, DC. http://data Colombia, Bogotá, Colombia. .worldbank.org/data-catalog/world-development Zhou, N. 2014. “Task Report: Land Use -indicators. Mapping and Analysis with VHR Images.” Yepes, T. 2008. “Un Modelo de Ciudad con Background paper for this report, World Bank, Economías Externas en el Consumo.” Ph.D. Washington, DC. Providing Aff ordable Land and Housing 5

Key messages

At least 26 percent of South Asia’s urban • Urban plans, development controls, and the population, an estimated 30 million households, efﬁ ciency of land markets (including informal lives in informal settlements (slums). Both low- markets) are improved. income and middle-income households live in • Cities provide and maintain sufficient these slums. Between 2010 and 2050, the region infrastructure to meet anticipated urban will require an additional 203 million housing growth. units, mostly targeted to low- and middle-income households, to accommodate projected urban • The residential construction industry produces population growth without further expanding affordable housing for all households. the slum population. This challenge can only be • Capital markets channel more funding into met if the following changes occur: housing construction and mortgage ﬁ nance.

Introduction required if these markets are to meet the increasing demand for affordable urban Properly functioning urban land and housing housing. markets are critically needed to accommodate Many South Asian cities fail to enable the South Asia’s growing urban population. Even emergence of efficient and affordable housing though regional urbanization is advancing at markets. Some examples of cities that follow 1 a relatively steady pace, the absolute volume comparatively good practices exist, such as of the overall demand for good-quality hous- Ahmedabad in India and Colombo in Sri ing and serviced land is already daunting. As Lanka, but the vast majority have limited described in chapter 1, countering the adverse technical capacity, inadequate resources to impacts of congestion forces, particularly in finance infrastructure, and counterproductive land and housing markets, is an urgent prior- planning and development control regula- ity for the region. Fundamental reforms are tions (see chapters 3 and 4). The resulting 149 150 LEVERAGING URBANIZATION IN SOUTH ASIA

highly constrained land and housing markets upgrading to actually slowing the growth of exacerbate congestion forces and contribute the slum population in the first place. Offering to a lack of affordable housing and low liva- more options to low- and middle-income bility in cities across the region. South Asia’s households will increase prosperity, reduce municipal, state, and national governments poverty, and enhance the quality of life for must actively address all these limitations. urban dwellers across South Asia. Market and policy failures cause land and This chapter’s four sections discuss the housing prices to be high relative to most impact of urbanization on housing, the failure incomes, which in turn affects households’ of South Asian governments to provide ability to access shelter. For shelter to be affordable housing, strategies to enable affordable, a household should spend no more creation of affordable housing, and key than 25–30 percent of its monthly income on recommendations. housing costs, whether for rent, servicing a mortgage and paying property taxes, or Impact of urbanization on acquiring land and undertaking incremental housing construction and property improvement on a pay-as-you-go basis. Another accepted stan- Urban population growth, whether through dard for rental housing is that monthly rent natural increase or net migration, drives should not exceed weekly household income. urban household formation, generating When costs are higher, as they are in many demand for more housing units. The South Asian cities, households have difficulty variation in household formation—from either owning or renting. families to single-person households to Generally, although conditions vary from groups of unrelated individuals—affects the country to country, the supply of affordable actual type of unit demanded. Increases in urban land and housing lags far behind household income also raise demand for demand. This affordability gap forces house- housing, since higher-income households holds priced out of the formal market to live want larger units with better services. in slums and squatter settlements, move in Finally, if household size (persons per with extended family members, rent smaller household) continues its current gradual units, or move to the city periphery and com- downward trend, more units will be mute long distances. The lack of access to demanded for a given population. affordable housing is, in turn, an important Based on national census and World factor in the generally lagging levels of livabil- Urbanization Prospects data (UN 2012), in ity in South Asian cities (see “Livability of 2010 South Asia was home to an estimated South Asia’s cities” in chapter 1). 403 million households and a population of South Asian cities must both reform land 1.63 billion, thus averaging four persons per and housing policies and foster innovative household. Household size tends to be smaller housing finance. High housing prices can be in urban than in rural areas—research attenuated over time by relaxing land use and conducted by Bongaarts (2001) shows that development control regulations, building average urban household size for a sample of infrastructure to open up land for residential Asian countries (including Bangladesh, India, development, adopting efficient and easy-to- Nepal, and Pakistan) is 96.1 percent of over- use land-titling and land-registration systems, all average household size. Using urban popu- and increasing access to construction and lation projections and assuming (again based mortgage finance. Government regulations on Bongaarts [2001]) that average household also need to stimulate the supply of affordable size declines by 5 percent per 20-year period, rental housing. Finally, city and suburban it is possible to extrapolate household forma- governments need to change their policy tion trends in the eight South Asian countries. stance on informal housing from “curative” This exercise reveals that the compound to “preventive” measures—to go beyond slum annual growth rate in the number of urban PROVIDING AFFORDABLE LAND AND HOUSING 151

households averaged 2.8 percent during lower-middle-income households typically 1990–2010 and will average 2.7 percent dur- cannot afford developer-built housing ing 2010–30 and 2.1 percent during 2030–50. because they cannot mobilize down pay- These growth rates confirm that South Asia ments, obtain bank financing,2 or afford faces substantial affordable housing land in serviced subdivisions (UN-HABITAT challenges. Between 2010 and 2050, the 2011). Therefore, low- and lower-middle- region will add 203 million new households income households commonly build incre- to its cities—an average of 5.1 million house- mentally, improving their dwellings over holds a year. Of this increase, 171 million will time as they accumulate resources to buy be driven by urban population growth and building materials—a time-consuming and the remainder by the projected decline in laborious process. average household size. Low-income, not to mention many lower- The challenge appears to be even greater middle-income, households simply cannot when considering the need to reduce cur- afford to buy or rent formal housing—that rent overcrowding, provide shelter for the is, durable housing that is built on legally homeless, and upgrade dilapidated and pre- titled land, is constructed with proper plan- carious structures. In 2010, an estimated ning permission, and normally complies with 30 million urban households in South Asia building codes and standards. Land con- lived in slums (table 5.1). Therefore, across straints due to government land ownership, South Asia, between 2010 and 2050, at topography, limitations on infrastructure least 233 million housing units will be networks, and restrictive zoning drive up required to accommodate projected urban land prices and undermine low-income hous- growth and address the existing backlog of ing delivery. In some countries, public sector affordable housing. land development agencies restrict land sup- Delivering the necessary affordable hous- ply and target sales to middle-income public ing will be extremely challenging and will workers, ignoring the needs of the poor. And require making land and housing markets housing finance—for both construction and work more efficiently. Private sector real long-term funding—is inadequate to support estate developers tend to build for higher- the needs of low- and lower-middle-income income households, designing and construct- households, making credit expensive and ing housing on purchased lots. But low- and mortgage periods short.

TABLE 5.1 Estimated slum population and number of slum households in South Asian cities, 2010 (except as noted)

Urban Average Total urban Proportion of urban Urban slum Estimated slum population household households population living in population households Country (thousands) size (thousands) slums (percent) (thousands) (thousands) Afghanistan 7,300 7.2 1,014 88.6 6,468 898 Bangladesh 41,476 4.4 9,392 61.6 25,549 5,807 Bhutan 253 5.4 47 – – – India 378,775 3.6 106,637 17.4 65,907 18,307 Maldives 126 6.9 18 – – – Nepal 4,990 6.0 825 58.1 2,899 483 Pakistan 62,290 6.6 9,404 46.6 29,027 4,398 Sri Lanka 3,188 3.8 830 12.0 383 101 Total 498,398 3.9 128,167 26.1 130,233 29,994

Sources: UN World Urbanization Prospects: 2011 Revision; UNESCAP 2012, 126; UN-HABITAT 2013, 126–28; and Offi ce of the Registrar General and Census Commissioner 2013. Note: – = not available. Data on proportion of urban population living in slums are for the most recently available years, as follows: 2011 (India); 2009 (Bangladesh, Nepal, and Pakistan); 2005 (Afghanistan and Sri Lanka). 152 LEVERAGING URBANIZATION IN SOUTH ASIA

The consequences of failing to concerns have been expressed at multiple lev- provide aff ordable housing: els of government that India’s census approach At least one in four urban dramatically understates the country’s slum dwellers lives in slums population. UN-HABITAT (2013) estimates India’s slum population in 2009 as nearly one Slums and informal settlements are wide- in three urban residents. This potential addi- spread in South Asia and may house, at a tional slum population suggests that the num- minimum, a staggering 130 million people in ber of urban slum dwellers for the region as a nearly 30 million households. According to whole may have been as high as 157 million the most recently available estimates, about in 2011, equivalent to the entire population of 26 percent of regional urban development is Bangladesh. unplanned and informal, though this propor- However, not only the poor live in slums tion varies greatly across countries, from only and informal settlements. In Afghanistan, 12 percent in Sri Lanka to nearly 90 percent Bangladesh, Nepal, and Pakistan, the most in Afghanistan (see table 5.1). Estimates for recently available estimates show that Bangladesh, Nepal, and Pakistan range between 7 percent and 30 percent of urban between 46.6 percent and 61.6 percent of the residents live below the official national pov- urban population. In addition to Sri Lanka, erty line. But in each country, the share of the estimated share of the urban population the urban population living in slums is sig- living in slums is much lower for India. nificantly higher (figure 5.1). In Afghanistan, According to India’s 2011 census, the most extreme case, the estimated share of approximately one in six urban residents lives the urban population living in slums exceeds in slums. But since the 2001 census, serious

BOX 5.1 Estimating India’s slum population

According to India’s 2011 census, 17.4 percent administration, or local government, as well as of the country’s urban population—equivalent to housing and slum boards, that may not have 65.5 million people—lived in slum settlements. been formally notifi ed as “slum” under any act; This fi gure is little changed from the 18.3 percent and (3) identifi ed slums—compact areas of at reported by the country’s 2001 census. least 300 people (or about 60–70 households) of But concerns have arisen within India that the poorly built congested tenements, usually unhy- census dramatically underestimates the country’s gienic environments with inadequate infrastruc- true slum population. The UN estimates that in ture and lacking proper sanitary and drinking 2009, 29.4 percent of India’s urban population water facilities. was living in slums (UN-HABITAT 2013). Several problems with the defi nition of slums The 2001 census estimates were based on the and the way the slum population was enumer- defi nition of slums adopted by the Offi ce of the ated caused the 2001 census to dramatically Registrar General and Census Commissioner of understate India’s slum population: India (Census of India). This defi nition identifi es three types of slums: (1) notifi ed slums—all spec- • The definition excludes pockets with fewer ifi ed areas in a town or city notifi ed as “slum” than 60 households having slum-like features. by state, union territories administration, or In many places, slums may be found that have local government, under any act including a only 20–25 households. “Slum Act”; (2) recognized slums—all areas • The census excluded several smaller states: recognized as “slum” by state, union territories Arunachal Pradesh, Dadra and Nagar Haveli,

(continues next page) PROVIDING AFFORDABLE LAND AND HOUSING 153

BOX 5.1 Estimating India’s slum population (continued)

Daman and Diu, Himachal Pradesh, Although the 2011 census increased its coverage Lakshadweep, Manipur, Mizoram, Nagaland, to include some of the previously missing states, and Sikkim. it still only covered slums in the country’s 4,041 • In some states, district or town authorities did statutory towns and therefore failed to enumer- not report all the towns and enumeration ate the slum population of India’s 3,894 census blocks that needed enumeration. towns (settlements that the Indian census recog- • In cities and towns covered under the census, nizes as urban even though they are governed as district and town authorities did not consider rural areas). In addition, the 2011 census per- nonnotiﬁ ed or nonrecognized slums that were sisted in using the minimum 60–70 household the subject of land disputes. deﬁ nition of a slum.

FIGURE 5.1 Slum share exceeds poverty rates, 2005–11

100 90 80 70 60 50 40 30

Percent of urban population Percent 20 10 0 Afghanistan Bangladesh Nepal India Pakistan Sri Lanka Below poverty line Living in slums

Sources: UNESCAP 2012; UN-HABITAT 2013; Offi ce of the Registrar General and Census Commissioner 2013; and World Bank World Development Indicators data 2011. Note: Estimates are for the most recently available years, as follows: 2011 (India); 2009 (Bangladesh and Pakistan); 2005 (Afghanistan and Sri Lanka). For Nepal, poverty data are for 2010 and slum data for 2009. Bhutan and Maldives are not shown owing to lack of data.

that living below the poverty line by basically identical to the proportion living 58 percentage points; for Bangladesh, Nepal, below the poverty line. However, this state- and Pakistan the gap is about 40 percentage ment depends on accepting the 17.4 percent points. For Sri Lanka, based on the most estimate of India’s urban slum population recent (2005) estimates, the gap is much from the 2011 census. smaller—4.8 percentage points. India pres- This disparity means that vast numbers of ents an exception in that the estimated share people live in South Asian urban slums who of the urban population living in slums is are not poor by local standards, whose 154 LEVERAGING URBANIZATION IN SOUTH ASIA

household incomes put them above nation- in slums makes it harder for households to ally defined poverty lines. The implication is move out of poverty. Marx, Stoker, and Suri that factors beyond poverty—such as (2013) present compelling evidence that sev- poorly performing urban land and housing eral slum-related factors contribute to the per- markets, inadequate infrastructure, poor or petuation of poverty, including poor health expensive land titling, and lack of housing outcomes; an inability to access finance and, finance—are important contributory factors more generally, leverage property assets; lack in the formation and expansion of slums and of access to basic services; and difficulty in informal settlements. commuting to jobs. Fortunately, the share of the urban popula- Slum residents are subjected to low-quality tion living in slums in South Asia declined housing in often precarious areas, which during 2000–11 (table 5.2). But this positive adversely affects their health and quality of trend should be qualified. Much of the reduc- life. It is common for these settlements to be tion derives from the removal of one or more in areas prone to flooding and landslides. of the UN–HABITAT (2003a) deprivations Slum housing is on land that has been squat- (that is, nonpermanent shelters, insufficient ted on or has not been properly subdivided living space; no access to safe water; no access and titled; it is built without planning permis- to adequate sanitation) and rarely reflects the sion and does not comply with local building establishment of formal land and property codes. Construction therefore is often title. In other words, reductions tend to reflect unsafe—being, for example, more liable to improved infrastructure access rather than collapse in extreme weather conditions or in improved security of tenure. Also, while the the event of a natural disaster. Additionally, relative proportion of urban populations most informal settlements do not have full living in slums is falling, in most countries the access to infrastructure services such as water absolute number of urban slum dwellers is and sanitation, paved roads, and sidewalks. increasing. As noted by UN-HABITAT (2003b, 172), “In Living in slums puts enormous social, accessible parts of the city, the poor can often economic, and financial burdens on house- afford only precarious sites with insecure ten- holds and can lead to intergenerational ure…. Conversely, affordable sites that may poverty. Some analysts argue that slums are a have more secure tenure are more likely to be natural process of development and that they located in the less accessible periphery of the are simply a transition to modernization and city and involve higher commuting times and adequate housing (Glaeser 2011). But many costs.” For example, in Mumbai a dispropor- argue that they are a poverty trap—that living tionate share of the urban poor live on

TABLE 5.2 Proportion of urban population living in slums for South Asian countries Percent

Country 2000 2001 2005 2007 2009 2011 Afghanistan 98.5 – 88.6 – – – Bangladesh 77.8 – 70.8 70.8 61.6 – Bhutan 44.1 – – – – – India: Census – 18.3 – – – 17.4 UNESCAP/UN-HABITAT 41.5 – 34.8 32.1 29.4 – Nepal 64.0 – 60.7 59.4 58.1 – Pakistan 48.7 – 47.5 47.0 46.6 – Sri Lanka 13.6 – 12.0 – – –

Sources: UNESCAP 2012 (2000, 2005, 2007 data); UN-HABITAT 2013 (2009 data). Note: – = not available. UNESCAP = United Nations Economic and Social Commission for Asia and the Pacifi c; UN-HABITAT = United Nations Human Settlements Programme. Maldives is not included owing to lack of data. PROVIDING AFFORDABLE LAND AND HOUSING 155

the periphery in the poorly connected eastern for tackling South Asia’s affordable housing part of the city (Baker and others 2005). predicament: Worse, as formal urban development continues across the region, slum dwellers face • A short-term approach that is curative or considerable eviction pressures because they remedial and aimed at upgrading and typically lack property title. For example, regularizing existing slum settlements UN-HABITAT (2007) reports that, between • A long-term approach that is preventive 1995 and 2005, 1.12 million persons were and intended to prevent both the expan- evicted in India and more than 242,000 people sion of existing slum settlements and the in Bangladesh. When slums are on private emergence of new ones. land, owners may take back their land to either sell it at high prices or develop it Upgrading of slums is a palliative that will profitably. Nor are slums on public land improve slum conditions but not resolve the exempt from these pressures: governments basic contradictions that cause informal hous- are increasingly relocating slum dwellers to ing to arise in the fi rst place. Therefore, cities build infrastructure or to rectify environmen- must combine short-term upgrading with tally hazardous areas. But because low- longer-term measures to reduce or reverse the income urbanites lack the funds to commute expansion of informal settlements. This dual long distances to work, they often prefer to approach requires concerted efforts from the live as close as possible to their workplaces, region’s governments to create enabling envi- frequently in center-city areas where land is in ronments, including effi cient property regis- high demand and thus expensive. ters and permissive urban planning and When they are evicted, slum dwellers’ development controls that encourage provi- social and economic networks can be severely sion of affordable rental and owner-occupied disrupted, depending on the place of their housing. Reforms are needed at all levels to relocation. As a recent UNESCAP report increase the supply of developable land, states, “Evicting slum households might be an expand urban infrastructure, establish sound effective way of clearing land for other uses, governmental and financial institutions, but almost all evictions, directly or indirectly, develop new fi nancing mechanisms, and nur- result in increased poverty” (UNESCAP ture formal rental housing markets. 2012, 14–15). The threat of eviction also reduces incentives for households to upgrade Upgrading and regularizing informal their housing in place. settlements Overcoming housing supply The critical element slums lack is formal own- constraints: A two-pronged ership, whether by a resident or a rental prop- approach erty owner. Informal settlements in stable areas that are not obstructing infrastructure devel- Urban land and housing markets are highly opment should be recognized (that is, residents complex and require a fl exible supply of land should be provided with security of tenure and developed housing stock to meet grow- through titling) and simultaneously upgraded. ing demand. If markets are operating effec- Such tenure should adapt to local condi- tively, housing supply should expand to tions and cultures. Cities should also develop accommodate increasing demand, easing land-regularization and land-readjustment upward pressure on prices and facilitating programs to improve infrastructure networks provision of accessible options to households and rationalize street patterns. In some cases, of different income levels. Improving the cities may need to rejuvenate and improve responsiveness of land and housing markets underutilized space—including existing slums to demand is difficult, but not impossible. and old derelict areas—by redeveloping and Two complementary approaches are required readjusting land parcels that are too small and 156 LEVERAGING URBANIZATION IN SOUTH ASIA

irregular to support higher-density modern much land is needed to meet the projected development (see also chapter 4). demand for housing arising from urbaniza- Much attention has focused on tion and household formation trends through land-readjustment policies. Nepal, to cite one 2050. Assuming that most of the existing example, has claimed that its land readjust- backlog of 30 million units will be accommo- ment process is effective at redeveloping irreg- dated on existing urbanized land, the answer ular areas. Closer examination, however, depends on the population density of new suggests that the pace has been very slow, development. with fewer than 20 projects in 15 years (Karki The analysis in chapter 2 concludes that 2004). Experience in India also shows that South Asian cities and metropolitan areas are land readjustment is very complex and time rapidly sprawling and observes an overall consuming (Ballaney 2008). As a conse- trend toward declining urban population quence, the country struggles to keep pace densities. During 1999–2010, urban land with slum proliferation. Although land read- area within South Asia grew a little more rap- justment has laudable goals, more streamlined idly than 5 percent a year, about twice the approaches are needed to foster rapid redevel- growth rate of the region’s urban population.3 opment. Even with such streamlining, how- Excluding Afghanistan and Maldives, urban ever, land readjustment by itself is unlikely to population densities decreased at an annual be a sufficient strategy to combat slum average rate of 2.6 percent.4 This rate is con- proliferation. sistent with Angel, Sheppard, and Civco (2005), who estimate that global urban popu- Estimating land requirements for lation densities declined by 2.7 percent a year meeting housing demand by 2050 between 1990 and 2000. Based on a similar analysis using pro- In a long-term, preventive approach, planning jected data, table 5.3 illustrates worst- and and land-management processes need to care- best-case urban land requirement scenarios fully balance projected housing demand and for South Asian countries for the period population growth with land and housing 2010–50. The worst-case scenario assumes supply or production capacity. Before consid- that urban population densities continue to ering what governments should do to carry decline at the same rate as in 1999–2010, the out this approach, it is useful to estimate how best-case scenario that they remain constant

TABLE 5.3 Urban land requirement scenarios, 2010–50 Square kilometers

Best-case Worst-case Urban area Urban area Change in urban Urban area Change in urban Country (2010) (2050)a area 2010–50 (2050)b area 2010–50 Afghanistan 1,969 8,928 6,959 23,560 21,591 Bangladesh 4,865 11,889 7,024 31,844 26,979 Bhutan 148 322 174 866 718 India 236,924 547,553 310,629 1,468,580 1,213,656 Nepal 742 2,536 1,794 6,736 5,994 Pakistan 47,956 118,537 70,581 317,405 269,449 Sri Lanka 4,695 10,512 5,817 28,216 23,521 Total 297,299 700,277 402,978 1,877,207 1,579,908

Sources: Based on UN World Urbanization Prospects: 2011 Revision and analysis of Defense Meteorological Satellite Program–Operational Linescan System nighttime lights data. Note: Urban area is measured as urban lit area with a nighttime light intensity of digital number = 13 or greater (see chapter 2, box 2.1). a. Assuming 2010 population density. b. Assuming 2.55 percent a year reduction in population density. PROVIDING AFFORDABLE LAND AND HOUSING 157

at their 2010 levels. If the worst-case sce- market principles. Berlin, New York City, and nario materializes, cities will need to plan for San Francisco have reallocated large public huge spatial expansion—almost 1.6 million land holdings (an old airport, rail yards, and square kilometers. Even in the best-case sce- a navy shipyard, respectively), increasing their nario, an additional 403,000 square kilome- housing stocks and earning windfall gains ters of land will be required, equivalent to from the sales. slightly more than 50 percent of the entire To carry out this transfer from public to land area of Pakistan.5 private ownership, South Asia’s cities desperately need efficient land tenure and owner- Reforming land management to ship record systems. Their absence prevents increase the supply of land private residential development from reaching a scale sufficient to accommodate urban- One of land management’s main functions is ization. Land titling and transfer procedures making land available for development. are particularly dysfunctional and expensive A critical concern for South Asia is to make in the region (though less so in Nepal; land management more effective. To increase table 5.4), enough so that people tend to rely the supply of developable residential land, cit- on unofficial and undocumented mecha- ies must carry out strategic and integrated nisms to transfer land, particularly when the planning to ensure housing requirements are risk of eviction or sanctions is low. As a met and that space is efﬁ ciently used for sus- result, private developers are reluctant to tainable urban development. Accomplishing acquire and assemble multiple parcels of this task will require cities to enhance their land for residential development because capacity to guide urban development and to they are unsure of the security of title. provide a framework for planning infrastruc- Financial institutions are similarly reluctant ture investments. to finance land development or to accept Many South Asian government agencies land as collateral (de Soto 2000). South Asia in fact own vast tracts of land, but as can learn from other regions to make these Bertaud (2009) points out, they often mis- procedures more efficient, systematic, accu- manage it and constrain urban land supply rate, transparent, and easy to navigate (see also Ballaney and others 2013 and (World Bank 2005a, 2005b). Dowall 1991). A good example comes from Karachi, where a large proportion of land is Eff ective urban planning and regulation controlled by parastatal organizations and to foster the supply of aff ordable land only 31 percent of the city’s land area is and housing under the control of the city district government, thereby restricting development Sound urban planning considers how spatial (figure 5.2). Many of Karachi Metropolitan development patterns are related to both Corporation’s land subdivisions remain accessibility and cost. Spatial patterns of vacant, even though the land was subdivided housing, employment, and services, as well as and serviced with infrastructure in the 1980s the speed, cost, and network density of transit (Karachi City District Government 2007). systems, exert an important influence on Further examples of public ownership urban residents’ access to jobs and services. imposing constraints on the supply of land Housing is the largest investment asset most are provided by Kanpur and Amritsar in households possess, and its location has a India (see chapter 4, box 4.8). huge impact on commuting times. South Cities with massive public ownership of Asian cities need to anticipate urban growth land, such as Karachi, should consider selling and provide adequate, effective, and afford- excess land for residential development. The able transportation services. land should be planned and zoned for resi- In most South Asian countries, urban dential development and priced based on planning regulations, including development 158 LEVERAGING URBANIZATION IN SOUTH ASIA

FIGURE 5.2 Large and fragmented public holdings of land in Karachi Percent

Cooperative Housing Societies, 1.9 Sindh Industrial Trade Estate, 0.6 Port Qasim, Cantonment Board, 2.1 1.5 Government of Pakistan, 0.5 Recent Allocations Railways, 0.4 (Industrial, Education, etc.), 2.7

Karachi Port Trust, 2.9

Private, 3.9 City District Government Karachi, 30.9 MDA, 3.9

Defence Housing Society, 5

LDA, 5.6

Government of Sindh, 17.4 Kirther National Park, 20.7

Source: Karachi City District Government 2007. Note: LDA = Lyari Development Authority; MDA = Malir Development Authority.

TABLE 5.4 Land registration and titling performance, by three measures, 2014

Cost of titling (as a percentage Country Number of procedures Time required to title (days) of property value) Afghanistan 9 250 5.0 Bangladesh 8 244 7.2 Bhutan 3 92 5.0 India 7 47 7.0 Maldives 6 57 16.2 Nepal 3 5 4.8 Pakistan 6 50 7.6 Sri Lanka 9 51 5.1

Source: Doing Business: Measuring Business Regulations, World Bank Group (http://www.doingbusiness.org/data/exploretopics/registering-property). Note: Number of procedures is defi ned as the total number of procedures legally required to register property. A procedure is any interaction of the buyer or seller, their agents (if an agent is legally or in practice required), or the property with external parties. PROVIDING AFFORDABLE LAND AND HOUSING 159 control and zoning regulations, are either Increasing the supply of serviced overly restrictive and inflexible or out of land through investments in urban date and unenforced. Consequently, urban infrastructure development often occurs in a haphazard Cities need to ﬁ nance the expansion of infra- and unplanned manner, resulting in the for- structure and engage in sustainable asset mation of slums and informal settlements. management to increase the supply of devel- As with land registration, complying with oped and accessible land. Infrastructure in planning regulations, obtaining develop- many South Asian cities has failed to keep ment permission, and acquiring build- pace with urbanization; large areas of many ing permits is complex and time consuming. cities lack basic urban services and have poor It is therefore common for builders to fail to livability. Expanding urban land supply is an comply with the formal process of obtain- important step toward increasing access to ing permits, again fostering informal affordable land and housing, but must be development.6 matched by increases in infrastructure ser- Land costs in most South Asian cities are vices if land is to be considered developable also very high relative to the incomes of poor (World Bank 2013). and many middle-income households. In the Chatterton and Puerto (2006) estimate new formal sector, households adjust to high land and replacement investment requirements for prices by living in high-rise structures, as a range of infrastructure: electricity, telecom- development control regulations allow. But munications, roads, rail routes, improved although high-rise construction lowers per water supply, and improved sanitation. They unit land costs, construction costs increase estimate that between 2006 and 2010, South with height and can easily jump fivefold— Asia would need to spend $88 billion (in con- meaning that the poor cannot afford to go stant 2004 U.S. dollars) a year on infrastruc- “high rise” (Bertaud 2010). However, it is ture if the region were to maintain a possible to design low- and mid-rise 7.5 percent GDP growth rate. Focusing only high-density residential structures that are on roads, water, and sanitation, the annual more affordable. Depending on building amount required for new investment and codes and practices, ground plus four-story replacement was estimated to be $44 billion— structures can often optimize density while approximately 50 percent of the region’s total maintaining affordability. infrastructure needs in 2006. Of this amount, Land and housing policy reforms in South $15 billion was required for water and Asia must therefore focus on a range of $10 billion for sanitation. interventions—rezoning more land for low- The per capita costs of road, water, and income housing, changing development con- sanitation infrastructure investments can be trol regulations to permit construction of estimated for South Asia. Assuming an more low-cost housing (that is, smaller units urban population density of 7,000 per- on smaller plots), ensuring that infrastruc- sons per square kilometer, two-lane roads ture services are expanded to support urban are estimated to cost $100 per capita (this is growth, and fostering the formation of both higher than Chatterton and Puerto’s esti- nonprofit and for-profit housing developers mates and reflects resettlement costs in to provide affordable housing. urban areas), improved water $400 per cap- Accommodative urban housing policies ita, and improved sanitation $700 per capita intended to increase the supply of develop- (World Bank 2001; Fay and Yepes 2003; able land and housing can make an enor- Yepes 2005). Table 5.5 presents estimates of mous difference in housing affordability and new and replacement infrastructure costs quality of life. Box 5.2 discusses how hous- for South Asian cities. These estimates ing supply elasticity can vary under rapid should be treated as rough approximations, growth conditions, depending on govern- being subject to variations in urbanization, ment housing policies. 160 LEVERAGING URBANIZATION IN SOUTH ASIA

BOX 5.2 Impact of regulations on housing supply elasticity

In a classic comparative analysis of the Republic Using an approach between a very restric- of Korea, Thailand, and Malaysia, Stephen Mayo tive Seoul and a more laissez-faire Bangkok, and Stephen Sheppard (1996) assess the effects of Kuala Lumpur’s regulations and land-use poli- land use and development control regulations on cies were moderately restrictive—following the the elasticity of housing supply. models used in England and Wales, which bal- In Seoul, development control regulations ance public versus private interests in the plan- were strict and effi ciently enforced. The most ning approval process. The system constrained binding regulation was the adoption of a green- residential development and increased developer belt that, once Seoul’s growth increased, dra- uncertainty, but allowed local authorities to pre- matically constrained housing production. Kim pare master and structural plans and use them and Mills (1988) identify national land use poli- to control development character and intensity. cies as a source of reduced elasticity of housing Mayo and Sheppard (1996) developed an supply and increased prices. econometric model to estimate price elasticities In Thailand, land-use regulations were rela- of housing supply across countries. Applying it, tively lax during 1970–2000 and enforcement they fi nd that price elasticities were signifi cantly was poor. In the 1980s, Bangkok was one of lower in Korea and Malaysia than in Thailand. the few Asian cities to experience a decline Assuming a midpoint income elasticity of in the share of housing classifi ed as informal demand for housing of 0.75, the price elasticity (Dowall 1989). This decline was largely attrib- of housing supply estimates ranged from –0.03 uted to a very high price elasticity of housing for Malaysia to –0.02 for Korea and +6.83 for supply that helped to contain house prices as Thailand—confi rming the impacts of land-use demand increased, due primarily to unenforced and development control regulations on housing regulations. supply.

TABLE 5.5 Urban infrastructure (new and replacement) investment requirements, 2010–50, South Asia Thousands of constant 2004 U.S. dollars

Absolute change in urban population Improved water Improved sanitation Total infrastructure 2010–50 Road investments, investments, $400 per investments, $700 per investment costs, 2010–50, (thousands) $100 per capita capita capita South Asian urban areas Afghanistan 25,802 2,580,200 10,320,800 18,061,400 30,962,400 Bangladesh 59,881 5,988,100 23,952,400 41,916,700 71,857,200 Bhutan 298 29,800 119,200 208,600 357,600 India 496,608 49,660,800 198,643,200 347,625,600 595,929,600 Maldives 140 14,000 56,000 98,000 168,000 Nepal 12,062 1,206,200 4,824,800 8,443,400, 14,474,400 Pakistan 91,677 9,167,700 36,670,800 64,173,900 110,012,400 Sri Lanka 3,888 388,800 1,555,200 2,721,600 4,665,600 Total 690,356 69,035,600 276,142,400 483,249,200 828,427,200

Sources: UN World Urbanization Prospects: 2011 Revision; Chatterton and Puerto 2006. PROVIDING AFFORDABLE LAND AND HOUSING 161 infrastructure costs, technological change, Establishing well-defi ned institutional and design innovation. With this caveat in arrangements and good governance mind, the infrastructure investment costs Developers need access to capital to ﬁ nance are staggering—more than $800 billion over land acquisition and construction, which in 40 years, an average of $20 billion per year, turn requires efficient banking systems to nearly three-quarters of which is required for mobilize funds from savers and disperse them India alone. to borrowers. Unfortunately, many develop- Improving access to affordable land and ing countries have inadequate financial housing will require massive investments in markets. Local governments must be empow- financially sustainable and affordable mod- ered and enabled to develop new financial els of infrastructure finance. One reason for tools for funding infrastructure investment the failure of housing markets to drive and operations. At the same time, reforms development is that developers (both pri- that remove disincentives to raise local reve- vate and public) are unable to afford the nues to fund infrastructure (for example, investment to build, operate, and maintain major trunk networks that straddle jurisdic- infrastructure without financial instruments tions) need to be expedited. Financial tools that allow them to recover their costs. should be supported by, at a minimum, Innovative approaches to infrastructure developer impact fees, vacant land taxes, provision should include an overhaul of user charges, beneficiary charges, land- existing policies and the introduction of readjustment tools, property taxes, and spe- tools to enable infrastructure financing, cial assessment districts. such as ad valorem property taxes, strategic Institutional arrangements are crucial in disposition of publicly held land, betterment service provision and delivery in cities. Their levies, developer exactions, impact fees, and importance extends beyond implementing public-private partnerships (Peterson and policies and programs to identifying issues at Annez 2007; Peterson 2009). Cross- the national and local levels. In South Asian subsidies between national and local gov- countries, ensuring smooth horizontal coordi- ernments could be introduced. Countries in nation (across local jurisdictions) and vertical the region could also learn from Indonesia, coordination (between different levels of gov- where the government has introduced ernment) is a major challenge. Financial and mechanisms such as viability gap funding to other arrangements between national and help mobilize the private sector for subnational governments need greater func- public-private projects that are economi- tional clarity. (Chapter 3 provides a detailed cally feasible but not yet financially viable.7 discussion of issues relating to the governance Tariff and rate structures for utilities and financing of South Asian cities.) should be reviewed and aligned with investment, operations, and maintenance costs for long-term financial sustainability along with Employing innovative fi nancing social equity. For example, studies of water mechanisms pricing in Bangalore, Chennai, Colombo, Dhaka, Hyderabad, and Kathmandu have Improving access to affordable land and hous- concluded that subsidies delivered through ing will not work without a robust housing tariffs often do not reach their intended finance system. Financing is important for beneficiaries—the urban poor—instead fre- both the demand and supply sides of the quently benefiting wealthier segments of market. Developer access to ﬁ nance for both society. In some cases, the poor actually pay land acquisition and construction is important a higher price for water than the rich. for helping to stimulate housing supply and Subsidies also threaten utilities’ financial sus- thereby keeping housing affordable in the face tainability, undermining equitable provision of growing demand. Meanwhile, even with (World Bank 2002). enhanced supply, households of all income 162 LEVERAGING URBANIZATION IN SOUTH ASIA

levels need access to mortgage fi nancing to be inclusionary zoning to provide developers able to afford housing. with density bonuses in return for building Although India is the undisputed leader in housing for low- and lower-middle-income housing finance in South Asia, mortgage bal- households. Such strategies have been very ances stood at a mere 3.97 billion rupees effective in the United States. Another supply- (equivalent to $62 million in current U.S. dol- side strategy would be to offer builders tax lars) in 2009, with housing finance limited to credits or incentives to build low-income units. upper-income, formally employed population groups (Nenova 2010). Enabling a sustainable, formal rental To improve market conditions, three ele- housing market ments are required: First, more capital should be mobilized in the housing sector, Relatively little research has been done into helping expand secondary mortgage markets rental housing markets in developing coun- and thus access to mortgage lending. Second, tries. Notable exceptions are Peppercorn and developers need more specialized lenders for Taffin (2013) and UN-HABITAT (2003b). construction and land acquisition finance. Peppercorn and Taffi n argue that all countries Third, specialized lending programs need to should be concerned about the development be developed to ensure that low- and of sustainable rental housing markets as part lower-middle-income households gain access of their overall housing strategies. As they to affordable credit. suggest (Peppercorn and Taffi n 2013, xvi), These changes will be difficult. Among the Enabling the development of a healthy main reasons for the lack of financing options formal rental housing sector is impor- for both developers and households are the tant for a number of reasons. First, the lack of an adequate supporting legal frame- rental sector is a natural outlet for those work, including poor foreclosure and eviction households that do not have sufficient procedures and land titling regimes; absence income to afford a home or have not of reliable property valuation; lack of a well- saved enough to meet down-payment functioning collateral system; a missing yield requirements. Second, because in many curve; and absence of long-term treasury countries, a good percentage of the instruments in some countries. Further com- income earned is informal, there are lim- plications include weak competition in the its to the share of the population that financial sector, poor transparency, underde- can qualify for mortgage loans. Third, veloped market structural features such as vibrant rental markets are necessary for second-tier lenders, and the lack of a level workers’ mobility. Fourth, home owner- playing field for financial institutions. ship produces greater urban sprawl. Funding distortions tend to handicap the This is particularly true as housing most dynamic and efficient actors; in some prices increase and people are forced to countries, a legacy of nonperforming housing move farther and farther away from the loans among state-owned banks plagues the city center. system to this day (Nenova 2010). Both demand- and supply-side approaches The supply of both public and private sector should be explored by governments to support rental housing in South Asia lags behind the poor in obtaining rental or owner-occupied demand. According to UNESCAP (2012), the housing—each country should pursue an overall share of urban rental housing across approach that best meets local conditions and South Asia is estimated to be 30 percent of capacity. On the demand side, governments the housing market. Most rental housing is could provide housing vouchers or other private, owned and operated by small-scale forms of targeted subsidies to help the poor property owners (Peppercorn and Taffin gain access to market-rate housing. On the 2013). Much of it is rented on an informal supply side, cities could implement basis with no lease or strong tenure security. PROVIDING AFFORDABLE LAND AND HOUSING 163

Such practices circumvent payment of income Key recommendations taxes and other government levies. In manufacturing zones, private rental housing is The key message of this chapter is that South often aimed at single tenants, not families. Asian cities are struggling to provide afford- Although most governments in the region able land and housing to accommodate urban operate rental-housing programs, they are growth. Four imperatives are critical to mak- small and grossly underfunded. ing housing affordable: Private rental housing in South Asian cit- • Land inventory must be managed to ies can be expensive and beyond the reach of ensure that housing demand and supply the poor as well as the middle class. are in balance, and an adequate stock of Historically, this situation has prompted affordable land must be provided to countries such as India to impose rent accommodate growth. controls. Although well intended, rent con- • Land must be provided with infrastruc- trols lead to disinvestment in the existing ture services to be developable; there- rental stock, resulting in its deterioration, fore, cities need to actively program and limit the production of new rental infrastructure investment to support housing, thus exacerbating the shortage of growth. decent, affordable rental housing. In • Land and housing development requires Mumbai, where rents are controlled by the scalable residential development that 8 Maharashtra Rent Control Bill, 20,000 meets the needs of all households and rental properties were abandoned by their income groups. owners. These units suffered deferred main- • Finance plays a critical role in supporting tenance, and each year many units became land and housing development, in con- uninhabitable (Keating, Teitz, and Skaburskis struction fi nance, and in mortgages. 1998).9 Recognizing these side effects, many South Asian cities (but not Mumbai) have Regardless of country, city size, or land and repealed their rent control policies or limited housing market conditions, policy makers their application to existing buildings (Arnott should follow a common sequence of reform 2008). Meanwhile, informal rental units activities. First, local and central govern- such as those in slums and informal settlements should loosen the constraints that bind ments may provide affordable accommoda- land market supply. Freeing up supply means tion, but often of poor quality, limited space, releasing public land holdings, reforming and weak tenure security (property owners land-use plans and regulations, and increas- can raise rents or evict tenants at will). ing the supply of land for residential develop- The government could also directly pro- ment. Publicly owned land should be sold to vide low-income rental housing. Although residential developers, a move that would practiced by many countries, the direct provi- both increase residential land supply and sion of low-income housing has fallen out of generate revenues from land sales or leases favor in developed and developing countries (Annez, Huet, and Peterson 2010). Second, alike for several reasons. First, production is cities should revise land-use regulations to expensive and generally cannot keep pace facilitate increased housing production. with demand. Second, large-scale projects Third, and related to the fi rst step, local and tend to concentrate the poor in “ghettos.” central governments should foster increased Third, most of these projects are on inexpen- investment in infrastructure to support land sive land far from jobs. In many documented development for housing. Local resources cases, housing recipients game the system by need to be mobilized to program and main- renting out their units to higher-income tain infrastructure systems. Fourth, real households willing to commute longer dis- estate construction fi rms need to be restruc- tances (because they can afford the higher tured and modernized to increase the quan- transportation costs). tity and affordability of their housing output. 164 LEVERAGING URBANIZATION IN SOUTH ASIA

Finally, construction and housing finance infrastructure investment and operations. needs to be broadened and deepened to fos- Tools that should be considered include, ter the production of affordable housing. at a minimum, developer impact fees, Crosscutting recommendations are as vacant land taxes, user charges, benefi- follows: ciary charges, land-readjustment tools, property taxes, and special assessment Land districts. • Improve the responsiveness of urban land and housing markets to housing demand Real estate industry by revising urban planning and development control regulations and establishing • Set annual targets for the production of practices that encourage private develop- affordable housing, including incremental ment of housing. Enabling actions include housing development. revising standards for dwelling units and • Convene taskforces, sponsored by both investing in local and national capacity to central and local governments, to formu- manage and control development. late recommendations for modernizing • Adopt reforms to improve the effi ciency of the construction industry. land and property registration systems. • Examine the following: removal of barri- • Divest excess publicly held land in cities ers to land assembly and subdivision, and reallocate it to residential use. land titling and registration, and access • Target opportune sites for affordable to infrastructure; impacts of urban plan- housing provision. ning and development control regulations on housing construction costs; and methods for improving housing and residential Infrastructure construction fi nance. • Improve the monitoring of housing • Encourage developers to invest in rental demand and supply conditions at city and housing for those who cannot afford metropolitan levels and adjust plans and owner-occupied housing or who prefer infrastructure programs. In large agglom- renting. erations fragmented across several juris- • Guide households that cannot afford for- dictions, enhance horizontal coordination mal housing in incremental development and collaboration in the provision of of shelter on the plots they occupy. infrastructure. • Develop financially sustainable and affordable models of infrastructure Finance fi nance that provide suffi cient funds for • Develop or expand existing housing building, operating, and maintaining finance institutions to channel more infrastructure. Adequate fi nancing will funding into housing. require substantial resource mobilization, • Consider underwriting, risk management, including user fees and taxes. and loan allocation policies to improve • Overhaul infrastructure provision and access to, and affordability of, long-term give greater consideration to public-private mortgages. Develop underwriting criteria partnerships. to provide developers with access to con- • Review tariff and rate structures for utili- struction loans. ties and realign with investment, operations, and maintenance costs. Policy makers should not assume that the • Develop new financial tools at the above recommendations are geared to local government level for funding work in all countries and in all sizes of cities. PROVIDING AFFORDABLE LAND AND HOUSING 165

The potential effectiveness of each action 7. See http://www.jica.go.jp/press/2012/ ku57pq should be evaluated locally based on cultural 000012e8t8-att/20130124_02_04.pdf. context and institutional and technical capac- 8. The Maharashtra Rent Control Bill was ity. In federally structured countries—India passed in 1999 and replaced the Bombay and Pakistan, for example—the central gov- Rents, Hotel and Lodging Housing Rates Control Act of 1947. The approaches to rent ernment should play a larger role in fostering control in the two acts closely resemble one efficient and affordable land and housing another (Gandhi and others 2014). market operations. In more decentralized 9. See Gandhi and others (2014) for a discus- countries, cities should consider options that sion of Mumbai’s rent control system and its they can implement effectively. adverse impacts on the city’s rental housing stock. Notes 1. However, as discussed in chapter 2, there are References important variations in the pace of urbanization across countries in the region. In Angel, S., S. Sheppard, and D. Civco. 2005. particular, the pace has been more rapid in The Dynamics of Global Urban Expansion. Maldives, Bhutan, Nepal, and Bangladesh Washington, DC: World Bank. than in the rest of the region. Annez, P. C., G. Huet, and G. Peterson. 2. Experience in the United States indicates that 2010. “Working with the Market: A New when a household devotes more than 40 per- Approach to Reducing Slums in India.” Policy cent of its income to housing, it is at risk of Research Working Paper 5475, World Bank, default. Washington, DC. 3. As mentioned in chapter 2 (see, in particu- Arnott, R. 2008. “Housing Policy in Developing lar, note 24), some care is, however, required Countries: The Importance of the Informal in comparing the estimated rate of expan- Economy.” Working Paper 13, Commission sion of urban land with the growth rate of on Growth and Development, World Bank, urban population. Although the estimated Washington, DC. rate of expansion is based on an analysis Ballaney, S. 2008. The Town Planning of nighttime lights data, the growth rate of Mechanism in Gujarat, India. Washington, urban population is derived from World DC: World Bank Institute. Urbanization Prospects: 2011 Revision data Ballaney, S., M.-A. Bertaud, P. Annez, and based on ofﬁ cial national deﬁ nitions of C. K. Koshy, B. Nair, B. Patel, V. Phatak, urban areas. and V. Thawakar. 2013. “Inventory of Public 4. Maldives has a limited island land area and Land in Ahmedabad, Gujarat, India.” Policy cannot expand except through reclamation. Research Working Paper 6664, World Bank, Consequently, it has an extremely high popu- Washington, DC. lation density of nearly 47,000 persons per Baker, J., R. Basu, M. Cropper, S. Lall, and square kilometer (470 persons per hectare). A. Takeuchi. 2005. “Urban Poverty and Instability in Afghanistan has also affected Transport: The Case of Mumbai.” Policy urban population density as households Research Working Paper 3693, World Bank, move to urban areas for better security. In Washington, DC. 1999–2000, Afghanistan had a very high Bongaarts, J. 2001. “Household Size and urban population density of 19,314 per- Composition in the Developing World.” sons per square kilometer. Working Paper 144, Population Council, 5. Pakistan has a land area of 770,880 square New York. kilometers according to data from the World Bertaud, A. 2009. “Urban Land Supply: Real Bank World Development Indicators. Resource Constraint or Man Made Shortage?” 6. Recommendations on improving urban plan- Presentation at World Bank, Washington, DC, ning that would also assist in stimulating the March 9–10. http://siteresources. worldbank supply of formal affordable housing are cov- .org/INTURBANDEVELOPMENT ered in depth in chapter 4. /Resources/336387-1235146842675 / bertaud.pdf. 166 LEVERAGING URBANIZATION IN SOUTH ASIA

———. 2010. “Land Markets, Government International Comparison.” Journal of Interventions, and Housing Affordability.” Housing Economics 5 (3): 274–89. Working Paper 18, Brookings Institution, Nenova, T. 2010. Expanding Housing Finance Wolfensohn Center, Washington, DC. to the Underserved in South Asia: Market Chatterton, I., and O. S. Puerto. 2006. Estimation Review and Forward Agenda. Washington, of Infrastructure Investment Needs in the DC: World Bank. South Asia Region. Washington, DC: World Office of the Registrar General and Census Bank. Commissioner. 2013. Primary Census de Soto, H. 2000. The Mystery of Capital: Why Abstract for Slum. New Delhi: Government Capitalism Triumphs in the West and Fails of India. http://www.censusindia .gov.in/2011- Everywhere Else. New York: Basic Books. Documents/Slum-26-09-13.pdf. Dowall, D. E. 1989. “Bangkok: A Proﬁ le of an Peppercorn, I. G., and C. Taffin. 2013. Rental Efficiently Performing Housing Market.” Housing: Lessons from International Urban Studies 26 (3): 327–39. Experience and Policies for Emerging ———. 1991. “The Karachi Development Markets. Washington, DC: World Bank. Authority: Failing to Get the Prices Right.” Peterson, G. 2009. Unlocking Land Values to Land Economics 67 (4): 462–71. Finance Urban Infrastructure. Washington, Fay, M., and T. Yepes. 2003. “Investing in DC: World Bank. Infrastructure: What Is Needed from 2000 to ———, and P. C. Annez. 2007. Financing 2010?” World Bank Policy Research Working Cities: Fiscal Responsibility and Urban Paper 3102, World Bank, Washington, DC. Infrastructure in Brazil, China, India, Poland Gandhi, S., V. Tandel, S. Patel, A. Pethe, and South Africa. Thousand Oaks, CA: Sage K. Agarwal and S. J. Liberio. 2014. “Decline Publications. of Rental Housing in India: A Case Study UN (United Nations). 2012. World Urbanization of Mumbai.” Marron Institute of Urban Prospects, the 2011 Revision. New York: UN Management Working Paper 10, New York Department of Economic and Social Affairs. University, New York. UNESCAP (United Nations Economic and Glaeser, E. 2011. Triumph of the City: How Our Social Commission for Asia and the Paciﬁ c). Greatest Invention Makes Us Richer, Smarter, 2012. The State of Asian Cities 2010/2011. Greener, Healthier, and Happier. New York: Bangkok: UNESCAP. Penguin Press. UN-HABITAT (United Nations Human Karachi City District Government. 2007. Karachi Settlements Programme). 2003a. The Strategic Development Plan 2020. Karachi: Challenge of Slums: Global Report on Human Master Plan Group of Offices. http://www Settlements 2003. Nairobi: UN-HABITAT. .shehri.org/2020.pdf. ———. 20 03b. Rental Housing: An Essential Karki, T. 2004. “Implementation Experiences of Option for the Urban Poor in Developing Land Pooling Projects in Kathmandu Valley.” Countries. Nairobi: UN-HABITAT. Habitat International 28 (1): 67–88. ———. 2007. Forced Evictions: Towards Keating, W. D., M. B. Teitz, and A. Skaburskis. Solutions? Second Report of the Advisory 1998. Rent Control, Regulation and the Group on Forced Evictions to the Executive Rental Housing Market. New Brunswick, Director. Nairobi: UN-HABITAT. NJ: Center for Urban Policy Research. ———. 2011. Affordable Land and Housing in Kim, K. H., and E. Mills. 1988. “Korean Asia. Nairobi: UN-HABITAT. Development and Urbanization: Prospects ———. 2013. Streets as Public Spaces and and Problems.” World Development 16 (1): Drivers of Urban Prosperity. Nairobi: 157–67. UN-HABITAT. Marx, B., T. Stoker, and T. Suri. 2013. “The World Bank. 2001. “Road Cost Knowledge Economics of Slums in the Developing World.” System (ROCKS).” World Bank, Washington, Journal of Economic Perspectives 27 (4): DC. http://www.worldbank.org/transport 187–210. / roads/rd _tools/rocks_main.htm. Mayo, S., and S. Sheppard. 1996. “Housing ———. 2002. Tariff Structures in Six South Supply under Rapid Economic Growth Asian Cities: Do They Target Subsidies Well? and Varying Regulatory Stringency: An Washington, DC: World Bank. PROVIDING AFFORDABLE LAND AND HOUSING 167

———. 2005a. “Conﬂ icts over Property Rights ———. 2013. Planning, Connecting and and Resolution of Disputes in Kabul.” Kabul Financing Cities—Now: Priorities for City Urban Policy Notes Series 6, Washington, DC. Leaders. Washington, DC: World Bank. ———. 2005b. “Will Formal Documents of Yepes, T. 2005. “The Expenditure on Title and the Courts Resolve All Land Infrastructure in East Asia Region, 2006–10.” Disputes?” Kabul Urban Policy Notes Series 5, Background paper for EAP Flagship, Washington, DC. World Bank, Washington, DC.

Sustaining Prosperity by Building Disaster-Resilient Cities 6

Key messages

Many South Asian cities are immensely vul- • Identify risk by developing urban risk assess- nerable to natural disasters because economic ment frameworks and hazard maps activity and residents are highly concentrated • Mitigate risk by planning critical and multi- in areas prone to natural risks. This risk is purpose infrastructure to be safe and resilient exacerbated by high poverty rates and weak • Build management capacity in urban resil- governmental readiness for natural hazards, ience among national and local institutions, which result in a high human impact of disas- and develop extensive risk data platforms and ters. To reduce hazard risks and plan for more make them publicly available resilient cities, policy makers need to do the • Establish a national catastrophe risk ﬁ nancing following: mechanism.

Introduction their exposure to natural risks. Megacities have larger populations at risk, while second- Historically, South Asia has been one of the ary cities have less capacity to manage natural least urbanized global regions. However, its disasters. urban population has grown steadily during Increasing urban resilience is particularly recent decades as the result of both natural vital because of the large numbers of people increase and rural-urban migration. Between at risk and because physical assets in cities are 2011 and 2030, 302 million inhabitants will concentrated in risk-prone areas. If South be added to South Asia’s urban population Asia is to leverage urbanization for economic (see chapter 1). So far, urban population growth, it must ensure that its cities can limit growth in both megacities and secondary the impacts of natural disasters. cities has to a large extent been unplanned, Disasters are a function of three inputs: haz- resulting in the concentration of the poor ards, exposure to hazards, and vulnerability. population in risk-prone areas and increasing Hazards are a fixed element in this equation

169 170 LEVERAGING URBANIZATION IN SOUTH ASIA

because of the geoclimatic characteristics of ranks South Asian countries as having the South Asian region. The presence of the medium (Bhutan, Nepal, and Sri Lanka) to Himalayan Mountains and the coasts of very high (Afghanistan, Bangladesh, India, the Indian Ocean, the Bay of Bengal, and the and Pakistan) potential mortality risk due to Arabian Sea interact to result in droughts and exposure to multiple disasters and extreme floods on the plains and cyclones that start in vulnerability (UNISDR 2009). The mountain- the Bay of Bengal and the Arabian Sea. Natural ous regions of Afghanistan, Bhutan, India, hazards are a regional phenomenon because Nepal, and Pakistan face risks from earth- geological formations and river basins are quakes and landslides; the coastal regions of shared among countries. Bangladesh, India, Maldives, and Sri Lanka The exposure of people and assets to haz- are at risk of cyclones, storm surges, and ards is significant because of unplanned coastal erosion; and coastal regions through- growth (see chapter 4) and the lack of risk- out South Asia as well as riparian regions2 of identification measures. Exposure to natural Nepal and Bhutan are highly prone to flood- disasters is a result of the concentration of ing. Drought, the disaster that affects the larg- people and assets in risk-prone areas such as est number of people in the region, is flood zones or areas vulnerable to earthquakes. especially important in most of Afghanistan, In South Asia, the number of people exposed India, and Pakistan, and parts of Nepal and to natural hazards is growing by 3.5 percent a Sri Lanka (Practical Action 2009; Gupta and year —the fastest growth rate in the world. By Muralikrishna 2010). 2050, 246 million South Asians will reside in The underlying natural risks combined cities in cyclone-prone areas, compared with with the region’s large population, inadequate 160 million people in member countries of the infrastructure, and socioeconomic conditions Organisation for Economic Co-operation and result in high vulnerability to hazard impacts Development (Bronkhorst 2012). (Bronkhorst 2012). From 1971 to 2009, Vulnerability to natural disasters is South Asia experienced 1,017 natural disas- extremely high because of socioeconomic fac- ters that affected more than 2 billion people, tors and lack of proper governance of hazard caused more than 800,000 deaths, and led to risk. Most of the population in risk-prone more than $80 billion in direct losses areas are the poor and underserved groups (Bronkhorst 2012). This exposure is shaped living in temporary structures and without in large part by the region’s geography as a access to either early warnings or means to major drainage basin of the Himalayas and evacuate quickly in the face of a natural disas- proximity to the monsoon and typhoon ter. The poor are concentrated in high-risk, trajectories. dense urban areas so that they can be close to South Asian countries are developing and jobs, which has increased the number of strengthening their institutions to mitigate the vulnerable people exposed to imminent dan- problems of natural hazard risk. However, ger. The number of fatalities due to disasters the local and national disaster risk manage- since 1900 in South Asia is second only to ment (DRM) institutions that have been that in East Asia and the Pacific, and hydro- established across the region have not been meteorological (hydromet for short) disasters allowed to influence overall planning and figure more prominently in South Asia than development programs. Despite being tasked in East Asia.1 with empowering different line ministries to The region faces high probabilities of natu- incorporate effective DRM practices, they ral hazard risk, of which earthquakes, floods, have not performed well because they lack cyclones, droughts, and landslides are among human and financial resources (Bronkhorst the five most significant (Gupta and 2012). Weak governance and ineffective Muralikrishna 2010). In its Global Assessment DRM institutions result in slow progress in Report, the United Nations International mitigating disaster risk and building resilience Strategy for Disaster Reduction (UNISDR) in South Asian cities. SUSTAINING PROSPERITY BY BUILDING DISASTER-RESILIENT CITIES 171

This chapter first provides an overview of exposed to fl oods, with about 45 percent of the risks and exposure to risks in the region’s urbanized extents in fl ood-prone areas and urban areas. It then outlines vulnerabilities of 14 percent in extremely flood-prone areas different countries based on risks and expo- (figure 6.2). Most city centers sit on river- sures, followed by a description of the impacts banks and coastal land, so most of the periph- of climate change and global warming in eral expansion has been in the hinterlands, exacerbating disaster risk in South Asia. It away from the largest waterways. Based on then classifies South Asian cities based on historical trends going back to 1970, as much size, urban risk exposure, and socioeconomic as half of all future urban expansion could be factors. Finally, four recommendations are in fl ood-prone areas, underlining the need for made for the first steps to be taken to increase anticipatory planning and mitigating urban resilience and DRM. infrastructure. Cities in Afghanistan, Nepal, Pakistan, and Urban risks and exposure northern India that lie along the Himalayan range are all at risk of earthquakes. Many Urban population growth and economic cities in northern India and Pakistan are also development have increased the exposure in at risk of heavy inland flooding. Almost all South Asia to natural hazards by concentrat- cities in Bangladesh are at some earthquake ing people and assets in risk-prone areas such risk, but the lack of earthquakes affecting as deltas, floodplains, coasts, and the cities in recent memory and the higher Himalayan belt (figure 6.1). Overlaying a frequency and impact of flooding and flood risk map3 (UNISDR 2009) and the cyclones have reduced people’s awareness of 2010 urban footprint map based on nighttime earthquake risk. Chittagong and Sylhet lights data (see chapter 2, “Rapid relative (in Bangladesh) are in the highest earthquake expansion of urban footprints and the rise of hazard zones, and although Dhaka lies in a the multicity agglomeration”) suggests that moderate zone, it has been rated among the 80 percent of major South Asian cities are top 20 most earthquake-vulnerable cities

FIGURE 6.1 Urban population growth by hazard risk, 2000 and 2050

Source: World Bank and UN 2010. Note: Add2050 = adding the 2050 forecasted population. 172 LEVERAGING URBANIZATION IN SOUTH ASIA

FIGURE 6.2 Extent of existing urbanization in risk-prone regions, 2010

Level of flood risk in existing urbanized extents

Sources: UNISDR 2009; and the 2010 urban footprints map based on nighttime lights data (see chapter 2). Note: The map covers all cities in South Asia with a population greater than 100,000 in 2000 (Brecht, Deichmann, and Wang 2013).

(World Bank 2012a). Two-thirds of the city is those risks, and their adaptive capacity to built on infill, making it severely prone to plan for and respond to systemic shocks liquefaction, and a major earthquake in (IPCC 2007; Adger 2006; Romero-Lankao, Dhaka could cause up to 1 million fatalities Qin, and Dickinson 2012). Although only (Ahmed and Ahmed 2010). 13 percent of the world’s hydromet disasters between 1975 and 2012 took place in South Urban vulnerability to natural Asia, the region accounts for 42 percent of hazards deaths and 30 percent of the total affected population worldwide. These numbers are Vulnerability is commonly defined as a signiﬁ cant given that South Asia’s share of the function of natural hazard risk, the level of world population is only 23.3 percent.4 exposure of physical assets and people to By some estimates, 64 percent of the global SUSTAINING PROSPERITY BY BUILDING DISASTER-RESILIENT CITIES 173

population exposed to ﬂ oods lives in South state of Kerala in southern India will Asia (Bronkhorst 2012). In many South Asian experience 3-sigma temperature anomalies6 cities, the combination of high natural risks, as often as seven months out of the year if the high concentration of resources and people in earth warms by 2 degrees Celsius (°C). If the risk zones, high rates of multidimensional planet warms by 4°C, these anomalies will poverty, low levels of governmental attention take place year round along the coasts of to natural hazards, and large populations of India, as well as in Bhutan, Nepal, and informal settlements translate into high vulnerability to natural disasters. South Asia’s urban population has FIGURE 6.3 Estimated urban economic loss and mortality loss rates in South Asia increased by 130 million just since the turn of the century. Fixed gross capital formation— a. Urban economic losses as a share of aggregate city GDP, 2000 investment in land improvements, industrial 80 machinery and equipment, buildings, and infrastructure—has surged. By allowing peo- 70 ple and assets to be concentrated in at-risk 60 areas, cities have increased their exposure to these natural hazard risks (Bronkhorst 2012; 50 Revi 2008). 40 The combination of risk, exposure, and vulnerability leads to high direct urban 30 economic and mortality losses from natural 20 hazards. Urban assets in the region are most

financially exposed to floods, whereas urban of city GDP (percent) as a share Losses 10 residents are most vulnerable to catastrophic 0 events such as earthquakes and cyclones (Brecht, Deichmann, and Wang 2013). In a India Nepal Pakistan 5 South Asia study of 233 South Asian cities, floods Afghanistan Bangladesh accounted for 64 percent of estimated urban Earthquake Cyclone Landslide Flood economic losses over 20 years, but only 16 percent of mortality; cyclones, affecting only 13 cities, contributed 13 percent of b. Proportion of deaths by disaster, 1998–2007 economic losses and 50 percent of mortality; 100 and earthquakes 23 percent and 32 percent, 90 respectively. Because of their size and level of 80 development, cities in India—and to a lesser 70 extent Pakistan—account for most of the losses in absolute terms, but as a share of 60 metropolitan gross domestic product (GDP) 50 and population, cities in Bangladesh suffer 40 much higher rates of loss (figure 6.3). 30 20 Impacts of climate change and 10 Proportion of deaths by disaster (percent) disaster by Proportion of deaths global warming 0

Natural disasters are likely to be made worse India Nepal Pakistan by global warming and climate change. South Asia Afghanistan Bangladesh According to a 2013 model by the Potsdam Earthquake Cyclone Landslide Flood Institute for Climate Impact Research and Climate Analytics, cities in Sri Lanka and the Source: Derived from Brecht, Deichmann, and Wang 2013. For more details see Shi 2013. 174 LEVERAGING URBANIZATION IN SOUTH ASIA

northeastern Afghanistan and Pakistan. They India except the northwest, as well as in all of will take place six months or more out of a Bangladesh, Bhutan, Nepal, and Sri Lanka, given year throughout Afghanistan, including regions that are also projected to Bangladesh, Pakistan, and most of India, have a decline in overall rainfall (Gupta and especially along the coast. These heat Muralikrishna 2010). Western and central waves, on top of urban heat island effects, India are expected to experience significantly will directly undermine urban respiratory more extreme floods, like those that struck health and make children and the elderly Mumbai in 2005 and Gujarat in 2005 and more vulnerable to heat stroke. Already, 2006. With greater glacial melt, flooding is many cities in South Asia are forced to re- expected to increase in the border region allocate long-term development budgets to between India, Nepal, and Bhutan, as well as rebuilding after disaster (Bronkhorst 2012). along the Pakistan and India border region By one estimate, climate impacts could cost (Revi 2008). the region 9–13 percent of GDP a year by For nine cities around the Bay of Bengal, 2100 (Gupta and Muralikrishna 2010). including Chennai, what is now a 100-year Global warming also influences precipita- storm event may occur as often as every two tion patterns, resulting in more chronic to five years by the end of the century. There droughts, especially in currently arid and is a high level of concurrence among models semi-arid regions. A higher incidence of on these projections. Projections for parts of drought is projected for Afghanistan, Gujarat northwest India and most of Pakistan and and central India, Maldives, Pakistan, and Sri Afghanistan, though uncertain, suggest that Lanka (Ramesh and Yadava 2005; Practical the frequency of severe weather may decline Action 2009). In addition, reduced glacial and that the region will instead face greater melt leads to lower year-round water avail- challenges of drought and water scarcity. ability in Afghanistan, Bhutan, Nepal, and Another impact of climate change is Pakistan. sea-level rise. The average sea level is projected Droughts affect not only drinking-water to increase by 30–115 centimeters by 2100, availability in cities but also rural-urban placing 6 million to 40 million people at risk in migration patterns. Research on urbaniza- South Asia (PICIRCA 2013; McGranahan, tion in Sub-Saharan Africa has shown that Balk, and Anderson 2007). One model predicts lower farm income due to water shortages the additional exposure of 136 of the world’s encourages migration to nearby cities largest cities to 100-year floods in 2050 due to (Henderson, Storeygard, and Deichmann sea-level rise and land subsidence (Hallegatte 2014). Growth planning for cities through- and others 2013). The model assumes certain out the region, therefore, must account for rates of population and economic growth, and the additional “spiky” population growth that storm frequencies and intensities remain caused by climate-driven rural population the same. By some estimates, Mumbai and displacement. Kolkata are the 5th and 14th cities in the list of However, precipitation is projected to 136 largest cities exposed to 100-year floods in increase significantly in eastern South Asia 2050 due to sea-level rise and land subsidence. and moderately across India (Revi 2008; Relative to city GDP, Mumbai (India), Khulna PICIRCA 2013). Warming waters in the Bay (Bangladesh), Kochi (India), and Surat (India) of Bengal may also lead to increased frequency are the 7th, 8th, 14th, and 17th most at risk and intensity of cyclones in Bangladesh and (Hallegatte and others 2013). on the eastern coast of India. In these areas, With 20 centimeters of sea-level rise by intense rainfall and severe floods are likely to 2050 (an optimistically low projection), and occur with greater frequency, even without considering adaptations that will restrict flood- accounting for sea-level rise and storm surges ing to current probabilities (for example, by (Hirabayashi and others 2013). Increasingly raising dikes by the amount of sea-level rise), frequent extreme rainfall is likely in most of Mumbai, Kolkata, Chennai, and Surat will SUSTAINING PROSPERITY BY BUILDING DISASTER-RESILIENT CITIES 175

become the top 2nd, 3rd, 13th, and 14th most 233 cities studied by Brecht, Deichmann, and exposed cities to flooding in the world. To Wang (2013) are in danger of flooding. retain current levels of average annual eco- Earthquake threatens 124, landslides 68, and nomic loss probability given 20 centimeters of cyclones 13. Two major cities in Bangladesh sea-level rise and land subsidence, Chennai, (Chittagong and Cox’s Bazar) are at risk of all Chittagong, Kochi, Mumbai, Surat, and four major hazards. All the region’s megaci- Visakhapatnam would need to raise dikes by ties (Dhaka, Kolkata, Delhi, Mumbai,9 and 20 centimeters, while Dhaka, Karachi, Khulna, Karachi) are highly at risk of earthquakes and and Kolkata would need to raise them by floods, with Dhaka, Kolkata, and Karachi more than 60 centimeters (Hallegatte and also at risk of cyclones and storm surge others 2013). (Gupta and Muralikrishna 2010). Disaster typologies are mapped in figure 6.4. Classifi cation of cities based on risk and vulnerability profi les Classifi cation based on earthquake risk To begin identifying packages of interventions About 12 percent of the 233 cities in the study for assistance to cities, this section classifi es sample (29 cities) are mainly at risk of earth- 241 of the largest cities in South Asia, based quakes and could have moderate to high on available data for risk, exposure, and vul- economic and mortality baseline losses due to nerability.7 The cities are arranged based on earthquakes, but low baseline losses due to their existing and projected risk due to earth- hydromet events. These cities are located in quakes and hydromet hazards, city size, and the Himalayan region.10 socioeconomic vulnerability.8 The regional Most of the cities in this category are also perspective provides a best-guess estimate of projected to experience moderate to severe ecological and socioeconomic circumstances risks of heat anomaly and flood. Five of the that may have direct and indirect effects on cities are in danger of increasing threat, cities even if the data are uncertain for specifi c given that the current high baseline earth- metropolitan areas. This classification may quake risk would be coupled with severe offer one potential strategy for prioritizing hydromet projections (Kathmandu, investment and climate action on a regional Jalandhar, and Dehradun) or moderate level, but before fi rm actions are taken, local hydromet projections (Meerut, Faisalabad) and subregional models should be developed in the medium or long term. The remaining to determine which approaches are appropri- cities in this group are exposed to moderate ate and how much disaster risk reduction and baseline earthquake risks, with some small adaptation investment is right for cities. to medium cities projected to experience Most cities in the region are exposed to moderate to severe hydromet impacts. For more than one hazard (table 6.1). Flooding more details and to see taxonomic trees of seems to be the most common: 187 out of these cities, see Shi (2013).

TABLE 6.1 Number of cities in South Asia affected by each of the four major hazards

Afghanistan Bangladesh India (144 total Nepal (5 total Pakistan Total (6 total cities) (28 total cities) cities) cities) (50 total cities) (233 cities) (%) (%) (%) (%) (%) (%) Earthquakes 6 (100) 24 (85) 40 (27) 5 (100) 49 (98) 124 (53) Flooding 4 (67) 26 (92) 120 (83) 3 (60) 34 (68) 187 (80) Cyclones 0 (0) 5 (17) 7 (4) 0 (0) 1 (2) 13 (5) Landslides 2 (30) 2 (7) 52 (36) 3 (60) 9 (18) 68 (29)

Source: Brecht, Deichmann, and Wang 2013. Note: Sri Lanka was not covered in the data set used by Brecht, Deichmann, and Wang 2013. 176 LEVERAGING URBANIZATION IN SOUTH ASIA

FIGURE 6.4 Map of disaster typologies

Sources: Based on Center for International Earth Science Information Network (http://www.ciesin.org); UN World Urbanization Prospects: 2011 Revision (http://esa.un.org/unpd/wup); Brecht and others 2012; Brecht, Deichmann, and Wang 2013; Hallegatte and others 2013; Hirabayashi and others 2013; OPHI 2013; PICIRCA 2013. For more information on data and methods, see Shi 2013.

Classifi cation based on face high baseline risks and severe projected hydrometeorological risk climate impacts by 2100 (for some, as early as the 2040s) and that are highly impoverished Baseline hydromet conditions threaten 87 cit- and socioeconomically vulnerable (Cox’s ies (36 percent of the sample), causing poten- Bazar is an example). Another 10 large and 5 tial moderate to high economic and mortality small to medium cities face similar risks and losses. These cities face low earthquake risk; have low to moderate vulnerability levels. most are in coastal India, Bangladesh, and Chennai and Kochi are among these. It is of Sri Lanka, as well as on the ﬂ oodplains of the particular concern that among cities facing Ganges, Mahanadi, Narmada, and other riv- high hydromet risks today, none are projected ers in central India. Of particular concern are to experience only moderate climate impacts. the 10 large and 9 small to medium cities that SUSTAINING PROSPERITY BY BUILDING DISASTER-RESILIENT CITIES 177

Of the other 53 cities facing moderate Policy implications baseline economic and mortality risk, only 5 This classification scheme suggests that are projected to experience more moderate policy makers should focus their DRM future climate impacts—the rest will experi- efforts on highly vulnerable large cities with ence severe impacts. These 48 cities include high present and future risk. Prioritization of the large cities of Bangalore, Bhilai, Indore, typologies beyond this first grouping and Mumbai and small to medium cities such becomes less obvious and depends on politi- as Birganj, Pabna, and Raipur. In each of cal interests, equity considerations, and these cities, more than half the population funding availability and fungibility. National lives in multidimensional poverty. governments and international banks and donor agencies often prioritize larger cities, Classifi cation of cities based on where investments can affect more people hydromet and earthquake risk and assets, but political and equity considerations also require distributed investments Some 55 cities (23 percent of the sample) across cities of all sizes. The vulnerability experience moderate to high economic and measure, though helpful, is not necessarily mortality losses from both earthquakes and indicative of priority rank: a large city with baseline hydromet events. All of these cities, low to moderate levels of poverty may still except for Kabul, are in India, Pakistan, and have larger urban poor populations than a Bangladesh. Most concerning in this category highly vulnerable small city, if absolute size are the eight large cities with high baseline is the target consideration. The level of risks, severe climate projections, and large urgency and degree with which future vulnerable populations, including Karachi projections affect present-day investments (Pakistan); Agra, Faizabad, Guwahati, and will depend on the context and projections Kolkata (India); and Chittagong, Dhaka, and for each city. An example of a metropolitan Rajshahi (Bangladesh). Another seven small investment project in resilience is discussed to medium cities with similar risk factors are in box 6.1. Quetta and Sukkur (Pakistan); Gorakhpur (India); and Brahmanbaria, Mymensingh, Narsingdi, and Sylhet (Bangladesh). Twelve Key recommendations cities in the more arid northwestern areas of South Asia experiencing high current losses The main responses of South Asian countries from hydromet events are projected to experi- to disaster have so far been reactive rather ence more moderate climate impacts. Another than proactive. Some, however, have begun 22 small to medium cities in the arid north- planning to increase their resilience and to west with moderate earthquake and baseline reduce negative impacts of disasters, includ- hydromet risks are projected to experience ing integrating DRM in their national plan- moderate to severe impacts under climate ning frameworks. However, DRM does not change.11 work as a standalone set of policies and Finally, 69 cities (29 percent of the sample) actions. Rather, it requires parallel policies on now face low earthquake and hydromet risks, spatial planning and the creation of function- but are projected to experience moderate to ing and formal land and housing markets, severe climate impacts from temperature which in turn boost urban resilience by reduc- change and flood risk. Only three large cities ing the prevalence of unplanned development (Aurangabad, Hubli, and Ranchi) fall into (see chapters 4 and 5). Because each of these this category, and most are small to medium policies by itself is insufﬁ cient for the develop- cities. Fifteen cities with majority vulnerable ment of resilient urban centers, the policy populations are at severe future risk, while recommendations presented here complement another 15 cities in this group are vulnerable, earlier ones to directly address urban resil- facing moderate risk. ience planning. 178 LEVERAGING URBANIZATION IN SOUTH ASIA

BOX 6.1 Investing in urban resilience: Colombo metropolitan area

During a 15-hour period in November 2010, in fl ood-prone areas and subject to the environ- nearly 500 millimeters of rain fell on Colombo, mental and health risks of fl oods. the capital city of Sri Lanka, causing unprec- The Metro Colombo Urban Development edented flooding across the city. Many houses Project was launched in 2012 to solve these and buildings were destroyed, and the parliament problems. The project supports the national building was in up to 1.2 meters of water. The government’s aim to reduce fl ooding in the fl oods caused high economic losses given that the catchment of the Colombo Water Basin and Colombo metropolitan area accounts for about strengthen the capacity of local authorities in 50 percent of Sri Lanka’s GDP. the metropolitan area to rehabilitate, maintain, The fl oods were particularly destructive and improve local infrastructure and services because of poor design and maintenance of through selected demonstration investments. drainage systems, illegal encroachments on The project has three main components: fl ood fl ood retention areas, and industrial pollution. and drainage management, urban development The metropolitan area is in a low-lying fl ood and infrastructure rehabilitation, and capac- plain and one of the urban areas in Sri Lanka ity building for metropolitan Colombo local most exposed to fl oods. authorities. The occurrence and damage of fl oods in The reduction in fl ood damage resulting the area have steadily increased because of a from the project is projected to directly benefi t combination of climate and nonclimate factors, the lives of about 232,000 people, primarily including rapidly changing climate patterns by avoiding the losses associated with damage resulting in frequent and more intense thunder- to residential and commercial property. A fur- storms. Rainfall frequency has almost doubled ther 2.5 million people are expected to benefi t in Colombo during the past 30 years, while the indirectly from avoiding losses caused by the area’s population has increased from 1.7 million more general disruption of economic activ- in 1981 to 2.5 million in 2010. Investment in ity resulting from fl ooding. Over the project the drainage system, however, has been sluggish, period, the amount of fl ood damage avoided is and storage capacity in the basin has declined put at $10 million in the fi rst year of the proj- greatly since 2000 as a result of uncontrolled ect, increasing to $91 million in 2051 when the encroachment on landfi lls and the fl ood plain impacts of climate change on the probability of by illegal settlements. In the city of Colombo fl ooding are taken into account. alone, 68,000 housing units are estimated to be in underserved pockets; most of this housing is Source: World Bank 2012b.

Even considering that planning for disaster • Develop a risk ﬁ nancing scheme to pro- risk is context based, a set of essential actions vide immediate liquidity in the aftermath should be considered by all countries: of disasters and to build financial resilience. • Identify risk by using an urban risk assessment framework • Mitigate risk by planning critical and Identifying risk multipurpose infrastructure that is safe and resilient As the impacts of climate change have • Build strong institutions and collect, become clearer in the past two decades, cities share, and distribute disaster data have begun to develop methods to assess its SUSTAINING PROSPERITY BY BUILDING DISASTER-RESILIENT CITIES 179

risk and potential harm. The first step in of Indonesia, using a free, open-source plat- developing a resilience strategy is to identify form (Open Street Map; http://www the risks at the national, subnational, and city . openstreetmap.org) engaged the community in levels. The classification scheme presented mapping 160,000 buildings. Exposure earlier provides a broad overview of only mapping has become easier because many gov- regional hazards. ernments have increased the accessibility of Risk assessments define the potential risks data on population, transportation, settle- and the characteristics (such as frequency ments, and so on. But estimating exposure still and severity) of potential hazards, and requires high-quality national statistical capac- identify vulnerabilities of communities ity, which may not always be present (see and potential exposure to given hazards. chapter 1, box 1.1). The increasing use of Development of a risk assessment framework advanced satellite technology and new also guides governments in their prioritiza- approaches to data collection have also facili- tion of risk management measures, consider- tated risk modeling at higher resolution. ing the probability and impact of potential Finally, vulnerability estimates include events, cost-effectiveness of preventive mea- physical and socioeconomic categories, sures, and resource availability (Dickson and mainly of potential damage or loss. However, others 2012). only rarely do risk assessment frameworks Risk assessment can be developed at vari- integrate cities’ socioeconomic characteristics ous scales, from multicountry or regional to because such data are patchy, adding to the national, urban, and even community levels. difficulty of quantifying potential losses. For this report, risk assessment experiences at Historical loss data usually include only the city and community levels were reviewed. direct, tangible losses caused by a disaster, Urban risk assessments usually aim to identify like damage to infrastructure and housing, critical infrastructure and develop early warn- but not intangible and indirect losses such as ing systems. Depending on scope, they may interruption to business and expenses linked be costly, since they require asset information to temporarily housing disaster victims. across wide geographic areas. On a smaller A global review of risk assessment frame- scale, community-based disaster risk assess- works by the Global Facility for Disaster Risk ments try to engage communities to promote Reduction recommends that the following local action and communicate the extent of elements be part of any risk assessment: risk and potential damage (GFDRR 2014). Because risk is a function of hazard, expo- • Clearly define the purpose of the risk sure, and vulnerability, a sound risk assess- assessment before analysis starts. ment framework must consider all three Successful assessments begin with clear elements. Vulnerability is the extent to which questions to answer and well-defi ned end a city is predisposed to experience the adverse users, that is, local government offi cials impacts of climate change (IPCC 2007).12 and the local community, to inform. A critical component of any risk assessment is Otherwise, they become mere engineer- the compilation of hazard data, which is nec- ing reports. essary for determining the possible scope and • Promote and enable ownership of the risk magnitude of natural hazards. However, lack assessment process and efforts to miti- of historical hydromet data has become an gate risk. For risk assessments to be obstacle to predicting risks in less developed accepted and trusted by the community, a countries, including South Asian countries, strong partnership between public offi - and to the potential modeling of such risks. cials and the community should be On the exposure side, however, the emer- formed from the very beginning of the gence of volunteer geospatial initiatives has cre- process. ated momentum for engaging communities in • Cultivate and promote open data prac- collecting data. For example, the government tices. Open data platforms have been 180 LEVERAGING URBANIZATION IN SOUTH ASIA

fundamental in gathering exposure data construction to decrease impacts. The latter in past disasters. Voluntarily gathered comprise policies and laws and the use of geospatial information and remote sens- knowledge, practices, and agreements to pre- ing products offer new opportunities to vent future vulnerability. Speciﬁ c examples of collect and update fundamental data. nonstructural measures include building • Make better communication of risk codes, land-use planning, public awareness, information an urgent priority. In doing and information (Bronkhorst 2012). so, make sure that the risk information Identifying risks and developing risk communicated is customized to the level assessment frameworks do, of course, take a of the audience’s technical knowledge. lot of time, and cities have to continue build- • Foster multidisciplinary, multi- ing infrastructure. To avoid putting develop- institutional, and multisectoral collabora- ment on hold, cities need to build new tion at all levels, from the international to infrastructure—transport infrastructure and the community level. Effective risk assess- water, sanitation, and power facilities—with ments bring together decision makers and optimum physical resilience. Cities often technical experts to agree on the purpose overlook these measures because of the extra and scope of the risk assessment. marginal costs of building resilient structures • Consider the broader risk context. As (Bronkhorst 2012). the risk classiﬁ cation of South Asian cit- The growth of South Asian cities has ies shows, successful risk assessments increased the number of people and assets in cover multiple hazards—an incomplete risk-prone areas. Relocating millions of peo- risk assessment will likely produce ple away from their homes and jobs is not maladaptation. realistic. Instead, with the help of urban plan- • Be aware of evolving risk. Risk assess- ners, engineers, and academics, cities can ments can suggest actions to be taken revisit urban design and ensure enforcement now to mitigate future and evolving risk, of building codes and land-use plans to mini- especially in the context of rapid urban mize or prevent further building in risk-prone development. areas and to reinforce structures so that they • Understand, quantify, and communicate are resilient to various hazards. As the con- the uncertainties and limitations of risk struction industry develops further in South information. The end users must be Asia, the region’s cities can halt substandard aware of the limitations of data and construction practices. City leaders should models. use policy tools and incentives to enforce • Ensure that risk information is credible building codes, which may impose extra costs and transparent. It is best to open data, but is an investment that may well more than models, and results for review by inde- offset the postdisaster costs that would other- pendent technical specialists to demon- wise be incurred. The problem is that South strate the credibility of risk assessments Asia lacks empowered city leaders (see (GFDRR 2014). chapter 3). Still, city leaders should, in planning to fill An example of a successful urban risk assess- the huge infrastructure gap, consider future ment is presented in box 6.2. risks and hazards, and ensure that the new infrastructure is not built in hazard-prone Mitigating risk areas and does not expose communities to additional risks. Leaders also need to identify City governments need to develop both struc- and plan for critical infrastructure, which tural and nonstructural measures to mitigate would be built with higher-than-usual risk. The former include dams, levies, and margins of safety, such as extra strength wave barriers and the retrofitting of build- (World Bank and United Nations 2010). Such ings, or any other physical adjustment or critical infrastructure must be identified now. SUSTAINING PROSPERITY BY BUILDING DISASTER-RESILIENT CITIES 181

BOX 6.2 Integrating risk assessment in development planning: Aqaba, Jordan

In 2001 the government of Jordan declared This assessment provided a broad view Aqaba a special economic zone, opening the door of losses, putting them at almost $2.4 billion to tourism and trade and boosting economic (almost 8 percent of the country’s GDP), given growth. Aqaba is a coastal city with a population that the earthquake would hit the country’s of 108,000 as of 2009 and has one of the high- only seaport and its gateway to international est population growth rates in the country. It is trade. The assessment predicted that the com- a major tourist attraction and the country’s only bination of earthquake-related disruption of seaport. port activities for three months and humanitar- However, the city is exposed to a high risk of ian activities could cost $420 million directly, intense earthquakes. The status of the city as a in addition to $300 million in losses due to special economic zone increased its exposure decreased tourism. to seismic risk. To assess and mitigate potential As a result of this risk assessment, a new losses from seismic hazards, in 2009 the Aqaba DRM master plan was prepared for Aqaba, and Special Economic Zone Authority (ASEZA) col- a DRM unit and multistakeholder coordination laborated with the United Nations Development committees were established within ASEZA to Programme and the Swiss Agency for Development integrate risk reduction into development plan- and Cooperation to launch a project to integrate ning and to coordinate stakeholders. The Aqaba seismic risk reduction considerations into Aqaba’s Development Company is using the fi ndings of economic development planning. the assessment as part of the decision-making To conduct the risk assessment and demon- process on construction projects and land allo- strate potential impact and losses, an impact cation for new businesses, to mitigate potential scenario from a maximum magnitude earth- economic losses due to earthquakes. quake of 7.5 on the Aqaba fault section was pro- Several lessons were learned when the duced. The risk assessment included expected assessment was produced. First, the focus of any losses and impacts on people, buildings, and the risk assessment should be on decision making. economy. These fi ndings were then integrated Second, local experts should be consulted and into development planning for the worst-case involved to ensure ownership and sustainabil- scenario. As an example, the assessment pointed ity (as were local institutions and universities out that the current hospital capacity of 206 in Aqaba). Third, fi ndings must be communi- beds among three hospitals in Aqaba would be cated to stakeholders. Last, this communication insuffi cient to handle potential casualties esti- should result in extensive engagement with mated to be in excess of 1,900. The assessment stakeholders through dissemination activities, also took into account critical infrastructure and workshops, and meetings, so that the com- transportation systems, predicting that main munity is aware of the risks and owns the risk and secondary roads would likely be disrupted assessment. for more than 40 days and wastewater systems disrupted for almost a month. Source: GFDRR 2014.

In Bangladesh, for example, the government multipurpose infrastructure is Kuala uses safe and structurally sound schools as Lumpur’s Stormwater Management and shelters during cyclones. Another suggestion Road Tunnel, which is a 9.7 kilometer tunnel, is to build multipurpose infrastructure that built on three levels. The upper two levels are can serve the community in normal times as for road traffic and the lowest level is for well as after a disaster. An example of such flood drainage, capable of transferring high 182 LEVERAGING URBANIZATION IN SOUTH ASIA

volumes of flood water to a storage reservoir scenario development and impact assess- and a bypass tunnel (World Bank and United ments. One role for local governments is to Nations 2010). update data after a disaster according to international standards. These data can then Building institutions and collecting data be banked and made available to the public through open platforms. Dissemination has All countries in the region have developed two benefits: (1) it allows residents to make national plans for DRM or have included resil- informed decisions, and (2) it allows markets ience in their national development plans— to function more efficiently, with real estate usually after major disasters such as the 2001 prices reflecting risk factors. However, these Gujarat earthquake or the 2004 Indian Ocean benefits are observed only in well-functioning tsunami. Most countries focus on disaster land and housing markets (see chapter 5). response rather than mitigation, but some Data range from primary risk maps to countries, such as Bangladesh, have developed loss scenarios and impact models, including more sophisticated plans for mitigating disas- land use and land cover, building area, build- ter risk by concentrating on local actions, ing cover, and vulnerability curves. More including disaster-preparedness education. sophisticated data enable disaster and cli- Countries with strong institutions for mate-risk modeling, which can be used by mitigating disaster risk can mainstream the international insurers and capital markets to collection and availability of data. South Asia develop premiums and assess risks to asset lacks historical loss data, but national institu- portfolios. The types of data required are tions have begun collecting data to conduct given in table 6.2.

TABLE 6.2 Characteristics of hazards and assessment data requirements

Type of disaster Measurement and characteristics Data required to assess hazard risk Flooding Intensity and frequency of fl oods Topography (digital elevation model), drainage patterns, built-up areas, land use and land cover, historical rain gauge data Cyclone and storm Maximum sustained wind and radius to Topography (digital elevation model), drainage surge maximum wind at landfall, central pressure patterns, bathymetry, land use and land cover, from water column, height of storm surge historical rain gauge data waves Earthquake Magnitude expressed on Richter scale, ground Information on soil, geology, and liquefaction shaking measured based on damage potential to develop geological, seismic, and soil maps Tsunami Wave height, inundation run-up Topography of coastal areas (digital elevation model), bathymetry, location and capacity of any fl ood-protection infrastructure Drought and water Water and food availability per capita Surface temperatures, precipitation, reservoir scarcity capacity, and actual volume stored Sea-level rise, tidal Wave height, horizontal pressure from water Topography of coastal areas (digital elevation fl ooding column model), bathymetry, tide gauge data, coastal land subsidence data Volcano eruption, lava Pyroclastic and ash fall, explosiveness of Topography, proximity of the volcano to people fl ow volcano, horizontal pressure of lava fl ow and assets Landslides, mud fl ows Failure of slopes with mass movements, Topography (digital elevation model), geological and lahars, rock and horizontal pressure of mud fl ows, vertical or data, land use and land cover rubble fall side impact of rock debris Fire High temperature and combustion Topography, wind, land use and land cover

Source: Dickson and others 2012. SUSTAINING PROSPERITY BY BUILDING DISASTER-RESILIENT CITIES 183

Regional cooperation in collecting hazard Developing risk fi nancing data is crucial, because many countries share How should governments ﬁ nance postdisas- river deltas and mountain ranges. For exam- ter response and reconstruction needs? ple, Bangladesh would benefit from coopera- Financing can be done after or before disas- tion with its neighbors in gathering data on the ter hits. After a disaster, governments usually Ganges-Brahmaputra Delta’s complex hydrol- have immediate access to various sources of ogy. Sharing data on hydromet conditions and funding that do not require advance plan- river flows in real time is vital to keeping or ning, including budget reallocations, domes- moving people out of harm’s way. Without tic and external credits, tax increases, and data on upstream water levels, Bangladesh donor assistance. But financing ahead of cannot predict floods with any accuracy. This time is, of course, a better approach. An situation has improved in recent years thanks advance financing plan should include to satellite data, which enable 10-day flooding reserves or calamity funds, budget contin- forecasts, but because the Brahmaputra is gencies, a contingent debt facility, and risk linked to the Ganges River’s flows, the accu- transfer mechanisms. Such instruments are racy and range of predictions would improve more sophisticated and require human if forecasting were done in cooperation with capacity and data. The instruments include India (World Bank and United Nations 2010). traditional insurance and reinsurance, para- Unfortunately, the two countries have yet to metric insurance, and alternative risk trans- work together on this problem. fer instruments such as catastrophe bonds. Central American countries have shared Insigniﬁ cant but recurring disasters are usu- their data on exposure to seismic fault lines ally covered by reserves or contingent credit. and hurricanes. Through the Central American High-impact, less frequent disasters can be Probabilistic Risk Assessment initiative, these insured in capital markets. countries together collect data on a set of National frameworks for disaster risk evaluation techniques and have structured a financing depend on “risk layering”—the risk communication platform to aid decision mak- is divided into low, medium, and high catego- ing. The platform is open to the public; if such ries, and each category is paired with appro- a platform were used in South Asia, it could be priate financing instruments ( figure 6.5). adjusted for use by each country (World Bank Financing needs for the low-risk layer and United Nations 2010).

FIGURE 6.5 Financial instruments to address diff erent layers of risk

High severity

International donor assistance

Insurance-linked securities Risk transfer Insurance and reinsurance

Contingent credit Risk Reserves retention Low severity

Low frequency High frequency

Source: Bronkhorst 2012. 184 LEVERAGING URBANIZATION IN SOUTH ASIA

(landslides, local floods) usually come from also sets aside about 4.5 percent of its annual contingent budgets, reserves, and annual budget for disaster response. Even low-capacity budget allocations; those for the medium-risk countries like Afghanistan have realized the layer (floods and lower-frequency, small importance of holding funds for disaster earthquakes) often come from contingent response and have developed national emer- credit; those for the high-risk layer (major gency funds (Bronkhorst 2012). disasters) have to be paired with catastrophic But the most extensive risk financing risk transfers such as parametric insurance efforts in South Asia are in Sri Lanka, the and catastrophe bonds (World Bank and first country to develop a “catastrophe United Nations 2010). draw down option”13 with assistance from A detailed risk assessment is necessary for the World Bank through a development designing a national catastrophe risk strategy; policy loan. Approved in April 2014, the the risk assessment should be followed by a loan provides a line of credit that can be hazard module for major perils. The next step drawn on partially or in full if the country in developing the catastrophe risk strategy declares a state of emergency after a natural would be to build a national georeferenced disaster. This line of credit is part of a pack- hazard exposure database that includes pub- age that includes a loan of $110 million for lic and private assets. This information is crit- a climate resilience improvement project for ical for insurers to be able to offer affordable financing short- and long-term flood and property catastrophe products. The strategy drought issues. This project will identify will help urban areas cope with disasters, but $1 billion of investments in comprehensive to reduce the financial risk of disasters even and sustainable basinwide flood and further, cities should develop frameworks for drought risk mitigation. These investments risk insurance independently and under a are expected to encompass both the mitiga- national umbrella. tion of physical structures and the improve- Some South Asian countries have developed ment of the country’s water management financing programs specifically for disasters. system. This facility will help Sri Lanka Nepal, for example, has developed a central access a rapidly available and flexible disaster relief fund as well as district, municipal, financial tool in the aftermath of a potential and village funds. Bangladesh has developed a major disaster rather than spending time natural disaster risk reduction fund, a contin- and resources trying to raise funds. gency line financed through government reve- Examples of risk financing from other nues ($15 million annually). The government regions are provided in box 6.3.

BOX 6.3 International experiences in risk fi nancing

Fund for Natural Disasters (FONDEN) in the public infrastructure and housing for the poor. It Ministry of Finance, Mexico. Mexico is exposed also conducts risk management studies and helps to many types of disasters, including seismic, vol- design risk transfer instruments. canic, and cyclonic. In 1996, the national govern- FONDEN has three components. The ment created FONDEN mainly as an instrument Revolving Fund fi nances emergency response to provide funds quickly after a natural disaster. after disasters. The FONDEN Program fi nances FONDEN’s main purpose is to provide imme- reconstruction of public infrastructure and diate financial support to federal agencies and restoration of housing for the poor. The FONDEN local governments, particularly for providing Trust Fund is managed by one of Mexico’s main relief supplies and fi nancing the reconstruction of public development banks. FONDEN receives (continues next page) SUSTAINING PROSPERITY BY BUILDING DISASTER-RESILIENT CITIES 185

BOX 6.3 International experiences in risk fi nancing (continued)

an annual allocation from the Ministry of Turkish Catastrophe Insurance Pool (TCIP). Finance and manages the ministry’s risk fi nanc- About 70 percent of Turkey’s population and ing strategy. It uses a layering strategy to transfer 75 percent of its industrial facilities are highly risk to the reinsurance markets for parametric exposed to earthquakes. The Marmara earth- coverage or the capital markets for catastrophe quake of 1999 caused 15,000 deaths and a huge bonds, after placing excess risk with the public fi nancial burden on the economy. The following insurer AGROASEMEX. year, the government (with assistance from the Caribbean Catastrophe Risk Insurance World Bank) developed a property catastrophe Facility (CCRIF). Small island states of the risk insurance mechanism, which was the fi rst Caribbean are at elevated risk of natural disas- national catastrophe insurance pool in World ters but have limited budgets and little access to Bank client countries. credit to absorb the fi nancial impacts of disas- TCIP is a legal entity that provides manda- ters. Because of their small levels of trade and tory insurance for properties built legally on reg- high transaction costs they do not have access to istered land. Benefi ting from economies of scale catastrophe insurance in international markets. through national risk pooling, it provides more CCRIF is the region’s fi rst multicountry risk affordable premiums. TCIP’s risk fi nancing pool; it was set up in 2007 to allow countries includes risk retention and reinsurance. Through to pool their individual risks into a joint reserve a World Bank contingency loan of $100 million, mechanism and provide insurance coverage at the government covers initial losses through its far lower cost to each. This facility has been well reserves and transfers excess losses to interna- received in the international reinsurance mar- tional reinsurance markets. The national gov- ket and has provided liquidity in the aftermath ernment commits to further covering losses ex- of disasters. CCRIF is not structured to cover ceeding the capacity of TCIP (enough to bear a all the costs associated with disasters, only the 1-in-350-year earthquake). estimated liquidity needs of the fi rst three to six months. Source: GFDRR 2011a, 2011b, 2011c.

Notes 6. A 3-sigma temperature anomaly has a three- standard deviation difference from the histor- 1. Data from EM-DAT: The International Disaster ical average temperature, and a probabilistic Database, Centre for Research on the Epide- return period of 720 years under baseline miology of Disasters. http://www.emdat.be. conditions (PICIRCA 2013). Recently, their 2. Areas between rivers or streams and land. frequency has been increasing, as seen in heat- 3. The UNEP/GRID-Europe fl ood map, pro- waves in 2012 (United States), 2010 (Russian duced for the Global Assessment Report Federation), and 2003 (Western Europe). (UNISDR 2009), was based on global data 7. This includes Brecht, Deichmann, and Wang’s for fl ood hazards and should not be inter- (2013) 233 cities (1,649 urban agglomera- preted as specifi cally indicative of fl ood risk tions from World Bank client countries with in a particular spot, as this is dependent on populations greater than 100,000 in 2000), building construction, fl ood mitigation infra- in addition to Thimphu (Bhutan) and seven structure, and other local issues. cities in Sri Lanka. Because of its size and 4. Based on World Bank data (http://data location, Malé, Maldives, is not covered by .worldbank.org/region/SAS). regional or global risk and climate maps; 5. South Asian cities with populations of more it is not included in the taxonomic analysis than 100,000, studied by Brecht, Deichmann, because of lack of information. For more and Wang (2013). information on data sources, see Shi (2013). 186 LEVERAGING URBANIZATION IN SOUTH ASIA

8. The classifi cation system depends on the Brecht, H., S. Dasgupta, B. Laplante, S. Murray, quality of the models at the city level for and D. Wheeler. 2012. “Sea-Level Rise present-day risks (Brecht, Deichmann, and Storm Surges: High Stakes for a Small and Wang 2013), on rough estimates of Number of Developing Countries.” Journal of future climate outlook that draw on global Environment and Development 21 (1): 120–38. and regional projections (Hirabayashi and Brecht, H., U. Deichmann, and H. G. Wang. 2013. others 2013; PICIRCA 2013, Brecht and “A Global Urban Risk Index.” Policy Research others 2012; Hallegatte and others 2013), Working Paper 6506, World Bank, Washington, and on district or provincial indexes for DC. multidimensional poverty (OPHI 2013). Bronkhorst, V. B. 2012. Disaster Risk Management Global and provincial data and projec- in South Asia: Regional Overview. Washington, tions were applied to the local conditions DC: World Bank. for this study. Local data in most cases Dickson, E., J. L. Baker, D. Hoornweg, and were unavailable. For more details, see Shi A. Tiwari. 2012. Urban Risk Assessments: (2013). Understanding Disaster and Climate Risk in 9. However, Brecht, Deichmann, and Wang Cities. Urban Development Series. Washington, (2013) calculate Mumbai as having a rela- DC: World Bank. tively low risk of earthquake. GFDRR (Global Facility for Disaster Reduction 10. The cities include Nepal (Lalitpur, and Recovery). 2011a. “Caribbean Catastrophe Kathmandu), Bhutan (Thimphu), northwest Risk Insurance Facility. Online case study.” India (the states of Uttaranchal Haryana, GFDRR, Washington, DC. http://www.gfdrr Punjab, Himachal Pradesh, Uttar Pradesh, .org/sites/gfdrr.org/files/documents/DRFI and Maharashtra), Pakistan (province of _CCRIF_Jan11.pdf. Punjab), and Afghanistan (Balkh). ———. 2011b. “FONDEN: Mexico’s National 11. Projections for northwest South Asia, how- Disaster Fund. Online case study.” GFDRR, ever, are highly uncertain, with models dis- Washington, DC. https://www.gfdrr.org/sites agreeing about whether major decreases in / gfdrr.org/files/documents/Mexico_FONDEN rainfall will occur. _fi nal_GFDRR.pdf. 12. This defi nition includes the adaptive capacity ———. 2011c. “Turkish Catastrophe Insurance of a city, which is the ability and willingness Pool. Online case study.” GFDRR, Washington, of all stakeholders and institutions to cope DC. http://www.gfdrr.org/sites/gfdrr.org/files with disaster impacts. It also covers physi- /documents/DFI_TCIP__Jan11.pdf. cal and socioeconomic elements, including ———. 2014. Understanding Risk: The Evolution population, density, quality of infrastructure, of Disaster Risk Assessment. Washington, DC: slum population, and governance (Mehrotra World Bank. http://www.drrinacp.org/sites and others 2009). /drrinacp.org/fi les/publication/_Understanding 13. This is a contingent credit line that provides _Risk-Web_Version-rev_1.7.3.pdf. immediate liquidity to member countries Gupta, S., and M. Muralikrishna. 2010. Synthesis after a natural disaster. It is part of a broad Report on SAR Countries Disaster Risks. spectrum of World Bank Group disaster risk Geneva: UNISDR. fi nancing instruments to assist countries in Hallegatte, S., C. Green, R. J. Nicholls, and planning disaster response. J. Corfee-Morlot. 2013. “Future Flood Losses in Major Coastal Cities.” Nature Climate References Change 3: 802–6. Henderson, J. V., A. Storeygard, and U. Deichmann. Adger, W. N. 2006. “Vulnerability.” Global 2014. “50 Years of Urbanization in Africa: Environmental Change 16 (3): 268–81. Examining the Role of Climate Change.” Policy Ahmed, S., and A. H. T. Ahmed. 2010. “Disaster Research Working Paper 6925, World Bank, Risks and Risk Mitigation in Dhaka and Washington, DC. Other Cities.” Presented at the South Asia Hirabayashi, Y., R. Mahendran, S. J. Koirala, launch of Global Campaign on Making Cities L. Konoshima, D. Yamazaki, S. Watanabe, Resilient, New Delhi, India, July 8. http://www H. Kim, and S. Kanae. 2013. “Global Flood . preventionweb.net/applications/hfa/lgsat/en Risk under Climate Change.” Nature Climate /image / href/429. Change 3: 816–21. SUSTAINING PROSPERITY BY BUILDING DISASTER-RESILIENT CITIES 187

IPCC (Intergovernmental Panel on Climate Change). Ramesh, R., and M. G. Yadava. 2005. “Climate 2007. Contribution of Working Group II to the and Water Resources of India.” Current Science Fourth Assessment of the Intergovernmental 89 (5): 818–24. Panel on Climate Change. New York and Revi, A. 2008. “Climate Change Risk: An Cambridge: Cambridge University Press. Adaptation and Mitigation Agenda for Indian McGranahan, G., D. Balk, and B. Anderson. Cities.” Environment and Urbanization 20 (1): 2007. “The Rising Tide: Assessing the Risks 207–29. of Climate Change and Human Settlements in Romero-Lankao, P., H. Qin, and K. Dickinson. Low Elevation Coastal Zones.” Environment 2012. “Urban Vulnerability to Temperature- and Urbanization 19 (1): 17–38. Related Hazards: A Meta-Analysis and Meta- Mehrotra, S., C. E. Natenzon, A. Omojola, Knowledge Approach.” Global Environmental R. Folorunsho, J. Gilbride, and C. Rosenzweig. 2009. Change 22 (3): 670–83. “Framework for City Climate Risk Assessment.” Shi, L. 2013. “Urban Hazards and Resilience in Presented at the Fifth Urban Research Symposium, South Asia.“ Background paper for this report, Marseille, France, June 28–30. http://siteresources World Bank, Washington, DC. .worldbank .org/INTURBANDEVELOPMENT UNISDR (United Nations International Strategy /Resources/336387-1256566800920/6505269 for Disaster Reduction). 2009. Global -1268260567624/Rosenzweig.pdf. Assessment Report on Disaster Risk Reduction. OPHI (Oxford Poverty and Human Development Geneva: UNISDR. Initiative). 2013. Global Multidimensional World Bank. 2012a. “Project Note on Bangladesh: Poverty Index. Online database. http://www Earthquake Risk Mitigation Program.” .ophi.org.uk/multidimensional-poverty-index. World Bank, Washington, DC. PICIRCA (Potsdam Institute for Climate Impact ———. 2012b. “Sri Lanka: Metro Colombo Urban Research and Climate Analytics). 2013. Development Project.” Appraisal Document. Turn Down the Heat: Climate Extremes, World Bank, Washington, DC. Regional Impacts, and the Case for Resilience. World Bank and United Nations. 2010. Natural Washington, DC: World Bank. Hazards, UnNatural Disasters: The Economics Practical Action. 2009. South Asia Disaster of Effective Prevention. Washington, DC: Report, Special Copenhagen Issue. Rugby, World Bank. U.K.: Practical Action. ECO-AUDIT Environmental Benefits Statement

The World Bank Group is committed to reducing its environmental footprint. In support of this commitment, the Publishing and Knowledge Division leverages 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. The Publishing and Knowledge Division follows 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://crinfo.worldbank.org/wbcrinfo/node/4.