Stéphane Hallegatte Jun Rentschler Julie Rozenberg LIFELINES SUSTAINABLE INFRASTRUCTURE SERIES LIFELINES The Resilient Infrastructure Opportunity Stéphane Hallegatte Jun Rentschler Julie Rozenberg © 2019 International Bank for Reconstruction and Development / The World Bank 1818 H Street NW, Washington, DC 20433 Telephone: 202-473-1000; Internet: www.worldbank.org Some rights reserved 1 2 3 4 22 21 20 19 This work is a product of the staff of The World Bank with external contributions. The findings, interpreta- tions, and conclusions expressed in this work do not necessarily reflect the views of The World Bank, its Board of Executive Directors, or the governments they represent. The World Bank does not guarantee the accuracy of the data included in this work. The boundaries, colors, denominations, and other information shown on any map in this work do not imply any judgment on the part of The World Bank concerning the legal status of any territory or the endorsement or acceptance of such boundaries. Nothing herein shall constitute or be considered to be a limitation upon or waiver of the privileges and immunities of The World Bank, all of which are specifically reserved. Rights and Permissions This work is available under the Creative Commons Attribution 3.0 IGO license (CC BY 3.0 IGO) http:// creativecommons.org/licenses/by/3.0/igo. Under the Creative Commons Attribution license, you are free to copy, distribute, transmit, and adapt this work, including for commercial purposes, under the following conditions: Attribution—Please cite the work as follows: Hallegatte, Stéphane, Jun Rentschler, and Julie Rozenberg. 2019. Lifelines: The Resilient Infrastructure Opportunity. Sustainable Infrastructure Series. Washington, DC: World Bank. doi:10.1596/978-1-4648-1430-3. License: Creative Commons Attribution CC BY 3.0 IGO Translations—If you create a translation of this work, please add the following disclaimer along with the attribution: This translation was not created by The World Bank and should not be considered an official World Bank translation. The World Bank shall not be liable for any content or error in this translation. Adaptations—If you create an adaptation of this work, please add the following disclaimer along with the attribution: This is an adaptation of an original work by The World Bank. Views and opinions expressed in the adap- tation 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 con- tained 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 compo- nent of the work, it is your responsibility to determine whether permission is needed for that re-use and to obtain permission from the copyright owner. Examples of components can include, but are not limited to, tables, figures, or images. All queries on rights and licenses should be addressed to World Bank Publications, The World Bank Group, 1818 H Street NW, Washington, DC 20433, USA; e-mail: pubrights@worldbank.org. ISBN (paper): 978-1-4648-1430-3 ISBN (electronic): 978-1-4648-1431-0 DOI: 10.1596/978-1-4648-1430-3 Cover design and graphs: Brad Amburn, Brad Amburn Creative, LLC Library of Congress Cataloging-in-Publication Data has been requested Contents Foreword. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xiii Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xv Abbreviations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xvii Overview. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 Infrastructure disruptions are a drag on people and economies. . . . . . . . . . . . . . . . . . . . . . . . . 3 More resilient infrastructure assets pay for themselves . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10 Making infrastructure more resilient requires a consistent strategy . . . . . . . . . . . . . . . . . . . . 15 Notes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21 References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21 1 Resilient Infrastructure: A Lifeline for Sustainable Development . . . . . . . . . . . . . . . 25 Objectives of this report. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27 Structure of the report. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29 References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29 I A Diagnosis: A Lack of Resilient Infrastructure Is Harming People and Firms. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31 2 Infrastructure Disruptions Are a Barrier to Thriving Firms . . . . . . . . . . . . . . . . . . . . . 33 Infrastructure services enable firms to thrive. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34 Infrastructure disruptions have direct and real costs for firms. . . . . . . . . . . . . . . . . . . . . . . . . 35 Firms employ costly measures to cope with unreliability. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40 Unreliable infrastructure leads to lower productivity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43 Notes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44 References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45 3 Infrastructure Disruptions Affect the Health and Well-Being of Households. . . . . . 49 Infrastructure provides households with essential services . . . . . . . . . . . . . . . . . . . . . . . . . . . 49 Power outages directly reduce the well-being of households. . . . . . . . . . . . . . . . . . . . . . . . . . 51 People’s health and well-being suffer when the water supply is unreliable . . . . . . . . . . . . . . 52 Transport disruptions lead to lost time, income, and access to services. . . . . . . . . . . . . . . . . . 53 Notes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55 References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55 v vi CONTENTS 4 Natural Shocks Are a Leading Cause of Infrastructure Disruptions and Damages. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57 The power sector is highly vulnerable to natural hazards . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58 Water systems are particularly vulnerable to climate change and can contribute to managing floods and droughts. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66 Natural hazards frequently disrupt and extensively damage transport infrastructure . . . . . . 70 When natural shocks disrupt telecommunications systems, whole countries can go offline . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75 Infrastructure sometimes creates or increases natural risks. . . . . . . . . . . . . . . . . . . . . . . . . . . 78 Notes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80 References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80 5 From Micro to Macro: Local Disruptions Translate into Macroeconomic Impacts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85 A survey confirms the cost of natural hazards for firms through infrastructure disruptions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87 Consequences spread through domestic and international supply chains . . . . . . . . . . . . . . . 88 Supply chain simulations enable better measurement of the macroeconomic impacts of disasters. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90 Notes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92 References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92 II A Matter of Design: Resilient Infrastructure Is Cost-Effective . . . . . . . . 95 6 More Resilient Infrastructure Assets Are Cost-Effective . . . . . . . . . . . . . . . . . . . . . . . 97 The additional up-front cost of more resilient assets depends on the asset and the hazard. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97 The additional up-front cost of more resilient assets could be offset by lower maintenance and repair costs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99 Improving maintenance and operations is an option for boosting resilience and reducing costs. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100 The cost of increasing resilience depends on the ability to spatially target strengthening. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101 Summing up . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104 References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107 7 From Resilient Assets to Resilient Infrastructure Services. . . . . . . . . . . . . . . . . . . . . 109 Using criticality analyses to prioritize interventions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109 Diversifying assets to increase network resilience . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115 Decentralizing and using new technologies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 116 Working across systems to capture synergies. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 116 Protecting infrastructure systems with dikes in dense areas. . . . . . . . . . . . . . . . . . . . . . . . . . 118 Combining infrastructure with nature-based solutions to reduce investment needs . . . . . . 118 Failing gracefully and recovering quickly. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 120 Notes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 124 References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 124 CONTENTS vii 8 From Resilient Infrastructure Services to Resilient Users. . . . . . . . . . . . . . . . . . . . . . 127 Reducing demand for infrastructure services by improving efficiency often builds resilience . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 127 Criticality depends on the end user: Some assets are critical for food security, others for competitiveness . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 129 End users need to prepare for infrastructure disruptions and design more resilient supply chains. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 129 Infrastructure affects the exposure of users to natural hazards. . . . . . . . . . . . . . . . . . . . . . . 134 Notes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 137 References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 137 A Way Forward: Five Recommendations for More Resilient III  Infrastructure. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 141 9 The Foundation for Resilient Infrastructure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 143 The obstacle: Many infrastructure systems are poorly designed, operated, or maintained. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 143 Recommendation 1: Get the basics right . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 148 Notes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 151 References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 151 10  Build Institutions for Resilience . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 153 The obstacle: Multiple political economy challenges and coordination failures impede public action on resilience . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 153 Recommendation 2: Build institutions for resilience. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 155 Notes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 160 References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 160 11  Create Regulations and Incentives for Resilience. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 163 The obstacle: Infrastructure providers often lack the incentives to avoid disruptions and to strengthen the resilience of users . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 163 Recommendation 3: Include resilience in regulations and incentives . . . . . . . . . . . . . . . . . 165 Note. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 171 References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 171 12 Improve Decision Making . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 173 The obstacle: Public and private actors often lack data, models, and capacity . . . . . . . . . . . 173 Recommendation 4: Improve decision making . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 174 Notes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 180 References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 180 13 Provide Financing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 183 The obstacle: The infrastructure sector faces affordability and financing constraints . . . . . . 183 Recommendation 5: Ensure financing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 183 References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 191 Appendix A Engineering Options to Increase the Resilience of Infrastructure Assets. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 193 viii CONTENTS Boxes 1.1 Resilience is central to achieving many international objectives. . . . . . . . . . . . . . . . . . . . 27 4.1 Exposure analysis of infrastructure assets is based on various hazard data sets. . . . . . . . 64 4.2 In hydropower, climate change adaptation is impaired by uncertainties . . . . . . . . . . . . . 68 5.1 When natural shocks affect firms, people suffer. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86 6.1 Infrastructure unit costs vary from country to country . . . . . . . . . . . . . . . . . . . . . . . . . . 98 6.2 Large investments in infrastructure will be necessary to close the service gap. . . . . . . . 102 7.1 Network topology and resilience . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110 7.2 Contingency planning for power utilities. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121 8.1 Building norms, urban forms, and behavioral changes can reduce energy demand during heat waves and prevent secondary impacts on power systems. . . . . . . 128 8.2 An energy management system to bridge power outages caused by disasters: The factory grid (F-grid) project in Ohira Industrial Park in Japan. . . . . . . . . . . . . . . . . 132 9.1 Data on infrastructure spending are scarce and limited . . . . . . . . . . . . . . . . . . . . . . . . . 145 10.1 A new hazard: Cyberdisasters and cyberattacks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 154 10.2 The structure of tariffs and targeted subsidies can help to ensure that the resilience of infrastructure services is not improved at the expense of access: The case of public transit . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 160 11.1 With climate change, when and where do standards need to be revised? . . . . . . . . . . 167 11.2 Public-private partnerships and their force majeure clauses. . . . . . . . . . . . . . . . . . . . . . 170 12.1 New technologies make data collection and processing easier . . . . . . . . . . . . . . . . . . . . 176 12.2 Preserving wetlands in Colombo minimizes the risk of regret . . . . . . . . . . . . . . . . . . . . 179 13.1 Many indicators have been developed to measure the sustainability of infrastructure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 189 Figures O.1 Poorer countries are hit hardest by inadequate infrastructure. . . . . . . . . . . . . . . . . . . . . . 3 O.2 Reliable access to electricity has more favorable effects on income and social outcomes than access alone in Bangladesh, India, and Pakistan. . . . . . . . . . . . . . . . . . . . . 6 O.3 Natural shocks explain a significant fraction of power outages. . . . . . . . . . . . . . . . . . . . . . 7 O.4 The vulnerability of the power network to wind is much higher in Bangladesh than in the United States. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8 O.5 Floods in Kampala severely restrict people’s access to health care facilities. . . . . . . . . . . . 8 O.6 Tanzanian firms report large losses from infrastructure disruptions. . . . . . . . . . . . . . . . . . 9 O.7 The resilience of infrastructure should be considered at several overlapping and complementary levels. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10 O.8 The incremental cost of increasing the resilience of future infrastructure investments depends on the spending scenario but remains limited in all cases . . . . . . . . . . . . . . . . . 11 O.9 Belgium’s and Morocco’s transport systems can absorb much larger road disruptions than Madagascar’s. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12 O.10 Spending more improves the reliability of the transport system, but only if governance improves as well. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15 O.11 Quality infrastructure requires providing for multiple funding needs . . . . . . . . . . . . . . . 16 O.12 Creating the right incentives for infrastructure service providers requires a consistent set of regulations and financial incentives . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18 1.1 Poorer countries experience more infrastructure disruptions. . . . . . . . . . . . . . . . . . . . . . 26 PI.1 A framework for analyzing how natural shocks affect people and firms through their impact on infrastructure systems. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31 CONTENTS ix 2.1 For all critical infrastructure sectors, poorer countries experience the highest utilization rate losses due to disruptions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37 2.2 In the most affected countries, utilization rate losses are a significant share of GDP . . . . 37 2.3 More frequent power outages tend to result in larger sales losses . . . . . . . . . . . . . . . . . . 38 2.4 Size of sales losses depends on more than the length of outages . . . . . . . . . . . . . . . . . . . 39 2.5 Generator ownership is more common for large firms and in countries with many power outages . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41 2.6 Increased power outages result in lower firm productivity in African countries. . . . . . . 43 3.1 Power outages hurt the well-being of households. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .51 3.2 There is large variation in people’s willingness to pay to avoid one hour without power . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52 3.3 Intermittent water supply poses major health risks in regions around the world . . . . . . 52 3.4 Water disruptions are linked with higher diarrheal risk. . . . . . . . . . . . . . . . . . . . . . . . . . 53 3.5 Transport disruptions can become life-and-death issues. . . . . . . . . . . . . . . . . . . . . . . . . . 54 4.1 Classification of causes of infrastructure disruptions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58 4.2 The share of power outages caused by natural shocks varies significantly across countries. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59 4.3 Power outages from natural shocks last much longer than those from other causes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60 4.4 Natural shocks only explain a fraction of power outages in Bangladesh . . . . . . . . . . . . . 61 4.5 The vulnerability of the power network to wind is much higher in Bangladesh than in the United States. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61 4.6 Storm-induced power outages are closely associated with the April–May nor’westers in Bangladesh. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62 4.7 Economies with the highest exposed generation capacity to multiple hazards . . . . . . . . 63 B4.2.1 Large changes in Africa’s hydropower revenues can be expected from climate change from 2015 to 2050. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68 4.8 Global exposure of transport infrastructure to multiple natural hazards. . . . . . . . . . . . . 71 4.9 Transport infrastructure damage first increases with income growth and then decreases. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73 4.10 Low- and middle-income countries bear the highest damage costs relative to their GDP . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73 4.11 Urban flooding affects a significant share of the road networks in Bamako, Dar es Salaam, Kampala, and Kigali. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75 5.1 Tanzanian firms report large losses from infrastructure disruptions. . . . . . . . . . . . . . . . . 87 5.2 Floods in Kampala cause transport disruptions and congestion. . . . . . . . . . . . . . . . . . . . 88 5.3 Supply chain disruptions are the main reason for delivery delays. . . . . . . . . . . . . . . . . . 89 5.4 Long-duration floods trigger disruptions in Tanzania, with cascading impacts on supply chains and households . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91 PII.1 The resilience of infrastructure needs to be considered at several overlapping and complementary levels. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95 6.1 Clearing vegetation around transmission and subtransmission electricity networks requires an easement. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100 6.2 The incremental cost of increasing the resilience of future infrastructure investments is significantly reduced if asset exposure is known. . . . . . . . . . . . . . . . . . . 105 6.3 Increasing the resilience of future infrastructure investments is cost-efficient— even more so with climate change. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106 6.4 The cost of inaction increases rapidly—even more so with climate change. . . . . . . . . . 106 x CONTENTS 7.1 Belgium’s and Morocco’s transport systems can absorb much larger road disruptions than Madagascar’s. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111 7.2 Increased redundancy can have high net benefits, if well targeted. . . . . . . . . . . . . . . . 112 7.3 Network topology can improve grid resilience . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114 7.4 Drought contingency plans in Spain use diverse water sources and are informed by historical drought threshold values . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123 B8.1.1 Behavioral policies are the most efficient way to reduce energy consumption during heat waves . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 128 8.1 Firms have a wide range of coping measures that they can use to mitigate the adverse effects of infrastructure disruptions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 131 9.1 Infrastructure quality correlates strongly with governance standards . . . . . . . . . . . . . . 144 9.2 Spending more improves the reliability of the transport system, especially if governance also improves . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 146 9.3 Potential savings on road spending from governance reforms. . . . . . . . . . . . . . . . . . . . 147 9.4 The full cost of infrastructure includes multiple cost components . . . . . . . . . . . . . . . . . 150 10.1 U.K. national risk matrix . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 156 11.1 Creating the right resilience incentives for infrastructure service providers requires a consistent set of regulations and financial incentives . . . . . . . . . . . . . . . . . . 165 B12.2.1 Preserving a large share of Colombo’s wetlands minimizes the potential for regret in 2030. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 179 12.1 Many low- and middle-income countries need to increase their enrollment in technical tertiary education. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 180 13.1 Countries need a layered risk financing strategy. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 186 Maps O.1 Africa and South Asia bear the highest losses from unreliable infrastructure . . . . . . . . . . 5 O.2 Investment priorities for Tanzania’s transport network will depend on its supply chains . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14 2.1 Firms in low- and middle-income countries are incurring high utilization rate losses due to infrastructure disruptions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36 2.2 Power outages are causing large sales losses in low- and middle-income countries, especially in Africa . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38 2.3 Additional costs of backup electricity generation are substantial in low- and middle-income countries. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42 4.1 Global exposure of power generation to multiple hazards. . . . . . . . . . . . . . . . . . . . . . . . 62 4.2 Some low- and middle-income countries face high annual damage and generation losses. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65 4.3 Dams and reservoirs face a high seismic risk. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69 4.4 Flooded segments of the road network (50-year return period), Inner Kampala . . . . . . 74 4.5 Simulation of air temperature in the streets of Greater Paris at 4 a.m., after nine days of a heat wave similar to that of 2003. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79 5.1 Mapping Tanzania’s supply chains onto its transport network reveals the impact of transport disruptions on Tanzanian households . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90 7.1 The criticality of a link can be measured by the additional road user cost resulting from its disruption . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110 7.2 Belgium’s transport network is much denser and offers greater redundancies than Madagascar’s. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111 7.3 The strengthening of infrastructure assets in Mozambique is prioritized based on risk levels and criticality . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113 CONTENTS xi 8.1 The criticality of a road depends on how it is used . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 130 8.2 The pattern of urbanization in Addis Ababa closely follows the major public transport lines. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 135 8.3 Risk-informed urbanization planning can help to accommodate the growing urban population of Fiji while limiting the increase in natural risks . . . . . . . . . . . . . . . 136 10.1 Different measures of natural risks in the Philippines highlight different priorities for interventions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 159 B11.1.1 In some U.S. states, revising stormwater infrastructure standards is urgent. . . . . . . . . . 167 Photo 7.1 A wetland park in Colombo helps to mitigate flood risk and offers recreational opportunities, such as birdwatching towers. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119 Tables O.1 Disrupted infrastructure services have multiple impacts on firms . . . . . . . . . . . . . . . . . . . 3 O.2 Disrupted infrastructure services have multiple impacts on households . . . . . . . . . . . . . . 6 O.3 Five recommendations to address the five obstacles to resilient infrastructure . . . . . . . . 21 2.1 Disrupted infrastructure services have multiple impacts on firms . . . . . . . . . . . . . . . . . . 34 3.1 Disrupted infrastructure services have multiple impacts on households . . . . . . . . . . . . . 50 4.1 Climatic events and their impacts on telecommunications infrastructure . . . . . . . . . . . . 76 B6.2.1 With the right policies in place, investments of 4.5 percent of GDP in infrastructure may be needed . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102 PIII.1 Key obstacles to more resilient infrastructure services and examples of underlying causes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 142 11.1 Examples of the presence (and absence) of incentives for resilience . . . . . . . . . . . . . . . 164 13.1 Cost multipliers vary across financial instruments for risk management. . . . . . . . . . . . 187 A.1 Engineering options to improve infrastructure asset resilience in the power sector. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 195 A.2 Engineering options to improve infrastructure asset resilience in the water sector. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 197 A.3 Engineering options to improve infrastructure asset resilience in the railways sector. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 198 A.4 Engineering options to improve infrastructure asset resilience in the roadway sector. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 199 Foreword Resilient infrastructure is about people. It is the natural hazards that are increasing due to about the households and communities for climate change. whom infrastructure is a lifeline to better But there is good news. Around the world, health, better education, and a better liveli- there are many examples of investments that hood. It affects people’s well-being, their eco- make infrastructure more resilient and more nomic prospects, and their quality of life. economically robust. Resilient infrastructure, is in part, about This report assesses, for the first time, the bridges that can withstand more frequent or cost of infrastructure disruptions to low- and stronger floods, water pipes that can resist middle-income countries and the economic earthquakes, or electric poles that are sturdier benefits of investing in resilient infrastructure. in the face of more intense hurricanes. And it is It examines four essential infrastructure also about making sure people will not lose systems: power, water and sanitation, trans- their jobs because they cannot get to work, port, and telecommunications. And the report that they can get urgent medical care, and that lays out a framework for understanding the their children can get to school. ability of infrastructure systems to function In developing countries, infrastructure dis- and meet users’ needs during and after natural ruptions are an everyday concern. When infra- shocks. structure fails, it undermines businesses, job We find that the extra cost of building resil- creation, and economic development. With ience into these systems is only 3 percent of rapidly growing populations and a changing overall investment needs. Thanks to fewer dis- climate increasing the frequency and intensity ruptions and reduced economic impacts, the of natural hazards, the need to adapt and invest overall net benefit of investing in the resilience in resilience should be an urgent priority. of infrastructure in developing countries would Disruption to infrastructure costs households be $4.2 trillion over the lifetime of new infra- and firms in low- and middle-income countries structure. That is a $4 benefit for each dollar at least $390 billion a year, and the indirect invested in resilience. effects place a further toll on households, busi- Finally, with a range of clearly defined rec- nesses, and communities. It is typically caused ommendations, the report lays out how to by poor maintenance, mismanagement, and unlock this $4.2 trillion opportunity. Rather xiii xiv FOREWORD than just spending more, the focus is on spend- There is no time to waste. With a rapidly ing better. The message for infrastructure changing climate, and large investments in investors, governments, development banks, infrastructure taking place in many countries, and the private sector is this: Invest in regula- business as usual over the next decade would tions and planning, in the early stages of proj- cost $1 trillion more. By getting it right, how- ect design, and in maintenance. Doing so can ever, we can provide the critical infrastructure significantly outweigh the costs of repairs or services—lifelines—that will spur sustained reconstruction after a disaster strikes. and resilient economic development. Kristalina Georgieva Chief Executive Officer The World Bank Acknowledgments The Lifelines report was prepared by a team led and Naho Shibuya contributed to the sections by Stéphane Hallegatte, with Jun Rentschler on public-private partnerships. The team at and Julie Rozenberg. It benefited from contri- Miyamoto International provided important butions from multiple teams working on dif- insights into the engineering solutions to build ferent sectors and topics. The power sector resilience. analysis was led by Claire Nicolas, with a team As World Bank Group peer reviewers, Greg composed of Christopher Arderne, Diana Browder, Marianne Fay, Vivien Foster, Hideaki Cubas, Mark Deinert, Eriko Ichikawa, Elco Hamada, Helen Martin, Shomik Mehndiratta, Koks, Ji Li, Samuel Oguah, Albertine Potter Artessa Saldivar-Sali, Alanna Simpson, and van Loon, and Amy Schweikert. The water Vladimir Stenek provided invaluable com- sector analysis was led by Zhimin Mao, work- ments and suggestions. Thanks also to external ing with Laura Bonzanigo, Xi Hu, Elco Koks, advisors: Carter Brandon, Jim Hall, Guillaume Weeho Lim, Raghav Pant, Patrick Ray, Clem- Prudent-Richard, Adam Rose, and Yasuyuki entine Stip, Jacob Tracy, and Conrad Zorn. The Todo. transport sector analysis was led by Julie Suggestions, comments, and data were pro- Rozenberg, with Xavier Espinet Alegre, vided by Anjali Acharya, Charles Baubion, Charles Fox, Stuart Fraser, Jim Hall, Elco Koks, Andrii Berdnyk, Moussa Blimpo, Marga Can- Mercedeh Tariverdi, Michalis Vousdoukas, and tada, Debabrata Chattopadhyay, Ashraf Dewan, Conrad Zorn. The telecommunication analysis Mirtha Escobar, Charles Esser, Scott Ferguson, was contributed by Himmat Sandhu and Matias Herrera Dappe, Martin Humphreys, Siddhartha Raja. The analysis of firm and Marie Hyland, Oscar Ishizawa, Asif Islam, Bren- household surveys was led by Jun Rentschler, den Jongman, Denis Jordy, Balázs Józsa, Shefali with Paolo Avner, Johannes Braese, Alvina Khanna, Brian Kinuthia, Shweta Kulkarni, Erman, Nick Jones, Martin Kornejew, Sadick Mathijs van Ledden, Jia Jun Lee, Richard Nassoro, Marguerite Obolensky, Samet Sahin, MacGeorge, Justice Tei Mensah, Jared Mer- and Eugene Tan. Shinji Ayuha, Célian Colon, cadante, Brian Min, Alice Mortlock, Sumati Etienne Raffi Kechichian, Maryia Markhvida, Rajput, Steven Rubinyi, Jason Russ, Peter Nah Yoon Shin, Shoko Takemoto, and Brian Sanfey, Guillermo Siercke, Ben Stewart, Shen Walsh contributed to the sections on resilient Sun, Janna Tenzing, Joshua Wimpey, Davida industries and supply chains. Sanae Sasamori Wood, and Fan Zhang. xv xvi ACKNOWLEDGMENTS Susan Graham of the World Bank Group’s The team thanks Julie Dana, manager of the Publishing Unit was the production editor. Edi- Global Facility for Disaster Reduction and torial services were provided by Sabra Ledent, Recovery (GFDRR), and Luis Tineo for their Laura Wallace, Nick Paul, Devan Kreisberg, support in the development of this project. Inge Pakulski, and Elizabeth Forsyth. Brad Finally, the team acknowledges the gener- Amburn designed the cover and created the ous support of the Japan–World Bank Program graphs. The team also thanks Aziz Gökdemir for Mainstreaming Disaster Risk Management and Jewel McFadden for their help in preparing in Developing Countries, the Climate Change the report for production. Visibility and launch Group of the World Bank under the leadership of the report were supported by Ferzina Banaji, of John Roome and Bernice Van Bronkhorst, with Uwimana Basaninyenzi, Joana Lopes, and the World Bank Sustainable Development Camila Perez, Mehreen Arshad Sheikh, and Practice Group led by Laura Tuck. Gerardo Spatuzzi. Abbreviations AMI advanced metering infrastructure APEC Asia-Pacific Economic Cooperation DEM digital elevation model EAD expected annual damage ESG environmental, social, and governance (principles) GDP gross domestic product GFDRR Global Facility for Disaster Reduction and Recovery ICT information and communication technology IPCC Intergovernmental Panel on Climate Change ISO International Organization for Standardization LPI Logistics Performance Index OECD Organisation for Economic Co-operation and Development PGA peak ground acceleration PPP public-private partnership RUC road user cost UNISDR United Nations Office for Disaster Risk Reduction WEF World Economic Forum WGI Worldwide Governance Indicators (World Bank report) WMO World Meteorological Organization xvii Overview F rom serving our most basic needs to enabling our most ambitious ventures in trade or technology, infrastructure services support our well-being and development. Reliable water, sanitation, energy, transport, and telecommunication services are uni- versally considered to be essential for raising the quality of life of people. Access to ba- sic infrastructure services is also a central factor in the productivity of firms and thus of entire economies, making it a key enabler of economic development. And in this time of rapid climate change and intensifying natural disasters, infrastructure systems are under pressure to deliver resilient and reliable services. By one estimate, governments in low- and ruptions in transport and energy networks, middle-income countries around the world are which in turn affect telecommunications and investing around $1 trillion—between 3.4 per- other essential services. The lack of resilient cent and 5 percent of gross domestic product sanitation systems also means that floods often (GDP)—in infrastructure every year (Fay et al. spread dangerous waterborne diseases. 2019).1 Still, the quality and adequacy of infra- The disruption of infrastructure services is structure services vary widely across countries. especially severe when considering more Millions of people, especially in fast-growing extreme natural shocks. For example, earth- cities in low- and middle-income countries, are quakes damage port infrastructure and slow facing the consequences of substandard infra- down local economies, as occurred in Kobe in structure, often at a significant cost. Under- 1995. Hurricanes wipe out electricity transmis- funding and poor maintenance are some of the sion and distribution systems, cutting people’s key factors resulting in unreliable electricity access to electricity for months, as occurred in grids, inadequate water and sanitation systems, Puerto Rico in 2017. In these examples, many and overstrained transport networks. people who did not experience direct damage Natural hazards magnify the challenges from the disaster still experienced impacts from faced by these already-strained and fragile sys- infrastructure disruptions. tems. Urban flooding, for instance, is a reality This report, Lifelines: The Resilient Infrastruc- for people around the world—from Amman, ture Opportunity, explores the resilience of four Buenos Aires, and Dar es Salaam to Jakarta essential infrastructure systems: power, water and Mumbai. Often exacerbated by poor drain- and sanitation, transport, and telecommunica- age systems, these floods cause frequent dis- tions. All of these systems provide critical ser- 1 2 LIFELINES vices for the well-being of households and the and supply chains—better able to manage productivity of firms, yet they are particularly disruptions. This report finds that investing vulnerable to natural hazards because they are $1 in more resilient infrastructure is bene- organized in complex networks through which ficial in 96 percent of thousands of scenar- even small local shocks can propagate quickly. ios exploring possible future socioeconomic Making them more resilient—that is, better and climate trends. In the median scenario, able to deliver the services people and firms the net benefit of investing in more resilient need during and after natural shocks—is criti- infrastructure in low- and middle-income cal, not only to avoid costly damage but also to countries is $4.2 trillion, with $4 in benefit minimize the wide-ranging consequences of for each $1 invested. Climate change makes natural disasters for the livelihoods and well- action on resilience even more necessary being of people. and attractive: on average, it doubles the net Building on a wide range of case studies, benefits from resilience. And because large global empirical analyses, and modeling exer- investments in infrastructure are currently cises, this report arrives at three main messages: being made in low- and middle-income countries, the median cost of one decade of • The lack of resilient infrastructure is harming peo- inaction is $1 trillion. ple and firms. Natural disasters cause direct • Good infrastructure management is the neces- damage to power generation and transport sary basis for resilient infrastructure, but targeted infrastructure, costing about $18 billion a year actions are also needed. Unfortunately, no sin- in low- and middle-income countries. This gle intervention will make infrastructure damage is straining public budgets and reduc- systems resilient. Instead, a range of coor- ing the attractiveness of these sectors for pri- dinated actions will be required. The first vate investors. But natural hazards not only recommendation is for countries to get the damage assets, they also disrupt infrastructure basics right—proper planning, operation, services, with significant impacts on firms and and maintenance of their assets—which people. Altogether, infrastructure disruptions can both increase resilience and save costs. impose costs between $391 billion and $647 However, good design and management billion a year on households and firms in low- alone are not enough to make infrastruc- and middle-income countries. These disrup- ture resilient, especially against rare and tions have a wide range of causes, including high-intensity hazards and long-term trends poor maintenance, mismanagement, and like climate change. To address these issues, underfunding. But case studies suggest that this report offers four additional recom- natural hazards typically explain 10 percent mendations: define institutional mandates to 70 percent of the disruptions, depending and strategies for infrastructure resilience; on the sector and the region. introduce resilience in the regulations and • Investing in more resilient infrastructure is incentive systems of infrastructure sectors, robust, profitable, and urgent. In low- and users, and supply chains; improve decision middle-income countries, designs for more making through data, tools, and skills; and resilient assets in the power, water and san- provide appropriate financing—especially itation, and transport sectors would cost for risk-informed master plans, asset design, between $11 billion and $65 billion a year and preparedness. Actions on these issues by 2030—an incremental cost of around 3 can be highly cost-effective and transfor- percent compared with overall investment mational, but they can nevertheless be chal- needs. And these costs can be reduced by lenging to fund in many poor countries, looking at services, not just assets, and mak- making them priorities for support from the ing infrastructure service users— households international community. OVERVIEW 3 INFRASTRUCTURE DISRUPTIONS indirectly, through their effects on the produc- ARE A DRAG ON PEOPLE AND tivity of firms, and directly, through their effects ECONOMIES on households’ consumption and well-being. This report begins by investigating how infra- structure disruptions—regardless of their origin— Infrastructure disruptions cost firms affect people and firms. The frequency of these more than $300 billion per year disruptions is generally closely linked to the level Unreliable infrastructure systems affect firms of economic development, as shown in figure through various impacts (table O.1). Most O.1 using GDP per capita as a proxy and electric- visible are the direct impacts: a firm relying on ­ ity and water outages from the World Bank’s water to cool a machine must halt production Enterprise Surveys. Disruptions cost people both during a dryout; a restaurant with an electric FIGURE O.1 Poorer countries are hit hardest by inadequate infrastructure a. Number of electricity outages per month b. Number of water outages per month 30 30 25 25 20 20 15 15 10 10 5 5 0 0 0 10,000 20,000 30,000 40,000 0 10,000 20,000 30,000 40,000 GDP per capita (US$) GDP per capita (US$) Source: Rentschler, Kornejew, et al. 2019, based on the World Bank’s Enterprise Surveys. Note: Panels a and b show the latest available survey data for 137 countries, but none older than 2009. Panel a only shows countries with up to 30 outages a month. Eight countries (all with GDP per capita below $9,000) report between 30 and 95 outages a month. TABLE O.1 Disrupted infrastructure services have multiple impacts on firms Sector Direct impacts Coping costs Indirect impacts Power • Reduced utilization rates • Generator investment ($6 billion ($38 billion a year) a year) • Sales losses ($82 billion a year) • Generator operation costs • Higher barriers to market entry and ($59 billion a year) lower investment • Less competition and innovation Water • Reduced utilization rates • Investment in alternative water due to lack of small and new firms ($6 billion a year) sources (reservoirs, wells) • Sales losses • Bias toward labor-intensive production Transport • Reduced utilization rates • Increased inventory ($107 billion a year) • More expensive location choices, • Inability to provide on-demand • Sales losses for example, in proximity to services and goods • Delayed supplies and deliveries clients or ports • Diminished competitiveness in international markets Telecommunications • Reduced utilization rates • Expensive location choices close • Sales losses to fast Internet Source: Rentschler, Kornejew, et al. 2019. Note: Highlighted in bold are the impacts for which original estimates are presented in this section. Estimates cover low- and middle-income countries. 4 LIFELINES stove cannot cook meals without power. Dis- to engage in productive, educational, and rec- ruptions leave production capacity unused, reational activities (Lenz et al. 2017). In South reduce firms’ sales, and delay the supply and Asia, Zhang (2019) finds that long power out- delivery of goods. Firms also incur costs for cop- ages are associated with a decrease in both per ing with unreliable infrastructure, such as for capita income and women’s labor force partici- backup power generation or water storage. The pation, probably because the lack of electricity indirect impacts of disruptions are less immedi- is associated with an increase in the time ate. They include effects on the long-term needed for domestic work (figure O.2). Studies investment and strategic decisions of firms and also identify a strong and consistent relation- on the composition, competition, and innova- ship between water outages and health tion of industries. Together, these effects figure impacts. In the Democratic Republic of Congo, in an economy’s ability to generate wealth and suspected cholera incidence rates increased in its international competitiveness (for details, 155 percent after one day of water disruption, see Braese, Rentschler, and Hallegatte 2019). compared with the incidence rate following Using a set of microdata on about 143,000 optimal water provision (Jeandron et al. 2015). firms, it is possible to estimate the monetary Infrastructure disruptions have many costs of infrastructure disruption for firms in impacts on households, and estimating the 137 low- and middle-income countries, repre- global cost is difficult (table O.2). For this analy- senting 78 percent of the world population sis, lower and upper bounds were established (map O.1).2 These data are used to assess the for power and water outages, based on studies impact of infrastructure disruptions on the assessing the willingness of households to pay to capacity utilization rates of firms—that is, to prevent such outages (see details in Obolensky compare the actual output of firms with the et al. 2019). For power outages, the estimates maximum output they can achieve using all range between 0.002 percent and 0.15 percent of their available resources—which is a good of GDP a year for low- and middle-income metric for firms’ performance. countries, which corresponds to between $2.3 The data reveal utilization losses from power, billion and $190 billion.3 In total, water inter- water, and transport disruptions of $151 billion ruptions are estimated to cost between 0.11 per- a year. (Unfortunately, a similar estimate for cent and 0.19 percent of GDP each year, which telecommunications is not possible because of a corresponds to a range of from $88 billion to lack of data.) In addition, firm data reveal sales $153 billion. Waterborne diseases stemming losses from electricity outages of $82 billion a from an intermittent water supply are estimated year and additional costs of self-generating to cause medical treatment costs and lost electricity of $65 billion a year. Although these incomes between $3 billion and $6 billion a figures highlight the significance of unreliable year. However, these results are highly uncer- infrastructure, they constitute lower-bound tain because of differences in methodologies estimates of the global costs of outages because and contexts. Similar assessments of the trans- neither all countries nor all types of impacts are port and telecommunications sectors were not covered in this analysis. possible due to data constraints. Infrastructure disruptions’ direct Natural shocks are among the leading impacts on people are worth at least causes of infrastructure disruptions $90 billion per year Taken together, the cost of infrastructure dis- Unreliable infrastructure services negatively ruptions ranges from $391 billion to $647 bil- affect the welfare of households. Frequent lion in the low- and middle-income countries power outages limit the ability of households for which data are available and for the types OVERVIEW 5 MAP O.1 Africa and South Asia bear the highest losses from unreliable infrastructure a. Countrywide average utilization rate losses from disruptions in electricity, water, and transport infrastructure b. Additional costs of firms’ backup electricity generation as % of GDP, including up-front investments and additional operating costs Source: Rentschler, Kornejew, et al. 2019. 6 LIFELINES FIGURE O.2 Reliable access to electricity has more favorable effects on income and social outcomes than access alone in Bangladesh, India, and Pakistan 40 37.0 35 31.2 30 28.0 25 24.2 23.0 21.1 % of change 20 17.1 16.7 15 13.8 11.7 9.6 10 5.8 6.5 5 2.3 2.0 0 Per capita Girls’ study Women’s Per capita Girls’ study Women’s Per capita Women’s income time labor force income time labor force income labor force participation participation participation Bangladesh India Pakistan  Increased access  Increased reliable access Source: Zhang 2019. Note: Estimates are based on household surveys in Bangladesh, India, and Pakistan. TABLE O.2 Disrupted infrastructure services have multiple impacts on households Sector Direct impacts Coping costs Indirect and health impacts Power • Diminished well-being • Generator investments • Higher mortality and morbidity (lack of access • Lower productivity of family • Generator operation costs to health care, air-conditioning during heat firms waves, or heat during cold spells) Willingness to pay to prevent outages: between $2.3 billion and $190 billion a year Water • Diminished well-being and • Investment in alternative • Higher incidence of diarrhea, cholera, and loss of time water sources (reservoirs, other diseases wells, water bottles) Willingness to pay to prevent outages: between $88 billion Medical costs and missed income: between and $153 billion a year $3 billion and $6 billion a year Transport • Greater congestion and loss • Higher cost of alternative • Air pollution and health impacts of time transport modes • Constrained access to jobs, markets, services • Higher fuel costs • People forced to live close to jobs, possibly on bad land Telecommunications • Diminished well-being • Inability to call emergency services Note: Highlighted in bold are the impacts for which original estimates are presented in this section. Estimates cover low- and middle-income countries. of impacts that can be quantified. Even though hazards play in these disruptions? While it is these estimates are incomplete, they highlight impossible to answer this question globally the substantial costs that unreliable infrastruc- and for all sectors, many case studies docu- ture impose on people in low- and middle- ment the role of natural hazards in infrastruc- income countries. But what role do natural ture disruptions. OVERVIEW 7 FIGURE O.3 Natural shocks explain a significant fraction of power outages 100 West Virginia Georgia Alabama 90 Share of power outages due to natural shocks (%) 80 Slovenia Croatia Belgium 70 Portugal 60 United States 50 Romania Italy France 40 Latvia Ireland Greece United Kingdom 30 Poland Germany Bangladesh (Dhaka) 20 Spain Sweden Lithuania 10 Czech Republic Bangladesh Canada Netherlands (Chittagong) Slovak Republic 0 0 10,000 20,000 30,000 40,000 50,000 60,000 70,000 80,000 GDP per capita (US$) Countries U.S. states Source: Rentschler, Obolensky, and Kornejew 2019. In the power sector, natural hazards—in partic- In many low- and middle-income countries, ular, storms—are a major cause of electricity natural shocks are responsible for a small frac- supply disruptions, as shown in figure O.3. In tion of power outages, although this does not Belgium, Croatia, Portugal, Slovenia, and the mean that resilience is not an issue. Indeed, United States, they are responsible for more power systems are more vulnerable to natural than 50 percent of all outages. By contrast, in shocks in poorer countries than in richer coun- Bangladesh, natural shocks account for a tries, and natural hazards can be responsible for smaller share of power outages—not because a large number of disruptions. In the power energy systems are more resilient, but because sector, aging equipment, a lack of maintenance, system failures and nonnatural factors are so rapid expansion of the grid, and insufficient frequent that energy users experience daily generation capacity are all factors that reduce outages. But this figure also underestimates the the reliability of service in general, while also role of natural hazards because outages caused increasing vulnerability to natural shocks. For by natural hazards tend to be longer and geo- example, storms of the same intensity are more graphically larger than other outages. In Europe likely to cause outages in Bangladesh than in between 2010 and 2017, natural hazard– the United States (figure O.4). On a day with induced outages lasted 409 minutes on aver- average wind speeds exceeding 35 kilometers age, making them almost four times as long as per hour, electricity users in Bangladesh are 11 outages caused by nonnatural causes. And in times more likely to experience a blackout than Bangladesh in 2007, Tropical Storm Sidr caused U.S. consumers. As a result of this vulnerability, the largest outage in national history: all 26 in 2013 in Chittagong, Bangladesh, users expe- power plants tripped and failed, leaving cus- rienced about 16 power outages due to storms tomers without power for up to a week (Rent- alone. This number corresponds to only 4 per- schler, Obolensky and Kornejew 2019). cent of all outages experienced, yet it is already 8 LIFELINES FIGURE O.4 The vulnerability of the power network to wind is much higher in Bangladesh than in the United States more than 15 times higher than the average 16 number of outages experienced by consumers  Bangladesh in New York City. Share of windy days during which at least 14.3 14  United States In the transport sector, floods and other haz- one outage was reported (%) 12 ards disrupt traffic and cause congestion, taking 10 a toll on people and firms in rich and poor 8.4 8 countries alike. In Kampala, the impacts of floods on urban transport reduce people’s 6 access to a health care facility, according to an 4 3.7 2.4 analysis undertaken for this report (Rentschler, 1.8 2 0.9 1.2 Braese, et al. 2019) (figure O.5). A network 0.3 0.4 0.6 0 analysis estimates that the mean travel time by >15 >20 >25 >30 >35 car to a hospital from nearly all locations in Wind speed (km/hour) Inner Kampala is less than 30 minutes. How- Source: Rentschler, Obolensky, and Kornejew 2019. ever, during a 10-year flood, disruption of the Note: Windy days are defined using different thresholds for recorded daily wind road network can increase travel times signifi- speeds. Wind speeds are obtained from the global ERA5 climate reanalysis model, which tends to underrepresent the highest local wind speeds. cantly, and about a third of persons living in FIGURE O.5 Floods in Kampala severely restrict people’s access to health care facilities a. Travel a. Mean time travel from times locations from across locations all Kampala in Inner b. b. Increase in Increases in travel time from travel times from locations across locations acrossInner Kampala of Inner Kampala health toto care health facilities care facilities to health Inner Kampala care facilities to hospitals during aflood in a 10-year 10-year flood Frequency density 0 10 20 30 40 50 60 Increase in travel time (%)  0–27  10-year flood extent Minutes  27–36  Bodies of water No flood  36–47 Area of analysis 10-year flood  47–70 Roads 50-year flood  > 70 Trips no longer possible Source: Rentschler, Braese, et al. 2019. Note: In panel a, the vertical line denotes the “golden hour” (the window of time that maximizes survival of a major health emer- gency), assuming that ambulances complete a return trip starting at a hospital. The curves show frequency densities that represent the distribution of travel times from all locations. The 10-year flood is the flood of a magnitude that occurs on average once every 10 years. OVERVIEW 9 Inner Kampala would no longer be able to FIGURE O.6 Tanzanian firms report large reach health facilities within the “golden losses from infrastructure disruptions hour”—a rule of thumb referring to the win- 350 325 dow of time that maximizes the likelihood of 300 Utilization losses (US$, millions) survival after a severe medical incident. Such flood-related transport disruptions are 250 46% 216 costly for firms. The same network analysis 200 estimates travel times between some 400 firms 47% 150 127 as a proxy for the impact of floods on interfirm connectivity and local supply chains. A moder- 100 ate flood in Kampala increases the average 50 travel time between firms by 54 percent. A sig- 0 nificant number of firms are affected even Transport Power Water more severely, with more than a quarter of  Losses due to disruptions caused by rain and floods firms facing an increase in average travel time  Losses due to disruptions caused by other factors of between 100 percent and 350 percent. As Source: Rentschler, Braese, et al. 2019. roads are flooded, people are unable to reach their workplace, and firms wait in vain for sup- plies, miss their deliveries, and lose sales. firms and households, local studies are needed In the water sector, assets and services are also to provide a detailed assessment. To support affected by natural hazards, even in the such an assessment, a survey was developed absence of physical damage to assets. The and piloted in Tanzania for a sample of 800 severe landslides that occurred in Lima in firms across the country. It found that Tanza- March 2017, interrupted the water supply for nian firms are incurring utilization losses of four days, as the city’s river filled with mud. $668 million a year from power and water The main water treatment plant could not outages and transport disruptions, which is handle the resulting turbidity and had to shut equivalent to 1.8 percent of the country’s GDP down (Stip et al. 2019). (figure O.6). Power alone is responsible for In the telecommunications sector, in December losses of $216 million a year, and 47 percent of 2006, the Great Hengchun Earthquake on the these losses are solely due to power outages island of Taiwan, China, and in the Luzon that can be attributed to rain and floods (equiv- Strait was one of the severest examples of dis- alent to $101 million, or 0.3 percent of GDP). ruptions to the submarine cable systems on As for transport disruptions, about 46 percent which international communications networks of utilization losses stem from disruptions depend. Submarine landslides caused 19 breaks caused by rain and floods (equivalent to $150 in seven cable systems, requiring repairs that million, or 0.4 percent of GDP). But the survey were carried out over 49 days. Meanwhile, does not find that rain and floods have a signif- traffic was quickly rerouted using undamaged icant impact on the incidence of water supply infrastructure, but the pressures on it resulted disruptions. in a lower quality of service and delays. Inter- In addition to these disruptions, natural haz- net connectivity in the region was seriously ards cause direct damage to infrastructure affected, and financial services and the airlines assets. This damage is critical, given that it bur- and shipping industries were significantly hurt dens public infrastructure budgets and detracts (Sandhu and Raja 2019). from the attractiveness of the infrastructure Although it is agreed that disruptions from sector for private investors. Based on a global natural hazards represent a significant cost for risk assessment performed for this report, 10 LIFELINES power generation and transport infrastructure estimate the total cost of a two-week blackout incur losses of $30 billion a year on average in Los Angeles at $2.8 billion—that is, 13 per- from natural hazards (about $15 billion each), cent of the total economic activity during the with low- and middle-income countries shoul- two weeks. Colon, Hallegatte, and Rozenberg dering about $18 billion of the total amount (2019) find that in Tanzania, the macroeco- (Koks et al. 2019; Nicolas et al. 2019). nomic impact of a flood disruption in the trans- Although these numbers remain manage- port sector increases nonlinearly with the dura- able on average and at the global level, losses tion of the disruption. A four-week disruption can reach high values after extreme events. In is, on average, 23 times costlier for households some vulnerable countries, they are high than a two-week disruption. Comprehensive enough to impede the provision of universal risk assessments need to account for these sec- access to infrastructure services. ondary impacts and look beyond asset losses to The severity of natural disasters is usually inform disaster risk management investments measured by the asset losses they provoke and policies properly and to guide decision (Munich Re 2019; Swiss Re 2019). But the sec- making on infrastructure design and operation. ondary consequences of direct asset losses on economic activities and output can often explain MORE RESILIENT INFRASTRUCTURE a large share of total disaster impacts, especially ASSETS PAY FOR THEMSELVES when infrastructure systems are affected (Halle- The resilience of infrastructure has three levels gatte 2013; Hallegatte and Vogt-Schilb 2016). (figure O.7): For example, Rose, Oladosu, and Liao (2007) • Resilience of infrastructure assets. In the nar- rowest sense, resilient infrastructure refers FIGURE O.7 The resilience of infrastructure should be to assets such as roads, bridges, cellphone considered at several overlapping and complementary levels towers, and power lines that can withstand external shocks, especially natural hazards. Here, the benefit of more resilient infra- High-quality infrastructure structure is that it reduces the life-cycle cost of assets. • Resilience of infrastructure services. Infrastruc- Resilience of infrastructure users ture systems are interconnected networks, Resilient infrastructure reduces the impact of and the resilience of individual assets is a natural hazards on people and economies poor proxy for the resilience of services pro- vided at the network level. For infrastruc- Resilience of infrastructure services ture, a systemic approach to resilience is Resilient infrastructure provides more reliable services preferable. At this level, the benefit of more resilient infrastructure is that it provides more reliable services. Resilience of infrastructure assets • Resilience of infrastructure users. Eventually, Resilient infrastructure is less what matters is the resilience of users. Infra- costly to maintain and repair structure disruptions can be catastophic or benign, depending on whether users— including people and supply chains—can cope with them. At this level, the benefit of more resilient infrastructure is that it OVERVIEW 11 reduces the total impact of natural hazards FIGURE O.8 The incremental cost of increasing the resilience on people and economies. of future infrastructure investments depends on the spending scenario but remains limited in all cases The resilience of infrastructure is one of the many determinants of high-quality infrastruc- 60 Average annual cost (US$, billions) ture. However, integrating resilience in the 50 design and implementation of infrastructure investments not only helps to manage natural 40 shocks but also complements the cost- 30 effectiveness and quality of infrastructure ser- 20 vices more generally. 10 Building more resilient infrastructure 0 assets in exposed areas is cost-effective Power Transport Water and Total The additional up-front cost of more resilient sanitation infrastructure assets ranges from negative to a Source: Hallegatte et al. 2019. Note: This figure shows the incremental annual capital cost for more resilient infra- doubling of the construction cost, depending on structure for 2015–30. The range comes primarily from the uncertainty on how much the asset and the hazard. Interventions to make will be invested on infrastructure during the period (and on the technologies chosen). assets more resilient include using alternative materials, digging deeper foundations, elevating methods are available. Improving the resilience assets, building flood protection around the of only the assets that are exposed to hazards asset, or adding redundant components. would increase investment needs in power, How much would it cost to implement these water and sanitation, and transport by between technical solutions? This report tackles this $11 billion and $65 billion a year (figure O.8). question with an analysis that begins with the Although not negligible, this range represents estimates by Rozenberg and Fay (2019) of how only 3 percent of infrastructure investment much low- and middle-income countries would needs and less than 0.1 percent of the GDP of have to spend on infrastructure to achieve their low- and middle-income countries. It would, development goals. The analysis then asks how therefore, not affect the current affordability much those estimates would change if infra- challenges that countries face. structure systems were designed and built in a However, making infrastructure more resil- more resilient manner (using one set of techni- ient by strengthening assets is realistic only if cal options from Miyamoto International 2019). the appropriate data on the spatial distribution Note that the solutions assessed here do not of natural hazards are available. Without infor- guarantee that assets cannot be damaged by mation on which locations are exposed to haz- natural hazards and do not include all possible ards, strengthening the whole system would options to reduce risks. Many high-income cost 10 times more, between $120 billion and countries like Japan implement technical solu- $670 billion, which suggests that the value of tions that go beyond—and are more expensive hazards data is orders of magnitude higher than—the set of solutions considered in this than the cost of producing the information. analysis. What are the returns on investments for Overall, the incremental cost of building the making exposed infrastructure more resilient resilience of infrastructure assets in low- and to natural disasters? The uncertainty pertaining middle-income countries is small, provided the to the cost of infrastructure resilience and the right data, risk models, and decision-making benefits in terms of both avoided repairs and 12 LIFELINES disruptions for households and firms make it of natural hazards and climate change. In 93 difficult to provide one single estimate for the percent of the scenarios, it is costly to delay benefit-cost ratio of strengthening exposed action from 2020 to 2030—and the median cost infrastructure assets. However, a set of 3,000 of a decade of inaction is $1 trillion. scenarios (which covers the uncertainty of all parameters of the analysis) can be used to From resilient infrastructure assets to explore the costs and benefits of making infra- resilient infrastructure services structure more resilient. Making assets more resistant is not the only The analysis shows that, despite the uncer- option for building resilience. Expansion of the tainty, investing in more resilient infrastructure analysis from infrastructure assets to infrastruc- is clearly a cost-effective and robust choice. The ture services reveals that the cost of resilience benefit-cost ratio is higher than 1 in 96 percent can be reduced further by working at the net- of the scenarios, larger than 2 in 77 percent of work and system level—looking at criticality, them, and higher than 6 in 25 percent of them redundancy, diversification, and nature-based (Hallegatte et al. 2019). The net present value of solutions as additional options. these investments, over the lifetime of new To illustrate the role of networks in infra- infrastructure assets, exceeds $2 trillion in 75 structure system resilience, a study conducted percent of the scenarios and $4.2 trillion in half for this report quantifies the resilience of transport of them. Moreover, climate change makes the networks, defined as the ratio of the loss of func- strengthening of infrastructure assets even more tionality to the loss of assets (Rozenberg et al. important. Without climate change, the median 2019b). A resilient road network, such as the benefit-cost ratio would be equal to 2, but it one in Belgium or Morocco, can lose many doubles when climate change is considered. assets (such as road segments) without losing The urgency of investing in better infrastruc- much functionality, whereas fragile networks ture is also evident. With massive investment in with little redundancy, such as the one in Mad- infrastructure taking place in low- and middle- agascar, become disfunctional even with slight income countries, the stock of low-resilience damage (figure O.9). Similar approaches can be assets is growing rapidly, increasing future costs mobilized in water systems, where the typical methodology consists of mapping all compo- nents of a network and assessing the conditions FIGURE O.9 Belgium’s and Morocco’s transport systems can under which they would fail, what the effects absorb much larger road disruptions than Madagascar’s of those failures would be, and how they would 100 affect service delivery. Loss of functionality of the network (%) Network effects create opportunities to 75 strengthen the resilience of services and users at a limited cost, either by strengthening criti- cal assets or by building in redundancy only 50 where there are choke points (Rozenberg et al. 2019a). For transmission and distribution net- 25 works, for example, resilience is often built up through redundancy, which does not necessar- 0 ily mean doubling or tripling key components 0 10 20 30 40 50 60 of the network. A more effective approach is Level of disruption (% links disrupted) usually to create “ringed” or meshed networks Belgium Madagascar Morocco that have multiple supply points for various Source: Rozenberg et al. 2019b. nodes in the grid. OVERVIEW 13 Diversification and decentralization also ture systems have to be stress-tested against a offer opportunities for more resilient services. range of events to minimize the risk of cata- The use of power generation with differentiated strophic failures (Kalra et al. 2014). Such stress vulnerabilities (for example, hydropower, tests have two goals: (1) identify low-cost which is vulnerable to drought, versus solar and options that can reduce the vulnerability of wind, which are vulnerable to strong winds) infrastructure systems to extreme events, even makes it more likely that a system will be able quite unlikely ones, and (2) prepare for failure to maintain a minimum level of service. Multi- in terms of managing infrastructure systems modal transport systems that rely on nonmo- (such as how to recover from a major failure) torized modes and public transit are more resil- and in terms of supporting users (such as how ient than systems that rely on private vehicles to minimize impacts on hospitals). Running sce- only. Distributed power systems using solar and narios of failures is the first and most critical batteries can harden a grid and make it more step in defining contingency plans. resilient. Minigrids and microgrids, because Finally, sometimes the best way to make an they do not rely on long-distance transmission infrastructure resilient is not to build it. Nich- wires, can provide useful backup generation in olls et al. (2019) find that coastal protection case of grid failure. During Hurricane Sandy, against storm surges and a rise in sea level the Co-Op City microgrid in New York City was would make economic sense only for about successfully decoupled from the main grid, and 22–32 percent of the world’s coastlines through it supported consumers during outages in the the 21st century. Thus, some communities may wider network (Strahl et al. 2016). have to retreat gradually or use lower-cost or Combining green and gray infrastructure nature-based approaches to coastal defense. can provide lower-cost, more resilient, and These communities are mostly in low-density more sustainable infrastructure solutions areas where the costs of protection are too high (Browder et al. 2019). In New York City, 90 to be affordable. In those areas, the best percent of water is from well-protected wilder- approach to resilience may be not to build new ness watersheds, making New York’s water infrastructure. This approach, however, has to treatment process simpler than that of other be complemented by a consistent strategy to U.S. cities (National Research Council 2000). manage retreat, while maintaining livelihoods According to Beck et al. (2018), without coral and community ties. reefs the annual damage from coastal flooding would double worldwide. They estimate that From resilient infrastructure services to Cuba, Indonesia, Malaysia, Mexico, and the resilient users and economies Philippines benefit the most from their reefs, In some cases, it can be easier and cheaper to with annual savings of more than $400 million manage service interruptions than to prevent for each country. In Colombo, preserving the them. This report explores the role of the users wetland system was found to be a cost- of infrastructure services and how their actions effective solution to reducing flooding in the can contribute to more resilient infrastructure city, even when accounting for land develop- systems. ment constraints (Browder et al. 2019). Often, a first option for building resilience is Limits to what is achievable in terms of to reduce demand by improving efficiency. In strengthening also need to be considered. No the face of growing populations and increas- infrastructure asset or system can be designed to ingly scarce water resources, a water utility can cope with all possible hazards. And great uncer- use demand management to reduce stress on tainty surrounds the probability and intensity of the city’s water supply. A recent example is the most extreme events. As a result, infrastruc- Cape Town, which had to take drastic measures 14 LIFELINES MAP O.2 Investment priorities for Tanzania’s transport network will depend on its supply chains a. Impacts of disruption on households’ consumption b. Impacts of disruption on international clients Source: Colon, Hallegatte, and Rozenberg 2019. Note: The width of the line overlaying a given road is proportional to the impacts that a one-week disruption of that road would trigger. Impacts, mea- sured in percentage of daily consumption, represent exceptional expenditures due to costlier transport and missed consumption due to shortages. Panel a shows these impacts for products consumed by households, and panel b shows these impacts for international buyers. to avoid reaching “Day 0”—the day the city considered most vulnerable or most important. would run out of water. The demand manage- For example, segments of the coastal trunk ment measures implemented by the city were road, located about 200 km south of Dar-es- extremely successful, reducing use by 40 per- Salaam, are critical for domestic consumption cent between 2015 and 2018 and preventing but rather irrelevant for international trade. For what could have been a major socioeconomic trade, the road east of Morogoro appears as a crisis. priority. This segment accommodates large Understanding the needs and capacities of freight flows between the port of Dar es Salaam users helps utilities to target better where to and landlocked countries, such as the Demo- invest and what part of the network to cratic Republic of Congo and Zambia. strengthen. A power distribution line to a hos- When preventing disruptions is not possible pital or a flood shelter is likely more important or not affordable, firms have many options for during and after an emergency than the aver- improving their own resilience to disruptions. age power line in a country. To investigate how Larger inventories will protect them against criticality depends on users and supply chains, a transport issues. Generators and batteries will study undertaken for this report combines a help them manage short power outages. Main- transport and a supply chain model to investi- taining a diversity of suppliers, from both local gate the criticality of the transport network in and distant locations, is another powerful safe- Tanzania (Colon, Hallegatte, and Rozenberg guard, especially against long disruptions. How- 2019). Map O.2 shows the most critical assets in ever, holding large inventories and managing the transport sector for two supply chains and multiple suppliers are financial burdens that reveals that investment priorities for strength- involve significant transaction costs, making ening assets depend on which supply chains are them most relevant for large firms. Because a OVERVIEW 15 static supply chain will never be able to cope FIGURE O.10 Spending more improves the reliability of the with a large-scale disaster and associated disrup- transport system, but only if governance improves as well tions, adaptability is critical and should be 5 Best Logistic Performance Index: Timeliness embedded in business continuity plans (Chris- topher and Peck 2004; Sheffi 2005). 4 MAKING INFRASTRUCTURE 3 MORE RESILIENT REQUIRES A CONSISTENT STRATEGY In many countries, infrastructure disruptions 2 are the symptoms of chronic shortcomings. Worst Power outages occur every day, water supply is 1 0 200 400 600 800 1,000 1,200 1,400 unreliable or unsafe, and congestion makes Annual public road spending per capita (constant 2009 US$) travel slow and unpredictable. In many places, these disruptions occur simply because infra-  Spending and governance improve together  Increase in spending alone structure systems are not designed to keep up Source: Kornejew, Rentschler, and Hallegatte 2019. with ever-rising demand or because system failures are the result of poor asset manage- Thus, poor governance of infrastructure sys- ment or maintenance. While natural hazards tems is the first obstacle that needs to be tack- can exacerbate these issues, the majority of led. If infrastructure is to be resilient to natural these disruptions reflect more fundamental shocks, countries first need to get the basics challenges related to infrastructure design and right for infrastructure management, with the management. This means that, to make infra- following three priority actions. structure systems resilient, the first step is to make them reliable in normal conditions Action 1.1: Introduce and enforce regulations, through appropriate infrastructure design, construction codes, and procurement rules operation, maintenance, and financing. Well-designed regulations, codes, and procure- ment rules are the simplest approach to Recommendation 1: Get the basics right enhancing the quality of infrastructure ser- Underperforming infrastructure systems are vices, including their reliability and resilience. explained largely by poor management and Effective enforcement in the infrastructure sec- governance, according to a recent analysis of tor requires a robust legal framework, but also countries across the world (Kornejew, Rent- strong regulatory agencies to monitor con- schler, and Hallegatte 2019). Using the World struction, service quality, and performance and Bank’s Logistic Performance Index as a proxy, to reward or penalize service providers for their figure O.10 shows how the performance of the performance. Currently, many regulators lack transport system depends on public spending the resources and capacity to enforce the exist- on roads. Performance increases rapidly with ing construction codes. spending per capita, but only if the quality of governance improves in parallel (dark blue Action 1.2: Create systems for appropriate line). If the quality of governance remains infrastructure operation, maintenance, and unchanged (light blue line), increased spending postincident response only yields marginal improvements in transport Improving maintenance and operations is a system performance and is not cost-effective. no-regret option (it generates benefits what- Similar analyses yield similar findings for power ever happens in the future) for boosting the and water systems. resilience of infrastructure assets while reduc- 16 LIFELINES ing overall costs. An analysis of member coun- Implementing these three basic measures tries of the Organisation for Economic Co- would contribute to more reliable infrastruc- operation and Development performed for this ture systems and establish a basic capacity to report suggests that each additional $1 spent cope with natural hazards and climate change. on road maintenance saves $1.5 in new invest- But they would not be sufficient to achieve ments, making better maintenance a very more ambitious objectives regarding resilience. cost-effective option (Kornejew, Rentschler, Without targeted actions to strengthen resil- and Hallegatte 2019). An important tool for ience, infrastructure assets will not be able to this purpose is infrastructure asset manage- cope with rarer events, such as hurricanes, ment systems, which include an inventory of river floods, or earthquakes. And without spe- all assets and their condition, as well as all of cific actions on climate change, these assets run the strategic, financial, and technical aspects of the risk of being designed for the wrong cli- the management of infrastructure assets across mate and environmental conditions. To build their life cycle. Such tools help to move toward resilience to these evolving natural hazards, it an evidence-based and preventive mainte- is necessary to tackle four additional obstacles nance schedule and away from a reactive that are specific to the resilience challenge. patch-by-patch approach to maintenance. Recommendation 2: Build institutions Action 1.3: Provide appropriate funding for resilience and financing for infrastructure planning, Political economy challenges and coordination construction, and maintenance failures impede the creation of a resilient infra- The quality of infrastructure services depends structure ecosystem. Governments, therefore, on many factors, from good planning to good need to play a coordinating role (OECD 2019), maintenance, but each of these comes at a cost with the following three priority actions. (figure O.11). If resources are insufficient to meet the need for any of these factors, the Action 2.1: Implement a whole-of-government quality of infrastructure services is likely to suf- approach to infrastructure resilience, building fer. Even if investment spending is appropriate, on existing regulatory systems insufficient resources for planning, designing, Analysts agree that governments play a key role or maintaining assets would result in low qual- in ensuring the resilience of critical infrastruc- ity and reliability. Dedicated funds and budget- ture and that they should adopt a whole-of- ary allocations can be used to ensure that government approach (Renn 2008; Wiener and enough resources are available to meet differ- Rogers 2002; World Bank 2013). A common ent needs, especially for maintenance. solution to improve the coordination of risk FIGURE O.11 High-quality infrastructure requires providing for multiple funding needs Cost to regulators and government Life-cycle cost to (public or private) infrastructure service • Master planning, and regulation providers Full design and enforcement • Project design and preparation infrastructure • Data and model development, • Up-front investment cost cost research, training, and education • Operational, maintenance, and repair costs • Decommissioning OVERVIEW 17 management across risks and across systems is structure disruptions. Therefore, governments to place an existing (or new) multiministry body need to include resilience in a consistent set of in charge of information exchange, coordina- regulations and financial incentives to align the tion, and possibly even implementation of risk interests of infrastructure service providers management measures for infrastructure. with the interests of the public (figure O.12), with the following three priority actions. Action 2.2: Identify critical infrastructure and define acceptable and intolerable risk levels Action 3.1: Consider resilience objectives in Criticality analyses are an important tool for master plans, standards, and regulations and identifying the most important infrastructure adjust them regularly to account for climate assets and their vulnerability. Once the critical change infrastructure assets and systems have been Standards and regulations need to account for identified, governments need to define risk a range of factors, including climate conditions, levels that are acceptable or intolerable. Each geophysical hazards, environmental and socio- infrastructure sector can use these risk levels to economic trends, local construction practices, design its own regulations and measures, and policy priorities. They also need to be ensuring consistency across systems. Definition revised more regularly than is the case today to of these risk levels needs to consider the local consider climate change and other long-term context, especially the resources that are avail- trends (Vallejo and Mullan 2017). In addition, able, and requires an open and participatory governments can use regulations to strengthen approach to ensure that risk management does the resilience of specific users of infrastructure not become an obstacle to development. services, not just providers. For example, hos- pitals could be required to maintain backup Action 2.3: Ensure equitable access to resilient generators, batteries, and water tanks. And infrastructure firms could be required to prepare business Decisions regarding resilience cannot be driven continuity plans to minimize the economic cost by economic considerations alone. The strength­ - of disasters and infrastructure disruptions. en­ing of infrastructure resilience should be guided by a more complete assessment of the Action 3.2: Create financial incentives potential risks and impacts of disruptions, espe- for service providers to promote resilient cially for vulnerable and marginalized popula- infrastructure services tion groups. New approaches enable more com- Rewards and penalties can be used as incen- prehensive assessments of spatial priorities. For tives for service providers to go beyond the example, estimates of well-being losses or socio­ mandatory standards and implement cost- economic resilience provide a balanced assessment effective solutions to improve resilience (Par- of the impacts of natural disasters on poor and dina and Schiro 2018). The Australian Energy rich households (Hallegatte et al. 2016; Walsh Regulator established the Service Target Perfor- and Hallegatte 2019). mance Incentive Scheme, which includes pen- alties and rewards calibrated according to Recommendation 3: Include resilience how willing consumers are to pay for improved in regulations and incentives service. Another example is payment-for- A third obstacle to more resilient infrastructure ecosystem-services schemes, which promote is that public and private decision makers tend the use of nature-based solutions to increase to have few incentives to avoid disruptions. resilience. In Brazil, water users pay a fee to the Too often, they only consider lower repair costs local water company that local watershed com- when deciding on investments in resilience; mittees use for watershed maintenance and they rarely consider the full social cost of infra- reforestation (Browder et al. 2019). 18 LIFELINES FIGURE O.12 Creating the right incentives for infrastructure service providers requires a consistent set of regulations and financial incentives 1 2 3 4 Government or Government or Government or Developer designs regulator defines regulator defines an regulator adds project above and enforces an “acceptable” level of incentives to align the minimum "intolerable" level of risk that can be the interest of standard risk through a tolerated ("force service providers minimum standard in majeure" event) with the public Intensity construction codes interest, with of hazards or procurement penalties and rewards based on social cost Major, rare events Acceptable risks: For rare events, infrastructure assets Government are expected to experience damage or disruptions that bears the risk need to be managed through contingent planning Force majeure Provider bears at least part of the risk (insurance may be required) Project-specific designs Minimum standard Infrastructure services Intolerable risks: Infrastructure should not be disrupted should resist frequent hazards below this level. Provider bears the risk Small, frequent hazards Action 3.3: Ensure that infrastructure Recommendation 4: Improve decision regulations are consistent with risk-informed making land use plans and guide development toward Even if regulators and providers of infrastruc- safer areas ture services have the right incentives to build Since infrastructure investments influence spa- more resilient infrastructure systems, they often tial development patterns, they can influence lack access to data and tools, as well as the skills people’s exposure to natural hazards. To ensure and competencies they need to make good deci- that new infrastructure contributes to the resil- sions. Governments, therefore, need to help all ience of users, regulations should be aligned stakeholders to improve their decision making, with risk-informed land use and urbanization with the following three priority actions. plans. And the choice of infrastructure localiza- tions needs to account for the potential invest- Action 4.1: Invest in freely accessible natural ments that a new infrastructure asset will hazard and climate change data attract and the implications for resilience. Even Investments in risk data and models (such as better, infrastructure localization choices can hydrological models, maps of flood hazards, dig- be used to support the implementation of land ital elevation models, and inventories of infra- use planning and promote low-risk spatial structure assets) can have extremely high development. returns by improving the design and mainte- OVERVIEW 19 nance of infrastructure assets. Producing digital structure resilience, their appropriate use elevation models for all urban areas in low- and requires skills that are not always available. middle-income countries would cost between Universities and research centers need to be $50 million and $400 million in total and make supported so that they can offer training, it possible to perform in-depth risk assessments develop new methodologies (or adapt them to for all new infrastructure assets, informing hun- the local context), and advise policy and deci- dreds of billions in investments per year. How- sion makers. When public sector expertise is ever, such data have public goods characteristics insufficient, bringing in the private sector— that discourage private actors from investing in through direct procurement or public-private them and require public support. To be useful, partnerships—can be a solution. risk and infrastructure data must be made avail- able (and affordable) to infrastructure service Recommendation 5: Provide financing providers and users. While privacy and security The fifth obstacle is linked to affordability and concerns can make it necessary to restrict access, financing constraints. Increasing resilience can it is preferable to make open access the default increase various components of the life-cycle situation for hazard and infrastructure data and cost of infrastructure, including the costs borne to create processes to restrict access for data by the government or regulators or the costs proven to be too sensitive. borne by infrastructure providers (figure O.11). At times, these costs can lead to affordability Action 4.2: Make robust decisions and minimize challenges, when resilience increases the full the potential for regret and catastrophic failures life-cycle cost of an asset or system. Solutions Often, large uncertainties make it impossible might include either an increase in funding to design “optimal” systems or assets. An (financed through higher taxes, user fees, or alternative is to seek robust designs that yield transfers) or a trade-off between the resilience good results across a wide range of futures, and quantity of infrastructure services (such as preferences, and worldviews, even if they fewer but safer roads). But more often, making may not be optimal for any particular future. infrastructure more resilient increases only the Decision makers can identify robust strategies costs of design, construction, or maintenance, through systematic stress-testing of possible while decreasing other costs such as repairs, so options for a variety of hazards and threats— that the overall life-cycle cost is reduced. The even highly unlikely ones—to ensure that the challenge in that case is linked to financing— residual vulnerabilities are acceptable and that is, transforming annual revenues or bud- manageable. These stress tests can help to gets into the resources needed at each stage of capture low-cost opportunities to build resil- the infrastructure project life cycle, with the ience to low-probability, high-consequence following three priority actions. events and prevent catastrophic failures. They can also support the development of contin- Action 5.1: Provide adequate funding to gency plans for service providers and business include risk assessments in master plans and continuity plans for users. early project design Even though hundreds of billions of dollars are Action 4.3: Build the skills needed to use data invested in infrastructure every year, it remains and models and mobilize the know-how of the difficult to mobilize resources for infrastructure- private sector sector regulations, risk-informed master plans, Even if infrastructure risk data and models are infrastructure risk assessment, or early-stage available to all those seeking to improve infra- project design. More resources tend to become 20 LIFELINES available when infrastructure projects are Action 5.3: Promote transparency to better mature, but at this stage most strategic deci- inform investors and decision makers sions have already been made, and most low- One way to ensure that resilient infrastructure cost options to increase resilience are no longer projects are adequately financed is to inform available (such as changing the location of an investors and decision makers about the risks asset or even the nature of the project). Sup- associated with projects. Multiple interna- porting and funding these activities is highly tional, regional, and national initiatives are cost-effective and can be transformational, seeking to make the physical risks associated especially in poorer countries, making them a with investments and assets more transparent. priority for international aid and cooperation Examples include the work of the Task Force (World Bank 2018). Dedicated organizations for Climate-Related Financial Disclosure, and project preparation facilities, such as the which recommends that firms and investors Global Facility for Disaster Reduction and report on physical risks and how they are man- Recovery or the Global Infrastructure Facility, aged. To contribute to this trend, the World are already active in these domains, but they Bank Group is committed to developing a resil- remain small compared with the magnitude of ience rating system to inform investors about the needs. the resilience of their infrastructure invest- ments and help them to select the most resil- Action 5.2: Develop a government-wide ient projects. financial protection strategy and contingency plans In sum, as illustrated by these five recom- In the aftermath of a disaster, governments are mendations and 15 actions (table O.3), no single typically required to raise significant financing measure can make infrastructure systems resil- for response and recovery measures. Several ient. Instead, governments need to define and instruments are available to do so, including implement a consistent strategy—in partnership reserve funds or budget reallocation, contin- with all stakeholders, such as utilities, investors, gent credit, or insurance or risk transfers. The business associations, and citizen organiza- choice of financial instruments is determined tions—to tackle the many obstacles to more by the risks that need to be covered, the cost of resilient infrastructure systems. One common the instrument, the speed of disbursement, and feature of these recommendations is a focus on the transparency and predictability of the the early stages of infrastructure system devel- resources (Clarke and Dercon 2016; World opment—the design of regulations, the produc- Bank 2017). After a disaster, however, the tion of hazards data and master plans, or the availability of financial resources is only half of initial stages of new infrastructure asset design. the story; just as important is the ability to These early stages are when small investments deliver resources effectively and rapidly to can significantly improve the overall resilience where they are needed, including to the firms of infrastructure systems and generate very large and households that are affected by infrastruc- benefits. In poor countries, however, mobilizing ture disruptions, even if they are not affected resources to invest in these actions may be chal- directly by the disaster. Financial instruments lenging, which makes targeted support from the therefore need to be combined with contin- international community necessary, transforma- gency plans and flexible delivery mechanisms— tional, and highly cost-effective. if possible, building on existing instruments, Although these recommendations are such as social protection systems. aimed at making infrastructure more resilient, OVERVIEW 21 TABLE O.3 Five recommendations to address the five obstacles to resilient infrastructure Recommendation Actions 1: Get the basics right 1.1: Introduce and enforce regulations, construction codes, and procurement rules 1.2: Create systems for appropriate infrastructure operation, maintenance, and postincident response 1.3: Provide appropriate funding and financing for infrastructure planning, construction, and maintenance 2: Build institutions for resilience 2.1: Implement a whole-of-government approach to resilient infrastructure, building on existing regulatory systems 2.2: Identify critical infrastructure and define acceptable and intolerable risk levels 2.3: Ensure equitable access to resilient infrastructure 3: Create regulations and incentives 3.1: Consider resilience objectives in master plans, standards, and regulations for resilience and adjust them regularly to account for climate change 3.2: Create economic incentives for service providers to offer resilient infrastructure assets and services 3.3: Ensure that infrastructure regulations are consistent with risk-informed land use plans and guide development toward safer areas 4: Improve decision making 4.1: Invest in freely accessible natural hazard and climate change data 4.2: Make robust decisions and minimize the potential for regret and catastrophic failures 4.3: Build the skills needed to use data and models and mobilize the know-how of the private sector 5: Provide financing 5.1: Provide adequate funding to include risk assessments in master plans and early project design 5.2: Develop a government-wide financial protection strategy and contingency plans 5.3: Promote transparency to better inform investors and decision makers most of them tackle market or government details, refer to chapter 2 and Rentschler, failures that are responsible not only for less Kornejew, et al. (2019). 3. The estimates summarized in this paragraph resilient infrastructure but also for less effi- cover up to 137 low- and middle-income cient, less inclusive, and costlier infrastructure. countries, although the exact country cover- As a result, taking these actions will contribute age varies across infrastructure sectors due to to more than infrastructure resilience and help data constraints. For details, refer to chapter create more productive, livable, and inclusive 3 and Obolensky et al. (2019). societies. REFERENCES NOTES Beck, M. W., I. J. Losada, P. Menéndez, B. G. 1. In this report, all dollar amounts are U.S. dol- Reguero, P. Díaz-Simal, and F. Fernández. lars, unless otherwise indicated. 2018. “The Global Flood Protection Savings 2. The data set covers 137 countries represent- Provided by Coral Reefs.” Nature Communi- ing 80 percent of the GDP of low- and middle- cations 9 (1): 2186. https://doi.org/10.1038 income countries, or 32 percent of global /s41467-018-04568-z. GDP. Due to data limitations, the exact coun- Braese, J., J. Rentschler, and S. Hallegatte. 2019. try coverage varies for different analyses. For “Resilient Infrastructure for Thriving Firms: A 22 LIFELINES Review of the Evidence.” Background paper Kalra, N. S. Hallegatte, R. Lempert, C. Brown, A. for this report, World Bank, Washington, DC. Fozzard, S. Gill, and A. Shah. 2014. “Agree- Browder, G., S. Ozment, I. Rehberger Bescos, ing on Robust Decisions: New Processes for T. Gartner, and G.-M. Lange. 2019. Integrat- Decision Making under Deep Uncertainty.” ing Green and Gray: Creating the Next Generation Policy Research Working Paper 6906, World Infrastructure. Washington, DC: World Bank Bank, Washington, DC. https://doi.org/doi:10 and World Resources Institute. .1596/1813-9450-6906. Christopher, M., and H. Peck. 2004. “Building the Koks, E., J. Rozenberg, C. Zorn, M. Tariverdi, M. Resilient Supply Chain.” International Journal Vousdoukas, S. A. Fraser, J. Hall, and S. Hal- of Logistics Management 15 (2): 1–14. https://doi legatte. 2019. “A Global Multi-Hazard Risk .org/10.1108/09574090410700275. Analysis of Road and Railway Infrastructure Clarke, D., and S. Dercon. 2016. Dull Disasters? Assets.” Forthcoming in Nature Sustainability. How Planning Ahead Will Make a Difference. Kornejew, M., J. Rentschler, and S. Hallegatte. Oxford: Oxford University Press. 2019. “Well Spent: How Governance Deter- Colon, C., S. Hallegatte, and J. Rozenberg. 2019. mines the Effectiveness of Infrastructure “Transportation and Supply Chain Resilience Investments.” Background paper for this in the United Republic of Tanzania.” Back- report, World Bank, Washington, DC. ground paper for this report, World Bank, Lenz, L., A. Munyehirwe, J. Peters, and M. Sievert. Washington, DC. 2017. “Does Large Scale Infrastructure Invest- Fay, M., I. L. Hyoung, H. Sungmin, M. Mastruzzi, ment Alleviate Poverty? Impacts of Rwanda’s and M. Cho. 2019. “Hitting the Trillion Mark: Electricity Access Roll-Out Program #555.” A Look at How Much Countries Are Spending World Development 89 (January): 88–110. on Infrastructure.” Policy Research Working Miyamoto International. 2019. “Overview of Paper 8730, World Bank, Washington, DC. Engineering Options for Increasing Infrastruc- Hallegatte, S. 2013. “Modeling the Roles of Het- ture Resilience.” Background paper for this erogeneity, Substitution, and Inventories in report, World Bank, Washington, DC. the Assessment of Natural Disaster Economic Munich Re. 2019. “Natural Catastrophe Review Costs.” Risk Analysis 34 (1): 152–67. https:// 2018.” Munich Re, January 8. https://www doi.org/10.1111/risa.12090. .munichre.com/en/media-relations/publica- Hallegatte, S., J. Rozenberg, C. Fox, C. Nicolas, tions/press-releases/2019/2019-01-08-press and J. Rentschler. 2019. “Strength­ ening New -release/index.html. Infrastructure Assets—A Cost-Benefit Analy- National Research Council. 2000. Watershed sis.” Background paper for this report, World Management for Potable Water Supply: Assessing Bank, Washington, DC. the New York City Strategy. Washington, DC: Hallegatte, S., and A. Vogt-Schilb. 2016. “Are National Academies Press. Losses from Natural Disasters More Than Just Nicolas, C., J. Rentschler, A. Potter van Loon, S. Asset Losses? The Role of Capital Aggregation, Oguah, S. Schweikert, M. Deinert, E. Koks, Sector Interactions, and Investment Behav- C. Arderne, D. Cubas, J. Li, and E. Ichikawa. iors.” Policy Research Working Paper 7885, 2019. “Stronger Power: Improving Power Sec- World Bank, Washington, DC. tor Resilience to Natural Disasters.” Sector note Hallegatte, S., A. Vogt-Schilb, M. Bangalore, and for this report, World Bank, Washington, DC J. Rozenberg. 2016. Unbreakable: Building the Nicholls, R. J., J. Hinkel, D. Lincke, and T. van Resilience of the Poor in the Face of Natural Disas- der Pol. 2019. “Global Investment Costs for ters. Washington, DC: World Bank. https://doi Coastal Defense through the 21st Century.” .org/10.1596/978-1-4648-1003-9. Policy Research Working Paper 8745, World Jeandron, A., J. M. Saidi, A. Kapama, M. Burhole, Bank, Washington. F. Birembano, T. Vandevelde, A. Gasparrini, B. Obolensky, M., A. Erman, J. Rozenberg, P. Avner, Armstrong, S. Cairncross, and J. H. J. Ensink. J. Rentschler, and S. Hallegatte. 2019. “Infra- 2015. “Water Supply Interruptions and Sus- structure Disruptions: Impacts on House- pected Cholera Incidence: A Time-Series holds.” Background paper for this report, Regression in the Democratic Republic of the World Bank, Washington, DC. Congo.” PLoS Medicine 12 (10): 1–16. https:// OECD (Organisation for Economic Co-operation doi.org/10.1371/journal.pmed.1001893. and Development). 2019. Good Governance for OVERVIEW 23 Critical Infrastructure Resilience. OECD Review of note for this report, World Bank, Washing- Risk Management Practices. Paris: OECD. ton, DC. Pardina, M. R., and J. Schiro. 2018. “Taking Stock Sheffi, Y. 2005. The Resilient Enterprise: Overcoming of Economic Regulation of Power Utilities in Vulnerability for Competitive Advantage. Cam- the Developing World: A Literature Review.” bridge, MA: MIT Press. Policy Research Working Paper 8461, World Stip, C., Z. Mao, L. Bonzanigo, G. Browder, and Bank, Washington, DC. J. Tracy. 2019. “Water Infrastructure Resil- Renn, O. 2008. “White Paper on Risk Gover- ience: Examples of Dams, Wastewater Treat- nance: Toward an Integrative Framework.” In ment Plants, and Water Supply and Sanitation Global Risk Governance, edited by O. Renn and Systems.” Sector note for this report, World K. D. Walker, 3–73. Stuttgart: Springer. Bank, Washington, DC. Rentschler, J., J. Braese, N. Jones, and P. Avner. Strahl, J., M. Bebrin, E. Paris, and D. Jones. 2016. 2019. “Three Feet Under: The Impacts of “Beyond the Buzzwords: Making the Specific Flooding on Urban Jobs, Connectivity, and Case for Community Resilience Microgrids.” Infrastructure.” Background paper for this ACEEE Summer Study on Energy Efficiency report, World Bank, Washington, DC. in Buildings, American Council for an Energy- Rentschler, J., M. Kornejew, S. Hallegatte, M. Efficient Economy, Washington, DC. Obolensky, and J. Braese. 2019 “Underuti- Swiss Re. 2019. “Preliminary Sigma Estimates for lized Potential: The Business Costs of Unreli- 2018: Global Insured Losses of USD 79 Billion able Infrastructure in Developing Countries.” Are Fourth Highest on Sigma Records.” Swiss Background paper for this report, World Bank, Re, December 18. https://www.swissre.com Washington, DC. /media/news-releases/nr_20181218_sigma_ Rentschler, J., M. Obolensky, and M. Kornejew. estimates_for_2018.html. 2019. “Candle in the Wind? Energy System Vallejo, L. and M. Mullan. 2017. Climate- Resilience to Natural Shocks.” Background Resilient Infrastructure: Getting the Policies paper for this report, World Bank, Washing- Right.” OECD Environment Working Papers, ton, DC. No. 121, OECD Publishing, Paris. Rose, A., G. Oladosu, and S. Y. Liao. 2007. “Busi- Walsh, B., and S. Hallegatte. 2019. “Measur- ness Interruption Impacts of a Terrorist Attack ing Natural Risks in the Philippines.” Policy on the Electric Power System of Los Ange- Research Working Paper 8723, World Bank, les: Customer Resilience to a Total Blackout.” Washington, DC. Risk Analysis 27 (3): 513–31. https://doi.org Wiener, J. B., and M. D. Rogers. 2002. “Com- /10.1111/j.1539-6924.2007.00912.x. paring Precaution in the United States and Rozenberg, J., and M. Fay. 2019. Beyond the Gap: Europe.” Journal of Risk Research 5 (4): 317–49. How Countries Can Afford the Infrastructure They World Bank. 2013. World Development Report 2014: Need While Protecting the Planet. Washington, Risk and Opportunity—Managing Risk for Devel- DC: World Bank. opment. Washington, DC: World Bank. https:// Rozenberg, J., X. Espinet Alegre, P. Avner, C. Fox, doi.org/10.1596/978-0-8213-9903-3. S. Hallegatte, E. Koks, J. Rentschler, and ———. 2017. Sovereign Catastrophe Risk Pools: World M. Tariverdi. 2019a. “From a Rocky Road to Bank Technical Contribution to the G20. Washing- Smooth Sailing: Building Transport Resilience ton, DC: World Bank. https://openknowledge to Natural Disasters.” Sector note for this .worldbank.org/handle/10986/28311. report, World Bank, Washington, DC. ———. 2018. Strategic Use of Climate Finance to Rozenberg, J., C. Fox, M. Tariverdi, E. Koks, and Maximize Climate Action. Washington, DC: S. Hallegatte. 2019b. “Road Show: Comparing World Bank. https://openknowledge.world Road Network Resilience around the World.” bank.org/handle/10986/30475. Background paper for this report, World Bank, Zhang, F. 2019. In the Dark: How Much Do Washington, DC. Power Sector Distortions Cost South Asia? South Sandhu, H. S., and S. Raja. 2019. “No Broken Asia Development Forum. Washington, DC: Link: The Vulnerability of Telecommunication World Bank. https://doi.org/doi:10.1596 Infrastructure to Natural Hazards.” Sector /978-1-4648-1154-8. Resilient Infrastructure: A Lifeline for Sustainable Development 1 I n Dar es Salaam, frequent urban flooding disrupts the city’s entire economy, even be- yond the directly affected flood zones. As roads become flooded, all traffic, including public transport, comes to a near standstill. People are unable to reach their workplaces, supply chains are interrupted, deliveries are missed, and sales are lost. The supply of electricity is often affected as well, resulting in power outages and halting economic ac- tivity. Because these incidents occur so frequently, businesses have to invest in expensive coping measures, ranging from buying diesel generators to keeping expensive backup inventories and contracting with backup suppliers. Overall, the lack of reliable transport and electricity systems is a defining factor of the urban economy of Dar es Salaam, influ- encing the investment and risk-taking behavior of everyone who lives and works there. But Dar es Salaam is by no means an excep- transport, water, electricity, and infrastructure tion. Cities and countries around the world are more generally, low- and middle-income coun- facing the challenging consequences of sub- tries tend to experience more disruptions and standard infrastructure, often at a significant have less reliable infrastructure than richer cost to people and firms. Worldwide, 940 mil- countries, with large differences across coun- lion people still lack access to modern electric- tries, even at similar income levels (figure 1.1). ity, let alone modern telecommunications ser- There is ample evidence that natural haz- vices; 2.1 billion have no access to safe drinking ards affect the functioning of infrastructure water; 4.5 billion lack adequate sanitation facil- systems in poor and rich countries alike. Floods ities; 1 billion live more than 2 kilometers away like those in Mozambique in 2019 destroy from an all-season road; and uncounted num- roads and isolate entire communities or regu- bers are unable to access work and educational larly paralyze public transit systems, as they do opportunities because transport services remain in Dar es Salam every rainy season. Earth- either unavailable or unaffordable. quakes like the one in San Francisco in 1989 Simply being connected to infrastructure damage bridges, slowing down local econo- networks does not guarantee reliable services. mies. Hurricanes wipe out electricity transmis- Many people and businesses experience fre- sion and distribution systems, cutting people’s quent power outages, intermittent water sup- access to electricity for months, as the 2017 ply, congested or regularly disrupted transport, storm in Puerto Rico did. Moreover, in the next or unreliable communication. In the areas of decades, many factors—including climate 25 26 LIFELINES FIGURE 1.1 Poorer countries experience more infrastructure disruptions a. Firms reporting major transport disruptions b. Transport infrastructure (quality index) 100 7 90 6 WEF Infrastructure Quality Index 80 70 5 60 % of firms 4 50 3 40 30 2 20 1 10 0 0 0 10,000 20,000 30,000 40,000 0 10,000 20,000 30,000 40,000 50,000 60,000 GDP per capita (US$) GDP per capita (US$) c. Number of electricity outages per month d. Electricity infrastructure (quality index) 30 7 25 6 WEF Infrastructure Quality Index 5 20 4 15 3 10 2 5 1 0 0 0 10,000 20,000 30,000 40,000 0 10,000 20,000 30,000 40,000 50,000 60,000 GDP per capita (US$) GDP per capita (US$) e. Number of water outages per month f. Overall infrastructure (quality index) 30 7 25 6 WEF Infrastructure Quality Index 5 20 4 15 3 10 2 5 1 0 0 0 10,000 20,000 30,000 40,000 0 10,000 20,000 30,000 40,000 50,000 60,000 GDP per capita (US$) GDP per capita (US$) Source: Rentschler et al. 2019, based on data from the World Bank’s Enterprise Surveys (panels a, c, and e), and the Infrastructure Quality Index in the Global Competitiveness Index of the World Economic Forum (panels b, d, and f). Note: Panel c only shows countries with up to 30 electricity outages a month. Eight countries (all with GDP per capita below $9,000) report between 30 and 95 outages a month. RESILIENT INFRASTRUCTURE: A LIFELINE FOR SUSTAINABLE DEVELOPMENT 27 BOX 1.1 Resilience is central to achieving many international objectives In the last decade, a series of international agree- Climate Change. It includes many objectives and ments have made resilience a central objective decisions for supporting more resilient devel- for development. opment. In particular, Article 7 establishes “the These agreements include the Sendai Dec- global goal on adaptation of enhancing adaptive laration and the Sendai Framework for Disaster capacity, strengthening resilience, and reducing Risk Reduction 2015–2030, which seeks “the vulnerability to climate change, with a view to substantial reduction of disaster risk and losses contributing to sustainable development.” in lives, livelihoods, and health and in the eco- The United Nations’ Sustainable Develop- nomic, physical, social, cultural, and environmen- ment Goals also relate to disaster risk. Target 1.5, tal assets of persons, businesses, communities, for example, aims “by 2030, [to] build the resil- and countries.” This goal is translated into mul- ience of the poor and those in vulnerable situa- tiple targets for 2030, such as reducing mortal- tions and reduce their exposure and vulnerabil- ity and direct economic loss in relation to global ity to climate-related extreme events and other GDP and increasing the availability of and access economic, social, and environmental shocks and to multihazard early warning systems and disas- disasters.” Target 13.1 aims to “strengthen resil- ter risk information and assessments.   ience and adaptive capacity to climate-related The Paris Agreement was approved in Decem- hazards and natural disasters in all countries.” ber 2015 at the 21st Conference of the Parties of The targets for food security and urban develop- the United Nations Framework Convention on ment are also relevant to reducing disaster risks. change, population and economic growth, and structure (Fay et al. 2019). Much more will be urbanization—will magnify many of these invested in the next decades. In low- and threats (IPCC 2012, 2014). middle-income countries alone, it is estimated When disasters affect infrastructure services, that new infrastructure could cost between 2 even the households and companies not percent and 8 percent of GDP a year to 2030, directly affected by the shocks experience or from $640 billion to $2.7 trillion a year impacts. People are sometimes left without (Rozenberg and Fay 2019). electricity or water for weeks or more. They are However, as countries continue to build also affected indirectly through impacts on their stock of infrastructure assets at a rapid businesses—such as reduced productivity and pace, they will have to emphasize the quality competitiveness—which in turn affect their of investments. Ensuring that infrastructure ability to provide the jobs, incomes, and goods services are reliable is critical for making large and services on which people depend. At the investments worthwhile. Infrastructure invest- macro level, infrastructure disruptions add to ments that do not meet these criteria are bound the already large impacts of natural disasters on to fail to deliver—not only financially but also people’s assets and livelihoods, thereby threat- in their ability to contribute to sustainable ening the achievement of many international socioeconomic development. objectives (box 1.1) (Hallegatte et al. 2016). How are low- and middle-income countries OBJECTIVES OF THIS REPORT responding? By one estimate, governments in This report explores the resilience of infrastruc- these countries are investing around $1 tril- ture—that is, the ability of infrastructure to lion—or between 3.4 percent and 5 percent of provide the services users need during and their gross domestic product (GDP)—in infra- after a natural shock. While natural hazards 28 LIFELINES are only one of the causes of infrastructure This report focuses primarily on four infra- disruptions, resilience is still an essential structure systems that are essential to eco- dimension of the overall reliability of infra- nomic activity and people’s well-being: structure systems. “Resilience” here is used in a broader sense • Power systems, including the generation, than the traditional definition in ecology, transmission, and distribution of electricity which refers to the ability of systems to recover • Water and sanitation systems, focusing on and bounce back. Boosting resilience, using water utilities (large-scale water storage, this broad definition, can be achieved in many hydropower, and irrigation systems are ways, including: considered, but are not the focus of the analysis) • Reducing the exposure of infrastructure • Transport systems, looking at multiple modes assets to natural hazards, such as by build- (including road, rail, waterways, and air- ing energy assets outside floodplains ports) and considering multiple scales • Reducing the vulnerability of assets, such (including urban transit and rural access) as by making roads able to cope with heavy • Telecommunications, including telephones precipitation or bridges able to resist strong and Internet connections. wind • Designing infrastructure systems so they These systems share two characteristics that are able to deliver services, even if some of make them worthy of in-depth investigation. their components have been damaged or First, they provide critical services for the destroyed well-being of households and the productivity • Ensuring that infrastructure systems do not of firms and are thus often referred to as “life- fail catastrophically, can recover quickly, lines” or “critical infrastructure systems.” Their and be repaired efficiently if damaged importance is clear over the short term, • Making the users of infrastructure services because they provide what are often consid- better able to cope with service disruptions, ered basic services, and over the long term, such as by installing batteries or generators because their reliability and quality are often in hospitals or ensuring that firms rely on considered a prerequisite for modern, produc- multiple suppliers. tive economies. Second, these systems—from transport sys- Building on a wide range of case studies, tems to electricity grids—are organized in net- global empirical analyses, and modeling exer- works, with direct implications for their vul- cises, the report explores how natural hazards nerability to natural hazards. For example, a and climate change reduce the reliability of localized shock can propagate very quickly infrastructure services—and in the process not through these networks, affecting households only diminish their socioeconomic and devel- or firms located even in a safe area. A network opment benefits but also make people and firms also creates some very specific challenges, in less resilient. The report also identifies technical that the vulnerability of one infrastructure solutions and policies for (a) building the resil- asset can be a poor proxy for the network’s ience of infrastructure services; (b) improving ability to perform during or after a shock. Only their reliability and quality; (c) strengthening a network-wide, system-wide view can pro- the resilience of infrastructure users; (d) maxi- vide a reliable assessment of vulnerability and mizing development benefits; and (e) reducing resilience, but doing so brings significant data the impacts of climate change. and methodological challenges. For this rea- RESILIENT INFRASTRUCTURE: A LIFELINE FOR SUSTAINABLE DEVELOPMENT 29 son, the report also describes existing or new more resilient infrastructure, including policy tools, data sets, and models, with examples of measures to ensure that the report’s suggested applications in specific case studies to demon- solutions can be implemented in practice. strate what is possible and to offer a toolbox not only for investigating infrastructure resil- REFERENCES ience but also for identifying the most promis- Fay, M., S. Han, H. I. Lee, M. Mastruzzi, and M. ing interventions. Cho. 2019. “Hitting the Trillion Mark: A Look at How Much Countries Are Spending on Infra- Other systems sometimes described as structure.” Policy Research Working Paper 8730, “infrastructure”—such as buildings, schools, or World Bank, Washington, DC. hospitals—are not discussed at length in this Hallegatte, S., A. Vogt-Schilb, M. Bangalore, and report. This is because they are not organized J. Rozenberg. 2016. Unbreakable: Building the primarily as networks, even though the lines Resilience of the Poor in the Face of Natural Disas- are sometimes blurred (for example, when an ters. Washington, DC: World Bank. https://doi .org/10.1596/978-1-4648-1003-9. ensemble of regional hospitals collaborates to IPCC (Intergovernmental Panel on Climate respond better to a shock). Change). 2012. Special Report on Managing the Risks of Extreme Events and Disasters to Advance Climate STRUCTURE OF THE REPORT Change Adaptation: Summary for Policymakers— This report is organized into three parts. Part I A Report of Working Groups I and II of the IPCC. establishes the scale of the problem, quantifies Geneva: IPCC. the total cost of infrastructure disruptions, and ———. 2014. Climate Change 2014: Synthesis Report: Contribution of Working Groups I, II, and III to the explores the role of natural hazards and cli- Fifth Assessment Report of the Intergovernmental Panel mate change in these disruptions. It also on Climate Change. Geneva: IPCC. demonstrates the adverse effects that a lack of Rentschler, J., M. Kornejew, S. Hallegatte, M. Obo- resilient infrastructure has on households and lensky, and J. Braese. 2019. “Underutilized firms, and how this lack may contribute to Potential: The Business Costs of Unreliable Infra- poverty and poor health. Part II identifies via- structure in Developing Countries.” Background paper for this report, World Bank, Washington, ble and affordable solutions to make infrastruc- DC. ture systems and their users more resilient and Rozenberg, J., and M. Fay. 2019. Beyond the Gap: How provides estimates of the costs and benefits of Countries Can Afford the Infrastructure They Need more resilient infrastructure systems. Part III While Protecting the Planet. Washington DC: World proposes concrete steps for the development of Bank. INFRASTRUCTURE DISRUPTIONS ARE A BARRIER TO THRIVING FIRMS 31 PA RT A Diagnosis: I A Lack of Resilient Infrastructure Is Harming People and Firms H ow are natural shocks, infra- structure systems, economic activities, and human well-being FIGURE PI.1 A framework for analyzing how natural shocks affect people and firms through their impact on infrastructure systems interconnected? To visualize these Natural shocks interactions, figure PI.1 provides a framework for part I of this report. It illustrates the channels through which natural hazards and shocks eventually affect people and firms via their impacts on infrastructure systems. The direct impacts of nat- ural shocks on firms and people Infrastructure are an important topic that has been discussed in past studies (for Power Water example, Hallegatte et al. 2016). This report focuses on how nat- ural shocks affect infrastructure Tele- systems, which in turn affect firms Transport communicatons and people. Part I begins with chapters 2 and 3, which analyze the effects and costs of unreliable infrastructure on firms and Firms People households—whether disruptions are provoked by natural shocks or other causes, such as technical failures (the white arrows in figure PI.1). Both chap- ters explore the high cost of infrastruc- ture disruptions on people, either di- rectly through their impacts on health Note: The dashed arrows represent the direct impacts of natural hazards on firms and people, effects that are treated elsewhere—see, for example, Hallegatte et al. (2016). and well-being or indirectly through This report focuses on the impact of natural shocks on infrastructure systems (red their impacts on firms, jobs, and income. arrow) and how this impact, in turn, affects firms and people (white arrows). 32 LIFELINES Then chapter 4 examines the impact of natural shocks—floods, storms, earth- quakes, and droughts—on infrastructure systems (the red arrow in figure PI.1). It shows that natural hazards account for a significant fraction of infrastructure service disruptions, at least in electricity and transport, and create large needs for recon- struction. Finally, chapter 5 provides a review of evidence from household and firm surveys and modeling exercises, which detail how infrastructure system disruptions and damages magnify the macroeconomic cost of natural disasters and the impacts on people’s well-being. REFERENCE Hallegatte, S., A. Vogt-Schilb, M. Bangalore, and J. Rozenberg. 2016. Unbreakable: Building the Resilience of the Poor in the Face of Natural Disasters. Washington, DC: World Bank. https://doi.org/10.1596/978-1-4648-1003-9. Infrastructure Disruptions Are a Barrier to Thriving Firms 2 L ack of access to infrastructure services can have severe economic consequences. Just as bad, however, can be the lack of reliability in infrastructure services: being con- nected to the electricity grid is of little use if the power is out. This chapter offers an overview of the real costs that infrastructure disruptions impose on firms and the result- ing damages at the macro level. Unreliable infrastructure systems affect firms ness of an industry. For example, a firm is through three key channels: less likely to upgrade its machinery to more productive technology if frequent black- • Direct impacts. These impacts are the most outs force it to revert regularly to manual visible, immediate consequences of infra- production. structure disruptions. For example, a firm relying on water to cool a machine cannot Together, these impacts take a big toll on an manufacture products during a drought; economy’s ability to generate wealth and likewise, a restaurant with an electric stove maintain international competitiveness. But cannot cook meals without power. Infra­ how high is this toll? What are the real costs structure disruptions interrupt firms’ activ- that infrastructure disruptions impose on firms, ities, force them to operate at less than full and what are the resulting damages at the production capacity, reduce their sales, and macroeconomic level? cause delays in the supply and delivery of This chapter addresses these questions by goods. looking at the impacts of disruptions in key • Coping costs. For example, a backup power infrastructure sectors (table 2.1). It presents generator reduces the direct impacts of estimates of the monetary costs of outages blackouts but has high operating costs and based on a set of microdata for more than requires an up-front purchase that prohibits 143,000 firms from the World Bank’s Enter- alternative, more productive investments. prise Surveys. This data set covers 137 coun- • Indirect impacts. These impacts are less vis- tries, representing 78 percent of the world’s ible and less immediate; they affect firms’ population and 80 percent of the gross domes- investment decisions, influence what prod- tic product (GDP) of low- and middle-income ucts can and cannot be produced, and countries. However, the various estimates influence the composition and innovative- reported in this chapter often use subsets of 33 34 LIFELINES TABLE 2.1 Disrupted infrastructure services have multiple impacts on firms Sector Direct impacts Coping costs Indirect impacts Power • Reduced utilization rates • Generator investment ($6 billion ($38 billion a year) a year) • Higher barriers to market entry and • Sales losses ($82 billion a year) • Generator operation costs lower investment ($59 billion a year) • Less competition and innovation Water • Reduced utilization rates • Investment in alternative water due to lack of small and new firms ($6 billion a year) sources (reservoirs, wells) • Sales losses • Bias toward labor-intensive production Transport • Reduced utilization rates • Increased inventory ($107 billion a year) • More expensive location choices • Inability to provide on-demand • Sales losses in proximity to, for example, services and goods • Delayed supplies and deliveries clients or ports • Diminished competitiveness in international markets Telecommunications • Reduced utilization rates • Expensive location choices close • Sales losses to fast Internet Source: Rentschler, Kornejew, et al. 2019. Note: Bolded are the impact channels for which original estimates are presented in this section, based on the World Bank’s Enterprise Surveys of 143,000 firms in 137 countries, representing 78 percent of the world’s population and 80 percent of the GDP of low- and middle-income countries. the full sample, due to missing data in some economies. In a prominent paper on the “com- countries. petitive advantage of nations,” Porter (1990) The analysis, detailed in a technical back- argues that the ability of a country to host ground study for this report by Rentschler, high-performing firms is supported by a wide Kornejew, et al. (2019), estimates that annual range of factors—including the availability of losses due to disruptions are substantial in low- reliable and efficient infrastructure systems. and middle-income countries (summarized in The importance of infrastructure for eco- table 2.1). Utilization rate losses due to power, nomic growth has been confirmed by a wide water, and transport disruptions amount to range of studies, reviewed by Braese, Rent- $151 billion a year, sales losses from electricity schler, and Hallegatte (2019). For example, outages amount to $82 billion a year, and the Calderón and Servén (2014) review the theo- additional costs of self-generating electricity retical and empirical literature on infrastruc- amount to $65 billion a year. At a total cost of ture and growth and conclude that, overall, the about $300 billion a year, these figures high- literature finds that infrastructure development light the significance of unreliable infrastruc- has positive effects on income growth and even ture. These are lower-bound estimates of the distributive equity. Bom and Ligthart (2014) global costs of outages because neither all conduct a meta regression of 68 quantitative countries nor all impact channels are covered studies, predominantly in high-income econo- by this analysis. To address these gaps, this mies, to quantify the impact of public infra- chapter also relies on examples from the exten- structure capital on GDP. Their assessment sug- sive literature on this topic (Braese, Rentschler, gests that on average a 1 percent increase in and Hallegatte 2019). public infrastructure capital is associated with a 0.1 percent increase in GDP. INFRASTRUCTURE SERVICES The same positive impact of infrastructure ENABLE FIRMS TO THRIVE investments emerges from studies investigating The availability of infrastructure systems is a individual countries at different income levels. key factor of production that determines the A prominent study by Aschauer (1989) finds competitiveness of firms and thus of entire that public investment in U.S. infrastructure INFRASTRUCTURE DISRUPTIONS ARE A BARRIER TO THRIVING FIRMS 35 has a significant positive effect on total factor internationally, the disruptions and lack of reli- productivity (TFP). In particular, investments ability have significant adverse impacts on the in “core” infrastructure—such as transport, performance of firms. electricity, gas, water, and sanitation—have the strongest explanatory power for productivity. INFRASTRUCTURE DISRUPTIONS In a 30-year-long panel of South African man- HAVE DIRECT AND REAL COSTS ufacturing firms, Fedderke and Bogeti´ c (2009) FOR FIRMS find that investments in different types of Frequent disruptions of electricity, water, or transport, telecommunications, and power transport infrastructure often mean that firms generation infrastructure have positive and sig- are unable to utilize all of their available pro- nificant impacts on measures of productivity, duction capacity. Capacity utilization is a com- output, and growth. mon measure of the effectiveness with which a Many more studies focus on the firm-level firm converts its resources into output. A firm benefits of the four critical infrastructure sec- that is frequently forced to halt production— tors studied in this report. Electricity infrastruc- for example, because of power outages or input ture has been shown to benefit both small shortages caused by transport disruptions or enterprises and industrial firms. Evidence from upstream production stops—will be operating Indonesia and South Africa shows that electri- below its full capacity. fication resulted in increased employment This section presents estimates of the (especially among women), incentivized the impacts of electricity, water, and transport dis- formation of new small and medium firms, ruptions using a pooled data set of firms from and enhanced productivity (Dinkelman 2011; low- and middle-income countries. This data Kassem 2018). Transport infrastructure has set is based on the World Bank’s Enterprise been found to yield similar benefits by creating Surveys, which provide harmonized firm-level employment, increasing productivity, lowering data on the operating conditions experienced production costs, and allowing firms to reduce by businesses worldwide. As part of the survey, inventory holdings (Duranton and Turner firms report on their capacity utilization rate 2012; Ghani, Goswami, and Kerr 2016; Gib- and on the quality of water and electricity bons et al. 2017; Volpe Martincus and Blyde infrastructure, including the average monthly 2012; Wan and Zhang 2018). Information and frequency and duration of service disruptions. communications technology infrastructure has Firms also report transport disruptions using a also been shown to generate growth through subjective ordinal scale. These data allow higher productivity and innovation. For exploration of how infrastructure disruptions instance, in an analysis of 45 countries in affect firms’ performance, controlling for a Sub-Saharan Africa from 1990 to 2014, Albi- range of other factors.1 (Unfortunately, the sur- man and Sulong (2016) find significant posi- vey does not include information on disrup- tive effects of mobile phones, the Internet, and tions in telecommunications.) telephone lines on economic growth. As with The results show that the annual losses due other types of infrastructure, the underlying to disruptions in low- and middle-income effect channels are increases in firm-level pro- countries are substantial (table 2.1). For the ductivity and innovation activities (Paunov 118 countries for which data are available, and Rollo 2015, 2016; Polák 2017). unreliable power, water, and transport infra- Overall, this rich evidence base highlights structure leads to utilization losses of $151 bil- why infrastructure disruptions are so detri- lion a year, which is equivalent to 0.59 percent mental to firms. As firms rely on infrastructure of the sample GDP (see map 2.1).2 These utili- services to operate effectively and compete zation rate losses can be separated into the 36 LIFELINES MAP 2.1 Firms in low- and middle-income countries are incurring high utilization rate losses due to infrastructure disruptions Source: Rentschler, Kornejew, et al. 2019. Note: Map shows countrywide average utilization rate losses from electricity, water, and transport infrastructure disruptions. three types of infrastructure covered by the countries. This persistence of transport losses in Enterprise Surveys (figure 2.1). The results richer countries and areas can explain their show that most utilization losses are caused by large contribution to the overall loss figure and disruptions in transport infrastructure, account­ suggests that the damage that unreliable trans- ing for losses of $107 billion annually, or 0.42 port infrastructure inflicts on an economy is percent of sample GDP. Disruptions in the elec- hard to eliminate. tricity supply account for $38 billion, and water Of course, firms are vulnerable not only to disruptions cause utilization rate losses of $6 disruptions that directly affect their facilities, billion a year. but also to disruptions in their wider region. Strikingly, some countries, especially For example, even if a firm is not directly low-incomes ones, face very high utilization affected by infrastructure breakdowns (for rate losses from unreliable power and water example, by experiencing a power outage infrastructure (figure 2.2, panels a and b). By onsite), it may still be forced to stop production contrast, most middle-income countries are as interruptions along the supply chain bring barely affected because of their much more input supply or output demand to a halt. reliable power and water systems. Transport To assess this issue, Rentschler, Kornejew, et disruptions show a different pattern. Although al. (2019) estimate how the utilization rates of poorer countries still tend to incur higher individual firms are affected indirectly by infra- transport-related utilization losses, the losses structure disruptions, proxied by region-level remain significant even for middle-income (instead of firm-level) disruptions. The results INFRASTRUCTURE DISRUPTIONS ARE A BARRIER TO THRIVING FIRMS 37 FIGURE 2.1 For all critical infrastructure sectors, poorer countries experience the highest utilization rate losses due to disruptions a. Electricity infrastructure b. Water infrastructure c. Transport infrastructure 2.0 2.0 2.0 1.5 1.5 1.5 Utilization rate loss (%) Utilization rate loss (%) Utilization rate loss (%) 1.0 1.0 1.0 0.5 0.5 0.5 0 0 0 0 10,000 20,000 30,000 40,000 0 10,000 20,000 30,000 40,000 0 10,000 20,000 30,000 40,000 GDP per capita (US$) GDP per capita (US$) GDP per capita (US$) Source: Rentschler, Kornejew, et al. 2019. Note: Data points represent 118 countries. FIGURE 2.2 In the most affected countries, utilization rate losses are a significant share of GDP Top 15 countries with greatest utilization rate losses, by type of infrastructure disruption a. Electricity infrastructure disruptions b. Water infrastructure disruptions c. Transport infrastructure disruptions Nepal Guatemala Kenya Congo, Dem. Rep. Paraguay West Bank and Gaza Niger El Salvador Congo, Rep. Dominican Republic South Sudan Tanzania Benin Solomon Islands Malawi Rwanda Ethiopia Iraq Gabon Sierra Leone Sudan Burundi Rwanda Pakistan Pakistan Niger Lesotho Gambia, The Jordan Costa Rica Yemen, Rep. Afghanistan Afghanistan Iraq Nicaragua Rwanda Nigeria Ghana Guinea Lebanon Côte d'Ivoire Papua New Guinea Congo, Rep. Congo, Rep. Burkina Faso 0 2 4 6 0 2 4 6 0 0.5 1.0 1.5 2.0 Utilization rate loss (%) Utilization rate loss (%) Utilization rate loss (%) Source: Rentschler, Kornejew, et al. 2019. suggest that the regional effects of infrastruc- ant than, the direct firm-level impacts. In fact, ture disruptions are significant. For water and for water disruptions (and depending on model transport, but not power, the regional effects specifications), the indirect losses can exceed may be as important as, or even more import- the direct losses by a ratio of 3 to 1. Although 38 LIFELINES MAP 2.2 Power outages are causing large sales losses in low- and middle-income countries, especially in Africa Source: Rentschler, Kornejew, et al. 2019. Note: Map shows average sales losses reported by firms, as country-level averages. regional disruption levels provide only an With regard to sales, the World Bank’s imperfect measure of indirect effects, it is evi- Enterprise Surveys also collect self-reported dent that considering just the direct impacts of data on sales losses due to power outages infrastructure disruptions misses a large part of for more than 80,000 firms from 122 mostly the economic cost. low- and middle-income countries (although the same data are not available for transport FIGURE 2.3 More frequent power outages tend to result in and water supply disruptions). The data show larger sales losses that power outages in these countries are 35 causing sales losses of $81.6 billion a year Share of sales lost due to power outages (%) Estimates based on Enterprise Surveys (map 2.2)—or more than twice the value of 30 Estimates based on literature capacity utilization losses (table 2.1). And 25 because a significant fraction of firms (about 20 15 percent) did not report their sales, the sales loss figure is likely to be a conservative esti- 15 mate. These results correlate closely with 10 those from other studies (see Braese, Rent- schler, and Hallegatte 2019), which find that 5 firms located in countries with frequent 0 power outages tend to incur large sales losses. 0.1 1 10 100 1,000 Monthly hours without power (log scale) Power outages are particularly hard on sales if Source: Rentschler, Kornejew, et al. 2019, based on the World Bank’s Enterprise they average more than 10 hours a month Surveys. (figure 2.3). INFRASTRUCTURE DISRUPTIONS ARE A BARRIER TO THRIVING FIRMS 39 Across countries, firms exhibit varying FIGURE 2.4 Size of sales losses depends on more than the capacities to deal with electricity disruptions. length of outages Take the case of the 15 countries for which Hours firms on average report the highest shares of 0 50 100 150 200 250 300 sales lost from outages. As figure 2.4 shows, Mali although all of these countries suffer significant Uganda power outages of more than 10 hours a month, Cameroon at this high level of outages, the sales losses are Nepal no longer clearly related to electricity down- Sierra Leone time. This finding indicates that the relation- Gambia, The ship is mediated by other factors that deter- Zambia mine firms’ vulnerability to electricity network Madagascar disruptions, such as the sectoral distribution of Angola firms, competition, and the energy intensity of Tanzania production. Moreover, the extent to which firms are affected by power outages is deter- Nigeria mined by their coping strategies, which are dis- Ghana cussed later in this chapter. Yemen, Rep. Central African Water supply infrastructure also plays an Republic important role in production. In agriculture, Pakistan the relationship between water availability— 0 5 10 15 20 25 30 determined by weather and irrigation technol- Percent ogy—and agricultural production is clearly  Power outage duration in average month (top axis) established (Damania et al. 2017). Iimi (2011)  Sales lost due to power outages (bottom axis) finds that if all water supply disruptions could Source: Rentschler, Kornejew, et al. 2019. be halted in Europe and Central Asia, firms Note: Figure shows the top 15 countries with the largest estimated sales losses due to power outages. would on average be able to reduce their costs by 0.5 percent. This effect would likely be sig- nificantly larger in low- and middle-income feres with their access to these production fac- countries with less reliable water infrastruc- tors (Weisbrod, Vary, and Treyz 2003). Sweet ture. Indeed, Islam and Hyland (2018), using (2013) finds that in 88 U.S. metropolitan areas, Enterprise Survey data for 103 countries, find a 1 percent increase in congestion—measured that water supply disruptions have adverse by daily traffic per freeway lane—not only impacts on firms in low- and lower-middle- affects economic growth but also leads to a income countries, but not in upper-middle- decrease in productivity growth per worker of and higher-income countries. In the first up to 0.033 percent. Using a panel similar in group, an additional water outage incident geographic scope, Jin and Rafferty (2017) find would lead to sales losses of about 8.2 percent that an increase in congestion growth of 1 per- for the average manufacturing firm. cent—here measured by an index of traffic Traffic congestion also causes significant delays—causes a decrease in employment economic losses and has been shown to have a growth of 0.08 percent. negative effect on economic growth (Sweet Traffic disruptions and congestion have neg- 2011). The evidence suggests that firms reliant ative productivity effects in low- and middle- on high-skilled labor, specialized inputs, and income countries as well. Based on a survey of geographically distributed markets are espe- commuters in Kumasi, Ghana, congestion has cially sensitive to congestion because it inter- been estimated to result in an average loss of 40 LIFELINES daily productive hours of 9 percent per worker from the loss of international Internet traffic. In (Harriet, Poku, and Emmanuel 2013). In the addition, the loss of an Internet connection in Greater Cairo Metropolitan Area in the Arab Australia would cut off the Internet connection Republic of Egypt, traffic congestion was in Papua New Guinea. By contrast, in Canada, accounting for direct costs of $5.1 billion a year the economic costs would be zero because alter- as of 2010, a number that is only expected to native overland connectivity is available to the increase (World Bank 2013). For a range of United States (APEC 2013). Sub-Saharan African cities, Rentschler, Braese, et al. (2019) show that urban flooding can be FIRMS EMPLOY COSTLY MEASURES an important driver of disrupted traffic flows, TO COPE WITH UNRELIABILITY thus reducing the connectivity between firms and supply chains (chapter 5). Self-generation of electricity Telecommunications and the Internet have Firms that are, or expect to be, heavily affected become essential to many types of economic by infrastructure disruptions can take measures activities, and telecommunications outages can to minimize the impact on their operations. present firms with large costs. Although the Although these actions reduce the costs of an global annual costs are not available, many additional disruptive event, they come with high-visibility events show the magnitude of their own costs as well. Such coping costs, the potential impacts, especially in businesses which can take various forms, are yet another that operate in real time and rely on data or aspect of the effects of unreliable infrastructure. online sales. For example, when Delta Airlines Self-generating electricity is a ubiquitous experienced a five-hour interruption in one of albeit costly strategy to adapt to frequent power its Atlanta data centers in 2016, some 2,000 outages. Facing frequent electricity outages, flights were grounded over the course of three firms often choose to operate their own backup days, costing the company an estimated $150 generator, usually powered by diesel. These million (Sverdlik 2016). Small events can also generators enable firms to bridge power out- be costly. Based on a survey of 49 organizations ages, but they also require firms to purchase, in 16 sectors, the Ponemon Institute (2016) install, maintain, and operate costly machin- has estimated the average cost of a data center ery. Generators tend, then, to be less affordable outage at more than $700,000, with the high- for smaller firms with limited cash reserves. As est cost reaching more than $2 million. It is no a result, generator ownership is significantly surprise that vulnerabilities are the highest in higher among large firms (figure 2.5, panel a) financial services, telecommunication services, for a panel of firms in low- and middle-income health care, and e-commerce. countries and in countries with an unreliable Internet disruptions affect not only individual electricity supply (figure 2.5, panel b). companies but also entire countries. In fact, for In addition to high up-front investments, many countries, their entire access to the Inter- operational costs also make self-generation sig- net depends on one or two submarine cables nificantly more expensive than conventional that are vulnerable to both natural and human- grid supply (Adenikinju 2003; Farquharson, made hazards, ranging from earthquakes to Jaramillo, and Samaras 2018). For example, fishing equipment and attacks. Disruptions of Steinbuks and Foster (2010) find for 25 African these cables or associated landing stations can be countries that self-generation is on average very expensive. For example, a fault in all land- three times more expensive than national elec- ing points in Australia would entail a direct cost tricity tariffs. (for cable repair) estimated at $2.2 million, and An analysis conducted for this report esti- an indirect economic cost of $3.2 billion, mostly mates the installed self-generation capacity and INFRASTRUCTURE DISRUPTIONS ARE A BARRIER TO THRIVING FIRMS 41 FIGURE 2.5 Generator ownership is more common for large firms and in countries with many power outages a. Generator ownership, by firm size b. Generator ownership, by outage duration 40 100 Share of firms owning a generator (%) Share of firms owning a generator (%) 35 90 80 30 70 25 60 20 50 15 40 30 10 20 5 10 0 0 1 2 3 4 5 6 7 8 9 10 0 20 40 60 80 100 Number of employees, by decile Average power outage duration per month (hours) Source: Rentschler, Kornejew, et al. 2019. the total annual cost of self-generation in the investments in other input factors (Mensah industrial sectors of 129 low- and middle-in- 2016). Furthermore, backup generation using come countries for which the required data are diesel generators significantly increases emis- available (Rentschler, Kornejew, et al. 2019). sions of air pollutants such as fine inhalable These estimates yield the costs of backup elec- particulates and carbon dioxide, thereby pro- tricity generation, accounting for both the ducing indirect costs in the form of health annualized up-front investment and the opera- impacts or climate change (Farquharson, Jara- tional costs. millo, and Samaras 2018). Overall, the estimates suggest that the costs of backup generation are substantial.3 Total Measures to conserve and reuse water up-front investments in backup generation Many firms rely on a dependable water supply amount to about $6 billion a year in low- and for their operations. When outages occur, firms middle-income countries. The annual operating can take measures to cope and decrease their costs of generators are estimated to add around reliance on the usual supply, often at signifi- $59 billion a year to total electricity costs for cant costs. Efforts to conserve water, such as firms in low- and middle-income countries. through automated control systems, allow Thus, because power is unreliable, firms firms to use less water in their operations and in the industrial sector spend an estimated also reduce stress on the water network (Rose additional $65 billion a year on backup self- and Krausmann 2013). The recycling and generation, corresponding to 0.28 percent of reuse of water within a firm can serve similar GDP of the 129 countries considered in this purposes. Such options appear to make sense analysis—with Africa bearing the highest costs even in the absence of disruptions, but the low (map 2.3). cost of water may make them uneconomical This large sum, however, does not capture until unreliability reaches a high level. These the full opportunity costs incurred by firms measures lessen the impact of disruptions on operating power generators. Although genera- firms, but do not make them independent of tors can mitigate short-term losses, they are the water grid and will not help in case of sus- also linked to lower longer-term productivity tained outages. Then, more costly options that because of the higher marginal costs that limit eliminate firms’ reliance on the water grid can 42 LIFELINES MAP 2.3 Additional costs of backup electricity generation are substantial in low- and middle-income countries Source: Rentschler, Kornejew, et al. 2019. Note: Map shows the cost of backup electricity generation as a percentage of GDP, including up-front investments and additional operating costs. be used, including adopting technologies that regions with poor infrastructure quality are give access to underground, river, or lake water bound to be less attractive to businesses, which or installing water storage facilities (Kajitani has implications for local economic activity and and Tatano 2009). employment. Having to increase an inventory to shield it Adapting to transport disruptions from low-quality transport infrastructure or Many studies in rich and poor countries show transport disruptions is costly for a firm. This that transport infrastructure has a significant adaptation measure comes with significant impact on firms’ location choices (see Arauzo- coping costs in the form of the opportunity Carod, Liviano-Solis, and Manjón-Antolín costs of capital bound in the inventory, the 2010). Other types of infrastructure affect loca- costs of storage, and the possible depreciation tion choices as well. For example, Kim and of stored goods. In a cross-country analysis, a Cho (2017) find that in the rural United States, decrease in an infrastructure quality indicator the availability of broadband connectivity sig- of 1 standard deviation increases raw material nificantly increases the chance that a firm will inventories by 11–37 percent (Guasch and choose a rural location. As a result of this influ- Kogan 2003). The importance of this effect is ence of infrastructure reliability on location also evident at the micro level in East Africa, as choice, firms may incur higher costs for real confirmed in an analysis of firms in Burundi, estate or face other difficulties such as lack of Kenya, Rwanda, Tanzania, and Uganda (Iimi, proximity to the labor force. In addition, Humphrey, and Melibaeva 2015). INFRASTRUCTURE DISRUPTIONS ARE A BARRIER TO THRIVING FIRMS 43 UNRELIABLE INFRASTRUCTURE FIGURE 2.6 Increased power outages result in lower firm LEADS TO LOWER PRODUCTIVITY productivity in African countries Clearly, the indirect impacts of infrastructure 4.0 7,000 disruptions are harder to quantify than the GDP per capita (current US$) 3.5 6,000 Productivity decrease (%) direct effects just described. It is not easy to 3.0 5,000 observe the influence of unreliable infrastruc- 2.5 4,000 ture on firm behavior and industry dynamics at 2.0 3,000 one moment in time because that influence 1.5 1.0 2,000 continually and often subtly alters firms’ deci- 0.5 1,000 sions. Nevertheless, ample evidence exists of its 0.0 0 importance, especially for firm productivity. An a o, Uga o ia Ni la Gh a Za a Ke a Ta oon Bu wa i a m a i m a l aw M . a ep ric ri an rk nd De nd bi Ca ny s an go M Fa Although the analysis for this report considers ge m al .R Af er nz a in h R ut power, water, and transport infrastructure, So ng most of the literature revolves around the indi- Co  GDP per capita (right scale) rect impacts of disruptions in electricity supply.  Decrease in productivity resulting from 1 percent increase in outages In Africa and Asia, power outages have been Source: Mensah 2016. shown to affect firms’ productivity significantly. Note: The left scale (dots) shows the percentage decrease in productivity resulting from a 1 percent increase in outages. The right scale (bars) shows GDP per capita of A study based on a firm panel of 23 African the countries analyzed (in current US$). countries estimates that a 1 percent increase in electricity outages would account for a loss in firms’ TFP of 3.5 percent on average (Mensah ity intensity adopt costly coping mechanisms to 2018). A similar study for 14 countries in a lesser degree and are therefore harder hit by Sub-Saharan Africa estimates that a 1 percent disruptions, whereas firms with high power increase in electricity outages would result in intensity do invest in adaptation measures, but productivity losses of between 1 and 3.5 per- they also must contend with the high costs of cent (figure 2.6; Mensah 2016). Considering self-generation, thus leading to greater losses absolute rather than relative effects, Bbaale (Gurara and Tessema 2018; Ramachandran, (2018) shows that one additional power outage Shah, and Moss 2018). in a typical month reduces productivity by The burden of unreliable infrastructure ser- 0.1–0.2 percentage point on average in 26 Afri- vices is particularly large for small firms, which can countries. Similarly, Zhang (2019) finds often are the economic foundation of people’s that manufacturing firms in Bangladesh would livelihoods in low- and middle-income coun- suffer TFP losses of 3–4 percent from an tries. For them, it is harder to deal with the increase in load shedding by 10 percent. In higher operational costs resulting from outages India, electricity deficits decrease the TFP of because they have weaker financial security or manufacturing firms by about 2 percent (All- less diversified income sources. Small firms in cott, Collard-Wexler, and O’Connell 2016). India, for example, are disproportionately The complexity of indirect impacts is high- affected by outages, and so they face produc- lighted by the relationship between the impact tion costs that are higher by 0.29 percent of of electricity disruptions and intensity of power revenue for every percentage point increase in usage. Disaggregating the impact of outages on electricity shortages (Zhang 2019). Such higher productivity shows that firms with very low operational costs then affect firms’ investment and very high power usage intensities suffer decisions and their productivity. In Indonesia, the most productivity losses from electricity the negative effect of electricity unreliability on disruptions. Intuitively, firms with low electric- firm productivity is more than 50 percent larger 44 LIFELINES for smaller manufacturing firms than for bigger Bank 2019). The inefficiency costs borne by ones (Poczter 2017). Furthermore, unreliable individual firms translate into disadvantages power networks can drastically increase the ini- for entire sectors and economies. Higher costs, tial investments required to start a business. In lower productivity, a lack of entrepreneurship Nigeria, small firms have to spend between 10 and innovation, and the absence of high- and 30 percent of their start-up costs on power growth sectors all negatively affect the chances self-generation (Adenikinju 2003, 2008). of a country finding success in increasingly Such large start-up costs and the prospect of internationalized markets. disproportionally high operational costs have Eventually, the impacts of outages on firms dire consequences for entrepreneurship. An are passed on to people—workers and con- analysis of Enterprise Survey data for 23 Afri- sumers—in the form of loss of income or can countries finds that power outages dimin- well-being. Workers may carry the burden ish the probability that individuals will start through lower employment and lower wages. their own business by 32 percent, an effect that A study of 23 African countries estimates that a rises to 44 percent when considering only the 1 percentage point increase in outages reduces nonfarm sector (Mensah 2018). This lack of the employment of low-skilled workers by 1.1 start-ups reduces competition and leads to effi- percent and of high-skilled workers by 0.35 ciency losses. Alby, Dethier, and Straub (2013) percent (World Bank 2019). For a sample of 21 analyze Enterprise Survey data for 77 countries countries in Africa, Mensah (2018) finds that and find that energy-intensive sectors such as living in a community with frequent electricity the chemical and textile industries have a sig- outages reduces the probability of being nificantly lower share of small firms in coun- employed by 35–41 percent on average. Rent- tries with frequent outages. schler and Kornejew (2017) find that when The unreliability of infrastructure can also manufacturing firms in Indonesia lack access to reduce firm efficiency and lead to inefficient reliable electricity, they switch to less efficient resource allocation at the national level. fuels and pass the higher prices down their Because most new technology relies on elec- supply chains to consumers and other firms. tricity, power outages reduce the adoption of And households are also affected directly by innovative means of production. Capital is thus infrastructure disruptions, as discussed in the directed toward more labor-intensive opera- next chapter. tions, which can be less productive. As a result, national economies are stuck in inefficient sec- NOTES toral allocations and miss out on certain high- 1. For the full methodology of this and the fol- growth sectors. For example, the provision of lowing analyses, see the work on firms and on-demand goods and services is complicated infrastructure published as a background paper or not possible because of unreliable infrastruc- for this report by Rentschler, Kornejew, et al. ture, and low power reliability makes it impos- (2019). 2. All monetary estimates have been converted to sible for countries to host large data centers real 2018 U.S. dollars. (World Bank 2019). 3. These estimates are based on a review of the The overall effect of these inefficiencies literature on self-generation, which suggests from unreliable infrastructure can also mean an annualized capital cost approximation of that firms must struggle to compete in interna- $0.032 per kilowatt-hour of self-generated elec- tricity (ESMAP 2007) and an approximate price tional markets. In 23 African countries, a 1 markup factor of 2 over the national electricity percentage point increase in power outage fre- tariff (Steinbuks and Foster 2010). Estimates of quency reduces the average firm’s share of self-generated electricity in the industrial sec- sales from exports by 0.12 percent (World tor are based on estimates from the Enterprise INFRASTRUCTURE DISRUPTIONS ARE A BARRIER TO THRIVING FIRMS 45 Surveys of electricity consumption by the nomic Surveys 28 (5): 889–916. https://doi.org industrial sector and self-reported shares of /10.1111/joes.12037. backup generation. See Rentschler, Kornejew, Braese, J., J. Rentschler, and S. Hallegatte. 2019. et al. (2019) for details. “Resilient Infrastructure for Thriving Firms: A Review of the Evidence.” Background paper for REFERENCES this report, World Bank, Washington, DC. Adenikinju, A. F. 2003. “Electric Infrastructure Fail- Calderón, C., and L. Servén. 2014. “Infrastructure, ures in Nigeria: A Survey-Based Analysis of the Growth, and Inequality: An Overview.” Policy Costs and Adjustment Responses.” Energy Policy Research Working Paper 7034, World Bank, 31 (14): 1519–30. https://doi.org/https://doi Washington, DC. .org/10.1016/S0301-4215(02)00208-2. Damania, R., S. Desbureaux, M. Hyland, A. Islam, S. ———. 2008. “Efficiency of the Energy Sector and Moore, A.-S. Rodella, J. Russ, and E. Zaveri. 2017. Its Impact on the Competitiveness of the Nigerian Uncharted Waters: The New Economics of Water Scar- Economy.” International Association for Energy Eco- city and Variability. Washington, DC: World Bank. nomics (Fourth Quarter): 27–31. https://doi.org/10.1596/978-1-4648-1179-1. Albiman, M. M., and Z. Sulong. 2016. “The Role of Dinkelman, T. 2011. “The Effects of Rural Electrifica- ICT Use to the Economic Growth in Sub-Saharan tion on Employment: New Evidence from South African Region (SSA).” Journal of Science and Tech- Africa.” American Economic Review 101 (7): 3078– nology Policy Management 7 (3): 306–29. https:// 108. https://doi.org/10.1257/aer.101.7.3078. doi.org/10.1108/JSTPM-06-2016-0010. Duranton, G., and M. A. Turner. 2012. “Urban Alby, P., J.-J. Dethier, and S. Straub. 2013. “Firms Growth and Transportation.” Review of Economic Operating under Electricity Constraints in Devel- Studies 79 (4): 1407–40. https://doi.org/10.1093 oping Countries.” World Bank Economic Review 27 /restud/rds010. (1): 109–32. https://doi.org/10.1093/wber ESMAP (Energy Sector Management Assistance Pro- /lhs018. gram). 2007. Technical and Economic Assessment of Allcott, H., A. Collard-Wexler, and S. D. O’Connell. Off-Grid, Mini-Grid, and Grid Electrification Technolo- 2016. “How Do Electricity Shortages Affect gies. Washington DC: World Bank. Industry? Evidence from India.” American Eco- Farquharson, D., P. Jaramillo, and C. Samaras. 2018. nomic Review 106 (3): 587–624. http://www “Sustainability Implications of Electricity Outages .aeaweb.org/articles?id=10.1257/aer.20140389. in Sub-Saharan Africa.” Nature Sustainability 1 APEC (Asia-Pacific Economic Cooperation). 2013. (10): 589–97. https://doi.org/10.1038/s41893 “Economic Impact of Submarine Cable Dis­ -018-0151-8. ruptions.” Report APEC#213-SE-01.2, APEC, Fedderke, J. W., and Ž. Bogeti´ c. 2009. “Infrastruc- S ingapore. http://publications.apec.org ture and Growth in South Africa: Direct and Indi- /Publications/2013/02/Economic-Impact-of rect Productivity Impacts of 19 Infrastructure -Submarine-Cable-Disruptions. Measures.” World Development 37 (9): 1522–39. Arauzo-Carod, J.-M., D. Liviano-Solis, and M. https://doi.org/10.1016/J.WORLDDEV.2009 Manjón-Antolín. 2010. “Empirical Studies in .01.008. Industrial Location: An Assessment of Their Ghani, E., A. G. Goswami, and W. R. Kerr. 2016. Methods and Results.” Journal of Regional Science “Highway to Success: The Impact of the Golden 50 (3): 685–711. https://doi.org/10.1111/j.1467 Quadrilateral Project for the Location and Perfor- -9787.2009.00625.x. mance of Indian Manufacturing.” Economic Jour- Aschauer, D. A. 1989. “Is Public Expenditure Pro- nal 126 (591): 317–57. https://doi.org/10.1111 ductive?” Journal of Monetary Economics 23 (2): /ecoj.12207. 177–200. https://doi.org/10.1016/0304-3932 Gibbons, S., T. Lyytik, H. Overman, and R. Sanchis- (89)90047-0. Guarner. 2017. “New Road Infrastructure: The Bbaale, E. 2018. “Infrastructure Quality and Firm Effects on Firms.” Journal of Urban Economics 110 Productivity in Africa.” World Journal of Entrepre- (March): 35–50. neurship, Management, and Sustainable Development Guasch, J. L., and J. Kogan. 2003. “Just-in-Case 14 (4): 367–84. https://doi.org/10.1108 Inventories: A Cross-Country Analysis.” Policy /WJEMSD-11-2017-0091. Research Working Paper 3012, World Bank, Bom, P. R. D., and J. E. Ligthart. 2014. “What Have Washington, DC. We Learned from Three Decades of Research on Gurara, D., and D. Tessema. 2018. “Losing to Black- the Productivity of Public Capital?” Journal of Eco- outs: Evidence from Firm Level Data.” IMF 46 LIFELINES Working Paper 18(159), International Monetary 1): S192–S204. https://doi.org/10.1093/wber Fund, Washington, DC. https://doi.org/10.1177 /lhv010. /00220345740530061001. ———. 2016. “Has the Internet Fostered Inclusive Harriet, T., K. Poku, and A. K. Emmanuel. 2013. “An Innovation in the Developing World?” World Assessment of Traffic Congestion and Its Effect on Development 78 (February): 587–609. https://doi Productivity in Urban Ghana.” International Jour- .org/https://doi.org/10.1016/j.worlddev.2015 nal of Business and Social Science 4 (3): 10–20. .10.029. Iimi, A. 2011. “Effects of Improving Infrastructure Poczter, S. 2017. “You Can’t Count on Me: The Quality on Business Costs: Evidence from Firm- Impact of Electricity Unreliability on Productivity.” Level Data in Eastern Europe and Central Asia.” Agricultural and Resource Economics Review 46 (3): Developing Economies 49 (2): 121–47. https://doi 579–602. https://doi.org/10.1017/age.2016.38. .org/10.1111/j.1746-1049.2011.00126.x. Polák, P. 2017. “The Productivity Paradox: A Iimi, A., R. M. Humphrey, and S. Melibaeva. 2015. Meta-Analysis.” Information Economics and Policy “Firm Inventory Behavior in East Africa.” Policy 38 (March): 38–54. https://doi.org/https://doi. Research Working Paper 7280, World Bank, org/10.1016/j.infoecopol.2016.11.003. Washington, DC. https://doi.org/10.1596/1813 Ponemon Institute. 2016. “Cost of Data Center Out- -9450-7280. ages.” Data Center Performance Bench­ marking Islam, A., and M. Hyland. 2018. “The Drivers and Series, Traverse City, MI. Impacts of Water Infrastructure Reliability: A Porter, M. 1990. “The Competitive Advantage of Global Analysis of Manufacturing Firms.” Work- Nations.” Harvard Business Review 90 (2, March- ing Paper 8637, World Bank, Washington, DC. April): 73–95. http://www.worldbank.org/research. Ramachandran, V., M. K. Shah, and T. Moss. 2018. Jin, J., and P. Rafferty. 2017. “Does Congestion Neg- “How Do African Firms Respond to Unreliable atively Affect Income Growth and Employment Power? Exploring Firm Heterogeneity Using Growth? Empirical Evidence from US Metropoli- K-Means Clustering.” CGD Working Paper 493, tan Regions.” Transport Policy 55 (C): 1–8. https:// Center for Global Development, Washington, DC. doi.org/https://doi.org/10.1016/j.tranpol.2016 https://www.cgdev.org/publication/how-do .12.003. -african-firms-respond-unreliable-power Kajitani, Y., and H. Tatano. 2009. “Estimation of Life- -exploring-firm-heterogeneity-using-k-means. line Resilience Factors Based on Surveys of Japa- Rentschler, J., J. Braese, N. Jones, and P. Avner. nese Industries.” Earthquake Spectra 25 (4): 755– 2019. “Three Feet Under: The Impacts of Flood- 76. https://doi.org/10.1193/1.3240354. ing on Urban Jobs, Connectivity, and Infrastruc- Kassem, D. 2018. “Does Electrification Cause Indus- ture.” Background paper for this report, World trial Development? Grid Expansion and Firm Bank, Washington, DC. Turnover in Indonesia.” Job Market Paper, Rentschler, J., and M. Kornejew. 2017. “Energy Department of Economics and STIRCED, London Price Variation and Competitiveness: Firm Level School of Economics, London, August. Evidence from Indonesia.” Energy Economics 67 Kim, K., and Y. Cho. 2017. “Estimation of Power (C): 242–54. https://doi.org/https://doi.org/10 Outage Costs in the Industrial Sector of South .1016/j.eneco.2017.08.015. Korea.” Energy Policy 101 (C): 236–45. https://doi Rentschler, J., M. Kornejew, S. Hallegatte, M. Obo- .org/10.1016/J.ENPOL.2016.11.048. lensky, and J. Braese. 2019. “Underutilized Poten- Mensah, J. T. 2016. “Bring Back Our Light: Power tial: The Business Costs of Unreliable Infrastruc- Outages and Industrial Performance in Sub- ture in Developing Countries.” Background paper Saharan Africa.” Paper presented at the Annual for this report, World Bank, Washington, DC. Meeting of the Agricultural and Applied Eco- Rose, A., and E. Krausmann. 2013. “An Economic nomics Association, Boston, July 31–August 2. Framework for the Development of a Resilience ———. 2018. “Jobs! Electricity Shortages and Index for Business Recovery.” International Jour- Unemployment in Africa.” Policy Research nal of Disaster Risk Reduction 5 (September): 73–83. Working Paper WPS 8415, World Bank, Wash- https://doi.org/https://doi.org/10.1016/j.ijdrr ington, DC (April). .2013.08.003. Paunov, C., and V. Rollo. 2015. “Overcoming Obsta- Steinbuks, J., and V. Foster. 2010. “When Do Firms cles: The Internet’s Contribution to Firm Devel- Generate? Evidence on In-House Electricity Sup- opment.” World Bank Economic Review 29 (Suppl. ply in Africa.” Energy Economics 32 (3): 505–14. INFRASTRUCTURE DISRUPTIONS ARE A BARRIER TO THRIVING FIRMS 47 https://doi.org/https://doi.org/10.1016/j.eneco Wan, G., and Y. Zhang. 2018. “The Direct and Indi- .2009.10.012. rect Effects of Infrastructure on Firm Productivity: Sverdlik, Yevgeniy. 2016. “Delta: Data Center Out- Evidence from Chinese Manufacturing.” China age Cost Us $150M.” DataCenterKnowledge, Sep- Economic Review 49 (C): 143–53. https://doi.org tember 8. /https://doi.org/10.1016/j.chieco.2017.04.010. Sweet, M. 2011. “Does Traffic Congestion Slow the Weisbrod, G., D. Vary, and G. Treyz. 2003. “Measur- Economy?” Journal of Planning Literature 26 (4): ing Economic Costs of Urban Traffic Congestion 391–404. https://doi.org/10.1177/0885412211 to Business.” Transportation Research Record 1839 409754. (1) : 98–106. https://doi.org/10.3141/1839-10. ———. 2013. “Traffic Congestion’s Economic World Bank. 2013. Cairo Traffic Congestion Study: Final Impacts: Evidence from US Metropolitan Report. Washington, DC: World Bank. Regions.” Urban Studies 51 (10): 2088–110. ———. 2019. Electricity Uptake for Economic Transfor- https://doi.org/10.1177/0042098013505883. mation in Sub-Saharan Africa. Africa Development Volpe Martincus, C., and J. S. Blyde. 2012. Shaky Forum. Washington, DC: World Bank. Roads and Trembling Exports: Assessing the Trade Zhang, F. 2019. In the Dark: How Much Do Power Sector Effects of Domestic Infrastructure Using a Natural Distortions Cost South Asia? South Asia Develop- Experiment. Washington, DC: Inter-American ment Forum. Washington, DC: World Bank. Development Bank. http://publications.iadb.org https://doi.org/doi:10.1596/978-1-4648-1154-8. /handle/11319/4650. Infrastructure Disruptions Affect the Health and Well-Being of Households 3 I nfrastructure disruptions and lack of reliability affect people as workers and consum- ers. But they can also impose direct costs on households through a variety of channels: direct impacts, coping costs, and indirect impacts. Each disruption can have real adverse impacts, including the direct short-term consequence of not having access to electricity, safe water, transport, or communication. For example, power outages can affect cooling and heating (which in turn may have health implications), economic activities and in- come, children’s educational outcomes, social and leisure activities, and regular house- hold tasks such as cooking and cleaning (World Bank 2019). Some of the negative consequences of unreli- global cost is difficult. This chapter sheds more able service may materialize only in the long light on the issue by looking at the impacts of term as a result of the prolonged high fre- disruptions in the power, water, transport, and quency of disruptions. For example, house- telecommunications sectors (table 3.1). It finds holds may decide not to invest in food refriger- that the willingness to pay for power outages ation or air-conditioning. In these cases, ranges between 0.002 percent and 0.15 per- individual outages may not carry a large cost cent of gross domestic product (GDP) per year because households give up on some types of for low- and middle-income countries (corre- energy use, but the long-term costs may be sponding to between $2.3 billion and $190 bil- substantial. Households also may be forced to lion). For water, the range is between 0.11 per- invest in expensive measures to mitigate the cent and 0.19 percent of their GDP per year impact of outages, such as a diesel generator (corresponding to between $88 billion and for a backup electricity supply, a water reser- $153 billion). voir for a backup water supply, or a vehicle to compensate for inadequate public transit. INFRASTRUCTURE PROVIDES Moreover, money spent on backup capacity HOUSEHOLDS WITH ESSENTIAL will not be available for more productive SERVICES investments that could help people to escape Infrastructure services not only help households poverty or grow a business. to meet their most basic needs but also enhance Infrastructure disruptions have many their quality of life in many ways. Indeed, many impacts on households, and estimating the studies have documented the extent to which 49 50 LIFELINES TABLE 3.1 Disrupted infrastructure services have multiple impacts on households Sector Direct impacts Coping costs Indirect and health impacts Power • Diminished well-being • Generator investments • Higher mortality and morbidity (lack of access • Lower productivity of family • Generator operation costs to health care, air-conditioning during heat firms waves, or heat during cold spells) Willingness to pay to prevent outages: between $2.3 billion and $190 billion a year Water • Diminished well-being and • Investment in alternative • Higher incidence of diarrhea, cholera, and loss of time water sources (reservoirs, other diseases wells, water bottles) Willingness to pay to prevent outages: between $88 billion Medical costs and missed income: between and $153 billion a year $3 billion and $6 billion a year Transport • Greater congestion and loss • Higher cost of alternative • Air pollution and health impacts of time transport modes • Constrained access to jobs, markets, services • Higher fuel costs • People forced to live close to jobs, possibly on bad land Telecommunications • Diminished well-being • Inability to call emergency services Source: Based on Obolensky et al. 2019. Note: The bolded terms in this table are the impact channels for which original estimates are presented in this section. Values are based on willing- ness-to-pay estimates in a few countries, applied to water and power outages from the World Bank’s Enterprise Surveys, covering 143,000 firms in 137 low- and middle-income countries. households rely on infrastructure services, as access to improved sanitation facilities was reviewed in detail by Obolensky et al. (2019). found to increase the average height-for-age of Electrification, for instance, has been shown children (Poder and He 2011). to facilitate entrepreneurship, education, and Studies also show that more efficient and female empowerment. Not only does it extend reliable transportation infrastructure reduces the length of an active day through lighting, travel times and transport costs (BenYishay but it also can free up time, especially for and Tunstall 2011). This reduction in time and women who can afford labor-saving electric costs, in turn, improves access to schools and appliances, and have positive impacts on health hospitals in rural areas and can raise productiv- through refrigeration and the replacement of ity and income (Levy 2004). Reduced transport polluting kerosene lamps. Moreover, electrifi- time and costs also enable workers to access cation may help to alleviate poverty because more distant employment opportunities (Gan- the poorest bear the largest opportunity costs non and Liu 1997) and stimulate economic of not being electrified (Samad and Zhang activity by increasing regional and interre- 2016; Zhang 2019). gional trade (Roberts et al. 2018; Volpe Martin- Water and sanitation infrastructure has cus and Blyde 2012). been shown to be particularly critical for good While infrastructure services benefit all peo- health. Access to in-house water and sanitation ple in modern economies, they can also increase services reduces the risk of exposure to germs the inclusion of disadvantaged population and the time households spend collecting groups, especially women. When modern infra- water and accessing public toilets. For exam- structure does not exist, women often have to ple, in India, the incidence of diarrhea in chil- perform time-consuming tasks at the expense of dren was found to be 21 percent lower for their education and livelihoods. For instance, households with access to piped water (Jalan without a centralized water supply, women and Ravallion 2003). The public health bene- carry the burden of collecting water from wells fits of improved access to sanitation facilities in 72 percent of cases (Birch 2011). As for sani- also grow in the long term. In Guatemala, tation, women living in poorly served settle- INFRASTRUCTURE DISRUPTIONS AFFECT THE HEALTH AND WELL-BEING OF HOUSEHOLDS 51 ments are typically responsible for disposing of FIGURE 3.1 Power outages hurt the well-being of households human waste or accompanying children to toi- 40 37.0 let facilities (Chant 2007). Furthermore, it is 35 31.2 now widely reported in a range of settings that 30 28.0 women and girls are at particular risk of attack 25 24.2 % of change 23.0 in and around toilet facilities located some dis- 21.1 20 tance from their homes (Cornman-Levy et al. 17.1 16.7 2011; McIlwaine 2013; Sommer et al. 2015). 15 13.8 11.7 Overall, the literature provides ample evi- 10 9.6 dence for why the well-being and livelihoods 5.8 6.5 5 of households depend so critically on the avail- 2.3 2.0 0 ability of quality infrastructure services. This Per capita Girls’ study Women’s Per capita Girls’ study Women’s Per capita Women’s income time labor force income time labor force income labor force evidence also explains why a lack of resilience participation participation participation and reliability of infrastructure services has Bangladesh India Pakistan direct adverse effects on the well-being of  Increased access  Increased reliable access households. Source: Zhang 2019. Note: Estimation is based on household surveys in Bangladesh, India, and Pakistan. The effects of electrification on girls’ study POWER OUTAGES DIRECTLY time and the effects of power outages on women’s labor force REDUCE THE WELL-BEING OF participation in Pakistan are not estimated because the data are not available. HOUSEHOLDS In the long run, an unreliable electricity supply has negative effects on household welfare. Fre- impacts that stem from loss of refrigeration quent outages limit households’ ability to (leading to food-borne diseases and vaccine engage in productive, educational, and recre- spoilage, among other things), heat, and higher ational activities during nighttime hours (Lenz levels of air pollution due to emissions from et al. 2017). Access to reliable electricity can backup power generation (Farquharson, Jara- help to mitigate inequality and promote social millo, and Samaras 2018). inclusion. An unreliable power network The total cost of outages has different com- increases the time needed for domestic work, ponents, the importance of which depends on mainly performed by women, and largely the context. In Pakistan, the total annual cost reduces the benefits from being connected to of outages for households adds up to 6.7 per- electricity networks (figure 3.1). In South Asia, cent of a household’s annual expenditures Zhang (2019) finds that long power outages (Pasha and Saleem 2013). The largest source of are associated with a decrease in women’s this cost is self-generation, making up 56 per- labor force participation. The persistence of cent of the total cost. Other costs include loss of electricity outages can constrain efforts toward well-being and forgone economic activity due economic transformation by reducing opportu- to outages, each of which accounts for 22 per- nities in nonagricultural sectors. cent of the total cost. Disaggregating by income Poor-quality electricity networks also affect levels reveals a very different picture: for public health. During extreme weather events, poorer households, monetization of utility loss power outages are common, and they affect makes up the largest source of losses—44 per- health by making it more difficult to access cent—because these households usually can- health care and maintain frontline services. not afford self-generation. After Hurricane Maria hit Puerto Rico, the dif- An analysis done for this report offers an ficulty in accessing health care was one of the estimate of the total well-being cost of power main causes of indirect deaths (Kishore et al. outages in low- and middle-income countries 2018). Power outages also cause indirect health (Obolensky et al. 2019). It suggests that the 52 LIFELINES FIGURE 3.2 There is large variation in people’s willingness to ruptions cause germs to settle in the water, pay to avoid one hour without power which increases the risk of the spread of water- 600 borne diseases. Even though these pathogens do not have a strong effect on mortality, they 500 are significant factors in morbidity. Bivins et al. Share of hourly GDP per capita (%) Canada, 1989 (2017) estimate that an intermittent water 400 supply causes several million infections and Italy, 2005 diarrhea cases every year in all parts of the 300 world, especially in South Asia and the West- Spain, 2013 ern Pacific (figure 3.3). Moreover, the impacts 200 Austria, 2013 of intermittent water supply are particularly significant in poor households because of their 100 US, 2004 higher dependency on tap water for their own Pakistan, 2004 Turkey, 2016 consumption (Ercumen et al. 2015; Jeandron Mexico, 1999 Sweden, 2008 0 et al. 2015; Nygård et al. 2007). 0 10,000 20,000 30,000 40,000 50,000 Case studies of specific water disruptions GDP per capita (US$, PPP) find consistently that households experiencing  Choice experiment  Macroeconomics  Contingent valuation water disruption and low water pressure are Source: Obolensky et al. 2019. more at risk of contracting diarrhea (figure Note: A country-year is matched to the closest nonmissing value of GDP per capita. See Obolensky et al. (2019) for additional details on the willingness-to-pay estimates. 3.4). For instance, several studies have docu- mented widespread diarrhea outbreaks, caused by cholera and Escherichia coli infections, in cost of power outages for households is the aftermath of floods (Ahern et al. 2005; between 0.002 percent and 0.15 percent of Qadri et al. 2005). GDP a year for 137 countries, which corre- sponds to between $2.3 billion and $190 bil- FIGURE 3.3 Intermittent water supply poses lion a year. This estimate is based on several major health risks in regions around the world studies that calculate the willingness to pay of 35 households to prevent power outages. This Annual infections range is so large because of uncertainty regard- Annual diarrheal cases 30 ing the willingness to pay to prevent power outages (figure 3.2). In fact, this value depends 25 on a variety of parameters—wealth of the 20 Millions respondents, quality of the power network, and timing and length of outages. It also 15 depends on the methodology, with contingent 10 valuation methodologies leading to higher esti- mates than choice experiments. 5 0 PEOPLE’S HEALTH AND an s a pe ia c al ica cifi ric As ob ne ro WELL-BEING SUFFER WHEN THE Af er Pa Gl Eu st rra Am Ea rn ite WATER SUPPLY IS UNRELIABLE te d ed an es M W h n ut Worldwide, 925 million people have an inter- er So st Ea mittent water supply—almost half in Southeast Source: Adapted from Bivins et al. 2017. Asia—with tremendous impacts on health, as Note: This figure shows the impact of an intermittent water supply documented by Bivins et al. (2017). Water dis- on health. The black lines represent 95 percent confidence intervals. INFRASTRUCTURE DISRUPTIONS AFFECT THE HEALTH AND WELL-BEING OF HOUSEHOLDS 53 Inadequate drainage systems aggravate the FIGURE 3.4 Water disruptions are linked with higher diarrheal situation, especially in overcrowded neighbor- risk hoods. In Dar es Salaam, water tends to stag- 6 nate and inundate neighborhoods during the 5 rainy season. Hospital records show that the incidence of waterborne illnesses increases sig- 4 Risk ratio nificantly during the rainy months when floods 3 are common, and this effect is stronger in 2.4 neighborhoods with a higher flood risk and 2 1.8 1.58 1.49 1.35 1.33 1.31 poor infrastructure (Picarelli, Jaupart, and Chen 1.08 1 2017). Dwellings situated downstream are the worst affected because sanitary waste overflows 0 2 4 6 7 8 15 15 06 when rains cause flooding. Cholera, fungus, 00 00 00 00 00 20 20 4– ,2 ,2 ,2 ,2 ,2 a, s, 00 ay ico za za za e di skin infections, and diarrhea are a common at ,2 rw Ga In Ga Ga ex St na No M d hi consequence for members of these households. ite ,C Un an The economic cost of the waterborne dis- iw a, am Ta ab eases caused by an intermittent water supply is Al difficult to determine, but it can be estimated Source: Adapted from Bivins et al. 2017. by combining the number of cases of illness Note: This figure compares the risk of contracting diarrhea in households with an intermittent water supply to the risk in households with a reliable water supply. To caused by an intermittent water supply (figure illustrate, in Mexico, a household with an intermittent water supply is 1.8 times more 3.3) with the estimated costs of treatment plus at risk of contracting diarrhea than a household with a reliable supply. the estimated costs associated with the loss of productive work for the sick or the caregiver.1 The financial cost is between $3 billion and $6 0.11 percent and 0.19 percent of GDP for 123 billion a year for the low- and middle-income countries, which corresponds to $88 billion countries covered in this analysis (Obolensky and $153 billion a year, respectively. Here, the et al. 2019). This relatively limited value stems uncertainty is probably larger than what is sug- from the low incomes of the people being gested by this range because available assess- affected and does not take into account how ments of the willingness to pay to improve being sick affects well-being. It should there- water distribution services have been con- fore be considered an underestimate. ducted only in high-income countries, where When a central water supply is disrupted, water-related health issues are less prevalent. people have no choice but to rely on alterna- tive sources of water, which can be 10–100 TRANSPORT DISRUPTIONS LEAD TO times more expensive than piped water (Kjel- LOST TIME, INCOME, AND ACCESS len 2000; UN-Habitat 2003). In most cities, TO SERVICES people have to rely on water kiosks, street ven- Transport disruptions are costly for households dors, or tanker trucks. Some households may because they give rise to longer travel times, be able to use their own well, but energy for wasted fuel, and missed work opportunities. In pumping can be expensive. In addition to these 2013 drivers in British, French, German, and monetary costs are the value of the time spent U.S. metropolitan areas spent on average 36 fetching water and the fact that such tasks are hours in gridlock (Cebr 2014). The time lost to usually performed by women, reinforcing gen- congestion increases threefold, to 111 hours, der inequality.2 when additional planning time is included.3 Willingness-to-pay estimates suggest a total According to these estimates, congestion across well-being cost for water outages of between Germany, the United Kingdom, and the United 54 LIFELINES States cost almost $450 billion in 2016, or $971 10-year flood—a flood that occurs on average per capita (INRIX Research 2018). every 10 years—disruptions of the road net- Capital cities in low- and middle-income work mean that travel times are significantly countries suffer the most from traffic disrup- longer. A common rule of thumb in emer- tions and congestion because roads and public gency responses is that the survival rate for transit systems in those cities have not kept life-threatening health incidents drops signifi- pace with population growth. In Thailand, cantly 60 minutes after an incident—the drivers lose an average of 56 hours a year to so-called golden hour (Campbell 2017). Road congestion at peak travel times. Indonesia and disruptions from a 10-year flood would mean Colombia are second and third, with 51 and 49 that, for residents of about a third of Inner hours, respectively (Cebr 2014). Kampala, travel times to a hospital would Transport disruptions can become life-and- exceed the golden hour.4 death issues when they affect people’s ability to reach hospitals and health facilities quickly. In sum, infrastructure disruptions are found Based on a network analysis, Rentschler, to affect households, both indirectly through Braese, et al. (2019) estimate that the mean their effects on firms and consequences on jobs travel time to a hospital from nearly all loca- and income and directly through people’s health tions in Inner Kampala is less than 30 minutes and well-being. Reducing these disruptions by car (figure 3.5). However, in the case of a should therefore be a policy priority, which in FIGURE 3.5 Transport disruptions can become life-and-death issues a. Mean travel times from locations in all b. Increases in travel times from locations across a. Travel time from locations across Inner Kampala b. Increase in travel time from locations across Inner Kampala of Inner Kampala to health care facilities Inner Kampala to hospitals in a 10-year flood to health care facilities to health care facilities in a 10-year flood Frequency density 0 10 20 30 40 50 60 Increase in travel time (%)  0–27  10-year flood extent Minutes  27–36  Bodies of water No flood  36–47 Area of analysis 10-year flood  47–70 Roads 50-year flood  > 70 Trips no longer possible Source: Rentschler, Braese, et al. 2019. Note: This figure shows the average travel times from inner Kampala to health care facilities during the different flood scenarios. In panel a, the vertical line denotes the “golden hour” (the window of time that maximizes survival of a major health emergency), assuming that ambulances complete a return trip starting at a hospital. Curves show frequency densities that represent the distribu- tion of travel times from all locations. INFRASTRUCTURE DISRUPTIONS AFFECT THE HEALTH AND WELL-BEING OF HOUSEHOLDS 55 turn requires a better understanding of their Research, Cebr, London. https://www.ibtta.org causes. Of particular interest for this report is /sites/default/files/documents/MAF/Costs-of- the role of natural hazards in causing these dis- Congestion-INRIX-Cebr-Report %283%29.pdf. Chant, S. 2007. Gender, Cities, and the Millennium ruptions, which is the topic of the next chapter. Development Goals in the Global South. London: London School of Economics. http://www.lse.ac NOTES .uk/collections/GenderInstitute. 1. Assuming that a diarrheal disease leads to Cornman-Levy, D., G. R. Dyrness, J. Golden, D. between four and seven days of loss of pro- Gouverneur, and J. A. Grisso. 2011. Transforming ductive work for the sick or the caregiver and Urban Environments. Women’s Health and the that treatment costs are between $2 and $4 World’s Cities. Philadelphia: University of Pennsyl- (Rozenberg and Hallegatte 2015). vania Press. 2. Data on the time spent to fetch water are usu- Ercumen, A., B. F. Arnold, E. Kumpel, Z. Burt, I. ally for rural households with no piped connec- Ray, K. Nelson, and J. M. Colford Jr. 2015. tion. No estimate could be identified of the time “Upgrading a Piped Water Supply from Intermit- needed for connected households that experi- tent to Continuous Delivery and Association with ence a water supply outage. Waterborne Illness: A Matched Cohort Study in 3. Planning time is the time lost due to uncertainty Urban India.” PLoS Medicine 12 (10): e1001892. in travel speed because drivers have to leave https://doi.org/10.1371/journal.pmed.1001892. earlier to make sure they arrive on time (here, Farquharson, D., P. Jaramillo, and C. Samaras. 2018. at least 95 percent of the time). “Sustainability Implications of Electricity Outages 4. Assuming that ambulances are based at hospi- in Sub-Saharan Africa.” Nature Sustainability 1 tals and one-way travel time is at least 30 min- (10): 589–97. https://doi.org/10.1038/s41893 utes, a round trip would exceed the 60-minute -018-0151-8. threshold. Gannon, C. A., and Z. Liu. 1997. Poverty and Trans- port. Washington, DC: World Bank. INRIX Research. 2018. “INRIX Global Traffic Score- REFERENCES card.” INRIX Global Traffic Scorecard. Kirkland, Ahern, M., R. S. Kovats, P. Wilkinson, R. Few, and WA, February. https://media.bizj.us/view F. Matthies. 2005. “Global Health Impacts of /img/10360454/inrix2016trafficscorecarden Floods: Epidemiologic Evidence.” Epidemiologic .pdf. Reviews 27 (1): 36–46. https://doi.org/10.1093 Jalan, J., and M. Ravallion. 2003. “Does Piped Water /epirev/mxi004. Reduce Diarrhea for Children in Rural India?” BenYishay, A., and R. Tunstall. 2011. “Impact Evalu- Journal of Econometrics 112 (1): 153–73. ation of Infrastructure Investments: The Experi- Jeandron, A., J. M. Saidi, A. Kapama, M. Burhole, ence of the Millennium Challenge Corporation.” F. Birembano, T. Vandevelde, A. Gasparrini, B. Journal of Development Effectiveness 3 (1): 103–30. Armstrong, S. Cairncross, and J. H. J. Ensink. https://doi.org/10.1080/19439342.2010.545892. 2015. “Water Supply Interruptions and Suspected Birch, E. L. 2011. Design of Healthy Cities for Women: Cholera Incidence: A Time-Series Regression in Women’s Health and the World’s Cities. Philadelphia: the Democratic Republic of the Congo.” PLoS University of Pennsylvania Press. Medicine 12 (10): 1–16. https://doi.org/10.1371 Bivins, A. W., T. Sumner, E. Kumpel, G. Howard, O. /journal.pmed.1001893. Cumming, I. Ross, K. Nelson, and J. Brown. Kishore, N., D. Marqués, A. Mahmud, M. V. Kiang, I. 2017. “Estimating Infection Risks and the Global Rodriguez, A. Fuller, P. Ebner, C. Sorensen, F. Burden of Diarrheal Disease Attributable to Inter- Racy, J. Lemery, L. Maas, J. Leaning, R. Irizarry, mittent Water Supply Using QMRA.” Environ- S. Balsari, and C. O. Buckee. 2018. “Mortality in mental Science and Technology 51 (13): 7542–51. Puerto Rico after Hurricane Maria.” New England https://doi.org/10.1021/acs.est.7b01014. Journal of Medicine 379 (2): 162–70. https://doi Campbell, J. 2017. International Trauma Life Support .org/10.1056/NEJMsa1803972. for Emergency Care Providers, Global Edition. New Kjellen, M. 2000. “Complementary Water Systems York: Pearson Education Limited. in Dar es Salaam, Tanzania: The Case of Water Cebr (Centre for Economics and Business Research). Vending.” International Journal of Water Resources 2014. “The Future Economic and Environmental Development 16 (1): 143–54. https://doi.org/10 Costs of Gridlock in 2030.” Report for INRIX .1080/07900620048626. 56 LIFELINES Lenz, L., A. Munyehirwe, J. Peters, and M. Sievert. Flooding on Urban Jobs, Connectivity, and Infra- 2017. “Does Large-Scale Infrastructure Invest- structure.” Background paper for this report, ment Alleviate Poverty? Impacts of Rwanda’s World Bank, Washington, DC. Electricity Access Roll-Out Program.” World Devel- Roberts, M., M. Melecky, T. Bougna, and Y. Xu. opment 89 (C): 88–110. https://doi.org/10.1016/j 2018. “Transport Corridors and Their Wider Eco- .worlddev.2016.08.003. nomic Benefits: A Critical Review of the Litera- Levy, H. 2004. “Rural Roads and Poverty Alleviation ture.” Policy Research Working Paper 8302, in Morocco.” Working Paper 30817, World Bank, World Bank, Washington, DC, January. https:// Washington, DC. doi.org/10.1596/1813-9450-8302. McIlwaine, C. 2013. “Urbanization and Gender- Rozenberg, J., and S. Hallegatte. 2015. “The Impacts Based Violence: Exploring the Paradoxes in the of Climate Change on Poverty in 2030 and the Global South.” Environment and Urbanization 25 Potential from Rapid, Inclusive, and Climate- (1): 65–79. https://doi.org/10.1177/0956247 Informed Development.” Policy Research Work- 813477359. ing Paper 7483, World Bank, Washington, DC. Nygård, K., E. Wahl, T. Krogh, O. A. Tveit, E. Samad, H. A., and F. Zhang. 2016. “Benefits of Elec- Bøhleng, A. Tverdal, and P. Aavitsland. 2007. trification and the Role of Reliability: Evidence “Breaks and Maintenance Work in the Water Dis- from India.” Policy Research Working Paper tribution Systems and Gastrointestinal Illness: A 7889, World Bank, Washington, DC. http:// Cohort Study.” International Journal of Epidemiol- documents.worldbank.org/curated/en/980911 ogy 36 (4): 873–80. https://doi.org/10.1093/ije 479147772730/Benefits-of-electrification-and /dym029. -the-role-of-reliability-evidence-from-India. Obolensky, M., A. Erman, J. Rozenberg, P. Avner, Sommer, M., S. Ferron, S. Cavill, and S. House. J. Rentschler, and S. Hallegatte. 2019. “Infra- 2015. “Violence, Gender, and WASH: Spurring structure Disruptions: Impacts on Households.” Action on a Complex, Under-Documented, and Background paper for this report, World Bank, Sensitive Topic.” Environment and Urbanization 27 Washington, DC. (1): 105–16. https://doi.org/10.1177/0956247 Pasha, H. A., and W. Saleem. 2013. “The Impact and 814564528. Cost of Power Load Shedding to Domestic Con- UN-Habitat (United Nations Human Settlements). sumers.” Pakistan Development Review 52 (4): 2003. Water and Sanitation in the World’s Cities: 355–73. Local Action for Global Goals. Abingdon-on- Picarelli, N., P. Jaupart, and Y. Chen. 2017. “Cholera Thames: Routledge. in Times of Floods: Weather Shocks and Health in Volpe Martincus, C., and J. S. Blyde. 2012. “Shaky Dar es Salaam.” IGC Working Paper C-4040- Roads and Trembling Exports: Assessing the TZA-1, International Growth Centre, London. Trade Effects of Domestic Infrastructure Using a Poder, T., and J. He. 2011. “How Can Sanitary Infra- Natural Experiment.” Inter-American Develop- structures Reduce Child Malnutrition and Health ment Bank, Washington, DC. http://publications Inequalities? Evidence from Guatemala.” Journal .iadb.org/handle/11319/4650. of Development Effectiveness 3 (4): 543–66. World Bank. 2019. Electricity Uptake for Economic Qadri, F., A. I. Khan, A. S. G. Faruque, Y. A. Begum, Transformation in Sub-Saharan Africa. Africa Devel- F. Chowdhury, G. B. Nair, M. A. Salam, D. A. opment Forum. Washington, DC: World Bank. Sack, and A.-M. Svennerholm. 2005. “Entero- Zhang, F. 2019. In the Dark: How Much Do Power Sector toxigenic Escherichia Coli and Vibrio Cholerae Distortions Cost South Asia? South Asia Develop- Diarrhea, Bangladesh, 2004.” Emerging Infectious ment Forum. Washington, DC: World Bank Diseases 11 (7): 1104–07. https://doi.org/10.3201 Group. https://doi.org/doi:10.1596/978-1-4648 /eid1107.041266. -1154-8. Rentschler, J., J. Braese, N. Jones, and P. Avner. 2019. “Three Feet Under: The Impacts of Natural Shocks Are a Leading Cause of Infrastructure Disruptions and Damages 4 S o far, this report has shown that the cost of infrastructure disruptions ranges from $391 billion to $647 billion in the low- and middle-income countries where data are available and for the types of impacts that can be quantified. Even though these estimates are incomplete, they highlight the substantial costs that unreliable infrastruc- ture imposes on people in low- and middle-income countries. But what role do natural hazards play in these disruptions? While it is impossible to answer this question globally and for all sectors, many case studies do document the importance of natural shocks in causing infrastructure disruptions. Infrastructure disruptions can have a range of quick to reach capacity constraints. Even causes. Conceptually, four categories of causes relatively minor external shocks can trigger can be distinguished: accidents that are human- failures. A lack of resilience to natural made external shocks, system failures during shocks is linked closely to a lack of reliability which parts of the functionality of infrastructure more generally—for example, from a lack systems break down, intentional external of investments in technical upgrades or attacks, and natural shocks (figure 4.1). maintenance. The importance of these types of shocks var- • In high-income countries, natural shocks are ies across different types of infrastructure and a leading cause of infrastructure disrup- different countries, and even from year to year. tions. Systems tend to be stable under nor- A lack of comprehensive data makes it difficult mal operating conditions, offering reliable to estimate accurately the share of infrastruc- services and suffering from relatively few ture disruptions that is caused by natural shocks. internal system failures. Yet external shocks Nevertheless, from the little data that are avail- still affect the functionality of systems, able, several general observations are possible: especially when maintenance is neglected. • Middle-income countries tend to be in a tran- • In low-income countries, the most frequent sition phase, which implies that the impacts cause of infrastructure disruptions tends to of infrastructure disruptions are particu- be system failure. Even under normal oper- larly large. The reliability and resilience of ating conditions, systems are inherently infrastructure systems may not be keeping fragile, prone to equipment failure, and up with rapid economic, urban, and demo- 57 58 LIFELINES FIGURE 4.1 Classification of causes of infrastructure disruptions Accidents System failures Attacks Natural shocks A critical error has occured. • Manmade external • Equipment failure • Vandalism • Storms shocks • Capacity • Terrorism • Floods constraints • Cyber attacks • Earthquakes •... graphic growth, which means that frequent on average about $15 billion in assets are at disruptions are causing widespread damage risk from natural hazards. to economic activity and well-being. Severe weather events—especially This chapter explores these issues in detail, storms—are among the main causes of focusing on the role of natural shocks. It also power outages explores the direct damage that natural haz- The high wind speeds produced by storms can ards inflict on infrastructure assets, which disturb the transmission and distribution of translates into repair and maintenance costs. electricity when flying debris hits lines or when The key findings are that in most countries transmission poles are damaged. Lightning can natural shocks are a significant and often lead- strike conductors and disconnect lines through ing cause of infrastructure disruptions. Further, short circuits, leading to voltage surges and a significant share of power, water, transport, damaging additional equipment (Panteli and and telecommunications infrastructure is Mancarella 2015). Falling trees are another located in areas exposed to natural hazards— major source of disruption. Reviewing almost and postdisaster repairs are a significant drag 20 years of power outage data for the United on the journey toward universal access to States, Rentschler, Obolensky, and Kornejew infrastructure services. (2019) find that states with dense forest cover are especially likely to experience outages THE POWER SECTOR IS HIGHLY during storms. VULNERABLE TO NATURAL The share of power outages from natural HAZARDS shocks can vary anywhere from 0 percent to In the power sector, analyses conducted for 100 percent—although most country-level this report find that storms are a major cause of estimates fall within the range of 10 percent to outages worldwide. They contribute to more 70 percent, according to evidence produced for than 50 percent of outages in Belgium, Croatia, this report (figure 4.2). Between 2000 and Portugal, Slovenia, and the United States, 2017, 55 percent of all recorded power outage stemming from damaged transmission net- events in the United States were caused by nat- works. Besides, a global risk analysis of power ural shocks and 44 percent were caused by generation infrastructure finds that every year nonnatural causes (figure 4.3).1 NATURAL SHOCKS ARE A LEADING CAUSE OF INFRASTRUCTURE DISRUPTIONS AND DAMAGES 59 FIGURE 4.2 The share of power outages caused by natural shocks varies significantly across countries 100 West Virginia Georgia Alabama 90 Share of power outages due to natural shocks (%) 80 Slovenia Croatia Belgium 70 Portugal 60 United States 50 Romania Italy France 40 Latvia Ireland Greece United Kingdom 30 Poland Germany Bangladesh (Dhaka) 20 Spain Sweden Lithuania 10 Czech Republic Bangladesh Canada Netherlands (Chittagong) Slovak Republic 0 0 10,000 20,000 30,000 40,000 50,000 60,000 70,000 80,000 GDP per capita (US$) Countries U.S. states Source: Rentschler, Obolensky, and Kornejew 2019. As for their duration, in the United States daily outages (figure 4.4). In Chittagong, a from 2000 to 2017, power outages caused by major coastal city in Bangladesh, storms are natural shocks lasted on average 2.5 days. This estimated to cause as few as 4 percent of all means that outages due to natural shocks outages (Rentschler, Obolensky, and Kornejew lasted more than twice as long as outages due 2019). In Dhaka, the World Bank’s Enterprise to nonnatural causes and three times as long as Surveys suggest that about two outages occur outages due to vandalism. In short, 74 percent a day on average throughout the year. How- of the total recorded outage time between ever, during the storm season in April and 2000 and 2017 was caused by natural shocks. May, outages are significantly more frequent In Europe between 2010 and 2016, climate- (although they do not necessarily occur in the induced outages lasted 409 minutes on aver- same areas of the city). In other words, a frag- age, making them almost four times as long as ile system is vulnerable not only to natural outages having nonnatural external causes. shocks but also to a host of other stressors and Over the period, natural shocks were responsi- shocks that include unmet demand, equip- ble for 37 percent of the total outage duration ment failure, and accidents. in the European countries considered. For most low- and middle-income countries, However, for developing countries such as limited data prevent quantification of the link Bangladesh, natural shocks account for a between power outages and storms. Still, in smaller share of power outages—not because those countries, a storm is likely to have a more their energy systems are more resilient, but severe impact than in high-income countries. because system failures and nonnatural factors Aging equipment, lack of maintenance, rapid are so frequent that energy users experience expansion of the grid, and excess demand due 60 LIFELINES FIGURE 4.3 Power outages from natural shocks last much more vulnerable to natural shocks in these longer than those from other causes countries than in richer countries, and natural Total power outage duration in the United States and 26 hazards can be responsible for a large number European countries, by cause of outages. For example, storms of the same a. Power outages, United States, b. Power outages, Europe, intensity are far more likely to cause outages in 2000–17 2000–16 Bangladesh than in the United States (figure 1,200 250 235 1,051 4.5). Wind speeds exceeding 25 kilometers an 1,000 200 hour lead to six times more outages in Bangla- 849 800 desh than in the United States. The gap 150 Number Number 600 116 becomes even wider when the threshold is 100 increased to winds over 35 kilometers an hour. 400 77 50 At that point, Bangladeshi consumers are 11 200 19 times more likely to experience a blackout 0 0 than U.S. consumers. ks s s ks n n The higher vulnerability of power systems ck ck ow ow oc oc ho ho kn kn sh sh ls ls Un Un in low- and middle-income countries means al al ra ra ur ur tu tu at at Na Na nn nn that even frequent events have large disrup- No No c. Average outage duration, d. Average outage duration, tive impacts. In Bangladesh, severe cyclones United States Europe damage power plants and power distribution 4,000 3,597 600 532 networks. Even relatively frequent storm 3,000 events, such as the nor’westers occurring each 400 Minutes Minutes 2,000 1,770 year during April and May, significantly 1,374 1,133 200 137 148 increase the incidence of power outages. These 1,000 78 storms, known for their localized but violent 0 0 gusts and lightning strikes, tend to cause sig- lis cks s s n ism ks n ng ck ck ow ow nificant damage to power transmission and oc ) di da o m ho ho al kn kn ed an sh sh nd ls ls Un Un sh l al Va t v ra No ura ra distribution systems—as illustrated by a recent ur ad tu ep tu t at Na xc na Na Lo nn (e n event in March 2019, after which 6,000 com- No munication towers lost access to power (Dhaka Source: Rentschler, Obolensky, and Kornejew 2019. Tribune 2019). In fact, these nor’westers Note: The European countries include the EU-28 (without Bulgaria, Denmark, and Hungary) plus Serbia. appear to be the main cause of storm-induced power outages (figure 4.6). to limited power generation capacity are all fac- tors that reduce reliability and increase the vul- Natural hazards also damage power nerability to natural shocks. However, these generation assets factors also mean that energy systems in What about the impact of natural shocks on low-income countries are typically character- power generation assets? In an analysis con- ized by frequent disruptions. As a result, out- ducted for this report, Nicolas, Koks, et al. ages from natural shocks can be expected (2019) demonstrate the significant exposure of to account for a smaller share of the overall power generation infrastructure to natural number of outages than in higher-income hazards. Using the Global Power Plant Data- countries. base of the World Resources Institute to pin- This does not mean that resilience to natural point the location of power plants, the study hazards is not an issue in low- and middle- assesses the exposure of plants to a large range income countries. Power systems are indeed of hazards (including cyclones, earthquakes, NATURAL SHOCKS ARE A LEADING CAUSE OF INFRASTRUCTURE DISRUPTIONS AND DAMAGES 61 FIGURE 4.4 Natural shocks only explain a fraction of power outages in Bangladesh Number of storm-induced outages in Chittagong and Dhaka, compared with annual average outages from all causes a. Chittagong, 2013 b. Dhaka, 2013 35 160 30 140 Outages per month Outages per month 25 120 100 20 80 15 60 10 40 5 20 0 0 Fe ary M y Oc ber ch ril ay ne ve r ce r r A ly pt st M ry Fe uary ch pt st ve r ril ay ne A ly Oc ber ce r r No obe De be be No obe De mbe be r Ju Se ugu Ju Ap Se ugu Ap ua ua ar M Ju ar M Ju nu em m m em m br br n t t Ja Ja Storm-induced outages Average outage rate per month (all causes) Source: Rentschler, Obolensky, and Kornejew 2019. floods, extreme heat, droughts, volcanic erup- FIGURE 4.5 The vulnerability of the power network to wind is tions, tsunamis, and wildfires). much higher in Bangladesh than in the United States A power plant is considered exposed if 16  Bangladesh Share of windy days during which at least (1) the type of asset is considered vulnerable to 14.3 14  United States a hazard (for example, a wind farm is never one outage was reported (%) 12 considered exposed to drought) and (2) the area where it is located has a “high” hazard 10 8.4 level for the relevant hazard in the ThinkHaz- 8 ard! database of the Global Facility for Disaster 6 Reduction and Recovery. The exposed capacity in 4 3.7 a country is calculated by totaling the capacity 2.4 1.8 of the plants exposed to each hazard and divid- 2 0.9 1.2 0.3 0.4 0.6 ing this sum by the total generation capacity in 0 the country. Exposed capacity can exceed 100 > 15 > 20 > 25 > 30 > 35 percent when power plants are exposed to Wind speed (km/hour) multiple hazards (such as when a wind farm is Source: Rentschler, Obolensky, and Kornejew 2019. exposed to storms and floods). Note: Windy days are defined using different thresholds for recorded daily wind speeds. Wind speeds are obtained from the global ERA5 climate reanalysis model, The study finds that a large fraction of the which tends to underrepresent the highest local wind speeds. km = kilometers. power generation capacity of many countries is exposed to hazards, often exceeding 100 per- cent due to the presence of multiple hazards For the most exposed countries, a large (map 4.1). Floods and coastal floods dominate share of the generation capacity is exposed to in most countries, cyclones dominate in most multiple hazards, and the dominant hazards island states as well as Mexico and the United are landslides, tsunamis, and earthquakes States, and extreme heat and water scarcity (figure 4.7). Earthquakes can inflict severe dominate in most of northern Africa and Asia. damage on power infrastructure. In the 2015 62 LIFELINES FIGURE 4.6 Storm-induced power outages are closely Nicolas, Koks, et al. (2019) then repeat the associated with the April–May nor’westers in Bangladesh exercise with high-voltage line infrastructure, 250 250 considering the three most devastating hazards  Power outages (left scale) for power lines: earthquakes, cyclones, and Lightning strikes (right scale) Number of lightning strikes Average number of outages 200 200 wildfires. They find that, as is the case for gen- eration, many countries are exposed to more 150 150 than one hazard. High-voltage infrastructure in countries such as Japan, Mexico, Mozambique, 100 100 Nepal, and New Zealand is heavily exposed to 50 50 various natural hazards. Notwithstanding, most of the Middle East and South Asia face 0 0 between 70 percent and 120 percent high- voltage line exposure. Fe ry ry ch st Oc er ve r ce r r ril ay ne ly e e be Ju gu Ap a b b ua b ar M Ju nu em to m m M Au br Ja The exposure of power generation to pt De No Se droughts is an overlooked risk that is, neverthe- Source: Rentschler, Obolensky, and Kornejew 2019. Note: Data shown are for 2000–17. Nor’westers are proxied by lightning strikes. less, increasing. Indeed, the vast majority of the Only outages due to natural shocks are included. Monsoon season accounts for the world’s electricity generation relies on either slight increase in outages in September. hydropower or thermoelectric power, both of earthquake in Nepal, for example, hydro- which are among the most water-intensive power plants accounting for 34 percent of the sources of electricity (Nicolas, Rentschler, et al. country’s capacity were damaged (Moss et al. 2019). Almost half of all global thermal power 2015). plant capacity is located in areas of water scar- MAP 4.1 Global exposure of power generation to multiple hazards Source: Nicolas, Koks, et al. 2019. Note: Map 4.1 shows the total exposed capacity for all hazards divided by the total installed capacity in each country. The value may exceed 100 percent because one power plant could be exposed to more than one hazard. Power plants are considered to be exposed to a hazard when they are in an area in which the hazard level is “high” in the ThinkHazard! database. The following hazards are considered: coastal flooding, earthquakes, floods, water scarcity, cyclones, volcanic eruptions, tsunamis, extreme heat, and wildfires. NATURAL SHOCKS ARE A LEADING CAUSE OF INFRASTRUCTURE DISRUPTIONS AND DAMAGES 63 FIGURE 4.7 Economies with the highest exposed generation capacity to multiple hazards Oman Pakistan Qatar United Arab Emirates Saudi Arabia Kuwait Bangladesh Sudan Fiji Nepal Costa Rica El Salvador Mozambique Honduras Nicaragua Japan Guatemala New Zealand Philippines Taiwan, China 0 100 200 300 400 500 600 Exposed capacity (% of total) Landslides Tsunamis Earthquakes Cyclones Extreme heat Water scarcity Floods Volcanic eruptions Source: Nicolas, Koks, et al. 2019. Note: The index represents the total exposed capacity for all hazards divided by the total installed capacity in the country. The value can exceed 100 percent, because one power plant could be exposed to more than one hazard. city, and 11 percent of hydroelectric capacity is power by 3.8 percent from the rates in an aver- located in such areas (Kressig et al. 2018; Wang, age year. Schleifer, and Zhong 2017). In India, 40 percent Based on the global exposure analysis, Nico- of thermal power plants are located in severely las, Koks, et al. (2019) then estimate the water-stressed areas. And between 2011 and expected annual damages (or repair costs) by 2016, 14 of the 20 largest plants were forced to considering the types of generation infrastruc- cease generating power at least once because of ture, hazard intensities, building standards used a water shortage, resulting in revenue losses of in the country, fragility curves, and infrastruc- $1.4 billion (Luo, Krishnan, and Sen 2018). ture investment costs, using data from Miya- Often, hydropower generation installations moto International (2019) and Schweikert et rely on a specific streamflow to function, but al. (2019). Damages are assessed only for the that streamflow cannot be maintained with most frequently recorded and costliest disas- low water availability (U.S. Department of ters—cyclones, earthquakes, surface flooding, Homeland Security and U.S. Department of river flooding, and coastal flooding—based on Energy 2017). Van Vliet et al. (2016) quantify the hazard data summarized in box 4.1.2 the relationship between water scarcity and The study finds that the total global expected power generation at the global level between annual damage (EAD) from all hazards totals 1981 and 2010. They find that droughts and about $15 billion, or around 0.2 percent of the warm years reduce the utilization rates for global value of the power generation infra- hydropower by 5.2 percent and thermoelectric structure. For low- and middle-income coun- 64 LIFELINES BOX 4.1 Exposure analysis of infrastructure assets is based on various hazard data sets Tropical cyclones. Tropical cyclones are repre- LISFLOOD-FP (Bates, Horritt, and Fewtrell 2010). sented by global cyclone hazard maps gener- Topographic information at 3-inch horizon- ated for the UNISDR Global Assessment Report tal resolution is available from the MERIT-DEM 2015 (Cardona et al. 2015). These maps show the model (Yamazaki et al. 2017). Inundation simu- distribution of cyclone wind speeds (peak wind lations take place at 90-meter resolution. More speed of 3-second gusts in kilometers per hour) details on inundation modeling can be found for five return periods between 50 and 1,000 in Vousdoukas et al. (2016). Flood simulations years. The maps are an output of probabilistic are forced by extreme sea levels obtained from cyclone analysis, based on perturbation of his- wave and storm surge reanalysis, combined torical cyclone tracks and wind-field modeling. with tidal information (Vousdoukas et al. 2018). Note that tropical cyclones are referred to as Waves are simulated using the WAVEWATCH-III hurricanes in the Atlantic, Caribbean Sea, and model (Tolman 2009), and storm surges are central and northeast Pacific; they are referred simulated using the DFLOW-FM model (Muis to as typhoons in the northwest Pacific. et al. 2016). Inland floods. River flooding (caused by rivers Earthquakes. Ground shaking hazard is repre- overtopping their banks) and surface flooding sented by the global earthquake hazard maps (caused by extreme local rainfall) are repre- produced for the UNISDR Global Assessment sented by the Fathom Global pluvial and fluvial Report 2015 (Cardona et al. 2015). These maps flood hazard data set (Sampson et al. 2015). This present the expected severity of ground shaking is a 3-arcsecond (~90 meters) resolution gridded as peak ground acceleration (PGA in centime- data set showing the distribution of maximum ters per square second), for five return periods expected water depth in meters. The hazard between 250 and 2,475 years. The hazard maps maps are for 10 return periods (5 to 1,000 years). are an output of probabilistic seismic hazard This analysis applies the “undefended” flood haz- analysis with global coverage. Because state of ard maps, which do not consider the effects of practice in situ testing for assessing liquefac- flood protection on inundation. The flood design tion potential is not feasible at the global scale, standards for road and rail are implemented from the geospatial prediction models of Zhu, Baise, the FLOPROS database (Scussolini et al. 2016). and Thompson (2017) are adopted. Liquefaction susceptibility is computed at a 1.2-kilometer grid Coastal flood s . Coastal inundation maps resolution based on a global data set (Worden are generated using the hydrological model et al. 2017). Source: Koks et al. 2019. tries, expected annual damages are $10 billion. tries. Those losses have been calculated consid- In some countries, annual losses exceed 1 per- ering not only the expected damage for each cent of the installed generation capital value plant but also an estimate of the restoration (map 4.2, panel a). Globally, losses are driven time in each country. Countries with the high- mainly by thermal plants for cyclones and by est production risks are those with power sys- hydropower plants for earthquakes. Map 4.2, tems that often are already under tight con- panel b, shows the resulting expected annual straints in terms of generation capacity. generation losses, which can represent up to 5 Several climate change–induced phenom- percent of the total generation in some coun- ena are likely to increase power sector risk. NATURAL SHOCKS ARE A LEADING CAUSE OF INFRASTRUCTURE DISRUPTIONS AND DAMAGES 65 MAP 4.2 Some low- and middle-income countries face high annual damage and generation losses Multihazard risk indicator for damage and lost production a. Expected annual damage to power plants (% of total capital value) b. Expected annual generation losses (% of total potential production) Source: Nicolas, Koks, et al. 2019. Note: Panel a shows the expected annual damage divided by installed capital values in the country. Panel b shows the expected lost generation (in megawatt-hours) divided by the total potential generation of the country. With more frequent droughts and higher tem- reduce power output by 0.45 percent to 0.8 peratures, the efficiency of nuclear and ther- percent (Mideksa and Kallbekken 2010). At mal power plants is likely to decrease. Research the same time, these events will affect substa- suggests that a 1°C temperature increase could tion equipment and the current rating of cables 66 LIFELINES and lines. They are also likely to increase sys- farms—water infrastructure is central to reduc- tem stress, because of the increased demand ing natural hazard risks related to floods and for air-conditioning. droughts. This infrastructure includes multi- In most regions, wind speed is likely to purpose reservoirs, river embankments, storm- increase with climate change, and atmospheric water drains, and coastal dikes, among others. icing (which negatively affects the performance A global analysis of the exposure of all water of wind turbines) is likely to decrease. Climate infrastructure to natural hazards was impossi- change will also affect flood frequency, river ble because of a lack of global data on water flows, and evaporation, with implications for sector assets. However, two partial assessments dam safety. In addition, climate change will were possible: (1) a crude global assessment of increase temperature, reducing the efficiency large dams, looking at their exposure to the of photovoltaic systems, which could drop two main natural hazards: high river inflows by about 0.5 percent for every temperature and earthquakes(Stip et al. 2019); and (2) a increase of 1°C (Patt, Pfenninger, and Lilliestam case study of China’s wastewater treatment 2013). Another impact of higher temperatures plants (WWTPs) to understand the level of could be increased transmission losses, because risks faced by this critical water infrastructure of the increased resistance of power lines. for river floods and earthquakes (Hu et al. Finally, climate change–induced sea-level 2019). The case study is based on a data set of rise may require power plant relocation. Sea- 1,346 WWTPs in China that includes the loca- level rise will be responsible not only for tion of assets and the size of the population increased flooding of coastal assets but also, dependent on each asset. combined with higher wind speeds, for more corrosion of these assets due to saltwater Dams are critical for reducing sprays. A study of potential impacts of climate downstream floods, but they can also change on the Bangladeshi power sector found create disasters if they collapse due to that around a third of power plants should be high river inflows relocated by 2030 to avoid inundations caused Dams’ reservoirs can be used for multiple pur- by sea-level rise (Khan, Alam, and Alam 2013). poses, depending on the context. These poten- Another 30 percent of Bangladesh power tial purposes include providing hydropower, plants will likely be affected by the increased supplying water for cities or irrigation, and salinity of cooling water and increased fre- reducing downstream flood risks. Dams are quency of flooding, while the northern region built with concrete spillways that release excess power plants will probably see a decrease in flows back into the river downstream from the output because of droughts. dam. The spillways are built with a specific design discharge to accommodate maximum WATER SYSTEMS ARE flows, typically ranging from 500-year to PARTICULARLY VULNERABLE 10,000-year or maximum probable discharges. TO CLIMATE CHANGE AND CAN If the discharge exceeds the spillway capacity, CONTRIBUTE TO MANAGING then water flows over the dam itself, which FLOODS AND DROUGHTS creates an emergency. If the dam is made from Water systems consist of reservoirs, ground­ earth, rock, or both, then the chances of dam water pumps, and transmission lines. They pro- collapse become quite high; if the dam is made vide different services, like bulk water provi- of concrete, then the chances of dam collapse sion, standard water supply and sanitation are lower, but it is still an emergency situation. services, irrigation, and drainage. In addition to If a dam collapses, it may have catastrophic supplying water—whether to cities, industry, or impacts on downstream communities. In NATURAL SHOCKS ARE A LEADING CAUSE OF INFRASTRUCTURE DISRUPTIONS AND DAMAGES 67 Henan Province, China, in 1975, the extreme river flood risk, representing around 21 percent rainfall produced by Typhoon Nina was beyond of the total global capacity. The actual risk of the design criteria of the Banqiao Reservoir- dam collapse depends on the design capacity of Dam. When exposed to such high levels of the spillway and the construction quality of the rainfall, the dam failed, killing tens of thou- dam. sands of people, with estimates reaching up to Until recently, climate change and its 171,000. impacts on hydrological flows have not been Spillway design standards are usually based considered in dam design. However, this is on the risk to downstream communities as changing quickly, and the latest example is the well as historical hydrological records. For Hydropower Sector Climate Resilience Guide (IHA example, a dam immediately upstream from a 2019), which was prepared under the auspices city typically has higher standards than a dam of the International Hydropower Association, in a rural area. However, over time the down- with technical and financial support from the stream populations may grow or a country’s World Bank and other international donors. risk tolerance may change, and thus there is a The risks associated with underdesigning for need to increase the spillway capacity and take future climates are multifold: dams will not be additional measures to ensure the structural able to provide reliable services to users—be it integrity of the dam. supplying water or power to cities or supplying The exposure analysis in this report is based irrigation water for agriculture—or help to mit- on the Global Reservoir and Dams Dataset igate flood and drought risks. Cervigni et al. (Version 1.01), which contains 6,862 records of (2015) stress how uncertainties about the reservoirs and associated dams with a cumula- future climate create a barrier to optimal dam tive storage capacity of 6,179 cubic kilometers. design (box 4.2). These only represent 20 percent of the dams registered by the International Commission on Water and wastewater treatment Large Dams, which lists more than 33,000 plants often face flood hazards, as they large dams. The Global Reservoir and Dams are typically in the lowest part of the Dataset is thus limited and likely biased toward network the high-income world; however, it is the only Wastewater collection systems typically work georeferenced record of dams (Lehner et al. by gravity to reduce energy costs, and treat- 2019) and was used for this exercise.  ment plants are generally located in low-lying The level of exposure of dams to high river flood-prone areas adjacent to the rivers, deltas, inflows—which could increase the chances of or lakes into which they discharge. For waste- exceeding spillway capacity and possible dam water systems with a combined sanitary and collapse—is difficult to assess at a global level. In storm drainage network, heavy precipitation this exercise, the “river flood risk” information can often overload the capacity of the net- from the ThinkHazard! database (2019) was work, resulting in combined sewer overflows used as an indirect proxy for considering river of untreated sewage into the environment. flows into a reservoir. If a dam is in an area clas- Constructing combined sewer overflow reten- sified as having a “high river flood risk,” this risk tion basins to store water temporarily and should indirectly and imprecisely correlate with then convey it back to the treatment plant is high river flows. The ThinkHazard! database one option that some cities are pursuing, but does not take future climate change into this approach is very expensive and only account, but rather relies on historical data. Of accessible to the richest cities. The case study the 6,862 dam sites in the Global Reservoir and on China conducted for this report finds that Dams Dataset, 15 percent are in areas of high climate change will significantly increase the 68 LIFELINES BOX 4.2 In hydropower, climate change adaptation is impaired by uncertainties Climate change will alter the amount of water FIGURE B4.2.1 Large changes in Africa’s available for important productive uses, such as hydropower revenues can be expected from hydropower and irrigation. But as illustrated by climate change from 2015 to 2050 a study of Africa, different climate models lead 140 to different results—with some projecting an 60 increase in available water and others projecting a Di erence from reference case (%) 40 decrease, making long-term planning particularly challenging (Cervigni et al. 2015). In the central 20 and southern Africa basins (Congo, Orange, and 0 Zambezi), depending on the climate scenarios –20 considered, the power and water sectors could underperform in many scenarios and overper- –40 form in others. In economic terms, the impacts of –60 climate change include lost revenue from under- –80 performing hydropower or irrigation infrastruc- Volta Niger Eastern Nile Zambezi Senegal Congo, ture in drier climate futures and, by contrast, the Nile equatorial Rep. lakes opportunity cost of not taking advantage of an abundance of exploitable water resources in wet- Maximum relative gain due to climate change Maximum relative reduction due to climate change ter climate futures. In simulations of the economic performance Source: Cervigni et al. 2015. of infrastructure in the climate scenario at the Note: The bars reflect the range of economic outcomes across all end of the range, the deviations from the results climate futures for each basin—that is, the highest increase (blue bars) and highest decrease (red bars) of hydropower revenues (dis- expected under a historical climate are dramatic. counted at 3%), relative to the no-climate-change reference case. In hydropower (figure B4.2.1), dry scenarios lead The outlier bar corresponding to the Volta Basin has been trimmed to revenue losses on the order of 10–60 percent to avoid distorting the scale of the chart and skewing the values for the other basins. Estimates reflect the range, but not the distribu- of baseline values, with the Nile (Equatorial Lakes tion, of economic outcomes across all climate futures. Each basin’s region), Senegal, and Zambezi basins being most results reflect the best and worst scenarios for that basin alone affected. Wet scenarios result in potential rev- rather than the best and worst scenarios across all basins. enue increases on the order of 20–140 percent, with the Eastern Nile, Niger, and Volta basins because power systems would have been planned having the largest gains. In some wetter climate in anticipation of lower than actual generation futures, infrastructure could perform better than from hydropower. As a result, the transmission expected because, for a given installed capacity, lines and power trading agreements needed to more hydropower or more crops could be pro- bring the extra hydropower to market may sim- duced with the extra water. However, many of the ply not be available. Without them, the gains from corresponding gains could be only potential ones, more abundant water might not be realized. exposure of Chinese WWTPs to floods, even For an event with a 30-year return period over the short term, with large potential under a scenario of moderate climate change, impacts on users (Hu et al. 2019). The sign of 35 percent of the WWTPs (472 out of 1,346 this effect is consistent in 10 out of the 11 plants) supplying 176 million people could climate models considered, although the experience significantly higher flood risk by magnitude of the impacts varies across models. 2035. By 2055, the number of exposed people NATURAL SHOCKS ARE A LEADING CAUSE OF INFRASTRUCTURE DISRUPTIONS AND DAMAGES 69 could rise by up to 208 million from the pres- reflection of the higher concentration of dams ent number. in richer economies and the large number of mega-dams in middle-income countries, par- Dams and wastewater treatment plants ticularly in Brazil and China. are also exposed and vulnerable to For China, Hu et al. (2019) find that earth- earthquakes quakes also pose a significant risk to waste­ Overtopping of dams can have large negative water treatment operations. In an earthquake consequences, whereas dam collapses caused event with a return period of 250 years in by earthquakes can be catastrophic—possibly China, 31 WWTPs are exposed to ground shak- triggering rapid flooding with many human ing of medium severity. More than half of casualties. In Japan, central China, the U.S. these plants are also in areas with high lique- West Coast, Southern Europe, and the Middle faction susceptibility, indicating their high vul- East, dams and reservoirs face the highest seis- nerability. Spatially, the western regions of mic risk (map 4.3) (Stip et al. 2019). About 2 mainland China and the surroundings of Bei- percent of the dams considered in this study jing are prone to the highest seismic risks.4 face very high peak ground acceleration (PGA) Dams and WWTPs are not the only water levels, with a return period of 2,475 years.3 assets that are exposed and vulnerable to natu- High-income countries have the largest num- ral hazards, so the analysis described here con- ber of dam sites exposed to earthquakes. How- siders only part of the vulnerability of the ever, upper-middle-income countries have the water system to climate change and natural largest capacity of dams exposed to the risk of hazards. That vulnerability arises as well from seismic shaking. This finding is probably a pumping stations or control centers subject to MAP 4.3 Dams and reservoirs face a high seismic risk Peak ground acceleration faced by dam sites for a 2,500-year earthquake event Peak ground acceleration level (g) < 0.092 0.092–0.18 0.18–0.34 0.34–0.65 > 0.65 Source: Stip et al. 2019. Note: See box 4.1 for a description of earthquake data. g = standard gravity acceleration. 70 LIFELINES collapse or flooding. In addition, a significant Meanwhile, warm summers with low pre- part of damage to water systems is caused by cipitation can affect inland waterway transport: breaking or leaking pipes from ground lique- in northwestern Europe, the dry summer of faction, landslides, and fault crossings (Kakderi 2013 resulted in low water levels and losses of and Argyroudis 2014). Furthermore, growing €480 million stemming from the inoperability water scarcity in many parts of the world will of some large vessels and a shift to other forms make it even more challenging to provide of transport (Jonkeren et al. 2014). water for many competing uses (Damania et al. Transport disruptions are costly. In the Euro- 2017). Although some studies show promise in pean Union, the total costs of the influence of identifying vulnerable sections of water infra- extreme weather events on the transport sys- structure (such as Bagriacik et al. 2018), they tem are an estimated €2.5 billion a year. Of rely on high-quality data describing the exist- these costs, about 72 percent are attributed to ing network, which limits their applicability to roads, 14 percent to air travel, and 12 percent low- and middle-income countries. to the rail sector. The remaining 2 percent are related, in descending order of magnitude, to NATURAL HAZARDS FREQUENTLY maritime transport, inland waterways, and DISRUPT AND EXTENSIVELY intermodal freight transport (Enei et al. 2011). DAMAGE TRANSPORT Looking at the next four decades in the Euro- INFRASTRUCTURE pean Union and using the same methodology, In the transport sector, weather events cause Doll, Klug, and Enei (2014) expect the road accidents, congestion, and delays. An analysis transport costs arising from extreme weather conducted for this report also finds that natural events to increase by 7 percent. Higher flood hazards cost about $15 billion a year on risks and less predictable winters could increase average in direct damage to global transport rail traffic costs by up to 80 percent. infrastructure. Natural hazards are responsible for Variations in the weather cause frequent large repair and maintenance costs in disruptions in all modes of transport road and rail networks Even in the absence of extreme natural shocks, How do natural hazards fit in? A new analysis weather can disrupt road, rail, water, and air conducted for this report demonstrates the sig- transport. In the United States, about 16 per- nificant exposure of transport infrastructure to cent of flight delays are caused by relatively natural hazards. With a resolution that is minor weather events and only about 4 percent unprecedented on a global scale, Koks et al. by extreme weather (Bureau of Transportation (2019) combine data on road and rail network Statistics 2018). And a survey of the empirical assets with information on the most significant literature finds that precipitation increases the types of natural hazards. This global study frequency of road accidents and increases con- assesses damaged network infrastructure at the gestion by reducing vehicle speeds (Koetse and asset level, such as individual road segments or Rietveld 2009). In the United States, about 15 bridge structures. The road and railway data percent of road traffic congestion is attributed used in this analysis are based on open-access to bad weather (Cambridge Systematics Insti- data from OpenStreetMap, which, thanks to tute and Texas Transportation Institute 2005). voluntary contributors, is a comprehensive Such effects are not limited to road networks; data set (Barrington-Leigh and Millard-Ball in Finland, 60 percent of freight train delays 2017; Meijer et al. 2018). between 2008 and 2010 were related to winter The exposure and risk of road and railway weather (Ludvigsen and Klæboe 2014). assets are assessed for the most frequently NATURAL SHOCKS ARE A LEADING CAUSE OF INFRASTRUCTURE DISRUPTIONS AND DAMAGES 71 recorded and costliest disasters: tropical This analysis finds that about 27 percent of cyclones, earthquakes, surface flooding, river all global road and railway assets are exposed flooding, and coastal flooding (see box 4.1 for to at least one hazard, and about 7.5 percent of hazard data sources). An asset is considered to assets are exposed to a 100-year flood event. be exposed only when the probability of occur- Road and rail networks are most exposed to rence of the hazardous event exceeds the surface flooding, followed by tropical cyclones, assumed design protection standards of the river flooding, and earthquakes (figure 4.8). asset. In this way, countries’ different resilience For earthquakes and surface flooding, richer standards can be incorporated in the analysis. countries with more assets are proportionally FIGURE 4.8 Global exposure of transport infrastructure to multiple natural hazards a. Cyclones b. Earthquakes c. Surface flooding d. River flooding e. Coastal flooding Annual exposed km (x1,000) 32 24 96 16 4.8 24 18 72 12 3.6 16 12 48 8 2.4 8 6 24 4 1.2 0 0 0 0 0 17 8 20 9 2 52 0. .18 0. .34 5 5 .50 .50 .50 –1 –1 –1 2 >2 2 >2 2m m 25 17 20 .6 .6 1– 1– 50 50 50 >2 >2 0 4– –0 >0 –0 –0 –0 –0 9– 1– 8– 2– 0. 0. 0. 15 34 18 25 25 25 09 0. 0. 0. 0. Wind speed Peak ground Water depth Water depth Water depth (km/hour) acceleration (g) (meters) (meters) (meters)  Low income  Lower-middle income  Upper-middle income  High income Source: Koks et al. 2019. Note: Map shows the hazard causing the highest transport infrastructure exposure in each region. The accompanying pie chart indicates the percentage of land area with the highest exposure to each hazard. Panels a to e present the exposure of the four country income groups to each hazard type and intensity. g = standard gravity acceleration; km = kilometers. 72 LIFELINES more exposed. But for river and coastal flood- as shown in figure 4.9, is caused by two key ing, high-income countries have fewer kilome- dynamics facing in opposite directions. At first, ters exposed because of their higher flood pro- states accumulate infrastructure as gross tection standards. For tropical cyclones and domestic product (GDP) increases, but this earthquakes, the large share of exposed infra- expansion is at the expense of higher disaster structure in upper-middle- and high-income exposure and greater damage. After they reach countries is related predominantly to the geo- a given level of income (in the middle- graphic distribution of the hazards. income category), they have enough resources The resulting total global expected annual to prioritize higher resilience. Thus they reduce damage from all hazards ranges from $3.1 bil- the exposure and vulnerability of their infra- lion to $22 billion, with a mean EAD of $14.6 structure through investments in more rigor- billion, depending on various assumptions ous design standards for transport assets and in about construction and reconstruction costs flood protection. and other uncertainties. Considering only low- In absolute terms, losses are the largest in and middle-income countries, the mean EAD big and wealthy countries, which is not sur- is $8 billion on average across scenarios. Of the prising. However, when EAD is considered in global damage, about 73 percent is caused by relation to GDP, infrastructure value, or infra- surface and river flooding, followed by coastal structure length, it appears that lower- and floods (16 percent), earthquakes (7 percent), middle-income countries are often more and tropical cyclones (4 percent) (Koks et al. severely affected (figure 4.10). In small island 2019). The results are driven mainly by pri- developing states, for example, the annual mary roads, which experience the highest rela- damage relative to the total infrastructure tive damage, and by tertiary roads, which rep- value is more than double the global average. resent the greatest cumulative length. At the global level, expected annual dam- But expected annual losses can hide the fact ages are small compared to the budget required that rare events cause devastating damage. for maintaining reliable transport networks Although earthquakes represent only 7 per- (0.2 percent to 1.5 percent). However, our cent of total annual losses, ground shaking or results reveal geographic disparities in expo- soil liquefaction can severely affect the func- sure and risk, and for several countries and tionality of transport infrastructure. Roads and regions, investing in transport asset resilience railroads can be blocked by fault ruptures, col- should be a priority (Rozenberg et al. 2019). lapsed buildings, or landslides; tunnels may Climate change will intensify the impacts of collapse; embankments can be displaced by soil natural hazards on transport infrastructure. For liquefaction; and bridges can collapse or example, in Mozambique, Kwiatkowski et al. become unstable (Argyroudis and Kaynia (2019) find that the risk of river flooding to 2014). In the 1995 earthquake in Kobe, acces- bridges under current conditions amounts to sibility as measured by the length of the open $200 million a year (1.5 percent of Mozam- network dropped by 86 percent directly after bique’s GDP) and could reach up to $400 mil- the shock for highways and by 71 percent for lion by 2050 in the worst-case climate change railways (Chang and Nojima 2001). scenario. Another interesting finding of Koks et al. (2019) is that as the wealth of countries Zooming in on urban flooding and road increases, the damage to their transport infra- networks structure first rises and then falls. This bell- Urban flooding is a major cause of transport shaped relationship between income and EAD, disruptions in cities across the world, and these NATURAL SHOCKS ARE A LEADING CAUSE OF INFRASTRUCTURE DISRUPTIONS AND DAMAGES 73 FIGURE 4.9 Transport infrastructure damage first increases with income growth and then decreases Expected annual damage (EAD) per hazard, by country income group a. Cyclones b. Earthquakes c. Surface flooding d. River flooding e. Coastal flooding f. Total risk 0.60 1.0 4.0 5 2.0 12 Expected annual damage 0.8 3.2 4 1.6 10 0.45 (US$ billions) 8 0.6 2.4 3 1.2 0.30 6 0.4 1.6 2 0.8 4 0.15 0.2 0.8 1 0.4 2 0 0 0 0 0 0 30.0 60 400 350 180 200 Expected annual damage 50 320 280 160 22.5 150 per km (US$) 40 240 210 120 15.0 30 100 160 140 80 20 7.5 50 10 80 70 40 0 0 0 0 0 0  Low income  Lower-middle income  Upper-middle income  High income Source: Koks et al. 2019. Note: Graphs show the expected annual damage (EAD) in absolute terms (top row) and per kilometer (km) of road (bottom row). FIGURE 4.10 Low- and middle-income countries bear the highest damage costs relative to their GDP Multihazard risk in expected annual damage (EAD), by country a. Expected annual damage in absolute terms b. Expected annual damage relative to GDP China Myanmar Japan Bolivia Indonesia Liberia United States Georgia Vietnam Lao PDR Philippines Somalia Brazil Belize India Vanuatu Myanmar South Sudan Russian Federation Madagascar Mexico Tajikistan Turkey Central African Rep. Bolivia Gambia, The Thailand Fiji Germany Vietnam France Afghanistan Chile Niger Iran, Islamic Rep. Papua New Guinea Argentina Mali Italy Sierra Leone 0.1 1.0 10.0 0.0 0.2 0.4 0.6 0.8 10.0 US$ (billions) Share of GDP (%) Source: Koks et al. 2019. Note: Panel a presents the 20 countries that have the highest multihazard EAD in absolute terms. Panel b presents the 20 countries that have the highest multihazard EAD relative to the country’s GDP. 74 LIFELINES MAP 4.4 Flooded segments of the road network (50-year return period), Inner Kampala Source: Rentschler et al. 2019. disruptions extend well beyond just the flood flood zones (map 4.4). In a 50-year flood, an zone. From Buenos Aires to Dar es Salaam, estimated 10 percent (11 kilometers) of all pri- Amman, Dhaka, and Jakarta, urban flooding is mary roads in Inner Kampala are flooded, and a frequent and devastating occurrence, espe- 8 percent of all primary roads are flooded at a cially in low- and middle-income countries. depth of more than 15 centimeters, thereby Open-source road network data reveal how preventing passage of most conventional cars.5 exposed urban road networks are to flooding. However, only 3 percent (45 kilometers) of For example, Rentschler et al. (2019) estimate residential roads are directly affected. that in Inner Kampala, about 4 percent of all In Dar es Salaam, the bus rapid transit lanes, roads are affected by a flood with a return the bus depot, and the port access road are period of 50 years. Primary roads (such as highly exposed to flooding by rainfall events motorways) are disproportionally located in with intensities as low as 4–6 millimeters per NATURAL SHOCKS ARE A LEADING CAUSE OF INFRASTRUCTURE DISRUPTIONS AND DAMAGES 75 hour over a 24-hour period, which currently FIGURE 4.11 Urban flooding affects a significant share of the occur every 2–10 years (ICF 2019). By 2050, road networks in Bamako, Dar es Salaam, Kampala, and Kigali all segments of Dar es Salaam’s bus rapid tran- 12 sit system will be exposed to routine flooding 10 Share of roads a ected (%) by events on the order of 4–6 millimeters an hour. Climate change will likely increase the 8 frequency and intensity of rainfall events and 6 thus lead to more frequent flooding. The story is much the same in other African 4 cities—such as Bamako and Kigali—where a 2 significant share of roads is affected by a flood depth of more than 15 centimeters in an event 0 Bamako Dar es Salaam Kampala Kigali with a 50-year return period (figure 4.11). Just as in Kampala, primary roads in Bamako and  Motorways and  Secondary  Tertiary  Residential and primary roads roads roads other roads Kigali are disproportionally affected by flood- ing. This high exposure of primary roads sug- Source: Rentschler et al. 2019. Note: Figure shows the percentage of roads affected by a flood depth of more than gests that urban floods have significant indirect 15 centimeters in an event with a 50-year return period. effects on the wider urban economy because they affect the linkages even between non- flooded areas. Indeed, infrastructure disrup- • Internet exchange points and other data tions are by no means limited to certain centers low-income neighborhoods—infrastructure • Wireless transmission infrastructure— systems are networks that transmit the disrup- towers and antennas. tions from urban flooding across wide areas. Chapter 3 of this report reveals how flooding Table 4.1 depicts the impacts of various cli- in a few locations of Kampala’s road network matic events on telecommunications infrastruc- can affect households’ access to health care ture based on studies commissioned by public across the entire city. sector agencies in the United Kingdom, the United States, and academia. As seen in the WHEN NATURAL SHOCKS DISRUPT table, acute events have a significant impact on TELECOMMUNICATIONS SYSTEMS, almost all forms of infrastructure, with earth- WHOLE COUNTRIES CAN GO quakes (high intensity) the most destructive OFFLINE across the spectrum of infrastructure. Telecommunications infrastructure, if dense Data centers and landing stations are partic- enough, has a certain level of resilience built ularly vulnerable to flooding because of the into its structure; but like other critical infra- large quantities of ICT equipment involved in structure, points of failure exist that are vul- their operations. As a result, submarine cable nerable to acute and chronic natural hazards. landing stations are the most vulnerable to a The core telecommunications infrastructure rise in sea level—one of the most direct impacts and information and communication technol- of long-term climate change. ogy (ICT) making up global networks can be Analyzing the climate risks to different types categorized as of ICT infrastructure using the broadband value chain can improve our understanding of • Submarine cables the impacts of damage to each type of asset. • Landing stations for submarine cables The broadband value chain comprises three • Terrestrial cables—underground and overland broad segments: 76 LIFELINES TABLE 4.1 Climatic events and their impacts on telecommunications infrastructure Inland and Sea-level High Water High winds Infrastructure coastal floods Earthquakes Tsunamis rise temperatures scarcity and storms Submarine cable L H M L L L L (deep sea) Submarine cable L H H L L L L (near shore) Landing station H H H H L L L Terrestrial cables M H L L L L L (underground) Terrestrial cables L M L L L L M (overland) Data centers H M L L M M L Wireless transmission L M L L L L H antennas Source: Adapted from Adams et al. 2014; Dawson et al. 2018; Fu, Horrocks, and Winnie 2016; and U.S Department of Homeland Security 2017. Note: L = low; M = medium; H = high. • First mile. International Internet connectiv- activity keep their submarine cable repair ity through submarine cables or terrestrial teams fairly busy. For example, off the coast of cross-border links eastern China, and particularly between Tai- • Middle mile. Domestic connectivity infra- wan, China, and mainland China, frequent structure linking sources of first-mile con- undersea earthquakes result in almost one nectivity to population centers—mostly cable break a week (Brandon 2013). The pres- cables running along existing connectivity ence of a highly active port contributes to more routes (transport and energy) frequent cable breaks, mostly from dropped or • Last mile. Infrastructure connecting indi- dragging anchors hitting the submarine cables viduals and premises to telecommunica- on the sea floor. tions networks—fiber or cable to the home Submarine cable systems are most at risk from local cabinets, mobile towers, or Wifi from earthquakes and landslides on the seabed transmitters. (table 4.1). This vulnerability also extends to landing stations, but modern construction First-mile infrastructure is critical—and techniques have improved the resilience of the the most vulnerable to earthquakes, buildings housing them. That said, coastal tsunamis, and landslides flooding and tsunamis can cause great damage First-mile infrastructure corresponds to more to landing stations, whereas the offshore cables than 370 submarine cable systems that connect themselves may remain protected. The great to terrestrial networks through landing stations Hengchun Earthquake on the island of Taiwan, in almost all coastal and island countries. These China, and in the Luzon Strait in December cable systems—the main arteries of the global 2006 was one of the severest examples of the Internet—carry the world’s information, disruption of submarine cable systems. Subma- including virtually all international financial rine landslides triggered by the earthquakes transactions. Although the number and fre- and the subsequent turbulent currents traveled quency of faults in submarine cable systems more than 300 kilometers, causing 19 breaks are low, the parts of the world prone to seismic in seven cable systems. Some of the damaged NATURAL SHOCKS ARE A LEADING CAUSE OF INFRASTRUCTURE DISRUPTIONS AND DAMAGES 77 cables were at depths of 4,000 meters. Repairs lines is perhaps most at risk because of expo- were carried out by 11 vessels over 49 days. sure to multiple hazards. The Internet connectivity of China; Japan; the Philippines; Singapore; Taiwan, China; and Last-mile infrastructure is most Vietnam was seriously affected, with all coun- exposed, but can be recovered quickly tries losing a portion of their international The infrastructure most prevalent in last-mile capacity. Financial services, airlines, and ship- access—poles and antennas—is physically ping industries were significantly affected, and quite resilient and can withstand significant cli- commerce in Taiwan, China, in general, came matic pressures. For example, mobile antennas to a halt. Traffic was rerouted rapidly using can withstand winds of up to 250 kilometers undamaged infrastructure, but the pressure on an hour, and terrestrial cables are either under- them resulted in lower-quality service, delays, ground in ducts or on wooden and metal poles and failures in those cable systems because of in urban centers. However, falling trees or dis- overloading. Following the earthquake, a sur- lodged debris can cause failures and are vey was conducted in China to estimate the unavoidable in these situations. Therefore, impact of the disruption, and the results were investments to ensure timely recovery of ser- staggering. It found that 97 percent of Chinese vices in the event of a disaster may be more Internet users faced issues visiting foreign web- effective than investments in protecting the sites, and 57 percent felt that their life and exposed last-mile assets. work were affected (APEC Secretariat 2013). Data centers are also vulnerable to Middle-mile infrastructure can be climatic conditions and extreme events protected by redundancy Another important element of the digital eco- The middle mile of broadband networks con- system—and now a core digital infrastructure— sists of telecommunications infrastructure con- are the data centers that host the various web- necting population centers within a country, sites, services, and applications used globally. similar to road highways. These connectivity According to a survey by the Ponemon Institute routes, which can connect internationally (2016) covering 63 data centers, between 10 across terrestrial borders, are part of the global percent and 12 percent of data center outages Internet. Countries with well-developed tele- are attributable to weather conditions. communications sectors have dense middle- In January 2015, thousands of residents of mile networks, with a larger number of routes Perth, found themselves suddenly discon- and carriers per route. Unlike electricity, the nected from the Internet. The Internet service transmission routes carry two-way traffic. provider had suffered cooling system failures Therefore, a break at one point of the network in part of its data center. Under normal cir- does not necessarily mean everything down- cumstances, the failure may not have led to stream is unconnected. Large parts of the net- network outages, but because the second- work can be revived by using the alternate hottest day of the summer that year had led to routes available, perhaps even originating fears of server failure, the Internet service pro- downstream from the failure point. vider shut down the servers affected by the The primary risks to middle-mile infrastruc- failed cooling system. Although annoying and ture—areal and underground cables—are cumbersome for residential customers, the earthquakes, landslides, and strong winds. economic impact of a loss of connectivity is Underground infrastructure is also at risk from felt almost immediately by businesses. Simple flooding. Infrastructure that runs along coast- tasks such as paying for goods with a credit 78 LIFELINES card become impossible without Internet drastically increase with climate change. connectivity. Extreme heat waves in cities lead to worse air Climate change can have far-reaching quality and numerous heat-related health impacts on telecommunications through issues and even death, especially among vul- chronic changes, particularly because cooling is nerable groups such as children and the elderly. a core requirement of data centers, which are Today, the impact of such events is significant: the foundation of the Internet. Rising tempera- in a 2015 heat wave in Delhi, more than 2,000 tures and lowering water tables will make cool- deaths were recorded. It is estimated that more ing increasingly challenging at industrial scales. than 350 cities and more than 200 million peo- The Uptime Institute, which tracks data center ple are regularly exposed to extreme heat, trends, estimates that today telecommunica- defined as a three-day period with average tions companies can spend almost 80 percent of maximum temperature of at least 35°C. By the cost of running their server on cooling them. 2050, the number of affected people will increase by 700 percent, to 1.6 billion, in more INFRASTRUCTURE SOMETIMES than 970 cities (Viguié et al. 2019). CREATES OR INCREASES NATURAL Urban infrastructure contributes to heat RISKS waves through the urban heat island effect and Not only are infrastructure assets exposed to air-conditioning systems. In the urban heat risks, but they also create risks and can increase island effect, cities are warmer than their sur- exposure. Sometimes, infrastructure directly rounding areas because they consist of built-up creates a hazard, such as when electricity trans- surfaces that absorb heat (map 4.5). The trans- mission and distributional lines trigger wildfires. portation infrastructure necessary for the func- In California in 2007, San Diego Gas and Elec- tioning of cities, such as paved roads, contributes tric was found liable for $2 billion in damages to this effect. To cope with high temperature from three fires that led to two deaths and the levels, residents and businesses resort to air- destruction of 1,300 homes (Daniels 2017). conditioning to maintain cool interiors. Air- Large reservoirs can increase the frequency conditioning systems, however, usually emit of earthquakes in areas of high seismic activity warm air to the outside and thus further and can cause earthquakes to happen in areas increase the overall urban heat island effect. that were thought to be seismically inactive. Heat waves stress infrastructure, especially Sometimes, infrastructure magnifies natural by increasing the demand for electricity. In a risks through so-called natech disasters (techno- case study of Paris, Viguié et al. (2019) simu- logical disasters triggered by a natural hazard). late the effect of more frequent and hotter heat For example, the Fukushima nuclear accident waves on air-conditioning. To maintain a tem- in Japan in 2011 was a technological accident perature of 23°C in all buildings, they project provoked by an earthquake and tsunami. At an average increase in final energy consump- times, infrastructure may not influence the haz- tion of 1.134 terawatt-hours a year. During a ard itself but may increase the exposure to the heat wave, the additional energy consumption hazard. An example is the development of from cooling corresponds to 81 percent of the transport, energy, or water infrastructure that current average daily electricity consumption attracts people and investment to risky areas. for offices and housing in Paris. Such addi- tional demand represents a significant chal- Urban infrastructure and lenge and can lead to outages, especially in air-conditioning worsen heat waves places where power systems are underdimen- Heat waves are already a big threat to well- sioned and struggle to keep pace with growing being and health in cities, and this threat will energy consumption. NATURAL SHOCKS ARE A LEADING CAUSE OF INFRASTRUCTURE DISRUPTIONS AND DAMAGES 79 MAP 4.5 Simulation of air temperature in the streets of Greater Paris at 4 a.m., after nine days of a heat wave similar to that of 2003 Source: Viguié et al. 2019. Note: The urban heat island effect is clearly visible, with a 6°C temperature difference between the center of Paris and the countryside. Infrastructure system disruptions in dependent systems and cause a interdependencies can amplify the cascading effect that greatly amplifies the impacts of a shock impact of the original event (Kadri, Birregah, Although it is important to understand how and Châtelet 2014). Such domino effects are specific shocks influence certain infrastructure difficult to anticipate because they consist of systems, such analyses done in isolation are the interactions of several highly complex sys- likely to underestimate an event’s actual tems, each of which is difficult to understand impact, because infrastructure systems are individually. Several approaches can be taken interconnected. These connections, or interde- to modeling interdependent infrastructure pendencies, can be classified as physical, cyber, systems differing in scope, complexity, and geographical, and logical. Physical interdepen- data requirements, but typically they are dency describes a system that is materially more useful for depicting the nature and dependent on another system. Cyber interde- direction of interdependencies than for accu- pendency refers to a system that is reliant on rately quantifying these relationships (Barker functioning information infrastructure. Geo- and Santos 2010; Ouyang 2014). In power graphical interdependency describes an envi- grids, for example, critical infrastructure inter- ronment that can simultaneously alter local dependencies link electricity infrastructure to systems. Finally, logical interdependency can transport, water supply, and ICT infrastruc- be used to classify all other connections ture, and the provision of oil and gas. Disrup- between two or more systems that cannot be tions in any of these components can cause described as being physical, cyber, or geograph- outages in all of the other systems and render ical (Rinaldi, Peerenboom, and Kelly 2001). restoration after a shock difficult. This again In the event of a shock, disruptions in one highlights the need for coordination among infrastructure system can thus translate into different actors in preparation and recovery 80 LIFELINES activities (Wender, Morgan, and Holmes Bagriacik, A., R. A. Davidson, M. W. Hughes, B. A. 2017). Bradley, and M. Cubrinovski. 2018. “Comparison of Statistical and Machine Learning Approaches This chapter has presented new analysis on to Modeling Earthquake Damage to Water Pipe- the exposure of infrastructure networks and lines.” Soil Dynamics and Earthquakes Engineering the damages they face due to natural disasters 112 (September): 76–88. and climate change. However, the impacts of Barker, K., and J. R. Santos. 2010. “A Risk-Based disasters go way beyond the direct damages to Approach for Identifying Key Economic and the infrastructure assets and can propagate to Infrastructure Systems.” Risk Analysis 30 (6): 962–74. regions and economic actors that were not hit Barrington-Leigh, C., and A. Millard-Ball. 2017. directly. This is the subject of the next chapter “The World’s User-Generated Road Map Is of this report. More Than 80% Complete.” PLoS One 12 (8): e0180698. Bates, P. D., M. S. Horritt, and T. J. Fewtrell. 2010. NOTES “A Simple Inertial Formulation of the Shallow 1. The cause of the remaining 1 percent of outage Water Equations for Efficient Two-Dimensional events is unknown. Flood Inundation Modelling.” Journal of Hydrology 2. Extreme heat days and droughts are also 387 (1-2): 33–45. high-frequency events, but their impact is Brandon, J. 2013. “Protecting the Cables That Wire mostly on service continuity during the event Our World.” Popular Mechanics, March 15. and very rarely affects the integrity of the asset. Bureau of Transportation Statistics. 2018. “Under- 3. Normally, the return period for maximum standing the Reporting of Causes of Flight Delays design earthquake is 10,000. However, since and Cancellations.” Bureau of Transportation the global data set only has return periods up Statistics, Washington, DC, March 29. to 2,475 years, this was taken as the basis for Cambridge Systematics Institute and Texas Transpor- assessment and should still indicate the approx- tation Institute. 2005. Traffic Congestion and Reli- imate earthquake hazard level. ability: Trends and Advanced Strategies for Congestion 4. The analysis looks at PGA hazard maps for Mitigation. Medford, MA: Cambridge Systematics return periods of 250, 475, 950, 1,500, and Institute; College Station, TX: Texas Transporta- 2,475 years. tion Institute. 5. Technically, flood depths of up to 20 centime- Cardona, O. D., G. A. Bernal, M. G. Ordaz, M. A. ters may be safe for most vehicles to navigate. Salgado-Gálvez, S. K. Singh, M. G. Mora, and However, in practice—and confirmed anecdo- C. P. Villegas. 2015. “Update on the Probabilis- tally—drivers cannot gauge exact water depths tic Modelling of Natural Risks at Global Level: and thus need to avoid submerged roads. Global Risk Model—Global Earthquake and Tropical Cyclone Hazard Assessment; Disaster REFERENCES Risk Assessment at Country Level for Earth- Adams, P., J. Steeves, B. Ashe, J. Firth, and B. Rabb. quakes, Tropical Cyclones (Wind and Storm 2014. “Climate Risks Study for Telecommunica- Surge), Floods, Tsunami, and Volcanic Erup- tions and Data Center Services.” Report for the tions.” Background paper for GAR15, CIMNE General Services Administration, Washington, DC. and INGENIAR Consortium, Barcelona and APEC (Asia-Pacific Economic Cooperation) Secretar- Bogotá. iat. 2013. “Economic Impact of Submarine Cable Cervigni, R., R. Liden, J. E. Neumann, and K. M. Disruptions.” APEC#213-SE-01.2, APEC Secre- Strzepek. 2015. Enhancing the Climate Resilience of tariat, Singapore. Africa’s Infrastructure: The Power and Water Sectors. Argyroudis, S., and A. M. Kaynia. 2014. “Fragility Washington, DC: World Bank. Functions of Highway and Railway Infrastruc- Chang, S. E., and N. Nojima. 2001. “Measuring ture.” In SYNER-G: Typology Definition and Fragil- Post-Disaster Transportation System Perfor- ity Functions for Physical Elements at Seismic Risk— mance: The 1995 Kobe Earthquake in Compar- Buildings, Lifelines, Transportation Networks, and ative Perspective.” Transportation Research Part A: Critical Facilities, edited by K. Pitilakis, H. Crowley, Policy and Practice 35 (6): 475–94. and A. M. Kaynia, 299–326. Dordrecht: Springer Damania, R., S. Desbureaux, M. Hyland, A. Islam, Netherlands. S. Moore, A.-S. Rodella, J. Russ, and E. Zaveri. NATURAL SHOCKS ARE A LEADING CAUSE OF INFRASTRUCTURE DISRUPTIONS AND DAMAGES 81 2017. Uncharted Waters: The New Economics of Water Homeland Security and Emergency Management 11 Scarcity and Variability. Washington, DC: World (2): 217–41. Bank. Kakderi, K., and S. Argyroudis. 2014. “Fragility Daniels, J. 2017. “Southern California Utilities Shut Functions of Water and Waste-Water Systems.” Off Power in ‘Extremely’ Windy Weather to Pre- In SYNER-G: Typology Definition and Fragility Func- vent Wildfires.” CNBC, December 13. tions for Physical Elements at Seismic Risk—Build- Dawson, R. J., D. Thompson, D. Johns, R. Wood, ings, Lifelines, Transportation Networks, and Criti- G. Darch, L. Chapman, P. N. Hughes, G. V. R. cal Facilities, edited by K. Pitilakis, H. Crowley, Watson, K. Paulson, S. Bell, S. N. Gosling, W. and A. M. Kaynia, 221–58. Dordrecht: Springer Powrie, and J. W. Hall. 2018. “A Systems Frame- Netherlands. work for National Assessment of Climate Risks Khan, I., F. Alam, and Q. Alam. 2013. “The Global to Infrastructure.” Philosophical Transactions of the Climate Change and Its Effect on Power Gener- Royal Society A: Mathematical, Physical, and Engineer- ation in Bangladesh.” Energy Policy 61 (October): ing Sciences 376 (2121): 20170298. 1460–70. Dhaka Tribune. 2019. “Nor’wester Knocks out 6,000 Koks, E., J. Rozenberg, C. Zorn, M. Tariverdi, Cell Towers.” April 1. http://www.dhakatribune M. Vousdoukas, S. A. Fraser, J. Hall, and S. Hal- .com/bangladesh/nation/2019/04/01/nor-wester legatte. 2019. “A Global Multi-Hazard Risk Anal- -knocks-out-6-000-cell-towers. ysis of Road and Railway Infrastructure Assets.” Doll, C., S. Klug, and R. Enei. 2014. “Large and Forthcoming in Nature Sustainability. Small Numbers: Options for Quantifying the Koetse, M. J., and P. Rietveld. 2009. “The Impact of Climate Change and Weather on Transport: An Costs of Extremes on Transport Now and in 40 Overview of Empirical Findings.” Transportation Years.” Natural Hazards 72 (1): 211–39. Research Part D: Transport and Environment 14 (3): Enei, R., C. Doll, S. Klug, I. Partzsch, N. Sedlacek, 205–21. J. Kiel, N. Nesterova, L. Rudzikaite, A. Papani- Kressig, A., L. Byers, J. Friedrich, T. Luo, and C. kolaou, and V. Mitsakis. 2011. Vulnerability of McCormick. 2018. “Water Stress Threatens Transport Systems: Main Report. Karlsruhe: Fraun- Nearly Half the World’s Thermal Power Plant hofer-Institute for Systems and Innovation Capacity.” World Resources Institute (blog), April Research. 11. https://www.wri.org/blog/2018/04/water Fu, G., L. Horrocks, and S. Winnie. 2016. “Explor- -stress-threatens-nearly-half-world-s-thermal ing Impact of Climate Change on UK’s ICT Infra- -power-plant-capacity. structure.” Infrastructure Asset Management 3 (1): Kwiatkowski, K., Y. Qiao, S. Young, E. Bell, and 42–52. R. Medina. 2019. “Quantifying Risk of Flood- Hu, X., R. Pant, C. Zorn, W. Lim, E. Koks, and Induced Bridge Failure in Mozambique Con- Z. Mao. 2019. “Methodology and Findings for sidering Climate Variability.” University of New the Exposure Analysis of the Chinese Wastewa- Hampshire. ter Sector to Flooding and Earthquakes Hazards.” Lehner, B., C. R. Liermann, C. Revenga, C. Background paper for this report, World Bank, Vörösmarty, B. Fekete, P. Crouzet, P. Döll, Washington, DC. M. Endejan, K. Frenken, J. Magome, C. Nilsson, ICF (Investment Climate Facility). 2019. “Strategic J. Robertson, R. Rödel, N. Sindorf, and D. Wisser. Assessment of the Climate Resilience of Dar es 2019. “Global Reservoir and Dam (GRanD) Data- Salaam Transport Infrastructure.” ICF, African base. Technical Documentation.” Global Water Development Bank, Abidjan. System Project, Bonn. IHA (International Hydropower Association). 2019. Ludvigsen, J., and R. Klæboe. 2014. “Extreme Hydropower Sector Climate Resilience Guide. London: Weather Impacts on Freight Railways in Europe.” IHA. Natural Hazards 70 (1): 767–87. Jonkeren, O., P. Rietveld, J. van Ommeren, and A. Luo, T., D. Krishnan, and S. Sen. 2018. “Parched te Linde. 2014. “Climate Change and Economic Power: Water Demands, Risks, and Opportunities Consequences for Inland Waterway Transport in for India’s Power Sector.” World Resources Insti- Europe.” Regional Environmental Change 14 (3): tute, Washington, DC. 953–65. Meijer, J. R., M. A. J. Huijbregts, K. C. G. J. Schot- Kadri, F., B. Birregah, and E. Châtelet. 2014. “The ten, and A. M. Schipper. 2018. “Global Patterns Impact of Natural Disasters on Critical Infrastruc- of Current and Future Road Infrastructure.” Envi- tures: A Domino Effect-Based Study.” Journal of ronmental Research Letters 13 (6): 064006. 82 LIFELINES Mideksa, T. K., and S. Kallbekken. 2010. “The Background paper for this report, World Bank, Impact of Climate Change on the Electricity Mar- Washington, DC. ket: A Review.” Energy Policy 38 (7): 3579–85. Rinaldi, S. M., J. P. Peerenboom, and T. K. Kelly Miyamoto International. 2019. “Overview of Engi- 2001. “Identifying, Understanding, and Analyz- neering Options for Increasing Infrastructure ing Critical Infrastructure Interdependencies.” Resilience.” World Bank, Washington, DC. IEEE Control Systems Magazine 21 (6): 11–25. Moss, R. E. S., E. M. Thompson, D. S. Kieffer, Rozenberg, J., X. Espinet Alegre, P. Avner, C. Fox, B. Tiwari, Y. M. A. Hashash, I. Acharya, B. R. S. Hallegatte, E. Koks, J. Rentschler, and M. Tari- Adhikari, D. Asimaki, K. B. Clahan, B. D. Collins, verdi. 2019. “From a Rocky Road to Smooth S. Dahal, R. W. Jobson, D. Khadka, A. McDonald, Sailing: Building Transport Resilience to Natu- C. L. Madugo, B. Mason, M. Pehlivan, D. Raya- ral Disasters.” Sector note for this report, World majhi, and S. Uprety. 2015. “Geotechnical Effects Bank, Washington, DC. of the 2015 Magnitude 7.8 Gorkha, Nepal, Earth- Sampson, C. C., A. M. Smith, P. B. Bates, J. C. quake and Aftershocks.” Seismological Research Neal, L. Alfieri, and J. E. Freer. 2015. “A High- Letters 86 (6): 1514–23. Resolution Global Flood Hazard Model.” Water Muis, S., M. Verlaan, H. C. Winsemius, J. C. J. H. Resources Research 51 (9): 7358–81. Aerts, and P. J. Ward. 2016. “A Global Reanalysis Schweikert, A. E., L. Nield, E. Otto, and M. Deinert. of Storm Surges and Extreme Sea Levels.” Nature 2019. “Resilience and Critical Power System Communications 7 (11969). Infrastructure: Lessons Learned from Natural Nicolas, C., E. Koks, A. Potter van Loon, C. Arderne, Disasters and Future Research Needs.” Back- C. Zorn, and S. Hallegatte. 2019. “Global Power ground paper for this report, World Bank, Wash- Sector Exposure and Risk Assessment to Natu- ington, DC. ral Disasters.” Background paper for this report, Scussolini, P., J. C. J. H. Aerts, B. Jongman, L. M. World Bank, Washington, DC. Bouwer, H. C. Winsemius, H. De Moel, and P. Nicolas, C., J. Rentschler, A. Potter van Loon, J. Ward. 2016. “FLOPROS: An Evolving Global S. Oguah, S. Schweikert, M. Deinert, E. Koks, Database of Flood Protection Standards.” Natu- C. Arderne, D. Cubas, J. Li, and E. Ichikawa. ral Hazards and Earth System Sciences 16 (May 3): 2019. “Stronger Power: Improving Power Sector 1049–61. Resilience to Natural Disasters.” Sector note for Stip, C., Z. Mao, L. Bonzanigo, G. Browder, and this report, World Bank, Washington, DC. J. Tracy. 2019. “The Resilience of Water Infra- Ouyang, M. 2014. “Review on Modeling and Sim- structure: Examples from Dams, Wastewater ulation of Interdependent Critical Infrastructure Treatment Plants, and Water Supply and Sanita- Systems.” Reliability Engineering and System Safety tion Systems.” Sector note for this report, World 121 (January): 43–60. Bank, Washington, DC. Panteli, M., and P. Mancarella 2015. “Influence of Tolman, H. L. 2009. “User Manual and System Extreme Weather and Climate Change on the Documentation of WAVEWATCH III TM Ver- Resilience of Power Systems: Impacts and Pos- sion 3.14.” Technical Note, MMAB Contribution sible Mitigation Strategies.” Electric Power Systems 276, U.S. Department of Commerce, National Research 127 (October): 259–70. Oceanic and Atmospheric Administration, Patt, A., S. Pfenninger, and J. Lilliestam. 2013. “Vul- National Weather Service, National Centers for nerability of Solar Energy Infrastructure and Environmental Prediction, Camp Springs, MD. Output to Climate Change.” Climatic Change 121 U.S. Department of Homeland Security. 2017. (1): 93–102. http://dx.doi.org/10.1007/s10584 “Threats to Undersea Cable Communications.” -013-0887-0. Public-Private Analytic Exchange Program, Ponemon Institute. 2016. “Cost of Data Center Out- Washington, DC, September. ages.” Ponemon Institute, Traverse City, MI. U.S. Department of Homeland Security and U.S. Rentschler, J., M. Obolensky, and M. Kornejew. Department of Energy. 2017. “Dams and Energy 2019. “Candle in the Wind? Energy System Resil- Sectors Interdependence Study: An Update to the ience to Natural Shocks.” Background paper for 2011 Study.” U.S Department of Homeland Secu- this report, World Bank, Washington, DC. rity and U.S. Department of Energy, Washington, Rentschler, J., J. Braese, N. Jones, and P. Avner. DC. 2019. “Three Feet Under: The Impacts of Flooding Van Vliet, M. T. H., J. Sheffield, D. Wiberg, and E. F. on Urban Jobs, Connectivity, and Infrastructure.” Wood. 2016. “Impacts of Recent Drought and NATURAL SHOCKS ARE A LEADING CAUSE OF INFRASTRUCTURE DISRUPTIONS AND DAMAGES 83 Warm Years on Water Resources and Electricity (blog), June 29. https://www.wri.org/blog/2017 Supply Worldwide.” Environmental Research Letters /06/no-water-no-power. 11 (12): 124021. Wender, B. A., M. G. Morgan, and K. J. Holmes Viguié, V., A. Lemonsu, S. Hallegatte, A.-L. Beau- 2017. “Enhancing the Resilience of Electricity lant, C. Marchadier, V. Masson, G. Pigeon, and Systems.” Engineering 3 (5): 580–82. J.-L. Salagnac. 2019. “Early Adaptation to Heat Worden, C. B., D. J. Wald, J. Sanborn, and E. M. Waves and Future Reduction of Air-Conditioning Thompson. 2017. “Development of an Open- Energy Use in Cities.” Unpublished manuscript. Source Hybrid Global Vs30 Model.” Seismological Vousdoukas, M. I., L. Mentaschi, E. Voukouvalas, M. Society of America Annual Meeting, Pasadena, Verlaan, S. Jevrejeva, L. P. Jackson, and L. Feyen. CA, April 21–23. 2018. “Global Probabilistic Projections of Extreme Yamazaki, D., D. Ikeshima, R. Tawatari, T. Yamagu- Sea Levels Show Intensification of Coastal Flood chi, F. O’Loughlin, J. C. Neal, C. C. Sampson, S. Hazard.” Nature Communications 9 (1): 2360. Kanae, and P. D. Bates. 2017. “A High-Accuracy Vousdoukas, M. I., E. Voukouvalas, L. Mentaschi, Map of Global Terrain Elevations.” Geophysical F. Dottori, A. Giardino, D. Bouziotas, A. Bianchi, Research Letters 44 (1): 5844–53. P. Salamon, and L. Feyen. 2016. “Developments Zhu, J., L. G. Baise, and E. M. Thompson. 2017. in Large-Scale Coastal Flood Hazard Mapping.” “An Updated Geospatial Liquefaction Model for Natural Hazards and Earth System Sciences 16 (8): Global Application.” Bulletin of the Seismological 1841–53. Society of America 107 (3): 1365–85. Wang, J., L. Schleifer, and L. Zhong. 2017. “No Water, No Power.” World Resources Institute From Micro to Macro: Local Disruptions Translate into Macroeconomic Impacts 5 T he severity of natural disasters is usually measured by the asset losses they provoke (Munich Re 2019; Swiss Re 2018). However, for many reasons such a metric is in- sufficient. The same loss can have very different impacts, depending on who is affected and their ability to cope with and recover from the loss (Hallegatte et al. 2016). But even without considering distributional impacts, asset losses do not capture the full macro- economic impact of a disaster. As shown by Hallegatte and Vogt-Schilb port, leading to a global supply disruption of (2016), $1 in asset loss can translate into more low-volume, high-value goods (such as elec- than $1 in output or consumption loss, espe- tronic components) and perishable goods (such cially if infrastructure assets are damaged. The as food and flowers) (BBC 2010). Similar indi- reason? Complementarities among assets in rect impacts affect households as well. When the economic system mean that the loss of one McCarty and Smith (2005) investigated the asset reduces the productivity of other assets impact of the 2004 hurricane season on house- (for example, the loss of a road makes a factory holds in Florida, they found that among the 21 less productive because workers cannot access percent of households forced to move after the it or goods cannot be delivered to or from it). disaster, 50 percent had to do so because of the These effects have been well identified theoret- disruption of utilities such as water supply; ically (Baqaee and Farhi 2017). only 37 percent had to move because of struc- The secondary effects of direct asset losses tural damage to their homes. on economic activities and output often repre- Various modeling studies have estimated the sent a large share of total disaster losses. Simu- macroeconomic impact of disaster-related infra- lations suggest that a major earthquake on the structure disruptions—see, for example, Cho et Hayward fault—a seismic fault line near San al. (2001); Gordon, Richardson, and Davis Francisco—could generate almost $40 billion (1998); Kroll et al. (1991); Rose and Wei (2013); in indirect losses and a drop in employment of and Tsuchiya, Tatano, and Okada (2007). By 36,700 employee-years, in addition to $115 simulating the reduction in the quality of ser- billion in asset losses (box 5.1). Disasters can vices delivered by the disrupted infrastructure, even reduce output without destroying any these studies were able to estimate the produc- assets. In 2010 the eruption of the Eyjafjalla- tion losses in various sectors of an economy and jökull Volcano in Iceland disrupted air trans- evaluate the macroeconomic losses using 85 86 LIFELINES BOX 5.1 When natural shocks affect firms, people suffer By destroying assets and infrastructure, earth- ance, and real estate (33 percent); and educa- quakes also destroy people’s jobs and economic tional services, health care, and social assistance opportunities. These effects are then transmit- (18 percent). However, the most vulnerable sec- ted across sectors and supply chains. A study tors in relative terms (largest losses relative to analyzing the likely consequences of a major their annual value added) are service industries earthquake in the San Francisco Bay Area in such as repair and maintenance services and California finds that the direct and indirect con- personal and laundry services, whose losses total sequences of such an event can be devastat- 74 percent of their annual value added. Some ing.  For example,  a major earthquake in the economic sectors are expected to increase their Hayward fault (moment-magnitude 7.2)  can production as a result of reconstruction demand, result in direct losses—losses associated with most notably the construction industry, with the cost of asset repair—valued on average an average value-added increase of $11 billion at $115 billion, or 15 percent of the Bay Area’s during the recovery period. gross domestic product (GDP). The majority of These changes in production in turn affect the damages (56 percent) occur in the housing employment and labor income across the Bay sector, followed by educational services, health Area. The average drop in employment is 36,700 care, and social assistance (7 percent); and man- employee-years over the recovery period, with ufacturing (6 percent). an initial drop of 8.7 percent of overall employ- Critically, the study finds that damage to infra- ment. The top industries affected by unemploy- structure and buildings in the private sector also ment are the service industries—in particular, causes significant indirect losses—mainly in the education, health care, and social assistance form of lower production as a result of damage (15,000 employee-years); professional and busi- to productive capital, supply constraints, and ness services (8,900 employee-years); and other changes in demand. The average losses in value services (8,700 employee-years). The effects on added total $39 billion, or an additional 5 per- unemployment are felt all across the Bay Area cent of the Bay Area’s GDP. (not just where the asset losses are concen- These losses accumulate over a recovery trated) because employment is related to the period of 10 years, but they are concentrated economic health of the entire region. Overall, this mainly in the first months and years following a study illustrates that it is essential to account for major shock. The industries suffering the largest the indirect consequences of infrastructure dis- absolute reduction in their value added are pro- ruption caused by natural disasters when quan- fessional and business services (37 percent of tifying the long-term impacts of disasters on total indirect losses), followed by finance, insur- households and individuals. Source: Markhvida et al. 2019. input-output or general equilibrium models. and Wei (2013) investigate the impact of a Such studies show that the impact of a disaster 90-day disruption at the twin seaports of Beau- can spread far beyond the businesses directly mont and Port Arthur, Texas, and find that such affected. Through input shortages, many more indirect losses alone could reduce regional gross firms suffer losses in production and sales, output by as much as $13 billion. resulting in reductions in workers’ incomes and In another study, Rose and Liao (2005) a drop in demand up the supply chain. Rose demonstrate how a major earthquake disrupt- FROM MICRO TO MACRO: LOCAL DISRUPTIONS TRANSLATE INTO MACROECONOMIC IMPACTS 87 ing the Portland water supply system would • What are the indirect impacts of disrupted change the composition of economic activity in infrastructure (such as on workers and jobs)? the affected regions. Several studies model the • What are the impacts on supply chains effect of blackouts on economic activity by trac- (suppliers, clients, and end users)? ing the initial impacts, such as damage to equip- • What adaptation strategies do firms use, ment and lost sales, through to further damage and what are their associated costs (such resulting from economic interdependencies as additional inventories, generators, own (Anderson, Santos, and Haimes 2007; Rose, water sources, and tanks)? Oladosu, and Liao 2007). These impacts, like those for disrupted transport infrastructure, The pilot survey was conducted in Tanzania include effects on firms up and down the supply for a sample of 800 firms, representing a wide chain (through the cancellation of orders and range of economic sectors. By comparing dis- lack of inputs), lower income for workers result- ruption levels during the dry and the rainy sea- ing in decreased consumption, and lower invest- sons, the survey was able to identify the role of ments because of the lower profitability of flooding in firm-level losses. affected firms. Rose, Oladosu, and Liao (2007) Overall, Tanzanian firms are incurring utili- estimate the total cost of a two-week blackout in zation losses of $670 million a year (or 1.8 per- Los Angeles at $2.8 billion, or 13 percent of the cent of the country’s GDP) from power and city’s total economic activity over that period. water outages and transport disruptions (figure This figure is, however, relatively limited, thanks 5.1).1 Power alone is responsible for $216 mil- to multiple “resilience factors.” For example, lion a year in losses. Of these losses, 47 percent some firms are able to find a substitute for elec- ($101 million, or 0.3 percent of GDP) are solely tricity or to reschedule production. due to power outages caused by rain and floods. The remaining 53 percent of utilization A SURVEY CONFIRMS THE COST losses are due to baseline power outages associ- OF NATURAL HAZARDS FOR FIRMS ated with causes other than rain and flooding THROUGH INFRASTRUCTURE (such as load shedding or equipment failures). DISRUPTIONS Although it is agreed that disruptions from nat- ural hazards represent a significant cost for FIGURE 5.1 Tanzanian firms report large firms and households, local studies are needed losses from infrastructure disruptions to provide a detailed assessment. But such sur- 350 325 veys are rare—and almost nonexistent in low- 300 Utilization losses (US$, millions) and middle-income countries—making it diffi- 250 46% cult for governments to assess the full economic 216 losses after a disaster or to identify and priori- 200 tize investments in more resilient infrastruc- 150 47% 127 ture. To address this gap, a dedicated question- 100 naire for firms was developed for this report, with the objective of providing insights on sev- 50 eral key questions: 0 Transport Power Water • What is the direct damage to firms from  Losses due to disruptions caused by rain and floods  Losses due to disruptions caused by other factors natural shocks (such as destroyed or dam- aged assets)? Source: Based on Rentschler, Kornejew, et al. 2019. 88 LIFELINES FIGURE 5.2 Floods in Kampala cause transport disruptions and congestion a. Travel times between firms in Inner Kampala b. Journey time increase due to flooding Frequency density Frequency density 5 15 25 35 45 0 50 100 150 200 250 300 350 400 Minutes Travel time increase (%) No flood 10-year flood 50-year flood Source: Rentschler, Braese, et al. 2019. Note: Curves show frequency densities that represent the distribution of all firm-to-firm travel times. Panel a shows the travel times between 400 firms surveyed in Kampala during road network disruptions due to urban flooding with 10- and 50-year return periods. Panel b shows the percentage increase in average travel times due to road network disruptions from urban flooding. For transport disruptions, about 46 percent of CONSEQUENCES SPREAD THROUGH utilization losses ($150 million, or 0.4 percent DOMESTIC AND INTERNATIONAL of GDP) are due to disruptions caused by rain SUPPLY CHAINS and floods. But the survey does not find that What about the role of supply chains? The rain and floods have a significant impact on the Tanzania survey also confirms the vital role of incidence of water supply disruptions. these chains. When firms were asked why Evidence from Kampala illustrates why they cannot deliver on time to their clients, floods have such significant impacts on firms, the most important factor cited by about a even though relatively few are flooded directly. third of all firms was delays in their supply By blocking road segments throughout the city, chain (figure 5.3). Interdependencies within floods significantly reduce the connectivity and across supply chains can magnify the eco- between firms and thus the ease with which nomic costs of a disaster. If a producer is hit by goods and services can be moved between a disaster and forced to interrupt its opera- them. For Kampala, Rentschler, Braese, et al. tions, customers may rapidly fall short of sup- (2019) estimate that a moderate flood increases ply, leading to disruptions that may spread fur- average travel times between firms by 54 per- ther down the chain.2 cent. Many firms are affected even more These effects are also observed in interna- severely: more than a quarter of firms would tional supply chains. In 2011 Thailand was face an increase in average travel time of from affected by the largest floods in 70 years. The 100 to 350 percent (figure 5.2). The prospect of country’s car manufacturing then fell by 50–80 such delays means that many firms will avoid percent. Strikingly, Toyota had the largest pro- undertaking trips altogether, resulting in duction loss of all carmakers, even though missed deliveries and halted production. In none of its plants were inundated. Its profit fact, just a few flooded intersections can affect loss, which amounted to over $1.35 billion, firms and their supply chains, and thus overall was triggered by the disruption of critical sup- economic activity. These results also indicate pliers in the flooded areas, which were unable that it does not take an extreme event to dis- to supply Toyota’s assembly lines (Haraguchi rupt supply chains significantly. and Lall 2015). Suppliers of damaged firms FROM MICRO TO MACRO: LOCAL DISRUPTIONS TRANSLATE INTO MACROECONOMIC IMPACTS 89 also may face sales losses, putting their FIGURE 5.3 Supply chain disruptions are the main reason for finances at risk. delivery delays Supply chain effects can cross borders and 35 have worldwide consequences. For example, a 32.0 consequence of the 2011 floods in Thailand 30 was a 30 percent decrease in the global produc- 24.9 25 tion of hard disk drives (HDDs) in the six Share of firms (%) months after the floods, causing a price spike 20 17.1 16.7 of between 50 and 100 percent (Haraguchi and 15 Lall 2015). This loss was caused not only by the flooding of HDD manufacturers in Thailand, 10 8.1 but also by the disruption of producers around 5 the world because of the missing parts from 1.0 0.3 Thailand (Chee Wai and Wongsurawat 2012). 0 s ns ny s s r s he Similarly, the 2011 Great Eastern Japan Earth- lay lay ge ge tio pa Ot ta ta de de up m ou ou co n No quake and the tsunami that followed were of isr ai er er td in ch w at ith Po W or ly global economic significance because their sw pp sp an Su m Tr le impacts spread well beyond the borders of ob Pr Japan (Boehm, Flaaen, and Pandalai-Nayar 2015; World Economic Forum 2012). Source: World Bank staff. The supply chain amplification of disasters has also been documented statistically beyond past decades. These corporate decisions have these specific case studies. Barrot and Sauvag- led to an unprecedented globalization and nat (2016) find that the sales of U.S. firms complexity of supply chains, resulting in firms affected by a natural disaster drop by about 5 becoming more specialized and interdependent percent, and the sales by their clients—even if (Baldwin and Lopez-Gonzalez 2015). Although not directly affected by the disaster—also fall by only a few firms will experience a disaster 3 percent up to 4 months after the event. directly, most firms are likely to be exposed to Kashiwagi, Matous, and Todo (2018) find that the indirect ripple effects of disasters. In other when the ripple effects of natural disasters words, supply chains globalize local disasters spread along supply chains within regions, the and generate systemic risks (Colon et al. 2017). large firms tend to be able to switch suppliers These risks are particularly hard to evaluate quickly, thereby containing the spread interna- because firms often lack a full understanding of tionally. However, studies also show that—sim- their own supply chains. Firms usually know ilar to the negative effects of disruptions—the their direct suppliers, but they often struggle to positive effects of postdisaster reconstruction keep track of their subsuppliers, from which subsidies can also propagate through supply about half of supply disruptions seem to origi- chains (Kashiwagi 2019; Kashiwagi and Todo nate (Business Continuity Institute 2014). 2019). Moreover, having a geographically Supply chain managers have to deal with diverse range of suppliers and clients can allevi- uncertainties, unknowns, and interdependent ate the indirect impacts of a disaster and accel- risks, making decision-making processes partic- erate recovery (Kashiwagi, Matous, and Todo ularly complex (Doroudi et al. 2018). 2018; Todo, Nakajima, and Matous 2015). Measures that firms commonly take to The risk of a wide-ranging spread of disrup- reduce costs and increase competitiveness can tions along supply chains are a by-product of also aggravate their supply chain risks. For the offshoring and outsourcing strategies of the example, reducing inventories and streamlin- 90 LIFELINES ing the supplier base are effective cost-cutting light on this issue, a new supply chain model measures that can be adapted to deal with fre- was developed for this report to evaluate the quent and lower-impact risks. However, firms impacts of transport disruptions on supply with low inventories and concentrated suppli- chains and household consumption in Tanza- ers are more exposed to low-probability and nia (Colon, Hallegatte, and Rozenberg 2019). It high-impact disasters because these strategies builds on Hallegatte (2013) and Henriet, Halle- reduce flexibility and backup capacity (Stecke gatte, and Tabourier (2012). and Kumar 2009). Similarly, custom-made The model maps the domestic and interna- supplies may help firms to offer innovative and tional supply chains of Tanzania onto its trans- distinctive products, but they increase the port network, using subnational and trade data domino effect when a disaster hits because (map 5.1, panel a). Firm-level data on coping they cannot be easily replaced by other suppli- strategies from the dedicated survey are used to ers (Barrot and Sauvagnat 2016). calibrate the model, including the level of reserve inventories or the number of suppliers. SUPPLY CHAIN SIMULATIONS The data clearly indicate that supply chains con- ENABLE BETTER MEASUREMENT OF nect firms not only across sectors but also across THE MACROECONOMIC IMPACTS OF the country and across borders. Physically, these DISASTERS connections take the form of freight flows on So how do supply chains and transport disrup- the road network between the main cities. Flows tions interact? The answer is key for assessing are particularly large around Dar es Salaam and the resilience of an economy. To shed more its port, which acts as a trade hub for shipments MAP 5.1 Mapping Tanzania’s supply chains onto its transport network (panel a) reveals the impact of transport disruptions on Tanzanian households (panel b) a. Weekly supply chain flows b. Household losses a. Weekly supply chain flows b. Household losses Mwanza Kenya Weekly supply Mwanza Kenya Household loss Rwanda chain flows Rwanda (% of national daily (US$, millions) consumption) Burundi 250 Burundi 8 Arusha 100 Arusha 4.2 10 0.5 Dodoma Dar es Dodoma Dar es Salaam Salaam Dem. Rep. Morogoro Dem. Rep. Morogoro of Congo of Congo Mbeya Mbeya Zambia Zambia Malawi Malawi Mozambique Mozambique Source: Colon, Hallegatte, and Rozenberg 2019. Note: Panel a maps weekly supply chain flows onto the road network. The width of the black lines is proportional to the monetary value of the flow. The widest lines are in the Dar es Salaam region and amount to $260 million a week. Panel b simulates the indirect impact of the 2016 Morogoro flood. The pink stars indicate the locations of disrupted roads. The size and color of the bubbles represent household losses, shown as a percentage of national daily household consumption. FROM MICRO TO MACRO: LOCAL DISRUPTIONS TRANSLATE INTO MACROECONOMIC IMPACTS 91 to and from neighboring landlocked countries FIGURE 5.4 Long-duration floods trigger disruptions in (including Burundi, the Democratic Republic of Tanzania, with cascading impacts on supply chains and Congo, Malawi, Rwanda, Uganda, and Zambia). households In monetary terms, these freight flows account Household consumption change (%) 0 for about 20 percent of the total flows. Exports and imports by Tanzanian firms also primarily transit through the port of Dar es Salaam, accounting for another 20 percent. –20 The model can be used to assess the conse- quences of a flood that in the spring of 2016 affected the Morogoro region, about 200 kilo- –40 meters west of Dar es Salaam. Disruptions 1 2 3 4 5 6 7 8 were long enough—about a month—to induce Time (weeks) shortages in local supply chains (map 5.1,  Agriculture panel b). Overall, the estimated indirect costs  Processed food and food services for households amount to about 0.5 percent of  Wholesale trade and retail annual consumption, mostly because of short- Source: Colon, Hallegatte, and Rozenberg 2019. The flood ages in three of the largest sectors in Tanzania: occurs from week 1 to week 4, but impacts on households con- tinue after the flood is over. agriculture, food (processed food and food-re- lated services), and wholesale and retail trade. Figure 5.4 depicts how these impacts evolve and the availability of financial resources after a through time and ripple across these sectors. disaster (see the discussion in the recommenda- First, consumption losses pile up during the tion chapters in part III). flood. Blocked shipments of agricultural prod- Sectors also differ in their vulnerability to ucts trigger production delays in the food sec- transport disruption. For example, because tor and induce product unavailability for agricultural products are primary products, wholesalers and retailers. After the flood, losses they are less dependent on suppliers, which remain sizable for another two weeks. In the reduces their vulnerability. Impacts on food flooded area, production recovery in agricul- products and manufacturing are, by contrast, ture and the food sectors is slowed down by magnified by supply chain issues. By applying missing inputs from wholesalers and retailers. this model, it is possible to assess the most vul- Particularly important is the fact that impacts nerable firms and municipalities, depending on households spread, with significant conse- not only on their location but also on their eco- quences far from the location of floods, such as nomic structure. It is then possible to target around Dar es Salam (map 5.1, panel b). interventions to strengthen the resilience of Applying this model allows disasters of dif- firms and supply chains where it matters the ferent magnitudes and locations to be simu- most, which can be far from where disruptions lated, revealing useful insights. For example, are the most likely to occur. simulating similar transport disruptions that Such analyses make it possible to assess the vary in duration shows that the macroeconomic relative importance of individual segments of impact increases nonlinearly with the duration infrastructure networks for enabling supply of the disruption. A four-week disruption is on chains and to identify the most vulnerable average 23 times costlier for households than a users of infrastructure services. By identifying two-week disruption. This result highlights the bottlenecks and vulnerability hotspots, they large benefit of responding rapidly to a disaster help to prioritize investments and develop and building back quickly, which depend on the resilience strategies—the topic of the next part systems in place for road system maintenance of this report. 92 LIFELINES NOTES Institute for Applied Systems Analysis (IIASA), 1. See chapter 2 for details on firms’ utilization Laxenburg, Austria. losses due to infrastructure disruptions. Colon, C., S. Hallegatte, and J. Rozenberg. 2019. 2. These ripple effects can even take place “Transportation and Supply Chain Resilience within a factory, if one segment of the pro- in the United Republic of Tanzania.” Back- duction process is impossible and therefore ground paper for this report, World Bank, interrupts the entire production. Washington, DC. Doroudi, R., R. Azghandi, Z. Feric, O. Mohad- desi, Y. Sun, J. Griffin, O. Ergun, D. Kaeli, P. REFERENCES Sequeira, S. Marsella, and C. Harteveld. 2018. Anderson, C. W., J. R. Santos, and Y. Y. Haimes. “An Integrated Simulation Framework for 2007. “A Risk-Based Input-Output Method- Examining Resiliency in Pharmaceutical Sup- ology for Measuring the Effects of the August ply Chains Considering Human Behaviours.” 2003 Northeast Blackout.” Economic Systems In Proceedings of the 2018 Winter Simulation Con- Research 19 (2): 183–204. ference, edited by M. Rabe, A. A. Juan, N. Mus- Baldwin, R., and J. Lopez-Gonzalez. 2015. tafee, A. Skoogh, S. Jain, and B. Johansson. “Supply-Chain Trade: A Portrait of Global Pat- Piscataway, NJ: IEEE Press. terns and Several Testable Hypotheses.” World Gordon, P., H. W. Richardson, and B. Davis. 1998. Economy 38 (11): 1682–721. “Transport-Related Impacts of the Northridge Baqaee, D. R., and E. Farhi. 2017. “The Macro- Earthquake.” National Emergency Training economic Impact of Microeconomic Shocks: Center, Emmitsburg, MD. Beyond Hulten’s Theorem.” 2017 Meeting Hallegatte, S. 2013. “Modeling the Roles of Het- Paper 479, Society for Economic Dynamics. erogeneity, Substitution, and Inventories in Barrot, J.-N., and J. Sauvagnat. 2016. “Input the Assessment of Natural Disaster Economic Specificity and the Propagation of Idiosyn- Costs.” Risk Analysis 34 (1): 152–67. cratic Shocks in Production Networks.” Quar- Hallegatte, S., and A. Vogt-Schilb. 2016. “Are terly Journal of Economics 131 (3): 1543–92. Losses from Natural Disasters More Than Just BBC (British Broadcasting Corporation). 2010. Asset Losses? The Role of Capital Aggregation, “Iceland Volcano: Nissan and BMW Suspend Sector Interactions, and Investment Behav- Some Production.” BBC News, April 20. iors.” Policy Research Working Paper 7885, Boehm, C., A. Flaaen, and N. Pandalai-Nayar. World Bank, Washington, DC. 2015. “Input Linkages and the Transmission Hallegatte, S., A. Vogt-Schilb, M. Bangalore, and of Shocks: Firm-Level Evidence from the 2011 J. Rozenberg. 2016. Unbreakable: Building the – To­ hoku Earthquake.” CES 15-28, Center for Resilience of the Poor in the Face of Natural Disas- Economic Studies, Washington, DC. ters. Washington, DC: World Bank. Business Continuity Institute. 2014. “Supply Haraguchi, M., and U. Lall. 2015. “Flood Risks Chain Resilience 2014: An International Sur- and Impacts: A Case Study of Thailand’s vey to Consider the Origin, Causes, and Con- Floods in 2011 and Research Questions for sequences of Supply Chain Disruption.” BCI, Supply Chain Decision Making.” International Caversham, U.K. Journal of Disaster Risk Reduction 14 (3): 256–72. Chee Wai, L., and W. Wongsurawat. 2013. “Crisis Henriet, F., S. Hallegatte, and L. Tabourier. 2012. Management: Western Digital’s 46‐Day Recov- “Firm-Network Characteristics and Economic ery from the 2011 Flood Disaster in Thailand.” Robustness to Natural Disasters.” Journal of Strategy & Leadership 41 (1): 34–38. Economic Dynamics and Control 36 (1): 150–67. Cho, S., P. Gordon, J. E. Moore II, H. W. Richard- Kashiwagi, Y. 2019. Enterprise Resilience to Disas- son, M. Shinozuka, and S. Chang. 2001. “Inte- ters: Who Needs Public Support? RIETI Discussion grating Transportation Network and Regional Paper 19-E-029. Tokyo: Research Institute on Economic Models to Estimate the Costs of a Economy, Trade, and Industry. Large Urban Earthquake.” Journal of Regional Kashiwagi, Y., P. Matous, and Y. Todo. 2018. Science 41 (1): 39–65. “International Propagation of Economic Colon, C., Å. Brännström, E. Rovenskaya, and Shocks through Global Supply Chains.” U. Dieckmann. 2017. “Fragmentation of Pro- WINPEC Working Paper E1810, Waseda Insti- duction Amplifies Systemic Risks in Sup- tute of Political Economy, Waseda University, ply Chains.” Interim Report, International Tokyo. FROM MICRO TO MACRO: LOCAL DISRUPTIONS TRANSLATE INTO MACROECONOMIC IMPACTS 93 Kashiwagi, Y., and Y. Todo. 2019. Propagation of Computable General Equilibrium Analysis of Positive Effects of Post-Disaster Policies through Sup- Water Service Disruptions.” Journal of Regional ply Chains: Evidence from the Great East Japan Science 45 (1): 75–112. Earthquake. RIETI Discussion Paper 19-E-030. Rose, A., G. Oladosu, and S.-Y. Liao. 2007. “Busi- Tokyo: Research Institute on Economy, Trade, ness Interruption Impacts of a Terrorist Attack and Industry. on the Electric Power System of Los Angeles: Kroll, C. A., J. D. Landis, Q. Shen, and S. Stryker. Customer Resilience to a Total Blackout.” Risk 1991. “Economic Impacts of the Loma Prieta Analysis 27 (3): 513–31. Earthquake: A Focus on Small Businesses.” Rose, A., and D. Wei. 2013. “Estimating the Berkeley Planning Journal 5 (1): 38–58  Economic Consequences of a Port Shutdown: Markhvida, M., B. Walsh, S. Hallegatte, and J. The Special Role of Resilience.” Economic Baker. 2019. “Well-Being Loss: A Comprehen- Systems Research 25 (2): 212–32. sive Metric for Household Disaster Resilience.” Stecke, K. E., and S. Kumar. 2009. “Sources of EarthArXiv. May 17. doi:10.31223/osf.io/6r93z. Supply Chain Disruptions, Factors That Breed McCarty, C., and S. K. Smith. 2005. “Florida’s Vulnerability, and Mitigating Strategies.” Jour- 2004 Hurricane Season: Local Effects.” Florida nal of Marketing Channels 16 (3): 193–226. Focus 1 (3): 1–13. Bureau of Economic and Swiss Re. 2018. “Preliminary Sigma Estimates for Business Research, University of Florida. 2018: Global Insured Losses of USD 79 Billion Munich Re. 2019. Natural Catastrophe Review 2018. Are Fourth Highest on Sigma Records.” Swiss Munich: Munich Re. Re, December 18. Rentschler, J., J. Braese, N. Jones, and P. Avner. Todo, Y., K. Nakajima, and P. Matous. 2015. 2019. “Three Feet Under: The Impact of “How Do Supply Chain Networks Affect Flooding on Urban Jobs, Connectivity, and the Resilience of Firms to Natural Disasters? Infrastructure.” Background paper for this Evidence from the Great East Japan Earth- report, World Bank, Washington, DC. quake.” Journal of Regional Science 55 (2): Rentschler, J., M. Kornejew, S. Hallegatte, M. 209–29. Obolensky, and J. Braese. 2019. “Underuti- Tsuchiya, S., H. Tatano, and N. Okada. 2007. lized Potential: The Business Costs of Unreli- “Economic Loss Assessment Due to Railroad able Infrastructure in Developing Countries.” and Highway Disruptions.” Economic Systems Background paper for this report, World Bank, Research 19 (2): 147–62. Washington, DC. World Economic Forum. 2012. New Models for Rose, A., and S.-Y. Liao. 2005. “Modeling Addressing Supply Chain and Transport Risk . Regional Economic Resilience to Disasters: A Geneva: World Economic Forum. PA RT A Matter of Design: Resilient II Infrastructure Is Cost-Effective P art I of this report highlights the high costs of infrastructure disrup- tions and damages—for infrastructure asset owners and governments, for firms, and for households—and the important role that natural hazards and climate change play in these costs. Part II investigates how these costs may be reduced through engineering and planning solutions that can help to make infrastructure more resilient. It does so by exploring the resilience of FIGURE PII.1 The resilience of infrastructure needs to be infrastructure at three levels (figure PII.1): considered at several overlapping and complementary levels Resilience of infrastructure assets. In the nar- 1.  rowest sense, resilient infrastructure re- High-quality infrastructure fers to assets such as roads, bridges, and power lines that can withstand external shocks, especially natural ones. Here, Resilience of infrastructure users the benefit of more resilient infrastruc- Resilient infrastructure reduces the impact of natural hazards on people and economies ture is a reduction in the life-cycle cost of assets. Resilience of infrastructure services Resilience of infrastructure services. Infra- 2.  Resilient infrastructure provides more reliable services structure systems are interconnected net- works, and the resilience of individual Resilience of assets is a poor proxy for the resilience infrastructure assets of services provided at the network level. Resilient infrastructure is less For infrastructure, a systemic approach to costly to maintain and repair resilience is preferable. At this level, the benefit of more resilient infrastructure is the provision of more reliable services. Resilience of infrastructure users. Eventually, what matters is the resilience of 3.  users. Infrastructure disruptions can be catastophic or more benign, depending on whether users—including people and supply chains—can cope with them. At this level, the benefit of more resilient infrastructure is a reduction in the total impact of natural hazards on people and economies. Resilience is one of the many determinants of high-quality infrastructure. However, integrating resilience into the design and implementation of infrastructure invest- ments not only helps to manage natural shocks but also complements the cost-effec- tiveness, efficiency, and quality of infrastructure services more generally. Following this framework, chapter 6 explores how infrastructure assets can be made more resilient (such as stronger bridges and better-designed power transmis- sion systems); provides an estimate of the additional cost of investing in more resilient assets, which is small compared with the cost of the assets; and offers a cost-benefit analysis of key options to increase resilience, which turn out to be cost-efficient. Then chapter 7 expands the analysis to examine the resilience of infra- structure services—demonstrating that the cost of resilience can be reduced by working at the network and system levels and considering nature-based solutions. Finally, chapter 8 explores the role of the users of infrastructure services and how their actions can contribute to more resilient economies and societies (such as through more resilient supply chains and business continuity plans). More Resilient Infrastructure Assets Are Cost-Effective 6 H ow do countries increase the resilience of their infrastructure systems? One solu- tion is to build assets that can withstand bigger shocks, such as cell phone towers with deeper foundations and roads with larger culverts. Doing so can prevent damage from natural hazards and generate significant benefits in terms of lower repair costs and maintenance needs over the life cycle of the asset. But to be resilient, assets not only need to be strong; they also need to be well maintained, which requires a steady flow of resources as well as processes and systems. This chapter discusses the options for enhanced for power plants, windmills, or water treatment resilience at the asset level. Using a set of sce- plants are often needed to protect them against narios of infrastructure investments developed earthquake liquefaction. Better materials for by Rozenberg and Fay (2019), it also assesses wind turbines, cell phone towers, and transmis- the additional costs of making all new infra- sion and distribution systems can increase their structure assets more resilient in low- and mid- resistance to strong winds and extend their life- dle-income countries. It finds that the cost of time. Increasing the redundancy of compo- building the resilience of infrastructure assets nents of water and wastewater treatment plants in these countries is small compared with total by adding backup components can improve the infrastructure needs, provided the right data performance of plants during earthquakes. and approaches are available. Building resil- Building higher dikes around water treatment ience does not affect current affordability chal- plants and nuclear plants is the best option for lenges and is robust and cost-effective. protecting them against floods. In a review performed for this report, Miya- THE ADDITIONAL UP-FRONT COST moto International (2019) presents a high-level OF MORE RESILIENT ASSETS assessment of the costs and benefits of these var- DEPENDS ON THE ASSET AND THE ious technical and engineering options (see HAZARD Appendix A for an overview of these options Interventions to develop more resilient assets and their performance). The additional cost of include using alternative materials, digging making assets stronger in the face of natural haz- deeper foundations, elevating assets, building ards depends on the hazard and the type of asset. flood protection around the asset, or adding Increasing the flood resilience of a road through redundant components. Deeper foundations bigger drainage pipes or trenches requires a 97 98 LIFELINES small percentage of the road’s construction cost, construction technology, some low-cost tech- while increasing the flood resilience of a railway nologies perform better than traditional by elevating it requires 50 percent of its costs. approaches. Meanwhile, advanced materials Similarly, protecting a hydropower plant against and methods are making infrastructure both earthquakes by installing the proper anchorage less expensive and more climate-resilient. One and seismic components requires 20 percent of example is modular bridge solutions that its construction cost, whereas protecting a encase the deck structure of a bridge in stain- hydropower plant against flooding through big- less steel. This approach results in a signifi- ger spillway capacity requires 3 percent of its cost cantly longer design life of up to 100 years with (Miyamoto International 2019). In Puerto Rico, lower maintenance costs—a performance well a study of the recovery after Hurricanes Irma beyond that achieved with the traditional in and Maria finds that the cost of building back situ reinforced concrete. Construction costs are better, when compared with baseline estimates, also lower because a standardized formwork varies greatly in magnitude, depending on the (including reinforcement) can be delivered to a component and the hazard against which it site in a container, with deck casting conducted should be protected. There is a 3–40 percent in a single pour, as opposed to the longer times increase in the cost to upgrade transmission and and complex formwork needed for traditional distribution infrastructure to withstand category in situ structures (World Bank 2017). 3 hurricanes and a 24–70 percent increase in Improved quality control is required to cost to upgrade to withstand category 4 hurri- ensure that an asset is actually built and main- canes (130 miles per hour sustained wind tained to expected standards. Miyamoto Inter- speeds). When wooden power poles (low wind national (2019) estimates that this quality con- speed design) are compared with tubular steel trol costs from 1 percent to 5 percent of the poles, for example, the cost may differ by as value for most assets and hazards, but it can much as 200 percent (Schweikert et al. 2019). cost up to 15 percent to ensure that drainage Some resilience-building interventions can systems can cope with earthquake motion and even lower the cost of assets. With advances in highway systems can cope with flooding. This BOX 6.1 Infrastructure unit costs vary from country to country Infrastructure unit costs vary widely across Democratic Republic to $65,000 per kilometer countries and over time. For example, the unit in Armenia (World Bank 2018). Unit costs also cost of sewage collection and treatment, as vary within countries. The unit cost of dikes var- examined by Hutton and Varughese (2016), is ies between $6 million and $17 million per meter less than $100 in Guinea, Nepal, and Somalia, height and kilometer width for urban areas in the but more than $1,000 in Costa Rica, Papua Netherlands, the United Kingdom, and Vietnam New Guinea, and Sudan. Similar spreads exist (Nicholls et al. 2019). for all of the technologies considered for water Many factors can explain these spreads—from and sanitation infrastructure. For rural roads, variations in the cost of local labor and materi- a single surface treatment can cost anywhere als to vast differences in the efficiency of public from $10,000 per kilometer in the Lao People’s spending, the prevalence of corruption, and the (Box continues next page) MORE RESILIENT INFRASTRUCTURE ASSETS ARE COST-EFFECTIVE 99 BOX 6.1 Infrastructure unit costs vary from country to country (continued) lack of competition in public procurement. In the competitive procurement procedures led to a sub- road sector, for example, collusion can increase the stantial reduction in electricity prices, whereas in per-kilometer cost of building a road by as much Pakistan, it saved more than Rs 187 million ($3.1 as 40 percent (Messick 2011). In South Africa, the million) for the Karachi Water and Sewerage Board difference between the price charged by a cement (World Bank and OECD 2017). cartel during collusive and noncollusive periods Although the lack of efficiency and competition was 7.5–9.7 percent. The total savings to South in public procurement can explain a large share of African customers from the breakup of the car- the spread in unit costs, understanding why build- tel was from $79 million to $100 million between ing infrastructure is far more expensive in some 2010 and 2013 (World Bank and OECD 2017). In countries than others would require extensive Bangladesh, the introduction of transparent and analysis that is beyond the scope of this report. quality control would accompany the good sion and distribution infrastructure resulted in procurement practices that are key to lower a $30 million to $50 million reduction in direct infrastructure construction costs (box 6.1). asset replacement costs (Kestrel Group 2011). Recently, the World Bank conducted a study THE ADDITIONAL UP-FRONT COST analyzing the impact of climate risks on the OF MORE RESILIENT ASSETS planned energy system expansion in Bangla- COULD BE OFFSET BY LOWER desh, a country highly vulnerable to climate MAINTENANCE AND REPAIR COSTS change. The analysis determined that account- The decision to invest more up-front in making ing for climate change in the design increases infrastructure assets more resilient should capital requirements by $560 million for addi- depend on many criteria, including the current tional flood protection but could save up to and future exposure of the asset, the conse- $1.6 billion (Oguah and Khosla 2017). quences of failure compared with the level of The use of earthquake-resistant pipes for risk acceptable to users, and the life-cycle cost water supply systems would pay off in areas savings generated by the higher up-front cost. exposed to earthquakes. A pilot project under- More resilient energy systems would reduce taken in Los Angeles, by the Los Angeles life-cycle costs. Schweikert et al. (2019) find Department of Water and Power revealed the that above-ground transmission systems are benefits of making up-front improvements the energy system component most commonly (Davis and Castruita 2013). The 1994 North­ affected by wind, debris, ice, fires, floods, ridge Earthquake caused numerous failures in earthquakes, and landslides. Wires buried the network, leading to repair costs of around below ground are affected by flooding, lique- $41 million. By contrast, the earthquake- faction, and landslides, but they are much less resistant ductile iron pipes (ERDIPs) used in vulnerable overall than those above ground. In Japan have survived many large earthquakes New Zealand, case studies following the earth- and have sustained several meters of perma- quakes in 2010–11 highlight the value of pre- nent ground deformation. Replacing the old emptive investment in transmission and distri- piping system in Los Angeles with ERDIPs bution infrastructure. According to the increased the total cost of the pilot project by estimates, $6 million spent to harden transmis- about 20 percent. 100 LIFELINES IMPROVING MAINTENANCE AND losses because of an effective strategic mainte- OPERATIONS IS AN OPTION FOR nance plan (European Union 2015). In addi- BOOSTING RESILIENCE AND tion, maintenance is critical for ensuring that REDUCING COSTS assets can withstand extreme events. The Improving maintenance and operations is a World Bank (2017) argues that better asset no-regret option for boosting the resilience of management systems and better maintenance infrastructure assets while reducing overall should be the number one priority for small costs. Rozenberg and Fay (2019) find that, island developing states in order to increase the without good maintenance, infrastructure cap- resilience of their transport systems. The report ital costs could increase 50 percent in the trans- finds that improved road maintenance could port sector and more than 60 percent in the reduce asset losses by 12 percent in Belize and water sector. An analysis of member countries 18 percent in Tonga. of the Organisation for Economic Co-operation For energy systems, good maintenance of and Development performed for this report the vegetation on each side of power transmis- suggests that every additional $1 spent on road sion lines is crucial to reducing vulnerability maintenance saves on average $1.50 in new to strong winds. Such maintenance requires investments, making better maintenance a very easements of 20–100 meters (figure 6.1). As cost-effective option (Kornejew, Rentschler, described in chapter 4, power outages during and Hallegatte 2019). storms occur especially in areas with forest There is indeed strong evidence that good cover. Indeed, during storms, flying debris and maintenance increases the lifetime of assets. In vegetation are the primary causes of pole dam- Salzburg, most water pipelines are more than age, not the strong winds themselves. There- 100 years old, but they suffer very low water fore, reinforcing poles is less efficient than FIGURE 6.1 Clearing vegetation around transmission and subtransmission electricity networks requires an easement Risk management zone Risk management zone Maximum height Area to be cleared for power lines (typically 20–100 meters) MORE RESILIENT INFRASTRUCTURE ASSETS ARE COST-EFFECTIVE 101 trimming trees. In September 2017, Hurricanes $5.8 billion a year, of which $2.6 billion are Irma and Maria severely damaged the power commercial losses. According to Kingdom, grid in Puerto Rico, largely because of trees fall- Liemberger, and Marin (2006, 4), it is “not ing on the transmission lines. As a result, 100 unrealistic to expect that the high levels of phys- percent of Puerto Rico Electric Power Author- ical losses could be reduced by half” through ity customers lost power for more than a week improved leak detection, pipe replacement, and after the storm, and the slow pace of recovery maintenance, thereby saving 8 billion cubic left many customers in the dark for several meters of treated water a year. Such programs months (U.S. Department of Energy 2018). lead to better-quality services, higher utility rev- Good forest maintenance can also prevent enues, and a positive financial flow that enables wildfires. Wildfires are a unique threat to trans- future investment in rehabilitation and mainte- mission and distribution infrastructure. Various nance, which in turn enhances resilience. case studies illustrate that during high-risk con- New technology can be deployed to improve ditions (droughts, high temperatures, high maintenance at a low cost. Sensors with telem- winds), curtailments are used to reduce the risk etry are already being deployed to monitor pres- of transmission infrastructure causing a wild- sure and flow, minimizing losses and improving fire. The potential risk was illustrated in Califor- system maintenance. The ePulse system was nia in 2007, when San Diego Gas and Electric used in Washington, DC, during pipe replace- was found liable for causing three fires that led ment works. Condition assessment found that to three deaths and the destruction of 1,300 32 kilometers of pipe were in good condition, homes. The utility ultimately paid out $2 billion numerous leaks were located, and $14 million in settlements (Daniels 2017). Recent wildfires in investments were saved. Miniaturized robots have put the large utility Pacific Gas and Electric are also being tested for deployment in pipes to under scrutiny due to $10 billion in liabilities identify leaks. Fiber-optic cable can be used to from fires in 2017 and unknown amounts from detect very small leaks by measuring variations fires in 2018 (McNeely 2018). in the signal in an external fiber, before the In water supply networks, good mainte- leaks develop into larger leaks and burst a pipe. nance reduces water losses. Lack of mainte- Finally, regular cleaning of canals and drain- nance often leads to deterioration of pipes and age systems is essential for ensuring the reli- failure of valves, which in turn leads to physical ability of flood protection systems. In many losses in the distribution system called nonreve- low- and middle-income countries, the current nue water. A 2006 study estimates that every flood protection systems do not deliver the year more than 32 billion cubic meters of intended protection, because canals and drain- treated water physically leak from the world’s age pipes are clogged by solid waste. Long-term urban water supply systems, with half of these solutions have to include solid waste manage- losses in low- and middle-income countries ment, but regular cleaning of canals would also (Kingdom, Liemberger, and Marin 2006). In increase the efficiency of the system. addition, when maintenance is irregular, a water system is less likely to be inspected and THE COST OF INCREASING thus well known by technicians, increasing the RESILIENCE DEPENDS ON THE likelihood that illegal connections will go unno- ABILITY TO SPATIALLY TARGET ticed and cause commercial losses (water that is STRENGTHENING treated and delivered to users but not billed). How much do low- and middle-income coun- The same study estimates total losses at 16 bil- tries need to spend on infrastructure to achieve lion cubic meters a year globally. In low- and their development goals? A new study by middle-income countries, the estimated loss is Rozenberg and Fay (2019) estimates that it 102 LIFELINES would take between 2 percent and 8 percent of sion) and spending efficiency (box 6.2). The low- and middle-income countries’ gross next question then becomes, by how much domestic product (GDP), depending on the would estimates change if infrastructure sys- countries’ objectives (in terms of service provi- tems were designed and built in a more resilient BOX 6.2 Large investments in infrastructure will be necessary to close the service gap In an effort to shift the debate on infrastructure depend on smart policies and good planning. investment needs away from spending more and Countries would take long-term climate goals toward spending better on the right objectives, into account now to avoid expensive stranded a recent study by Rozenberg and Fay (2019) assets later; they would combine transport plan- offers a new way forward. They use a systematic ning with land use planning, resulting in denser approach to estimate the funding needs (capi- cities and cheaper and more reliable public trans- tal and operations and maintenance) for closing port; and they would develop reliable railway sys- the service gap in water and sanitation, transport, tems that freight haulers would find attractive. electricity, irrigation, and flood protection by 2030. Decentralized technologies, such as minigrids for (Telecommunications is not included in their analy- electricity and water purification systems pow- sis because it is mostly privately funded.) ered by renewable energy, would be deployed in They estimate that new infrastructure could rural areas. cost low- and middle-income countries between 2 However, improving services requires much percent and 8 percent of their GDP a year to 2030, more than capital expenditures. Success will depending on the quality and quantity of infra- depend on ensuring a steady flow of resources structure services sought and the spending effi- for operations and maintenance. In the preferred ciency achieved to reach this goal (table B6.2.1). scenario, low- and middle-income countries would Moreover, with the right policies, investments of need to spend 2.7 percent of GDP a year to main- 4.5 percent of GDP could enable low- and middle- tain their existing and new infrastructure, in addi- income countries to achieve the infrastructure- tion to the 4.5 percent of GDP in new capital (table related Sustainable Development Goals and stay B6.2.1). Meanwhile, good maintenance generates on track to full decarbonization by the second half substantial savings, reducing the total life-cycle of the century. cost of transport and water and sanitation infra- The ambitious goals and high efficiency structure by more than 50 percent. of Rozenberg and Fay’s “preferred scenario” TABLE B6.2.1 With the right policies in place, investments of 4.5 percent of GDP in infrastructure may be needed Infrastructure spending on capital and maintenance needs in low- and middle-income countries between 2015 and 2030, by sector Share of GDP (%) US$ (billions) Sector Capital Maintenance Capital Maintenance Electricity 2.2 0.6 780 210 Transport 1.3 1.3 420 460 Water and sanitation 0.55 0.75 200 70 Flood protection 0.32 0.07 100 20 Irrigation 0.13 — 50 — Total 4.5 2.7 1,550 760 Source: Rozenberg and Fay 2019. Note: — = maintenance costs of irrigation infrastructure are included in the capital costs. MORE RESILIENT INFRASTRUCTURE ASSETS ARE COST-EFFECTIVE 103 manner, through the technical and engineering to the full network. In both scenarios, it is solutions identified in Miyamoto International assumed that future infrastructure assets are (2019)? These options—listed in Appendix A— exposed in a similar fashion to the existing have been selected because they are realistic infrastructure in each region (in other words, and can make assets more resilient in low- and on average, the space available for future infra- middle-income countries. However, they are structure location is exposed to the same level not necessarily the ones that will reduce risk of hazards as the space already used). Results the most, and they do not guarantee that assets for these two scenarios are compared here for cannot be damaged by natural hazards. Many three infrastructure systems: power, transport, high-income countries, like Japan, implement and water and sanitation. technical solutions that go beyond—and are more expensive than—the set of solutions con- Power sidered in this analysis. In the power sector, baseline investment needs, Because the incremental costs of making assuming current resilience levels, would range assets more resilient can be significant, it is from $298 billion to $1 trillion a year in low- important to target strengthening to areas and middle-income countries between 2015 where exposure to natural disasters is high. and 2030. This depends on energy efficiency Ideally, infrastructure standards and codes and the timing of the transition toward should be asset and localization specific. Road carbon-free power generation (which creates designs should account for the hydrological stranded assets, such as coal power plants that and hydraulic data and climate model results at need to be decommissioned before the end of the location of the road in order to account for their lifetime). In addition, between $106 bil- the range of impacts that climate change can lion and $282 billion a year would be needed have on the probability of flood events in the for maintenance. How would those costs future. For electricity distribution systems— increase to make power systems more because they are particularly vulnerable to resilient? wind from storms, hurricanes, and typhoons— historical data and model results on wind • Scenario 1. If only exposed assets are made velocities with hour-level resolution could be more resilient to hazards, the incremental cost used for a geospatial analysis to inform risk and would rise from $9 billion to $27 billion a design standards in many regions of the world. year, which represents a 3 percent cost Data on water availability for cooling is also increase on average across the spending range central to planning for electricity generation. and a 6 percent cost increase in the most The analysis used in this report explores two expensive case. These investments would extreme scenarios in terms of the knowledge reduce the damage risk by a factor of two to on the spatial distribution of natural hazards three for new infrastructure assets. and the ability to target strengthening to the • Scenario 2. If, instead, all new power assets places exposed to them (Hallegatte et al. 2019). were made more resilient to wind, floods, In the first scenario, it is assumed that the loca- and earthquakes, because data on natural tion and intensity of the hazard are perfectly hazards are not available, then an addi- known, now and in the future, and that differ- tional $96 billion to $296 billion a year ent standards can be applied in different loca- would be needed. This is a 30 percent tions, depending on the level of risk. In the sec- increase in capital cost on average over the ond scenario, it is assumed that the hazard is spending range and a 10-fold increase com- unknown, or is too uncertain to be acted on, pared with that in the scenario for which and that a uniform standard has to be applied hazard data are available. 104 LIFELINES Transport ture is less than 1—suggesting that, in the In the transport sector, baseline investment absence of hazard data, it is not cost-effec- needs—with current resilience standards— tive to strengthen all assets in transport could range from $157 billion to $1.1 trillion a systems. year between 2015 and 2030 in low- and middle-income countries. The exact value Water and sanitation would depend on the choice of mode (such as In the water supply and sanitation sector, the personal cars versus public transit in cities) and cost of providing universal access to safe water on the policies put in place to encourage and sanitation in low- and middle-income switching to rail and public transport. In addi- countries by 2030, with the current resilience tion, between $550 billion and $700 billion level, would range from $116 billion to $229 would be needed every year to maintain the billion a year for capital investments and from existing and new transport infrastructure in $32 billion to $69 billion a year for mainte- low- and middle-income countries by 2030, nance. How would those costs increase to make bringing the total annual spending needs to water and sanitation systems more resilient? between $700 billion and $1.8 trillion. How would those costs increase to make transport • Scenario 1. The cost of protecting new systems more resilient? exposed water assets would be between $0.9 billion and $2.3 billion a year (assum- • Scenario 1. The incremental cost of making ing that, on average, water and sanitation new exposed transport assets more resilient infrastructure has the same exposure to to floods and landslides lies between $860 earthquakes as transport and power assets). million and $35 billion. This is a 0.6 percent These investments would reduce the risk increase in cost, on average, across the of damage to new infrastructure by 50 spending range and potentially a 5 percent percent. increase in the most expensive case. These • Scenario 2. Instead, if all water assets were investments would reduce the risk of dam- made more resilient to floods, an additional age for new infrastructure by a factor of $2 billion to $5 billion a year would be two. According to Koks et al. (2019), these required. This estimate represents a 1.1 per- investments would pay for themselves cent to 2.2 percent increase in capital costs. through the lower repair costs for about 60 Increasing the resilience of these assets to percent of roads exposed to a 1/100-year earthquakes would require an additional flood event (4.5 percent of the network). $8 billion to $20 billion a year (or between • Scenario 2. If, instead, all new transport 5 percent and 9 percent of capital invest- assets were made more resilient to floods ment needs). and landslides regardless of their exposure, the incremental cost would range from SUMMING UP $8 billion to $350 billion a year. This is a Unfortunately, similar estimates were not pos- 5.5 percent increase in cost, on average, sible for telecommunications, and these num- across the spending range, but potentially bers only include low- and middle-income a 17 percent increase in the most expensive countries. Yet this exercise provides three case for many rail investments. Consider- important insights. ing the benefit of upgrades to be only the First, there is huge value in knowing the spatial cost of repairs after a disaster, Koks et al. distribution of natural hazards, including climate (2019) estimate that the benefit-cost ratio change. Focusing the strengthening of infrastruc- of strengthening all transport infrastruc- ture assets on exposed assets reduces total MORE RESILIENT INFRASTRUCTURE ASSETS ARE COST-EFFECTIVE 105 annual costs from between $120 billion and FIGURE 6.2 The incremental cost of increasing the resilience $670 billion to between $11 billion and $65 bil- of future infrastructure investments is significantly reduced if lion (figure 6.2). The savings from targeting the asset exposure is known infrastructure assets most exposed appear to be 700 orders of magnitude larger than the costs of data 600 Average annual cost (US$ billions) collection and modeling that would be required to improve knowledge of current and future 500 hazards. Indeed, a global platform like Think 400 Hazard! (Fraser 2017), which compiled most of 300 the hazards data that were used for the risk assessments in this report (see chapter 4), costs 200 a few million dollars to create and maintain. At 100 most, creating high-resolution digital elevation 0 models and hazard maps for all cities in low- Power Transport Water and Total and middle-income countries would cost a few sanitation hundred million dollars (Croneborg et al. 2015).  Upgrade all assets  Upgrade exposed assets Second, the cost of building the resilience of infra- Source: Hallegatte et al. 2019. structure assets in low- and middle-income countries Note: “Cost ” here is the average annual capital investment cost between 2015 and is small compared with total infrastructure needs, 2030. The circles represent the median, and the vertical bars represent the full provided the right data and approaches are avail- range of possible incremental costs. able. Increasing the resilience of only the assets exposed to hazards would increase investment the cost of infrastructure resilience and the needs in power, transport, and water and sani- benefits in terms of both avoided repairs and tation by between $11 billion and $65 billion a disruptions for households and firms make it year. While not negligible, this is only around 3 difficult to provide a single estimate for the percent of baseline infrastructure investment benefit-cost ratio of strengthening exposed needs and less than 0.1 percent of the GDP of infrastructure assets. To manage this uncer- low- and middle-income countries. Therefore, tainty, an analysis performed for this report making infrastructure more resilient does not explores the benefit-cost ratio in 3,000 scenar- affect current affordability challenges for new ios (Hallegatte et al. 2019). These scenarios infrastructure, and it would decrease the risk of combine uncertainties regarding the cost of the damage for new infrastructure by a factor of technical options to increase resilience, the cur- between two and three. rent and future exposure of infrastructure Third, these investments to increase infrastructure assets to natural hazards, the current and future resilience are cost-effective. In transport, where an role of natural hazards in infrastructure disrup- asset-per-asset cost-benefit analysis is possible, tions, and their full social costs to firms and 60 percent of the exposed assets are worth households. In addition, the analysis considers strengthening, even if the only benefits various assumptions about economic growth, included are the avoided repair costs. However, the depreciation rate of the existing infrastruc- avoided repairs are far from being the only ture stock, and the impacts of climate change benefits of strengthened assets (chapters 2, 3, on natural hazards. and 5 of this report review the costs associated Results suggest that strengthening infra- with disruptions). structure assets exposed to natural hazards is a What are the returns on investments for very robust investment. The benefit-cost ratio making exposed infrastructure more resilient to is higher than 1 in 96 percent of the scenarios, natural disasters? The uncertainty pertaining to larger than 2 in 77 percent of them, and higher 106 LIFELINES FIGURE 6.3 Increasing the resilience of future infrastructure ure 6.3). Moreover, climate change makes the investments is cost-efficient—even more so with climate strengthening of infrastructure assets even change more important. Without climate change, the 0.25 median benefit-cost ratio would be equal to 2, but it is doubled when climate change is considered. 0.20 The 4 percent of scenarios with a benefit- cost ratio below 1—meaning that strengthen- 0.15 ing new assets is not desirable—are scenarios in which all estimates are consistently biased in 0.10 the same direction. In other words, the cost of strengthening is at the top of the range, the impact of hazards on infrastructure assets and 0.05 disruptions is at the bottom of the range, the socioeconomic consequences of disruptions are 0 the lowest, and climate change barely affects 0 5 10 15 20 25 30 natural hazards. Overall, strengthening infra- Net present value of more resilient infrastructure (US$, trillions) structure assets seems a very robust and attrac-  No climate change  With climate change tive solution: it is very likely to be cost- Source: Hallegatte et al. 2019. effective, has a high likelihood of generating Note: A net present value higher than 0 means that benefits are higher than costs. very large benefits, and cannot generate mas- sive losses, even in the worst-case scenarios. FIGURE 6.4 The cost of inaction increases rapidly—even more The urgency of designing better infrastruc- so with climate change ture is also evident in the simulations (figure 0.7 6.4). In 93 percent of the scenarios, it is costly to delay action from 2020 to 2030. The median 0.6 cost of delaying action to 2030 is $1.0 trillion. The only scenarios in which delaying action is 0.5 beneficial are scenarios in which strengthening infrastructure assets has a benefit-cost ratio 0.4 below, or very close to, 1. Here again, climate 0.3 change makes action more urgent: climate change almost doubles the median cost of 0.2 delaying action by 10 years. This analysis underestimates the desirability 0.1 of investing in more robust infrastructure assets. 0 The options considered here to strengthen –1 0 1 2 3 4 5 6 infrastructure assets against natural hazards Cost of delaying action to 2030 (US$, trillion) would also make them more resistant to other  No climate change  With climate change types of shocks, such as technical failures. Source: Hallegatte et al. 2019. Thus, there are large co-benefits in terms of avoided disruptions, going beyond the haz- than 6 in 25 percent of them (Hallegatte et al. ard-related ones explored here. However, as 2019). The net present value of these invest- discussed later in this report, infrastructure ments, over the lifetime of new infrastructure owners and operators often bear only a frac- assets, exceeds $2 trillion in 75 percent of the tion of the social cost of infrastructure disrup- scenarios and $4.2 trillion in half of them (fig- tions and damages. As a result, their incentive MORE RESILIENT INFRASTRUCTURE ASSETS ARE COST-EFFECTIVE 107 to build resilient assets is largely reduced, Can Help—A Look at Performance-Based Service unless specific regulations and policies are Contracting.” Water Supply and Sanitation Sec- implemented, a subject covered in more detail tor Board Discussion Paper 8, World Bank, Wash- ington, DC. in part III of this report. Koks, E., J. Rozenberg, C. Zorn, M. Tariverdi, M. This chapter has explored how to make Vousdoukas, S. A. Fraser, J. Hall, and S. Halle- infrastructure systems more resilient through gatte. 2019. “A Global Multi-Hazard Risk Anal- more robust assets, leaving aside all system- ysis of Road and Railway Infrastructure Assets.” level instruments to build resilience (or even Forthcoming in Nature Sustainability. Kornejew, M., J. Rentschler, and S. Hallegatte. 2019. something as simple as building assets in safer “Well Spent: How Governance Determines the areas). The next chapter explores how looking Effectiveness of Infrastructure Investments.” at systems and services instead of assets opens Background paper for this report, World Bank, new ways to build resilience at a low cost. Washington, DC. McNeely, A. 2018. “PG&E Credit Cut to Brink of REFERENCES Junk by Moody’s on Wildfire Risk.” Bloomberg Croneborg, L., K. Saito, M. Matera, D. McKeown, News, November 15. and J. van Aardt. 2015. “Digital Elevation Models: Messick, R. 2011. “Curbing Fraud, Corruption, and A Guidance Note on How Digital Elevation Mod- Collusion in the Roads Sector.” World Bank, els Are Created and Used—Includes Key Defini- Washington, DC. tions, Sample Terms of Reference, and How Best Miyamoto International. 2019. “Overview of Engi- to Plan a DEM-Mission.” World Bank and Global neering Options for Increasing Infrastructure Facility for Disaster Reduction and Recovery, Resilience.” Background paper for this report, Washington, DC. World Bank, Washington, DC. Daniels, J. 2017. “Southern California Utilities Shut Nicholls, R. J., J. Hinkel, D. Lincke, and T. van der Off Power to Prevent Wildfires.” CNBC, Decem- Pol. 2019. “Global Investment Costs for Coastal ber 13. Defense through the 21st Century.” Policy Davis, C. A., and J. Castruita. 2013. “Los Angeles Research Working Paper 8745, World Bank, Pilot Project Using Japanese Earthquake Resis- Washington, DC. tant Joint Ductile Iron Pipe.” American Water Oguah, S., and S. Khosla. 2017. “Disaster Prepared- Works Association, Denver. ness Offers Big Payoffs for Utilities.” Live Wire European Union. 2015. “EU Reference Document 2017/84, World Bank, Washington, DC. Good Practices on Leakage Management.” WFD Rozenberg, J., and M. Fay. 2019. Beyond the Gap: How CIS WG PoM: Case Study Document, European Countries Can Afford the Infrastructure They Need Union, Brussels. While Protecting the Planet. Washington, DC: World Fraser, S. 2017. “Think Hazard! Methodology Report. Bank. Version 2.0.” World Bank, Washington, DC. Schweikert, A. E., L. Nield, E. Otto, and M. Deinert. Hallegatte, S., J. Rozenberg, C. Fox, C. Nicolas, and 2019. “Resilience and Critical Power System Infra- J. Rentschler. “Strengthening New Infrastructure structure: Lessons Learned from Natural Disasters Assets—A Cost-Benefit Analysis.” Background and Future Research Needs.” Background paper paper for this report, World Bank, Washington, for this report, World Bank, Washington, DC. DC. U.S. Department of Energy. 2018. “Energy Resil- Hutton, G., and M. C. Varughese. 2016. “The Costs ience Solutions for the Puerto Rico Grid.” U.S. of Meeting the 2030 Sustainable Development Department of Energy, Washington, DC. Goal Targets on Drinking Water, Sanitation, and World Bank. 2017. Climate and Disaster Resilient Trans- Hygiene.” Water and Sanitation Program Techni- port in Small Island Developing States: A Call for cal Paper, World Bank, Washington, DC. Action. Washington, DC: World Bank. Kestrel Group. 2011. “Resilience Lessons: Orion’s ———. 2018. Road Costs Knowledge System (ROCKS)— 2010 and 2011 Earthquake Experience: Indepen- Doing Business Update. Washington DC: World dent Report.” Report for Orion New Zealand Ltd., Bank. Kestrel Group, Wellington. World Bank and OECD (Organisation for Economic Kingdom, B., R. Liemberger, and P. Marin. 2006. Co-operation and Development). 2017. A Step “The Challenge of Reducing Non-Revenue Water Ahead: Competition Policy for Shared Prosperity and in Developing Countries: How the Private Sector Inclusive Growth. Washington, DC: World Bank. From Resilient Assets to Resilient Infrastructure Services 7 T he resilience of infrastructure assets is only a small part of the overall picture of resilience described in this report. Because the cost of disruptions exceeds the cost of repairs, infrastructure services offer a better perspective on resilience. For networked infrastructure, a look at services requires a systemic view of both the resilience of the full system, including supporting systems such as ecosystems and wider river basins, and the full cost of failures. This chapter explores how the high costs of abilities, from strengthening critical assets to infrastructure disruptions and damages may be creating redundancy in the system. reduced by focusing on infrastructure services rather than assets and working at the network USING CRITICALITY ANALYSES TO and system levels. It finds that countries can PRIORITIZE INTERVENTIONS increase the resilience of their networks at a A simple approach to giving priority to inter- cost that is even lower than what chapter 6 ventions is to assign a level of criticality to suggests. But they must assign priority to the assets based on the quantity of services they assets that are critical to users or the function- provide—that is, their capacity. For example, ing of their economic system. Identification of construction standards are often higher for pri- critical assets allows utilities and planners to mary roads such as highways and freeways hedge against disruptions by strengthening than for tertiary roads that have a much lower these assets, adding redundant components to volume of traffic. Power generation plants or the networks to reduce their criticality, devel- water reservoirs can also be ranked as a func- oping contingency plans by simulating what tion of their capacity. Although this is a useful happens when they fail, or using network- first assessment for criticality, it is limited in informed solutions to boost resilience. that it does not include information on the Not all assets need to be made more resil- type of service the asset provides (for example, ient. By looking at the system that supports a freeway that provides access to a tourist area infrastructure services, it is possible to identify is less critical than one that leads to the main the most critical parts of a network and assess port or hospital) or the role that the asset plays the performance of options that reduce vulner- in overall network functionality. Sophisticated 109 110 LIFELINES BOX 7.1 Network topology and resilience Infrastructure systems can be represented by an into higher accessibility and lower probability abstract network of nodes and connecting links. of node isolation. Connectivity and accessibility A network establishes and maintains connectivity metrics from graph theory can thus be directly between these nodes to facilitate a flow between employed to gauge the coping capacity of sys- them. A flow is the movement of people, goods, tems. Connectivity metrics describe basic net- material, energy, and services through the sys- work characteristics such as the ratio of links to tem. The vulnerability of the system can therefore nodes or the maximum possible number of links. be linked to the network connectivity that guar- Accessibility metrics describe the best possible antees an available and functional path between flow conditions. Such a measure is, for example, intended origin-destination (O-D) pairs. the network diameter, defined as the maximum The shape of a network contributes to its cop- distance among all shortest distances between ing capacity. Because the shape of an infrastruc- all O-D pairs in the network. Accessibility met- ture system network is static in practice (after all, rics can also be used to identify critical nodes (or a new road cannot be built in an instant), the net- links) in the network. For example, a node that work topological attributes are viable indications is crossed by many of the shortest paths in the of its coping capacity in the face of disruptive network (a node with the largest betweenness events. Overall, networks with a higher number centrality) is likely to have higher importance of interconnection paths between O-D pairs have for maintaining the functioning of the network a greater redundancy, which generally translates (Kwakkel et al. 2019). MAP 7.1 The criticality of a link can be measured by the approaches to prioritizing infrastructure assets additional road user cost resulting from its disruption model infrastructure systems as a network of Example from Zambezia Province, Mozambique nodes and links (box 7.1). Cabo Delgado Namapa Transport systems Me mba Criticality can be assessed by systematically Nassa Lalaua Nacaroa simulating disruptions in a network and esti- Nacala-a-Velha Nacala Me cuburi mating the resulting loss of functionality. Links Ribaue Monapo Ma lema Ra pale Me conta and nodes can be removed one by one—or Nampula Nampula Moss uril several at a time—and the network functional- Gu rue Mu rrupula Alto Mol ocue Liupo ity (such as for transport, travel time, and cost) Nametil Namarroi can be recalculated in the absence of these ele- Milange Ille Gile Angoche ments. Doing so enables identification of the Lugela most critical links as the ones that lead to the Zambezia Mocuba Moma highest loss of functionality when they are Tete removed (map 7.1). Maganj a Pebane Morrumbala Additional RUC Networks should also be stress-tested against Tete Nicoadala due to disruption (US$/vehicle) realistic shocks that include multiple simultane- Quelimane Mopeia Ihas sunge 0–10 ous disruptions (also referred to as n-p), not just 10–20 against shocks to a single component. For 20–30 Sofala Chinde 30–60 example, Kwakkel et al. (2019) test the vulner- > 60 No redundancy ability of the Bangladeshi transport network to Agriculture routes past flood events by examining the spatial cor- Source: Espinet Alegre et al. 2018. relation of disruptions. They find that the best FROM RESILIENT ASSETS TO RESILIENT INFRASTRUCTURE SERVICES 111 solutions for increasing the resilience of the net- FIGURE 7.1 Belgium’s and Morocco’s transport systems can work depend on the set of events that are used absorb much larger road disruptions than Madagascar’s for simulating the disruptions. If data on the full Examples of functionality loss in a transport system as a function of the percentage of links disrupted distribution of possible events are not available, it may be more robust to invest in improve- 100 Loss of functionality of the network (%) ments that will increase the resilience of the network to a wide range of random events. 75 These approaches make it possible to mea- sure the resilience of a network, defined as the 50 ratio of the loss of functionality to the loss of assets (Rozenberg et al. 2019). A highly resilient transport network can lose many assets (such 25 as road segments) without losing much func- tionality. Figure 7.1 represents the functionality 0 0 10 20 30 40 50 60 loss (expressed as isolated trips—that is, when Level of disruption (% links disrupted) travelers can no longer reach their destination) from the disruption of random transport links Belgium Madagascar Morocco as a function of the percentage of links dis- Source: Rozenberg et al. 2019. rupted. It shows that, thanks to their redun- dancy, the transport networks in Belgium and MAP 7.2 Belgium’s transport network is much denser and Morocco exhibit much more resilience than the offers greater redundancies than Madagascar’s network in Madagascar. For low levels of dis- ruption (below 20 percent of links), functional- a. Transport network, Belgium b. Transport network, Madagascar ity losses are mostly negligible in Belgium, whereas in Madagascar they quickly rise to 80 percent. The key here is that, because Madagas- car’s network has much less redundancy than Belgium’s network (map 7.2), a disruption of critical roads can paralyze the whole network. This type of analysis provides more valuable information than static network metrics because it allows identification of the extreme cases in which even a small number of dis- rupted links can lead to high functionality loss. Such criticality analyses can help to identify investments that increase the redundancy of a network and have positive economic returns. In Peru, Rozenberg et al. (2017) show that tar- geted investments to increase the redundancy Source: Based on OpenStreetMap data. of the road network around Carretera Central, a strategic export route for agricultural prod- and duration of climate-related events; the ucts, could be justified on the sole basis of the structural impact of water levels on the road; annual user losses from floods and landslides the amount of traffic to be rerouted when a avoided. This measure yields a positive return flood or landslide hits; and the time and total in almost all possible scenarios, combining cost of reconstructing a road after a disaster (fig- uncertainty regarding the intensity, frequency, ure 7.2). In transport, redundancy can also be 112 LIFELINES FIGURE 7.2 Increased redundancy can have Power systems high net benefits, if well targeted Network resilience can also play a role in Net benefits of four interventions across power transmission and distribution system hundreds of scenarios, Carretera Central, Peru planning. Vulnerable or critical parts of the network can be identified using a network Flood-proof analysis with a single-element contingency first-best road (n – 1), a double-element contingency (n – 2), or even a p-element contingency (n – p). It is Targeted increase in redundancy important to understand how the network will behave should 1 to p elements fail, as the criti- Large-scale increase cality of the remaining nodes changes when in redundancy the most critical node is removed (Carlotto and Grzybowski 2014). Unfortunately, because of More frequent the complexity of power flow analysis models, maintenance even the n – 2 contingency analysis is often –800 –600 –400 –200 0 200 400 600 very difficult to perform. Thus, an n – p contin- Performance of interventions gency analysis at the system level is often not (US$, millions) possible, and studying how the system would Source: Rozenberg et al. 2017. behave if p elements were to fail is only possi- Note: The net benefits focus on avoided losses and do not ble in a designated area or for a selection of include benefits from interventions regarding reduced road user costs in the absence of disasters. Each cross in this those p elements. graph is a different scenario, with various assumptions about Veeramany et al. (2018) illustrate how these the intensity, frequency, and duration of climate-related events; the structural impact of water levels on the road; the approaches can identify opportunities for inter- amount of traffic to be rerouted when a flood or landslide hits; ventions with very high returns. They perform and the time and total cost of reconstructing a road after a disaster. a network criticality analysis for seismic risks in the state of Washington in the U.S. Northwest. Working on a subset of the transmission net- built through multimodal systems so that users work assumed to be vulnerable to seismic haz- can switch between modes after a disruption. ards, they consider 40 potential seismic events In Mozambique, Espinet Alegre et al. (2018) and run 200,000 scenarios to assess the behav- prioritize interventions in the rural road net- ior of the system during an earthquake. They work based on a combination of criticality and are able to identify the most critical combina- risk to infrastructure—that is, the expected tion of assets, finding that hardening one asset annual damage based on hazards and vulnera- or adding redundancy to “double” this asset bility (map 7.3). They define criticality using would reduce risk by 88 percent. not only the loss of functionality if a road is Power transmission and distribution net- damaged but also information on users, includ- works are built with some level of redundancy ing the poverty and agricultural potential in to allow them to cope with the disruption of the province served by the road. They find that one network element by rerouting power, in provinces with a high risk of floods and low thereby reducing the curtailment of plants and redundancy, the direct benefits of investments limiting disruptions to consumers. Such levels in new culverts and stronger bridges are rela- of redundancy are included in the planning tively small. However, the indirect benefits, and construction standards of most utilities. expressed in lower expected annual costs for These standards are often more stringent at the road users due to flood disruptions, are four transmission level than at the distribution times larger and justify the investments under level, because the risk of widespread outages is most of the scenarios considered. higher at the transmission level. FROM RESILIENT ASSETS TO RESILIENT INFRASTRUCTURE SERVICES 113 MAP 7.3 The strengthening of infrastructure assets in Mozambique is prioritized based on risk levels and criticality Priority Nassa Very low Nampula Low Gurue Medium Alto M olocue High Very high Namarroi Gi le Mi lange Ile Lugela Za mbezia Pebane Mo rrumbala Mocuba Maganja da Costa Tete Namacurra Nicoadala Mo peia Risk to infrastructure Inhassunge 1 2 3 4 5 1 1 1 1 1 2 Chinde Sofala Criticality 2 1 2 2 2 3 3 2 2 3 3 4 4 3 3 4 4 5 5 3 4 4 5 5 Source: Espinet Alegre et al. 2018. Note: Criticality is defined as a combination of the poverty in the province served by the road, the agricultural potential of the prov- ince served by the road, and the loss of functionality if the road is removed. Identifying critical assets via an n – 1 or n – 2 has multiple links, so that if one fails, an alter- contingency analysis does not necessarily mean native power supply route is available. This spi- doubling or tripling key components of the der’s web approach greatly increased Orion’s network (for n – 1 and n – 2, respectively) or ability to restore power promptly after the placing lines underground. A more effective 2010 and 2011 earthquakes. It meant that approach is usually to create “ringed” or power stayed on unless all the multiple links meshed networks that provide multiple supply into an area failed. If all of the links were dam- points to various nodes in the grid (figure 7.3). aged, Orion could fix the link that was the eas- A meshed network reduces the exposure to iest and quickest to repair. outages along corridors. It also enables net- works to switch loads quickly between feeders Water systems or supply points. This approach is used more In water systems, the typical methodology for and more for distribution networks that used assessing criticality in a network calls for carry- to be star-shaped (the traditional radial distri- ing out a failure mode, effects, and criticality bution) but are now becoming increasingly analysis (Stip et al. 2019). This analysis consists meshed, like most transmission networks. of mapping out all of the components of the Orion is one of the largest electricity distri- network and assessing under which conditions bution companies in New Zealand, providing they would fail, what the effects of that failure power to remote rural areas, regional towns, would be, and how they would affect service and the city of Christchurch. Rather than oper- delivery. Based on the latter, the “criticality” of ating a single line or cable into an area, Orion that component can be ranked and a rating 114 LIFELINES FIGURE 7.3 Network topology can improve grid resilience a. Tree-like distribution network b. Distributed generators c. Meshed network Generator Load center Source: Stöcker 2018. Note: The meshed distribution network contains distribution feeders that are linked by open switches during normal operations to maintain the radial characteristic of the distribution network. These switches are closed to provide alternative paths for electricity when a distribution feeder is disconnected. recorded accordingly. In the Netherlands, maintenance plan for a given budget constraint, breakdowns are ranked by level: a level 1 prioritizing the most critical elements. breakdown should never occur because it Japan and the Netherlands have built would significantly disrupt service, a level 2 redundancy into their water distribution sys- breakdown is allowed to occur every three tems through “loops,” so that if one element of years, and a level 3 breakdown can happen the network breaks down, other elements of every year because it is not vital to operations the system can always be reached via an alter- (Wright-Contreras 2018). Based on this cate- nate route (Wright-Contreras 2018). In the gorization, a regular maintenance regime can Netherlands, this redundancy is also found in be implemented to check the elements linked storage systems and water treatment plants. to level 1 breakdowns, while spare parts can be For example, river intakes can be shut down if stored for those elements in a level 3 break- the water quality of the river worsens, and so down, which is expected more often. These the reservoir, because it has a storage supply of levels are also determined by whether the asset water for five to six months, will usually be is essential to providing service to more than able to “flush” the river water from that pollu- 1,000 households or to a hospital or providing tion event. Water treatment plants themselves other services such as firefighting. are built with storage, which not only improves In Cutzamala, Mexico, a sensitivity assess- water quality at the inlet through sunlight and ment that examined the city’s water system for retention but also provides a water source for a lack of maintenance of major system compo- given amount of time if the intake has to be nents identified elements that would have the closed or the plant malfunctions. severest negative impacts on maintaining accept- able performance of the system in different sce- Telecommunications narios (Ray and Brown 2015). This knowledge Telecommunication networks can be designed could then be used to develop an optimized to have high redundancy—physical and logi- FROM RESILIENT ASSETS TO RESILIENT INFRASTRUCTURE SERVICES 115 cal—that protects users against extreme works, diversifying, decentralizing, and work- events. The Great East Japan Earthquake in ing across systems. March 2011, measuring 9.0 on the Richter In addition to redundancy and strengthen- scale, and the resulting tsunami, damaged ing of critical assets, other system-level inter- submarine cable systems along the Japanese ventions can be envisaged to increase the resil- coast. The effects of this disaster on Internet ience of infrastructure services, ranging from connectivity was, however, limited because diversification to decentralization and cross- the level of redundancy in Japan’s interna- system analyses. tional connectivity was adequate. Japan’s diversity of submarine cable system routes DIVERSIFYING ASSETS TO ensured that the overall capacity landing in INCREASE NETWORK RESILIENCE the country was not significantly curtailed. As The benefits of diversifying generation sources a result, although there was some disruption in the power sector were particularly evident from the cable breaks, international connec- in Texas following Hurricane Harvey. Nuclear tivity was remarkably robust considering the power was able to operate at full capacity scale of the disaster. However, the right level throughout the event. Wind farms were cur- of redundancy is hard to determine. In 2013 a tailed during the event, but most immediately diver intentionally cut the South East Asia– came back online, compensating for much of Middle East–Western Europe 4 (SE-ME-WE-4) the production deficit from reductions in the cable system. The presence of eight submarine generation facilities located on the coast, which cables between the Arab Republic of Egypt were affected by storm surges and flooding for and Europe meant that Egypt’s Internet longer time periods (Conca 2017). Schweikert should not have been affected significantly. et al. (2019) recommend a power mix that However, four cable systems reported faults or does not depend fully on water to reduce the breaks during the same week, resulting in risk of power shortages during droughts or overloads and congestion on the active cable extreme heat events (Alvaro 2018). Indeed, systems.1 Internet speeds crashed by 60 per- thermal generation facilities and nuclear power cent, with impacts felt by all telecom opera- plants, which rely on water for cooling, often tors in the country. need to be curtailed or closed when intake Commercial agreements between owners water exceeds the permitted temperatures and users of telecommunications infrastructure (approximately 24°C in most cases). allow for “logical redundancy” in networks, In transport, diversification is done through significantly reducing the risk and impact of multimodal transport planning. Urban plan- damage to physical infrastructure. The industry ning, for example, can include nonmotorized has also adopted cost- and risk-sharing busi- modes, such as walking and cycling, and mass ness models, where telecom-ready infrastruc- transit. If transport planning is accompanied ture—such as poles, underground ducts and by policies that incentivize a higher urban channels, fiber-optic cables, and pylons—is density, this mode diversification can reduce shared between telecom operators and other traffic density and the need to build an sectors such as energy and transport. While increasingly large number of roads, thereby these infrastructure-sharing models have reducing obstacles to water flow and mitigat- enabled rapid and cost-efficient network roll- ing floods. In addition, by reducing the need outs, the risk associated with aggregated infra- to build more roads, urban planning can structure also increases. Therefore, there are reduce the scale of the exposure and vulnera- trade-offs between cost sharing and adequate bility of the transport sector to disasters. These investment in the physical redundancy of net- alternative modes can also provide resilient 116 LIFELINES forms of transport during an emergency forced outage and blackout or whether the (World Bank 2015). minigrid should remain “islanded.” Rainwater harvesting and decentralized DECENTRALIZING AND USING NEW water treatment can contribute to more flexible TECHNOLOGIES hybrid water systems (Stip et al. 2019). New Distributed power systems that rely on solar containerized treatment systems for waste­ energy and batteries can harden a grid and water treatment and for drinking water pro- make it more resilient. Minigrids and micro­ duction, using ultrafiltration technology with grids, because they do not rely on long-dis- low-fouling hollow-fiber membranes, can pro- tance transmission wires, can provide useful vide high-quality water effluent (Georges et al. backup generation in case of grid failure. 2018). The units, which are modular and plug Indeed, most electricity outages result from and play, are easy to transport because they are damage to transmission lines and transformers installed in shipping containers. A basic decen- rather than generation facilities. During Hurri- tralized rainwater harvesting system can pro- cane Sandy, the Co-Op City microgrid in New vide nonpotable water for toilet flushing and York successfully decoupled from the central other nonpotable requirements to reduce the grid and supported consumers during outages demand for potable water. A more advanced on the wider network (Strahl et al. 2016). system could collect harvested rainwater sup- In the future, sensors will allow power dis- plemented by rainwater harvesting and storage tribution management systems to be pro- at the customer’s property as well as use storm- grammed to reconfigure networks to distribute water retention and treatment systems to sup- loads after isolating faulty segments of the net- plement raw water resources. Such systems work. Sensors within components of power would increase resilience to droughts, bursts, systems will allow power plants and substa- and the pollution of raw water sources, and tions to communicate with one another as well they would reduce the risk of urban flooding. as with the grid operator. The grid control sys- Container-based sanitation, in addition to tem will take into account real-time conditions providing low-cost sanitation services, is more that affect the locational marginal price (chang- resilient to floods and droughts than other ing the costs to operate and the congestion solutions (Georges et al. 2018). In Haiti, users costs). Substations may communicate with of container-based sanitation services reported control systems on the distribution end, send- that they were able to use their toilets during ing signals about pricing to end-use customers’ floods, whereas traditional latrines were unus- devices. Those signals may modulate and mod- able. In Nairobi, some service users found the erate customers’ power demands accordingly— waterless nature of Fresh Life toilets to be a dis- for example, by sending price signals to a tinct advantage. In that water-scarce environ- building’s air-conditioning system. The same ment, there is no piped water, and conse- types of sensors and optimizations could also quently water for household use is costly and function on a minigrid, perhaps even with has to be hauled over considerable distances— more value because a minigrid’s operation typ- typically by women. ically has fewer degrees of freedom than a cen- tral grid. The algorithms may help the minigrid WORKING ACROSS SYSTEMS TO operators to coordinate with the central grid CAPTURE SYNERGIES (in cases where a connection is possible) and to The criticality of an infrastructure asset also decide whether the minigrid should supply depends on complex interdependencies and power to help a central grid return from a possible cascading failures, including trans- FROM RESILIENT ASSETS TO RESILIENT INFRASTRUCTURE SERVICES 117 boundary ones. Mapping interdependencies during Hurricane Sandy (U.S. Department of between critical infrastructure assets and sec- Energy 2018). tors is increasingly important to understanding Managing services together or through potential cascading consequences. Interdepen- effective collaboration platforms enhances dencies between infrastructure sectors can be efficiency and can generate cost savings, physical, cyber, geographical, or logical, and which can later be reinvested in the system. In they can be between sectors or between assets. Orange County, California, joint planning All infrastructure sectors tend to be highly between the Orange County Water Depart- dependent on electricity. Contingency plans for ment (in charge of the bulk water supply) and water utilities should include ways to prevent the Orange County Sanitation Department (in or recover quickly from power outages at charge of sanitation) helped to identify waste- pumping stations, reservoirs, and storage tanks. water reuse as a key cost saver for both the Similarly, telephone, cellular, e-mail, or dedi- sanitation district (due to the avoided costs of cated broadband networks cannot function seawater outfall) and the water district (by without electricity, and so telecommunications securing a new drought-proof source of water) facilities usually have reserve power—battery (World Bank 2018). In general, a utility that banks—for short-duration outages. In North manages both water supply and sanitation America, these battery banks have power for together may reduce the transaction costs three to eight hours, which is appropriate for associated with coordination, while being bet- frequent disruptions but perhaps insufficient ter placed to identify opportunities to close the for large-scale disasters. It is essential that key water cycle. telecommunications facilities have a backup Roads make a major imprint on hydrology power generator and secure fuel storage by blocking and guiding water, concentrating arrangements for a prolonged power outage. runoff, interfering with subsurface flows, and The power sector itself depends on other changing flooding patterns. However, there is a infrastructure sectors. Unpassable roads are beneficial connection between road planning one of the main obstacles that electric utilities and building and water management. Water is face in repairing transmission lines. By work- considered the prime enemy of road infrastruc- ing with road agencies, utilities can ensure that ture and the single greatest factor in road dam- they have the right information on accessible age. Therefore, a strong case can be made for routes. Such an arrangement would also managing water around roads better and for ensure that the roads needed to repair the considering roads as an integral part of the power system would receive priority for watershed and landscape in which they are sit- reopening. Generation technologies that uated. Such an integrated approach will pre- require on-demand fuel delivery, such as natu- serve road infrastructure and reduce the bur- ral gas, oil, and coal-fired systems, also rely den of maintenance, contributing to greater heavily on the transport network. In Puerto infrastructure productivity, while providing Rico following Hurricanes Irma and Maria, water supply and flood protection. Van Steen- port closures resulted in the loss of an esti- bergen et al. (2019) describe how the negative mated 1.2 million barrels of petroleum a day impact of roads on the surrounding landscape for 11 days, which directly affected the major can be turned around and how roads can generation stations that relied exclusively on become instruments of beneficial water man- imported fuel (U.S. Department of Energy agement. For example, in arid areas the water 2018). Similar closures occurred in Texas in intercepted by road bodies can be guided to 2017, as well as in New Jersey and New York recharge areas or surface storage or applied 118 LIFELINES directly on the land. On floodplains and in high benefits, the development of appropriate coastal areas, roads also play a role in flood pro- institutions and governance mechanisms to tection. Roads can double as embankments and deliver maintenance as well as the necessary provide evacuation routes and flood shelters. In funding streams is essential. Failure to do so low-lying wetland areas and on floodplains, would increase risk and could result in cata- roads and bridges affect the shallow groundwa- strophic failures, putting lives, not just assets, ter tables and have enormous consequences for at risk. Absent a credible commitment to reli- land productivity. The way in which a road is able maintenance, a combination of nature- built and, for example, the height of bridge sills based protection, land use planning, and and culverts will have considerable influence retreat should be favored. on the quality of the wetland on either side of the road (Van Steenbergen et al. 2019). COMBINING INFRASTRUCTURE WITH NATURE-BASED SOLUTIONS PROTECTING INFRASTRUCTURE TO REDUCE INVESTMENT NEEDS SYSTEMS WITH DIKES IN DENSE Combining green and gray infrastructure can AREAS provide lower-cost, more resilient, and more One option to reduce coastal and river flood sustainable infrastructure solutions (Browder risk is to protect infrastructure systems with et al. 2019). The filtration services provided by dikes, which are part of water systems. Water healthy forests saved Portland, Maine, between systems act both as water service providers and $97 million and $155 mil­ lion over 20 years by as protection against water-related hazards. canceling out the need for a water filtration Dikes can be a cost-efficient strategy in high- plant (Gartner et al. 2013). In the Philippines, density areas and would reduce the exposure mangroves, reefs, and other natural systems of other infrastructure systems (Rozenberg and prevent more than $1 billion in annual disaster Fay 2019). However, dikes cannot protect losses (Tercek 2017). Meanwhile, 90 percent of against all possible events, and they need to be New York City’s water is provided by well- accompanied by clear communication cam- protected wilderness watersheds, so that New paigns on residual risk, as well as contingency York’s water treatment process is simpler than plans in case of failure. While dikes can protect that of other U.S. cities (NRC 2000). assets, appropriate early warning systems and Good catchment management can increase evacuation plans remain important for manag- the availability of freshwater and reduce the ing the risk of large human losses in case of cost of treatment. Floating wetlands can be dike failure or overtopping. used for in situ treatment of elevated nutrient Rozenberg and Fay (2019) assess the invest- concentrations. And riparian planting can be ment in coastal and river flood protection used to lessen the rate of runoff, erosion, and infrastructure (using dikes and storm surge nutrient reduction and to increase the quantity barriers) needed to protect cities in low- and of water captured for use. middle-income countries by 2030, under a In Suva, Fiji, the RISE Program is working range of socioeconomic and climate change in communities exposed to tidal flooding and scenarios. They find that, depending on accept- forced to rely on poor sanitation solutions that able risk levels and construction unit costs, allow the spread of fecal contamination from total costs could go from $23 billion to $335 latrines in each flooding event.2 The proposed billion per year. Although these costs are again interventions would mix simplified sewerage low compared with total infrastructure invest- to contain the waste, with wetlands and walk- ment needs, and although dikes can generate ways that separate the community from flood- FROM RESILIENT ASSETS TO RESILIENT INFRASTRUCTURE SERVICES 119 ing and filter the water as it flows in and out of more than $400 million for each country. In the area. Colombo, preserving the wetlands system Working with nature also means closing the proved to be a cost-effective solution to water cycle. In 1968, faced with a severe reduce flooding in the city, even when taking drought, Windhoek, Namibia, became one of into account land development constraints the first cities in the world to introduce full- (Browder et al. 2019)—see photo 7.1. Roads scale waste­ water reclamation for use as drink- are especially vulnerable to landslides, and ing water (World Bank 2018). The wastewater different forest management practices can is treated to potable level and injected directly have large implications for landslide suscepti- into the water supply, and it now provides 25 bility. According to Dhakal and Sidle (2003), percent of Windhoek’s water. The aquifer in partial cutting produces fewer landslides and Orange County, California, is also used as a lowers the volume of landslides by a factor of buffer during dry conditions. Stormwater infil- 1.5 compared with clear-cutting. tration is promoted through canals and inflat- Power transmission lines are very vulnera- able dams, while highly treated wastewater is ble to falling trees during high wind events (see injected to recharge the aquifer. This managed chapter 4). But by preventing certain high veg- aquifer recharge increases the drinking water etation from encroaching on the rights-of-way available to Orange County service providers, alongside these lines, some utilities in the while also serving as a barrier to seawater United States are encouraging native low- intrusion. growth vegetation. The result is that, in some A noteworthy example of such integration areas, the scrubby habitat under some trans- of the water cycle with city infrastructure is mission lines becomes the best place to find China’s sponge cities (State Council of China wild bees. As the scrub vegetation grows in, it 2015). Under this ambitious program, the excludes many taller trees, and over a few country seeks to reduce the effects of flooding years, mowing costs drop dramatically. Such through a mix of low-impact development vegetation management thus comes at a lower measures and urban greenery and drainage infrastructure, and to have 80 percent of urban PHOTO 7.1 A wetland park in Colombo helps to mitigate areas reuse 70 percent of rainwater by 2020. flood risk and offers recreational opportunities, such as This approach is similar to what Australia’s bird-watching towers Cooperative Research Centre for Water Sensi- tive Cities calls its vision of the “city as a water catchment.” Nature-based solutions are also used for flood protection, reducing the need for hard infrastructure like dikes. The fact that man- groves and coral reefs protect coastlines against floods and storm surges is well known. According to Beck et al. (2018), coral reefs halve the annual global damages from flood- ing and divide by three the costs from fre- quent storms. They estimate that the coun- tries benefiting the most from reefs are Cuba, Indonesia, Malaysia, Mexico, and the Philip- pines, with annual expected flood savings of Photo credit: Matthew Simpson. 120 LIFELINES cost to the utilities and can create a network of Develop and update contingency plans wildlife corridors under transmission lines Contingency plans set out the measures to be (Conniff 2014). taken by a service provider in the event of an emergency or unforeseen incident. Transport FAILING GRACEFULLY AND operators could, as part of their contingency RECOVERING QUICKLY plan, focus on restoring connections to critical It is sometimes more cost-effective to replace nodes such as hospitals and ports (Benavidez infrastructure after an event than to make it and Mortlock 2018). Water utilities could have strong enough to resist everything, such as a standing contract with water tankers to pro- antennas in the telecommunications sector. vide water if the water system fails in an emer- Investing in the protection of these assets gency situation. In the Philippines, after a would not yield proportional returns, as typhoon, water tankers were contracted to opposed to investing in backups and resto- ensure service continuity despite infrastruc- ration preparedness. In addition, no infrastruc- ture damage. ture asset or system can be designed to cope In the power sector, contingency planning with all possible hazards. Because there is great needs to be carried out more frequently to uncertainty about the probability and intensity understand the extent of widespread black- of the most extreme events, infrastructure sys- outs, simulate various restoration procedures, tems should be stress-tested against events that and incorporate outputs into operational and go beyond the likely ones. Such a stress test training manuals for system operators. Contin- would have two goals: gency analysis could also be extended to include demand-side management. For exam- • Identify low-cost options that can reduce ple, predetermined loads could be disconnected the vulnerability of infrastructure systems to to avoid a loss of grid stability and avert possi- extreme events, even if those events are ble widespread outages (box 7.2). considered extremely unlikely. For exam- New technologies can help to achieve ple, the Fukushima nuclear incident quicker recovery in the power sector. Smart demonstrated that, even if large dikes are grids and advanced metering infrastructure supposed to protect a nuclear power plant (AMI) improve situational awareness and sup- against all possible tsunamis, a “what-if” port rapid restoration after disasters. AMI is an scenario exercise would be useful, consider- integrated system of smart meters, communi- ing the possibility that some unexpected cations networks, and data management sys- event exceeds the level of protection. Such tems that enables two-way communication an exercise could produce additional vul- between utilities and customers. This informa- nerability-reducing options, such as elevat- tion is vital to system operators, who otherwise ing a plant’s backup generators in case are blind to rapid changes in the energy sys- flooding occurs despite the dikes. tem, and thus help to improve resilience in the • Understand the consequences of an unex- grid (GridWise Alliance 2013; White House pected failure to prepare for the required 2013). AMI was used after Hurricane Sandy by response—both in terms of management of the Potomac Electric Power Company, which the infrastructure system (such as how to serves the Washington, DC, metropolitan area. recover from a major failure) and support The utility received “no power” signals from for users (such as how to minimize impacts meters that enabled it to pinpoint outages and on hospitals). Running scenarios of failures dispatch teams to specific areas instead of is the first and most critical step in defining scouting wider areas to locate problems (Oguah contingency plans. and Khosla 2017). FROM RESILIENT ASSETS TO RESILIENT INFRASTRUCTURE SERVICES 121 BOX 7.2 Contingency planning for power utilities Power utilities could recover quickly from disas- from other utilities. Preselected vendors for ters by taking the following specific steps: cars, ships, and helicopters could be utilized to deliver materials and spare parts in a timely • Information gathering. Once a disaster occurs, manner. Furthermore, staff should secure a it is critical that utilities gather and share infor- place to store these materials, cooperating with mation in a timely manner. Crucial informa- municipalities if needed. tion includes (1) meteorological and terrestrial Cooperation with external institutions. Cooper- •  phenomena, (2) damage to power facilities, ation with the central government and munici- (3) blackouts, (4) affected staff, and (5) the palities should include information sharing and traffic situation. staff deployment. The military may offer staff • I nformation distribution. To help users man- as well as the tools needed to restore affected age disruptions, it is important to publicize the facilities in the affected areas. Although utilities information via television, radio, newspapers, are often competitors, they frequently coop- and the Internet, especially information on erate in disaster recovery periods by sharing blackouts and the restoration of power and fur- staff, equipment, or spare parts. To ensure ther expected hazards. cooperation, utilities are increasingly entering • S ecuring of staff. Staff who are assigned to into mutual aid agreements that describe pos- deal with disaster recovery should be present sible ways of cooperation (Lindsey 2008). even during holidays and at nighttime. These Q uick recovery tools for power facilities. The •  staff oversee the disaster recovery operations resources required for the recovery period and are responsible for deploying staff to the may include (1) alternate offices with appropri- affected sites as well as for cooperating with ate access to information and communication; external organizations until normal operations (2) special vehicles such as mobile substations are restored. and a generator vehicle; (3) alternative gener- • Securing of materials and spare parts. Utilities ation options (such as hydrogen, storage bat- should confirm whether materials and spare tery, co-generation, microgrids, or diesel emer- parts are sufficient for recovery and, if insuffi- gency stations); and (4) helicopters for access cient, seek means to procure them, including to damaged assets if roads are closed. In the telecommunications sector, as high- unavailable in the immediate aftermath of the lighted in chapter 4, natural shocks are likely event and making recovery challenging. to cause damage to exposed assets (such as In the water sector, so-called slow-onset towers and antennas) and underground assets events—hazards that happen over a long (such as ducts and cables). Utility operators period—offer opportunities to react as the should be prepared and have the assets to event unfolds. For example, in Spain the restore services as soon as possible. As part of Aigües de Barcelona’s Drought Management their contingency plans, they also need to Plan tracks key indicators of water system per- ensure that the assets required to restore ser- formance and helps the service provider to vices are protected from hazards. For example, respond through measures taken to guarantee the 2011 earthquake and resulting tsunami in drinking water supply and mitigate economic Japan that damaged submarine cable systems impacts (World Bank 2018). Based on surface along Japan’s coast also damaged most of the storage levels, drought thresholds are estab- submarine cable repair vessels, rendering them lished for the sources from which the utility 122 LIFELINES will draw (figure 7.4). In a dry event, the more ural hazards than before the disaster, but it also expensive sources (reuse and desalination) are greatly enhanced the quality of services and used first, followed by strategic buffer sources access to essential public services (including (the aquifer). As a last resort, the city taps into water, sanitation, roads, health, and educa- water normally reserved for environmental tion). For example, 300 roads were rebuilt flows to the water supply. or renovated to new seismic standards and upgraded through the addition of modern traf- Build back better fic management and drainage systems. Building back better is a central part of disaster recovery. Hurricane Sandy caused catastrophic Sometimes, the best approach is not damage in New York City, with kilometers of to build copper cables rendered useless. Verizon lost not One way to reduce risk—or at least to mini- only its carrier vaults in Manhattan (two vaults, mize increases in risk—is to ensure that no each with a volume of more than 90,000 cubic new assets are located in at-risk areas. For feet) but also multiple manholes in the city. example, to avoid the impact of heat waves on Estimating the loss at approximately $1 billion, data center cooling, new large data centers are Verizon did not see the value in repairing the being built near the Arctic Circle to keep the existing network. Instead, it replaced the cop- servers as cool as possible, which in turn is per networks with fiber-optic cables, which are reducing significantly the energy consumption more resilient to water damage (Adams et al. for cooling and avoiding disruptions. Infra- 2014). Verizon also undertook other resil- structure can also guide households and firms ience-enhancing measures to protect its critical toward low-risk areas if it is properly planned infrastructure. The carrier vaults, as well as fuel and future construction plans are communi- storage and pump rooms, were made water- cated to the public (see chapter 8). tight, with submarine doors to ensure continu- Sometimes, retreat is a better option than ity of operations. protection, especially considering long-term In 2008 a major earthquake struck south- climate change trends and impacts on sea level western China. With more than 69,000 fatali- or water scarcity. For instance, Nicholls et al. ties, 374,000 people injured, and about 18,000 (2019) find that coastal protection against missing, it was one of the deadliest earthquakes storm surges and sea-level rise would only in recent history (Hallegatte, Rentschler, and make sense for about 22–32 percent of the Walsh 2018). In addition to the human toll, the world’s coastlines throughout the 21st century, disaster destroyed or severely damaged 34,000 depending on assumptions about economic kilometers of highways; thousands of schools, growth and sea-level rise. Thus, communities hospitals, and wastewater systems; and more located adjacent to at least 68 percent of coast- than 4 million homes. In response to the disas- lines may have to retreat gradually or use ter, the government of China adopted a build low-cost ecosystem-based or nature-based back stronger approach. It ensured that the approaches to coastal defense. These areas are reconstruction of affected infrastructure mostly low-density areas with a small stock of adhered to higher seismic standards and flood assets, and the costs of protection are too high risk management codes, while ensuring a bal- to be affordable. In those areas that cannot be ance between reconstruction activities and lay- realistically protected against long-term sea- ing a foundation for the longer-term sustain- level rise and coastal floods, not building new able economic recovery and development of infrastructure may be the best approach to the affected areas. Not only was the restored resilience. This approach should, however, be infrastructure built to be more resilient to nat- complemented by a consistent strategy to man- FROM RESILIENT ASSETS TO RESILIENT INFRASTRUCTURE SERVICES 123 FIGURE 7.4 Drought contingency plans in Spain use diverse water sources and are informed by historical drought threshold values a. Drought threshold values 700 600 500 Water storage level (hm3) 400 300 200 100 0 00 02 04 06 08 10 12 14 16 82 84 86 88 90 92 94 96 98 80 20 20 20 20 20 20 20 20 20 19 19 19 19 19 19 19 19 19 19 January of year shown Drought level Alert Exceptionality Emergency I Emergency II Emergency III b. Water source mix Normal Alert Exceptional Emergency Surface water Reused water Desalinated water Ground water Source: World Bank 2018. age retreat while maintaining livelihoods and resilience at a lower cost than strengthening community ties. assets. The next chapter brings users into the This chapter has highlighted how consider- equation, because it is sometimes easier and ing infrastructure services—instead of infra- cheaper to enable users to cope with infra- structure assets—and looking at the system and structure disruptions than it is to prevent all network levels can offer opportunities to build possible disruptions. 124 LIFELINES NOTES Uncertainty in Transport Operations: Prepara- 1. “Undersea Cables Off Egypt Disrupted as Navy tion and Appraisal of a Rural Roads Project in Arrests Three,” Guardian, March 23, 2013. Mozambique under Changing Flood Risk and https://www.theguardian.com/technology Other Deep Uncertainties.” Policy Research /2013/mar/28/egypt-undersea-cable-arrests. Working Paper 8490, World Bank, Washing- 2. See https://www.rise-program.org/about. ton, DC. Gartner, E. T., J. Mulligan, R. Schmidt, J. Gunn, A. Reference, and N. Price. 2013. “Natural Infra- REFERENCES structure Investing in Forested Landscapes for Adams, P., J. Steeves, B. Ashe, J. Firth, and Source Water Protection in the United States.” B. Rabb. 2014. “Climate Risks Study for Tele- World Resources Institute, Washington, DC. communications and Data Center Services.” Georges, K., B. Hall, T. McNally, and B. Tam. Report prepared for the General Services 2018. “The Smart Water Utility: 21st Century Administration, Riverside Technology, Fort Water Utility (Isles Utilities).” ICMA (Interna- Collins, CO. tional City/County Management Association) Alvaro, L. 2018. “The Pitfalls of Hydroelectric (blog), July 3. Power in Drought-Prone Africa.” Grantham GridWise Alliance. 2013. “Improving Electric Grid Institute Climate and Environment at Peril Reliability and Resilience: Lessons from Super- (blog), August 21. storm Sandy and Other Extreme Events.” Beck, M. W., I. J. Losada, P. Menéndez, B. G. GridWise Alliance, Washington, DC, June 12. Reguero, P. Díaz-Simal, and F. Fernández. Hallegatte, S., J. Rentschler, and B. Walsh. 2018. 2018. “The Global Flood Protection Savings Building Back Better: Achieving Resilience through Provided by Coral Reefs.” Nature Communica- Stronger, Faster, and More Inclusive Post-Disa- tions 9 (2186): n.p. ster Reconstruction. Washington, DC: World Benavidez, M. J., and A. M. Mortlock. 2018. Bank. https://openknowledge.worldbank.org “Transport Sector Recovery: Opportunities /handle/10986/29867 to Build Resilience.” GFDRR Guidance Note, Kwakkel, J., B. A. Jafino, A. Verbraeck, Global Facility for Disaster Reduction and S. Eker, and J. Rozenberg. 2019. “Resilience Recovery, Washington, DC, October. of the Transport Network in Bangladesh: A Browder, G., S. Ozment, I. Rehberger Bescos, Graph-Theoretical and Simulation-Based T. Gartner, and G.-M. Lange. 2019. Integrat- Approach.” World Bank, Washington DC ing Green and Gray: Creating the Next Generation (forthcoming). Infrastructure. Washington, DC: World Bank Lindsey, K. 2008. “Guidelines for Emergency and World Resources Institute. Resource Planning for Overhead Transmission Carlotto, T., and J. M. V. Grzybowski. 2014. Line Asset Owners.” Institute of Electrical and “Self-Organized Criticality and Cascading Fail- Electronics Engineers (IEEE), New York. ures in the Topological Model of Power Grids: Nicholls, R. J., J. Hinkel, D. Lincke, and T. van A Case Study.” Proceedings of the Brazilian Soci- der Pol. 2019. “Global Investment Costs for ety of Applied and Computational Mathematics Coastal Defense through the 21st Century.” (SBMAC) 2 (1): n.p. Policy Research Working Paper 8745, World Conca, J. 2017. “Hurricane Harvey Makes the Bank, Washington, DC. Case for Nuclear Power.” Forbes, September 1. NRC (National Research Council). 2000. Water- Conniff, R. 2014. “Electric Power Rights of Way: shed Management for Potable Water Supply: Assess- A New Frontier for Conservation.” Yale Envi- ing the New York City Strategy. Washington, DC: ronment 360 (blog), October 16. https://e360 National Academies Press. .yale.edu/features/electric_power_rights_of Oguah, S., and S. Khosla. 2017. “Disaster Pre- _way_a_new_frontier_for_conservation. paredness Offers Big Payoffs for Utilities.” Live Dhakal, A. S., and R. C. Sidle. 2003. “Long-Term Wire 2017/84, World Bank, Washington, DC. Modelling of Landslides for Different Forest Ray, P. A., and C. M. Brown. 2015. Including Management Practices.” Earth Surface Processes Climate Uncertainty in Water Resources Planning and Landforms 28 (8): 853–68. and Project Design: The Decision Tree Initiative. Espinet Alegre, X., J. Rozenberg, K. S. Rao, Washington, DC: World Bank. and S. Ogita. 2018. “Piloting the Use of Net- Rozenberg, J., C. Briceno-Garmendia, X. Lu, L. work Analysis and Decision-Making under Bonzanigo, and H. Moroz. 2017. “Improving FROM RESILIENT ASSETS TO RESILIENT INFRASTRUCTURE SERVICES 125 the Resilience of Peru’s Road Network to Cli- American Council for an Energy-Efficient mate Events.” Policy Research Working Paper Economy, Pacific Grove, CA, August 21–26. 8013, World Bank, Washington, DC. Tercek, M. 2017. “Mangroves: A Star Player in Rozenberg, J., and M. Fay. 2019. Beyond the Gap: the Coastal Protection Game.” Nature Conser- How Countries Can Afford the Infrastructure They vancy (blog), August 7. Need While Protecting the Planet. Washington, U.S. Department of Energy. 2018. “Hurricanes DC: World Bank. Nate, Maria, Irma, and Harvey Situation Rozenberg, J., C. Fox, M. Tariverdi, E. Koks, and Reports.” Office of Cybersecurity, Energy Secu- S. Hallegatte. 2019. “Road Show: Comparing rity, and Emergency Response, U.S. Depart- Road Network Resilience around the World.” ment of Energy, Washington, DC. Background paper for this report, World Bank, Van Steenbergen, F., T. Alemayehu, K. Woldeare- Washington, DC. gay, and M. Agujetas Perez. 2019. Guidelines: Schweikert, A. E., L. Nield, E. Otto, and M. Green Roads for Water—Road Infrastructure in Deinert. 2019. “Resilience and Critical Power Support of Water Managements and Climate Resil- System Infrastructure: Lessons Learned from ience. Hertogenbosch, the Netherlands: Meta Natural Disasters and Future Research Needs.” Meta Research. Background paper for this report, World Bank, Veeramany, A., G. A. Coles, S. D. Unwin, T. B. Washington, DC. Nguyen, and J. E. Dagle. 2018. “Trial Imple- State Council of China. 2015. “Guideline to Pro- mentation of a Multihazard Risk Assessment mote Building Sponge Cities.” State Council of Framework for High-Impact Low-Frequency China, Beijing. Power Grid Events.” IEEE Systems Journal 12 Stip, C., Z. Mao, L. Bonzanigo, G. Browder, and (4): 3807–15. J. Tracy. 2019. “The Resilience of Water Infra- White House. 2013. “Economic Benefits of structure: Examples from Dams, Wastewater Increasing Electric Grid Resilience to Weather Treatment Plants, and Water Supply and San- Outages.” Executive Office of the President, itation Systems.” Sector note for this report, Washington, DC. World Bank, Washington, DC. World Bank. 2015. “Disaster Risk Management Stöcker, C. 2018. “Software Defined Digital Grid in the Transport Sector: A Review of Concepts on a P2P Network: On Systems of Autono- and International Case Studies.” Working mous Energy Cells.” Medium (blog), March 12. Paper 98202, World Bank, Washington, DC. Strahl, J., M. Bebrin, E. Paris, and D. Jones. 2016. ———. 2018. Water Scarce Cities: Thriving in a Finite “Beyond the Buzzwords: Making the Specific World. Washington, DC: World Bank. Case for Community Resilience Microgrids.” Wright-Contreras, L. 2018. “VEI and Dawaco: Paper presented at the 2016 ACEEE Sum- Water Operators’ Partnership Case Study.” mer Study on Energy Efficiency in Buildings, UN-Habitat, Nairobi. From Resilient Infrastructure Services to Resilient Users 8 S o far in part II, this report has shown how infrastructure networks could be made more resilient by employing a combination of interventions in assets (strengthen- ing) and in networks (redundancy, diversification, and working across systems). These strategies offer the benefits of lower life-cycle costs of assets and more reliable services. Yet ultimately what matters is not the resilience of the supply of infrastructure services, but the resilience of the end users—the topic of this chapter. After all, infrastructure disruptions can be cata- extremely low rainfall (1/590-year events) in strophic or more benign, depending on 2014–16 and was forced to step back and take whether users—including people and supply stock of its water situation. After much deliber- chains—can cope with them. At this level, the ation, Cape Town decided to focus initially on benefit of more resilient infrastructure is a demand management (Kaiser 2018). The mea- reduction in the total impact of natural hazards sures implemented reduced water usage by 400 on people and economies. This chapter explores million liters a day (40 percent of usage) ways to reduce the vulnerability of users and to between 2015 and 2018, making it possible to make supply chains more resilient. avoid a major socioeconomic crisis. Las Vegas—by no means a low water con- REDUCING DEMAND FOR sumer, at 284 net liters per capita per day— INFRASTRUCTURE SERVICES BY has managed to reduce its residential con- IMPROVING EFFICIENCY OFTEN sumption by 40 percent since 2002 (World BUILDS RESILIENCE Bank 2018), despite its population growth With growing populations and increasingly and its 40 million visitors a year. At the other scarce—or fought-over—water resources, utili- end of the spectrum, Zaragoza, with per capita ties must manage demand to reduce stress on consumption of 99 liters per person per their city’s water supplies. A recent example is day, one of the lowest in the country and Cape Town, which had to take drastic measures worldwide, achieved a 30 percent decrease to avoid reaching “Day 0”—the day the city in consumption levels in the early 2000s, would run out of water. After relying primarily when the city launched ambitious efficiency on surface water resources for two centuries improvement programs. This reduction was through an elaborate network of elevated lakes, achieved through a combination of water the city was hit by three consecutive years of pricing adjustment, network rehabilitation, 127 128 LIFELINES and public outreach and education (World might be. Study results demonstrated that a Bank 2018). descriptive social norm measure using a neigh- Demand management recognizes that ser- borhood comparison (through stickers on vice customers are at the center of efforts to water bills) was most effective among high- build resilience in the water supply and sanita- consumption users and more effective than city- tion system. In Belen, Costa Rica, customers of wide comparisons. Among low-consumption the water utility helped to identify low-cost users, an intervention that gave customers the demand reduction measures to be tested in a information they needed to devise their own study (Datta et al. 2015). These focus group dis- water use reduction plan—with targets, mea- cussions revealed that customers generally sures, and milestones—was most effective. agreed about the importance of conserving Including users in program design through both water but did not necessarily think that they focus group and field testing yielded important themselves should reduce use and knew little findings that the Belen service provider could about what high or low water consumption incorporate into future programming. BOX 8.1 Building norms, urban forms, and behavioral changes can reduce energy demand during heat waves and prevent secondary impacts on power systems Heat waves, which are becoming increasingly much greater impact on the use of energy for air- intense and frequent, can have severe effects on conditioning in buildings (–17 percent). Finally, power systems. Urban forms and building char- behavioral change (increasing the thermostat set- acteristics can contribute to heat waves through ting) has the largest impact on energy consump- the “urban heat island” (Lemonsu et al. 2013; tion for air-conditioning (–43 percent). The higher Stone, Hess, and Frumkin 2010). Socioeconomic effects of this action highlight the importance, vulnerability and the vulnerability of power sys- beyond changes in infrastructure, of actions tar- tems, therefore, depend on the choices made geting behavioral change (figure B8.1.1). during urban planning. Viguié et al. (2019) consider three broad cat- FIGURE B8.1.1 Behavioral policies are the most egories of actions to reduce the vulnerability of efficient way to reduce energy consumption during cities to heat waves: (1) a large-scale urban recon- heat waves figuration policy, leading to the addition of many Numbers show reduced electricity consumption parks and green spaces; (2) a building-scale pol- from air conditioning during a heat wave in Paris icy, in which strict building insulation rules and 43 GWh the use of reflective materials for walls and roofs might be applied to all buildings in urban areas except historical buildings; and (3) behavioral changes in the use of air-conditioning to maintain 28°C in residential buildings and 26°C in offices 17 GWh instead of 23°C in a reference scenario. The addition of parks and green spaces across a city decreases air temperature mainly through 3 GWh evapotranspiration. However, this effect is not big enough to have a significant impact on elec- Creation of parks Stricter building Moderate use of and green insulation and air conditioning tricity consumption for air-conditioning (–2 per- spaces reflective cent). Improvements in building insulation have a materials Source: Viguié et al. 2019. FROM RESILIENT INFRASTRUCTURE SERVICES TO RESILIENT USERS 129 Similar examples can be found in the power and to identify the parts of the network to sector, where demand management can help strengthen. The importance of a bridge or a in responding to crises (box 8.1). Demand power distribution line depends on who is response is defined by the U.S. Federal Energy using it. A power distribution line that con- Regulatory Commission as changes in custom- nects a hospital or a flood shelter is likely more ers’ normal electricity consumption in response important during and after an emergency than to changes in the price of electricity over time the average distribution line in the country. A or to incentive payments designed to induce road or a bridge that is used by an on-demand lower electricity use. This mechanism can be supply chain with no inventory (such as for useful during disasters because it can help to fresh food) cannot tolerate short disruptions, reduce stress on the network. Indeed, Carlotto whereas industries with large inventories can and Grzybowski (2014) show that the size and tolerate long disruptions. scope of blackouts in a network grow with its To investigate how criticality depends on utilization rate, meaning that the closer a net- users and supply chains, Colon, Hallegatte, and work is to its operational limit, the larger the Rozenberg (2019) combine a transport and a blackouts. supply chain model to investigate the criticality Demand management was used in Texas in of the transport network in Tanzania. As 2014 when two power plants went down expected, the most critical road segments because of the cold, suddenly forcing 1,800 depend on the types of products considered. megawatts offline. Because of the extreme Map 8.1 shows how a one-week disruption in weather, the grid was already under stress, so certain roads in the Tanzanian transport net- the Electric Reliability Council of Texas work would affect four different users or sup- (ERCOT), the state’s grid operator, had to call ply chains: (1) the economy as a whole (panel for a demand response across the state to avoid a); (2) food supply chains only (panel b)— rolling blackouts. At the time, ERCOT relied a food security issue; (3) the manufacturing predominantly on large industrial customers to sector (panel c); and (4) exports (panel d), reduce their consumption of electricity. Cou- which are important for trade competitiveness pled with the use of all available power sources, (and the profitability of the port). the demand response proved to be the solution Comparison of the maps reveals that invest- to the two power plant failures. Currently, ment priorities depend on policy objectives. For automated demand response programs are example, segments of the coastal trunk road being implemented in some countries. More located about 200 km south of Dar-es-Salaam traditional ways can also be used to implement are critical for food security but rather irrele- this solution. Television, the Internet, radio, vant for manufacturing and trade. For the latter and newspapers, as well as automated phone purpose, improving the road east of Morogoro or text messaging, can be used to let customers is a priority. This segment carries large freight know when they need to reduce demand in flows moving between the port of Dar es thefaceofanextremeweatherevent(Brown,Prudent- Salaam and landlocked countries such as the Richard, and O’Mara 2016). Democratic Republic of Congo and Zambia. CRITICALITY DEPENDS ON THE END END USERS NEED TO PREPARE FOR USER: SOME ASSETS ARE CRITICAL INFRASTRUCTURE DISRUPTIONS FOR FOOD SECURITY, OTHERS FOR AND DESIGN MORE RESILIENT COMPETITIVENESS SUPPLY CHAINS Understanding the needs and capacities of End users can take a range of measures to users helps utilities to target investments better mitigate the adverse impacts of infrastructure 130 LIFELINES MAP 8.1 The criticality of a road depends on how it is used Source: Colon, Hallegatte, and Rozenberg 2019. Note: In all four panels, the width of the line overlaying a given road is proportional to the impacts that a one-week disruption of the road would trigger. Impacts are measured in % of daily consumption in the considered sector. They represent exceptional expendi- tures due to costlier transport and missed consumption due to shortages. Panels a, b, and c depict the products used by Tanzanian households. Panel d depicts the exceptional expenditures and missed purchases from international buyers. These impacts relate specifically to exports and transit flows. disruptions. Dormady et al. (2017) identify sures can be summarized along the main the coping measures that firms affected by components of a firm’s production function— Hurricane Sandy in the United States most that is, they relate to a firm’s decisions about commonly applied. The study uses survey its capital and assets and its labor, inputs, and data to estimate the costs and effectiveness of production technology (figure 8.1). In prac- the measures. The most common coping mea- tice, the measures applied by infrastructure FROM RESILIENT INFRASTRUCTURE SERVICES TO RESILIENT USERS 131 FIGURE 8.1 Firms have a wide range of coping measures that they can use to mitigate the adverse effects of infrastructure disruptions Capital Labor Inputs Technology Replace assets Hire Use more (or less) Increase e ciency Relocate Fire Substitution Back up machinery Work overtime Switch suppliers Excess inventories Financial coping measures Reduce profit margins Borrow money Obtain insurance Source: Adapted from Dormady et al. 2017. users depend on their local options and operational cost, compared with the cost of constraints. electricity from the grid (box 8.2 discusses how In areas with frequent infrastructure service Japanese firms have reduced these costs). disruptions, end users can ramp up their own These generators tend to be less affordable for resilience by investing in backup resources smaller firms with limited cash reserves (see such as generators or water and gas tanks. In chapter 2). addition, they can fill emergency generators with fuel and contact fuel suppliers with antic- Firms need to be prepared for shocks ipated needs for deliveries after the storm has that affect them indirectly through passed, as well as ensure that their business supply chains emergency supply kit is fully stocked. The U.S. Firms also need to manage supply chain issues, Federal Emergency Management Agency has which include not only transport disruptions prepared comprehensive emergency prepared- but also problems with suppliers and clients. ness materials for use in preparing for a disaster Indeed, a firm that is not affected by disasters (FEMA 2014). directly or through disrupted infrastructure In Vietnam, a recent firm-level survey indi- services may still be unable to produce because cated that firms that purchase water equip- its suppliers cannot provide the required ment such as tanks or pumps in preparation for inputs, or because its clients are not able to water outages face no impacts on production continue buying. A broader view of the full costs when water service is disrupted, com- supply chain is needed to assess disaster-related pared with an increase in production costs of production risks. 8.24 percent otherwise (Hyland et al. 2019). Firms exposed to transport or supply chain However, unlike water tanks, an electricity disruptions tend to rely on large inventories for backup capacity provided by diesel generators protection. In fact, firms in low- and middle-in- is associated with significant and additional come countries have already adapted to poor 132 LIFELINES BOX 8.2 An energy management system to bridge power outages caused by disasters: The factory grid (F-grid) project in Ohira Industrial Park in Japan Before the earthquake in eastern Japan in 2011, In February 2013, nearly two years after the Toyota’s automotive plant in Ohira village, Miy- earthquake, Toyota, in partnership with 10 corpo- agi Prefecture, north of Fukushima, had relied rations and organizations located in the industrial entirely on the Tohoku Electric Power Company park, established a limited liability partnership. for energy. However, the earthquake shut down The objective was to establish a comprehensive the power supply to the plant for two weeks, energy management system that contributes which led to considerable economic losses for to improved energy efficiency in the industrial Toyota and other companies in the surrounding park during normal times, as well as serves as industrial park as well as disruption of the supply a backup power supply system during disaster chain. To avoid such losses in the future, compa- times. Through the onsite generation of elec- nies in the industrial park sought to secure energy tricity and heat, as well as use of the commu- during power outages and shortages by building nity energy management system to balance the their own minigrid system with a comprehensive power supply optimally in the industrial park, energy management system. F-grid achieved a 24 percent increase in energy However, creating a backup power system to efficiency and a 31 percent reduction in carbon be used only during emergencies or a natural dioxide emissions in 2016, compared with those disaster is extremely costly. The companies thus of industrial parks similar in size. Overall, the recognized that they needed to build a power F-grid system not only helps the industrial park system that would be useful in both normal and to bridge power outages caused by natural disas- disaster times. They also recognized that strong ters (or other reasons) but also helps to reduce collaboration among firms to consolidate power energy costs thanks to increased efficiency. demand within the industrial park would be criti- cal to creating demand for minigrid power during normal operations. Source: World Bank 2019. infrastructure and tend to hold larger invento- dens—costly to maintain and, in some cases, ries than firms in high-income economies such as perishable goods, a source of significant (Guasch and Kogan 2003). Simulations for losses. Tanzania show that if firms maintain two Maintaining a diversity of suppliers from weeks of inventories instead of one, the costs both local and distant locations is another pow- of disaster-related transport disruptions are erful safeguard, especially in long transport or reduced by 80 percent (Colon, Hallegatte, and supply chain disruptions. Relying on a single Rozenberg 2019). In disaster-prone areas supplier is a critical vulnerability. For example, where transport disruptions are frequent but in 2011, many automakers used a paint pig- relatively short, holding larger inventories can ment called Xirallic that was produced at only be a cost-effective coping solution (Schmitt one factory in the world, the Onahama plant 2011). Firms with large inventories still suffer near the Fukushima-Daiichi nuclear power from higher transport costs due to disruptions, station in Japan.1 When the factory was evacu- but they have to interrupt their own produc- ated and closed after the earthquake, many tion processes only for long disruptions. How- automakers realized that they had no alterna- ever, excessive inventories are financial bur- tive suppliers and had to restrict sales of some FROM RESILIENT INFRASTRUCTURE SERVICES TO RESILIENT USERS 133 colors. Maintaining a diversity of suppliers, if business continuity plans and have no special- possible in different areas and using different ist in recovery following a disaster. delivery routes that cannot be simultaneously A static supply chain cannot cope with a hit by a shock, strengthens supply chains. The large-scale disaster and disruptions. Adaptabil- total benefits of more diversity in suppliers ity in supply chains is critical and should be could be large. For example, the modeling embedded in business continuity plans. For this exercise described in this report suggests that reason, a pillar of supply chain resilience is the for Tanzania sourcing critical inputs from two development of organizational capacities to suppliers instead of one reduces the indirect handle unexpected disruptions across firms costs of transport disruptions by about 70 per- (Blackhurst et al. 2005; Christopher and Peck cent. However, managing multiple suppliers 2004; Sheffi 2005). Decentralized decision creates significant transaction costs, which making and increasing communication explains why recent supply chains have tended between firms are essential for resilience (Sheffi to reduce the number of suppliers (Bakos and 2005). Specific actions include developing Brynjolfsson 1993; Berger, Gerstenfeld, and internal business continuity plans and rescue Zeng 2004; Goffin, Szwejczewski, and New plans with suppliers and collocated companies. 1997). Thus, there is a trade-off between the Since the 2011 earthquake in Japan, several efficiency of supply chains in normal times and firms have come together to redesign their their resilience to various shocks. evacuation protocols and emergency commu- Local supply chains are more robust to nication procedures and to develop new shared transport disruption, but they are more vulner- backup solutions for critical utilities (World able to direct shocks. Sourcing from local part- Bank 2019). ners decreases the reliance on transportation Business continuity plans can also be cali- and significantly reduces the risks of incurring brated by performing stress tests and exploring the indirect damages of a distant disruption. In “what if” scenarios to identify bottlenecks and Tanzania, simulations suggest that having sup- particularly vulnerable points (Chopra and pliers twice as close reduces impacts by 20 per- Sodhi 2004). Such plans should be updated cent. At the same time, local supply chains are regularly, incorporating lessons from any new more often directly affected by a shock, which disruptions (Hamel and Välikangas 2003). And makes recovery more difficult. Maintaining they should rely on sophisticated data manage- relationships with distant partners helps firms ment practices. After the 2011 earthquake in to recover when their facilities and those of Japan, which caused large production disrup- nearby partners are directly affected by a disas- tions, Toyota created a new database, Rescue, ter (Kashiwagi, Todo, and Matous 2018; Todo, for the inventories held by 650,000 suppliers Nakajima, and Matous 2015). In this way, worldwide.2 This information is being used to affected firms can receive support and help locate available resources more easily and to from their nonaffected clients and suppliers prevent bottlenecks in production processes. and do not suffer a disaster-related drop in demand that makes recovery more challeng- Critical users during disasters: the ing. One extreme example of support to and special case of hospitals from suppliers is Toyota, which in 2011 paid its Hospitals are both critical to the response to a employees to work at its suppliers so they disaster and highly vulnerable to its impacts could restore production as fast as possible. (Tariverdi et al. 2019). A disaster in a heavily This type of support makes a large difference, populated area can lead to a sudden surge in especially for small and medium enterprises, demand for regional health care services. which do not have the resources to prepare Simultaneously, health care services may be 134 LIFELINES diminished because of structural damage, loss Baum-Snow (2007a, 2007b) provides both of critical support systems such as power or empirical and theoretical evidence that post– water supply, or a reduced workforce because World War II suburbanization in the United of transport network disruptions. Regional States was driven largely by investments in response planners need to ensure that the highways that reduced travel times. Moreover, operations of critical support infrastructure are transit infrastructure investments can guide restored in a timely fashion. spatial development and influence land use, Resilience-enhancing options can be divided land use intensity, land values, and employ- into two main groups: health care alternative ment and population densities (map 8.2). Typ- operations and infrastructure improvements. ically, transit-oriented development invest- Alternative operations include (1) collaborative ments have a unique ability to influence the regional responses (such as transferring resilience of communities, because they inher- patients between hospitals) and individual hos- ently lead to concentrations of people and pital-based operational modifications (such as businesses around transit stops (Salat and Olli- increasing bed capacity by using buffered vier 2017). However, if these investments are capacity); (2) changing roles, such as nurses not made strategically, taking into account taking on roles ordinarily assigned to doctors; information on the exposure of areas to natu- (3) increasing efficiency by speeding up patient ral hazards, the outcome could be an increase care; and (4) applying alternative but lawful in vulnerability to disasters. standards of care to the discharge and transfer Infrastructure investments can be used to of patients. These measures assume that hospi- support the implementation of risk-informed tals take the necessary steps for preparedness, land use and urbanization plans and to prevent such as providing onsite family care to facilitate unplanned developments. In cities in low- and maintaining the required staff levels in a disas- middle-income countries, a large share—if not ter event, establishing relationships across hos- the vast majority—of households flock to pital units, and developing interhospital informal settlements, often on the periphery of agreements. urban areas, because they are priced out of the Infrastructure improvement includes (1) poten- narrow, formal housing market. Often these tial mitigation, such as requesting that the informal neighborhoods are located in disas- poles of power lines to the hospital be strength- ter-prone areas, because that is where land ened; (2) redundancy in access to key health tends to be available. For example, informal care facilities; (3) preparedness, such as prepo- settlements on the outskirts of Dakar, grew sitioning water reservoirs and generators and when droughts in the 1970s sparked mass medical warehouse management; (4) repairs, migration from rural areas. However, these such as reconstructing damaged facilities or land plots proved to be highly exposed to lifelines; and (5) responses, such as refueling floods (a fact only obvious once the droughts generators for an uninterrupted power supply. had ended), with the result that between 100,000 and 300,000 people were affected by INFRASTRUCTURE AFFECTS THE floods every rainy season, particularly during EXPOSURE OF USERS TO NATURAL the destructive episodes in 2009 (World Bank HAZARDS 2016). In Conakry, Guinea’s narrow peninsula Because infrastructure localization decisions capital, land is so scarce that many urban drive urbanization patterns and the exposure dwellers live in the lowest-lying areas, increas- of populations and assets to risks, they should ing their exposure to storm surges and floods, be coordinated with land use and urban plans. or directly in the mangroves, increasing the FROM RESILIENT INFRASTRUCTURE SERVICES TO RESILIENT USERS 135 MAP 8.2 The pattern of urbanization in Addis Ababa closely follows the major public transport lines Population density Transport lines Source: World Bank staff. city’s exposure to floods in the process (World ture in the 1960s before the area was occupied. Bank, forthcoming). Once these neighbor- The layout of roads created small accessible hoods have reached a critical mass, relocating blocks that would later be filled by residential households becomes very difficult. Similarly, structures. Today, a 160-square-meter house in retrofitting these neighborhoods with basic this neighborhood (which was a slum not so infrastructure and adapting them to the risk of long ago) costs $180,000 (Angel 2017). Similar natural hazards are expensive, lengthy, and models were applied to sites-and-services proj- sensitive processes. ects in India (Owens, Gulyani, and Rizvi 2018) A solution lies in equipping low-risk areas and Tanzania (Michaels et al. 2017), consisting with basic infrastructure to guide the localiza- of the provision of basic infrastructure and tion choices of people before they arrive. Such services. investments attract populations to areas that Areas to be given priority for infrastructure are relatively safe from natural hazards. Only development can be identified using simple the most basic infrastructure is needed in the geographic information system approaches. early days to guide development while pre- The goal is to identify “good” land that is safe serving the possibility of upscaling in the and close to opportunities, jobs, and the exist- future, and it is essential at the outset to secure ing network infrastructure. the rights of way for roads and sewage systems. In Fiji, the city of Nadi sought to identify This approach was followed in the Comás where future settlements and investment in squatter community in Lima, where volunteer infrastructure should be located to minimize engineering students laid out the basic struc- exposure to natural risks and the cost of devel- 136 LIFELINES opment (Government of Fiji and World Bank Digital elevation models and flood maps are 2017). Nadi Town is the third-largest urban useful as a first screen for identifying areas that center in Fiji, with a population of around might be suitable for development. In map 8.3, 52,800 (in 2016). The town is growing at the the low-lying areas of Nadi that are highly relatively fast rate of 2.5 percent a year, driven exposed to coastal and river floods are indi- by tourism, transport, and high-value real cated in red, blue, and orange. The areas that estate developments. It is acting as an eco- are considered at high or extreme risk of flood nomic magnet, and in the absence of forward in a 100-year return flood risk map are purple, planning for low-income groups, informal set- already-developed areas are gray, and areas tlements have mushroomed: 17 settlements with steep slopes are white.3 The light pink (home to 18 percent of the town’s population) areas are potentially suitable for future devel- are in unplanned areas, particularly in the opment, although further studies should be urban boundary and periurban areas. The city conducted to confirm this simple assessment, is expected to maintain this growth into the and more investment in drainage could make next decade, and regularizing the existing some of the flood-prone, low-lying areas suit- unplanned settlements and planning for the able for development. absorption of future growth are an urban man- At this point, about 4.3 square kilometers are agement and land use challenge. not developed within the town boundary (see MAP 8.3 Risk-informed urbanization planning can help to accommodate the growing urban population of Fiji while limiting the increase in natural risks Sources: Government of Fiji and World Bank 2017. FROM RESILIENT INFRASTRUCTURE SERVICES TO RESILIENT USERS 137 inset). If additional investments were made to These insights raise the question of how to improve drainage in the area, this land could be implement these solutions. What concrete a priority for future development. With future steps are necessary? What institutional systems densities of between 10 dwellings per hectare and what types of incentives, capacities, and (today’s values) and 15 dwellings per hectare, financial instruments are required to build the available area within the town boundary more resilient infrastructure? These questions could host 4,300–6,500 households. In view of are the subject of the next part of this report. the current backlog of about 2,000 units in Nadi and 300 new households a year (2.5 percent NOTES growth rate), this land could accommodate 1. “Automakers Face a Paint Shortage after Nadi’s urban growth for 8–15 years. JapanQuake,”Reuters,March25,2011.https:// Over the longer term, areas beyond the www.reuters.com/article/us-japan-pigment /automakers-face-paint-shortage-after town boundary should be considered—possi- -japan-quake-idUSTRE72P04B20110326. bly combined with an expansion of the bound- 2. “How Toyota Applied the Lessons of the ary. More than 45 square kilometers are avail- 2011 Quake,” Automotive News, April 25, able close to Nadi, but outside the town 2016. https://www.autonews.com/article boundary. That area could accommodate from /20160425/OEM/304259956/how-toyota -applied-the-lessons-of-2011-quake. 45,000 to almost 70,000 households—which is 3. The flood map is based on a flood risk assess- enough to manage rural-urban migration for ment of Nadi, Fiji, by the Pacific Community several decades. Use of this land, however, and the National Institute of Water and would require addressing issues of land tenure Atmospheric Research. and ownership and expanding networks, espe- cially for water and sanitation. REFERENCES Part II of this report has shown that building Angel, S. 2017. “Urban Forms and Future Cities: more resilient infrastructure assets is often A Commentary.” Urban Planning 2 (1). doi: http://dx.doi.org/10.17645/up.v2i1.863. costlier, but that the additional cost is small, Bakos, J. Y., and E. Brynjolfsson. 1993. “Informa- especially if countries spend well, and the tion Technology, Incentives, and the Optimal investment in more resilience is highly cost- Number of Suppliers.” Journal of Management effective. Building up the resilience of infra- Information Systems 10 (2): 37–53. structure services and users at a low cost is Baum-Snow, N. 2007a. “Did Highways Cause Suburbanization?” Quarterly Journal of Econom- possible, provided the right interventions are ics 122 (2): 775–805. implemented in the right places—for example, ———. 2007b. “Suburbanization and Transpor- by targeting investments using hazards data tation in the Monocentric Model.” Journal of and criticality analyses, by capturing synergies Urban Economics 62 (3): 405–23. across sectors using smart and robust deci- Berger, P. D., A. Gerstenfeld, and A. Z. Zeng. 2004. sion-making processes, by using nature-based “How Many Suppliers Are Best? A Decision- Analysis Approach.” Omega 32 (1): 9–15. solutions, and by capturing resilience benefits Blackhurst, J., C. W. Craighead, D. Elkins, and from diversification and decentralization. This R. B. Handfield. 2005. “An Empirically Derived part of the report has also shown that the ben- Agenda of Critical Research Issues for Manag- efits of resilient infrastructure systems can go ing Supply-Chain Disruptions.” International beyond reliable infrastructure services; well- Journal of Production Research 43 (19): 4067–81. Brown, R., G. Prudent‐Richard, and K. O’Mara. designed infrastructure system can reduce the 2016.  Enhancing Power Sector Resilience: Emerg- exposure of populations and firms to natural ing Practices to Manage Weather and Geological hazards and increase their overall resilience Risks. Energy Sector Management Assistance without having to cost more. Program. Washington, DC: World Bank Group. 138 LIFELINES Carlotto, T., and J. M. V. Grzybowski. 2014. Policy Working Paper, World Bank, Washing- “Self-Organized Criticality and Cascading Fail- ton, DC (forthcoming). ures in the Topological Model of Power Grids: Kaiser, G. 2018. “Planning for Water Security in A Case Study.” Proceeding Series of the Brazilian Cape Town.” PowerPoint presentation, World Society of Applied and Computational Mathematics Bank, Washington, DC. 2 (1): n.p. Kashiwagi, Y., Y. Todo, and P. Matous. 2018. Chopra, S., and M. S. Sodhi. 2004. “Managing “International Propagation of Economic Risk to Avoid Supply-Chain Breakdown: By Shocks through Global Supply Chains.” Understanding the Variety and Interconnect- WINPEC Working Paper E1810, Waseda Insti- edness of Supply-Chain Risks, Managers Can tute of Political Economy, Waseda University, Tailor Balanced, Effective Risk-Reduction Tokyo. Strategies for Their Companies.” MIT Sloan Lemonsu, A., R. Kounkou-Arnaud, J. Desplat, Management Review 46 (Fall): 53–62. J. L. Salagnac, and V. Masson. 2013. “Evolu- Christopher, M., and H. Peck. 2004. “Building the tion of the Parisian Urban Climate under a Resilient Supply Chain.” International Journal of Global Changing Climate.” Climatic Change 116 Logistics Management 15 (2): 1–14. (3-4): 679–92. Colon, C., S. Hallegatte, and J. Rozenberg. 2019. Michaels, G., D. Nigmatulina, F. Rauch, T. Regan, “Transportation and Supply Chain Resilience N. Baruah, and A. Dahlstrand-Rudin. 2017. in the United Republic of Tanzania.” Back- “Planning Ahead for Better Neighborhoods: ground paper for this report, World Bank, Long Run Evidence from Tanzania.” CESifo Washington, DC. Working Paper 6680, IFO Institute for Eco- Datta, S., J. J. Miranda Montero, L. D. C. Zoratto, nomic Research (CESifo), Munich. O. Calvo-Gonzalez, M. Darling, and K. J. O. Owens, K. E., S. Gulyani, and A. Rizvi. 2018. Lorenzana. 2015. “A Behavioral Approach “Success When We Deemed It Failure? Revisit­ to Water Conservation: Evidence from Costa ing Sites and Services Projects in Mumbai and Rica.” Policy Research Working Paper 7283, Chennai 20 Years Later.” World Development World Bank, Washington, DC. 106 (June): 260–72. Dormady, N., A. Rose, H. Rosoff, and A. Roa- Salat, S., and G. P. Ollivier. 2017. Transforming the Henriquez. 2017. “Estimating the Cost- Urban Space through Transit-Oriented Develop- Effectiveness of Resilience to Disasters: Survey ment: The 3V Approach. Washington, DC: World Instrument Design and Refinement of Pri- Bank. mary Data.” In Handbook on Resilience of Socio- Schmitt, A. J. 2011. “Strategies for Customer Ser- Technical Systems. Cheltenham: Edward Elgar. vice Level Protection under Multi-Echelon FEMA (Federal Emergency Management Supply Chain Disruption Risk.” Transportation Agency). 2014. “Every Business Should Have Research Part B: Methodological 45 (8): 1266–83. a Plan.” FEMA, Washington, DC. Sheffi, Y. 2005. The Resilient Enterprise: Overcoming Goffin, K., M. Szwejczewski, and C. New. 1997. Vulnerability for Competitive Advantage. Cam- “Managing Suppliers: When Fewer Can Mean bridge, MA: MIT Press. More.” International Journal of Physical Distribu- Stone, B., J. J. Hess, and H. Frumkin. 2010. tion and Logistics Management 27 (7): 422–36. “Urban Form and Extreme Heat Events: Are Government of Fiji and World Bank. 2017. Sprawling Cities More Vulnerable to Climate “Climate Vulnerability Assessment: Making Change Than Compact Cities?” Environmental Fiji Climate Resilient.” World Bank, Washing- Health Perspectives 118 (10): 1425–28. ton, DC. Tariverdi, M., H. Fotouhi, S. Moryadee, and Guasch, J. L., and J. Kogan. 2003. “Just-in-Case E. Miller-Hooks. 2019. “Health Care Sys- Inventories: A Cross-Country Analysis.” Policy tem Disaster-Resilience Optimization Given Research Working Paper 3012, World Bank, Its Reliance on Interdependent Critical Life- Washington, DC. lines.” Journal of Infrastructure Systems 25 (1): Hamel, G., and L. Välikangas. 2003. “The Quest 04018044. for Resilience.” Harvard Business Review 81 (9): Todo, Y., K. Nakajima, and P. Matous. 2015. “How 52–56. Do Supply Chain Networks Affect the Resil- Hyland, M., J. Russ, R. Damania, and A. Khalil. ience of Firms to Natural Disasters? Evidence 2019. “A Firm Commitment to Water Delivery from the Great East Japan Earthquake.” Jour- in Vietnam: A View from the Private Sector.” nal of Regional Science 55 (2): 209–29. FROM RESILIENT INFRASTRUCTURE SERVICES TO RESILIENT USERS 139 Viguié, V., A. Lemonsu, S. Hallegatte, A.-L. Beau- ———. 2018. Water Scarce Cities: Thriving in a Finite lant, C. Marchadier, V. Masson, G. Pigeon, World. Washington, DC: World Bank. and J.-L. Salagnac. 2019. “Early Adaptation ———. 2019. Resilient Industries: Ensuring Competi- to Heat Waves and Future Reduction of Air- tiveness in the Face of Natural Disasters. Washing- Conditioning Energy Use in Cities.” Unpubli- ton, DC: World Bank Group. shed manuscript. ———. Forthcoming. “Republic of Guinea: Plan- World Bank. 2016. Perspectives urbaines: Villes ning, Connecting, Financing in Conakry.” emergentes pour un Sénégal emergent. Washing- World Bank, Washington, DC. ton, DC: World Bank. PA RT A Way Forward: III Five Recommendations for More Resilient Infrastructure S o far, this report has shown that increasing the resilience of infrastruc- ture services and users is possible, thanks to a set of cost-effective and readily available options—from using stronger materials to adopting redun- dancy or nature-based solutions. This raises the inevitable question of why these options are not always implemented in practice and why infrastruc- ture systems so often are unable to cope with natural hazards. Part III of this report explores the obstacles The next five chapters consider these five that prevent those who design, build, operate, obstacles and propose recommendations for and maintain infrastructure assets and systems actions to tackle them and to improve the resil- from taking advantage of all available opportu- ience of infrastructure systems and users. nities to boost resilience. It then identifies a set Chapter 9 starts from the fundamental chal- of five recommendations that can serve as a lenge of infrastructure systems that are not starting point from which to develop a country- resilient because they are poorly designed or specific strategy to enhance infrastructure mismanaged, such as when assets are not resilience. maintained adequately. It recommends that These obstacles differ in importance and rel- governments put in place some basic institu- evance across countries: they depend on the tions, processes, and financing for managing level of income and wealth, the current extent infrastructure systems better—that is, that they and condition of the infrastructure systems in “get the basics right.” Better infrastructure gov- various sectors, and the institutional and tech- ernance is not only a prerequisite for function- nical capacity to design, build, and maintain ing and reliable infrastructure systems, but also infrastructure assets. To identify the appropriate can help to increase the resilience of infrastruc- recommendations in a country, decision mak- ture systems and their ability to cope with, and ers need to account for the local context recover from, shocks, regardless of their origin. through a country-specific process. Neverthe- But while these basic principles of good less, the most common obstacles to resilient infrastructure design and management are infrastructure can be identified (table PIII.1), important, they are by no means sufficient to and general recommendations to tackle these make infrastructure resilient—especially to obstacles can be proposed. The first obstacle rarer and higher-intensity events, such as hurri- impairs infrastructure management in general, canes, earthquakes, and major floods. More- while the other four obstacles are about infra- over, good infrastructure management does not structure resilience in particular. guarantee that climate change and other long- TABLE PIII.1 Key obstacles to more resilient infrastructure services and examples of underlying causes Obstacles to good infrastructure management Obstacles to infrastructure resilience Poor design, operation, Political economy Lack of Inadequate data, and maintenance of challenges and incentives to models, skills, Affordability and infrastructure systems coordination failures increase resilience or tools financing constraints • Absence of local standards, • Invisibility of • Infrastructure service • Lack of data, • Lack of resources codes, and regulations (or resilience benefits providers not bearing methodologies, for risk-informed lack of enforcement) • Interdependency the full cost of or technical skills planning and risk • Underfinanced or of infrastructure disruptions • Designs often based assessment at early understaffed regulators systems • Lack of incentives on historical data and stages of project • Insufficient resources for the • Synergies and trade- to protect or restore not on future hazards design early-stage design of the offs across different ecosystems and climate change • Lack of resources infrastructure system and risks or infrastructure • Overconfidence in in postdisaster assets systems model results and situations • Borrowing constraints and • Narrow mandates of historical data • Lack of information affordability issues institutions • Insufficient and transparency on • Lack of financing and consideration of low- infrastructure asset capacity for asset probability scenarios resilience maintenance term environmental and socioeconomic trends hazards and climate change; improving deci- will be planned for. To address these issues, sion making and minimizing the potential for decision makers need to tackle four more obsta- catastrophic failures; and building the skills cles that are specific to resilience to natural haz- needed to use the data and models. ards and climate change. These four obstacles Chapter 13 examines affordability and lead to four additional recommendations. financing issues for resilience. It recommends Chapter 10 explores the challenges of politi- providing adequate funding to include risk cal economy and the coordination failures that assessments in master plans and early project impede the creation of a resilient infrastructure design, developing government-wide financial ecosystem. It recommends creating a whole-of- protection strategies and contingency plans, government coordination mechanism for resil- and promoting transparency to better inform ient infrastructure—along with identifying investors and decision makers. critical infrastructure, defining acceptable (and These recommendations are not indepen- intolerable) risk levels, and ensuring equitable dent. They need to be coordinated and designed access to resilient infrastructure. together. For example, a new institution in Chapter 11 examines why public and pri- charge of infrastructure resilience needs to vate decision makers often do not have suffi- have appropriate incentives, capacity, and bud- cient incentives to create more resilient infra- get to be effective. And a financing initiative, structure systems. It recommends including such as a disaster risk financing strategy, can be resilience consideration in regulations and used to create the right institutions or to build financial incentives to align the interest of capacity. Thus, a comprehensive approach to infrastructure service providers with the public these obstacles and recommendations is interest and updating them regularly to account necessary. for climate change and other long-term trends. The good news is that these measures would Chapter 12 focuses on the lack of data, also contribute to better management of infra- models, and tools that make it difficult for structure systems in general and, therefore, do infrastructure service providers to implement more than increase resilience. They would resilience-building solutions. It recommends enhance the quality of infrastructure systems investing in freely accessible data on natural and make them more efficient and reliable. The Foundation for Resilient Infrastructure 9 OBSTACLE RECOMMENDATION ACTIONS  oor design, operation, P Get the basics right 1.1: Introduce and enforce regulations, construction •  and maintenance of codes, and procurement rules infrastructure systems 1.2: Create systems for appropriate operation, •  maintenance, and postincident response 1.3: Provide appropriate funding and financing •  for infrastructure planning, construction, and maintenance THE OBSTACLE: MANY clogging of existing systems by solid waste INFRASTRUCTURE SYSTEMS ARE explain the regular occurrence of flooding POORLY DESIGNED, OPERATED, OR in many large cities. This problem largely MAINTAINED explains the recurrent transport disruptions in In some countries, infrastructure disruptions Dar es Salaam during the rainy season. Struc- are chronic events: power outages occur every tural weaknesses sometimes cause bridges to day, water supply is intermittent, and conges- collapse without any external shock, as hap- tion makes travel slow and unpredictable. And pened in Genoa, in 2018. On August 14, 2003, in many places, a large fraction of these disrup- a large blackout occurred in northeastern tions is not explained by any external factor North America due to a combination of line like a natural hazard. Sometimes, infrastruc- contacts with overgrown trees and a technical ture systems are simply insufficient to meet failure of the alarm system in the utility con- demand, leading to disruptions. This is the case trol room (North American Electric Reliability when a lack of power generation forces utilities Council 2004). to shed load, which causes regular outages, or How can governments build more resilient when a lack of public transit and transport infra- infrastructure systems? The first step is to make structure leads to massive congestion in fast- them reliable in normal conditions by ensur- growing cities. ing appropriate design, operation, and main- Sometimes, disruptions occur due to sys- tenance. A power system with tight capacity tem failures that are the result of poor asset constraints and regular load shedding cannot design or insufficient maintenance. For exam- be made resilient to storms or heat waves. ple, inadequate drainage infrastructure and Maintenance is particularly critical: culverts 143 144 LIFELINES FIGURE 9.1 Infrastructure quality correlates strongly with governance standards a. Overall infrastructure b. Electricity 2.5 Governance index (control of corruption) Governance index (control of corruption) 2.5 2 2 1.5 1.5 1 1 0.5 0.5 0 0 –0.5 –0.5 –1 –1 –1.5 –1.5 –2 –2 1 2 3 4 5 6 7 1 2 3 4 5 6 7 Worst Best Worst Best WEF Infrastructure Quality Index WEF Infrastructure Quality Index Source: Kornejew, Rentschler, and Hallegatte 2019. Note: The World Economic Forum (WEF) Infrastructure Quality Index measures the quality of infrastructure services from 1 (extremely underdeveloped) to 7 (well developed and efficient by international standards). The Worldwide Governance Indicators report estimated governance standards for more than 200 countries and territories. For each country, these governance indicators reflect surveys of a large number of enterprises, citizens, and experts, based on more than 30 individual data sources. Control of cor- ruption captures perceptions of the extent to which public power is exercised for private gain, including both petty and grand forms of corruption, as well as “capture” of the state by elites and private interests. cannot protect a road if they are blocked by ity exists for roads (Kornejew, Rentschler, and solid waste, transmission lines fail if nearby Hallegatte 2019). Similar patterns also appear vegetation is not properly maintained, and for different WGI subindexes, such as for reg- leaking water pipes increase a water system’s ulatory quality and government effectiveness. vulnerability to droughts. Quality infrastructure does not have to be But a government’s ability to implement reserved for rich countries. The data behind resilience-building options depends on whether figure 9.1 suggest significant differences in it has effective systems in place to implement, infrastructure quality for countries at the same finance, manage, and maintain infrastructure income level. At low income levels, the dif- assets. Strong institutions, clear assignment of ference is particularly large. For example, the responsibilities, and transparent and reliable reliability of electricity in Bhutan, whose gross financing mechanisms are all essential to ensur- domestic product (GDP) per capita is $2,500, ing the effective provision of public services. In is comparable to that of many middle- and other words, good governance matters greatly high-income economies, whereas Nigeria, for infrastructure quality. whose GDP per capita is $2,476, has some of the Data show a clear correlation between gov- most frequent power outages of all countries. ernance and infrastructure quality. Figure 9.1 This difference in governance and quality of presents the relationship between the World- investments may explain why infrastructure wide Governance Indicators (WGI, World Bank, projects do not always deliver the intended n.d.) subindex on corruption and the Infra- benefits. For example, investments in electric- structure Quality Index of the Global Compet- ity infrastructure have a mixed track record. itiveness Report (WEF 2018). Both panel a on Some studies show that electrification leads to infrastructure in general and panel b on the a significant increase in school enrollment or electricity sector show that as the quality of years of schooling, while other studies estimate governance improves, so does the quality of that it has no or few impacts on educational infrastructure.1 The same positive correlation outcomes.2 Regarding the impact on health, between corruption and infrastructure qual- Brass et al. (2012) and Samad et al. (2013) find THE FOUNDATION FOR RESILIENT INFRASTRUCTURE 145 the same lack of conclusive evidence. Mean- ship between the reliability of and investment while, several impact assessments of electrifi- spending on transport infrastructure. Transport cation projects show significant increases in reliability is proxied by the timeliness subindi- household income and female employment, cator of the Logistics Performance Index (LPI), while others do not.3 Thus, simply spending and transport investment data are from the money on infrastructure does not always yield Organisation for Economic Co-operation and benefits to users: investments need to be well Development (OECD) (box 9.1). designed and well implemented. The results show that when spending and To explore the relationship between spending governance improve together, higher spend- more and spending better, Kornejew, Rentschler, ing significantly improves transport reliability and Hallegatte (2019) explore the relation- (figure 9.2). In fact, doubling current spend- BOX 9.1 Data on infrastructure spending are scarce and limited Data on infrastructure investment and mainte- in an effort to estimate harmonized spending in nance are rarely available (Fay et al. 2019). Few low- and middle-income countries, although their countries have common or harmonized public estimates do not differentiate among infrastruc- accounting standards, and the relevant expen- ture sectors. diture items can be mixed in with other types Data on the performance and reliability of of expenditures. Especially in low- and middle- infrastructure services are also limited. The World income countries, the level of transparency Bank’s Enterprise Surveys provide data on the varies, and so does the definition of infrastructure total annual duration of blackouts and the total maintenance spending. annual duration of water supply disruptions, as Moreover, disaggregating public spending well as a subjective index (1–4) of the severity of data by sector is difficult because the organiza- transport problems in overall business operations tion of public budgets makes it difficult to dis- (see chapter 2). No similar information is given on tinguish public spending on water, electricity, telecommunications disruptions. and transport infrastructure. Rather than rely on In this chapter, transport reliability is mea- nationally published budget figures, Kornejew, sured using the LPI, which is a benchmarking tool Rentschler, and Hallegatte (2019) use two main created to help countries identify the challenges international sources of investment data, ensur- and opportunities they face in their performance ing consistency and comparability. on trade logistics (World Bank 2018a). LPI 2018 The OECD International Transport Forum allows comparisons across 160 countries and data­b ase provides transport infrastructure offers country-specific scores along six dimen- investment spending for 57 middle- and high- sions: (1) customs, (2) infrastructure, (3) inter- income countries, covering the years 1995–2016 national shipments, (4) logistics competence, (OECD 2018). These data are supplemented by (5) tracking and tracing, and (6) timeliness. The public infrastructure investment data from the infrastructure subindicator aggregates a qual- World Bank’s BOOST Initiative, which are avail- ity scoring of ports, railroads, roads, and infor- able through the Open Budgets portal (World mation technology. The timeliness subindicator Bank 2018b). Together, these data sources yield measures reliability rather than quality per se. a panel of 603 individual country-year obser- By scoring the timeliness of shipments in reach- vations from 85 countries, covering all income ing their destination within the scheduled deliv- groups. However, because of the OECD’s focus ery time, this subindicator is a measure of unex- on high-income countries, infrastructure spend- pected transport disruptions rather than average ing data for low- and middle-income countries performance. All LPI indicators are scored on a remain patchy. Fay et al. (2019) address this gap scale from 1 (worst) to 5 (best). 146 LIFELINES FIGURE 9.2 Spending more improves the reliability of the these results have to be considered cautiously, transport system, especially if governance also improves because of the small number of countries (only 5 33) for which data are available and the fact Best Logistic Performance Index: Timeliness that public spending in the energy sector is an imperfect proxy for total (public and private) 4 spending in the power sector. In the water sec- tor, for the set of 32 countries for which data 3 are available, the average daily hours of unin- terrupted water supply are explained in part 2 by governance indicators and, as in the power sector, public water spending does not seem to Worst 1 influence water outages. 0 200 400 600 800 1,000 1,200 1,400 What would be the benefits of improving Annual public road spending per capita (constant 2009 US$) the quality of infrastructure spending? Korne-  Spending and governance improve together  Increase in spending alone jew, Rentschler, and Hallegatte (2019) consider Source: Kornejew, Rentschler, and Hallegatte 2019. an ambitious but feasible governance reform. Under the WGI subindicator for govern- ment effectiveness, 10 percent of the sample’s ing levels is estimated to increase transport country-year observations achieved at least infrastructure performance (as measured by a 0.23 index point increase over a three-year the LPI indicator for timeliness) by roughly period (for example, Ecuador between 2010 0.27 index points. For example, this improve- and 2013). By assuming the same ambitious ment corresponds to improving Mozam- governance improvement in every country, bique’s transport service reliability to equal the model described in this report yields an that of Cambodia. estimate of how much transport infrastructure But if governance quality is held constant, spending could be reduced without reducing the impact of spending more is largely muted. the quality of transport service.4 With governance unchanged, the benefit of The analysis suggests that the potential sav- spending $1 on transport reliability remains ings from improved governance could be sub- larger than zero, but it is reduced by a factor stantial (figure 9.3). Specifically, such an ambi- of six. In other words, increasing spending tious but realistic governance reform could and improving governance in parallel enhance allow countries to cut their road expenditures transport system performance on average six by 30–90 percent over the long term with- times faster than increasing spending alone. out reducing the performance of their trans- Statistically, the results given in figure 9.2 sug- port system. Relative savings are the highest gest that only about 8 percent of the variation for countries with poor governance quality in transport reliability across countries can be but relatively high levels of per capita spend- explained by investment spending, whereas ing, such as Haiti and Sierra Leone. Savings about 44 percent is explained by a country’s are small in countries with good governance governance quality. (such as New Zealand) and in countries with Similar results can be found for energy low spending on roads (such as Niger). These and water. In the power sector, governance findings highlight the importance not only of explains most of the difference in the annual spending enough on infrastructure but also of duration of blackouts across countries, while spending well to ensure that infrastructure ser- public energy spending explains very little. But vices perform well. FIGURE 9.3 Potential savings on road spending from governance reforms Haiti Guinea-Bissau Sierra Leone Togo Afghanistan Gabon Liberia Paraguay Lesotho Solomon Islands Guatemala Myanmar Angola Moldova Ukraine Cameroon Benin Mongolia Mozambique Fiji Ethiopia Azerbaijan São Tomé and Príncipe Peru Bangladesh Uganda Pakistan Romania Dominican Republic Saint Lucia Russian Federation Armenia Kenya Mali Tunisia Senegal Albania Bulgaria El Salvador Bhutan China Greece North Macedonia Argentina Burkina Faso Namibia Italy Turkey Hungary Cabo Verde Mexico Tanzania Croatia Slovak Republic Georgia Korea, Rep. Brazil Mauritius Uruguay Slovenia Chile Guinea India Estonia Czech Republic Ireland Montenegro Costa Rica Latvia Luxembourg Norway Australia Switzerland Japan Malta Lithuania Spain Poland France United States Austria Belgium Denmark Netherlands Finland Sweden United Kingdom Iceland Germany Canada New Zealand Burundi Portugal Niger 0 20 40 60 80 100 % of estimated savings from feasible governance improvements Source: Kornejew, Rentschler, and Hallegatte 2019. 147 148 LIFELINES RECOMMENDATION 1: To make the best use of resources like GET THE BASICS RIGHT these, countries should rely on local organiza- What are the solutions for coping with these tions to translate international standards into challenges for the design, operation, and main- standards relevant for the country context. tenance of infrastructure? Beyond the usual In particular, countries at different income recommendations on regulation and gover- levels—or with different preferences in terms nance, which are well treated in other reports, of reliability—will want to design regulations this report identifies three basic actions that and codes that are adapted to their needs. In are essential for better managing infrastructure the absence of a standardization body and systems. centers of technical expertise, many low- and middle-income countries use standards from Action 1.1: Introduce and enforce high-income countries, which do not take regulations, construction codes, and into account the local context. For example, procurement rules Mozambique’s National Administration of Well-designed regulations, codes, and pro- Roads designs and builds roads using the 2001 curement rules are the simplest approach to draft standards of the South African Transport enhancing the quality of infrastructure services, and Communication Commission (SATCC) including their reliability and resilience. In as a guide. The SATCC standard is in turn an the most widely applied solution, the govern- adaptation of U.S. and European standards to ment defines the level of service expected from the South African context. To adapt standards public or private infrastructure providers and to their contexts, most countries have a stan- applies it through its procurement rules (when dardization agency that is a member of the the asset is publicly owned—for example, ISO. However, the capacity of these national roads), its market regulations (when private agencies to adapt international standards and actors provide services such as electricity), or a their scope of work vary widely, so they can- contractual engagement (for example, through not always develop the local standards that are performance indicators for the procurement needed for the development of construction and monitoring of public-private partnerships). codes. Regardless of the financial model, strong pro- A particularly important issue is quality curement rules and appropriate performance control and the enforcement of construction indicators in tender processes can ensure a codes. In an analysis conducted for this report, minimum level of service and reliability. Miyamoto International (2019) points out that Countries can define construction codes enforcing construction codes and standards and regulations based on existing international is costly and more challenging than defining standards. Organizations such as the Interna- them. Enforcement in the infrastructure sector tional Organization for Standardization (ISO) requires a robust legal framework and strong and the American Society for Testing and regulatory agencies able to monitor construc- Materials (ASTM) International are creat­ ing tion and service quality and performance and international standards for the components of to reward or penalize service providers based infrastructure systems.5 For example, a stan- on their performance. Many regulators lack dard from the ASTM subcommittee on steel the resources and capacity to enforce exist- reinforcement (A01.05) provides a tool to pro- ing construction codes. As a result, expensive mote the long-term strength of bridges and infrastructure systems may be designed with support the production of high-performance, inappropriate materials or technologies, lead- corrosion-resistant steel (A1055/A1055M). ing to very high costs over the long term. THE FOUNDATION FOR RESILIENT INFRASTRUCTURE 149 Action 1.2: Create systems for A complex asset management system also appropriate operation, maintenance, documents the functional context in which the and postincident response infrastructure delivers its services. It identifies Operations and maintenance are critical to the related infrastructure systems that affect ensure the performance of infrastructure sys- its ability to deliver the services required, the tems and to reduce investment costs (see contact people, and the details of collaborative part II of this report). Poor maintenance can maintenance. Whatever form it takes, effective increase infrastructure investment needs by asset management relies on stakeholder com- 50 percent in the transport sector and by more mitment, effective institutions, and adequate than 60 percent in the water sector (Rozen- resources. berg and Fay 2019). And an analysis focusing One solution that is widely used for the on OECD countries performed for this report maintenance of transport infrastructure, espe- suggests that each additional $1 spent on road cially roads, is performance-based contracts maintenance saves on average $1.50 in new (PBCs) (Iimi and Gericke 2017; Lancelot 2010). investments, making better maintenance a These contracts explicitly link payment of con- very cost-effective option (Kornejew, Rent- tractors to the performance of assets, provid- schler, and Hallegatte 2019). ing a powerful incentive for the contractors How can proper maintenance be ensured? maintaining or operating an asset to ensure An important tool is the infrastructure asset that its reliability is accounted for in all deci- management system, which utilities can use sions. However, designing and implementing to better manage their operations. Such a sys- PBCs requires capacity on behalf of both the tem includes an inventory of all assets and government and contractor, and allocating too their condition, as well as all of the strategic, much risk to the contractor can have signifi- financial, and technical aspects of the man- cant impacts on costs or place the PBC at risk of agement of infrastructure assets across their failure (Henning, Hughes, and Faiz 2018). life cycle. The objective is to move toward an Even with preventive maintenance, the evidence-based and preventive maintenance capacity to respond quickly to incidents and to schedule and to move away from a reactive dispatch teams and resources to repair dam- approach to maintenance. aged or failing assets is critical for a reliable A simple infrastructure asset management infrastructure system. Chapter 7 describes in system focuses on each asset, independent of detail the need for emergency management or the system in which it functions. The system contingency plans for postdisaster situations, includes how much assets cost, who is respon- but these plans need to extend to smaller, sible for maintaining them, their condition and isolated incidents that can easily propagate functionality, and when they require rehabil- through an infrastructure network and have itation. A more complex asset management significant system-scale impacts. Thus, utili- system includes photographs and plans of ties and agencies need information-gathering all assets, their component parts, their main- systems and contingency plans, clear attri- tenance schedules, and details of all actions bution of responsibility in case of incidents, involving the asset since it was designed. It and an appropriate stock of parts and emer- includes an estimate of the life-cycle costs of gency equipment. Countries that are unable to the asset, the actual depreciation each year, respond quickly to isolated system failures are amortization details, and possible development obviously unable to deal with natural disas- to better align the current components with ters, where the spatial scale of the damages is the changing needs of users and their clients. usually much larger. 150 LIFELINES Action 1.3: Provide appropriate funding Moreover, the design and preparation of and financing for infrastructure infrastructure projects are very expensive. planning, construction, and During the early stages of project preparation, maintenance mobilizing resources is particularly challeng- The quality of infrastructure services depends ing. As a result, preparation budgets tend to be on many factors—from good planning to good small, making it difficult to conduct the sophis- maintenance—and each of these factors has a ticated analyses needed, even if they can gen- cost (figure 9.4). If resources are insufficient erate massive savings over the lifetime of an to meet the needs for any of these factors, the infrastructure asset. quality of infrastructure services is likely to Funding of maintenance is also often chal- suffer. Therefore, countries need to provide lenging. Underinvestment in operation and sufficient resources to meet their objectives in maintenance is common, because it is gener- terms of infrastructure services and resilience, ally easier to raise resources to finance new and they have to distribute these resources investments or a major rehabilitation than to appropriately across the various needs. Even if cover continuous operation and maintenance total spending is appropriate, allocating insuf- costs. Maintenance is also less visible than ficient resources for planning, designing, or new investments and can usually be delayed, maintaining assets would lead to low quality which makes it an easy target for budget cuts and reliability. (Briceno, Estache, and Shafik 2004; Regan Underfunded and understaffed regulators 1989). Appropriate and reliable budgetary allo- or agencies are unlikely to design and enforce cations—or the use of contracts that effectively efficient regulations or create the master plans precommit adequate maintenance expen- that maximize the reliability of infrastructure ditures, such as private-public partnerships systems. Increasing the resources available for and PBCs—are necessary to ensure that good infrastructure regulators can therefore have maintenance can actually happen. transformational impacts. These impacts may Financial constraints can push countries include enhancing the regulation of energy toward solutions that have lower up-front and water utilities, boosting the capacity of costs, even if these options have higher life- road agencies, or creating open-data portals cycle costs or major social costs. Countries with and asset management systems. Thus, appro- fragile infrastructure systems often spend large priate funding of enforcement agencies is a amounts to repair and maintain this infrastruc- priority. ture, compounding the challenge of limited fis- FIGURE 9.4 The full cost of infrastructure includes multiple cost components Cost to regulators and government Life-cycle cost to (public or private) infrastructure service • Master planning, and regulation providers Full design and enforcement • Project design and preparation infrastructure • Data and model development, • Up-front investment cost cost research, training, and education • Operational, maintenance, and repair costs • Decommissioning THE FOUNDATION FOR RESILIENT INFRASTRUCTURE 151 cal space to finance an investment that could shocks and adapting infrastructure systems to improve reliability and reduce vulnerability. changing climate conditions involve additional Escaping this vicious circle of high fragility, high obstacles and require additional actions. They maintenance, and low investment requires a are the topic of the next four chapters. temporary increase in spending. But governments in both low- and middle- NOTES 1. The Worldwide Governance Indicators report income countries and high-income countries (World Bank, n.d.) estimates governance stand- already struggle to finance the infrastructure ards for more than 200 countries and territories investment needed to meet demand. Many along six dimensions: (1) voice and accounta- infrastructure systems struggle to meet normal bility, (2) political stability and absence of vio- demand, with inadequate power generation lence, (3) government effectiveness, (4) regula- tory quality, (5) the rule of law, and (6) control capacity, unreliable Internet services, or highly of corruption. For each country, these govern- congested public transit and urban roads, even ance indicators reflect surveys of a large num- in normal times. Systems that cannot satisfy ber of enterprises, citizens, and experts, based normal demand are naturally highly vulner- on more than 30 individual data sources. 2. For details, see Bensch, Kluve, and Peters able to any shock that reduces supply—for (2011); Khandker, Barnes, and Samad (2009); instance, the failure of one power plant or Kumar and Rauniyar (2011); and Lee, Miguel, transmission line or the closure of a road. and Wolfram (2016). Where governments struggle to raise finance 3. For details, see Dinkelman (2011); Grogan and for economically and financially viable invest- Sadanand (2013); Lee, Miguel, and Wolfram (2016); Rud (2012); and van de Walle et al. ments in infrastructure, one option is to turn (2017). to the private sector. Private investors may raise 4. Specifically, the model described in the previous finance on the basis of future cash flows gen- section is extended by interacting log per capita erated by the asset itself (project finance) or, road spending with the subindicator for gov- in the case of utility companies, their own bal- ernment effectiveness. This process is necessary to generate meaningful variation across coun- ance sheet (corporate finance). Either approach tries—see Kornejew, Rentschler, and Hallegatte reduces the burden on the government balance (2019) for details. sheet—although not entirely, since almost all 5. For an example for highways, see https://www infrastructure investment creates contingent .astm.org/ABOUT/OverviewsforWeb2015 liabilities for the government where private /HighwaysOvrvwApril2018.pdf. investors cannot or will not bear the risks and maybe even direct liabilities where subsidies REFERENCES Bensch, G., J. Kluve, and J. Peters. 2011. “Impacts of continue to be required. Moreover, attracting Rural Electrification in Rwanda.” Journal of Devel- private investment—and making sure those opment Effectiveness 3 (4): 567–88. https://doi.org investors are incentivized to deliver high- /10.1080/19439342.2011.621025. quality, efficient, and resilient infrastructure— Brass, J. N., S. Carley, L. M. MacLean, and E. Bald- also depends on the quality of the governance win. 2012. “Power for Development: A Review of Distributed Generation Projects in the Devel- and regulatory environment. oping World.” Annual Review of Environment and Resources 37 (1): 107–36. https://doi.org/10.1146 Implementing these recommendations /annurev-environ-051112-111930. would improve infrastructure system design, Briceno, C., A. Estache, and N. T. Shafik. 2004. “Infrastructure Services in Developing Countries: management, and maintenance, and do much Access, Quality, Costs, and Policy Reform.” Pol- to improve the quality of infrastructure ser- icy Research Working Paper 3468, World Bank, vices and their resilience to frequent shocks. Washington, DC. https://doi.org/10.1596/1813 However, building resilience to more intense -9450-3468. 152 LIFELINES Dinkelman, T. 2011. “The Effects of Rural Electri- Miyamoto International. 2019. “Overview of Engi- fication on Employment: New Evidence from neering Options for Increasing Infrastructure South Africa.” American Economic Review 101 (7): Resilience.” Background paper for this report, 3078–108. World Bank, Washington, DC. Fay, M., S. Han, H. I. Lee, M. Mastruzzi, and M. North American Electric Reliability Council. 2004. Cho. 2019. “Hitting the Trillion Mark: A Look “Technical Analysis of the August 14, 2003, at How Much Countries Are Spending on Infra- Blackout: What Happened, Why, and What Did structure.” Policy Research Working Paper 8730, We Learn?” North American Electric Reliability World Bank, Washington, DC. Council, Princeton, NJ. Grogan, L., and A. Sadanand. 2013. “Rural Electrifi- OECD (Organisation for Economic Co-operation and cation and Employment in Poor Countries: Evi- Development). 2018. “International Transport dence from Nicaragua.” World Development 43 (C): Forum—OECD.Stat.” OECD, Paris. 252–65. Regan, E. V. 1989. “Holding Government Officials Henning, T. F. P., J. Hughes, and A. Faiz. 2018. Accountable for Infrastructure Maintenance.” “Incorporating Criticality and Resilience into Per- Proceedings of the Academy of Political Science 37 (3): formance Based Contracts.” Technical Report, 180. https://doi.org/10.2307/1173761. World Bank, Washington, DC. Rentschler, J., M. Kornejew, S. Hallegatte, M. Obo- Iimi, A., and B. Gericke. 2017. “Output- and Per- lensky, and J. Braese. 2019. “Underutilized formance-Based Road Contracts and Agricul- Potential: The Business Costs of Unreliable Infra- tural Production: Evidence from Zambia.” Pol- structure in Developing Countries.” Background icy Research Working Paper 8201, World Bank, paper for this report, World Bank, Washington Washington, DC. DC. Khandker, S. R., D. F. Barnes, and H. A. Samad. Rozenberg, J., and M. Fay. 2019. Beyond the Gap: How 2009. “Welfare Impacts of Rural Electrification: Countries Can Afford the Infrastructure They Need A Case Study from Bangladesh.” World Bank, While Protecting the Planet. Washington, DC: World Washington, DC. https://doi.org/10.1596/1813 Bank. -9450-4859. Rud, J. P. 2012. “Electricity Provision and Industrial Kornejew, M., J. Rentschler, and S. Hallegatte. 2019. Development: Evidence from India.” Journal of “Well Spent: How Governance Determines the Development Economics 97 (2): 352–67. Effectiveness of Infrastructure Investments.” Samad, H. A., S. R. Khandker, M. Asaduzzaman, and Background paper for this report, World Bank, M. Yunus. 2013. “The Benefits of Solar Home Washington, DC. Systems: An Analysis from Bangladesh.” Pol- Kumar, S., and G. Rauniyar. 2011. “Is Electrification icy Research Working Paper 6724, World Bank, Welfare Improving? Non-Experimental Evidence Washington, DC. from Rural Bhutan.” MPRA Paper 31482, Har- van de Walle, D., M. Ravallion, V. Mendiratta, and vard University, Cambridge, MA; Asian Develop- G. Koolwal. 2017. “Long-Term Gains from Elec- ment Bank, Manila, February 20. trification in Rural India.” World Bank Economic Lancelot, E. 2010. “Performance Based Contracts in Review 31 (2): 385–411. the Road Sector: Towards Improved Efficiency WEF (World Economic Forum). 2018. The Global in the Management of Maintenance and Reha- Competitiveness Report 2017–2018. Cologny: WEF. bilitation—Brazil’s Experience.” Transport Paper World Bank. 2018a. Logistics Performance Index. TP-31, World Bank, Washington, DC. Washington, DC: World Bank. Lee, K., E. Miguel, and C. Wolfram. 2016. “Exper- ———. 2018b. “BOOST Initiative.” Open Budgets imental Evidence on the Demand for and Costs portal, World Bank, Washington, DC. http://boost of Rural Electrification.” NBER Working Paper .worldbank.org/boost-initiative. 22292, National Bureau of Economic Research, ———. No date. Worldwide Governance Indicators. Cambridge, MA. http://info.worldbank.org/governance/wgi. Build Institutions for Resilience 10 OBSTACLE RECOMMENDATION ACTIONS Political economy Build institutions for 2 .1: Implement a whole-of-government approach •  challenges and resilience to resilient infrastructure, building on existing coordination failures regulatory system 2 .2: Identify critical infrastructure and define •  acceptable and intolerable risk levels 2 .3: Ensure equitable access to resilient •  infrastructure THE OBSTACLE: MULTIPLE ience in the power sector raised the distribution POLITICAL ECONOMY CHALLENGES price of electricity by up to 30 percent in some AND COORDINATION FAILURES regions, sparking strong public and political IMPEDE PUBLIC ACTION ON reactions and adjustment of the Electricity Mar- RESILIENCE ket Act (OECD 2019). Policy makers’ incentives to invest in more Lack of coordination among actors is also a resilient infrastructure systems are weakened challenge. Coordination is needed to ensure by the asymmetry in the visibility of the costs that actions by stakeholders are consistent and and benefits of such investments. Catastrophes synergetic. For example, a public-private insur- make headlines and produce in-depth analyses ance scheme regulated by the ministry of of what went wrong, but extreme natural finance at the national level cannot be designed events that lead to no or minor damage do not without considering risk reduction measures attract much interest. Understanding the bene- such as land use plans and building norms at fits of adding resilience to infrastructure the local level (tasks often led by local authori- requires identifying the crises that are avoided ties). Power outages can have secondary effects (thanks to previous policies or measures), on telecommunications, water treatment, and which is hard to do and even harder to com- urban transit systems—and power generation municate to the public. By contrast, the cost of utilities, especially coal power plants, can be making infrastructure more resilient is easy to dependent on the transport system for supplies. identify and contest. In Finland, increased resil- Kunreuther and Heal (2003) explore this inter- 153 154 LIFELINES BOX 10.1 A new hazard: Cyberdisasters and cyberattacks Modern infrastructure systems are programmed being integrated into electric grids. Because all and controlled by computer systems, making of these devices are active participants in grid them vulnerable to cyberattacks. Take the case operations, they open vulnerable new access of Ukraine in December 2015, when a large-scale points for cyberattacks on the grid (Cleveland power grid hack left 230,000 people without and Lee 2013). The vulnerability of water and power. In typical fashion, the hackers first gained transport networks may also increase with the access to control systems and then launched an rise of smart transport solutions and the increas- attack and blocked attempts at quick fixes to ing digitization of infrastructure systems. For reinstate services (Wagner 2016). In this attack, example, Zou, Choobchian, and Rozenberg the hackers gained access to the business net- (2019) warn that autonomous mobility systems, works of the utilities using a phishing program. which move people and goods around using They eventually managed to obtain a worker’s self-driving vehicles, are particularly vulnerable credentials for the control system, which enabled to cyberthreats. them to understand the programs that controlled Overall, it is important to weigh efficiency the electricity networks. The next step was to gains from smart systems against the vulnerabil- overwrite operational programs with malicious ities they may create. Smart grid and consumer versions that would stop operators from reclos- demand response technologies may allow utili- ing tripped circuit breakers. In this way, the hack- ties to balance electrical loads more effectively, ers were able to manipulate multiple utilities but these benefits should be weighed against and substations and then to trip all of the circuit the higher risk of cyberattacks on the grid. Even breakers simultaneously (Zetter 2016). though a growing literature has investigated the Past cyberattacks have been limited mostly vulnerability of smart grids to cyberattacks (Aloul to attacks on power systems, which is why et al. 2012), the trade-offs remain complex. The electricity systems are the focus of cybersecu- inadequacy of existing cybersecurity measures rity efforts. More and more Internet of Things may mean that the efficiency gains from smart devices (such as smart meters) and distributed grid and consumer demand response systems energy resources (such as small-scale battery are not fully justified because of the increase in storage systems or photovoltaic systems) are risks from cyberattacks. Source: Eugene Tan. dependency theoretically, showing that, in the power outages—both of which can increase absence of cooperation mechanisms, individual vulnerability to cyberattacks (box 10.1). Even actors may prefer not to invest in resilience. when considering a single risk, reducing the One major challenge in risk management is impact of a frequent occurrence can lead to an to look across risks and threats, even beyond increase in vulnerability to rarer and more dan- natural risks and climate change, to capture gerous events. For example, when dikes pre- synergies and avoid instances in which reduc- vent frequent floods, people wrongly assume ing one vulnerability increases another. Many that floods are now impossible. Such a wrong solutions seem attractive if one risk is consid- impression can lead to more investment in the ered, but then increase vulnerability to other protected area and greater vulnerability to risks. An example is the use of smart grids and floods that exceed the level of the dikes (Burby consumer demand management to prevent et al. 2001; Burby, Nelson, and Sanchez 2006). BUILD INSTITUTIONS FOR RESILIENCE 155 An increase in protection can even lead to a net There is a consensus among experts that increase in average annual losses, because addi- governments have a key role to play in ensur- tional protection can lead to much larger ing the resilience of critical infrastructure and investments in at-risk areas (Hallegatte 2017). that they should adopt a whole-of-government One particular trade-off between short-term approach. This approach involves the sectoral and long-term risks is “maladaptation”—that ministries and agencies overseeing infrastruc- is, measures that reduce the short-term level of ture services delivery and regulation in multi- risk but increase the longer-term vulnerability ple critical sectors, as well as those responsible to climate change. Examples include the for resilience to hazards and threats. It also increased use of groundwater pumping and involves local authorities, especially municipal- irrigation to manage droughts, which can lead ities that, in many countries, are responsible to long-term vulnerabilities. Uncharted Waters: for supplying drinking water and managing The New Economics of Water Scarcity and Variabil­ urban transit and transportation. ity (Damania et al. 2017) finds that in arid areas The most common solution for improving and in low-income countries, the presence of the coordination of risk management is to irrigation infrastructure can exacerbate the place an existing multiministry body (or, if nec- impact of shocks on agricultural yields because essary, a new body) in charge of the exchange it encourages farmers to adopt more water- of information, coordination, and perhaps the intensive crops that are even more vulnerable implementation of risk management measures. to droughts. Preventing maladaptation requires Many countries have agencies in charge of systematic exploration of the long-term impli- coordinating disaster risk management or cations of measures to reduce short-term risks. national security issues, and these agencies can also tackle issues related to infrastructure resil- RECOMMENDATION 2: BUILD ience. For example, in France, the General Sec- INSTITUTIONS FOR RESILIENCE retariat for Defense and National Security What are the solutions for coping with these under the prime minister coordinates resilience challenges of political economy and coordina- policy for critical infrastructure across eight tion? They include creating institutions to line ministries, using a multihazard approach. manage infrastructure resilience and defining The body in charge of critical infrastructure the vision that can help actors to coordinate can be given special powers to collect informa- their actions. tion, perform assessments, and impose certain actions and ban others. For example, the recent Action 2.1: Implement a whole-of- Australian Security of Critical Infrastructure government approach to infrastructure Act, which is aimed at protecting the country resilience, building on existing from sabotage and espionage, mandates the regulatory systems creation of a registry of critical infrastructure Different countries take different approaches to assets. It also gives the minister of the Depart- infrastructure resilience, but common princi- ment of Home Affairs the right to request infor- ples have been widely applied. These princi- mation about these assets to determine whether ples—discussed in detail in Good Governance for any risk to national security is associated with Critical Infrastructure Resilience, which was issued an asset. The minister can impose or prohibit by the Organisation for Economic Co-opera- certain actions if there is “a risk of an act or tion and Development (OECD 2019)—are con- omission that would be prejudicial to security.” sistent with typical recommendations on the A body in charge of infrastructure resilience governance of risks.1 needs to be appropriately staffed and funded. 156 LIFELINES However, it cannot, and should not, replace tion” (OECD 2019, 47). This requires assessing the regulatory bodies in charge of sectors, the vulnerability of critical infrastructure assets which should be a priority in low-capacity and systems and the consequences of possible countries (see chapter 9). Various decisions or disruptions, so the government can prioritize regulations need to be coordinated across sec- actions. Chapter 7 illustrates the use of critical- tors, but their design and practical implemen- ity analyses to identify the most important tation are better conducted by each sector reg- assets in an infrastructure network or to iden- ulator to ensure consistency with other tify additional infrastructure that would do the regulations and to prevent conflicts. In prac- most to build resilience of the network. tice, implementation will vary, depending on The United Kingdom and many other coun- whether the regulation of an infrastructure tries conduct regular national risk assessments sector is carried out directly by the govern- (figure 10.1) to assess the main risks they face, ment, by an independent agency, or through a regardless of the type and origin of risk (natu- contract (Eberhard 2007). ral, technological, terrorist, or other). The assessments are based on similar approaches: Action 2.2: Identify critical identifying risks, generating scenarios, assess- infrastructure and define acceptable ing the probability or plausibility and impacts and intolerable risk levels of the risks, and enabling the construction of a The task of building infrastructure resilience at national risk matrix. an acceptable cost begins with identifying the A risk matrix summarizes the main risks critical infrastructure—that is, the “systems, and organizes them according to their likeli- assets, facilities, and networks that provide hood and severity of impact. Regular national essential services for the functioning of the risk assessments can also be used to assess the economy and the well-being of the popula- quality of risk management of various agencies FIGURE 10.1 U.K. national risk matrix Natural hazards Diseases More 5 Coastal flooding Pandemic influenza River flooding Emerging infectious diseases Surface water flooding Animal diseases 4 Storms and gales Major accidents Impact severity Cold and snow Widespread electricity failures 3 Heat waves Transport accidents Droughts Industrial/urban accidents Solar flares 2 System failures Volcanic eruptions Poor air quality Societal risks Less 1 Employee strikes Earthquakes Wildfires Public disorder 1 2 3 4 5 Less More Likelihood of occuring in the next five years Source: U.K. Cabinet Office 2017. BUILD INSTITUTIONS FOR RESILIENCE 157 and organizations (including local authorities In contrast, a risk is considered intolerable if and their land use plans) through risk audits its likelihood and potential impact are too high and benchmarking. The results can be reported and the cost of prevention is affordable. For annually to the country’s legislative body, rais- example, major transport infrastructure—such ing policy makers’ awareness of critical infra- as a large bridge or tunnel—cannot be suscep- structure issues. tible to failure and collapse from storms or Next, the government needs to define a moderate earthquakes because the human and shared vision of the level of risk that is consid- economic impacts of such events would be ered acceptable, based on its potential impact unacceptable. If a major highway is forced to and likelihood, and to identify the resources close several times a year because of local that are available for disaster risk management flooding, the disruptions would have a major and infrastructure financing. Indeed, with the economic impact. These unacceptable vulnera- significant interdependencies across systems, bilities can also be reframed as a minimum applying a consistent level of resilience to vari- expected level of service. ous components of the infrastructure system is The definitions of acceptable and intolerable more cost-effective. It would not make sense risks for infrastructure systems need to adhere for a government to invest major resources in to four important principles. making the power system highly resilient if the First, the approach to defining these risks water supply or transport system cannot cope should be open and participatory, characterized with frequent hazards. Instead, for systems by close cooperation between scientists, infra- that interact, it is more efficient to target a sim- structure service providers, infrastructure ser- ilar level of resilience. Using a target level of vice users, and policy makers. Scientists and resilience is a more practical way of allocating other experts alone cannot define what risks investments efficiently across sectors than try- are acceptable; they lack the legitimacy to do ing to equalize the rates of return of various so. Nor can policy makers; they usually lack investments. the technical expertise. Relying on a participa- Acceptable risks are usually those with conse- tory approach ensures that the appropriate quences that can be managed or those that data and concerns are given due consideration cannot be prevented at an affordable cost. and helps to raise awareness of, and form a When these risks materialize, it is expected consensus on, the vision to anchor the decision that an infrastructure system will be damaged making of various independent actors. and its service disrupted, with consequences Second, risk taking sometimes yields bene- for the rest of the economy. For instance, sig- fits that justify the risk taken. Risk manage- nificant residual risk often needs to be accepted ment should not become an obstacle to devel- for tertiary roads, especially in low-income opment (World Bank 2013). In some rural countries, since their length would make it areas, proximity to water offers cheaper trans- unaffordable to strengthen them beyond a cer- port and regular floods increase agricultural tain point. The fact that these risks are accept- productivity (Loayza et al. 2012). People may able does not mean that they can be ignored. settle in risky coastal areas to benefit from job Governments need to be prepared for the opportunities in industries driven by exports. disasters and disruptions that cannot be Better jobs and services may attract people to avoided, including by having an emergency cities, even if the cities are more exposed than response and possibly financial support in place rural areas to some threats. Innovation gener- for infrastructure operators, households, and ates growth, but almost always involves risk. businesses. Risk taking is one of the drivers of economic 158 LIFELINES growth and should not be suppressed indis- and impacts of disruptions—especially for vul- criminately (Hallegatte 2017). nerable and marginalized population groups. Third, definitions of these risks should con- First and foremost, the potential loss of life sider the local context and the cost of resil- is important to include in any risk analysis. Of ience—compared with the resources available. course, infrastructure disruptions can be life Countries at different income levels are unlikely threatening for some people, such as for those to be able to afford the same level of resilience, who depend on electricity-powered medical and not all countries can aspire to the same devices or pregnant women who need urgent level of resilience over the short term. Also, access to medically assisted delivery (see chap- countries with different exposures may aim to ter 3). In the Netherlands, the Dutch Water Act achieve different resilience levels; a small island of 2015, which sets out standards for flood regularly affected by hurricanes is likely to ded- defenses (along the coast, lakes, and rivers), icate a larger share of its resources to resilience explicitly considers loss of life. The starting than the average country. Highly exposed, point of the new flood defense standard is that wealthy countries like Japan and the Nether- every citizen should be able to rely on the same lands spend much larger amounts on flood pro- (minimum or basic) level of protection. This tection and resilience-enhancing initiatives and level of protection is expressed in the “local regulations than other countries at the same individual risk”—that is, the chance that an income level. individual permanently present at a specific Fourth, the definition of acceptable risk location will die as a result of flooding. The levels for infrastructure assets should look far legal basis for considering these risks is in Article into the future. With economic growth and 21 of the Dutch Constitution, which imposes a technological change, resources, preferences, duty on government “to ensure the habitability and standards will change, possibly leading to of the land and the protection and improve- stronger demand for resilience and a lower risk ment of the environment.” of failure. These changes will be a challenge for Economic losses also hide the impact of the design of long-lived infrastructure. An disasters on poor people (Hallegatte et al. acceptable level of risk at the time an infra- 2017). Because the wealthy have more assets structure asset is designed and built may prove and income to lose, their interests dominate in unacceptable 30 years later, when the asset will assessments of economic losses. If informed be only at its half-life. In the design of long- only by potential economic losses, decisions lived systems, the potential for regret is a critical about the resilience of infrastructure or invest- metric and may justify fortifying an asset ments to reduce natural risks will tend to favor beyond the point the current situation would the richest areas of a country or a city. Although suggest.2 the poor often have very little to lose, they lack the resources and tools to smooth income Action 2.3: Ensure equitable access to shocks while maintaining consumption and resilient infrastructure coping with infrastructure disruptions. Thus, Decisions on resilience cannot be driven by after disasters, they are more likely than the economic considerations alone. Indeed, eco- wealthy to forgo the consumption of food, nomic losses are only one part of the many health services, and education. impacts of disasters, including infrastruc- To ensure that resilience is distributed fairly ture-mediated ones. Thus, the strengthening of across the population, one option is to measure infrastructure resilience should be guided by a the impacts of disasters and infrastructure dis- more complete assessment of the potential risks ruptions using a metric that accounts for the BUILD INSTITUTIONS FOR RESILIENCE 159 MAP 10.1 Different measures of natural risks in the Philippines highlight different priorities for interventions a. Annual asset risk Numberof b.Number b. peoplefalling ofpeople falling c. Socioeconomic resilience d. Annual well-being risk a. Annual asset risk poverty ininto every poverty year every year c. Annual well-being d. Socioeconomic resilience 3 6 9 12 15 0 0.2 0.4 0.6 0.8 1 20 30 40 50 60 5 10 15 20 25 30 Philippine pesos (billions) Share of population (%) Percent Philippine pesos (billions) Source: Walsh and Hallegatte 2019. socioeconomic status of the affected popula- of national risk and identification of critical tions (Hallegatte et al. 2017). A recent analysis infrastructure need to account for multiple pol- in the Philippines employed a multimetric icy objectives and, therefore, use a set of met- assessment of disaster risks at the regional level rics that goes beyond asset losses. using (1) traditional asset losses; (2) poverty- Affordability issues are a direct threat to related measures such as the poverty head- universal access to infrastructure services and count; (3) well-being losses for a balanced esti- the achievement of the 2030 Sustainable mate of the impact on poor and rich house- Development Goals—and the higher costs aris- holds; and (4) socioeconomic resilience, an ing from resilience-enhancing investments indicator that measures the ability of the popu- may negatively affect these important policy lation to cope with and recover from asset objectives. In some places, the retail price of losses (Walsh and Hallegatte 2019). electricity is already high, leaving some house- Priority interventions—in both spatial terms holds connected to the grid but unable to (where to act?) and sectoral terms (how to afford electricity. In other places, infrastructure act?)—are highly dependent on which metric services are heavily subsidized to ensure afford- for disaster severity is used (map 10.1). In the ability, but these subsidies can have unin- Philippines, the most important interventions tended consequences for the ability of service will take place in the Manila area if asset losses providers to invest in new, and maintain exist- are the main measure of disaster impacts. ing, infrastructure assets. Any large increase in Other regions become priorities if the policy prices that would be triggered by regulations or objectives are expressed in terms of poverty financial incentives for more resilient infra- incidence and well-being losses. Assessments structure systems could magnify this problem 160 LIFELINES BOX 10.2 The structure of tariffs and targeted subsidies can help to ensure that the resilience of infrastructure services is not improved at the expense of access: The case of public transit Public transit is a sector in which price subsidies or differential discounts, depending on the char- are common, justified by the positive externality acteristics of individual trips such as time of day of increasing access to jobs and services and the or type of route. benefits of public transit for congestion and air In February 2014, Bogotá rolled out a “pro- quality. In the United States, the median revenue poor” transport subsidy program. The program from fares for major transit systems is approxi- builds on the progressive adoption of smart mately 35 percent of total expenses, mean- cards by Bogotá’s public transit systems and on ing that the remaining 65 percent of operating national experience with other poverty-targeting expenses must be covered elsewhere. initiatives (such as conditional cash transfer pro- Today, advanced targeted subsidy schemes grams) that use the country’s poverty-targeting can rely on modern electronic fare systems and system and database (Sistema Nacional de Selec- sophisticated methodologies for defining and ción de Beneficiarios, or SISBEN). Beneficiaries targeting beneficiary populations. In particular, defined as “SISBEN 1 and 2 users” can receive a the use of smart cards has allowed governments public transit subsidy, effectively amounting to a to structure subsidies that target demand rather 40 percent discounted fare capped at 21 trips a than supply. Smart cards can be personalized, month. Well-targeted subsidies make it easier to and subsidies delivered via smartcard can take fund urban transit with more cost recovery, with- on different structures. Examples are a flat rate out threatening access for the poorest. Source: Mehndiratta, Rodriguez, and Ochoa 2014. and potentially attract criticism and opposition. Burby, R. J., A. C. Nelson, D. Parker, and J. Hand- However, as discussed in part II of this report, mer. 2001. “Urban Containment Policy and the increase in infrastructure service costs Exposure to Natural Hazards: Is There a Connec- tion?” Journal of Environmental Planning and Man­ to achieve higher resilience is expected to be agement 44 (4): 475–90. https://doi.org/10.1080 limited and thus will not radically change the /09640560120060911. existing trade-off between affordability and Burby, R. J., A. C. Nelson, and T. W. Sanchez. 2006. cost recovery. As a result, the usual recommen- “The Problems of Containment and the Promise dations for managing this trade-off would of Planning.” In Rebuilding Urban Places after Dis­ aster: Lessons from Hurricane Katrina, 47–65. Phila- apply, as in the case of public transit (box 10.2). delphia: University of Pennsylvania Press. Cleveland, F., and A. Lee. 2013. “Cyber Security for NOTES DER Systems.” Electric Power Research Institute, 1. See, for example, Renn (2008, 2017); Wiener National Electric Sector Cybersecurity Organiza- and Rogers (2002); and World Bank (2013). tion Resource, Palo Alto, CA. 2. The use of regret as a decision-making crite- Damania, R., S. Desbureaux, M. Hyland, A. Islam, rion can help in planning for the future (see S. Moore, A.-S. Rodella, J. Russ, and E. Zaveri. chapter 12). 2017. Uncharted Waters: The New Economics of Water Scarcity and Variability. Washington, DC: World REFERENCES Bank. https://doi.org/10.1596/978-1-4648 Aloul, F., A. R. Al-Ali, R. Al-Dalky, M. Al-Mardini, and -1179-1. W. El-Hajj. 2012. “Smart Grid Security: Threats, Eberhard, A. 2007. “Infrastructure Regulation in Vulnerabilities, and Solutions.” International Journal Developing Countries: An Exploration of Hybrid of Smart Grid and Clean Energy 1 (1): 1–6. and Transitional Models.” Working Paper 4, BUILD INSTITUTIONS FOR RESILIENCE 161 World Bank, Washington, DC. http://documents Renn, O. 2008. “White Paper on Risk Governance: .worldbank.org/curated/en/5288014681763 Toward an Integrative Framework.” In Global Risk 32574/Infrastructure-regulation-in-developing Governance, edited by O. Renn and K. D. Walker, -countries-an-exploration-of-hybrid-and 3–73. Stuttgart: Springer. http://link.springer -transitional-models. .com/chapter/10.1007/978-1-4020-6799-0_1. Hallegatte, S. 2017. “A Normative Exploration of the ———. 2017. Risk Governance: Coping with Uncertainty Link between Development, Economic Growth, in a Complex World. London: Routledge. and Natural Risk.” Economics of Disasters and Cli­ U.K. Cabinet Office. 2017. “U.K. National Risk Reg- mate Change 1 (1): 5–31. https://doi.org/10.1007 ister of Civil Emergencies.” U.K. Cabinet Office, /s41885-017-0006-1. London. Hallegatte, S., A. Vogt-Schilb, M. Bangalore, and Wagner, D. 2016. “Threat of Cyber-Attacks on Criti- J. Rozenberg. 2017. Unbreakable: Building the cal Infrastructure.” International Risk Manage- Resilience of the Poor in the Face of Natural Disasters. ment Institute, Dallas. Washington, DC: World Bank. https://doi.org Walsh, B., and S. Hallegatte. 2019. “Measuring Nat- /10.1596/978-1-4648-1003-9. ural Risks in the Philippines.” Policy Research Kunreuther, H., and G. Heal. 2003. “Interdependent Working Paper 8723, World Bank, Washington, Security.” Journal of Risk and Uncertainty 26 (2–3): DC. https://doi.org/10.1596/1813-9450-8723. 231–49. https://doi.org/10.1023/A:1024119 Wiener, J. B., and M. D. Rogers. 2002. “Comparing 208153. Precaution in the United States and Europe.” Loayza, N. V., E. Olaberría, J. Rigolini, and L. Christi- Journal of Risk Research 5 (4): 317–49. https://doi aensen. 2012. “Natural Disasters and Growth: .org/10.1080/13669870210153684. Going beyond the Averages.” World Development World Bank. 2013. World Development Report 2014: 40 (7): 1317–36. https://doi.org/10.1016/j Risk and Opportunity—Managing Risk for Develop­ .worlddev.2012.03.002. ment. Washington, DC: World Bank. https://doi Mehndiratta, S. R., C. Rodriguez, and C. Ochoa. .org/10.1596/978-0-8213-9903-3. 2014. “Targeted Subsidies in Public Transport: Zetter, K. 2016. “Inside the Cunning, Unprecedented Combining Affordability with Financial Sustain- Hack of Ukraine’s Power Grid.” Wired, March 3. ability.” World Bank, Washington, DC. Zou, B., P. Choobchian, and J. Rozenberg. 2019. OECD (Organisation for Economic Co-operation and “Autonomous Mobility Systems and Cyber Resil- Development). 2019. Good Governance for Critical ience: Threats, Vulnerabilities, and Possible Resil- Infrastructure Resilience. OECD Review of Risk ience-Enhancing Strategies.” Background paper Management Practices. Paris: OECD. for this report, World Bank, Washington, DC. Create Regulations and Incentives for Resilience 11 OBSTACLE RECOMMENDATION ACTIONS Lack of incentives to Create regulations and 3.1: Consider resilience objectives in master •  increase resilience incentives for resilience plans, standards, and regulations, and adjust them regularly to account for climate change 3.2: Create financial incentives for service •  providers to promote resilient infrastructure services 3.3: Ensure that infrastructure regulations are •  consistent with risk-informed land use plans and guide development toward safer areas THE OBSTACLE: INFRASTRUCTURE ruptions and the incentives that service provid- PROVIDERS OFTEN LACK ers face (table 11.1): THE INCENTIVES TO AVOID DISRUPTIONS AND TO STRENGTHEN • For a road agency that operates with a fixed THE RESILIENCE OF USERS budget, there is an incentive to build roads Ideally, the providers of infrastructure services in a way that minimizes the maintenance and the entities that design, build, operate, and repair costs, but no incentive to account and maintain infrastructure assets would bear for the full cost of transport disruptions, the full cost of infrastructure disruptions. This such as the impact on businesses and supply would include covering the cost of the repairs chains. and additional maintenance needed after nat- • For the operator of a toll road public-private ural shocks, such as floods and storms, as well partnership (PPP) or the owner of a private as the full cost of disruptions for the users power generation plant, there are typically of infrastructure services. Service providers stronger incentives to incorporate resilience would then have the incentives needed to (although the ultimate decision will depend minimize disruptions, including from natural on the exact contractual structure). They hazards. want not only to minimize repair and main- But the reality is different. Take the follow- tenance costs but also to avoid revenue losses ing cases, which highlight the existence of a when the asset cannot be used. Even so, the gap between the full cost of infrastructure dis- cost borne is less than the real impact. For 163 164 LIFELINES TABLE 11.1 Examples of the presence (and absence) of incentives for resilience Resilience of Resilience of infrastructure assets infrastructure services Resilience of infrastructure users Costs that should be Postdisaster repair Loss of revenue during Well-being or revenue Impact of infrastructure internalized in asset design, costs (and increased disruptions (when user losses for infrastructure assets on people’s construction, maintenance, maintenance costs) fees are used to fund users (or other and firms’ exposure and operation due to natural hazards the asset) infrastructure systems) to risk Rural road built and maintained by a public road agency  X X X Toll road built and operated by a private actor through a public-   X X private partnership Solar power plan owned and operated by a private firm   X X Land that could be reforested to reduce landslide risks for nearby roads and water utilities X X X X downstream Note:  = presence; X = absence. example, the social cost associated with a Another problem is that budgetary and con- 1-kilowatt-hour power outage is at least 80 tractual arrangements can further reduce the percent higher than the loss in revenue from incentive to minimize life-cycle costs and an interruption (see chapter 3). disaster-related losses. For example, in a public road agency, the investment and maintenance Making matters worse, there is no incentive budgets may be separate, making it difficult for for infrastructure owners or operators to lower maintenance and repair costs to com- account for how the infrastructure will affect pensate for higher investment costs. The usual the risk exposure of other people and firms or budgeting processes are also an obstacle, their ability to manage infrastructure disrup- because the benefits of lower maintenance and tions (table 11.1). For example, a road agency repair costs can take decades to materialize, may build a road on a floodplain, ensuring that well beyond the time horizon of even plurian- the road can cope with floods, but not consider nual budgeting. Another example is procure- that the road will attract new settlements, busi- ment and contracting models, in which the pri- nesses, and investments. Although the road vate contractor building or operating an asset itself may be resilient, it can still reduce the does not own it and thus would not incur any resilience of the community and those who repair costs resulting from a disaster. Such a sit- build their livelihoods and economic activity uation could arise in standard procurement or around the new infrastructure. Without risk- in PPPs following the “build, transfer, operate” informed land use planning, infrastructure pro- model. Even when the private contractor owns viders are unlikely to recognize these risks. And the asset (such as in “build, operate, transfer” even if they do, they are unlikely to bear the models), the transfer of the asset to the govern- long-term costs of risky spatial development. ment usually takes place long before the end of Further, while nature-based solutions and the asset’s lifetime. And even when longer green infrastructure can efficiently reduce the concessions are possible (for example, the cost of infrastructure services and increase resil- Tours-Bordeaux high-speed rail line in France ience, there is often little incentive to protect or was built under a 50-year concession contract), restore ecosystems. the high discount rate observed in the private CREATE REGULATIONS AND INCENTIVES FOR RESILIENCE 165 sector means that long-term natural hazard stake. Governments will always support and and climate change risks will have a minimal facilitate the recovery of these infrastructure impact on decision making. systems, and this fact needs to be acknowl- Further, the expectation that the govern- edged and taken into account in the design of ment will provide ad hoc support if a disaster regulations and incentives. occurs can diminish the incentives to act. In the aftermath of a disaster, governments usu- RECOMMENDATION 3: INCLUDE ally provide people, firms, and infrastructure RESILIENCE IN REGULATIONS AND owners and operators with support. But the INCENTIVES mere possibility of public aid after disasters can What are the solutions for coping with the create moral hazard, discouraging risk manage- challenges that arise from a lack of incentives ment and the purchase of insurance—and this to avoid disruptions and strengthen the resil- moral hazard is simply unavoidable. Providing ience of users? The answer lies in governments support during and after crises is one of the designing a consistent set of regulations and main missions of governments, and it is not financial incentives to align the interests of realistic to expect them to withhold support infrastructure service providers with the public from an area affected by a disaster just to avoid interest, as illustrated in figure 11.1. moral hazard, especially when basic services How is this done? First, for each hazard and (such as electricity, water, and transport) are at infrastructure system, governments or regula- FIGURE 11.1 Creating the right resilience incentives for infrastructure service providers requires a consistent set of regulations and financial incentives 1 2 3 4 Government or Government or Government or Developer designs regulator defines regulator defines an regulator adds project above and enforces an “acceptable” level of incentives to align the minimum "intolerable" level of risk that can be the interest of standard risk through a tolerated ("force service providers minimum standard in majeure" event) with the public Intensity construction codes interest, with of hazards or procurement penalties and rewards based on social cost Major, rare events Acceptable risks: For rare events, infrastructure assets Government are expected to experience damage or disruptions that bears the risk need to be managed through contingent planning Force majeure Provider bears at least part of the risk (insurance may be required) Project-specific designs Minimum standard Infrastructure services Intolerable risks: Infrastructure should not be disrupted should resist frequent hazards below this level. Provider bears the risk Small, frequent hazards 166 LIFELINES tors need to define a minimum standard of ability, on the other. Very specific standards resistance—that is, a hazard intensity below applied to infrastructure—such as the amount which infrastructure assets should not suffer of rainfall (in millimeters) that road culverts any damage or disruption. For example, all should be able withstand are simple to apply roads should be able to cope with a 20-year and enforce, but they are not context specific return period rainfall event. Second, they need and can be locally inappropriate. To ensure to define the level of the force majeure or that infrastructure assets can cope with local acceptable risk—that is, the level at which hazards, standards can be developed for each infrastructure failures have to be tolerated. region and for broad categories of assets—for Beyond this level, the risk from a natural haz- example, primary versus secondary versus ter- ard is usually supported by the public sector. tiary roads. All primary roads in a mountain- Below this level, at least part of the risk is usu- ous area could be required to be able to cope ally supported by the owner or operator of the with a certain amount of rainfall. Even better, infrastructure asset. Third, they need to create the standard could be asset and location spe- the right incentives to align the interest of the cific. For example, roads could be required to infrastructure asset owner or operator with the withstand events of a certain return period, public interest. This can be achieved by penal- with the precise level based on the criticality of izing an infrastructure operator for disruptions, the road. for example, at an amount calibrated on the With climate change and other long-term social cost of these disruptions. Based on these environmental trends, standards and codes regulations, incentives, and risk allocations, need to be revised regularly (box 11.1). project developers and asset owners can deter- According to Vallejo and Mullan (2017), mine their strategy, the desired resilience level approximately one-third of Organisation for of their assets, and an appropriate design. Economic Co-operation and Development (OECD) countries are revising at least one Action 3.1: Consider resilience mandatory national infrastructure standard to objectives in master plans, standards, account for climate change adaptation, but and regulations, and adjust them similar processes are lacking in low- and middle- regularly to account for climate change income countries. For instance, Sweden Agencies responsible for infrastructure services updated its road drainage standard in 2008, usually undertake regular master-planning introducing a climate safety factor to cope with exercises, typically on a five-year cycle, to for- the anticipated increase in future rainfall due mulate their investment program and financial to climate change. Similarly, the European needs. These master-planning exercises should Commission mandated the Centre Européen explicitly consider the resilience of the plans de Normalisation to include climate change in and options available to reduce the vulnerabil- the European civil engineering technical stan- ity of their systems to various natural hazards dards (the Eurocodes), especially for transport and climate change. and energy infrastructure (European Commis- Standards and regulations need to account sion 2014). Several national standards organi- for climate conditions, geophysical hazards, zations have produced risk management and climate change and other environmental guidelines that include climate change and and socioeconomic trends. Resilience-related resilience considerations for infrastructure standards can be expressed in many ways, with (British Standards Institution 2011; Council of trade-offs between simplicity and enforceabil- Standards Australia 2013; U.S. National Insti- ity, on the one hand, and specificity and adapt- tute of Standards and Technology 2015). And CREATE REGULATIONS AND INCENTIVES FOR RESILIENCE 167 BOX 11.1 With climate change, when and where do standards need to be revised? In the United States, stormwater infrastructure each climate region to assess the relative strin- is designed using government documents on gency of each state’s requirements. Using these precipitation frequency, informed by states’ index values, the observed change in precipita- department of transportation (DOT) guidelines tion frequency estimates, and each state’s design that balance risks and costs. However, both manual publication date, this research identifies the government precipitation documents and the states that need to prioritize revision of their states’ DOT guidelines are updated infrequently, stormwater standards to maintain the originally which increases the risks in areas where patterns intended design performance over time (see map of precipitation have changed over time. Lopez- B11.1.1 for the 25-year return period event). When Cantu and Samaras (2018) review DOT design considering all return periods, eight states are manuals for the 48 contiguous U.S. states and found to require an immediate revision of their the District of Columbia and find wide variations stormwater standards. In addition, these states in the design of return period standards rec- should assess whether the existing infrastructure ommended for similar roadways and types of requires additional adaptive capacity to manage infrastructure. observed precipitation increases. Patterns of precipitation and intensities used Looking to 2050, under a scenario of climate in various design manuals have been changing change, the priority for such a revision becomes over time, indicating that stormwater infrastruc- more urgent for all states. Although local assess- ture installed prior to the latest update of precip- ments comparing the infrastructure cost of itation frequency documents could be underde- increasing the stringency of standards with the signed for present and future climate conditions. expected cost of future damages remain nec- Comparing states’ DOT design storm values for essary, revising stormwater standards to incor- each roadway and type of infrastructure, Lopez- porate observed precipitation increases is a Cantu and Samaras (2018) develop an index for no-regret option. MAP B11.1.1 In some U.S. states, revising stormwater infrastructure standards is urgent Priority of revising standards  1 (low) 2 3  4 (high)  No data Source: Lopez-Cantu and Samaras 2018. Note: Map shows priority (1 lowest, 4 highest) assigned to each state to revise storm- water infrastructure standards, according to the observed changes in 25-year return period. As of January 2018, states in gray remain uncovered by the National Oceanic and Atmospheric Organization’s Atlas 14 of precipitation frequency and thus are not included in the analysis. 168 LIFELINES in 2015, the International Standards Organisa- and develop a BCP to minimize the conse- tion (ISO) created the Adaptation Task Force to quences of shocks. The 2012 Japan Revitaliza- develop standards for vulnerability assessment, tion Strategy sets BCP establishment targets for adaptation planning, and adaptation monitor- 100 percent of large firms and 50 percent of ing and evaluation (ISO 2015). small and medium enterprises by 2020. Resilience-related regulations can also be Households can do much to be better pre- based on outcome indicators, using observed pared to cope with infrastructure disruptions. performance. For example, electricity utilities Basic disaster supply kits are widely available, can be required to limit power outages to and most disaster management agencies and below a certain number of hours a year. In organizations, such as the Red Cross, provide France, the electricity distribution company is guidance. 1 Traditional recommendations committed to limiting power outages to below include having 72-hour reserves of emergency an average of one long outage (longer than supplies, such as water, canned food (with a three minutes) and five short outages a year. manual can opener), extra batteries, candles, The main advantage of outcome-based regula- pet supplies, and copies of important personal tions is that they outsource the risk assessment documents (like passports, land titles, bank to infrastructure operators and should auto- account information, and insurance contracts). matically adjust for climate change. However, Vulnerable people and groups (like people with such observed outcomes cannot be applied to disabilities or chronic diseases, the elderly, and rare shocks—such as a 100-year return period young children) have specific needs that hurricane—because such an event cannot should also be accounted for (including pre- be regularly observed. Input-based or pro- scription medicines and baby formula and dia- cess-based approaches founded on construc- pers). Well-equipped households provide more tion codes are the only ones that can be applied room for utilities, agencies, and governments to exceptional events. to restore services, while avoiding the worst It is sometimes easier to enable the users of impact on people’s health and well-being. infrastructure services to manage disruptions than to prevent all disruptions (see chapter 8). Action 3.2: Create financial incentives As a result, regulations can also apply to specific for service providers to promote users of infrastructure services, not just to sup- resilient infrastructure services pliers. For example, hospitals can be required to A common limit of codes and regulations is that invest in generators and water tanks so that regulators and governments may not have all they can cope with power and water outages the information they need on the costs and for a certain period of time, mitigating the con- benefits of all options for building resilience. sequences of infrastructure disruptions that This limit can be overcome with economic and would be too expensive to prevent. financial incentives. Two options are particu- Firms also can adopt business continuity larly common: (1) rewards and penalties for plans (BCPs) to reduce the cost of infrastruc- infrastructure service providers, pushing them ture disruptions. For example, Japan’s policy to go beyond the code and capture further oppor- and institutional framework for industry resil- tunities to build resilience, and (2) payment for ience, the Basic Disaster Response Plan, ecosystem service schemes to promote nature- requires companies to recognize the role that based solutions. they are expected to play when disaster strikes, Rewards and penalties can motivate service to understand their own risk from a natural providers to implement cost-effective solutions disaster, and to implement risk management to improve resilience beyond the mandatory CREATE REGULATIONS AND INCENTIVES FOR RESILIENCE 169 (Pardina and Schiro 2018). The Australian In the many countries where the power sector Energy Regulator established the Service Tar- is not financially viable (Kojima and Trimble get Performance Incentive Scheme (STPIS) to 2016), for instance, financial penalties may incentivize electricity providers to improve the exacerbate existing challenges for utilities and quality of their services, including their reliabil- may not support better design or maintenance ity and resilience (Pardina and Schiro 2018). of infrastructure systems. The goal of STPIS is to prevent providers from Another instrument is payment for ecosys- achieving cost reductions at the expense of ser- tem services (PES) schemes, which can be used vice quality. Penalties and rewards distributed to create an incentive to promote nature-based by STPIS are calibrated according to how will- solutions to increasing resilience. These schemes ing consumers are to pay for improved service. entail a user fee that those who benefit from This arrangement aligns distributors’ incentives ecosystem services pay for protection or resto- for efficient price and nonprice outcomes with ration of the ecosystem (Browder et al. 2019). the long-term interests of consumers (Austra- And the fee can be applied to nature-based lian Energy Regulator 2014). solutions that reduce the cost or increase the Similarly, in Finland, the 2013 revision of resilience of infrastructure services, with pay- the Electricity Market Act sets compulsory ment originating from dedicated service fees, resilience targets for weather hazards with government revenues, or specialized funds. which operators must comply by the end of Infrastructure service providers can, with 2028 (OECD 2019). It specifies that the longest the approval of regulators, create a dis­ tinct fee acceptable interruption time is six hours in to support nature-based solutions. Some U.S. urban areas and 36 hours in rural areas. water utilities have “watershed protection fees” Enforcement is ensured by economic incen- that are reinvested in water­ shed protection tives that also encourage service providers to measures. In Brazil, water users pay a federally reach higher than minimum levels of security mandated fee to the local water company that of supply. The compensation paid by electricity local watershed committees use for watershed distribution operators to their customers in the maintenance and reforestation. event of a long outage reaches up to 200 per- If a specific fee is not possible, government cent of the yearly average electricity fee—up to revenues (or the reallocation of other subsidies) a maximum of €2,000—when the disruption can be earmarked to fund nature-based solu- exceeds 12 days. The new scheme was first tions. In the 1990s, the power supply of Costa activated during the January 2018 winter Rica was threatened by unsustainable farming storm that left 40,000 people in northern Fin- practices that accelerated the siltation of hydro- land without electricity, some of them for up to power reservoirs. Using revenues from fuel and a week. As a result, 10,000 customers received water taxes and grants and loans from multilat- compensation totaling €5 million. eral donors, the government created a PES pro- But financial incentives are challenging to gram that gives landowners incentives to implement (see box 11.2 regarding PPPs). First, restore and conserve forestland (Blackman and calibrating the value of the reward or penalty Woodward 2010). As a result, siltation is being can be difficult due to a lack of data, even reduced, helping to preserve the country’s elec- though a financial incentive for resilience does trical power generation infrastructure. not need to be precisely calibrated to improve In addition, nature-based solutions can be the situation. Second, such instruments will be encouraged by removing some of the obstacles effective only if the infrastructure operators to their implementation. One obstacle is the have the capacity to respond (see chapter 12). fact that the mandate of most infrastructure 170 LIFELINES BOX 11.2 Public-private partnerships and their force majeure clauses Governments should develop a legal framework majeure for each event category and to define and institutional structure to ensure that disaster force majeure as applicable only in extreme resilience is incorporated into PPP projects. Many cases. Ideally, a third party would decide after governments have a disaster risk framework and an event whether the return period or intensity a PPP framework, but the two frameworks rarely of the event was sufficient to trigger the force interact. Even in Japan, where PPPs are well majeure clause. The contract can then determine developed and natural hazards are well managed, the allocation of risk in terms of both missed rev- guidelines for including resilience in PPPs exist, enues and restoration costs, ensuring that the but they are not mandatory. private operator always bears a significant share The incentives for operators to incorporate of the cost. Mandatory or voluntary insurance resilience in their assets depend on the type of could also ensure the sustainability of the infra- contract, with “build, operate, transfer” models structure services—protecting the private opera- creating a stronger incentive than “build, trans- tor against losses, while minimizing the cost for fer, operate” models. However, contracts can be the public sector and maintaining the incentive to weakened by excessively broad force majeure build more resilient assets and systems. clauses, which transfer the risks from the private However, the design of PPPs needs to account to the public sector. When they are too broad, for many context-specific factors, including the force majeure clauses reduce the incentives for maturity of the PPP market, the risk tolerance actors to build and operate an infrastructure of private sector players, and other risk factors asset in a way that accounts for low-probability such as vulnerability to commodity price shocks. risks. For example, many force majeure clauses These factors will determine how much risk can include “acts of God (such as fires, explosions, be transferred to private operators, creating earthquakes, droughts, tidal waves, and floods).” trade-offs for governments between incentivizing Force majeure clauses are essential for estab- resilience and mobilizing private sector finance. lishing PPPs at a reasonable cost, and they can When the private sector is unable to bear the be designed to minimize the negative impact on risks from natural hazards, it becomes even more incentives for resilience. One solution is for a con- important to use alternative tools, such as strong tract to include a quantified definition of the force construction codes and procurement rules. Source: World Bank 2019. service regulators and operators often includes protect ecosystems and the services they pro- hard infrastructure systems, but not the envi- vide are often reduced by subsidies in the ronment and ecosystems that support these water, agriculture, energy, or housing sectors. systems. For example, water utilities are usu- For example, some agricultural subsidies favor ally responsible for the systems of pipes, the extension of farming at the expense of for- pumps, and treatment stations needed to pro- ests and wetlands. And tax incentives for con- vide households and firms with high-quality, struction, intended to improve housing afford- reliable water, but they have no mandate to act ability or create economic activity and jobs, can on the upstream ecosystems that are so essen- similarly lead to urban sprawl at the expense of tial to the provision of quality water. A second the natural areas that play a key role in miti- obstacle is that the incentives to preserve and gating floods (Brueckner and Kim 2003). CREATE REGULATIONS AND INCENTIVES FOR RESILIENCE 171 Action 3.3: Ensure that infrastructure Because infrastructure investments make regulations are consistent with risk- land more attractive by improving its accessi- informed land use plans and guide bility or the amenities it includes, they also development toward safer areas lead to an increase in land values. This increase Infrastructure regulators or operators have lit- can be “captured” through tax instruments or tle incentive to account for the effects of their specific fees and used for infrastructure devel- actions on the resilience of users. To ensure opment. Land value capture finance is a process that new infrastructure does not increase expo- whereby part or all of the value created sure and vulnerability to natural hazards, through public interventions or investments infrastructure regulations should be aligned and accruing to private agents is recuperated with risk-informed land use and urbanization by the public sector and used to finance public plans. Infrastructure localization choices need goods (Huxley 2009). Such financing can fund to account for the public and private invest- resilience-enhancing infrastructure systems, ments that a new infrastructure asset will creating an incentive for infrastructure service attract and their implications for resilience. For providers to build the resilience of the example, a new road on a floodplain may be a community. bad idea if it attracts people to this flood-prone area who will not be able to build resilient NOTE housing, even if the road itself is designed to 1. See, for instance, the Red Cross emergency cope with all possible natural disasters. To pre- supply kit at https://www.redcross.org/get-help vent such outcomes, infrastructure risk assess- /how-to-prepare-for-emergencies/survival-kit -supplies.html. ments have to consider induced investments, not just the infrastructure asset itself. Even better, infrastructure localization REFERENCES Australian Energy Regulator. 2014. “Overview of the choices can be used to implement land use Better Regulation Reform Package.” Australian planning and promote low-risk spatial devel- Energy Regulator, Melbourne. https://www.aer opment. Indeed, infrastructure investments .gov.au. can actively guide the development of evolv- Blackman, A., and R. T. Woodward. 2010. “User ing spatial patterns, and thus they should be Financing in a National Payments for Environ­ mental Services Program: Costa Rican Hydro- embedded in land use and regional planning. power.” Ecological Economics 69 (8): 1626–38. Take the case of Nadi, Fiji, where land that is British Standards Institution. 2011. “Climate Change safe and well connected to jobs and services Adaptation—Adapting to Climate Risks Using can be given priority for future development ISO 9001, ISO 14001, BS 25999, and BS 31100.” (see chapter 8). People and developers can be British Standards Institution, London. Browder, G., S. Ozment, I. Rehberger Bescos, attracted to this priority land by means of early T. Gartner, and G.-M. Lange. 2019. Integrating investments in transport, water and sanitation, Green and Gray: Creating Next Generation Infrastruc- and electricity infrastructure. Simple commu- ture. Washington, DC: World Bank and World nication can then help to drive urbanization Resources Institute. toward safer areas. In Tunis, the wide dissemi- Brueckner, J. K., and H.-A. Kim. 2003. “Urban nation of simple plans indicating where the Sprawl and the Property Tax.” International Tax and Public Finance 10 (1): 5–23. https://doi.org government was planning infrastructure /10.1023/A:1022260512147. extensions helped to guide urban develop- Council of Standards Australia. 2013. “Australian ment (Lozano-Gracia and Garcia Lozano Standards: Climate Change Adaptation for Settle- 2018). ments and Infrastructure: A Risk Based 172 LIFELINES Approach.” AS 5334—2013, Council of Stan- Lozano-Gracia, N., and M. Garcia Lozano. 2018. dards Australia, Sydney. https://infostore “Haitian Cities: Actions for Today with an Eye on .saiglobal.com/store/details.aspx?ProductID= Tomorrow.” World Bank, Washington, DC. 1631218. OECD (Organisation for Economic Co-operation and European Commission. 2014. “Commission Imple- Development). 2019. Good Governance for Critical menting Decision of 28.5.2014 on Deciding to Infrastructure Resilience. OECD Review of Risk Make a Standardisation Request to the European Management Practices. Paris: OECD. Standardisation Organisations.” European Com- Pardina, M. R., and J. Schiro. 2018. “Taking Stock of mission, Brussels. www.etsi.org/images/files Economic Regulation of Power Utilities in the /ECMandates/M526.pdf. Developing World: A Literature Review.” Policy Huxley, J. 2009. “Value Capture Finance: Making Research Working Paper 8461, World Bank, Urban Development Pay Its Way.” Urban Land Washington, DC. https://doi.org/10.1596/1813 Institute, Washington, DC. -9450-8461. ISO (International Standards Organisation). 2015. U.S. National Institute of Standards and Technology. “Practical Tools for Addressing Climate Change.” 2015. Community Resilience Planning Guide for ISO, Geneva. www.iso.org/iso/climatechange_ Buildings and Infrastructure Systems. Vols. I and II. 2015.pdf. Washington, DC: U.S. Department of Commerce. Kojima, M., and C. Trimble. 2016. “Making Power www.nist.gov/el/resilience/guide.cfm. Affordable for Africa and Viable for Its Utilities.” Vallejo, L. and M. Mullan. 2017. “Climate-Resilient World Bank, Washington, DC. https://open Infrastructure: Getting the Policies Right.” OECD knowledge.worldbank.org/handle/10986/25091. Environment Working Paper 121, OECD Publish- Lopez-Cantu, T., and C. Samaras. 2018. “Temporal ing, Paris. http://dx.doi.org/10.1787/02f74d61-en. and Spatial Evaluation of Stormwater Engineer- World Bank. 2019. “Technical Brief on Resilient ing Standards Reveals Risks and Priorities across Infrastructure Public-Private Partnerships (PPPs)— the United States.” Environmental Research Letters Contracts and Procurement.” World Bank, Wash- 13 (7): 074006. https://doi.org/10.1088/1748-9326 ington, DC. /aac696. Improve Decision Making 12 OBSTACLE RECOMMENDATION ACTIONS Insufficient consideration Improve decision 4 .1: Invest in freely accessible natural hazard and •  of natural hazards and making climate change data climate change 4 .2: Make robust decisions and minimize the •  potential for regret and catastrophic failure 4 .3: Build the skills needed to use data and models •  and mobilize the know-how of the private sector THE OBSTACLE: PUBLIC AND tools and to support improvements in how PRIVATE ACTORS OFTEN LACK decisions are made. DATA, MODELS, AND CAPACITY Major “data bottlenecks” impair the design So far, the three recommendations given in of more resilient infrastructure. One example is part III of this report have stressed the need for the high-resolution digital elevation model standards and regulations, plans, and financial (DEM), a data set that provides the topography incentives to build the resilience of infrastruc- of a given area. Such data are the basis for ture systems (see chapters 9–11). But a major many hazard models and assessments, includ- challenge is to include natural hazards and cli- ing hydrological and flood models and land- mate change in these regulations, plans, and slide susceptibility analysis. These data are fre- incentives. And in the absence of natural haz- quently generated using a LIDAR installed on a ard and climate change data and models, plane. Recent DEMs have been generated at a well-meaning operators are unable to improve lower cost using drones—for example, in Dar resilience, and regulators are unable to create es Salam, through collaboration between the smart, efficient regulations and incentives. World Bank and the Red Cross—but this Thus, a package of resilient infrastructure poli- approach has its limits when a large area needs cies should include investments to ensure that to be mapped.1 Although a DEM is generally stakeholders have access to the right data and available for all urbanized areas in high- 173 174 LIFELINES income countries, such data are often unavail- tion codes and standards, for example, depends able in low- and middle-income countries, on what the users of infrastructure services can making it impossible to create the required easily handle. The introduction of compensa- flood and landslide hazard maps. Investing a tion systems for power or water outages (such few million dollars in a DEM would help to as in Finland, where power distribution com- improve the design of billions of dollars of resil- panies have to compensate users for outages) ient infrastructure. requires an estimate of the economic cost of Another example of a data bottleneck is the these outages. This information relies, in turn, lack of long time series of hydrometeorological on collecting data from infrastructure users as data. These data are needed to design infra- well as understanding how disruptions affect structure, but they may not be available households’ daily lives and the productivity because the data have never been collected, or and effectiveness of businesses. However, this because they have never been digitized, or knowledge is very patchy, especially in low- because they are accessible only at a prohibitive and middle-income countries, and far more price. Remote sensing using satellites, drones, systematic household and business surveys are and progress in computing has made it easier needed to better map and understand users’ and cheaper to monitor and model environ- needs (see chapter 2 and 3). mental conditions. However, these new tools cannot fully replace the networks of well- RECOMMENDATION 4: IMPROVE maintained weather- and water-monitoring DECISION MAKING stations and data processing that are still miss- What are the solutions for coping with these ing in many low-income countries. This situa- challenges of insufficient data, models, skills, tion stems in part from the low capacity and and competencies, which result in insufficient lack of resources of national hydrometeorologi- consideration of natural hazards and climate cal services in most of the world (Rogers and change? Besides investing in freely accessible Tsirkunov 2013). data on these issues—aided by new technolo- Because of climate change and other long- gies—it is important to make an appropriate term trends—from land use artificialization to use of these data, identify robust decisions in soil degradation—historical data are now insuf- the face of great uncertainties, and minimize ficient predictors of future risks. Many studies the potential for regret and catastrophic fail- have shown that infrastructure design and ures. At the same time, governments should management cannot assume that the future build the skills needed to create and use the will resemble the past (McCarl, Villavicencio, data and models and mobilize the know-how and Wu 2008; Milly et al. 2008). Today, proper of the private sector. risk assessment should include the effects of climate change, using the many climate models Action 4.1: Invest in freely accessible now available. However, outputs of climate natural hazard and climate change data models are very different from observations of Investments in data and models can provide the historical climate (Hallegatte 2009; Kalra et extremely high returns on investments by al. 2014), and using them requires accounting improving the design of billions of dollars of for the uncertainty in these results. long-lived infrastructure assets. The scenarios Data on natural hazards and climate change developed in chapter 6 suggest that strength- are not enough. Also needed are data on how ening infrastructure is much cheaper if invest- users are managing disruptions. The minimum ment is targeted to the most exposed and most level of resilience that can be set in construc- critical assets. The ability to target investments IMPROVE DECISION MAKING 175 is estimated to divide the cost by 10, reducing tems. For example, power plants are usually it from between $120 billion and $670 billion curtailed before the arrival of a hurricane to to between $11 billion and $65 billion. This minimize the potential for damage and cascad- finding suggests that the value of precise haz- ing events. Before Hurricane Sandy hit New ard information is orders of magnitude higher York City in 2012, the Metropolitan Transit than the cost of generating these data and Authority moved its trains out of flood-prone modeling results. Other studies have con- areas, minimizing the impact on its equipment cluded that investing in hazard data and fore- and allowing it to restore services relatively casting capacities is very cost-effective (Halle- rapidly after the storm. gatte 2012; Rogers and Tsirkunov 2013; WMO To be useful, data need to be available to et al. 2015). those making decisions. Multiple initiatives to Specific investments in data bottlenecks improve access to hazard and risk data contrib- could be transformational and generate large ute to improved decision making—such as the benefits. For instance, producing a DEM would Global Facility for Disaster Reduction and cost between $120 and $200 per square kilo- Recovery (GFDRR) ThinkHazard! platform, meter with a plane-installed LIDAR, with other which provides a simple estimate of hazard options (such as stereo photography) ranging exposure everywhere on the globe and a link between $30 to $100 per square kilometer. to the underlying data necessary to conduct Producing DEM for all urban areas in low- and more in-depth assessments. middle-income countries would cost between Also, there are increasing calls to adopt open- $50 million and $400 million in total—making data policies across government and academic it possible to perform an in-depth risk assess- research to ensure that these data generate as ment for all new infrastructure assets and, in much benefit as possible. Open-data licensing the process, inform hundreds of billions in supports transparency, efficiency, and participa- investment per year. In addition to better- tion in government; peer review of science; and designed infrastructure systems, these data more widespread and effective use of data for would allow risk-informed land use and urban- decision making in general. The Open Data for ization planning, which would also generate Resilience Initiative (OpenDRI) of the GFDRR large benefits. And if the data were available has been working on these issues in relation to for free, one could expect private sector actors disaster and climate risk assessment since 2011. to use them in innovative ways to improve risk OpenDRI has partnered with national govern- management for the whole economy. Overall, ments, universities, and community-based the benefits would be at least an order of mag- organizations to launch data-sharing platforms, nitude larger than the cost of generating and such as the Sri Lanka Disaster Risk Information distributing these data, which is also continu- Platform. The goal is to support community- ously declining, thanks to new technology to mapping projects for disaster risk assessment collect and process data (box 12.1). and to build tools for communicating complex To improve data availability, some organiza- risk information to diverse stakeholders. GFDRR tions have undertaken “data rescue”—in par- (2016) has compiled a list of key principles to ticular, digitizing the paper-based records of use in applying open-data approaches to disas- hydrometeorological data (WMO 2016). In ter risk management. addition to historical data, weather forecasting To make sure that these data influence and and early warning systems can play a key role support decision making, the asset manage- in anticipating extreme weather events and ment systems that are so useful to improve the mitigating their impacts on infrastructure sys- design and maintenance of infrastructure 176 LIFELINES BOX 12.1 New technologies make data collection and processing easier Satellite images and image processing. Today, flood maps of each community and publishes an urban development, transport networks, and atlas that includes community maps of the 21 power generation plants can be mapped through most flood-prone wards (World Bank 2018). the post-treatment of satellite images. For exam- ple, Graesser et al. (2012) have used satellite Social network analyses. Social media data images and machine learning to map informal include a wealth of information, but the challenge settlements in Caracas, Kabul, Kandahar, and here is to manage the huge volume of data and La Paz, helping to identify areas with poor infra- extract what is useful for decision making. For structure services and supporting the prioritiza- example, FloodTags uses Twitter information, tion of investments. However, difficult challenges natural language processing, and mapping tools remain, as illustrated by the relatively low per- to support postflood emergency management formance of algorithms for mapping the electric by identifying the location of flood emergencies. grid. a In postdisaster situations, aerial and satel- This tool has been piloted in the Philippines by lite images can also be used to assess the dam- the Red Cross. Although these approaches will age rapidly and to prioritize emergency actions not replace direct data collection, they provide (GFDRR 2019). useful complementary insights into a crisis and help to guide action. Crowdsourcing. This term describes the ways in which the Internet and mobile telephones facili- Traditional surveys. These surveys continue to tate outsourcing data collection tasks to the pub- play an important role in collecting data and lic. Crowdsourcing can be used to collect large understanding the importance of infrastructure amounts of data in real time at potentially lower services and their disruptions. Part I of this report costs than traditional approaches (UNISDR 2017). gives multiple examples of business or household OpenStreetMap, which has been used extensively surveys that provide information on the cost of in the risk assessments discussed in parts I and II infrastructure disruptions. These surveys sup- of this report, is one of the best-known examples port design of the right incentives for infrastruc- of an open database built by crowdsourcing, but ture service providers. For example, they help to it is not the only one. In Dar es Salam, the Ramani determine the appropriate magnitude of penal- Huria community maps identify flood-prone ties when service disruptions exceed regulations, areas. The project, initiated in 2015, trains stu- or rewards when performance goes beyond the dents and community members to create detailed construction code. a. See https://code.fb.com/connectivity/electrical-grid-mapping/. assets (see chapter 9) can easily be upgraded to and eventually the repair costs after heavy incorporate climate change and natural disas- rainfall. ter risks into decision-making processes. In Making data broadly available faces import- particular, the data recorded can include the ant challenges. One is that the collection of exposure of each asset to various hazards, data on the individual users of infrastructure which would inform the prioritization of main- services may pose privacy issues. As data col- tenance actions. For instance, keeping the cul- lection increases its spatial resolution and verts exposed to frequent floods free of waste devices such as smartphones collect individual can be prioritized, reducing their vulnerability data, it becomes possible to link risk informa- IMPROVE DECISION MAKING 177 tion to specific individuals. Concerns include and current research suggest that any ability to the use of these data for purposes beyond risk predict the future is limited at best (Kahneman management (for example, to target advertis- 2011; Silver 2012). Compounding the prob- ing). Moreover, flood exposure data can also lem, parties to a decision often have competing create issues, as households often fear expro- priorities, beliefs, and preferences. These condi- priation without due process and compensa- tions create deep uncertainty, which occurs tion if their home is in a flood zone. The trade- when parties to a decision do not know or can- off between the efficiency of risk management not agree on (1) the models that describe the and privacy concerns needs to be taken seri- key processes that shape the future; (2) the ously, and any data collection and distribution probability distribution of key variables and should come with clear and well-enforced reg- parameters in these models; or (3) the value of ulations regarding how the data can be used. alternative outcomes (Lempert, Popper, and Another challenge is the need to balance Bankes 2003). access to data with security considerations. The What is certain is that a cascade of uncer- data needed to identify critical infrastructure tainties plague climate change, and these and the priorities for strengthening networks uncertainties preclude prediction of the precise are the same data needed to plan the most nature, timing, frequency, intensity, and loca- damaging attacks on these networks. Since the tion of climate change impacts. Uncertainty 9/11 attacks in the United States, these consid- about the future rise in sea levels and about erations have led to the removal of much data temperature, precipitation, and other climate from the public domain in many countries. factors has tremendous implications for the One of the roles of the public agencies in near-term choices of decision makers. Exam- charge of risk management and critical infra- ples are where to locate key infrastructure such structure is to determine which data can be as airports, how to protect coastal areas from made publicly available, which data can be flooding, and how to ensure water security. shared among infrastructure service providers But climate-related uncertainty is not the with some conditions on their use and dissem- only issue; socioeconomic changes, political ination, and which data should be considered factors, disruptive new technologies, and too sensitive to be shared beyond specialized behavioral changes also create major uncer- agencies. Although these considerations are tainties that affect infrastructure-related deci- legitimate and create real trade-offs, national sion making. For example, the future perfor- security should not be used as a blanket excuse mance and cost of electric cars or the to restrict access to data. Best practices suggest availability of self-driving vehicles could sig- making open access to data the default situa- nificantly affect how cities develop. Because tion and creating strict processes to restrict the potential of these technologies is still being access for data proven to be too sensitive. debated, urban planners and developers face these uncertainties as well. Action 4.2: Make robust decisions and In the presence of such deep uncertainty, minimize the potential for regret and traditional methods—such as least-cost catastrophic failures approaches or cost-benefit analyses—are Regardless of the quality of the data and mod- unable to point to the preferred infrastructure els available, the long lifetime of assets and design. Traditional methods tend to search for deep uncertainty about the future exacerbate an optimum, which requires considering all the challenges of sound decision making in possible scenarios and knowing their probabil- infrastructure risk management.2 Past evidence ity. In situations of deep uncertainty, the prob- 178 LIFELINES abilities of future scenarios are difficult to • What is the potential for regret in the estimate, and this difficulty often leads to dis- ­ future? agreement. Faced with disagreement and deep • What should be done in case of failure? uncertainty, traditional decision-making approaches are vulnerable to bias and gridlock. Selecting robust solutions can usually be They are also vulnerable to reaching brittle achieved by selecting options that minimize decisions—ones that are optimal for a particu- the potential for regret. Here, regret is defined lar set of assumptions but that perform poorly, as the difference between what a given deci- or even disastrously, under other assumptions. sion would achieve—for instance, in terms of An alternative to seeking the “optimal” financial performance—and what the best solution is to look for a robust decision—one decision could have achieved. For instance, that performs well across a wide range of there is regret from having strengthened a futures, preferences, and worldviews, although bridge to resist a strong earthquake, if no such it may not be optimal in any particular one. strong earthquake occurs during the lifetime of New methods such as robust decision making, the bridge. Similarly, there is regret from not decision trees, and adaptive pathways have having strengthened the bridge if a strong been developed in the search for more robust earthquake does happen and the bridge is options (Haasnoot et al. 2013; Lempert and destroyed. Groves 2010; Ray and Brown 2015). These This metric and approach are used in chap- methods are sometimes also called “con- ter 6, where an exploration of the uncertainty text-first” (Ranger et al. 2010) or “agree-on- regarding the costs and benefits of strengthen- decisions” (Kalra et al. 2014). ing infrastructure assets shows that such These methods begin by stress-testing the strengthening is a robust action that is highly available options under a wide range of plausi- unlikely to lead to significant regret. This ble conditions, without requiring a decision or approach has also been applied to efforts to agreement on which conditions are more or identify the best design for future investments less likely. They evaluate the decision options in hydropower in Africa (Cervigni et al. 2015) repeatedly, under many different sets of and to assess the effort to preserve wetlands in assumptions, including low-likelihood but Colombo (box 12.2). In addition, the results of high-consequence events. stress tests can be used to create “failure sce- This process promotes consensus around narios” that can serve as a starting point for decisions and can help in the management of preparing for disruptions and creating contin- deep uncertainty. Analyses performed in this gency plans. way help decision makers to debate important questions: Action 4.3: Build the skills needed to use data and models and mobilize the • Are the conditions under which an option know-how of the private sector performs poorly likely enough to result in Even if all actors have access to data and mod- the choice of a different option? els, using them appropriately requires skills • What trade-offs should be made between and competencies that are not always avail- robustness and, for example, cost? able. For a government, utility, or agency, out- • Is it possible to add safety margins to a proj- sourcing the production of hazard and climate ect to hedge against surprises? change data without building the skills to use • Which options offer the most flexibility for these data appropriately in a robust decision- responding to unexpected changes in the making framework is unlikely to lead to signif- future? icant improvements in resilience. IMPROVE DECISION MAKING 179 BOX 12.2 Preserving wetlands in Colombo minimizes the risk of regret A study of Colombo’s floods, conducted amid FIGURE B12.2.1 Preserving a large share of large uncertainties about climate change and Colombo’s wetlands minimizes the potential for urban development, evaluates various choices regret in 2030 for the preservation of wetlands (Browder et al. 2019). The study looks specifically at how much regret decision makers would experience in 2030, Regret in 2030 (RS, billions) 200 comparing the realized level of risk and the best possible outcome. 150 The study finds that all conservation levels could lead to zero regret. In other words, for each conservation level measured on a scale of between 100 0 and 100 percent, there is at least one scenario in which the level is optimal (figure B12.2.1). But if a 50 small share of the wetlands is preserved and the rest is developed, the potential for regret is high: in 0 scenarios with a major increase in rainfall and river runoff, high population growth, and high building 0 30 50 90 100 Share of wetlands conserved (%) vulnerability, the development of wetlands would lead to substantial flood losses. And because wet- Source: Browder et al. 2019. lands are difficult and costly to recreate, these losses would be largely irreversible, resulting in wetlands is a reversible solution, if decision makers high regret. By contrast, the conservation of wet- want to avoid experiencing regret by 2030, they lands cannot lead to high regret because the main may prefer to conserve Colombo’s urban wetlands cost of this option is the opportunity cost of not for now, wait for more information on how climate developing the wetland areas, which is less uncer- and urbanization will evolve, and reconsider their tain than the cost of floods. Because conserving position in a decade or two. This is a major challenge in a low-income, ters can play an important role, training people low-capacity environment in which skilled in the right skills, developing new methodolo- engineers are scarce (figure 12.1). But it is also gies or adapting them to the local context, and a common challenge in local authorities and advising policy and decision makers. This sup- municipalities, which are often in charge of port is required not only for suppliers of infra- managing water or transport infrastructure. structure services, but also for users who can Small cities, even in rich countries, do not have prepare for infrastructure disruptions and min- the resources to hire specialized staff and need imize their costs—for example, with business to rely on external advisers and service provid- continuity plans. ers or on support from national or regional When public sector expertise is insufficient, agencies. bringing in the private sector—as advisers or Infrastructure sectors at all scales benefit through direct procurement or public-private from the support and expertise of local consult- partnerships (PPPs)—can be a solution. Both ing and engineering firms and other expertise domestic or international firms may have the centers based in universities, think tanks, and capacity and know-how to implement innova- research centers. Universities and research cen- tive solutions. This capacity is particularly 180 LIFELINES FIGURE 12.1 Many low- and middle-income countries need to design flaw will not lead to its collapse during a increase their enrollment in technical tertiary education stronger storm. Share of students in tertiary education enrolled in engineering, Where performance cannot be easily mea- manufacturing, and construction in 2016 sured and verified, labels and certifications can 25 provide the clients of a particular industry with some level of protection. Such labels can be 20 provided through self-regulation (such as when professional organizations create the 15 label) or assigned by the public sector (such as Percent when the ministry of construction certifies pri- 10 vate companies as capable of performing cer- tain tasks). Designing such certification is 5 tricky, however. If too strict, a certification pro- cess can easily slow down innovation, create 0 0 5,000 10,000 15,000 20,000 barriers to entry for new players, and reduce Country GDP per capita (US$) competition. If too lax, a label or certification Source: World Bank staff, based on data from the United Nations Educational, can help low-quality players—or even crooks— Scientific, and Cultural Organization Institute for Statistics. to enter the game. NOTES important in mobilizing the new technologies 1. See https://drones.fsd.ch/wp-content/uploads that are mostly developed in, and implemented /2016/03/Case-Studies-Dar-es-Salaam-Final2 -1617045.pdf. by, the private sector—for example, those related 2. This section is adapted from Kalra et al. (2014). to finance, telecommunications, and cybersecu- rity. However, cooperation between the public REFERENCES and private sectors is often difficult to establish Browder, G., S. Ozment, I. Rehberger Bescos, because of differences in culture and work hab- T. Gartner, and G.-M. Lange. 2019. Integrating its, issues related to privacy and commercial Green and Gray: Creating Next Generation Infrastruc- secrecy, and the risks of capture and rent-seek- ture. Washington, DC: World Bank and World ing behaviors from private actors, especially Resources Institute. Cervigni, R., R. Liden, J. E. Neumann, and K. M. where public agencies have limited capacity. Strzepek, eds. 2015. Enhancing Climate Resilience of Experts supporting long-term planning Africa’s Infrastructure: The Power and Water Sectors. would benefit from the introduction of labels Washington, DC: World Bank. https://doi and certifications. In the areas of climate .org/10.1596/978-1-4648-0466-3. change and resilience to low-probability events, GFDRR (Global Facility for Disaster Reduction and customers and users cannot easily observe the Recovery). 2016. “Open Data for Resilience Ini- tiative.” GFDRR, Washington, DC. performance of various engineering and con- ———. 2019. “Machine Learning for Disaster Risk sulting firms. The effectiveness of a recommen- Management.” GFDRR, Washington, DC. dation on climate change adaptation (for exam- Graesser, J., A. Cheriyadat, R. R. Vatsavai, V. Chan- ple, for the construction of a road) will not be dola, J. Long, and E. Bright. 2012. “Image Based evident until decades after the asset is built. The Characterization of Formal and Informal Neigh- borhoods in an Urban Landscape.” IEEE Journal resilience of infrastructure cannot be observed of Selected Topics in Applied Earth Observations and until it is tested by severe weather or an earth- Remote Sensing 5 (4): 1164–76. quake. The fact that a bridge performs as Haasnoot, M., J. H. Kwakkel, W. E. Walker, and expected for decades does not mean that a J. ter Maat. 2013. “Dynamic Adaptive Policy IMPROVE DECISION MAKING 181 Pathways: A Method for Crafting Robust Deci- A Decision-Making Process.” Centre for Climate sions for a Deeply Uncertain World.” Global Envi- Change Economics and Policy, University of ronmental Change 23 (2): 485–98. Leeds; Grantham Research Institute on Climate Hallegatte, S. 2009. “Strategies to Adapt to an Uncer- Change and the Environment, London. tain Climate Change.” Global Environmental Ray, P. A., and C. M. Brown. 2015. Confronting Cli- Change 19 (2): 240–47. https://doi.org/10.1016/j mate Uncertainty in Water Resources Planning and .gloenvcha.2008.12.003. Project Design: The Decision Tree Framework. Wash- ———. 2012. “A Cost-Effective Solution to Reduce ington, DC: World Bank. https://doi.org/10.1596 Disaster Losses in Developing Countries: /978-1-4648-0477-9. Hydro-Meteorological Services, Early Warning, Rogers, D., and V. Tsirkunov. 2013. Weather and Cli- and Evacuation.” Policy Research Working Paper mate Resilience: Effective Preparedness through 6058, World Bank, Washington, DC. https://doi National Meteorological and Hydrological Services. .org/10.1596/1813-9450-6058. Vol. 62. Washington, DC: World Bank. https:// Kahneman, D. 2011. Thinking, Fast and Slow. New doi.org/10.1596/978-1-4648-0026-9. York: Farrar, Strauss and Giroux. Silver, N. 2012. The Signal and the Noise: Why So Kalra, N., S. Hallegatte, R. Lempert, C. Brown, A. Many Predictions Fail—but Some Don’t. New York: Fozzard, S. Gill, and A. Shah. 2014. “Agreeing on Penguin Press. Robust Decisions: New Processes for Decision UNISDR (United Nations Office for Disaster Risk Making under Deep Uncertainty.” Policy Reduction). 2017. “Words into Action Guide- Research Working Paper 6906, World Bank, lines: National Disaster Risk Assessment— Washington, DC. https://doi.org/doi:10.1596 UNISDR.” UNISDR, Geneva. https://www.unisdr /1813-9450-6906. .org/we/inform/publications/52828. Lempert, R. J., and D. G. Groves. 2010. “Identifying WMO (World Meteorological Organization). 2016. and Evaluating Robust Adaptive Policy Responses “Guidelines on Best Practices for Climate Data to Climate Change for Water Management Agen- Rescue.” WHO, Geneva. https://public.wmo.int cies in the American West.” Technological Forecast- /en/resources/library/guidelines-best-practices ing and Social Change 77 (6): 960–74. -climate-data-rescue. Lempert, R. J., S. W. Popper, and S. C. Bankes. 2003. WMO (World Meteorological Organization), World Shaping the Next One Hundred Years: New Methods Bank Group, GFDRR (Global Framework for for Quantitative, Long-Term Policy Analysis. Santa Disaster Reduction and Recovery), and USAID Monica, CA: RAND Corporation. (U.S. Agency for International Development). McCarl, B. A., X. Villavicencio, and X. Wu. 2008. 2015. “Valuing Weather and Climate: Economic “Climate Change and Future Analysis: Is Station- Assessment of Meteorological and Hydrological arity Dying?” American Journal of Agricultural Eco- Services.” WMO, Geneva. https://public.wmo nomics 90 (5): 1241–47. https://doi.org/10.1111/j .int/en/resources/library/valuing-weather .1467-8276.2008.01211.x. -and-climate-economic-assessment-of-meteoro Milly, P. C. D., J. Betancourt, M. Falkenmark, R. M. logical-and-hydrological. Hirsch, Z. W. Kundzewicz, D. P. Lettenmaier, and World Bank. 2018. “Ramani Huria: The Atlas of R. J. Stouffer. 2008. “Stationarity Is Dead: Whither Flood Resilience in Dar es Salaam.” Working Water Management?” Science 319 (5863): 573–74. Paper 125586, World Bank, Washington, DC. https://doi.org/10.1126/science.1151915. http://documents.worldbank.org/curated Ranger, N., A. Milner, S. Dietz, S. Frankhauser, A. /en/200421524092301920/Ramani-Huria-the Lopez, and G. Ruta. 2010. “Adaptation in the UK: -atlas-of-flood-resilience-in-Dar-es-Salaam. Provide Financing 13 OBSTACLE RECOMMENDATION ACTIONS Affordability and financing Provide financing 5.1: Provide adequate funding to include risk •  constraints assessments in master plans and early project design 5.2: Develop a government-wide financial protection •  strategy and contingency plans 5.3: Promote transparency to better inform investors •  and decision makers THE OBSTACLE: quantity of infrastructure services (such as THE INFRASTRUCTURE SECTOR fewer but stronger schools). FACES AFFORDABILITY More often, however, making a project AND FINANCING CONSTRAINTS more resilient does not increase its life-cycle Efforts to increase the resilience of infrastruc- cost, but will increase its design cost, construc- ture can increase various components of its full tion cost, or maintenance cost, linking the cost, including the cost borne by the govern- challenge to financing. If the life-cycle cost is ment or regulators or the cost borne by the pro- manageable within the available resources, viders of infrastructure services (see figure 9.4). then the challenge is to allocate resources In some cases, making an infrastructure toward the early project stages to ensure good asset more resilient leads to an absolute design, toward the substantial up-front invest- increase in its life-cycle cost and thus to afford- ments required for resilient infrastructure, and ability challenges. For example, when struc- toward a regular flow of resources to ensure tural reinforcements of schools are needed to good maintenance. prevent the loss of life during an earthquake, solutions might include either an increase in RECOMMENDATION 5: ENSURE funding (such as from higher taxes, higher user FINANCING fees, or larger regional or international trans- What are the required measures that can help fers) or a trade-off between the resilience and to address affordability challenges and financ- 183 184 LIFELINES ing constraints? This section highlights three risk assessment to inform master-planning key actions: the need for adequate funding to exercises and early infrastructure project devel- include risk assessments in master plans and opment. For this reason, disaster and climate early project design; the need for a government- risk assessment is one of the areas that has wide financial protection strategy to aid recov- been identified as a priority for the use of cli- ery from disasters that cannot be prevented; mate finance (World Bank 2018) and one of and the need for greater transparency of infra- the main focuses of the 2018–21 strategy of structure investments to ensure that investors the Global Facility for Disaster Reduction and and decision makers have the information they Recovery (GFDRR). need to select the best, and most resilient, Dedicated organizations and project prepa- projects. ration facilities can support the inclusion of risk assessment in master planning, regulation Action 5.1: Provide adequate funding design, and early stages of infrastructure proj- to include risk assessments in master ects. For example, the Global Infrastructure plans and early project design Facility—a partnership of governments, multi- Regulators often have limited budgets, thus lateral development banks, private sector making it difficult for them to design the right investors, and financiers—supports the prepa- codes and regulations or to enforce them (see ration, structuring, and implementation of chapter 9). Further, master-planning exercises complex infrastructure projects. In particular, it are critical to capture the system-level options supports preliminary work to prioritize invest- for resilience (see chapter 7), but these exer- ments and test a project concept through “pre- cises often lack the resources to conduct a full- feasibility” analysis. It also supports, if needed, fledged risk assessment. the legal, regulatory, and institutional reforms Similarly, budgets tend to be small at the required to enable the successful development early stage of preparation of an infrastructure or participation of long-term private capital in asset project, making it difficult to conduct the the financial structure of a project. These inter- sophisticated risk and resilience analyses ventions could easily have some resilience- needed (such as those recommended in chap- related aspects, such as considering resilience ter 12)—even if they can generate massive sav- in a prefeasibility study or developing the legal ings over the lifetime of an infrastructure asset. and regulatory environment to ensure that cli- When projects are more mature and financing mate and disaster risk are considered in the is easier to access, more resources become development of public-private partnerships. available. However, at this stage most strategic decisions already have been made, and most Action 5.2: Develop a government- low-cost options to increase resilience are no wide financial protection strategy and longer available, including, for example, contingency plans for natural hazards changing the location of an asset or even the This report has identified the need for infra- nature of a project. Providing the financing and structure asset operators to have the capacity technical support needed to include risk analy- to respond quickly to incidents, so that inci- sis at the early stages of project design can, dents that cannot be avoided can be managed therefore, be extremely cost-effective. (see chapter 9). However, disruptions caused Governments or international organizations by natural hazards have specific characteristics: can achieve a transformational impact, with they tend to be larger, last longer, and be cost- relatively limited resources, by financing the lier than those caused by system failure. For generation of appropriate data and dedicated instance, hazard-related power outages in PROVIDE FINANCING 185 Europe last four times longer than those due to needed. Doing so requires that governments be nonnatural causes. And while Bangladesh prepared before a disaster hits, with the right experiences almost daily power outages, a instruments, institutions, and capacities in tropical cyclone caused the largest outage in place. The measures that can ensure rapid the country, during which all 26 power plants recovery and reconstruction include (1) contin- stopped operating. Similarly, while congestion gency plans to ensure that the coordination of linked to car accidents is a daily occurrence in recovery and reconstruction efforts is effective all countries in the world, an earthquake can and that responsibilities are clearly allocated damage hundreds of bridges at once. among government agencies; (2) contingent To manage hazard-related asset losses and financial arrangements—such as contingent disruptions, countries need additional instru- credit lines or insurance products—to ensure ments, with specific contingency plans and a that financing is immediately available and not financial protection strategy. Resources to help delayed by budgetary procedures; (3) prear- countries to build such a strategy can make ranged contracts to accelerate procurement— countries more resilient—a topic generating for example, ensuring that debris can be increasing interest from the international com- removed as soon as possible to facilitate recon- munity. For instance, the Global Risk Financ- struction; and (4) international cooperation to ing Facility (GRiF) is a recently established share the cost of the staff and equipment financing mechanism that supports the devel- needed for the recovery and reconstruction. opment of risk-informed financial planning Governments usually finance most infra- across different sectors and the continuity of structure recovery and reconstruction. They critical public services (such as electricity, trans- can ensure liquidity during these phases in port, and water). three ways: (1) maintaining sufficient reserve When a disaster damages infrastructure, funds, (2) arranging for contingent credit facil- resources are needed to manage the disrup- ities, or (3) using insurance schemes or trans- tions both on the supply side (to recover, ferring risk. Governments can structure these repair, and reconstruct) and on the demand financial instruments along “risk layers,” with side (to help users to manage the disruptions). different instruments covering different types Postdisaster needs usually have two phases: of risks (figure 13.1). By using a layered disas- recovery and reconstruction. The recovery ter risk financing strategy, countries such as phase refers to the weeks and months follow- Mexico and the Philippines have prepared ing a disaster, during which relatively limited themselves for a wide range of contingencies. resources are urgently required. In this phase, Reserve funds are used to manage low-cost, timeliness is critical. The reconstruction phase high-probability events, whereas contingent refers to the longer period, during which infra- financing and sovereign risk transfer instru- structure and buildings are repaired or rebuilt. ments are used for high-cost, low-probability This phase often involves massive funding and events (Ghesquiere and Mahul 2007; Mahul financing needs, but with less urgency so that and Ghesquiere 2010; World Bank 2017). traditional funding and financing mechanisms Sometimes, infrastructure assets can be can be mobilized. insured directly by their owners or operators, But the availability of financial resources is whether private or public entities. For exam- only half of the story. The capacity of a govern- ple, the Kenyan government has implemented ment to support postdisaster recovery and requirements for mandatory disaster risk insur- reconstruction depends greatly on its ability to ance coverage in power purchase agreements. deliver these resources to where they are Insurance helps to finance repairs and recon- 186 LIFELINES FIGURE 13.1 Countries need a layered risk financing strategy Low frequency High intensity Sovereign risk transfer • Insurance (including risk pools) Insurance of public assets • Derivatives • Catastrophe bonds Contingent financing Postcrisis financing Hazard • World Bank, IDB, JICA: Deferred Drawdown Option (DDO) • Emergency lending • Contingent Emergency Response Components (CERC) • Bilateral or multilateral • IDA Crisis Response Window (CRW) financing High frequency Low intensity Budgetary instruments • Sovereign reserve funds • Contingent budget • Government reserves • Budget reallocation Short-term liquidity Long-term financing needs Time Source: World Bank 2017. Note: IDB = Inter-American Development Bank; JICA = Japan International Cooperation Agency; IDA = International Development Association. struction after a shock and also creates a pow- The speed at which funds can be obtained is erful incentive for infrastructure asset owners determined by the underlying legal and admin- and operators to reduce risks in order to pay a istrative processes (Mahul and Ghesquiere lower premium for more resilient assets. Usu- 2010). However, the cost multipliers and speed ally, an insurance policy requires appropriate of fund disbursement may vary on a case-by- maintenance as a condition of the insurance case basis, depending on the type of hazard, payment, which should incentivize govern- the frequency of the payout, or the institu- ments to maintain infrastructure properly and tional and management capacity in the to mitigate disaster risks. The feasibility and country. desirability of insurance depend on the matu- Other important considerations include the rity of the domestic insurance markets, the fea- transparency and predictability of the resources sibility of accessing the global reinsurance mar- (Clarke and Dercon 2016). Rule-based instru- kets or other capital market instruments, and ments—such as index insurance products and cost considerations. risk transfer mechanisms based on measurable The choice of financial instruments is deter- indicators—provide governments, technical mined not only by their functionality but also agencies, local authorities, and firms and by their cost and speed of disbursement. Table households with a predictable amount of sup- 13.1 provides an indicative cost multiplier for port and enable them to design their own different financial risk instruments. The cost response (such as taking out their own insur- multiplier is defined as the ratio of the cost of a ance contract). From a government perspec- financial product (such as the premium of an tive, rule-based instruments also help to build insurance product or the expected net present discipline regarding how postdisaster resources value of the cost of a contingent debt facility) are mobilized and used. to the expected payout over its lifetime. A ratio Contingent financing can also help users to of 2 indicates that the overall cost of the finan- cope with, and recover from, infrastructure cial product is likely to be twice the amount of disruptions. Indeed, as emphasized in this the expected payout over a long period of time. report, reconstruction costs make up only a PROVIDE FINANCING 187 TABLE 13.1 Cost multipliers vary across financial instruments for risk management Indicative cost Disbursement Amount of funds Instrument multiplier (months) potentially available Ex post financing Donor support (humanitarian relief) 0–1 1–6 Uncertain Donor support (recovery and reconstruction) 0–2 4–9 Uncertain Budget reallocations 1–2 0–9 Small Domestic credit (bond issues) 1–2 3–9 Medium External credit (e.g., emergency loans, 1–2 3–6 Large bond issue) Ex ante financing Budget contingencies 1–2 0–2 Small Reserves 1–2 0–1 Small Contingent credit 1–2 0–1 Medium Parametric insurance 1.3 and up 0–2 Large Alternative risk transfer (for example, cat bonds, 1.5 and up 1–6 Large weather derivatives) Traditional (indemnity-based) insurance 1.5 and up 2–12 Large Source: Mahul and Ghesquiere 2010. fraction of the full cost of a lack of resilience. adaptive social protection system, supple- After major disruptions, small firms will have mented by ad hoc postdisaster transfers for peo- lost clients and sales, and households will have ple who are not covered by existing systems had to spend more to buy bottled water and (Hallegatte et al. 2017). Firms in the formal sec- batteries or will have lost income after mem- tors can usually be supported through tax bers are unable to go to work. And the firms breaks, ad hoc transfers, or subsidized loans. For and households that are affected by infrastruc- example, in 2007 the Shizuoka Prefecture ture disruptions can be located far from the Credit Guarantee Association in Japan devel- areas directly hit by natural hazards, and the oped a postdisaster guarantee program for small distribution of postdisaster support may have and medium enterprises. Through the program, to cover a much larger spatial area than the small and medium firms with business continu- disaster itself (Colon, Hallegatte, and Rozen- ity plans can submit preapplications for a post- berg 2019; Rentschler et al. 2019). disaster credit guarantee, and the guarantee fee In some countries, the regulations governing is waived if a business borrows after a disaster. infrastructure services call for compensating Often, small businesses in the informal sector users affected by outages—especially those in are the most difficult to support in the face of the power, telecommunications, and water sec- legal and technical obstacles, and ad hoc action tors. In the absence of a compensation system, may sometimes be necessary. government may want to help firms and house- holds to manage infrastructure disruptions in Action 5.3: Promote transparency to the same way that it helps them to manage the better inform investors and decision reconstruction of dwellings and replacement of makers assets. Typical instruments for households (and One way to ensure that financing is directed to households’ individual enterprises) include an more resilient infrastructure projects is to 188 LIFELINES ensure that investors are informed about the bonds. The number of U.S. institutional inves- risks attached to projects. They may, then, pre- tors considering ESG factors in their decisions fer the more resilient ones. Such an approach almost doubled between 2013 and 2018, from requires transparency on every project’s expo- 22 to 40 percent, according to the Callan Insti- sure and vulnerability to various hazards in a tute, but the inclusion of resilience within ESG way that is currently not available. considerations remains limited. Multiple international, regional, and There are many indicators for measuring national initiatives have been designed to the sustainability of infrastructure, using an increase the transparency of the physical risks ESG lens (box 13.1). However, resilience is also attached to investments in assets. Examples a key driver of performance. One challenge for include the Task Force for Climate-Related the tools that inform investors and decision Financial Disclosure (TCFD), which recom- makers is how to identify the performance mends that businesses (and the financial actors dimension (how will natural risks and climate that invest in them) report physical risks and change affect the return on a financial prod- how they are managed. Recognizing the chal- uct?) and the ESG dimensions (how will a lenges of standardizing the disclosure of physi- financial product contribute to economic, cal risks, the European Bank for Reconstruc- social, and environmental sustainability?). tion and Development and the Global Centre To identify these dimensions and comple- of Excellence on Climate Adaptation launched ment the existing measurement systems, the a study that highlights the need to perform for- World Bank Group is committed to developing ward-looking assessments of climate-related a resilience rating system, which would aim at risks using various scenarios. Titled “Advancing better informing investors and decision makers TCFD Guidance on Physical Climate Risk and on the resilience characteristics of their proj- Opportunities,” it recommends that firms and ects. This rating system would not create new financial institutions report on the exposure of information or data. Instead, it would translate their assets to natural hazards and provide the highly technical information already exist- qualitative information on how they manage ing in project documents into a simple rating them, thereby facilitating assessment of the that can be of use to people without an engi- impacts of climate risk on corporate perfor- neering background. It will rate projects along mance and credit risks. two dimensions of resilience: The decision making of infrastructure inves- tors is increasingly including consideration of • Dimension 1: resilience of investments and proj- environmental sustainability (Bennon and ects. This dimension measures the extent to Sharma 2018), mainly through the adoption of which a project has taken climate and disas- environmental, social, and governance (ESG) ter risks into consideration. The rating, principles or a responsible investment expressed in grades from A+ to D, charac- approach. The United Nations–supported Prin- terizes the confidence in a project’s ability to ciples for Responsible Investment Program has avoid financial, environmental, and social been endorsed by more than 2,000 organiza- underperformance. A high rating, for exam- tions (including asset owners, investment ple, denotes higher confidence that the managers, and other financial service provid- expected rate of return of an investment ers). And although infrastructure equity accounts for the possible negative impacts of investments have led the way in taking ESG natural hazards or climate change on the principles into consideration, the fixed-income investment. With a low rating and every- space is also beginning to include ESG princi- thing else being equal, the expected rate of ples—for example, green bonds and social return is unlikely to be achieved and would PROVIDE FINANCING 189 BOX 13.1 Many indicators have been developed to measure the sustainability of infrastructure Today, various sets of standards for sustain- Civil Engineering Environmental Quality Assess- •  able infrastructure are including considerations ment and Awards (CEEQUAL) scheme, an of resilience or at least governance dimensions assessment scheme launched in 2003, with related to resilience. What follows is a summary fees ranging from less than $6,500 for very of five of these standards: small projects to more than $58,000 for large projects: Standard for Sustainable and Resilient Infra- •  T he scheme assesses nine categories of a   structure (SuRe), a project certification stan- project’s environmental management and dard developed by the Global Infrastructure impacts. Basel Foundation in Switzerland, in collabora- Each category consists of a series of point-   tion with Natixis, a French investment bank: scored questions that can be applied to dif-  S uRe’s cost for certification is between ferent management practices and perfor- $30,000 and $60,000, depending on the size mance indicators. of the project and its stage of development. T he scheme includes simple questions    T he certification applies 61 criteria across related to the consideration of and response 14 themes. The criterion dedicated to “resil- to expected changes in climate conditions. ience planning” requires that a vulnerabil- ity assessment be conducted for the proj- I nternational Finance Corporation (IFC) Per- •  ect’s life cycle, that resilience measures formance Standards (and Equator Principles), be reported, and that a risk-monitoring a methodology with eight performance stan- system be included in the project. It also dards for projects financed by the IFC, derived includes components regarding emergency from the World Bank Group’s environmental, response preparedness and supply chain health, and safety project guidelines: vulnerabilities.  T he first standard relates to the “Assess- ment and Management of Environmental Envision, a rating system developed jointly by •  and Social Risks and Impacts” and includes the Institute for Sustainable Infrastructure and requirements for emergency management. the Zofnass Program for Sustainable Infrastruc-  Performance standards focus on the risk cre- ture at Harvard University: ated by the project without in-depth explora- P rojects can opt for verification, with fees   tion of the risks to the project. ranging from $11,000 to $56,000. The rating system includes 60 sustainability   GRESB (ESG Benchmark for Real Assets) Infra- •  criteria, or credits, in five categories: quality structure, a project-level and a portfolio-level of life, leadership, resource allocation, natural assessment tool for asset owners, fund manag- world, and climate and risk. ers, contractors, and asset managers: The climate and risk category includes many    Data collected include management practice subindicators related to resilience: the devel- indicators regarding sustainability planning, opment of a comprehensive impact assess- eight categories of environmental perfor- ment and adaptation plans; consideration mance indicators, and project performance of long-term trends such as climate change; metrics. preparation for long-term adaptability; and  A resilience module focuses on preparation the management of short-term threats and for disruptive events and long-term trends heat island effects. such as climate change. Source: Bennon and Sharma 2018. 190 LIFELINES need to be adjusted to account for disasters fits and profits. It also accounts for the broader and climate change. This metric provides impacts on communities and economies. This information not on whether the project is tool aims to help investors select the best proj- likely to fail, but on whether the risk of fail- ects and contribute to a more productive and ure (which can be low or high, depending resilient future. on the case) is considered in the economic If such a resilient rating system becomes or financial analysis that justifies the proj- common practice, it could allow investors to ect. As a result, a project with a high risk of impose a minimum standard in terms of how failure can be highly rated, provided this new infrastructure projects account for natural risk is accounted for in the analysis. The hazards and climate change. It would help to project may in fact be attractive in spite of translate high-level commitments to support this risk, if the potential returns in the more resilient societies into changes in prac- absence of failure are extremely high. tices and designs in the real economy. And if • Dimension 2: resilience building through invest- such a system becomes embedded in govern- ments and projects. Targeted investments, or ment budgetary processes, it could also influ- specific components of investments, are ence public spending—which represents the often designed with the objective of build- large majority of investments in infrastruc- ing the resilience of beneficiaries. Examples ture—and support a broad transition toward of this are a seawall or a drainage system more resilient infrastructure systems. needed to manage storm surges or heavy More transparency regarding the resilience precipitation in cities or a new road that of private and public investments would also connects an isolated village to markets, provide a strong incentive for implementation building food security. Such investments of the other recommendations presented in thus support moving toward greater resil- this report and help to manage the political ience against current and future risks. economy challenges highlighted earlier. Today, The distinction between this dimension there is little immediate reward for a govern- and the first is important: although all ment that provides the right hazards and cli- projects should be resilient, not all pro- mate change data to its infrastructure operators jects seek to improve the resilience of the or that regularly updates its construction codes broader community or country. Thus, the and infrastructure sector regulations. More second dimension helps to prioritize and transparency on the resilience of infrastructure promote those investments that are key to projects would give visibility to these actions, climate-resilient development and longer- by improving the rating of the investments tak- term resilience development pathways. The ing place in a country. If aggregated, it could rating conditions for this category—also even help to build a country-level measure of expressed with letter grades—are by neces- the resilience of new investments. sity less technical than those of the first, and they depend, other things being equal, These synergies are only a fraction of the on beneficiaries and related vulnerabilities. many synergies that exist between the recom- mendations made in this report. Indeed, no The objective of this two-dimensional rating single intervention can make all infrastructure system is to ensure that each and every invest- systems resilient. Instead, governments will ment made by the private or the public sector need to define and implement a consistent gives due consideration to natural disaster and strategy—in partnership with all stakeholders climate change risks, examining its own resil- such as utilities, investors, business associa- ience and ability to deliver the expected bene- tions, and citizen organizations—to tackle the PROVIDE FINANCING 191 many obstacles to more resilient infrastructure Policy Research Working Paper 4345, World systems. And while doing so will be challeng- Bank, Washington, DC. https://doi.org/10.1596 /1813-9450-4345. ing and take time, this report highlights the Hallegatte, S., A. Vogt-Schilb, M. Bangalore, and potential benefits of doing so and of doing it J. Rozenberg. 2017. Unbreakable: Building the without delay. According to the analysis pre- Resilience of the Poor in the Face of Natural Disasters. sented in this report, each decade of inaction Washington, DC: World Bank. https://doi.org may cost the world trillions of dollars. /10.1596/978-1-4648-1003-9. Mahul, O., and F. Ghesquiere. 2010. “Financial Pro- REFERENCES tection of the State against Natural Disasters: A Bennon, M., and R. Sharma. 2018. “State of the Primer.” Policy Research Working Paper 5429, Practice: Sustainability Standards for Infrastruc- World Bank, Washington, DC. https://doi.org ture Investors.” SSRN Electronic Journal, October. /10.1596/1813-9450-5429. https://doi.org/10.2139/ssrn.3292469. Rentschler, J., M. Kornejew, S. Hallegatte, M. Obo- Clarke, D. J., and S. Dercon. 2016. “Dull Disasters? lensky, and J. Braese. 2019. “Underutilized How Planning Ahead Will Make a Difference.” Potential: The Business Costs of Unreliable Infra- World Bank, Washington, DC. https://doi.org structure in Developing Countries.” Background /10.1017/CBO9781107415324.004. paper for this report, World Bank, Washington, Colon, C., S. Hallegatte, and J. Rozenberg. 2019. DC. “Transportation and Supply Chain Resilience in World Bank. 2017. Sovereign Catastrophe Risk Pools. the United Republic of Tanzania.” Background Washington, DC: World Bank. https://open paper for this report, World Bank, Washington, knowledge.worldbank.org/handle/10986/28311. DC. ———. 2018. Strategic Use of Climate Finance to Maxi- Ghesquiere, F., and O. Mahul. 2007. “Sovereign Nat- mize Climate Action. World Bank, Washington, DC. ural Disaster Insurance for Developing Countries: https://openknowledge.worldbank.org/handle A Paradigm Shift in Catastrophe Risk Financing.” /10986/30475. AP P E N D I X Engineering Options to A Increase the Resilience of Infrastructure Assets T he tables in this appendix are adapted from Miyamoto In­ ternational (2019), which provides and the costs of such improvements depend on specific applications, site conditions, and other variables. more information on each of the op­ Unless stated otherwise, the im­ tions, including sources for the esti­ provement costs are typically for mates provided. The data in many enhancements implemented during of the cells rely on past experience the construction of new units. For and engineering judgment. The data most applications, the costs of retro­ are intended to be a representative fitting improvements are similar. sample. The actual improvements, For more details, see Miyamoto In­ benefits from such improvements, ternational (2019). TABLE A.1 Engineering options to improve infrastructure asset resilience in the power sector Damage Incremental Natural hazard Critical system/component probability cost (including Engineering Quality quality Type Hazard Intensity Component improvement improvement Baseline Improved control) Thermal EQ motion Mw 7 PGA Items Seismic component and Construction inspection, 0.25 0.02 0.20 power plants 0.4g and their anchorage testing (coal, natural attachments gas, oil) Liquefaction ND Substrate Soil improvement Geotechnical report and 0.30 NS 0.20 Deep foundation testing Wind 100 mph Building Stiff braced structures Welding quality control, 0.40 0.1 0.10 structures Helical strake inspection, testing stacks Flood 2- to 3-ft. Entire Floodwall, sheet piling Ensure watertight 0.05 NS 0.02 inundation facility construction, inspection Hydropower EQ motion Mw 7 PGA Gateway, lift Design for stronger Inspection during 0.7 0.4 0.2 plants 0.4g joints, intake events, use proper construction, periodic towers anchorage and seismic inspection components Liquefaction N/A Wind N/A Flood Large Spillways, Increased spillway Proper drenching, 0.1 0.05 0.03 rainstorms, dam crest capacity underwater inspection 200- to overtopping 500-year flood Solar farms EQ motion Mw 7 PGA Support Adequate anchorage, Inspection, maintenance 0.1 0.02 0.05 0.4g structure proper design and bracing Liquefaction N/A Wind 100 mph Uplift support Proper anchorage Ensure tested 0.2 0.08 0.15 support for platform components use, perform random sampling Flood ND Pole If scour concern, use Periodic maintenance NS NS NS foundation riprap Wind farms EQ motion PGA 1.0g Monopole Use seismically robust Maintenance, obtain 0.1 0.08 0.05 (large event) unit manufacturer testing and certificates Liquefaction ND Monopole Use deep foundations Inspection during 0.2 NS 0.3 foundation installation Wind Design wind Blade Optimize blade Periodic inspection, 0.2 0.1 0.05 70 to 100 configuration report any crack mph Use material with initiation on blades or higher fatigue life connections Conservatism in design Flood N/A (Table continues next page) 195 196 LIFELINES TABLE A.1 Engineering options to improve infrastructure asset resilience in the power sector (continued) Damage Incremental Natural hazard Critical system/component probability cost (including Engineering Quality quality Type Hazard Intensity Component improvement improvement Baseline Improved control) Nuclear EQ motion Large Main Seismic isolation of Testing, inspection, 0.3 0.02 0.05 power events structures, main building, construction plants interior Flexible connections documentation components Seismically rated components; pipes, cable racks, etc. Liquefaction NA Wind NA Flood Large Reactor Improved dike Shutdown drills, 0.1 0.07 0.05 events ground, construction, extreme document review, cooling event flood design including geotechnical, towers, hydrological, and buildings construction documents Substations EQ motion Mw 7 PGA Bushings, Component anchorage, Review all test 0.8 0.3 0.1 0.4g switches, use of seismic documents, ensure circuit components redundancy, spares breakers Liquefaction Mw 7 PGD Switches, Deep foundation Geotechnical report, pile 0.6 NS 0.2 300 mm elevated load testing components Wind Design wind Elevated More robust components Testing, inspection 0.3 0.1 0.2 70 to 100 components mph Flood 2- to 3-ft. Transformers, Elevate components Review construction 0.1 NS 0.1 inundation buildings, reports, inspections ground- mounted equipment Transmission EQ motion Mw 7 PGA T&D systems Use seismic components Periodic inspection 0.02 0.01 0.02 and 0.4g distribution Liquefaction ND Lattice Use deep foundation Construction inspection 0.2 NS 0.15 lines support Wind Design wind Tower Use steel, concrete, or Construction inspection, 0.3 0.07 0.2 70 to 100 composite towers Use tested components mph Use vibration dampers Flood N/A LIST OF ENGINEERING OPTIONS TO INCREASE THE RESILIENCE OF INFRASTRUCTURE ASSETS 197 TABLE A.2 Engineering options to improve infrastructure asset resilience in the water sector Damage Incremental Natural hazard Critical system/component probability cost (including Engineering Quality quality Type Hazard Intensity Component improvement improvement Baseline Improved control) Reservoirs EQ motion PGA 0.6g Embankment Design for higher Drenching, maintenance 0.15 0.05 0.05 (impounding) seismic design forces Liquefaction ND Embankment Restressed concrete Geotechnical report, 0.2 0.02 0.20 piling inspection during construction and pile driving Wind N/A Flood Large event Embankment Design for higher Maintenance, drenching 0.2 0.05 0.05 crest freeboard (taller structure) Reservoirs EQ motion Mw 7 Tank Thicker tanks (ground) Construction inspection, 0.2 0.02 0.05 (storage PGA 0.4g elevated Perform seismic random testing during tanks) support design and use larger erection members and adequate connections (elevated) Liquefaction ND Tank support Use pile foundation Geotechnical testing and 0.4 0.1 0.5 pile inspection Wind Large events Elevated Design for higher wind Keep tank full during 0.2 0.05 0.1 tank force storms Flood N/A Water and EQ motion Mw 7 Pumping Higher threshold Improving anchoring 0.7 0.4 0.15 wastewater PGA 0.4g system seismic design system and introducing treatment seismic protective plants devices Liquefaction ND Sewage Higher threshold for Improving the backfilling 0.7 0.4 0.2 system permanent ground displacement Wind N/A Flood Large event Pumping Elevating Improve construction 0.5 0.2 0.05 system quality Distribution EQ motion Mw 7 Joints Higher threshold in Replace joints with 0.7 0.4 0.2 pipes PGA 0.4g seismic design flexible joints with higher displacement and rotation capacities Liquefaction Large event Joints and Higher threshold for Replace the sections and 0.7 0.4 0.55 sections permanent ground joints to accommodate displacement very large differential displacement and rotation demand Wind N/A Flood Large event Pipelines Higher threshold for Keep the pipes filled 0.2 0.1 0.02 large pipe displacement with water to mitigate buoyancy effects (Table continues next page) 198 LIFELINES TABLE A.2 Engineering options to improve infrastructure asset resilience in the water sector (continued) Damage Incremental Natural hazard Critical system/component probability cost (including Engineering Quality quality Type Hazard Intensity Component improvement improvement Baseline Improved control) Sewage EQ motion MMI VII to VIII Pumping Equipment anchorage Apply higher level of 0.56 0.39 0.25 network (equiv. PGA = station retrofit quality assurance emissaries 0.3g) Liquefaction PGD Buried pipe Soil improvement/ Apply higher level of NA NA 0.55 compaction quality assurance Wind PGWS = 90 WTP building Roof-wall connection Apply higher level of 0.04 0.03 0.15 mph retrofit and Bldg. quality assurance envelopes replacement Flood FID = 3.3 ft. Pumping Elevation and watertight Apply higher level of 0.08 0.01 0.40 station barrier installation quality assurance Water EQ motion PGV 0.5 m/ Canal walls Use reinforced concrete Construction inspection, 0.2 0.05 0.2 conveyance sec liner cylinder testing, rebar systems PGD 0.15 m placement (canals) Liquefaction Based on Canal wall Geomembrane liners, Construction inspection, 0.2 0.01 0.03 small and base soil densification geotechnical testing segment of long canal Wind N/A Flood Large events Gates and Use proper gates, dry Periodic maintenance, 0.1 0.02 0.15 locks channels adjacent construction inspection Drainage EQ motion MMI VII to VIII Drainpipe Drainpipe replacement Apply higher level of NA NA 1.05 systems (PGA or PGV) quality assurance Liquefaction PGD Drainpipe Soil improvement/ Apply higher level of NA NA 0.55 compaction quality assurance Wind N/A Flood N/A TABLE A.3 Engineering options to improve infrastructure asset resilience in the railways sector Damage Incremental Natural hazard Critical system/component probability cost (including Engineering Quality quality Type Hazard Intensity Component improvement improvement Baseline Improved control) Railways EQ motion MMI VII to Bridge pier Pier jacketing retrofit Apply higher level of 0.12 0.05 0.25 (diesel and VIII (equiv. quality assurance electric) PGA = 0.3g) Liquefaction PGD = 12 in. Tracks/ French drainage and Apply higher level of 0.16 0.01 0.45 roadbeds drainpipe installation quality assurance Wind PGWS = 90 Railway Roof-wall connection Apply higher level of 0.04 0.03 0.15 mph stations retrofit and building quality assurance envelopes replacement Flood FID = 3.3 ft. Fuel/DC Elevation and watertight Apply higher level of 0.03 0.01 0.50 substations barrier installation quality assurance LIST OF ENGINEERING OPTIONS TO INCREASE THE RESILIENCE OF INFRASTRUCTURE ASSETS 199 TABLE A.4 Engineering options to improve infrastructure asset resilience in the roadway sector Damage Incremental Natural hazard Critical system/component probability cost (including Engineering Quality quality Type Hazard Intensity Component improvement improvement Baseline Improved control) Highways EQ motion PGD 0.5 m Embankment Provide geogrid Construction inspection, 0.1 0.05 0.1 (on grade) reinforcement use of approved material Liquefaction ND Embankment Soil improvement Geotechnical testing, 0.1 0.05 0.05 construction inspection and testing Wind N/A Flood N/A Landslide ND Road surface Add retaining wall, Construction monitoring 0.2 0.02 0.1 stabilize sloe, shotcrete, soil nails Highway EQ motion Mw 7 Bridge Use CA or Japan seismic Construction inspection, 0.4 0.05 0.1 bridges PGA 0.4g superstructure, design, columns as fuse testing, qualify column, contractors foundation Liquefaction PGD Bridge Use pile foundation Geotechnical testing, 0.3 0.05 0.2 250 mm foundation construction inspection Wind Small events Steel bridge Use details with longer Inspection of welded 0.05 0.01 0.05 members and fatigue life during bridge connections, reduce connections design life section loss by corrosion prevention Flood Large floods Bridge Use riprap Hydrological report, 0.05 0.02 0.05 foundation construction inspections Landslide PGD = 14 in., Bridge Soil improvement Apply higher level of 0.5 0.16 0.15 7 in. foundation quality assurance Secondary EQ motion Mw 7 Road Provide seismic Use earthquake resistance 0.1 0.05 0.05 urban roads PGA 0.4g surface and reinforcement, compact foundations (on grade) underlying the underlying material material Liquefaction Large PGD: Road surface Provide reinforcement Soil improvement, 0.1 0.05 0.05 more than and underlying against large ground avoid areas subjected 0.3 m material displacement vulnerable to liquefaction Wind N/A Flood Large floods Road surface Provide barriers, Construction inspection, 0.1 0.05 0.03 improve drainage testing, qualify contractors Landslide N/A (Table continues next page) 200 LIFELINES TABLE A.4 Engineering options to improve infrastructure asset resilience in the roadway sector (continued) Damage Incremental Natural hazard Critical system/component probability cost (including Engineering Quality quality Type Hazard Intensity Component improvement improvement Baseline Improved control) Urban EQ motion Mw 7 Bridge Use CA or Japan Construction inspection, 0.35 0.04 0.2 (roadway) PGA 0.4g superstructure, seismic design, testing, qualify bridges abutments, columns as fuse contractors footings Liquefaction PGD Bridge H pile or prestressed Geotechnical testing, 0.4 0.1 0.3 250 mm foundation pile foundation construction inspection Wind Small events Connection Reduce dissertation- Inspection of welded 0.1 0.03 0.05 of diaphragms induced fatigue cracking, connections, reduce to steel girders redundant nonfracture section loss by corrosion critical design prevention Flood Large events Pier and Mitigation of local scour, Regular inspection, 0.03 0.02 0.01 abutment use rocks or pier walls construction quality foundations control Landslide N/A Unpaved EQ motion Mw 7 Road surface Provide seismic Use earthquake-resistant 0.1 0.05 0.1 tertiary PGA 0.4g and underlying reinforcement, compact foundations roads material the underlying material Liquefaction Large PGD: Road surface Provide reinforcement Soil improvement, avoid 0.1 0.05 0.05 more than and underlying against large ground areas vulnerable to 0.3 m material displacement liquefaction Wind N/A Flood Large floods Road surface Provide barriers, Maintain the roads 0.1 0.05 0.03 improve drainage Landslide ND Road surface Add retaining wall, Construction monitoring 0.2 0.02 0.05 stabilize slope, shotcrete, soil nails Wooden EQ motion Accelera- Wood bridge Truss strengthening Apply higher level of 0.35 0.03 0.20 bridges tion = 0.4g trusses and connection quality assurance retrofit Liquefaction PGD = 10 in. Bridge Pile addition Apply higher level of 0.44 0.13 0.30 foundation (foundation retrofit) quality assurance Wind Connection Truss Connection retrofit/ Apply higher level of 0.15 0.05 0.10 fatigue connections replacement quality assurance category Flood Flood return Foundation Scour mitigation by Apply higher level of 0.06 0.02 0.03 period (1,000 ground ground strengthening quality assurance to 100 yr.) (riprap, rock, etc.) Landslide PGD = 14 in., Bridge Soil improvement Apply higher level of 0.63 0.25 0.25 7 in. foundation quality assurance Source: Miyamoto International 2019. Note: FID = flood inundation depth; N/A = denotes hazards that are not considered critical for the given infrastructure; ND = designates hazard for which intensity is not defined explicitly; NS = designates small or negligible; NA = specific damage probability is not available; Mw = moment magnitude scale; PGA = peak ground acceleration; PGD = permanent ground deformation. ECO-AUDIT Environmental Benefits Statement The World Bank Group is committed to reducing its environmental foot- print. In support of this commitment, we leverage electronic publishing options and print-on-demand technology, which is located in regional hubs worldwide. Together, these initiatives enable print runs to be lowered and shipping distances decreased, resulting in reduced paper consumption, chemical use, greenhouse gas emissions, and waste. We follow the recommended standards for paper use set by the Green Press Initiative. The majority of our books are printed on Forest Stewardship Council (FSC)–certified paper, with nearly all containing 50–100 percent recycled content. The recycled fiber in our book paper is either unbleached or bleached using totally chlorine-free (TCF), processed chlorine–free (PCF), or enhanced elemental ­ c hlorine–free (EECF) processes. More information about the Bank’s environmental philosophy can be found at http://www.worldbank.org/corporateresponsibility. From serving our most basic needs to enabling our most ambitious ventures in trade and technology, infrastructure services support our well-being and development. Reliable water, sanitation, energy, transport, and telecommunication services are universally considered to be essential for raising and maintaining people’s quality of life. Yet millions of people, especially in low- and middle-income countries, are facing the consequences of unreliable electricity grids, inadequate water and sanitation systems, and overstrained transport networks. From floods and storms to earthquakes and landslides, natural hazards magnify the challenges faced by these fragile systems. This book, Lifelines: The Resilient Infrastructure Opportunity, lays out a framework for understanding infrastructure resilience—the ability of infrastructure systems to function and meet users’ needs during and after a natural shock—and it makes an economic case for building more resilient infrastructure. Building on a wide range of case studies, global empirical analyses, and modeling exercises, Lifelines provides an estimate of the impact of natural hazards on infrastructure. It looks at not only the repair costs but also the consequences for users—from households to global supply chains. It also reviews available options to make infrastructure assets, systems, and users more resilient and better able to cope with natural disasters. Assessing the costs and benefits of these options, the book demonstrates the economic value of investing in more resilient infrastructure, especially in low- and middle-income countries. Lifelines concludes by identifying five obstacles to resilient infrastructure and offering concrete recommendations and specific actions that can be taken by governments, stakeholders, and the international community to improve the quality and adequacy of these essential systems and services, and thereby contribute to more resilient and prosperous societies. SKU 211430