D I S A S T E R R I S K M A N A G E M E N T S E R I E S N O . 4 Understanding the Economic and Financial Impacts of Natural Disasters Charlotte Benson and Edward J. Clay THE WORLD BANK Disaster Risk Management Series Understanding the Economic and Financial Impacts of Natural Disasters Disaster Risk Management Series Understanding the Economic and Financial Impacts of Natural Disasters Charlotte Benson Edward J. Clay © 2004 The International Bank for Reconstruction and Development / The World Bank 1818 H Street, NW Washington, DC 20433 Telephone: 202-473-1000 Internet: www.worldbank.org E-mail: feedback@worldbank.org All rights reserved. 1 2 3 4 07 06 05 04 The findings, interpretations, and conclusions expressed herein are those of the author(s) and do not necessarily reflect the views of the Board of Executive Directors of the World Bank 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. Rights and Permissions The material in this work is copyrighted. Copying and/or transmitting portions or all of this work without permission may be a violation of applicable law. The World Bank encourages dissemination of its work and will normally grant permission promptly. For permission to photocopy or reprint any part of this work, please send a request with complete information to the Copyright Clearance Center, Inc., 222 Rosewood Drive, Danvers, MA 01923, USA, telephone 978-750-8400, fax 978-750-4470, www.copyright.com. All other queries on rights and licenses, including subsidiary rights, should be addressed to the Office of the Publisher, World Bank, 1818 H Street, NW, Washington, DC 20433, USA, fax 202-522-2422, e-mail pubrights@worldbank.org. Cover photo credit: © Reuters NewMedia Inc./CORBIS Library of Congress Cataloging-in-Publication Data has been applied for. ISBN 0-8213-5685-2 Contents Preface vii Abbreviations and Acronyms viii Summary 1 1. Introduction 3 Objectives 3 Selection of Countries and Issues for Investigation 4 Concepts and Definitions 5 Method of Investigation 6 2. Disasters and the Macroeconomy 9 The Dynamic Nature of Vulnerability 9 Overview of the Factors Determining Vulnerability 15 Natural Hazards 16 Economic Structure 17 Stage of Development 18 Prevailing Socioeconomic Conditions 19 The Macroeconomic Impact of Disasters 19 Lessons Learned 26 3. Public Finance and Disasters 29 Background 29 The Broad Fiscal Impact of Disasters 30 Disaggregated Reexamination of Public Finances 31 External Aid 34 Is Reallocation an Appropriate Solution? 35 Risk Reduction Activities 38 Long-Term Policy Consequences of Disasters 39 Lessons Learned 39 4. Information on Natural Hazards and Disaster Reduction 43 Information and Public Action 43 Hazard Information as a Public Good 43 v vi Contents Climatic Forecasting in Southern Africa 44 Tropical Storms 46 Failures in the Provision of Information as a Public Good 47 Findings and Conclusions 49 5. Financing the Cost of Future Disasters 53 Risk Transfer Tools 53 Potential Obstacles 54 Creative Solutions 57 Promoting Mitigation 58 Conclusions 59 6. Findings of the Study and Implications for Policy and Research 61 Findings 61 Policy Implications 63 Directions for Future Research 66 Appendix A. Dominica: Natural Disasters and Economic Development in a Small Island State 69 Appendix B. Bangladesh: Disasters and Public Finance 79 Appendix C. Malawi and Southern Africa: Climatic Variability and Economic Performance 91 Notes 99 References 107 Index 113 Boxes 2.1 Measuring vulnerability 20 2.2 Saying so does not make it so: poverty reduction strategies 21 2.3 Funding rehabilitation: the implications for long-term growth 24 3.1 Fiscal impacts of drought in Sub-Saharan Africa 31 4.1 Evidence-based volcanology: application of Bayes' rule to the situation in Dominica in 1998 49 5.1 Insuring banana growers against disaster: the WINCROP scheme 56 B.1 Uncertainties in postdisaster economic forecasting in Bangladesh 85 Figures 2.1 Dominica: real annual fluctuations in agricultural, nonagricultural, and total GDP, 1978­99 10 2.2 Bangladesh: real annual fluctuations in agricultural, nonagricultural, and total GDP, financial years 1965­2000 12 2.3 Malawi: real annual fluctuations in agricultural, nonagricultural, and total GDP, 1980­2001 13 2.4 Southern Africa: cereal production and El Niño events, 1972­99 15 Preface Disaster prevention and mitigation are integral to devel- Paul Freeman, Rodney Lester, Simon Maxwell, John opment activities. In February 2000, the World Bank's Roberts, Malcolm Smart, and Dirk Willem te Velde. Disaster Management Facility (now called Hazard The study team extends its thanks to Nemat T. Shafik, Management Unit) initiated a three-year study on the World Bank Infrastructure vice president and head of economic and financial consequences of natural disas- network; Maryvonne Plessis-Fraissard, director of the ters, with the support of the U.K. Department for Transport and Urban Development Department (TUD); International Development (DfID), provided through and John Flora, former director of TUD, for supporting its Conflict and Humanitarian Aid Department (CHAD) thisstudy.WealsothankcountrydirectorsCarolineAnstey, under the umbrella of the ProVention Consortium. Orsalia Kalantzopoulos, Darius Mans, Hartwig Schafer, The principal researchers for this three-year study, Fred Temple, and Christine I. Wallich for their support which began in February 2000, were Charlotte Benson and inputs to this report. They also provided detailed and Edward J. Clay of the Overseas Development Institute comments on the individual case studies, which were (ODI), London. Study team members from the World published as part of the Disaster Management Facility's Bank's Disaster Management Facility included Alcira Working Paper Series. The papers in this series can be Kreimer, Margaret Arnold, Jonathan Agwe, Hager Ben- accessed on the Internet at http://www.worldbank.org/ Mahmoud, Maria Eugenia Quintero, and Zoe Trohanis. hazards. The team also thanks World Bank staff who pro- The study consists of a state-of-the art review and vided helpful contributions to the study, including Imti- three country case studies: on Dominica, a small island azuddin Ahmad, Tercan Baysan, Bernard Becq, Sarwat economy (Benson and Clay 2001); on disasters and Chowdhury, Robert Epworth, Arnaud Guinard, Rumana public finances in Bangladesh (Benson and Clay 2002a); Huque, Reazul Islam, Kapil Kapoor, Chingboon Lee, and on climatic variability in southern Africa, with a Ashoka Mody, John Pollner, S.A.M. Rafiquzzaman, Con- country study of Malawi (Clay and others 2003). This stantineSymeonides-Tsatsos,andClaudioVisconti.Finally, synthesis report draws together the new findings and the full cooperation extended by officials of the govern- evidence from the researchers' previous studies and from ments of Bangladesh, Dominica, and Malawi was essen- other relevant literature. tial to the successful completion of the country studies. This report was prepared by Charlotte Benson and They and the many others who provided information Edward J. Clay, with editorial assistance from Alice Baker andadviceareacknowledgedinthecountrystudyreports. on the appendixes. Among those who contributed to There is scope for further work on the economic con- the three country studies were Enrique Blanco de Armas, sequences of natural disasters, and it is hoped that this Louise Bohn, Jim Dempster, P. Dalitso Kabambe, Franklyn report will provoke discussion on both analytical and V. Michael, Clement Peris, Alistair W. Robertson, and policy issues and will stimulate others to undertake fur- Hardwick Tchale. Mavis Clay provided editorial and ther investigations. The authors, of course, accept full bibliographical assistance. responsibility for all errors and omissions in this report. The authors benefited considerably from comments The opinions expressed are those of the authors and on the draft of this report by Willy Aspinall (who also do not necessarily represent the views of the World Bank contributed box 4.1), Stephen Biggs, Hugh Brammer, or the DfID. vii Abbreviations and Acronyms ADMARC Agricultural Development and HIV/AIDS human immunodeficiency Marketing Corporation (Malawi) virus/acquired immune deficiency ADP Annual Development Program syndrome (Bangladesh) IADB Inter-American Development Bank CARICOM Caribbean Community and Common IBRD International Bank for Market Reconstruction and Development CDERA Caribbean Disaster Emergency IDA International Development Response Agency Association (World Bank) CHAD Conflict and Humanitarian Aid IDNDR International Decade for Natural Department (DfID, United Disaster Reduction (United Nations) Kingdom) IFPRI International Food Policy Research CRED Centre for Research on the Institute Epidemiology of Disasters IFRC International Federation of the (Belgium) Red Cross and Red Crescent Societies DfID Department for International IIASA International Institute for Applied Development (United Kingdom) Systems Analysis DMF Disaster Management Facility IMF International Monetary Fund (World Bank) IPCC Intergovernmental Panel on Climate EC European Commission Change EC$ Eastern Caribbean dollars IPG international public good ECCB Eastern Caribbean Central Bank IRI International Research Institute for ECLAC Economic Commission for Latin Climate Prediction America and the Caribbean MFA Multi-Fibre Agreement EM-DAT Emergency Events Database MFI microfinance institution (CRED) MTEF medium-term expenditure ENSO El Niño­Southern Oscillation framework ESCOM Electricity Commission of Malawi NGO nongovernmental organization F$ Fiji dollars NPG national public good FCDI flood control, drainage, and O&M operations and maintenance irrigation program (Bangladesh) OAS Organization of American States FDI foreign direct investment ODA official development assistance GDP gross domestic product ODI Overseas Development Institute GEF Global Environment Facility (United Kingdom) GIS Geographic Information System OECF Overseas Economic Cooperation GNP gross national product Fund (Japan) ix x Abbreviations and Acronyms OECS Organisation of Eastern Caribbean SOE state-owned enterprise States SRU Seismic Research Unit, University of PML probable maximum loss the West Indies, Trinidad PRSP poverty reduction strategy paper Tk taka (Bangladesh currency) R&R relief and rehabilitation UNDP United Nations Development RMSM Revised Minimum Standard Model Programme RPG regional public good UNISDR United Nations International RS Richter scale Strategy for Disaster Reduction SADC Southern African Development WINCROP Windward Islands Crop Community Insurance Ltd. SAP structural adjustment program WMO World Meteorological Organization SARCOF Southern African Regional Climate WTO World Trade Organization Outlook Forum Z$ Zimbabwe dollars Summary The study described here examines the short- and long- is the primary fiscal response to disaster. Disasters term economic and financial impacts of natural disas- have little impact on trends in total aid flows. ters. It relies in part on in-depth case studies of overall sensitivity to natural hazards in the small island econ- omy of Dominica; public finance consequences of dis- Public Policy Implications asters in Bangladesh; and the economic consequences of climatic variability and the use of climatic forecast- A full reassessment of the economic and financial impacts ing in Malawi and southern Africa. Policy implications of a major disaster should be made 18 to 24 months are drawn, and, where appropriate, recommendations after the event. It should be taken into account in review- are made. Finally, directions for future research and ing the affected country's short-term economic per- cooperation are outlined. formance and the assistance strategy for the country. Governments need appropriate risk management strategies for future disasters, including medium-term Economic and Financial Impacts financial planning covering 8 to 10 years. The basis of funding has to be broadened, using a range of insur- Major natural disasters can and do have severe nega- ance and other mechanisms for different layers of loss. tive short-run economic impacts. Disasters also appear Natural hazard risk management should be integrated to have adverse longer-term consequences for economic into longer-term national investment policies and devel- growth, development, and poverty reduction. But neg- opment strategies and appropriately reflected in the allo- ative impacts are not inevitable. cation of financial resources. Vulnerability is changing quickly, especially in coun- High-quality, reliable scientific information is a nec- tries that are experiencing economic transformation-- essary condition for effective disaster risk management. rapid growth, urbanization, and related technical and The international community should support global and social change. In the Caribbean area and in Bangladesh, regional research and information systems on risk. It there is evidence of declining sensitivity to tropical storms should also ensure that there are adequate complementary and floods and increased resilience as a result of economic monitoring and dissemination programs at the national transformation and public measures for disaster reduc- level. Priorities include climatic variability, regional and tion. The largest concentration of high-risk countries-- national flood forecasting, and geophysical hazards. which are increasingly vulnerable to climatic hazards--is in Sub-Saharan Africa. Risks emanating from geophys- icalhazardsneedtobebetterrecognizedinhighlyexposed Economic Research on Natural Disasters urban areas across the world, as the potential costs are rising exponentially with economic development. Vulnerability to natural hazards is determined by a com- Naturaldisasterscausesignificantbudgetarypressures, plex, dynamic set of influences that include the country's with both narrowly fiscal short-term impacts and wider economic structure, stage of development, and prevail- long-term implications for development. Reallocation ing economic and policy conditions. To understand and 1 2 Understanding the Economic and Financial Impacts of Natural Disasters assess the economic consequences of natural hazards The eclectic approach adopted in this study, which and the implications for policy, it is necessary to con- employed largely qualitative methods, is particularly sider the pathways through which different types of useful in exploring the many complex and dynamic hydro-meteorological (climate-related) and geophysi- pathways through which extreme hazard events influ- cal hazard affect an economy, the different risks posed, ence an economy and its financial system, as well as for and the ways in which societies and economies adapt identifying areas and issues where further investigation, to or ignore these threats. including quantification, would be worthwhile. Chapter 1 Introduction Between the 1950s and the 1990s, the reported global to atmospheric and hydrological processes, are recur- cost of natural disasters increased 15-fold. Major nat- rent events, striking a country at infrequent intervals. ural catastrophes in the 1990s caused economic losses Such recurrent shocks can have cumulative effects on estimated at an average US$66 billion per year (in both the rate and the pattern of development (Benson 2002 prices). In 1995, the year of the Kobe earthquake and Clay 2000). By contrast, earthquakes and volcanic in Japan, record losses of about US$178 billion were eruptions, which are very uncommon and better fit the recorded, the equivalent of 0.7 percent of global gross idea of a one-off catastrophic event, accounted for only domestic product (Munich Re 2002). 11 percent of reported natural disasters in the 1990s Such widely cited figures have triggered a growing (IFRC 2002). The potential differences in the economic awareness of the potential damage from natural haz- consequences of these different types of natural hazard ards. There is, however, less recognition of their broader are examined in this study. macroeconomic significance and the problems they These biases and limitations of economic assessment could pose for longer-term development. One reason have severely restricted the information available to pol- is that assessments of the economic impacts of disas- icymakers on the nature and scale of the vulnerability ters have typically concentrated on the most easily meas- of many economies to natural hazards. This lack of infor- ured direct losses--the financial costs of visible physical mation may in turn have contributed to what many see damage. This focus on losses arises from a drive to as a widespread failure to address natural hazards as a meet the short-term humanitarian needs of affected possibly serious threat to sustainable development, and people in the aftermath of a disaster and from pres- a general lack of appreciation of the potentially high sures to rapidly determine replacement investment economic and social returns to disaster reduction. Clearly, requirements and the extent of insured losses. It also how disasters are conceptualized and impacts are assessed reflects the practical difficulties of isolating and meas- within the framework of economic analysis merits fuller uring the indirect and secondary impacts that result as and more systematic review. This investigation seeks to the effects of a disaster shock spread through the econ- contribute to knowledge of these issues and to under- omy. Such impacts may affect, for example, flows of standing how they can be analyzed. goods and services, the balance of payments and gov- ernment budgets, and, ultimately, economic growth, income distribution, and the incidence of poverty. Objectives A further limitation of the existing body of evidence is that most of the relatively few studies to have exam- The broad objectives of the study are to increase ined indirect and secondary effects focus on the impact understanding of the wider economic and financial of a single, recently occurring event. The longer-term, impacts of natural disasters through detailed analysis cumulative effects of a series of disasters on a particu- of the impacts of disasters, the factors determining the lar country's development are more difficult to deter- vulnerability of hazard-prone economies, the opportu- mine and are typically ignored, apart from speculative nities for improving the management of risk, and the comments. Yet in reality, most disasters, being linked hindrances to the adoption of such measures. The study 3 4 Understanding the Economic and Financial Impacts of Natural Disasters focuses primarily on experiences in developing coun- Dominica: Natural Disasters and Economic tries. The findings are intended to contribute to the Development in a Small Island State development of guidelines on the assessment of vul- The study of Dominica explores the overall vulnerabil- nerability to natural hazards from an economic per- ity of an economy to natural hazards. It considers the spective. It was recognized at the outset that the complexity of the factors determining broad sensitivity subject is complex and multifaceted and that the study and the dynamic nature of that sensitivity, focusing on would probably identify many areas that are beyond the disaggregated impacts of natural hazards on differ- its scope but are worthy of future, separate research. ent sectors of the economy. Dominica offers an inter- The investigation adopts a country case study approach esting case that exemplifies the experience of many for exploring economywide disaster impacts. In doing small, open island economies. Such economies face spe- so, it builds on previous research and evaluations by cial disadvantages associated with their size, insularity, the primary investigators, including work on drought and remoteness (Briguglio 1995), which make them in Sub-Saharan Africa (Benson and Clay 1998; Clay and highly sensitive to economic shocks in any form, includ- others 1995; Thomson, Jenden, and Clay 1998) and ing natural hazards. Indeed, they are often perceived studies of the impacts of disasters in five countries in as being among the countries of the world most vul- the Asia and Pacific and the Caribbean regions: Fiji nerable to natural hazards.1 (Benson 1997a), the Philippines (Benson 1997b), Vietnam (Benson 1997c), Zimbabwe (Benson 1998), and Montserrat (Clay and others 1999). Three new coun- Bangladesh: Natural Disasters and Public Finance try studies, on Bangladesh, Dominica, and Malawi, were Disasters can have potentially significant implications completed for this study. This synthesis report thus for public finance, increasing expenditure and simul- reflects the findings and cumulative experience from taneously reducing domestic revenue, in turn resulting undertaking, over a period of 10 years, eight country in increased domestic or external borrowing, substan- studies and regional investigations on the economywide tial alterations to existing investment, and recurrent consequences of natural disasters. expenditure plans or monetary expansion. Natural haz- ards also impose additional pressures on public finances to the extent that governments undertake mitigation Selection of Countries and Issues and preparedness measures. for Investigation Data on aggregate revenue and expenditure typi- cally do not reveal the severity of the budgetary impact The three case study countries were selected to repre- of disasters, as previous work by the principal researchers sent a range of hazard experiences in economies of vary- has clearly shown. The public financial consequences ing size and complexity in different regions of the world, of natural disasters are seldom explored systematically, as well as distinct but complementary methodological except in the narrow context of a single major disaster.2 and policy issues. The study on Dominica, one of sev- After examining these issues for the open, structurally eral highly hazard-prone small island Caribbean states, less complex Dominican economy, it was decided that is an economywide exploration of the impact of disas- this theme should be the central focus of the case study ters (Benson and Clay 2001). That on Bangladesh, a of Bangladesh, with the aim of shedding more light on large, hazard-prone Asian economy, concentrates, in these issues. particular, on public finance (Benson and Clay 2002a). Malawi, a low-income southern African economy, is the Malawi and Southern Africa: Climatic Variability and subject of the third study, which focuses on the use of Economic Performance scientific information, particularly short-term climatic forecasting, in disaster mitigation and the value of this The extreme regionwide drought in southern Africa information from an economywide and sectoral per- in 1991/92 was quickly followed by further droughts in spective (Clay and others 2003). 1993/94andin1994/95.Thesedroughtswereassociated Introduction 5 with an extended and intense El Niño event--a development issues (Apthorpe 1984; Harriss 2002). It reversal of ocean currents across the southern Pacific is therefore necessary, at the beginning of an investiga- that is associated with extreme global climatic effects.3 tion that covers the less explored aspects of natural The droughts had severe impacts on agriculture, as disasters as an economywide or macroeconomic phe- well as wider social and economic consequences. The nomenon, to state clearly what the authors mean by concern engendered by these experiences, as well as the use of specific terms and concepts. The report awareness of the scientific evidence linking events in seeks to adopt widely accepted definitions of key con- southern Africa to global climatic variability and, cepts (hazard, disaster, vulnerability, risk), but this is possibly, to climatic change, created a widespread sen- not always possible where there is no agreed standard timent in favor of strengthening climatic forecasting usage. and promoting the use of the information to support A natural hazard is a geophysical, atmospheric, or food security and improved management of agricul- hydrological event that has potential for causing tural and other renewable natural resources through- harm or loss. Usually, these events are both uncom- out the region. It was also envisaged that climatic mon and extreme, in the perspective of the range of forecasting and information could help improve natural phenomena such as rainfall, tropical storms, resilience to longer-term global climatic change and flooding, and seismic tremor or earthquake. Hence, to the likely associated increase in the frequency and there is a need to determine risk. This is understood severity of extreme events. Recognition of the sever- to be "a combination of the probability, or frequency, of ity of the economic impacts of drought simultaneously occurrence of a defined hazard and the magnitude of heightened interest in taking the risks of climatic shocks the consequences of the occurrence" (Royal Society into account in the management of national economies 1992: 4). and in structural adjustment programs (Benson and A natural disaster is the occurrence of an abnormal Clay 1998). or infrequent hazard that affects vulnerable communi- The third case study therefore focuses on Malawi and ties or geographic areas, causing substantial damage, on the wider southern African area. It reassesses, in the disruption, and perhaps casualties and leaving the affected light of experience during the 1990s, the economic con- communities unable to function normally. From an eco- sequences, on both regional and country scales, of cli- nomic perspective, a disaster implies some combination matic variability, and it examines the status of and of losses, in human, physical, and financial capital, and prospects for climatic research and forecasting as they a reduction in economic activity such as income gener- relate to these levels. It reviews the range of potentially ation, investment, consumption, production, and useful products in the light of recent experience, employment in the "real" economy. There may also be examines meteorological and other institutional capac- severe effects on financial flows such as the revenue and ity to utilize fully the potential of forecasting knowledge expenditure of public and private bodies (Benson and and expertise, and assesses the financing issues posed Clay 1998). The losses in stocks of capital and inven- by strengthening climatic forecasting. tory and the reductions in short-term economic flows are sometimes confounded in reporting the costs of dis- asters.4 Stock losses and short-term flow effects may be Concepts and Definitions so extreme as to alter the medium-to-longer-term tra- jectory or development path of an enterprise, region, Natural disasters are an area of multidisciplinary research or national economy. and policy analysis. A problem of discourse arises because Vulnerability is the potential to suffer harm or loss, basic terms in the language of disaster research and prac- expressed in terms of sensitivity and resilience or of tice seem to be common to the various disciplines but the magnitude of the consequences of the potential in fact often reflect subtle differences in conceptualiza- event.5 The sensitivity of economic behavior to a disas- tion by natural scientists, social scientists, and practi- ter shock is reflected at a macroeconomic or sectoral tioners. This problem of discourse is typical of most level in the deviation of economic aggregates from the 6 Understanding the Economic and Financial Impacts of Natural Disasters trends that were expected without taking the effects of Method of Investigation the event into account. Because economic activity is sensitive to many influences, including other sources Isolation of the economic impacts of natural hazards of shock, in practice it can be difficult to identify pre- from other internal and external factors poses consid- cisely the impacts of a specific disaster or disasters. erable methodological difficulties. The study adopts an The primary objective of our studies has been to iso- eclectic approach, employed in previous work by the late and understand these short- and long-term conse- authors, that involves the construction of a historical quences of natural disasters. narrative of disasters for the case study country or Resilience is the speed of recovery of economic activ- region. Disasters are not treated as "black box" eco- ity, which may involve repair and replacement of lost nomic shocks. Rather, care is taken to establish, through and damaged capital. People seek to cope with shocks close consultation with scientists in relevant fields, within a range of responses that will not jeopardize their the precise nature of each hazard type, including the survival or lifetime aspirations. Communities and formal frequency and characteristics of extreme events. A mix- public and private institutions seek to manage the effects ture of formal quantitative and qualitative analysis is of a shock without jeopardizing their envisaged longer- employed to examine the economic impacts of natu- term plans. ral hazards at an economywide level (Benson 1997a, The disaster management literature commonly dis- 1997b, 1998; Benson and Clay 1998). The quantita- tinguishes between rapid-onset disasters such as storm tive investigations are partial, involving a combina- surges and earthquakes, which cause immediate loss tion of regression analysis, the use of charts to examine and disruption, and slow-onset events, notably drought. movement around trends, and "before-and-after" com- In our empirical investigations of economic conse- parison of disaster impacts and of forecast and actual quences, we have found it useful to distinguish hydro- economic performance. The implied null hypothesis meteorological hazards (atmospheric or climatic hazards is that there is no direct link between disaster shocks and the related riverine and coastal hydrological haz- and the relevant aspect of economic performance. Such ards) from geophysical hazards because of the different analysis cannot always be definitive, but the results at character of the risks involved.6 least provide a basis for further reflection and investi- Hydro-meteorological hazards present threats of vary- gation. If impacts are not apparent at an aggregate level, ing intensity that are usually recognized at a local or the analysis moves on to consider possible effects within national level, and there is consequently some form of the composition of the relevant economic indicator. A adaptation in economic behavior and in the technology qualitative political-economic analysis is also employed in which capital (productive capital, housing and in a complementary way to place quantitative results habitat, and infrastructure) is embodied. The economic within the specific economic and social policy context and the wider social consequences of individual of each case study country. Where similar qualitative hydro-meteorological events appear to be susceptible results repeatedly emerge from previous and current to investigation for most lower- and middle-income studies, this is taken to be preliminary evidence of a developing countries. By contrast, potentially cata- more general finding about the economic consequences strophic geophysical hazards may occur very rarely. Even of natural disasters. in high-risk geographic regions, there may have been The country studies were constrained by the very no extreme event in living memory or perhaps within limited resources and time available and, in some respects, the historical record. Consequently, such hazards pose by substantial data limitations. Moreover, the deliber- quite different problems of risk perception and eco- ately simple methodological approach, which is, after nomic behavior. But a global phenomenon--satellite all, only an extension of the approach typically employed television and increasingly widespread availability of in looking at a single shock, relied heavily on judgment. media information--may be changing perceptions of And, most obviously, it was necessary to select the "major" risk associated with these types of hazard, too. natural hazard events to be included in the analysis. Introduction 7 For each case study, a country visit was made to col- "major" hazard events. Relevant country program offi- late data and conduct interviews with selected current cers at the World Bank in Washington, D.C., were con- and former officials and administrators; representatives tacted and were met, where possible. The case studies of civil society and private sector managers who had also entailed a review of the available official docu- been involved in specific hazard events; and environ- mentation and the recent literature. As noted in the Pref- mental scientists with direct experience of the country. ace, local researchers or researchers in the region Interviewees were also consulted about the selection of contributed to each of the country case studies. Chapter 2 Disasters and the Macroeconomy This chapter explores the overall vulnerability of an Frederick, both in 1979; Hurricane Allen in 1980; Hur- economy to natural hazards. Case study evidence about ricane Hugo, another Category 4 storm, in 1989; the the dynamics of vulnerability leads to a more general cumulative impact of three tropical storms in 1995; and discussion of the sources of vulnerability. The macro- Hurricane Lenny, also Category 4, in 1999. Hurricane economic impacts of disasters are then reviewed in terms David directly hit the island and was extremely devas- of short-term and long-term effects. The impacts of tating, having severe environmental and demographic disasters on development strategies are considered, and, consequences. Landslides triggered by storms are finally, lessons are drawn. common in Dominica and can cause substantial eco- nomic damage, as well as potential loss of life. There are geophysical hazards, too. Although there The Dynamic Nature of Vulnerability has been only one volcanic eruption in Dominica's recorded history, the island is now experiencing a period The vulnerability of an economy to natural hazards of increased seismic activity. The risk of volcanic activ- depends on a complex set of influences. This section ity remains relatively high, particularly in the south of briefly presents evidence from the three country cases, the island, where the capital city and most of the key which typify more general country situations. The cases infrastructure are located. highlight the dynamic, rapidly changing sensitivity of Dominica's small, very open economy still relies heav- economies to natural hazards in the present era, focus- ily on a single export crop, bananas, which repre- ing on the interaction of developmental, economic, and sented a third of total merchandise export earnings in societal factors with natural hazards. Some common 1997. Although agriculture's share of gross domestic influences are at work, along with country- and region- product (GDP) fell from 37 percent in 1977­78 to 20 specific factors.7 In addition, in the longer term, climatic percent in 1997­98, it remains the main productive change is altering the frequency and intensity of hazard sector and is the principal source of livelihoods. Despite events, with implications for the scale and nature of vul- some growth in the nonagricultural private sector nerability. This is an issue that the case study findings since the mid-1970s, other private sector activity remains suggest should be explored separately. small. Between 1997 and 1998, manufacturing output rose from 3.9 to 8.2 percent of GDP, and there has been promising growth in the offshore financial services indus- Dominica try. Along with tourism, which by the late 1990s Dominica, a small Caribbean island, is susceptible to a accounted for an estimated 35 percent of external wide range of natural hazards. The most common, most earnings, these activities have helped meet the sub- probable, and historically most significant are extreme stantial deficit in the external visible trade account. climatic events: tropical storms and hurricanes. The The close association between the fluctuations in series of disasters since 1978 includes Hurricane David, Dominica's banana exports, the country's agricultural, an extreme Category 4 storm with sustained winds in nonagricultural, and total GDP, and the incidence of excess of 210 kilometers per hour, and Hurricane severe storms demonstrates the substantial impact that 9 10 Understanding the Economic and Financial Impacts of Natural Disasters Figure 2.1 Dominica: real annual fluctuations in agricultural, nonagricultural, and total GDP, 1978­99 Hurricanes Tropical Tropical David and Hurricane Tropical Storm Storm Three Storm Hurricane Frederick Allen Klaus Hurricane Hugo Debbie Storms Hortense Lenny 30 20 10 cent) (per 0 change -10 -on-year earY-20 -30 -40 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 Total GDP Nonagricultural GDP Agricultural GDP Source: Benson and Clay (2001); reprinted from Kreimer, Arnold, and Carlin (2003). natural hazards have had on the island's economic per- of bananas in total agricultural output because banana formance since 1978, as shown in figure 2.1. The analy- cultivation offered a fast, low-investment means of restor- sis also suggests, however, that the economy is becoming ing agricultural livelihoods in an assured export market.9 relatively less sensitive to extreme climatic events. These The rapid recovery in export production after Hurricane shifts in vulnerability to natural hazards are related both Hugo in 1989 again demonstrated the resilience of to increasing development and capital investment in the the banana economy. In this case, the compulsory island and to changes in the structure and composition WINCROP banana crop insurance scheme (described of economic activity. The economy was most vulnera- inbox5.1inchapter5),whichwasintroducedin1987­88 ble to extreme climatic events in the years 1975­85, by the banana marketing boards of four Windward Island shortly before and after Independence in 1978.8 states, helped encourage replanting of bananas by Beginning in the 1950s, bananas, grown largely by offering partial financial protection in the event of a dis- smallholders, had progressively displaced plantation aster. The dominance of bananas in Dominica, and of tree crops as the principal commodity exported to the similar monocrop agricultural sectors in other small United Kingdom and then to the European Union under island economies, exemplifies a progressive adaptation a preferential access agreement. These changes in the to a specific external economic environment and is often type and structure of production increased the overall accompanied by institutional innovation. vulnerability of both the agricultural sector and the The wider economy's vulnerability to natural hazards wider economy to natural hazards. has changed over the past two decades as a consequence Hurricane David and, in rapid succession, Hurri- of changes in the sectoral composition of GDP--a devel- canes Frederick and Allen demonstrated that vulnera- opment accelerated by the World Trade Organization bility, causing severe damage to banana plantings. Yet (WTO) process. From the mid-1990s on, external fac- the hurricanes led directly to an increase in the share tors resulted in a decline in export-oriented banana Disasters and the Macroeconomy 11 production, as real prices fell and guaranteed prefer- from multiple forms of natural hazard. Riverine floods, ential access to the European market ended. Counter- tropical cyclones (sometimes accompanied by devas- intuitively, the resulting more diversified agricultural tating storm surges), flash flooding, erosion, and drought sector is more sensitive to both natural hazards and have caused severe economic and social disruption other risks. But agriculture's share of GDP has been and considerable loss of life in recent decades. Fur- declining; by 1997 it had dropped to only 19 percent, thermore, Bangladesh is in a zone of very high seismic half its 1977 level, while manufacturing, tourism, and activity. financial services grew and increased their share of GDP. A decade of severe disasters began in the mid- The services sectors are less sensitive to anything short 1960s. In November 1970 a catastrophic cyclone killed of a catastrophic event, such as Hurricane David, and over 300,000 people, and in 1971 came the War of Inde- so their growth implies a reduction in the vulnerabil- pendence and its aftermath, when 12 million people ity of the economy as a whole. were displaced. These events resulted in massive damage The development of the island's infrastructure shows to infrastructure and institutional disruption. The decade how long-term changes in vulnerability are linked to culminated in 1974­75 with a famine linked to extreme overall levels of development and to changes in the floods, hyperinflation, and a bloody political crisis. These structure and composition of economic activity. Between events created a worldwide perception in the mid-1970s 1950 and 1978, Dominica was transformed from an of Bangladesh as not just a disaster-prone country but, underdeveloped plantation-cum-subsistence colony into in the insensitive words of the U.S. secretary of state at an independent, middle-income economy. Key to this the time, a nonviable "basket case." achievement was rapid development of infrastructure. With no further major disasters, the Bangladesh econ- Because of the severe financial constraints, this devel- omy recovered rapidly in the late 1970s. Annual growth opment took place at the lowest possible construction of per capita GDP averaged 1.7 percent in the 1980s costs. The investments followed more than 20 years and 3.3 percent in the 1990s; the latter rate reflected without any major hurricane impacts. As a result, ade- both higher GDP growth and declining population quate disaster mitigation was not built into the con- growth. At the same time, the structural composition struction design, and this omission had devastating of the economy has changed: agriculture's share of consequences when Hurricane David struck.10 All the GDP has declined, while the industrial and services sec- key infrastructure systems were devastated. Except for tors have expanded, resulting in a sharp shift in the airports, these systems were again partly disrupted by composition of the country's exports. A gradual process Hurricane Lenny in 1999. Their vulnerability to natu- of structural adjustment and trade liberalization, along ral hazards now varies, reflecting the amount of hazard with more disciplined monetary management, resulted mitigation investment that has taken place and the asso- in the 1990s in an inflation rate that stayed in the ciated practical and funding issues.11 single digits and an annual current account deficit below Dominica is part of the Eastern Caribbean dollar 2.5 percent of GDP. The reforms have also helped increase (EC$) area. The currency is carefully and conservatively private sector development and foreign direct invest- managed by the Eastern Caribbean Central Bank (ECCB). ment (FDI). Fiscal policy has been less successful: the That framework of monetary stability reduces financial country has had large fiscal deficits, a low tax-to-GDP uncertainty for the private sector and lessens the poten- ratio, and relatively poor-quality public spending. tial destabilizing financial impacts of a disaster shock A simple assessment of the sensitivity of Bangladesh's (see chapter 3). economic performance to major disasters, as measured by fluctuations in GDP and in growth rates of agricul- tural and nonagricultural sector products, highlights Bangladesh some key issues: Most of Bangladesh's densely settled population of 130 · In the period 1965­75, extreme volatility in the still million people lives in the delta of the great Ganges largely agricultural economy was clearly linked to and Brahmaputra river systems and is at significant risk catastrophic natural disasters. 12 Understanding the Economic and Financial Impacts of Natural Disasters · With the notable exception of the 1998 floods, major the external and domestic grain trade have also played disasters have led to downturns in the agricultural a role. As Bangladesh approached self-sufficiency in rice, sector's annual rate of growth. internal prices for this national staple displayed reduced · The short-term impact of disasters on the nonagri- seasonal volatility and moved closer to import parity cultural sector is much less significant, but the longer- price levels with the liberalization of the grain import term impacts of disasters are not reflected in interannual trade. After the floods of 1998, large-scale private sector fluctuations. If resources are diverted from produc- imports covered the greater part of the temporary food tive investments to disaster response, the pace and gap, limiting pressures on prices and public finance (del nature of development will be adversely affected. Ninno and others 2001). · The sensitivity to natural hazards of both the agri- Investment in structural flood control has been another cultural and the nonagricultural components of GDP factor contributing to increased resilience. Urbanization appears to be declining over time, suggesting is rapidly creating large urban and periurban zones, and greater resilience (figure 2.2). the capital, Dhaka, is becoming a sprawling, minimally The improvement in the economy's resilience is partly planned megacity with weak, overstretched infrastruc- attributable to structural change in the agricultural sector. ture. However, since the severe floods of the late 1980s, Following the 1987 and 1988 floods, a relaxation of there has been a de facto shift of flood control invest- restrictions on private agricultural investment and on ment and operations and maintenance (O&M)--the imports of equipment--initially, to encourage recovery-- main area of public expenditure on disaster mitigation-- was associated with a rapid expansion of dry-season from rural and agricultural to urban and industrial (winter) irrigated rice, displacing highly flood-prone protection. These changes in priority have met with deep-water rice and jute and carrying a much lower some success. The 1998 floods, of longer duration and risk. Increased rice production and liberalization of with higher river levels than those of 1987 and 1988, Figure 2.2 Bangladesh: real annual fluctuations in agricultural, nonagricultural, and total GDP, financial years 1965­2000 Floods, 1988 Floods, Cyclone, Floods, Drought, Flood, Floods, Cyclone, Drought, Floods, 1966 1970 1974 1979 1984 1987 1991 1994 1998 25 War of 20 Indep- endence 15 and cent) after- (per 10 math 5 change 0 -5 -on-year earY-10 -15 -20 66/ 67/ 68/ 69/ 70/ 71/ 72/ 73/ 74/ 75/ 76/ 77/ 78/ 79/ 80/ 81/ 82/ 83/ 84/ 85/ 86/ 87/ 88/ 89/ 90/ 91/ 92/ 93/ 94/ 95/ 96/ 97/ 98/ 99/ 00 /20 1965 1966 1967 1968 1969 1970 1971 1972 1973 1974 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 Total GDP Agricultural GDP Nonagricultural GDP Source: Benson and Clay (2002a); reprinted from Kreimer, Arnold, and Carlin (2003). Disasters and the Macroeconomy 13 did not severely affect the Greater Dhaka metropolitan financial stability in the aftermath of subsequent disas- area or some secondary towns that had received enhanced ters. Remittances by Bangladeshis working abroad have protection. played an important role in financing economic Changes in the composition of productive activity growth and in providing postdisaster financial support; have also altered the vulnerability of the economy. Rapidly these remittances increased by 18 percent in the finan- expanding, export-oriented garment manufacture has cial year that included the 1998 floods. Finally, Bangladesh been the primary motor of export growth, as inward has been a leader in developing microfinance for the FDI and some local industrialists exploited the trading rural poor and, more recently, the urban poor. Micro- niche offered by the Multi-Fibre Agreement (MFA). During finance had a significant, although still limited, role in the 1998 floods, there was some disruption of supply enabling the poor to cope with the costs of the 1998 and export chains, but the industry, which was largely floods (del Ninno and others 2001). The central bank based in urban zones less affected by flooding, proved was able to protect this important financial sector through resilient. Again, however, it appears that the risks have massive refinancing. changed rather than being simply reduced. The indus- try's markets are far from assured and could be lost if there Malawi were a major disaster-related disruption. Manufacturing in coastal Chittagong is potentially exposed to an extreme On current evidence, since around 1999 some countries cyclone and storm surge, such as the one in 1991. in southern Africa have experienced increased economic Building standards in new industrial and commercial volatility linked with climatic variability (figure 2.3). developments (with short life expectancies) and in rap- This apparent increase in vulnerability has occurred idly expanding housing largely ignore seismic hazard. during a period of complex, interacting developments The financial system has seen some major develop- in the region. Some of these developments have been ments too, including important innovations, that positive--for example, the political reintegration of again have strengthened resilience to natural hazards. South Africa and the end of the conflict in Mozambique. After the chaotic hyperinflation that contributed to the Others are negative, such as the increasing problems famine of 1974, the government has maintained relative with governance in Malawi, Zambia, and Zimbabwe and Figure 2.3 Malawi: real annual fluctuations in agricultural, nonagricultural, and total GDP, 1980­2001 Drought Drought 60 50 40 30 (percent) 20 change 10 0 -10 Year-on-year-20 -30 -40 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 Total GDP Agricultural GDP Nonagricultural GDP Source: Clay and others (2003); reprinted from Kreimer, Arnold, and Carlin (2003). 14 Understanding the Economic and Financial Impacts of Natural Disasters the HIV/AIDS epidemic, which undermines capacity At least six factors are contributing to increasing eco- to cope with shocks. Malawi provides an illustration of nomic sensitivity in Malawi: these trends. · Unsustainable agricultural practices. The stagnation in Malawi, a small, landlocked country, had a popula- cereal production over more than two decades is a tion of 10.8 million in 2000. It is one of the poorest problem throughout southern Africa (figure 2.4). It countries in Africa, with per capita GDP of US$170 in has been linked with failure to follow cropping pat- 2000. Health and social indicators are among the lowest terns that maintain nutrient levels and to compen- anywhere, and Malawi is one of the countries most sate for lost nutrients with increased fertilizer severely affected by HIV/AIDS. The loss of human cap- applications. Demographic pressures are contribut- ital and the prevalence of ill health among the eco- ing to the decrease in the size of smallholdings, nomically active population are probably making the but other factors must be invoked to explain why country more disaster-prone. smallholder farmers are unable to address technical Malawi still has a largely rural economy; 89 percent constraints. of the economically active population is classified as · Structural change in agriculture. Structural change has rural. Agriculture accounted for about 40 percent of been brought about by deliberate land redistribution GDP in 2000, compared with 44 percent in 1980. The and by economic processes, both of which are influ- share of agriculture in GDP, which had been declining, enced by policy. The decades-long marginalization rose again in the 1990s as the industrial sector stagnated of the small farmer and the switch in the 1990s toward and the public services sector contracted. Export earn- encouraging small farmers to take up tobacco pro- ings are dominated by agricultural commodities--largely, duction have increased volatility. Meanwhile, attempts rainfed tobacco--making the economy sensitive to cli- to establish a viable credit system, input supply, and matic variability and commodity price shocks. supportive marketing structure for smaller produc- Despite internal liberalization and reductions in tar- ers have not been fully effective. iffs, the economy has become relatively less open over · Institutional weaknesses in agriculture. Institutional time. Exports as a proportion of GDP declined from 28 weaknesses reflect the shortcomings of many sectoral percent in 1980 to 24 percent in 2000, and imports as and structural adjustment programs. Previously dom- a share of GDP decreased from 43 to 40 percent. inant or monopolistic parastatal marketing, credit, The main source of natural hazards is climatic vari- and input supply organizations have been weakened, ability. The principal food staple, rainfed maize, which but effective and efficient commercial enterprises accounts for over 70 percent of energy intake, is extremely have not emerged to fill the gap. The regional crisis sensitive not just to drought or low rainfall but also to of 2002 confirms that too little has been achieved. erratic rainfall during the growing season and to · Political instability. Since the early 1990s, political abnormally high rainfall. There were only two clearly instability and governance problems have weak- defined droughts in the 20th century--one associated ened the government's capacity to manage the fiscal with the famine of 1949 and the other in 1991/92, when and monetary aspects of shocks (see chapter 3). maize production fell by around 60 percent. But relatively · Short-term variability in external aid levels. Donors' unfavorable conditions such as the sparse and erratic country-specific policies--which are influenced by rainfall in 1993/94, the extremely high rainfall in 2000/01, political and governance issues, as well as by directly and locally erratic rains in 2001/02 pose increased food economic and humanitarian considerations--have security threats and wider economic threats to a more been an important factor in the volatility of public vulnerable, less resilient economy. finances. Riverine flooding is an annual, relatively predictable · The effects of HIV/AIDS on human resources. The effects hazard in the southern districts, where population den- of HIV/AIDS are insidious, and although much dis- sity is lower. Even in 2001, flooding did not have a wide- cussed, they are only gradually being understood, spread, catastrophic impact. There are apparently no quantified, and seriously addressed. The loss of human other significant natural hazards. capital and the incidence of ill health among the Disasters and the Macroeconomy 15 Figure 2.4 Southern Africa: cereal production and El Niño events, 1972­99 El Niño El Niño El Niño El Niño El Niño El Niño El Niño 30 200 25 150 tons 20 100 (percent) metric 15 50 of change 10 0 Millions Year-on-year 5 -50 0 -100 1972 1973 1974 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 Cereal production Growth in cereal production Source: Clay and others (2003). economically active population are undermining and climatic change. In Bangladesh and Dominica, sen- coping strategies and making the country more dis- sitivity to climate-related hazards seems to have aster-prone (Haacker 2002). peaked in the 1970s, but Malawi and some other south- Extreme events such as the 1991/92 drought in ern African economies are currently showing increas- southern Africa (see figure 2.4) and the floods in Mozam- ing sensitivity. By contrast, risks emanating from bique in 2000/01 frequently evoke warnings about the geophysical hazards appear to have risen with urban- effects of climatic change. There is as yet no conclusive ization and with the growth of the secondary (indus- evidence that Malawi, southern Africa as a whole, or trial) and tertiary (services) sectors in both Bangladesh other regions of Sub-Saharan Africa are experiencing and Dominica. either more frequent extreme events or long-term arid- ification (Hulme and others 2001), but both are antic- ipated for parts of southern Africa as a consequence of Overview of the Factors climatic change (IPCC 2001; Fischer, Shah, and Van Determining Vulnerability Velthuizen 2002). Economic vulnerability is not a static condition that reflects location-specific environmental hazards. The Comparison of the Experiences scale and nature of the economic impacts of a natural of the Case Study Countries hazard event depend, as well, on influences that are time- The three case study countries demonstrate striking con- specific. The country studies, and evidence from earlier trasts, not just in their increasing or decreasing vulner- case studies, highlight five basic factors that determine ability but also in the changing forms of vulnerability. broad macroeconomic vulnerability to natural hazards: These developments belie somewhat simplistic notions · The type of natural hazard of general and rapidly increasing vulnerability to nat- · The overall structure of an economy, including nat- ural hazards associated with global economic growth ural resource endowments 16 Understanding the Economic and Financial Impacts of Natural Disasters · The geographic size of a country Natural Hazards · The country's income level and stage of development · Prevailing socioeconomic conditions, including the The inductive nature of case study investigation gives policy environment and the state of the economy. rise to preliminary findings and hypotheses that need Each of these influences is considered in greater detail to be more widely explored. The authors' earlier study in the sections that follow. of drought in Sub-Saharan Africa (Benson and Clay There are also other factors affecting vulnerability. 1998) suggested the hypothesis that vulnerability is Vulnerability is time-dependent; the country's stage of linked in a nonlinear way with development from a sim- socioeconomic development matters, as does its state pler economic structure to a more complex one char- of technical and scientific advancement. For example, acterized by increasing intersectoral linkages and since the early 1990s, considerable advances have integration of financial flows. Further investigations, been made in seasonal climatic forecasting for south- reported in this study, lead to a second hypothesis: that ern Africa and other regions of Sub-Saharan Africa. This the changing patterns of economic vulnerability differ information can be used to inform private and public for two broad types of hazard--hydro-meteorological decisions on the management of water resources, the (drought, flood, cyclones, and hurricanes) and geo- choice of crops, and the level of grain exports and imports, physical. If that hypothesis is provisionally accepted, it and these decisions in turn alter the relationship between will have important implications for further investiga- climatic variability and economic performance. tions and for disaster risk reduction policy. The application of technical and scientific develop- ments in economic activity can affect vulnerability. For example, flood-tolerant cultivars used in deep-water Hydro-meteorological Hazards rice cultivation in South and Southeast Asia have grad- ually been displaced by shorter-stemmed cultivars that Climatic variability and drought. Both abnormally low require more controlled water management (often, or erratic rainfall (usually characterized as drought) and through irrigation) but also permit more intensive pro- abnormally high rainfall are likely to affect agricultural duction. In Bangladesh this intensification, which is performance adversely. The effects of abnormally high associated with a switch to dry-season, irrigated, high- rainfall may be less severe but are still serious for a highly input rice, has reduced overall variability in crop pro- vulnerable rural economy. Drought, in particular, can duction and vulnerability to climatic hazards. cause heavy crop and livestock losses over wide areas, As discussed in the country studies, disaster man- often affecting several countries simultaneously (the agement, narrowly defined in terms of specific mitiga- covariance aspect of disaster risk), as happened in south- tion, preparedness, and postdisaster responses, also ern Africa in the early 1990s. Extreme climatic events contributes to determining the level and nature of may persist (auto or serial correlation); in the Sahel, there vulnerability. is a significant risk of a succession of very dry years. In Another critical factor is environmental change. In southeastern Africa, extreme events occurring within a Bangladesh the destruction of the Sundarbans mangrove well-established, long quasi cycle of drier and wetter forests may affect the impact of cyclones. Upper ripar- periods lasting approximately a decade may be ampli- ian water management and use may alter the risks of fied (as illustrated by drought in the early 1990s and extreme flooding and the consequences of drought. floods at the turn of the millennium), or they may be Deforestation also probably increases the risks of land- dampened, as with drought in the 1970s. Such quasi- slides, which could be triggered by climatic or seismic cyclical phenomena, which modify the potential impact events, as seen in Dominica. In Malawi farming prac- of an event considered in isolation, have implications tices and social pressures on wasteland and forestland not only for agriculture but also for other water- accelerate soil erosion and increase sensitivity to cli- related, hydrologically sensitive sectors of an economy, matic extremes. Global climatic change could also such as hydroelectricity and domestic water supply alter the frequency and scale of climatic hazards. (Hulme and others 2001). Disasters and the Macroeconomy 17 Riverine floods. Abnormally severe flooding is likely to depending on their intensity. The droughts in Malawi and damage infrastructure and productive capacity, as well the extreme floods in Bangladesh during the 1980s and as directly reduce output, particularly by destroying 1990s, which had clear macroeconomic impacts, were standing crops and by disrupting economic and social also 5- to 50-year events, according to level of severity. activity. These effects can be widespread, as in Bangladesh, The recurrent nature of hazards leads to adaptations in or very restricted, as in Malawi. economic and social activities, such as agriculture, housing, and water supply, at the micro and sectoral levels. Tropical cyclones and hurricanes. Tropical storms pose The historical climatological and hydrological records a considerable threat to human life, especially when asso- also allow formal assignment of risks within probability ciated with storm surges, and can have devastating impacts bands that can be taken into account in larger-scale public on the productive economy. The economic impacts may and commercial investment and in production decisions. be less widespread than those of drought or riverine flood, These risks, however, are nonrandom and are potentially but the storms can leave a path of destruction and dis- subject to quasi-cyclical and secular change.12 Conse- ruption across a whole region, as in the case of hurricanes quently, experiences based on (climatically) relatively in the Caribbean. Storm impacts are likely to have a local- short periods may be profoundly misleading guides to ized impact in larger economies (coastal Bangladesh and the formation of expectations.13 the Philippines) but to be overwhelmingly devastating By contrast, from the viewpoint of most public and for smaller economies (Dominica, Fiji, and Montserrat). private investment decisionmaking, geophysical hazards Severe storms are likely to be associated with and are to be regarded as random, stochastic events of uncer- intensify localized hazards such as flash flooding and tain and mostly low probability. Extreme geophysical landslides. This happened with Hurricane David in 1979 events, with the potential to cause severe damage and in Dominica and Hurricane Mitch in 1998 in Central disruption, are very rare--a 1 percent risk or less in any America (IFRC 1999). year. Even in relatively high-risk zones, the probability of there having been disastrous events within living Geophysical Hazards memory or even in the historical record is low. Conse- quently, until the recent advent of rapid media dissem- Earthquakes can cause widespread destruction of infra- ination of information about events elsewhere, risks had structure and other productive capacity over relatively little effect on private or even public decisions about large areas. These events have little impact on standing the location of activities or about construction standards crops, except for localized losses resulting from land- for the built environment. Only formal regulation or risk slides. The greatest risk of catastrophic macroeco- assessments that are required for internationally funded nomic consequences is when the event occurs in a major investments are likely to introduce risk reduction into urban center or in the metropolis (e.g., Tokyo in economic decisions. But on a global scale, the disaster- 1923). The possibility of such an extreme outcome is, related costs of geophysical hazards are rising rapidly, as with all hazards, in part a function of the economy's and in countries with a significant risk, the potential cost size; thus, a single volcanic eruption completely dis- is growing exponentially with economic development. rupted Montserrat's economy. Volcanic eruptions and tsunamis also usually have localized direct impacts. Economic Structure Geophysical and Climate-Related Hazards Compared The interaction between the types of natural hazard risk The area of danger in a geophysical event is usually to which a country is exposed and the basic structure more restricted than for the most extreme climatic events. of its economy at a particular moment in time plays a The other important difference between climate-related significant role in determining broader macroeconomic and geophysical hazards is the form of risk associated vulnerability. The economic structure is reflected in the with events likely to cause severe economic impacts. In relative importance of the various sectors and subsectors; Dominica damaging storms have been 5- to 50-year events, in patterns of ownership and systems of production; 18 Understanding the Economic and Financial Impacts of Natural Disasters in intersectoral linkages; and in the role and nature of effects from the immediately affected regions or sectors productive capital. through the rest of the economy may be fairly limited, Both the Dominica and Bangladesh cases show how and much of the cost of relief and rehabilitation may reductions in broad macroeconomic vulnerability are be met through external grants and concessional assis- related to structural change--specifically, to a relative tance. Nevertheless, the physical impact of a disaster decline in agriculture, commonly the most vulnerable can be severe, causing widespread destruction and some- sector. Even within the agricultural sector, however, var- times high loss of life, and disasters can exacerbate exist- ious factors work to determine the precise nature and ing problems of indebtedness and poverty. level of vulnerability. These include capacity to recover But economic growth and development per se may from a hazard event and the degree of resilience. Banana not reduce vulnerability. In Bangladesh and Dominica, producers, for example, can suffer enormous damage for instance, development is altering sensitivity to nat- as a consequence of disasters, but their capacity to recover ural hazards rather than reducing it. Poor and socially has been strong, as is seen in the case of Dominica. The disadvantaged groups can become differentially vul- relative importance of different socioeconomic groups nerable.16 Although this much-researched subject is out- in production can also significantly influence vulnera- side the scope of this study, its importance should be bility, as in Malawi and Zimbabwe (Benson 1998). fully acknowledged. Similar patterns may be mirrored Dominica and Zimbabwe provide contrasting exam- at a macroeconomic level. ples of the importance of intersectoral linkages in deter- An economy at an intermediate stage of develop- mining hazard sensitivity.14 Some countries have a ment is typically more integrated, sectorally and geo- high degree of dualism, with a large capital-intensive graphically, than a simple one, and this increases the extractive sector that features significantly in the trade multiplier effects of adverse performance in a particu- account but is weakly linked with other sectors of the lar sector or region. For example, climatic hazards may economy. The effects of drought on the macroeconomy affect the growing manufacturing sector, as well as the and on the trade account were modest in Zambia in agricultural and livestock sectors, particularly where, the 1980s, before the collapse of copper mining, and as is commonly the case, the initial growth of the in Botswana and Namibia during the 1990s.15 manufacturing sector is based primarily on agropro- cessing. Meanwhile, the government is likely to meet a larger share of the costs of relief and rehabilitation Stage of Development efforts itself rather than rely almost entirely on inter- national assistance (see chapter 3). In such an econ- Economic development is widely understood as a process omy, the financial sector is also likely to have a more of increasing complexity that is reflected in the grow- important role in shaping the impact of a natural dis- ing proportion of GDP accounted for by the secondary aster. Intermediate economies typically have more devel- and tertiary sectors of the economy. The stage of devel- oped economywide financial systems for the flow of opment of an economy, measured in terms of the degree funds, including small-scale private savings and of sectoral, geographic, and financial integration, transfers, and this diffuses the impact of disasters levels of economic specialization, and government rev- more widely. For example, in Zimbabwe following the enue-raising capabilities, is likely to influence vulner- 1991/92 drought, the transfer of remittances from urban ability to natural hazards (Benson and Clay 1998). to rural regions was facilitated by the well-articulated Least-developed or simple, less complex economies system for small savings. These transfers not only mit- are typically perceived as most vulnerable, even igated the impact of the drought in rural areas but also though their absolute losses as a consequence of a par- spread the effects more widely (Hicks 1993). ticular disaster may be small relative to the levels reported In the later stages of development, the relative scale in developed countries. It is true that because of weak of the economic impacts of disasters is likely to decline intersectoral linkages, a high degree of self-provisioning, again, as is suggested by evidence on the reported and, often, poor transport infrastructure, the multiplier costs of disaster damage (e.g., Alexander 1997) and Disasters and the Macroeconomy 19 the effects of drought shocks on GDP (Benson and 2000.) Such factors are contingent, reflecting a range of Clay 1998). In part, this relationship reflects the decline dynamic influences. A probably incomplete list includes: of the share in GDP of the particularly hazard-vulnerable · Domestic macroeconomic policies such as stabiliza- agricultural sector, and its decreasing importance as a tion or structural economic reform programs. source of employment, a provider of inputs to other sec- · Medium-term economic and social strategies such as tors, and an end-user. More developed economies are poverty reduction (see box 2.2). also typically more open and have fewer foreign exchange · Domesticsectoralpoliciessuchasthoseconcernedwith constraints, implying that any disaster-related increases food marketing, foreign exchange management, or in imports will not displace normal imports. Other stockpiling of cereals--although there is no clear pat- factors that contribute to lower vulnerability include tern relating to the implications of particular policies.17 higher investment in risk reduction, improved envi- · Deliberatechangesinpolicyinresponsetoadisaster-- ronmental management, and lower levels of poverty. forinstance,tocontrolinflation,encouragereinvestment, Moreover, a larger share of economic assets is likely to or generate revenue to meet the costs of rehabilitation. be held by the private sector and adequately insured · The external policy environment, to the extent that against disaster, and a higher proportion of damage sus- it influences the pattern of productive activities and tained by individual households will be covered by thus affects underlying vulnerability. In Dominica insurance. Thus, the scale and cost of relief and reha- banana production has been encouraged by prefer- bilitation programs will be limited and less likely to ential trade arrangements. Malawi is experiencing necessitate a substantial increase in government domes- deindustrialization as South Africa is reintegrated tic or external borrowing. Despite all this, a small seg- into the regional economy. ment of the affected population may be severely affected · Coincidental fluctuations in primary export and by loss of income, assets, and savings. import prices (e.g., of cereals or oil), which may lessen Within this broad framework, very small island or exacerbate impacts on the balance of payments economies represent a special case--a point that is illus- and inflation. trated in case studies and is underlined by recent attempts · A country's significance in specific export markets. to develop national indices of vulnerability (see box 2.1). The shift from agricultural to manufacturing exports, A cyclone (Hurricane David in Dominica) or a volcanic and thus to an apparently less hazard-vulnerable form eruption (Montserrat) whose impact would be relatively of economic activity, may not in fact have reduced localized in a larger country can be catastrophic to a small the potential sensitivity to disasters of Bangladesh's island country--it can disrupt the whole economy; destroy export earnings.18 much of the transport, power, and communications · The timing and nature of other shocks, especially networks, as well as productive capacity and social conflict--for example, the independence struggle infrastructure; and precipitate an exodus of human cap- in Bangladesh and the war in Mozambique that desta- ital. A loss that "would be merely a local transfer in a bilized Malawi's transport links and brought an influx larger economy"--for example, the temporary or per- of refugees. manent displacement of markets for a country's outputs-- · The incidence of health hazards. The HIV/AIDS pan- could be a devastating setback to a small island economy demic and the resurgence of malaria and tuberculo- (Handmer and Thompson 1997: 15). sis are undermining coping strategies, eroding human capital, and placing potentially considerable strain on public finances in southern and eastern Africa. Prevailing Socioeconomic Conditions The Macroeconomic Impact of Disasters A myriad of other factors, either coincidental or the consequence of deliberate public policy, can also act to The wide range of views on the economic consequences dampen or amplify the economic impact of a hazard of natural hazards justifies a careful reexamination of event. (For a fuller discussion, see Benson and Clay the theoretical arguments in order to set the context 20 Understanding the Economic and Financial Impacts of Natural Disasters Box 2.1 Measuring vulnerability Several attempts have recently been made, essentially for purposes of international development policy, to assess coun- tries' relative vulnerability to natural hazards. The results typically generate a numerical ranking of countries. Attach- ing a single quantitative number to a country's vulnerability is an appealing notion. Just as the poverty level provides an immediate indicator of the likely importance of poverty reduction efforts in a country and the extent to which poverty factors should be considered and addressed in broader policy, a summary indicator of vulnerability could highlight a possible need to address vulnerability to natural hazards and perhaps bring risk concerns into the heart of government thinking. The recent flurry of interest in indices of vulnerability to natural hazards is part of a wider attempt to measure the extent of the vulnerability of individual countries, particularly small states, to external economic shocks. This interest is partly fueled by the fact that many small states have relatively high levels of per capita income, suggesting their eco- nomic strength rather than--as is often the case--economic fragility. The focus on their relative wealth, rather than on economic stability, has limited these countries' access to concessional aid resources, generating concern as to whether they may require differential treatment by the international development community. Vulnerability indices could become an important determinant of a country's graduation status within the United Nations and Bretton Woods systems (World Bank 2002). The usefulness of vulnerability indices ultimately hinges on their success in capturing meaningful differences in sensitivity to shocks. The vulnerability indices that have been developed so far have been based on a (sometimes weighted) range of components designed to capture various aspects of vulnerability, including that relating to natural hazards. Vulnerability to natural hazards, in turn, has been measured on the basis of some form of historical record of impacts--for example, total damage from significant disaster events, defined as events causing damage that exceeds 1 percent of gross national product (GNP) over the period 1970­89 (Briguglio 1995); the percentage of the popula- tion affected by natural disasters over the period 1970­96 (Atkins, Mazzi, and Easter 2000); or the total number of natural disasters over the period 1970­96, expressed in relation to total land area (Atkins, Mazzi, and Easter 2000). The United Nations Development Programme (UNDP) is currently working on a more sophisticated form of vulner- ability indicator which incorporates political, social, and economic factors that could influence current and future vul- nerability. Even with that more refined measure, the scores and relative ranking of different countries will be approximate, at best. Differences in definition, compounded by incomplete and inaccurate data, have resulted in significant differences in the ranking of countries--although all rankings highlight the apparent special vulnerability of small island economies to natural hazards. For example, Bangladesh ranked 15th as measured by total damage as a percentage of GNP, 2nd according to population affected, and 23rd according to total number of disasters relative to land mass. The corre- sponding rankings for Dominica were 2nd, 13th, and 7th, and for Malawi, 51st, 23rd, and 44th (a tie). More fundamentally, any ranking is sensitive both to the period of analysis and to the type of hazard faced by a par- ticular country. Different types of hazard imply different types of risk. A country might have a high score because of a single extreme geophysical event with very low probability of recurrence, or a very low score because it has not expe- rienced such an event. Moreover, each index has been based entirely on historical impact, however defined, and has ignored the dynamic nature of vulnerability, which demands a forward-looking perspective. The impact of past disasters is significant, and comparison of hazard events across countries and within countries helps build understanding of the factors contributing to vulnerability. Past impacts, however, cannot be directly equated with future vulnerability. The nature and scale of vulnerability, whether viewed in economic or other terms, are themselves highly dynamic and in constant flux, since human actions continually influence vulnerability, at both the household and macroeconomic levels. The insidious, progressive effect that the HIV/AIDS pandemic in southern Africa is having on vulnerability to natural hazards, which was fully exposed only when unfavorable weather triggered a humanitarian crisis in 2002, highlights these issues. Source: Benson (2003b). Disasters and the Macroeconomy 21 Box 2.2 Saying so does not make it so: poverty reduction strategies During the 1990s, international commitments were made to what eventually became the United Nations Millennium Development Goals for poverty reduction. The commitment process was initiated in response to the disappointing outcomes of efforts to reduce poverty in earlier decades. As a consequence, since the late 1990s, preparation of a poverty reduction strategy paper (PRSP) or a similar policy document has become a major focus of government-donor dialogue on development strategies. To some extent, poverty reduction has superseded structural adjustment--which was the focal issue of development policy dialogue in the early 1990s, when the authors' initial studies were conducted (Benson and Clay 1998). Clear evidence of this change in priorities toward poverty reduction emerged in the course of the three country case studies. At the time of the country investigations in mid-2000, the government of Dominica, a small lower-middle-income country, had not yet prepared a formal poverty strategy. Some aid officials saw the omission as an obstacle to support for disaster mitigation. In the aftermath of Hurricane Lenny in 1999, certain donors wanted to support only compo- nents of reconstruction that accorded with their poverty reduction priorities--for example, funding repairs to and hazard-proofing of parts of the vulnerable west coast road in the relatively poorer northern extremities of the island. In Bangladesh the government published its draft strategy on poverty reduction and development while this study was in progress. Disaster reduction is not incorporated in any systematic way into the new strategy; as in previous medium-term planning exercises, it is treated as a separate sector of public action (Bangladesh 2002). The draft Malawi PRSP is organized around a framework for sustainable improvement in the livelihoods of the rural poor, and the negative impacts of drought shocks on livelihoods are recognized (Malawi 2001). Before the food crisis in 2002, however, actions for mitigating drought and climatic shocks were not accorded any priority in the proposed strategy. These cases suggest a potentially unsatisfactory feature of the PRSP process that merits further investigation. Link- ages between disasters and poverty receive increasing recognition; indeed, exposure to risk and income shocks, including shocks emanating from natural hazards, is identified as one of the four basic dimensions of poverty in the PRSP handbook (Klugman 2002). Nevertheless, evidence from the three country studies suggests that disaster reduc- tion is not being treated as a significant, integrated, cross-cutting priority within the new development strategies. Ways need to be found to ensure that disaster mitigation and preparedness are mainstreamed within the PRSP framework. for a presentation of case study findings. This section ECLAC and IADB 2000; UNISDR 2002).There has, reviews the available evidence on the economic impact however, been little wider appreciation outside the field of disasters, in both the short and the longer term, pri- of disaster management of the potentially serious impli- marily as captured in aggregate measures of GDP per- cations of natural hazards. formance. Disasters may also have implications for Some economists, in fact, question the adverse effects sustainable development more broadly defined, in terms of natural hazards and even argue the opposite--that of livelihoods, human development, and poverty reduc- disasters can be a positive shock: tion, but investigation of such effects was beyond the scope of the study. The development costs of war are greater than the destruc- tion associated with natural disasters, for two reasons. First, natural disasters such as floods, hurricanes, and Theoretical Perspectives earthquakes tend to destroy housing and transport infra- structure, but have less effect on productive capacity and The study is based on the premise that natural hazards leave human capital (other than those killed, of course) potentially have significant adverse macroeconomic con- intact. Second, as organizational and social capital remains sequences and so require serious consideration by pol- intact and natural disasters tend to be of relatively short duration, investment quickly recovers and may even have icymakers and decisionmakers. A number of other studies a positive multiplier effect on the economy as a whole. have also recognized the economywide significance of (Stewart and others 2001: 15­16) natural hazards and the problems they pose for long- term development (see, for example, Jovel 1989; This argument relies heavily on a multicountry empir- Anderson 1991; Gilbert and Kreimer 1999; UNDP 1999; ical analysis (Albala-Bertrand 1993) of the impact of 22 Understanding the Economic and Financial Impacts of Natural Disasters single disaster events on overall levels of economic in the aftermath of a geophysical disaster, potentially growth and other broad macroeconomic indicators. creating a construction boom that can last several years, Albala-Bertrand examined the impact of 28 disasters while agriculture--which in many developing coun- in 26 countries occurring over the period 1960­79. tries is still a major source of livelihoods--is seldom He concluded that there are no such things as economic severely affected by geophysical events. Albala-Bertrand "national calamities" and that the overall rate of growth also treated disasters as isolated, one-off events rather of GDP improves after a disaster. than as recurring shocks with potentially cumulative A theoretical explanation for this apparently contra- economywide impacts. That is a reasonable assump- dictory finding is provided by Aghion and Howitt's (1998) tion for a low-probability event such as an earthquake endogenous Schumpterian model of growth through a or tsunami, but in order to conclude that natural dis- process of creative destruction. In this model, growth is asters are beneficial for long-term development, it would generated by a random sequence of quality-improving at least be necessary to undertake analysis over a innovations resulting from research activities (Benson much longer period and to include a careful examina- 2003a). Replacement capital is likely to embody tech- tion of climatic hazards of relatively frequent occur- nical changes that raise factor productivity and thus com- rence, such as floods in Bangladesh or drought in petitiveness. A disaster would be the random event that southern Africa. results in the adoption of such improved innovations. Thus, each disaster would force an economy upward to a new, scaled-up version of itself. Short-Term Impacts In reality, the process is not that simple. Postdisas- ter investment resources, for instance, are not neces- Disasters commonly cause a short-term decline in sarily additional (see chapter 3), and some technological GDP. For example, Charvériat's study (2000) of 35 dis- advances may occur at the expense of others that were aster events in Latin America and the Caribbean between already scheduled. Moreover, postdisaster reconstruc- 1980 and 1996 found that real growth rates fell in the tion efforts are not necessarily well planned or care- year of the disaster in 28 of the cases examined and fully orchestrated, and so technological and other typically rose sharply in the two succeeding years. This improvements may not occur. Financial capital may and other studies have, however, found wide varia- also move its place of business, or fresh FDI may tions in the level of interannual fluctuations in GDP, select other sites, so that the disaster-affected area loses reflecting not simply the scale of direct losses but also capital assets to less risky locations. And where factor other variables at work. In-depth analysis of individual mobility exists, there may be loss of human capital country experiences is necessary in order to identify the through labor migration.19 factors underlying these variations and to understand A careful reexamination of Albala-Bertrand's data the nature of their interplay with natural disasters. set suggests that his findings may not entirely contra- In considering individual countries, it is useful to dict this study's conclusions on the negative impact of distinguish between three groups: hydro-meteorological disasters. Most of the countries · Very small countries where hazard impacts may be that he found to have achieved higher GDP growth in economywide the two years after a disaster, as compared with the two · Larger countries where substantial areas are directly years before the event, had experienced earthquakes. affected by individual events Ten of the 12 countries with lower postdisaster growth · Larger countries that typically experience individ- had been affected by other types of hazard (Benson ual disasters in a geographically limited area. 1994). These contrasting results imply that the net Small island economies are particularly vulnerable positive effects reported were largely associated with to natural hazards. The impacts of hazard events typi- geophysical hazards, the potential impacts of which cally show up clearly as interannual fluctuations in are somewhat different from those of climate-related economic performance, as seen in Dominica, Fiji, events. Large reconstruction programs may be required and Montserrat. The sensitivity of Dominica's sectoral Disasters and the Macroeconomy 23 economic performance to hurricanes over the period The Philippines is one of the most hazard-prone coun- 1978­98 was tested more formally in the case study, tries in the world, experiencing all major types of nat- using regression analysis. The analysis confirmed the ural hazard. Because of its large geographic size, however, negative impact that major hurricanes have had on over- individual hazard events (except for droughts) typically all short-term economic performance, although it also affect only a small part of the country, implying that indicated that the effects of hurricanes have become rel- their impact may not be reflected in fluctuations in atively less severe with the decline of agricultural national economic performance.20 Measurement prob- sector product as a share of GDP. Fiji displays similar lems are compounded by the frequent occurrence of evidence of extreme sensitivity to natural disasters, as tropical cyclones and extreme floods in some parts of confirmed by regression analysis of sectoral economic the country, which makes it difficult to establish a coun- performance from 1971 to 1994 (Benson 1997a). terfactual (nondisaster) scenario. Indeed, government Montserrat is the extreme case of a hazard-prone accounts of national economic performance refer only microeconomy. After the devastation of Hurricane Hugo to major disasters. Yet annual losses as a consequence in 1989, GDP returned to its predisaster level only in of tropical storms alone may be equivalent to as much 1994. Then, after the volcanic eruption began in 1995, as a 0.3 percent reduction in GDP growth (Benson GDP fell by half over three years. Even these dramatic 1997b). figures underestimate the true extent of economic down- turn: since the onset of the volcanic crisis, the econ- Long-Term Impacts omy has been sustained only by substantial inflows of assistance, which have offset to some extent the mas- Theories of development place considerable emphasis sive decline in autonomous private sector economic on the roles of capital and labor growth and produc- activity (Clay and others 1999). tivity (see, for example, Solow 1956; Denison 1967). In geographically larger countries, the short-term Capital assets and other resources can be severely effects of disasters are discernible when severe natural damaged by natural disasters, and the productivity of disasters affect extensive areas or the whole country. undamaged capital and labor can be reduced by asso- This was broadly the case in southern Africa in 1991/92. ciated disruptions of infrastructure and markets. Except GDP contracted sharply in every country except Botswana in the case of drought, significant direct capital losses and Namibia, whose economies are dominated by can result. All major types of disaster (including drought) mineral exports. The recovery in 1993/94 failed to com- can disrupt longer-term investment plans for both phys- pensate for the departure from predrought trends before ical and human capital, in several ways. Governments the next shock occurred in 1994/95. These regional may divert resources away from planned investments findings are confirmed by country-level statistical analy- to fund relief and rehabilitation. Public reconstruction sis for Malawi and Zimbabwe. efforts may also be funded through domestic or exter- In Bangladesh there is also a clear relationship between nal borrowing, increasing future debt-servicing pay- extreme volatility in the economy between 1965 and ments. Even if disaster-related external assistance is 1975 and catastrophic natural disasters. The disaster extended, it may not be entirely additional; instead, shocks were compounded by conflict and internal dis- because of some combination of limited donor resources order during and after the War of Independence in 1971. and constraints on local counterpart funding, it may in The amplitude of disaster shocks considerably dimin- part replace development investment aid flows (see ished after initial postconflict reconstruction. Reinte- chapter 3). Other damage may be covered by insur- gration of the economy was completed in the early 1980s, ance policies, but even this implies opportunity costs and subsequent structural transformation of the econ- related to the payment of premiums (see chapter 5). omy has further reduced vulnerability. Some destroyed assets may not be replaced at all. In An earlier examination of the Philippines highlights the longer term, disasters and related risks can also con- potential difficulties in estimating the short-term effects tribute to economic instability and to an atmosphere of of disasters in a geographically extensive economy. uncertainty, deterring potential investors. 24 Understanding the Economic and Financial Impacts of Natural Disasters Modeling the long-term impact of disasters. Research country's indebtedness. A recursive Keynesian growth into the effects of natural disasters on longer-term eco- model was used, with disaster shocks introduced in nomic growth and development has focused primarily the form of a reduction in private and public capital and on the implications of impacts for capital accumulation. an increase in government expenditure on emergency For instance, a study by the International Institute for relief. It was assumed that the increased government Applied Systems Analysis (IIASA) modeled the poten- deficit would be financed entirely through external bor- tial impact of disasters in three countries (using the rowing. Cochrane concluded that disasters can lower a World Bank's Revised Minimum Standard Model pro- country's credit rating; increase interest rates on exter- jection tool) and found that postdisaster financial resource nal borrowing, in turn dampening investment and reduc- gaps reduce future growth (box 2.3). Currently, pro- ing long-term growth; and increase debt stocks. jections of future economic performance and the iden- The empirical reality of ascertaining precisely how tification of existing and prospective resources to meet countries have in fact financed relief and rehabilitation growth objectives typically do not take into account is far from straightforward, however. As described in the impact of potential disasters. chapter 3, a considerable proportion of postdisaster In a similar vein, Cochrane (1994) explored the impor- public reconstruction efforts involves reallocated tant but rarely considered impact of disasters on a resources, but these resources are typically only partly Box 2.3 Funding rehabilitation: the implications for long-term growth The International Institute for Applied Systems Analysis (IIASA), in conjunction with the World Bank, has developed a planning tool for incorporating future probabilistic losses resulting from natural disasters into macroeconomic fore- casting models and quantifying the implications--in particular, for growth objectives--of various options for finan- cing rehabilitation. In essence, this tool is based on a simple model that focuses on the impact of disaster-related capital losses on rates of national economic growth. To illustrate the use of the planning tool and the nature of the findings that it can generate, the IIASA applied it to three case studies, of Argentina, Honduras, and Nicaragua, under varying assumptions about sourcing of postdisaster funding. The results clearly demonstrate that the ability to finance losses following a catastrophe is crucial to recovery and affects how quickly a country can resume its growth path. In the case of Argentina, it was assumed that all relief and reconstruction financing would be met out of private consumption and foreign savings and that the country would still achieve its growth projections by making sufficient financial resources available to replace damaged capital stock and fund needed future investment. The model was used to estimate the increased government expenditure and import requirements consequent on a disaster and the implied rise in external debt and decline in private consumption. The Honduras study considered a situation in which private consumption and foreign savings (external borrow- ing) are not reliable sources of postdisaster reconstruction and relief funds (in the case of private consumption, because of the high incidence of poverty). The model forecast the impact of a disaster on the Honduran economy, assuming no access to foreign assistance to meet reconstruction needs. Under this scenario, investment resources were diverted into private and government disaster-related consumption, leading to chronic underinvestment. The model indicated that this would lead to stagnation in future expected economic performance. Nicaragua, like Honduras, currently depends on external funds to sustain infrastructure investment, including postdisaster reconstruction. In this case, the impact of decreased economic growth as a consequence of natural disas- ters was translated into implications for poverty, using a household-level model to supplement the World Bank's Revised Minimum Standard Model (RMSM). The results indicated that the inability to finance probable losses would stall or defeat poverty reduction measures. The conclusion from the analysis was that risks emanating from natural hazards should be incorporated into eco- nomic projections, for three reasons. First, there are high opportunity costs associated with the diversion of scarce financial resources into postdisaster relief and reconstruction efforts. Second, disasters can wreak havoc on the already complicated budgetary planning process. Third, disasters place high demands on international aid resources, divert- ing resources away from development uses (MacKellar, Freeman, and Ermolieva 1999; Freeman and others 2002). Disasters and the Macroeconomy 25 reported in official accounts, making it difficult to deter- factors for development, there has been little empirical mine total reallocations or the related opportunity costs. analysis of the historical evidence on the impact of dis- There are further complications in disentangling the asters on long-term growth.21 Benson (2003a) attempts roles played by disasters and other factors in determining to address this issue, examining comparative cross- levels of domestic and external borrowing or of any sectional data on real GDP performance for 115 countries monetary expansion. over a 34-year period, 1960­93. The results suggest that It is also typically difficult to establish exactly how countries experiencing a higher incidence of disasters much is ultimately spent, even by the public sector, on over the period of analysis tended to achieve lower rates postdisaster response. Reconstruction projects are not of growth than countries experiencing fewer disasters. necessarily labeled as such and may not correspond There are fundamental problems in undertaking such with the needs identified in postdisaster assessments, an analysis. First, countries experiencing a lower inci- as the study of Bangladesh illustrates. dence of disasters were typically among the more devel- As for the private sector, it is difficult to get much of oped countries by the latter half of the 20th century. In a handle on total private expenditure on reconstruc- other words, the subset of less developed countries tion, let alone how it is financed. Aggregate consump- includes a disproportionate number of more disaster- tion, saving, and lending data reveal very little, as prone countries. So, the finding that slower growth demonstrated by case studies of Dominica, Fiji, and and higher incidence of disasters are associated may the Philippines. Data for some countries with large com- simply reflect Quah's (1993) broader finding of a munities of overseas nationals, such as Bangladesh, do polarization toward a bimodal distribution, with indicate considerable postdisaster increases in remit- countries already at the higher end of the income dis- tances, but how much of this is for consumption and tribution likely to experience further increases in income. how much for investment purposes cannot be deter- Second, the pace of growth and the level of develop- mined without detailed investigation. Postdisaster pri- ment of an economy are obviously not determined by vate household investment would, anyway, entail the incidence of disasters alone, as is clearly demon- considerable social capital. strated by the vast body of evidence on other causes of These constraints make it extremely difficult to apply development, underdevelopment, and economic growth. models such as those developed by the IIASA and Nevertheless, the basic finding, if tentative, of a neg- Cochrane to draw out the implications of past rehabil- ative association between disasters and development itation funding decisions for long-term growth. Neverthe- at a cross-country level is paralleled by qualitative evi- less, the models do indicate ways of planning financial dence on differential development within individual responses more strategically in the future, through, for countries. The poorer regions of a country are often example, the incorporation of probable losses into the more hazard-prone. Charvériat (2000), for instance, development planning process. These models could notes that the northeastern part of Brazil and coastal be of significant value in, among other things, deter- areas of Ecuador and Peru are typically poorer than mining the appropriate level of external assistance, as less hazard-prone parts of the same countries. Kelly and part of planning a more rational, coordinated response. Khan Chowdhury (2002) note a concentration of poverty The IIASA model is currently restricted to RMSM- in more hazard-prone districts in Bangladesh. In part, modeled countries.Manyverysmallcountriesdonothave such patterns reflect differences in opportunities for RMSM models, implying that partial analysis, building on growth and development, as determined by the rela- the types of statistical method used in this study, would tive risks faced by different communities.22 There is be required. This approach should give reasonable approx- scope for further research on the differential influence imate answers because small economies tend to be that various forms of hazard exposure and related risks highly open and to have weak intersectoral linkages. may have on regional development. Empirical evidence. Despite the recent revival of interest Qualitative evidence. The authors, in this and earlier in the significance of environmental and other geographic studies, have adopted a less formal, eclectic approach 26 Understanding the Economic and Financial Impacts of Natural Disasters that combines various forms of partial statistical analy- impoverishment of poor households and communities-- sis with other qualitative evidence. An extended coun- an issue that can be systematically explored only at a try-specific narrative is constructed that includes an micro level.27 assessment of the effects of a sequence of disasters. Although no attempt has been made to formally model or quantify the longer-term impacts of disasters in any Lessons Learned of the three case study countries, there is strong qual- itative evidence that disasters have affected the pace and Various lessons emerge from this study relating to method- nature of growth in these and in previously researched ological issues in exploring the potential impact of nat- countries. ural hazards. The evidence also provides some insights In Dominica disasters appear to have had a ratchet into opportunities for reducing the economic effects of effect in determining shifts in the structure of the econ- disasters. omy. Agricultural sector product and agriculture's share in GDP fell successively with each major natural disas- ter shock, in 1979­80, 1989, and 1995.23 Complementary Approaches to Impact Assessment Natural disasters can hamper the pace of investment in basic infrastructure, with implications for long-term A mixture of quantitative and qualitative analysis has growth. As the government of Bangladesh recognized been used to examine the macroeconomic impacts of in a memorandum to the Bangladesh Development natural hazards. The evidence assembled indicates Forum, "inadequate infrastructure has been a constraint that major disasters have both short-term and longer- on investment in productive activities as well as on uti- term impacts for economic growth and development. lization of installed capacity" (Bangladesh 2000: 17). This finding has broad policy implications. But such The memorandum notes that electric power interrup- evidence cannot, by itself, be used directly as a basis tions, transport bottlenecks, limitations of port facili- for formulating disaster reduction strategies. Quantita- ties, and so forth are well-known difficulties.24 The fact tive data are needed to facilitate risk analysis of the that expenditure related to natural disasters has largely potential financial and economic returns to specific displaced planned capital investment and normal recur- policy options and measurement of relative opportu- rent expenditure must be part of the problem. nity costs in the use of financial resources. At the level Similarly, in the Philippines efforts to improve the of the individual project (whether for preparedness or country's transport systems have been only modestly for hazard-proofing), cost-benefit or investment analy- successful, and increasing difficulties have been encoun- sis is required. tered in meeting the social infrastructural needs of the Nevertheless, the case study approach sheds light rising population. An important factor is that a large on the factors that determine the nature of vulnerabil- proportion of the available public resources earmarked ity and the outcomes of specific events. The impact of for such purposes has had to be redirected in response a particular event is determined by a complex and chang- to calamities (Benson 1997b). ing combination of factors. Thus, outcomes of similar Other case studies provide further glimpses of the events are, as the case studies demonstrate, likely to be longer-term impacts of disasters on various aspects of significantly different even when separated by only a economic life. An example is the loss of human capital few years in time, and single-event analysis is likely to through emigration, as in Dominica after 1979 and be an unsatisfactory basis for policy. Detailed, disag- Montserrat since 1995­98.25 Furthermore, the poor gregated, and careful assessments such as those under- draw on their financial and material savings to cope taken in this study can generate considerable with disasters, thus incurring opportunity costs, as is understanding of both the forms of vulnerability faced well documented for droughts in Sub-Saharan Africa.26 by an economy and the possibilities for mitigation and This behavior, combined with disaster-related losses of preparedness. They can also provide information that productive assets, can contribute to the progressive allows governments and civil society to consider how Disasters and the Macroeconomy 27 they can incorporate risk reduction concerns into national boundaries and may provide an inflated measure of economic strategies and policies and, ultimately, how the national economic cost of a disaster. to contribute to sustainable development. The detailed case study approach in the current study Opportunities for Reducing Macroeconomic sheds light on the forms that preemptive action might Consequences of Disasters take by advancing understanding of the complex fac- tors that determine underlying vulnerability so that it The evidence presented in this chapter suggests that can be appropriately reduced. Cost-benefit analysis and natural disasters can and often do have negative impacts investment analysis determine the economic efficiency on both short-term and long-term growth. But it also of individual risk reduction actions. A formal broader demonstrates striking contrasts, not just in changes in economic planning tool, such as that developed by the levels of vulnerability but also in the changing forms of IIASA, provides a means of assessing remaining risk vulnerability. These developments belie somewhat sim- more broadly by quantifying its potential implications plistic notions of a general decline in vulnerability to for economic growth and the requirements for recon- natural hazards as an economy grows. Instead, a more struction funding. Such findings could act as a wake- sophisticated perspective needs to be adopted and applied up call to governments and the international community in undertaking detailed risk assessments from a macro- to take further preventive action. Models of this kind economic standpoint. can facilitate exploration of the implications of differ- Using such assessments, risks emanating from nat- ent public choices concerning reconstruction financing ural hazards need to be incorporated into broad devel- and financial planning for disasters. Thus, the various opment policies and plans. In doing so, a distinction approaches are complementary. has to be made between potential short-term and longer- But a methodological caveat is in order. The case term hazard impacts. In addition, the possible trade- study countries were selected purposively; they are all offs between the two and the interlinkages of impacts disaster-prone. The economywide and sectoral analy- need to be recognized. Sometimes, direct short-term ses undertaken show that in the short term, these impacts are severe but recovery is rapid, as is illus- economies are highly sensitive to natural disaster shocks, trated by the case of banana production in Dominica. the impacts of which are visible in national income Or, the immediate impacts may be entirely indirect-- and trade flows and in physical measures of produc- for instance involving a reduction in availability of inputs tion such as crop yields. The more formal statistical to industry--but recovery may be slow. analysis has quantified the effects of the most extreme Risk management strategies also need to recognize events, where the footprint of a disaster is economy- that disasters are not a single, homogeneous form of wide or impacts are diffused through linkage and mul- economic shock. The two broad categories of hazard-- tiplier effects. For geographically larger countries, climatic and related hydrological hazards, and geo- regional- or provincial-level, as well as national analy- physical hazards--appear to be associated with different sis, may be required, particularly where relatively local- patterns of economic vulnerability and so entail differ- ized natural disasters are a frequent occurrence. An ent options for reducing risk. These differences partly earlier case study of the Philippines (Benson 1997b) relate to differences in the probability of occurrence. highlighted the problems of isolating the impacts of dis- Climatic hazards occur more frequently, and it is eco- asters and differentiating them from annual fluctuations nomically worthwhile (and is recognized as such) to in key economic indicators in situations where the adapt productive activities--for instance, agricultural benefits of an entirely disaster-free year cannot be directly practices--so as to reduce risk. It is also worthwhile to measured, even though nationwide disasters are rare. take appropriate structural and related measures per- Thus, for large countries such as Brazil, Indonesia, and taining to the design and location of buildings and other Nigeria, a more disaggregated regional analysis may be infrastructure. By contrast, options for reducing vul- appropriate. Regional analysis, however, does not take nerability to geophysical events are largely restricted to into account redistributional impacts across provincial physical structural and locational factors. Given rapid 28 Understanding the Economic and Financial Impacts of Natural Disasters urbanization in many countries and increasing invest- resources are involved, direct replacement of a parti- ment in physical infrastructure, the potential impor- cular road or power facility, or support for the recov- tance of such measures cannot be overemphasized. ery of existing agricultural and industrial activities, may In addition, risk management strategies should take not require detailed cost-benefit analysis or environ- into account the challenges and opportunities presented mental impact assessments before a project is approved. by increasing trends toward globalization. Globaliza- Preplanning of changes that could be implemented after tion has expanded the opportunities for risk diversifi- a disaster to reduce future vulnerability could be useful: cation, and for nations as a whole, it is probably a positive the political will for change would already have been trend from a natural hazards perspective, since it may marshaled and preinvestment analysis completed, so open up opportunities for smaller enterprises and con- that the changes could be made without jeopardizing sumers, as well as larger corporations and governments. the objective of rapid recovery. This point is discussed But whether globalization ultimately exacerbates or in greater detail in chapter 3. reduces the sensitivity of particular economies and of Finally, lessons drawn from particular disasters individual households is a complicated matter that need to be assessed, and action should be taken as appro- depends on specific country circumstances, including priate. Disasters can induce policy changes and insti- public actions to reduce vulnerability (Benson and Clay tutional innovations that are ultimately beneficial, not 2002b). only in reducing vulnerability but also in supporting Risk management necessarily involves the private sector economic growth and development. Food policy reforms as well as the public sector. The private sector should be in Bangladesh were directed at preventing a recurrence encouraged and supported in enhancing its awareness of the 1974 flood-related famine (Clay 1985) and then and understanding of risks posed by natural hazards containing the financial costs of subsequent floods and in adopting appropriate risk management tools, both (Ahmed, Chowdhury, and Haggblade 2000). The dereg- structural and nonstructural. As part of this process, it is ulation of agricultural investment after the floods in important to ensure that sufficient investment is made 1987 and 1988 encouraged the rapid expansion of in risk-mapping, monitoring, assessment, and dissemi- disaster-reducing irrigation. The microfinance revolu- nation and that this information is provided in an easily tion was in part a response to the 1974 famine (Yunus understood and usable form. Both the services indus- 1998). These examples take the discussion into the tries, including financial institutions, and productive sec- realm of political economy and serve as a reminder tors should be included in such initiatives. that an economy is not a purely technical system that Postdisaster reconstruction efforts need to be well operates through mechanistic responses; there is also planned and carefully orchestrated and should seek to conscious learning, with potentially profound long-term maximize the economic-structural, technological, and consequences. An example of such adaptation is the other improvements that can be made when rebuild- application of climatic forecasting to reduce the impacts ing an economy. Currently, the emphasis often seems of climatic variability in southern Africa, as is described to be on restoring the status quo, which is politically in chapter 4. Another area in which learning is taking and administratively the easiest approach and satisfies place concerns financial mechanisms for managing risk, pressures for rapid recovery. Even where external which are taken up in chapter 5. Chapter 3 Public Finance and Disasters Little, if any, systematic examination has been conducted running down foreign exchange reserves, increasing of the public finance consequences of natural disasters levels of domestic or external borrowing, or expanding and related risk management, except in the narrow con- the money supply. These financing options, in turn, text of a single disaster event.28 This chapter seeks to have potentially significant knock-on effects. The cre- address the gap. It is based in the first instance on find- ation of base money is inflationary. Domestic borrow- ings and issues raised by the Bangladesh case study. ing exerts upward pressure on interest rates and can result in a credit squeeze. Foreign borrowing can lead to an appreciation of the exchange rate, reducing the Background price of imports and increasing that of exports, and it can place future strains on the economy via higher debt- Natural disasters can have important implications for servicing costs. Natural disasters can trigger an increase public expenditure. Disasters are likely to result in addi- in interest rates charged on external debt by raising the tional expenditure or the partial reallocation of already risk premiums associated with a country's assets. The committed financial resources (or both), to meet the option of running down foreign exchange reserves is costs of repair and rehabilitation of public property limited by the size of those reserves and also entails an and to provide support to the victims. These actions appreciation in the exchange rate, with possible asso- may lead to the postponement or abandonment of ciated risks of capital flight and a balance of payments planned investments, reductions in the provision of crisis (Fischer and Easterly 1990). public services, and deferment of wage and salary Disasters can also impose continuing pressures on increases and of staff appointments. Implementation public finance to the extent that governments under- of ongoing projects may be delayed, increasing their take mitigation and preparedness measures. These are ultimate cost. Such problems can be exacerbated by costs that governments in less hazard-prone countries additional pressures on administrative capacity in the do not have to bear. aftermath of a disaster. On the positive side, postdisaster investment may Public revenue may also be affected. Disasters can result in high levels of economic activity. Rehabilita- cause government revenue to fall, since lower levels of tion and reconstruction also provide an opportunity for economic activity, including possible net declines in necessary but neglected repairs and for the upgrading imports and exports, imply reduced direct and indi- of facilities. Needless to say, the specific circumstances rect tax revenue. Although such losses may be partly in which a major disaster occurs, such as the Bangladesh offset by increased flows of official external assistance, floods of 1998, will influence the outcome. they are unlikely to be entirely compensated for. Pub- The challenge is to respond to disasters and related licly owned enterprises may experience disaster-related risk in such a way that economic losses and disrup- losses, placing an additional burden on government tion, including fiscal impacts, are minimized, rapid eco- resources. nomic recovery is encouraged, poorer and more In consequence, a government may face increasing vulnerable sectors of society receive priority support budgetary pressures that it will be obliged to meet by in keeping with poverty reduction strategies, and the 29 30 Understanding the Economic and Financial Impacts of Natural Disasters attainment of longer-term development objectives is not a further allocation of F$0.1 million set against the emer- significantly delayed. This is a tall order. gency transport of water (Benson 1997a). In Bangladesh This chapter examines actual evidence on the the process is less formal, with some disaster-related public finance consequences of natural disasters. It is expenditure apparently met in part by drawing on exist- based, in the first instance, on findings and issues raised ing unallocated resources (block allocations) in the rev- by the Bangladesh case study, but it also draws on evi- enue budget.30 In countries that experience localized dence from Dominica and Malawi, as well as on previ- disasters such as storms, landslides, and flooding ous work by the authors and others. The details of the every year, there is a strong case for such preassign- Bangladesh, Dominica, and Malawi experiences are doc- ment of funds to meet a substantial share of costs (see umented in the case studies and are summarized in the "Financing Hazard-Related Costs" in the "Lessons appendixes. Learned" section, below). The fiscal impact of more sudden disasters may be obscured by lagged effects. For example, a close exam- The Broad Fiscal Impact of Disasters ination of the 1988 flood in Bangladesh reveals that some effects of the disaster were delayed.31 The Major disasters can have significant budgetary impacts. impacts of a disaster with lagged effects are not readily Yet when broad fiscal aggregates, such as the central captured in a statistical analysis or modeled at a highly government's recurrent and capital expenditure, its rev- aggregated level. enue, and the budgetary deficit, are examined, disas- The consistently contrary cases in which fiscal impacts ters are found to have little discernible impact in many of disasters are readily discernible are drought-affected instances. Bangladesh is a case in point: a superficial Sub-Saharan African economies. Five of the six economies review of overall budgetary aggregates in the 1980s and compared in Benson and Clay (1998) showed a sharp 1990s suggests that major disasters, including the increase in government borrowing in response to drought. most extreme floods in 1987, 1988, and 1998 and the The sixth, Zimbabwe, was studied more closely. Over devastating cyclone of 1991, had little impact on cen- the period 1980/81 to 1993/94, fluctuations in both tral government finances. Total revenue and expendi- total central government and direct tax revenue gener- ture increased gradually over the two decades, while ally mirrored overall trends in Zimbabwe's economic the overall budget deficit remained fairly stable until performance, including drought-related economic down- the 1998 flood, when it rose markedly.29 Indeed, an turns. The 1982/83, 1986/87, 1991/92, and 1992/93 examination of Bangladesh's overall expenditure and droughts all resulted in an increase in the public deficit revenue forecasts and performance suggests that the (whether foreign grants were excluded or included). public financial impacts of even the 1998 flood were The droughts of the 1980s also led to higher borrow- very limited. ing and lower debt repayment than had been planned, In three other countries examined--Dominica, Fiji, although the extent of borrowing was apparently con- and the Philippines--it is similarly difficult to discern strained by a deliberate government policy of contain- much impact of natural disasters on government finances, ing the budget deficit. apart from annual allocations of small tranches of bud- Malawi experienced near chaos, fiscally, in the period getary resources for use in the event of a disaster. In the of the two droughts in the early 1990s. In the context Philippines, for example, budgetary resources are annu- of violently fluctuating GDP--which fell and recov- ally allocated to a Calamity Fund, which totaled 0.4 to ered twice, in swings of about 10 percent--public 0.7 percent of total annual government expenditure and expenditure grew rapidly, rising by 30 percent in real 0.9 to 1.6 percent of discretionary spending over the terms between 1992/93 and 1994/95. Revenue, by period 1991 to 1994 (Benson 1997b). In Fiji budget- contrast, declined by 9 percent in 1992/93 and again ary resources to the tune of 2 million Fiji dollars (F$), by 11 percent in 1993/94. The decreases reflected the or US$1.4 million at the 1994 rate of exchange, have been effects of drought on exports and imports (other than earmarked each year for emergency relief activities, with emergency food), accentuated by fiscal laxity in an Public Finance and Disasters 31 election year. The deficit increased by 23 percent over Disaggregated Reexamination of Public Finances three years. Confounding influences were also at work: the election and the incoming government's efforts A fuller understanding of the fiscal effects of disasters to honor its commitments, especially to universal free requires a careful, more disaggregated examination of primary education. individual country experiences. Rather than focusing It is unwise to generalize from a small sample. Nev- on budgetary aggregates, budgetary impacts and related ertheless, evidence for low-income African economies government responses should be analyzed in the con- suggests that drought has distinct fiscal impacts. Part text of overall budgetary performance, recent govern- of the explanation may be that the adverse macroeco- ment policies and budgetary targets, and the structure nomic consequences of droughts can be particularly of government revenue and expenditure. Indeed, a more severe (see chapter 2). Other factors, noted in box 3.1, detailed examination of the budgetary impact of disas- may also play a role. ters begins to reveal a somewhat different picture, as Public resources in a very small economy can be over- has been noted in the case of Bangladesh following the whelmed by the sheer scale and duration of a disaster, 1988 flood. as in the extreme case of the British overseas territory Onekeyissueisthereallocationofexpenditure.Another of Montserrat. In 1989 Hurricane Hugo wrecked the is the role of funding sources. In some countries, exter- economy's infrastructure, necessitating massive exter- nal assistance finances a significant part of public expen- nal assistance for reconstruction until 1994. Then a vol- diture, as is discussed in "External Aid," below. canic eruption that began in 1995 forced 90 percent of the population to relocate; over 60 percent chose vol- Expenditure untary and officially assisted emigration. The volcanic crisis resulted in a massive budgetary deficit as rev- Because of reallocations of budgetary resources, the enue and increased expenditure contracted sharply, and impacts of disasters may be much greater than first the government lost its financial autonomy.32 apparent. Country case studies and anecdotal evidence Box 3.1 Fiscal impacts of drought in Sub-Saharan Africa The fiscal impacts of disasters are readily discernible in drought-affected Sub-Saharan African economies. The case studies, by identifying some specific factors underlying these impacts, help inform policies on the financing of disas- ters, both ex ante and ex post. These factors relate both to the nature of the impact of drought--the region's principal type of hazard--and to specific macroeconomic circumstances. · The macroeconomic impacts of droughts can be particularly severe, directly causing a sharp fall in productive activ- ity. These impacts are exacerbated in economies that are already weak, as are those of many Sub-Saharan African countries. Recovery may be relatively slow, particularly where strong intersectoral linkages exist between agricul- ture and other sectors. Unlike types of disaster that cause extensive physical damage, droughts do not trigger a poten- tial postdisaster boost to the economy in the form of increased, largely nontradable, postdisaster construction activity. · Because public resources are severely limited, less money is available for reallocation, and the limited capacity of government agencies may restrict ability to absorb costs. These factors effectively make relief and rehabilitation expenditure more visible by involving additional resources--for instance, in the form of external borrowing or aid. · Relief and rehabilitation needs associated with droughts are typically very pressing and must be met immediately if further potential losses, including loss of human life, are to be avoided and if rapid recovery on the return of favorable rains is to be possible. This urgency puts severe short-term pressures on budgetary resources and creates a need for additional financing. · Droughts have a direct impact on the principal source of livelihood for a significant share of the population in Sub- Saharan Africa. In combination with lower per capita incomes, they can force a large proportion of the population into dependency on public relief programs. 32 Understanding the Economic and Financial Impacts of Natural Disasters suggest that postdisaster reallocation of resources is of GDP and more than 6 percent of total expenditure. common. The brunt of these reallocations, at least as Yet real expenditure fell by 4.9 percent year-on-year, as they involve financial resources, appears to fall prima- part of broader government efforts to reduce the budget rily on capital expenditure, which typically is largely deficit under a structural adjustment agreement. The discretionary. drought apparently forced a change of emphasis in the In Bangladesh reallocations occur in most years public sector investment program, with some planned because of consistently overoptimistic revenue projec- projects brought forward and new, previously unplanned tions (Rahman and others 2000) and underestimation projects introduced. of recurrent expenditure. Additional postdisaster-related With a few exceptions, such reallocations are typi- pressures on resources then force further reallocation, cally poorly documented and cannot be easily quanti- as happened in 1998/99. fied. In the Philippines, for example, circumstantial In Dominica the apparent insensitivity of budget- evidence suggests that postdisaster reallocations are sig- ary aggregates to disaster shocks also partly reflects post- nificant, but they often occur within broad budget head- disaster reallocations to support relief and rehabilitation ings and so are not apparent in published expenditure efforts. Indeed, the practice of reallocating expenditure reports. inthiswayisapparentlyanannualoccurrence,withunan- There may also be substantial unrecorded realloca- ticipated expenditure on landslide and storm damage tions within the recurrent budget. Freeman and others crowding out routine maintenance. (2003) suggest that such reallocations are likely to be In Fiji there has been significant redeployment of relatively limited because a large part of the recurrent resources in the aftermath of major disasters. In early budget (wages, debt servicing, and operational over- 1993, for instance, Fiji experienced a severe cyclone head) may be nondiscretionary. But in reality, consid- (Kina), necessitating a government rehabilitation pro- erable redeployment in kind (for example, of government gram that was equivalent to almost a third of the staff, vehicles, and equipment and supplies of drugs and annual capital budget and 5.3 percent of total allo- other items) may occur. According to the Bangladesh cated expenditure. Nevertheless, the government government (Bangladesh 1999a: 5), during the 1998 remained intent on containing expenditure and accord- floods, "the entire civil administration was deployed in ingly redeployed resources to meet the cyclone-related relief operations." Such reallocations can be even more costs, holding a special meeting to determine realloca- difficult to track because funds are often spent under the tions. As a consequence, total annual expenditure was same line item--salaries, maintenance of road-building only 0.5 percent higher than budgeted. Recurrent oper- equipment, and so on. In some cases, reallocation of ating expenditure, however, increased by 7.1 percent recurrent resources is facilitated by the negotiation of year-on-year, while capital expenditure fell by 3.4 per- temporary moratoriums on debt servicing. cent, to around only 75 percent of the original alloca- In many countries, a substantial part of development tion.Someinvestmentprojectsweresuspended,including expenditure is funded by external assistance but with a number of small rural projects and a rural road a local counterpart funding commitment. If local program. funds are not available, aid disbursements are delayed, Malawi's experience again confirms that fiscal as happened in Bangladesh and Dominica, with con- effects are more extreme when disaggregated. The com- sequences for non-disaster-related expenditure. position of expenditure, in broad sectoral terms and Disasters also have longer-term impacts that extend between recurrent and development expenditure cate- beyond the crisis year, again squeezing non-disaster- gories, became extremely volatile in the early 1990s. related expenditure. Some reconstruction projects may There were substantial reallocations of expenditure to not begin for months or even years after an event. Mea- agriculture in drought-affected years. surement of the longer-term impacts of disasters on The Zimbabwe government incurred almost 600 mil- patterns of expenditure is again hindered by reporting lion Zimbabwe dollars (Z$) in additional drought-related practices. Disaster-related projects are often not reported expenditure in 1991/92, equivalent to about 2 percent as such, as was seen in Bangladesh after the 1998 floods.33 Public Finance and Disasters 33 Attributing expenditure to a particular disaster can be 1982/83 drought in Zimbabwe, a temporary 5 percent complicated because some infrastructure may have drought levy was imposed on all individuals except those already been in a state of disrepair, as in Dominica imme- inthelowertaxbrackets(Benson1998).Afterthe1986/87 diately before Hurricane David in 1979.34 In some drought, Zimbabwe introduced a 2.5 percent surcharge instances, the upgrading of services in the course of on company taxes for the tax year April 1987 to March reconstruction exacerbates measurement difficulties. 1988 to help finance drought-related expenditure. Other Nevertheless, it remains important to ascertain how efforts target particular industries, as in Fiji, where, fol- much disasters cost in order to inform strategies and lowing two hurricanes in 1985, tax payable on reinsur- policies on optimal risk management, including ance premiums remitted overseas was waived to assist appropriate forms and levels of financial risk transfer. recovery in the insurance industry. Decisions on postdisaster fiscal changes are often made rapidly, with little time for analysis. Yet such issues Revenue require careful retrospective analysis in order to The impact of a disaster on government revenue depends inform future decisions. It is also important to explore in part on the structure (including the relative impor- in greater detail the impacts of disasters on different tance) of taxation and other forms of government rev- types of tax, with the aim of understanding how the enue. In Bangladesh a substantial part of government overall sensitivity of the tax structure to disasters and revenue is generated from import earnings, and the rev- other economic shocks can be reduced and what are enue base has become more resilient to natural haz- the income distributional impacts of tax changes. ards since the early 1980s as a result of changes in the More in-depth analysis of the effects of disasters on other country's principal sources of foreign exchange (see sources of government income would also be useful. chapter 2). In Dominica the insensitivity of local rev- Finally, it should be borne in mind that disasters may enue to natural disasters in part reflects the absence of affect the timing of tax payments. Disasters can disrupt direct taxation on agricultural production, which is par- the smooth inflow of revenues through the year, perhaps ticularly sensitive to natural hazards, and the relative creating increased bunching of inflows and necessitat- unimportance of export taxes, which since 1979/80 ing greater short-term borrowing, as in Bangladesh in have accounted for less than 1 percent of local rev- 1998. enue. The yield on general consumption taxes will fall markedly only if widespread personal income losses Policy Context occur--an exceptional outcome observed only in extreme cases such as the 1990s volcanic crisis in Montserrat. The budgetary impact of disasters also needs to be Postdisaster changes in taxation can also affect understood in the context of prevailing government levels of revenue. To address the economic consequences policies and priorities. Pursuit of various budgetary of natural disasters, a government may offer certain tax goals can play a role in obscuring the impact of disas- reductions as an incentive to economic recovery, or it ters. As noted above, increased disaster-related expen- may increase taxation to meet additional disaster-related diture may be offset by cutbacks elsewhere. The declining expenditure. Some governments choose both courses fiscal deficit in Bangladesh in the flood years 1987/88 of action, as happened in Bangladesh following the 1988 and 1988/89, for instance, reflected deliberate efforts and 1998 floods. to reduce the fiscal deficit following the 1979 oil price Theavailableevidencesuggeststhatwhenfiscalchanges shock and the collapse of international jute prices in are made, there is often some attempt to achieve greater the early 1980s. In fact, in the prevailing policy envi- equity and support those hardest hit. In Fiji an existing ronment, which typically emphasizes careful budget- tax exemption on farming income was extended for ary management (perhaps partly because it may be an another five years following two cyclones in 1985, partly aid conditionality), natural disasters are often more in recognition of the particular difficulties faced by likely to force reallocation of monies than substantial cane farmers (Benson 1997a). In the aftermath of the increases in expenditure. 34 Understanding the Economic and Financial Impacts of Natural Disasters Monetary objectives also influence budgetary out- case in Dominica. In exploring the impact of disasters comes. The Bangladesh government deliberately made on public sector enterprises, it is important to consider only modest use of deficit financing in the aftermath of all such enterprises rather than make assumptions about the 1988 flood, with the objective of keeping the expan- likely impacts. For instance, the impact of the 1991/92 sion of domestic credit and broad money supply drought in Malawi and Zimbabwe was not confined to within the desired limit. In consequence, spending for agricultural parastatals but affected utilities and trans- the year on the Annual Development Program (effec- port parastatals as well.39 tively, capital and some recurrent expenditure) was almost 20 percent lower than budgeted, despite flood- related expenditure.35 Such policies can delay rehabil- External Aid itation and curtail other investments, contributing to lower long-term rates of growth (see chapter 2).36 It is widely believed that the international community Disasters themselves can trigger changes in policy responds to disasters by increasing assistance, particu- that affect budgetary outcomes. Temporary adjustments larly in the form of emergency relief and food aid. This in fiscal or monetary policy are one response, but dis- has given rise to concerns about moral hazard, in the asters may also prompt more fundamental changes sense that "provision of post-disaster assistance creates that have longer-term budgetary implications. Follow- disincentives for recipient countries to ensure the phys- ing the 1987 and 1988 floods, for example, the gov- ical protection of their assets through disaster preven- ernment of Bangladesh embarked on deregulation of tion and mitigation measures" (Freeman and others 2002: private agricultural investment. This shift in policy 35). But is disaster-related assistance really additional? had a direct impact on production and, because of the The response by donors to a particular disaster needs government's active involvement in the food economy, to be seen in the context of their normal noncrisis on public finances in 1989/90 and beyond.37 activity. Aid typically provides support for development, including investment and (as case studies confirm) elements of recurrent spending associated with proj- Reporting and Accounting Practices ects. In some policy contexts, bilateral donors and inter- Budgetary reporting practices can obscure the impact national financial institutions provide broader of a disaster. The Bangladesh government's method of programmatic and budgetary support. When a subse- reporting, which is unusual in several respects, makes quent disaster shock puts pressure on the country's it less likely that the public accounts will reveal the full public finances and creates foreign exchange difficul- extent of the impact of disasters on public finance. The ties, an appropriate crisis response is low-cost, rapidly quasi-fiscal deficit of the central bank is not yet recog- disbursing, additional financial aid that is focused on nized as part of the public deficit, even though the cen- meeting the direct costs of disaster response and coun- tral Bangladesh Bank absorbs effective losses on teracting the recessionary effects of the shock. The appro- subsidized credit programs that the financial sector is priate balance of aid instruments, including food aid obliged by the government to offer and even though it and other relief supplies, depends on the precise counterguarantees the government's external borrow- nature of the shock and the circumstances of the affected ing.38 Analyzing the performance of the Bangladesh gov- country. Both governance and the sheer scale of the ernment's separate Food Account is difficult, as budgetary assistance required can influence donor decisions. In reporting does not cover stock changes. some cases, it is not feasible to provide all assistance in More generally, it is not easy to obtain information the form of financial aid to the government; direct emer- about the impact of disasters on autonomous and gency relief through international and nongovernmen- semiautonomous government agencies. Information tal organization (NGO) channels may be required (Benson concerning public enterprise finances and transfers to and Clay 1998). and from such agencies may not be included in govern- The case studies for Bangladesh, Dominica, and ment financial statistics or budget estimates--as was the Malawi considered external assistance only in terms of Public Finance and Disasters 35 aggregate commitments and disbursements and the and wider policy priorities. The requirement to demon- related fiscal and macroeconomic implications. The find- strate acceptable internal rates of return resulted in under- ings are consistent, despite very different country cir- investmentinstorm-proofingofportfacilitiesinDominica; cumstances. In all three countries, external aid has been the potential cost of storm damage had not been ade- the main source of funding for development. In low- quately factored into the calculations (see note 34). income Bangladesh and Malawi, external assistance also More recently in Dominica, some donors' preference for supports social safety nets directly through the provi- investing in poorer areas increased the difficulty of obtain- sion of food aid and indirectly via poverty reduction ing coordinated funding for protecting economically activities of NGOs. In Malawi aid also finances small key sectors of the island's coastal road network. farmer inputs. Aside from emergency assistance in Some of these policy decisions were unfortunately 1979/80 following Hurricane David, lower-middle- timed. At the time of the 1992 drought in southern income Dominica has received development project aid Africa, bilateral development aid to Malawi had been and programmatic support in response to export earn- frozen because of human rights abuses, and this exac- ings shortfalls and in the context of stabilization and erbated the financial effects of the drought. After a adjustment policies. considerable surge in pre-Independence colonial aid The available data suggest that disasters have little to Dominica in 1978, the newly independent country impact on trends in aid flows in these countries. Many was beset with governance problems and poor finan- donors appear to respond to disaster crises by reallo- cial management. Most donors and international agen- cating resources and bringing forward commitments cies had not established effective working relations with under existing multiyear country programs and budget Dominica when its most extreme 20th century disas- envelopes. Reflecting these practices, total aid com- ter, Hurricane David, struck in 1979. In 1974 the United mitments do typically increase in the year of, or imme- States halted food aid to flood-affected, famine-stricken diately following, a major disaster. Reallocations can Bangladesh because of a breach of the U.S. trade embargo be an appropriate way of responding, in the sense that on Cuba. they involve a lesser administrative burden than the These are not just old issues. Donor-related difficul- negotiation of fresh aid commitments. By the same token, ties over governance and poor financial management however, development spending, which is largely aid delayed the commitment of emergency assistance to supported, tends to fall as aid and counterpart local Malawi in the lead-up to the 2002 food crisis. The first funds are shifted to emergency assistance. The reallo- famine deaths in Malawi since 1941 were reported in cated resources are typically not made good subse- 2002 (Devereux 2002; IDC 2003). Low-income coun- quently; instead, aid commitments fall back after the tries are at their most vulnerable, financially and eco- crisis, with total aid receipts in line with longer-term nomically, to a disaster shock when there are problems trends. of governance and poor fiscal and monetary manage- After a disaster, a substantial gap often opens up ment. The international community needs to be espe- between projected and actual aid disbursements, reflect- cially alert to a potential disaster in such circumstances. ing management constraints such as procedural diffi- culties, procurement delays, and lack of local counterpart finance. (These same problems can also delay disburse- Is Reallocation an Appropriate Solution? ment of development aid.) But rapid disbursement of food aid commitments, in particular, may be vital; As described above, disasters can trigger considerable even relatively short delays can prejudice postdisaster reallocation of budgetary resources. Such reallocations agricultural recovery and cause food system­related help limit overall levels of expenditure and, thus, any financial pressures, as in Bangladesh in 1989 and Malawi widening of fiscal deficits. On one level, such actions in 1993. would appear beneficial, minimizing some of the poten- Aid flows are also strongly influenced by other con- tial adverse, longer-term impacts of disasters such as siderations, such as changing donor views on governance increases in public sector borrowing and monetary 36 Understanding the Economic and Financial Impacts of Natural Disasters expansion, and perhaps satisfying aid conditionalities This practice informally goes back at least to the mid- regarding the size of budget deficits. To some extent, 1970s, and a formal core investment program was intro- reallocations may also be a natural choice. Some planned duced in 1983. In practice, the system does not appear expenditure cannot go ahead because of the damage to function particularly well. Following the 1987 and and disruption created by a disaster, effectively releas- 1988 floods, available funds were basically spread more ing resources for reallocation. thinly across existing projects instead of allocations to Ideally, the reallocation of funds after a disaster should individual projects being adjusted according to prior- occur through a formal process, in the context of a care- ity. After the 1998 floods, funding allocations to par- ful strategic review, rather than on an ad hoc basis or ticular projects were adjusted according to priority, but via across-the-board restrictions on flows of funds to they were determined not on a long-term strategic basis spending agencies. This, in turn, requires that proce- but depending on the projects' stage of implementa- dures be in place preventing the transfer of funds. tion (that is, whether they had been approved) and Donors, for their part, respond to disasters by reallo- whether they received external aid. cating funds as a rapid, practical expedient, working Such problems may well be generic, a recurrent fea- within an overall country aid program. This is ration- ture of budgeting practice. Writing about responses to alized as drawing on commitments to projects that are shortfalls in budgetary resources more generally, not performing well. But the point about the need for Foster and Fozzard (2000: 18) comment on a ten- a formal and, in this case, joint review process holds dency for across-the-board cuts to be made "on the here too. grounds that this is `fairer' to the various spending agen- A number of further conditions then need to be sat- cies and easier to apply."41 isfied to ensure that reallocations are rational and cost- There is some evidence suggesting that social sec- effective. These are discussed below, with illustrative tors may come off particularly badly in postdisaster real- evidence from Bangladesh. locations. For example, the available data on budgeted, 1. A clearly defined and applied policy framework. Ide- revised, and actual allocations for the flood year 1998 ally,reallocationsshouldbecarefullyplannedtominimize in Bangladesh suggest that social sectors, particularly disruption to and delays in the attainment of key policies health, population, and family welfare, fared poorly. and goals. A clear understanding of the implications of This pattern seems to run counter to the stated objec- particular decisions for the achievement of longer-term tives and pro-poor policies of many governments. goals and objectives is essential. That, in turn, requires 3. Up-to-date, reliable information on the current avail- well-functioningplanningandcontrolinstrumentslinked ability of resources. Good accounting and monitoring sys- to carefully defined, achievable objectives and outcomes, tems are needed to provide timely information on the as well as a system for prioritizing individual projects financial status of line agencies and the availability of (see point 2). In reality, however, planning processes are resources, including external assistance, in the aftermath often weak, particularly where government capacity is of a disaster. Immediate analysis is also required of the limited, hindering rapid, appropriate decisionmaking in likely impact of disasters on future flows of public rev- the immediate postdisaster context.40 enue, again including pledges and flows of external assis- 2. A system of prioritization of individual investment tance, both on a monthly basis and for the remainder projects. The development budget, including most public of the budget year. In reality, many countries may have capital investment, often bears the brunt of postdisas- problems satisfying this condition. In the three case study ter reallocation of resources. Ideally, any reallocations countries, information on actual revenue and expendi- should entail the movement of resources away from ture becomes available only after considerable delay. lower-priority projects. For such a process to be imple- In fact, Foster and Fozzard (2000: 17) find that vir- mented rapidly and effectively, projects already need tually all developing countries operate a cash account- to have been ranked according to priority. ing system in which transactions are registered only after In Bangladesh certain "core" projects are indeed given payment has been made. This makes it difficult to obtain priority under the Annual Development Plan (ADP). a clear picture of outstanding liabilities and payment Public Finance and Disasters 37 arrears at any given time, and thus of the resources avail- 5. Good information on the broader macroeconomic able for reallocation after a disaster. Problems can be impact of disasters. Government and other economic exacerbated where donors deal directly with line agen- decisionmakers face considerable uncertainty in respond- cies in a country and the ministry of finance is not ing to a disaster and planning appropriate, cost-effective informed about postdisaster assistance. interventions. They have to consider not only the kind 4. Up-to-date and reliable information on the demand of direct assistance required but also the resources avail- for resources. Information is also needed on the impact able to the government and, underlying this, issues such of disasters on planned spending and the new expen- as the most appropriate fiscal and monetary policies. diture requirements they generate. Timely production There are inevitably tradeoffs, and the relative benefits of comprehensive and accurate damage assessments is and costs of various options have to be considered. For essential, together with proper appraisals and reliable example, should a government expand credit availability cost estimates of postdisaster rehabilitation and recon- to support productive recovery, or tighten monetary struction projects. growth to stem possible disaster-related inflationary Preliminary damage assessments are often pro- pressures? The answer will depend on prevailing cir- duced relatively rapidly after a disaster. These assess- cumstances, which means that accurate, reliable, up- ments, however, are commonly problematic. In 1991 to-date information is needed on many aspects of the Economic Commission for Latin America and the economic performance (which themselves may be Caribbean (ECLAC) developed a formal methodology affected by the disaster), as well as on the direct costs for assessing damage, and the ECLAC is working in con- of the disaster. In reality, it is difficult to obtain such junction with the ProVention Consortium to promote information, and expectations of economic performance its wider use (ECLAC and IDNDR 1999).42 But many may be revised substantially in the months succeeding governments, particularly outside the ECLAC region, a disaster, as in Bangladesh in 1998­99 and Malawi still do not apply comprehensive guidelines in estimating in 2001­02. losses. This lack of guidance can result in nonscientific sampling procedures and incorrect valuation of damage. To conclude, the conditions under which decisions There are also often gaps and discrepancies in cover- on postdisaster reallocations of budgetary resources age. Damage assessments are typically undertaken by are made are likely to be far from ideal; indeed, disas- a range of government and other agencies, each with ters can result in budgetary chaos and confusion. Where its own concerns, and so some impacts may go unre- disaster-related expenditure occurs regularly, resources ported. Moreover, as in Bangladesh following the 1998 should be specifically earmarked for postdisaster response floods, and in Dominica in 1999 after Hurricane as part of the annual budgetary process, and efforts Lenny, there is typically no considered reassessment of should be made to ensure that appropriate mitigation losses once a situation has stabilized, and no final loss and preparedness measures are properly funded and figure is produced. undertaken. Failure to do this constitutes poor bud- A comparison of the 1988 and 1998 floods in getary planning and management. This discussion of real- Bangladesh highlights the benefits of fuller assess- location processes and of the extent to which various ment, as well as the critical need for standardized guide- preconditions are likely to be satisfied is also further lines for reporting disaster damage. The coverage of evidence of the need for financial risk transfer instru- impacts was still somewhat arbitrary in the 1998 assess- ments that can help meet the cost of larger rehabilita- ment, but the consequences of the floods were more tion programs, alleviating some of the pressure on closely examined, providing a better understanding of budgetary resources following a disaster. the effects of the disaster and creating the opportunity In fact, the time frame for budget preparation, exe- for improved risk management in the future. The cution, and finalizing of accounts is typically three years. Bangladesh authorities also made more information Preparation starts perhaps a year or more before the available for external assessment, so that the manage- budget year, but even this may not be enough. So, ment of the 1998 crisis was a more transparent process. there may be insufficient time to analyze and adequately 38 Understanding the Economic and Financial Impacts of Natural Disasters address the impacts of a disaster, even in the budget forecasting, creation of early-warning systems, prepa- for the succeeding year.43 ration of emergency plans, and stockpiling of food, drugs, and relief supplies. Funding for such activities is often contained within overall budgetary allocations Risk Reduction Activities to individual departments. These departments are given only loose directives about their preparedness and In seeking to manage risk, a government needs to explore mitigation responsibilities, and relevant measures and the precise nature and forms of vulnerability. On the actions are not necessarily clearly labeled. Thus, it is basis of this analysis, it should develop an appropriate, nearly impossible to ascertain with any accuracy the integrated risk strategy that covers various aspects of total amounts spent on mitigation and preparedness. vulnerability and draws on a range of responses (regu- This lack of data has further policy implications because latory, fiscal, investment, and strategic policy). A system it is more difficult to present a possible case for main- of monitoring is also required to explore the vulnera- taining or increasing expenditure if the current level of bility implications of particular policies and expendi- spending is not known.48 ture decisions and modify them as appropriate. The fact that governments and donors typically do Some elements of a strategy do not require substan- not report expenditure on mitigation and prepared- tial financial resources; others may. Even if the funds ness suggests that such spending has low political pri- required are modest, budgeting is invariably very tight, ority. Measurement can be a problem, particularly where with many demands competing for limited funds. Accord- a risk reduction measure is included as a small com- ingly, the implementation of a risk strategy ideally requires ponent of a project rather than entailing dedicated proj- a long-term budgetary framework to ensure that ects that are clearly identifiable as mitigation or longer-term goals are not repeatedly subordinated to preparedness. Examples are cyclone-proofing or earth- shorter-term, perhaps populist, spending decisions.44 quake-proofing buildings as part of a school building Fozzard and others (2001) observe that in reality, even project and drought-proofing a water supply. Other ini- countries with a tradition of five-year plans have not tiatives in areas such as poverty reduction, agricultural been successful in integrating these development plans extension, and microfinance can also contribute to reduc- with the annual budget. There is, however, at least increas- ing vulnerability, and this further complicates any attempt ing recognition of the need to capture the long-term at calculating spending. Aggregate expenditure on implications of current spending decisions more gener- mitigation and preparedness, even on dedicated proj- ally, making for a more strategic approach that would ects, is rarely, if ever, reported in annual public accounts. benefit agendas that generate longer-term benefits, The nature and level of recurrent expenditure also such as the promotion of risk reduction measures.45 have implications for vulnerability. Many countries, In the meantime, current evidence suggests that in including the three case study countries, have a dual many countries expenditure on structural mitigation budgeting system under which recurrent and invest- may be lower than is economically justified. In Dominica ment expenditure plans are the responsibilities of dif- public expenditure on disaster mitigation and pre- ferent ministries, or of separate departments within paredness was relatively low, at least until the 1990s.46 the ministry of finance. Mitigation is sometimes per- An Organization of American States (OAS) study pro- ceived as involving major structural investment proj- vides examples from a retrospective analysis of infra- ects and so is of little concern to those preparing recurrent structure damaged by disaster, illustrating how small budgets. Yet levels of recurrent expenditure in partic- incremental increases in initial investment costs would ular areas, most obviously operations and mainte- have avoided far higher postdisaster repair costs.47 nance (O&M), have substantial implications for It is difficult to obtain more precise data on levels of vulnerability. Badly maintained infrastructure can increase expenditure for disaster mitigation and preparedness vulnerability to natural hazards and postdisaster costs. measures such as hazard-proofing of infrastructure, con- In Bangladesh the high costs of rehabilitation follow- struction of physical defenses, scientific monitoring and ing the 1987 and 1988 floods and the 1991 cyclone Public Finance and Disasters 39 partly reflected inadequate O&M and poor past invest- Other disaster legacies include increased indebted- ment planning. Inadequate O&M has been a major con- ness, which has potential long-term development impli- straint on the effective functioning of flood control, cations, primarily relating to the opportunity cost of drought, and irrigation infrastructure as well, despite future debt-servicing and repayment costs.52 Disasters huge capital investments. can exacerbate external debt pressures to the extent that Indeed, it is widely held that expenditure on O&M they destroy infrastructure and other assets funded with is too low in most developing countries.49 Greater inte- outstanding external loans.53 gration of recurrent and capital budgets, as embodied The financial effects of disasters can have long-term in the medium-term expenditure framework (MTEF; implications for broader economic policy. In Dominica see note 45), would be beneficial in this regard, help- these severe budgetary effects have been a factor in the ing to tie mitigation objectives into the recurrent budget. adoption of major reforms. In the aftermath of Hurri- If risk management concerns were more fully integrated cane David, Dominica undertook certain reforms into the project cycle, this could influence both the levels under a program supported by the International Mon- of expenditure on O&M and the design standards for etary Fund (IMF). A subsequent structural adjustment construction of infrastructure. The road network in program (SAP) in 1986/87 sought to address the prob- Dominica, which is vulnerable to coastal storm damage lem of slow recovery from the 1979­80 disasters and as well as to annual flooding and landslides, exempli- the effects of later storm damage on exports. There was fies these issues (see chapter 2). furtherpressureforadjustmentfollowingthe1995storms. But on some occasions disasters have been used as Long-Term Policy Consequences of Disasters an excuse for fiscal problems partly attributable to other causes, thus deflecting pressures for policy change. In To what extent do successive disasters thwart the attain- Zimbabwe the 1991/92 drought was in part blamed ment of long-term policy goals? Country case evidence for difficulties which in fact were the result of the gov- suggests that disasters have had some effect but that it ernment's SAP that had been adopted in the previous may be difficult to demonstrate clearly. In Bangladesh year. Such tactics are potentially dangerous, perhaps poverty reduction is widely acknowledged as the buying a government and other interested parties time country's central development challenge (see, for exam- but implying that efforts to address other underlying ple, Bangladesh 2002; World Bank 2002). Successive causes of budgetary difficulties might be postponed. governments have attached high priority and devoted More positively, the existence of Zimbabwe's SAP ensured considerable resources to poverty reduction, focusing that an effective macroeconomic monitoring system was on economic growth and human resource development already in place, providing early indications of the addi- as the main ways of achieving this goal. Disasters, tional economic difficulties that emerged as a conse- however, have played a role in thwarting the attainment quence of the drought (Benson 1998).54 of targets and have hindered efforts to develop the coun- try's infrastructure. As demonstrated above, disaster- Lessons Learned related expenditure displaces planned investment and normal recurrent expenditure. In Dominica, too, dis- The primary lessons to be drawn from the analysis of aster spending has displaced planned investment by the interaction between public finance and disasters creating more urgent needs, although it is difficult to concern budgetary analysis and reporting, options for measure the precise impact on the overall level of cap- financing hazard-related costs, and governance. ital expenditure.50 Efforts to improve the efficiency of the Philippines' transport systems are reported to have Analyzing and Reporting Budgetary been only moderately successful because a large part of Impacts of Disasters the available resources has been redirected in response to calamities, with knock-on implications for the pace This chapter has shown that a disaggregated approach of improvement of rural transport linkages.51 to exploring the public finance impacts of disasters, and 40 Understanding the Economic and Financial Impacts of Natural Disasters better documentation of expenditure on all aspects of given stagnant aid resources and increasingly stringent risk management and postdisaster response, would donor conditionality. Such reviews could provide a useful aid public decisionmaking. input into policy on taxation and other forms of rev- A superficial review of broad fiscal aggregates is enue generation in a postdisaster situation. likely to be hugely deceptive, in that it suggests that even 3. As part of efforts to improve information, attempts severe disasters have very limited budgetary conse- to finalize disaster loss figures would be helpful. It would quences. The initial attempt in this study to explore more also be useful to tag postdisaster relief and reconstruc- closely the budgetary effects of disasters and related deci- tion funding. Currently, postdisaster assessments pro- sionmaking implies a somewhat different picture. Dis- duced in the turmoil and disruption immediately asters were found to have significant narrowly fiscal following an event are rarely revised, and actual relief impacts in the short and longer terms and broader impli- and rehabilitation activities may not directly correspond cations for development. They caused widespread--if with those identified as required immediately after a largely nontransparent--immediate and interannual real- disaster. locations of funds by governments and aid donors. Dis- 4. Finally, for longer-term consideration, the sensi- aster responses can and do crowd out other expenditure. tivity of revenue generation to disasters and the effects In all the countries studied so far by the authors, it that different forms of revenue raising, such as indirect was impossible to ascertain actual levels of expendi- taxes or school fees, have on affected groups in a post- ture on either postdisaster responses or mitigation and disaster situation are also issues for fiscal policy. preparedness. Expenditure is made from a number of budget lines held by various departments and levels of Financing Hazard-Related Costs government and is often not explicitly identifiable as disaster related. These limitations of fiscal reporting lead Policies on the future funding of postdisaster relief and to several specific recommendations. reconstruction would benefit from better information 1. Expenditure on disasters and other related risk on the impact and cost of disasters. Many countries lack reduction activities should be measured more explic- any explicit policy on financing postdisaster response, itly. Spending on mitigation and preparedness needs to beyond making some relatively minor annual budget- be properly reported. Risk management measures, how- ary allocations for use in the event of a disaster. Instead, ever, take many forms and cut across virtually all sec- they implicitly rely on postdisaster reallocations to meet tors of government. It is not proposed that dedicated a large share of relief and rehabilitation costs, without funds should be established for such purposes but, exploring the long-term developmental implications rather, that relevant activities and related expenditure of the various funding options (reallocation, borrow- be tagged so that total expenditure in this area can be ing, insurance, and so on). In many developing coun- estimated. Spending levels could then be assessed in tries, it is assumed that additional external assistance relation to government risk reduction policies and goals. will be forthcoming in the aftermath of a disaster, although 2. Improved information on the fiscal impact of dis- the reality is somewhat less straightforward (see asters is required. This will help ensure the adoption of "External Aid," above). Such approaches to the financ- appropriate mitigation measures in the design of new ing of disaster relief and rehabilitation constitute poor projects and the implementation of dedicated structural planning. Moreover, in many countries, because of the projects, where necessary. More broadly, it will facilitate escalating cost of disasters, they may not be sustain- theintegrationofriskmanagementconcernsintomedium- able in the long term without causing significant eco- and long-term economic and financial planning. nomic damage. Better information on the costs and Careful and detailed review of the fiscal implica- impacts of disasters is required in order to persuade pol- tions of individual disasters would also help identify icymakers to change existing practices. Emerging real- appropriate postdisaster policy responses. The reasons ization of these facts has at least triggered increased underlying any failure to reach planned targets would interest in alternative ways of financing disaster costs, be more transparent--which is ever more pertinent, as described in chapter 5. Public Finance and Disasters 41 In countries experiencing localized disasters, such Postdisaster reconstruction should be better planned, as storms, landslides, and flooding, almost every incorporating mitigation needs, as well as being con- year, there is a strong case for the annual preassign- sistent with development objectives. An option for gov- ment of funds to meet a substantial share of relief and ernments experiencing frequent, large-scale disasters rehabilitation costs, as well as for mitigatory measures. would be to preplan possible reconstruction and reha- Annual budgetary allocations help strengthen finan- bilitation programs on the basis of a series of disaster cial planning and fiscal discipline. But such alloca- scenarios and, within those, to identify critical projects tions, by themselves, may not be the most cost-effective that should receive priority in the allocation of funds strategy. When disasters do not occur, the contingency after a disaster. These exercises should also seek to iden- provisions may well be dissipated in wasteful, last- tify key areas of capital and recurrent expenditure that minute, unplanned supplementary allocations to vari- must be protected in a crisis. Possible policy options, ous expenditure heads. By making budgetary allocations such as specific fiscal changes and monetary measures, for use in the event of disaster, governments are also could be explored through scenarios to develop guide- raising the general level of budget expenditure or lines for responding to disasters. The guidelines could squeezing planned allocations to programs.55 More- be presented in the form of computer simulations. over, contingency funds of this nature are unlikely to For improved financial management of disasters, it be adequate in the event of a major disaster, particu- is also necessary to look at how the frequency of haz- larly where the same budget line is required for other ards and the magnitude of their consequences may purposes.56 change over time. The main factors likely to affect occur- Longer-term disaster contingency reserves that are rence are global climatic change and more localized rolled over from year to year raise different problems. environmental changes. In Bangladesh the costs of urban These reserves can be held either domestically or abroad, flood protection, including drainage and continuous ideally in highly liquid accounts allowing immediate protection against erosion, are inevitably rising, but access to funds. Such accounts, however, offer only a the rethinking of the rural disaster reduction strategy 5 to 6 percent rate of return, compared with the 16 points to the possibility of less costly, more localized percent rate of return frequently attributed to invest- solutions in the future. The effects of climatic change, ment in development projects, an alternative use of and how these effects will interact with the dynamics resources (Freeman and others 2003). In theory, annual of land and water use, also need to be taken into account. contributions to contingency reserves should be equal In Dominica the decline of agricultural exports makes to annual expected losses, depending on the risk the protection of communications infrastructure, which is fund is designed to cover. But, Freeman and others essential to the services sector, increasingly important. argue, the time required to accumulate a sufficient In Malawi evidence of increased sensitivity to climatic fund would be so long that the fund would not effec- variability resulting from a number of factors, includ- tively protect against large events occurring in the first ing the wider HIV/AIDS pandemic, will require a rethink- years of accumulation. In addition, there may be polit- ing of financing for information systems. ical difficulties in keeping up adequate commitments and protecting accumulated funds if there is a long run Governance of disaster-free years. Instead, hazard-prone countries need to undertake analysis to determine optimal bun- Donors and civil society organizations draw attention dles of financing options. They should consider all to the critical role of good governance, which includes possible instruments and the funding needs raised by a transparent and accountable fiscal regime. The coun- more extreme events and by more frequently occur- tries studied were certainly most vulnerable to natural ring lesser events (see box 2.3 and chapter 5). They disasters during periods of conflict and weak gover- should also try to determine appropriate forms of mit- nance characterized by extreme uncertainty and inse- igation and preparedness and adequate levels of fund- curity, as in Bangladesh from 1969 to 1975 and Dominica ing for them. in 1978­80. Governments were ineffective in managing 42 Understanding the Economic and Financial Impacts of Natural Disasters relief, and public finances were chaotic and plagued by problems of governance persist. In Bangladesh the Min- corruption. Dominica's case is relatively straightforward; istry of Finance slowly, and not without reverses, estab- the country returned to more disciplined fiscal man- lished a more stable budgetary envelope and, with the agement during the 1980s. In Malawi the government's central Bangladesh Bank, achieved more disciplined failure, since the early 1990s and despite democratiza- monetary management in the 1980s and 1990s. The tion, to manage the fiscal and monetary aspects of shocks government allowed, and sometimes actively encour- has contributed to increased sensitivity to disasters. But aged, the growth of microfinance institutions, with much such a simple, generalized explanation of failure is too support from international donors. Such developments facile, and it gives those working to ensure good gover- eventually contributed to reduced sensitivity to disas- nance a means of escaping responsibility. ters. Nevertheless, political instability continues, and There are other influences on vulnerability, and there micro-decisionmaking within the government remains are things that can be done within a country and severely compromised by short-term rent-seeking behav- internationally to reduce vulnerability, even if serious ior (Sobhan 1998). Chapter 4 Information on Natural Hazards and Disaster Reduction Information on natural hazards plays a key role in the disruption, was suffered by port facilities, coastal roads, management of disaster risk. This chapter employs the and other infrastructure that still had little protection conceptual framework of international public goods to against wave action in a severe storm. The risks were explain sources of weakness in hazard information understood, but funding constraints had curtailed systems. It considers provisionally, on case study evi- protective measures. dence, whether funding of hazard information for devel- The case studies also provide evidence that weak- oping countries, as an international and national public nesses in the systems that generate and disseminate good, is adequate. information on natural hazards and their likely effects have been a factor in the potential for extreme natural events to become social and economic disasters. The Information and Public Action floods in southwest Bangladesh in 2000 and, notably, the 2002 food crisis in Malawi and in the wider south- The availability of good-quality, trustworthy data is a ern African area are recent examples. necessary condition for effective management of natu- An organizing framework is required to make sense ral disaster risk. Evidence from the case study coun- of these experiences with good and bad practice. This tries provides clear examples of how strengthening of chapter therefore considers the usefulness of the inter- information systems and the application of information national public goods concept in explaining sources of in risk management have reduced the economic and weakness in hazard information systems. Research human suffering inflicted by extreme events. into global climatic processes and forecasting of the The cyclone warning system in Bangladesh is widely likely weather on regional and smaller scales are clas- recognized as a successful instance of disaster reduc- sic examples of public goods. The case studies shed light tion. The meteorological forecasts are able to provide on the applicability of the public-good framework but earlier and more precise advice on imminent storms. also highlight situations in which it breaks down, as This information has been translated into warnings, when there is rivalry among users of information (for with increasing effectiveness (notably in 1997), allow- example, between countries that share a river system). ing precautionary evacuations of at-risk people to shelters and of livestock to protective mounds. In Dominica reassessment of the coastal protection Hazard Information as a Public Good levels required to minimize damage and economic dis- ruption from wave action during severe hurricanes is The generation and dissemination of scientific informa- reflected in the more limited impacts of recent storms. tiononnaturalhazardswouldappeartobeaclassicexam- In 1999 Hurricane Lenny, a Category 4 storm, caused pleofapublicgood.57 Ifagood'sbenefitsarebothnonrival limited damage to infrastructure that had been rein- and nonexcludable, the good is a pure public good. forced or constructed to higher specifications in the A good's benefits are nonrival when one agent's con- 1990s--for example, the sea defenses in the capital, sumption or use of the good does not detract in the Roseau. By contrast, costly damage, causing extended slightest from the consumption opportunities of other 43 44 Understanding the Economic and Financial Impacts of Natural Disasters users of the same unit of good. Nonexcludability requires Public choice analysis suggests that different IPGs that no one can be excluded from the use or consump- are likely to require different funding arrangements. tion of a good once it has been provided. Theoretically, These arrangements will depend on the extent to which "where benefits are non-rival, it is inefficient to exclude IPGs are pure public goods and on the practicalities of anyonewhoderivesapositivebenefit,becauseconsumption exclusion and country financial capacity (Sandler 2002). to more users creates benefits that cost society nothing" Grant funding may be required for the generation of core (Kanbur, Sandler, and Morrison 1999: 61). But the pro- IPGs, and complementary activities that are largely coun- visionofagoodisinherentlyproblematic:institutionscan try-specific are appropriately funded by governments, restrictaccess,andthereisanincentivetofree-ride,which to the extent that NPGs are involved. NPGs may justify cannot be prevented. In practice, goods may not exhibit and more easily allow the use of traditional country- eithernonrivalryornonexcludabilityinapureform.Nev- based aid instruments, including loan finance (Ferroni ertheless,informationonenvironmentalhazardsandenvi- and Mody 2002). The level of aid funding for IPGs is ronmentalchange(i.e.,changesinhazardrisk)isgenerally difficult to estimate precisely, but it is currently at least regarded as being a public good. 10 percent of official development assistance (ODA), Meteorological information on tropical storms is close and the share is generally agreed to be increasing.58 to the theoretical paradigm case of a public good. Such Nonexcludability is a common source of problems information typically includes climatic risk assessments, with coordination and financing because of the incen- which may be the annualized probability of a storm above tive to free-ride (Ferroni and Mody 2002). All countries a specified wind speed or the probable levels and inten- benefit from an IPG (in the case of an RPG, all coun- sities of wave action associated with specific wind speeds. tries in the region benefit), and all should contribute to Forecasts are the other typical products--real-time infor- the costs of provision. But there are problems of valu- mation on the likely spatial range, duration, and inten- ation, and also differences in ability to pay, as reflected sity of a storm when it is threatened or actually happening. in the experiences of the case study countries. Both types of information have made important contri- butions to reduction of disaster risk in Bangladesh and Dominica, as shown in the case studies. Climatic Forecasting in Southern Africa A public good may be international, regional, or national, depending on the potential spread of benefits. Global climatic processes, especially the forcing mech- Some hazard information is country-specific (e.g., assess- anisms that underlie variability and climatic change, are ment of landslide hazard risk in Dominica). An example the subject of intense investigation. The El Niño­South- of a regional public good (RPG) is a tropical storm warn- ern Oscillation phenomenon is widely regarded as a key ing in the Caribbean. The El Niño­Southern Oscillation indicator of global processes, providing a basis for long- (ENSO) is a phenomenon with global consequences, and lead forecasting (Stockdale and others 1998; Zebiak information on it is being explicitly generated and dis- 1999). This research is widely regarded as generating seminated as an international public good (IPG). Even an international public good. International coordina- where programs are country-centered, benefits may tion takes place through the World Meteorological Orga- spill over to others; for example, seismic monitoring in nization (WMO), and the findings and predictions of India registers events that occur in Bangladesh. a substantial number of forecasting models are regu- Practically, the World Bank (2001) has found it useful larly and openly reported--for example, by the Inter- in considering IPGs to distinguish between two cate- national Research Institute for Climate Prediction (IRI) gories of activities: at Columbia University, New York. · Core activities that aim to produce public goods In southern Africa, the close association between El · Complementary activities that prepare countries to Niño events and the regional droughts in 1982/83, consume the IPGs that core activities make available 1991/92, and 1994/95 provided the impetus for the while at the same time creating valuable national development of regional long-lead or seasonal fore- public goods (NPGs). casting (see figure 2.4). By 1997/98, a formal process Information on Natural Hazards and Disaster Reduction 45 for consensus-based long-lead forecasting had emerged, to be taken in circumstances of more than usual managed through the Southern African Regional Cli- uncertainty about rainfall. mate Outlook Forum (SARCOF) and involving national The international effort to understand and predict meteorological agencies from all the Southern African global climatic processes is a pure public good. The gen- Development Community (SADC) countries (Thom- eration of this public good has been extended through son, Jenden, and Clay 1998). The process was funded, international support to regional forecasting as an RPG to begin with, as part of a wider global initiative to in southern Africa and elsewhere in Sub-Saharan Africa. strengthen regional climatic forecasting. SARCOF Because of the wider consequences of food security crises forecasts rely heavily on forecasts from global models in the region, the international community finds itself that also strongly reflect the behavior of ENSO. Addi- with a contingent responsibility to respond, and regional tional detail takes account of regional topography. During forecasting in Sub-Saharan Africa therefore continues the 1997/98 El Niño, conditions in the closer southern to be a core IPG activity. The effectiveness of these efforts, Indian Ocean were locally influential, resulting in higher- however, depends on decisionmakers' capacity to use than-anticipated rainfall in some countries, especially this information at the country level and below. Malawi and much of Mozambique. Much less well under- Country-specific forecasts can alert international and stood oceanic-atmospheric interactions in the Indian national agencies and civil society to the need for pre- and South Atlantic Oceans are now recognized as impor- cautionary measures to safeguard food security and tant influences on rainfall patterns and have become water supplies, and they can reduce the cost of poten- the focus of internationally supported research (Spencer, tially financially destabilizing crisis measures. The study Slingo, and Davey forthcoming). of Malawi has confirmed the value of forecasts at a coun- The precision of SARCOF forecasts is still very lim- try level and the need for complementary activities, ited; broad probability bands are assigned to zones that but it has also highlighted problems (as discussed above) encompass parts or the whole of several countries. that at present limit the value of forecasts. Currently, The forecasts are difficult to downscale and are impre- only some commercial farmers are able to respond to cise as to the risks of erratic rainfall patterns that are more specific seasonal forecasts. Smallholders lack the critical to crop performance.59 But the greater atten- technical options and resources to modify significantly tion now paid to forecasting and weather monitoring their choice of crop, seed variety, or traditional plant- throughout the season ensures that scientific data on ing practices. These differences in capacity to exploit a 10-daily basis are more rapidly available to inform information have implications for income distribution assessment and decisions. Global climatic develop- and poverty. Pressures for cost recovery at a national ments are also closely watched, and assessments are level could result in a focus on product development quickly disseminated through the Internet. A real prob- and dissemination as "club goods" for those with the lem is that decisionmakers would like very clear pre- incentive and ability to pay. The use being made of cli- dictions ("This climatic event will lead to this pattern matic forecasting is promising, but institutional strength- of weather in the coming months"), particularly when ening, technical capacity building, more systematic food security depends on good crop-growing condi- application of current scientific knowledge, and invest- tions. The reality is that because of the complexity of ment in data and equipment are needed. weather patterns and impacts, forecasts often have to The World Bank, using IDA credits, has invested in reflect a lack of certainty. For example, in 2002, as a the development of climatic forecasting in Malawi, fol- food crisis affected six southern African countries, fore- lowing closely the distinction between core and com- casters indicated the high probability of a moderate El plementary activities (Clay and others 2003).60 This was Niño event and then confirmed its occurrence. How- done within the wider context of supporting the strength- ever, there was great uncertainty about what this might ening of the national meteorological agency. The limi- mean for the 2002/03 wet season and whether it would tation of this approach is that forecasting has a high continue. In effect, the models are saying that decisions proportion of recurrent fixed costs, posing a problem about an already difficult food security situation have of sustainability. In Malawi these costs are well beyond 46 Understanding the Economic and Financial Impacts of Natural Disasters the level of expenditure funded by the central govern- and public and private users need time and, probably, ment. Meanwhile, the possibilities for cost recovery in assistance to understand how climatic information can providing relatively generalized information are severely improve risk management. limited. In such a situation, a cost-benefit calculation is required to indicate whether this IPG and comple- mentary activities at the country level should be con- Tropical Storms tinuing priorities for international support. Although it is difficult to assign a robust value to cli- The case studies of Bangladesh and Dominica offer exam- matic forecasting for southern Africa, qualitatively its ples of the uses of forecasting and risk assessment in potential usefulness is clear. Efforts to strengthen cli- dealing with risks from hurricanes, cyclones, and matic forecasting and the associated dissemination activ- other tropical storms. ities have: · Provided a process for scientific consensus Bangladesh · Integrated and strengthened meteorological sys- tems in the region Cyclone protection became an intermittent priority for · Established systems for closer monitoring and report- official donors and international NGOs in response to ing of weather throughout the year specific extreme events. After the 1970 cyclone in · Identified priorities for further research to improve Bangladesh, in which at least 350,000 lives were lost, forecasting ability the Bangladesh Red Crescent managed programs for · Created systems for assessing climatic risk that can cyclone shelter construction, with increasing success. feed into decisionmaking processes. Although 140,000 people lost their lives in the May The strengthening of forecasting systems is yielding 1991 cyclone, 340,000 were evacuated and were housed improved real-time data on rainfall and on weather more in purpose-built shelters and public buildings. In May generally that can inform decisionmaking after a crisis 1997 a similar cyclone claimed fewer than 200 lives, has begun. So far, however, benefits in the shape of and 1 million people were evacuated to shelters (IFRC demonstrable improvements in decisionmaking have 2002). Since 1991, cyclone shelters have been designed yet to be reaped. for multipurpose use, mainly as schools (with budget- The financial costs attributable to the whole fore- ary implications for O&M). Even so, underprovision casting effort for southern Africa are around US$5 mil- of shelters persists. For that reason, and because the lion, spread across services and research institutions buildings are in multiple use, the principle of nonri- within and outside the region. There is a large fixed- valry is compromised. Mitigation investments and cost element in sustaining climatological monitoring humanitarian aid, unless provided at levels that satu- capacity, and forecasting at the national level is a joint rate demand (so that they become a free good) are pri- product whose costs are difficult to isolate. These vate goods, subject to rivalry and excludability, and costs, however, are modest compared with the economic this situation raises intense problems of competitive costs imposed by climatic variability in the region, which access. Institutional arrangements are required to ensure are estimated to be equivalent to at least US$1 billion that the poorest are not excluded (Pantelic and others a year. Even a small reduction in losses through improve- 2000).62 ments in public decisions on food system management A critical factor in the success of cyclone protection and in private risk management would justify both is the considerable progress that has been made in coastal past and continued investment in better forecasting.61 cyclone warning since the devastating storm of May Although long-lead forecasting is still in its infancy, 1991. Meteorological observation of storms by remote climatic research is making rapid progress. Regional cli- sensing has become, in effect, an internationally funded matic forecasting needs to be sustained as a learning public good. The complementary cost of using this infor- process. Scientists need to understand and then attempt mation is relatively modest, and an institutional vehi- to respond to more specific information requirements, cle has been found to undertake the task.63 Cyclone Information on Natural Hazards and Disaster Reduction 47 protection has been facilitated by the gradual emergence Transnational River Systems of large NGOs that provide further institutional sup- Riverine flood warning in Bangladesh is still primitive port to affected communities, including the poorest, in and provides very limited advance warning. Bangladesh crisis situations. effectively received only a 24-hour warning, in the form of "rivers at danger level" advice, from its upper ripar- Dominica ian neighbor, India, at the time of the 2000 floods in southwestern Bangladesh. On Bangladesh's part, river Tropical storm warnings in the Caribbean region have modeling, which is crucial for risk assessment, began beenprogressivelystrengthened.Theregionbenefitsfrom to be seriously undertaken only in 1989, under the its proximity to the United States, the lead nation in offi- Flood Action Plan. Even this work has been restricted cial support for forecasting as a public good. In addition, by the attempt to model only for Bangladesh, which commercial media and Internet-based communication contains the larger part, but not all, of the Brahmapu- havecometoprovidereal-timeinformationonstormrisks. tra-Gangetic delta system.65 But, as the unpredicted course of the destructive Hurri- This weakness partly reflects very restricted regional caneLennydemonstratedin1999,eventhemostadvanced cooperation in meteorology and the monitoring of hydro- prediction models may be found wanting. logical systems. The long-running dispute over the use The recurring damage to Dominica's key infrastruc- of the Farakka Barrage upstream in India to control the ture highlights deficiencies in risk assessment that are flow of the Ganges River illustrates the problems that only slowly being remedied. Following independence arise when the assumption of nonrivalry in consump- in many Caribbean states, wider regional arrange- tion of information does not hold. ments for scientific research on strengthening disaster The floods in Mozambique in 1999­2000 highlighted management are emerging under the auspices of the similarproblemswithhazardwarningsbetweenupperand OAS, supported with international and bilateral (U.S.) lower riparian countries. There are indeed externalities financial and human resources. These projects have con- in water use and control, but the generation and regional tributed to enhancing scientific hazard assessment and dissemination of knowledge on hydro-meteorological monitoring. For example, concerns about sea-level systems can help address the problem of significant exter- rise in the Caribbean Sea and the absence of reliable nalities and provide a basis for cooperation. benchmarks have highlighted the past lack of sea-level and wave monitoring within the region. Dominica itself Geophysical Hazards has no capacity to undertake such monitoring inde- pendently. Consequently, only qualitative assessments The Bangladesh and Dominica case studies reveal the of the coastal sea conditions associated with the difficulties in ensuring that an adequate system of mon- impact of individual storms, including Hurricane Lenny itoring is established and sustained for geophysical haz- in 1999, are available. To provide benchmarks for deter- ards. The problem appears to stem from the typically mining the effects of climate change, the OAS has low probabilities of severe or catastrophic events. The launched a regional program for sea-level monitoring, issues have to do with awareness of hazard risk and supported by the Global Environment Facility (GEF).64 understanding of the appropriate risk management response to hazard assessment. Failures in the Provision of Information as a Bangladesh. All of Bangladesh lies within a high earth- Public Good quake risk region. Minor tremors are common, and one of the most extreme earthquake events ever Examples from Bangladesh, Dominica, and southern recorded--a quake in 1897 that registered 8.8 on the Africa highlight how rivalry and exclusiveness under- Richter scale (RS)--had its epicenter in the nearby Shil- mine the public-good character of forecasting and infor- long Plateau in the Indian state of Meghalaya. The mation provision. 1950 Assam earthquake brought huge sediment loads 48 Understanding the Economic and Financial Impacts of Natural Disasters into the Brahmaputra River, raising the river bed for a hazards in Dominica. Scientific hazard monitoring and number of years (which probably contributed to the information dissemination have been organized in the serious floods in the mid-1950s) and creating large areas Caribbean area at a regional level in ways that reflect of new alluvial land in the Meghna estuary. colonial history. For seismic-volcanic monitoring, Assessment of earthquake hazard in Bangladesh has Dominica contributes to and relies on the Seismic been relatively neglected. This again reflects an insti- Research Unit (SRU), based in Trinidad.67 There is a tutional failure at a regional level to provide scientific convention of extreme caution in making potentially information on a serious natural hazard as a public sensitive information available, at least in those former good.66 The historical record implies an average of more European colonies that have not had a tradition of than one extreme (RS 7.0+) event every 50 years. But open government. The centralized authority in colonial the fragmentation of seismic monitoring systems in 1947 times had a general responsibility--a contingent lia- and again in 1971, as political boundaries changed, bility, in insurance terms--in the event of a disaster, and left Bangladesh with little monitoring capacity--a single nonexclusion of potential beneficiaries is still seriously seismographic station--and so unable to triangulate the qualified by administrative reluctance to make infor- epicenter of tremors. The danger was highlighted in mation available. February 2001, when a tremor of RS 4.2 was identified Dominica is seismically extremely active. The SRU by Indian sources as being centered in northeast successfully monitored volcanic alerts in the 1970s and Bangladesh. 1980s and initiated risk assessment and risk-mapping Because of the low awareness of seismic hazard and activities. When, however, a new volcanic alert began of weaknesses in enforcing building regulations, little in September 1998, the monitoring arrangements regard has been given to earthquake-proofing during were found not to have been properly maintained and the rapid urban expansion that has been going on also to be inadequate, given that an alert has occurred since the late 1980s. Earthquake-proofing has been almost every decade. The monitoring network had to undertaken, however, in major internationally funded be refurbished as well as enhanced, with additional construction projects such as the Jamuna Bridge and emergency external support, to provide the appropri- the Ashuganj fertilizer factory. ate level of seismic monitoring. This suggests that the Reinterpretations of the causes of the 1897 Megha- SRU network was insufficiently funded to enable it to laya earthquake show the challenges in undertaking rel- conduct enhanced crisis monitoring. atively accurate risk assessment in the region, which The way the 1998­99 alert was handled raises the may have large buried faults. On the positive side, the difficult but important issue of how scientific informa- recent Gujarat earthquake in western India, which was tion should be disseminated to the wider public to ensure extensively reported and shown on television in that both public and private sector institutions make Bangladesh, appears to have transformed attitudes within rational decisions about natural hazard risk. Despite the government and civil society about earthquake previous alerts, a volcanic emergency plan had to be hazard. This has implications for public expenditure on specially prepared. Emergency exercises were carried seismic monitoring and risk assessment as public goods. out, but little precise information was made publicly Incorporating earthquake resistance into design available on the nature and extent of the risks. There will increase investment costs for public and private was considerable uncertainty in the private sector and construction of key infrastructure and commercial prop- among civil society organizations about the precise erty. (The insurance option for managing risks is dis- nature and level of risk, how the crisis might evolve, cussed in chapter 5.) Decisions on the tradeoffs between and the appropriate responses. The result was a con- costs and variability would be better informed if improved fused range of reactions. For example, some insurance risk assessment were publicly available. companies temporarily stopped taking on new business in the southern part of the island, a few declined to Dominica. Institutional issues have to be addressed in renew existing annual policies, and others continued ensuring adequate support for monitoring volcanic to provide coverage. But positively, and presumably as Information on Natural Hazards and Disaster Reduction 49 a result of the experience gained in this crisis and the issues arise with respect to all categories of hazard risk example of the continuing volcanic eruption in Montser- because decisionmakers want clear predictions. rat, the SRU has begun to make its risk assessments widely and freely available, both on the Internet and through public meetings. Findings and Conclusions The gap between formal science and the perspectives of most of those making public and private decisions There are several recurring issues relating to the gen- is highlighted by the problem of giving an objective eration and use of information on natural hazards as a assessment of the risks of a volcanic eruption in ways public good. that can be easily understood (box 4.1). If the experts 1. The public-good framework is helpful in clarify- draw attention to a nonzero risk but an eruption does ing issues for further consideration. The distinction not occur, or is "unfulfilled," the general public may between core and complementary activities is especially misinterpret the warning as a false alarm. Similar useful in distinguishing between the generation and the Box 4.1 Evidence-based volcanology: application of Bayes' rule to the situation in Dominica in 1998 Bayes' rule provides a formula for updating estimations of the relative odds of scenario A and an alternative scenario B, given new evidence about either or both scenarios. The formula for the new (or posterior) odds is: Pr (scenario A | evidence) = Pr (scenario A) Pr (evidence | scenario A) . Pr (scenario B | evidence) Pr (scenario B) Pr (evidence | scenario B) where Pr (evidence | scenario) is the conditional probability of seeing the evidence given the scenario occurs, and the ratio of the two conditional probabilities in the above expression is referred to as the likelihood ratio of the evidence. Pr (scenario A) / Pr (scenario B) expresses our prior odds (or belief) and is sometimes referred to as the base rate. (As exemplified in some notorious law cases, ignoring the base rate can lead to fallacious conclusions about the cor- rect weight to attribute to evidence.) In the case of Dominica, where a local earthquake swarm started in 1998: · Scenario A might be the hypothesis that there will be an eruption within a short interval (e.g., the next six months). · Scenario B would be the hypothesis that no eruptive activity will occur in the next six months. The prior odds ratio, Pr (eruption) / Pr (no_eruption) might be estimated from the information that the last eruption in Dominica was about 685 years ago (SRU 2000) and that there have been three eruptions on the island in the past 1,200 years. From the latter information, the current likelihood of an eruption in any six-month period might be estimated as about 1 in 800, on average. (This estimate could be reduced if we had reason to believe that the chance of observing an interval of 685 years or longer is unusual in the distribution of eruption intervals for Dominica.) So, for present purposes, the prior odds of an eruption (in any six months) might be estimated at 1/800, or, in betting terms, about 799 to 1 against. If a seismic crisis has already begun, we should now use Bayes' rule to estimate new_odds = prior_odds * likelihood_ratio where the second factor is a ratio of likelihoods that can be derived from additional information about volcano-seismic crises and eruptions in Dominica. There have been eight seismic crises in Dominica in the past 50 years (before the pres- ent episode) and at least 12 in the past 200 years, none of which resulted in an eruption. So, extrapolating over the past 685 years, there might have been between 40 and 110 such noneruption seismic crises, and perhaps more. If 75 is used as an average estimate, the equivalent "false alarm" probability is about 0.05 in any one six-month period. (continued) 50 Understanding the Economic and Financial Impacts of Natural Disasters Box 4.1 Evidence-based volcanology: application of Bayes' rule to the situation in Dominica in 1998 (continued) Almost all strong eruptions are preceded by some form of seismic activity, which would be detectable by modern instruments, suitably deployed. Assuming that there is no more than a 2 percent chance of not detecting precursory seismic activity before the next eruption in Dominica, Pr (seismic_crisis | eruption) = 0.98 then the likelihood ratio, calculated as Pr (seismic_crisis | eruption) Pr (seismic_crisis | no_eruption) is approximately 0.98/0.05 = 19.6, and new_odds (eruption) = 1/800 * 19.6 1/41 or, in betting terminology, 40 to 1 against an eruption in the next six months. Even at the much lower noneruption crisis rate implied by the known 12 episodes in 200 years, the new odds would still be about 24 to 1 against an ensu- ing eruption. So, even though a notable volcano-seismic crisis has started in Dominica, it is still extremely unlikely, solely on the basis of the evidence that an earthquake swarm has begun, as considered here, that there will be an eruption in the next six months. The numbers can be reworked for other forecast periods, and other evidence--such as the duration, inten- sity, or seismic characteristics of the events involved, or separate independent indicators such as changes in geother- mal conditions--might cause us to modify the probability of this particular episode being a unfulfilled alarm. These low odds for an eruption can be contrasted with a view that is commonly expressed under such circum- stances. As noted above, almost all sudden-onset volcanic eruptions we know about have some form of detectable pre- cursory seismic activity, and, making allowance for incomplete knowledge, we might guess that this happens in at least 98 out of 100 eruptions of volcanoes of the type found in Dominica. Thus, the equivalent odds of there being no detectable seismic activity before an eruption are less than 1/50 (Pr ~ 2 percent)--a "statistic" of such low probability that it leads many people to believe, mistakenly, that an eruption must be almost inevitable once an earthquake swarm develops under a volcano. (In evidential matters, this is an example of what is known as the transposed conditional, or the prosecutor's fallacy.) Source: Derived from Aspinall and others (2002). dissemination of information or knowledge. Dissemi- public good that is relevant to land use regulation and nation is often not a pure public good because fore- building standards. Furthermore, unless a disaster has casts or hazard assessments can be provided to restricted occurred recently, the status of hazard monitoring and groups. the use of information in risk management are likely to 2. Information is most likely to be generated where be the accidental consequences of history. the hazard is recognized as a global issue. Developed 4. Public funding of hazard monitoring and dissem- countries are taking the lead in generation of informa- ination is worthwhile, although precise cost-benefit cal- tion and promotion of its use, as an international public culations are difficult. The case studies suggest that there good. This is most obvious for climatic variability and are considerable potential benefits but that they are extreme weather events such as tropical cyclones. not smoothly distributed. Costs are relatively modest 3. Geophysical hazards are a more uncertain area. but are largely recurrent. The combination of patterns The funding of basic research, a core activity, seems of benefits and costs makes for difficulties in sustain- not to be in question--again, because of the risks posed ing national funding in countries with severe budget- in developed countries. Hazard risks, however, are ary constraints. region- or even country-specific, and so the provision 5. The use of information is apparently unsatisfactory and dissemination of information is more often a national in many cases. This has partly to do with the probabilistic Information on Natural Hazards and Disaster Reduction 51 nature of the information generated by scientific research 6. In relation to crisis-affected Malawi and to southern and monitoring (see box 4.1), and it raises a series of Africa more generally, there is a need to avoid creating questions: excessive expectations. Put simply, "At this time, weather · What forms of information are appropriate to make forecasts can prompt governments and international available to various stakeholder groups? agencies to move to a higher level of preparedness, but · How can scientific information be disseminated cannot effect significant changes in production." 68 in an easily understandable form? 7. There are encouraging examples of improved prac- · How should scientific information be used, and tice. The concerns about global climatic change have what are the implications of its use, bearing in contributed to ensuring interest in climatic processes mind that it will be probabilistic and thus diffi- and funding of research on them. Progress in using infor- cult to take into account? mation to reduce the impacts of tropical storms is impres- · What role should scientists play in informing the sive. There is evidence, too, of institutional learning general public and other stakeholders directly about the need to ensure wider dissemination of infor- about natural hazard risk and uncertainty? mation, such as seismic risk assessments in the Caribbean. Chapter 5 Financing the Cost of Future Disasters Historically, there has been heavy reliance on aid to meet instruments, thus offering investors a way of diversifying disaster-related relief and reconstruction costs in the capital market risk. developing world. But today, global aid flows are stag- The potential advantages of risk transfer mechanisms nant, while the annual cost of disasters is increasing. include: Global climate change could imply a further rise in the · A reduction in postdisaster pressure on fiscal and incidence and cost of natural disasters. Questions are external balances therefore beginning to be asked about alternative ways · Increased government control over the financing of of meeting disaster-related costs. This chapter reviews disasters, possibly including the immediate and timely some ideas and options that are being explored, and in availability of funds69 some cases applied, relating to risk transfer tools and · Greater capacity for a government to set its own pri- their potential relevance and scope for application to orities in the management of relief and rehabilitation developing countries. · Increased transparency in the delivery of relief and reconstruction funding · Promotion of mitigation, by making provision of Risk Transfer Tools mechanisms conditional on particular structural meas- ures being in place or by offering discounts where Risk transfer mechanisms shift financial risk from one they are. party to another. The two basic tools for catastrophic risk An increase in public insurance, in whatever form, are insurance and instruments for spreading risk directly may also stimulate more extensive and fuller private to the capital market. An insurance policy provides cash coverage. payoutsintheaftermathofadisasterinreturnforthepay- In developed countries, there are well-established ment of monthly ex ante premiums; insurance compa- markets for insurance against a wide range of hazards, nies, in turn, redistribute their risk to global reinsurers. including the major natural hazards of floods, droughts, Overthepastfiveyears,novelinstruments,entailingsome cyclones, earthquakes, and volcanic eruptions. Newer form of hedging transaction in the capital market, have hedging instruments are also gaining some popularity. been developed in response to dramatic increases in the Weather derivatives have been written on indices of cost of insurance and reinsurance, with an entrepreneur- earthquake intensities, temperature levels, cumulative ialpushfromNewYorkinvestmentbankers.Weatherderiv- precipitation over a specified period, and wind speed. atives involveautomaticandimmediatepayouts(typically Some catastrophe bonds have been issued in the United available within 72 hours) on the occurrence of a prede- States and Japan. termined trigger event, irrespective of the scale or nature In some developing countries, some basic form of of damage. Catastrophe bonds provide attractive pay- insurance is often necessary to secure formal sector mentstoinvestors,butifthespecifiedcatastropheoccurs, loans. In Dominica, for example, such policies are the principal or interest (or both) on the bond is reduced required for obtaining mortgages and must cover all or even canceled. Performance of these new mechanisms natural hazards, including windstorms and volcanic is largely uncorrelated with that of other capital market risk. In many other developing countries, catastrophe 53 54 Understanding the Economic and Financial Impacts of Natural Disasters insurance has to be purchased separately and is often require direct verification of loss, and the structure of not specifically required in order to secure loans. The the insurance industry. Each is taken up in turn in this availability of business interruption insurance and agri- section. cultural insurance is also typically severely restricted. Fire and hail insurance for commercial farmers and agribusiness is generally available from the markets, but Affordability securing multiperil crop insurance is often difficult The current and future affordability of any risk trans- except where it is provided under government schemes. fer scheme and the related issue of price instability are Overall, it is estimated that less than 1 percent of the immediate and obvious potential constraints on losses from natural disasters is insured in the world's greater insurance coverage, particularly in developing "poorest" countries (Freeman and others 2002: 2). countries where opportunity costs are higher. (Because In the developed world, most insurance is taken out of the lower capital stocks in these countries, insur- by the private sector. Governments the world over typ- ance premiums compete with the potentially higher ically retain all risks associated with their investments, returns to investments as a use for resources.) including those relating to natural hazards. Whereas Catastrophe insurance premiums can be several times governments in richer nations have the resources to higher than the actuarially determined expected losses absorb risks by adjusting internal funds, their coun- (Froot 1999). This reflects the high variance associated terparts in many developing countries frequently do with expected loss. An insurer needs to have sufficient not. Instead, these governments often take the view that capital to support the risk underwritten and to be able if a disaster occurs, international assistance will be forth- to meet claims should the most extreme event covered coming and that financial risk transfer mechanisms are occur. not required. In reality, this may not always be the Because of their high transaction costs, capital case, nor does this approach guarantee the best response market instruments are even more expensive than in terms of speed or of a government's ability to deter- insurance--perhaps twice as expensive, according mine the allocation of reconstruction funds. Moreover, to some estimates (Swiss Reinsurance Company 1999). aid resources are finite, and natural disasters are plac- Consequently, these instruments have been prima- ing increasing demands on them. The evidence sum- rily purchased by reinsurers to add to their higher- marized in chapter 3 suggests that there is little level capacity, increasing their ability to provide additionality in aid over the medium term, and so dis- coverage against very high losses resulting from extreme aster assistance largely diverts funds from development. events. As a consequence of these factors, there has been a The cost of insurance is negotiated annually (unlike recent growth of interest in exploring ways to support the case of most capital market instruments), and so developing countries in gaining greater access to inter- rates fluctuate widely, often reflecting the annual nationalrisktransfermarkets.Thischapterbrieflyexplores global scale of bills incurred by the industry rather the potential obstacles to greater coverage. It then reviews than more localized factors. In 1992, for example, pre- some solutions that have been pursued to overcome these miums rose three- or fourfold when Hurricane Andrew obstacles and explores the use of risk transfer mecha- generated record claims. Volatility in rates is particu- nisms in loss reduction. The discussion draws, in par- larly high in some regions, such as the Caribbean, where ticular, on evidence from the three country studies. some 80 to 85 percent of gross property insurance pre- miums are transferred to reinsurers. Any fluctuations in reinsurance costs, whether caused by local, regional, Potential Obstacles or global factors, are passed directly on to insurees in the region. Country-specific risk factors seem to come The main obstacles to coverage of disaster risk in into play only when local risks are perceived to be developing countries are affordability, demand, deter- particularly high--often, after disasters occur in quick mination of parametric insurance triggers that do not succession. Financing the Cost of Future Disasters 55 Demand perceived need for private businesses and homeowners to secure their own coverage. Uptake of available risk transfer mechanisms offered in The rapid expansion of microcredit programs has a particular market depends on demand--that is, will- given rise to an important new category of at-risk insti- ingness and ability to pay--and demand depends in tutions, with large portfolios of potentially highly vul- part on the cost (price) of the instruments. The extent nerable clients. Considerable refinancing requirements of risk aversion, which is often partly determined by for microcredit were borne by the Bangladesh Bank as subjective factors, also influences demand. In addi- lender of last resort after the 1998 floods. That disas- tion, income levels play a role in determining demand ter, which threatened to destabilize the country's high- for private insurance, and budgetary constraints affect est-profile initiative in poverty reduction, has generated demand for public insurance.70 considerable interest in incorporating insurance into In developing countries, willingness to pay may be microcredit. Some large microcredit providers, such as influenced by expected flows of external assistance in BRAC and the Grameen Bank, could emerge as signif- response to future disaster events. Some observers see icant insurance providers. But, as Brown and Churchill a problem of moral hazard in this regard because the (1999) state, "[F]or mass, covariant risks that occur fre- international community accepts a contingent liability quently in the same region, such as floods in Bangladesh, in a postdisaster situation. As noted by Freeman and the cost of (reinsurance) coverage will likely be prohib- others (2002: 35), "Insurance is only an effective tool to itively high." These insurance portfolios will require reduce risk if the party concerned is willing to pay for continued acceptance of contingent liability by the the insurance. In the case of catastrophes and develop- Bangladesh Bank and aid donors. Another possibility ing countries, this party would be the affected poor coun- might be some form of risk-pooling arrangement with tries that currently rely on post-disaster assistance." microcredit institutions in other parts of the world, effec- Postdisaster assistance is often highly concessional, while tively spreading risks geographically. catastrophe insurance is expensive. Freeman and others estimate, for instance, that for Honduras, insur- ance for flood and storm risk for all events occurring less Determination of Parametric Insurance Triggers frequently than once in 10 years would cost US$100 million annually. They conclude that the only reason the Derivatives,orparametricinsurance,areintuitivelyappeal- government of a poor country might seek to take out ing. Payouts can be fast and need not be dependent on insurance would be an anticipated reduction in the avail- thecompletionoflengthydamageassessmentprocedures. ability of timely postdisaster external assistance. These instruments do require a careful assessment of Private sector demand for insurance in developing the nature of the hazard faced, however, including suf- countries is also apparently low, at least as measured by ficient high-quality historical data to enable computa- the volume of insurance actually written. Businesses, as tion of probabilities and thus the rate of premium charged. well as private homeowners, may require insurance in To be economically sensible, the trigger event must be order to access financial loans, but the extent to which highly correlated with economic losses, and this requires some form of catastrophe coverage has to be included some understanding of the relationship between types varies among countries, in part depending on awareness of risk and socioeconomic vulnerability--for example, and understanding of levels of hazard risk. Low levels how a particular hazard would affect the production of of coverage in part reflect lower per capita incomes and specific crops (box 5.1). perhaps cultural attitudes as well (World Bank 2002). Commercial,large-scaleagricultureisawell-established In addition, willingness to pay in the private sector may insurance user, as confirmed by the three country case to some extent reflect expectations about likely govern- studies. The challenge is to find simple instruments with ment postdisaster assistance. Government grants may low transaction costs for smallholders that require min- be available to reconstruct homes and businesses, whether imal direct verification. The World Bank has undertaken in a direct form or indirectly (for instance, through tax initial investigations of the scope for providing sup- breaks, loan deferrals, and write-offs), reducing the port to governments to develop such instruments. It has 56 Understanding the Economic and Financial Impacts of Natural Disasters Box 5.1 Insuring banana growers against disaster: the WINCROP scheme Windward Islands Crop Insurance (1988) Ltd. (WINCROP) insures growers of export bananas against damage by "windblows" and tropical storms. The scheme, which covers the entire export crop in Dominica, Grenada, St. Lucia, and St. Vincent and the Grenadines, is owned by the banana marketing organizations in the four countries. In 1999 there were over 12,900 active growers producing 131,000 metric tons, for an average of about 10 tons per grower. The WINCROP scheme only provides coverage against a small proportion (20 percent) of losses, but since own labor and delivery often form a large part of the cost of production, payouts are sufficient to enable growers to reha- bilitate their plantations quickly. Collection of premiums is simple, and payment is assured, since all growers for export are obliged to market through the organizations, which deduct at source. Verification of losses is also easy, as the scheme only covers one crop against one hazard, with known probability of incidence, intensity, and impact on the crop. Losses are assessed by a 5 percent physical survey of affected growers to obtain the proportion of damaged plants. The benefit is based on 75 percent of average deliveries over the preceding three years and a value per plant of about 25 percent of delivery price. The verification system is not strictly the same as those involving parametric insur- ance, but it is similar. The scheme does entail problems of covariant risk because the geographic spread of risks is insufficient. For instance, in 1999, although claims were made against 16 loss events, almost 90 percent of the claims were settled against a single event, Hurricane Lenny, which caused damage to several islands. Furthermore, reserves are not adequate to cover the part of the liability that has not been underwritten. Premium income is, on average, too low because of covariant risk. The premiums reflect the low incomes of the producers and the political difficulty of setting adequate rates. There is partial reinsurance, but it is constrained by costs and, again, by premium income. WINCROP has been unable to extend coverage to other crops or to other businesses on behalf of banana growers. There are legislative restrictions, and the rates quoted by reinsurers have been discouraging. The scheme's monocrop focus could ultimately threaten its viability, as declining banana exports and a squeeze on grower profitability create pressures to keep premiums below a level adequate for covering payouts and the operational costs of the scheme, and as the ratio of overhead rises because of falling exports. also explored the potential for provision of some form In Malawi it would be practically too difficult to estab- of parametric weather insurance to individual farmers lish a trigger event for payouts to smallholders. The sen- and associated traders. sitivity of smallholder agriculture to climatic variability In Bangladesh, however, there are great practical is extreme and may be increasing, and there is consid- difficulties in defining a trigger event. Flooding in erable variability in rainfall patterns within the coun- Bangladesh is hugely complex; it cannot be measured try. Maize, the staple crop, is sensitive to low rainfall but simply by the amount of rainfall at particular weather also to erratic or excessive rainfall. These relationships stations or by river flow or depth. Drought insurance, require further agro-meteorological research. Liberal- too, poses complex problems. The rapid expansion of ization of grain markets has been poorly carried out, irrigation may invalidate the attempt to infer losses from and market imperfections pose risks to food security. historical data. Bangladesh exhibits considerable agro- At present, smallholder food production, much of it hydrological complexity, and even within relatively small for own consumption, is being partially sustained by areas there are likely to be large numbers of both losers programs of free targeted inputs. Efforts are also going and gainers from interacting weather and hydrological into strengthening social safety nets. With the ability conditions. Furthermore, enormous problems having to pay low, and cost recovery prospects minimal, it would to do with landholding titles, sharecropping, and extreme be difficult to distinguish parametric insurance from fragmentation hamper verification and determination postdisaster relief. Even the data on agricultural pro- of entitlements to compensation. Some of these prob- duction have been discredited. These are, for the moment, lems, in particular relating to the determination of trig- unpromising conditions in which to look for a simple ger events, exist in many other countries as well. alternative risk-spreading mechanism. Financing the Cost of Future Disasters 57 Overall, these experiences suggest that the most industry in the Organisation of Eastern Caribbean States promising circumstances are when the hazard is read- (OECS) and has drafted new insurance legislation aimed ily measured (i.e., is of a known intensity), insurance at providing disincentives to small players and encour- is specific to a crop or a livelihood activity, and cost aging amalgamation across countries. recovery is simple, transparent, and assured. Good gov- ernance is a factor that will determine the transparency and credibility of any public scheme with widespread Creative Solutions coverage. The preceding list of potential obstacles to increased uptake of risk transfer mechanisms in developing coun- Structure of the Insurance Industry tries, on both the demand and supply sides, contains Successful insurance requires both that risk is spread few surprises. Many of the problems are well recog- and that insurers are sufficiently capitalized to bear nized, and various ways of overcoming them have potential claims. The country case studies revealed poten- been tried. Some of these involve initiatives on the part tial structural constraints, which may well be common of particular private sector interests to obtain coverage to other developing countries as well. at more favorable prices. In other cases, governments In Dominica and in the wider Caribbean area, there and international organizations, in particular the World are concerns about the efficiency and underlying strength Bank, have sought to promote increased utilization of of the insurance industry, relating to the proliferation risk transfer mechanisms and so provide alternative of property and casualty insurance players in the region. sources of disaster financing (as has been discussed There is apparently strong competition for property above in relation to weather derivatives). insurance, motivated by the desire to capture reinsur- In the private sector, one of the more common ance commission revenues. The widespread competi- responses to the high cost of insurance premiums has tion for direct fees, however, discourages primary been to establish disaster reserves. In some countries, domestic insurers from accumulating reserves and, governments deliberately encourage this development together with tax disincentives on the sector, results in through tax breaks. Some governments also make annual a high-dividend-paying industry, high dependence on budgetary allocations to some form of calamity fund, foreign reinsurance, and continued fractionalization. which could at least limit budgetary reallocations in The sharp rise in reinsurance premiums in recent the event of a disaster. For countries experiencing local- years has led to higher commissions, attracting even ized hazards every year, such practices are highly pru- more insurers and agents into the Caribbean market. A dent, helping to strengthen broader financial planning World Bank report (1998b: 20) states that "the prolif- and fiscal discipline.71 eration of small insurers is cause for concern regarding Logically, a related measure is to try to purchase insur- efficiency . . . but even more regarding safety. Are these ance for losses only in excess of a particular level, or small companies sufficiently capitalized for the 15% of for a specific layer of coverage (that is, for losses exceed- the risk they retain? Are they sufficiently careful in choos- ing $x but not for losses over $y, where x is less than ing reinsurers that can be relied upon to pay up their y). In years of more severe disaster, substantially larger 85% share? Regulation in this sector needs to be sub- resources may be required, beyond the scope of calamity stantially strengthened . . . ." The report suggests, among funds, but by covering restricted levels of loss them- other things, that tougher standards, including an abil- selves, businesses or governments can secure some ity to cover maximum probable losses consistent with reduction in premiums. Some larger and special-risk international industry practice, are needed if domestic categories in the Caribbean, such as power utilities, have companies are to improve their safety, particularly given been unable to obtain full, affordable insurance in recent the stochastic nature of catastrophic events. To this years. Some of them have therefore voluntarily devised end, the Eastern Caribbean Central Bank (ECCB) has high self-insurance deductibles, seeking insurance only reviewed the regulatory framework of the insurance against higher levels of loss (World Bank 2000). 58 Understanding the Economic and Financial Impacts of Natural Disasters Risk-spreading within a group is another solution, supplemented by various insurance and capital market and one that has been tried or at least explored in var- instruments. ious guises. For example, members of the Caribbean Hotel Association have created a risk management firm for their own exclusive use. It is based on a prob- Promoting Mitigation able maximum loss (PML) profile of members' proper- ties which indicated that risks were sufficiently diversified Risk transfer mechanisms can be used to help promote to allow a regional insurance company to survive a 1.3 risk reduction--that is, to modify the behavior of percent probability of a major storm. In Dominica, how- those insured. As noted in IADB (2001: 11), "insur- ever, the historical record suggests a substantially higher ance is an important part of risk management strategy, incidence--at least a 4 percent a year probability of a and should be promoted not for its own sake, but because direct hit by a Category 4 hurricane. it can be a powerful tool to promote risk awareness Also in the Caribbean area, as described in box 5.1, and enforce risk mitigation measures." banana growers' organizations in four countries have The issuance of catastrophe insurance policies can pooled risk from storms through the Windward be made conditional on the implementation of specific Islands Crop Insurance (WINCROP) scheme, which is loss reduction measures and on adherence to building extremely important in providing a risk-spreading mech- and land use zoning codes. Catastrophe coverage may anism for the growers. The success of group or club be required before a business or home loan can be solutions such as WINCROP suggests a model for micro- obtained, to ensure that the loan can be paid off in the credit organizations: the scheme is transparent to mem- event of damage to or destruction of the business or bers, and the group addresses the problem of moral property. Weather derivatives also indirectly encour- hazard by obliging all to participate and by exerting peer age actions to minimize losses. If the trigger event occurs, pressure. But WINCROP is a highly context-specific those covered will receive payouts regardless of levels mechanism that is confined solely to wind damage to of loss, and thus it is in their interest to minimize bananas, and extension of coverage to a wider range of losses. risks has proved difficult. In reality, the record on the use of insurance in pro- There has been debate about the creation of a more moting risk reduction is rather disappointing. In general regional risk management tool for the Caribbean Bangladesh, for example, premium reductions are cur- area, possibly involving some form of contingent fund- rently not formally offered for measures to reduce ing. In a recent initiative, the World Bank developed a risk, although reductions can be negotiated case by proposal for the East Caribbean that favors an inter- case. country insurance-pooling arrangement. The arrange- In Dominica catastrophe coverage (covering all nat- ment would aim to utilize reinsurance and risk-financing ural hazards) is mandatory for securing a mortgage. A resources more effectively by reaping economies of scale differential premium structure also exists to some extent, and improving capacity to accumulate and retain cap- with at least some companies offering discounts for ital reserves. In the earlier years of the pool, its full cap- hazard-proofing. More widespread discriminatory pric- italization would require guarantee financing, a ing practices, in Dominica and the Caribbean area at contingent line of credit for quick disbursement from large, are discouraged by low retention of risk combined a multilateral institution, or, alternatively, a long-term with the reinsurance industry's blanket-pricing policy. bond issue in the capital markets. Large geographic areas are placed in the same PML cat- Other efforts to help promote insurance have involved egory, "without regard for the topographical features support to governments to establish some form of manda- and structure resistance distinctions propounded by tory insurance, as in Turkey after the Marmara earth- regional and international experts" (World Bank 2000: quake.72 Several countries are also reported to have 57­58). Meanwhile, individual insurance companies are expressed an interest in establishing some form of manda- reported as fearing that significant premium discounts tory catastrophe insurance that would involve pools for their better-protected risks cannot be balanced by Financing the Cost of Future Disasters 59 surcharging poorer risks (OAS 1996). That being so, considerable scope for greater use of risk transfer tools insurance companies (with one notable exception) in promoting mitigation. typically follow the reinsurance lead--in effect, doing Some practical obstacles have to be overcome little to promote hazard mitigation in the region.73 More before insurance coverage can be increased substan- widespread discriminatory premium pricing would tially. In a number of developing countries, there is a require comprehensive hazard-mapping, as well as need to reform the structure and the legal and regula- inspections, to determine the vulnerability of indivi- tory framework of the insurance industry, including dual properties. Fiji is an apparent exceptional case removal of barriers to entry. The cost of insurance has among developing countries because of its relatively to be both affordable and stable. At the same time, insur- high rate of cyclone insurance coverage in urban areas.74 ers must remain sufficiently capitalized to bear any losses, which will require detailed scientific information on current and future risks. This is certainly a challenge, Conclusions given the uncertainty about the precise implications of climatic change for the incidence and severity of haz- To date, insurance and capital market instruments ards and the evidence on increasing vulnerability, as in have played a relatively small role in the transfer of risk Malawi. And making insurance available to the poor is in developing countries. Risk transfer mechanisms seem not simply a matter of affordability; there are adminis- to work best at middle-range risk frequencies (World trative and legal problems, too, involving, for example, Bank 2002). High-frequency hazards require annual land titles. allocations of funds to meet their anticipated conse- The case of Montserrat, although extreme, serves as quences, as well as mitigation and preparedness a reminder that international insurers can withdraw at efforts. At the other extreme, there is probably little short notice should the perceived risks become too great. demand for insurance against geophysical events that The insurance sector is a for-profit business, not a provider are likely to occur centuries apart. There is certainly of a public good. In addition, globalization is changing merit, however, in exploring the potential scope for risks but not necessarily reducing them.75 the increased use of risk transfer mechanisms for well- In view of all the obstacles, creative solutions need defined hazards in the middle-risk range, such as extreme to be sought and tested. Application of some combi- tropical storms, especially in view of the significant pres- nation of tools for different layers of loss coverage is a sures that such disasters can place on government, donor, possible way forward. The bundling of various types of and private resources. Indeed, interest in financial risk hazard under one contract could also reduce premium transfer mechanisms is growing as appreciation of the rates to the extent that better-understood risks such as potential benefits increases. cyclones are bundled with less well understood ones A larger challenge that is just beginning to be addressed (e.g., earthquakes). In some situations, as in rural Malawi, is how to make insurance and related instruments avail- more conventional administrative, publicly funded, sub- able to the poor. Insurance is being explored as a sidized livelihood programs and safety nets are still the means of protecting local-level saving and microcredit appropriate way to tackle growing vulnerability and schemes, such as those run by NGOs in Bangladesh. provide basic food and health security. Perhaps the starting point for microcredit insurance is The success of insurance and risk transfer options coverage against idiosyncratic shocks such as ill health assumes a favorable financial environment in which and the death of the borrower. The covariant risk asso- transparency is the norm and there is confidence that ciated with natural hazards makes refinancing of losses contracts will be honored and that continuing arrange- difficult to avoid, and continued acceptance of contin- ments will not be prejudiced by short-term political gent liability by public bodies and aid organizations will considerations and rent-seeking behavior. Institu- be required. This is especially so where a microcredit tional stability is required so that insurance and other scheme is already expanding with the support of con- schemes can be expanded and tested in practice. These cessional or grant funds from aid sources. There is also preconditions raise more general issues of governance. 60 Understanding the Economic and Financial Impacts of Natural Disasters Developing country governments need to acknowl- disaster-related external assistance may not be addi- edge explicitly that a decision not to make risk man- tional but instead may displace funding for develop- agement a priority issue is a policy stance on the handling ment. Moreover, the costs of disasters are rising, while of risk in society and the economy (Freeman and others aid resources are, at best, stagnant. Governments need 2002). At first glance, it might appear that in the event to be persuaded to take a medium-term perspective on of a disaster, additional external assistance will be forth- appropriate risk management, including financial plan- coming. As this study has shown, however, ultimately ning for future catastrophes. Chapter 6 Findings of the Study and Implications for Policy and Research The broad objective of this study has been to increase encourages adaptation in economic and social activity. understanding of the wider economic and financial Historical climatological and hydrological records allow impacts of natural disasters through the detailed formal assignment of risk to inform investment and pro- analysis of three countries. Factors that determine the duction decisions that are then likely to favor less haz- vulnerability of hazard-prone economies have been ardous locations and risk-spreading and to encourage described, opportunities for improving the management disaster mitigation. Geophysical hazards, by contrast, of risk have been noted, and obstacles that inhibit the are mostly low-probability, seemingly random events, adoption of such measures have been identified. This and their risks are often almost wholly discounted. Yet final chapter reviews the findings of the study and these risks need to be recognized in highly exposed their policy implications and highlights areas that merit countries and regions because their potential costs are further investigation. rising exponentially with economic development. 3. Within the timescale of this study (20­35 years), environmental change has not had a discernible impact on hazard risk, as measured by the frequency Findings and intensity of extreme events likely to have severe economic consequences. In particular, the widely antic- The findings from the study are broken down into the ipated effects of global climatic change are not yet appar- macroeconomic consequences of natural disasters, their ent in the hazard data. budgetary impacts, the contribution of hazard infor- 4. The vulnerability of a country to natural hazards mation generation and dissemination in reducing the is determined by a complex, dynamic set of influences effects of disasters, and options for the financial trans- relating to factors such as economic structure, stage of fer of risk. development, and prevailing economic and policy conditions. Where vulnerability has been reduced, this improvement can be linked to appropriate invest- Broad Macroeconomic Consequences of Disasters ments in disaster mitigation and favorable developments 1. Major natural disasters can and do have severe in the structure of the economy, in production tech- negative short-run impacts. Disasters, especially when nologies, and in the wider economic and domestic policy they occur frequently, also appear to have negative longer- environment. Similarly, negative trends, notably the term consequences for economic growth, development, increasing sensitivity of southern Africa to climatic vari- and poverty reduction, although these effects are more ability (encompassing not just drought but also erratic difficult to isolate and quantify. rainfall, extremely high rainfall, and related floods), 2. The two broad categories of hazard--hydro- reflect a complex mix of factors. meteorological and geological--are associated with dis- 5. Catastrophic events can induce conscious responses, tinct patterns and forms of economic vulnerability, in part relating to technical progress, policy changes, and insti- reflecting differences in their frequency of occurrence. tutional innovations, that may ultimately increase an The recurrent nature of hydro-meteorological hazards economy's resilience. 61 62 Understanding the Economic and Financial Impacts of Natural Disasters Budgetary Impacts The Contribution of Information to Disaster Reduction In view of the lack of existing research data on the fiscal consequences of disasters, these impacts, includ- 1. The international public good framework is help- ing effects on public expenditure and revenue, were ful for exploring the issues surrounding the generation carefully examined. of knowledge and the dissemination of information 1. Natural disasters can create significant budgetary about hazard risks. pressures, with potential narrow fiscal impacts in the 2. Knowledge is most likely to be generated, and short term and wider long-term implications for devel- monitoring to be sustained, where the hazard is recog- opment. nized as a global issue, as is the case for climatic vari- 2. The behavior of broad fiscal aggregates such as ability and extreme weather events such as tropical total recurrent and capital expenditure, revenue, and cyclones. The adequacy of information on geophysical the budgetary deficit can be misleading because it may hazards is less assured because the risks are regional or suggest that disasters have little discernible impact. even country-specific. This is especially so where no (Drought, however, does have severe impacts on the disasters have occurred recently. macroeconomic performance of economies largely dom- 3. There are encouraging examples of improvements inated by agriculture, as is seen in low-income Sub- in the use of hazard information. Concerns about Saharan Africa.) The apparent general insensitivity to global climatic change have helped ensure interest in cli- hazards reflects the successful efforts of many coun- matic processes and funding of research on them. Progress tries to remain within already established budgetary in reducing the impacts of tropical storms is impres- envelopes. Disasters often result in widespread, if largely sive. There has been institutional learning about the need nontransparent, immediate and interannual realloca- to ensure wider dissemination of information. tions of resources. 4. Nevertheless, the use of scientific information in 3. Reallocation is the primary fiscal response to a dis- public and private sector risk assessment is unsatisfac- aster. The study found that: tory in many cases. This has partly to do with the · The brunt of financial reallocations appears to probabilistic nature of the information generated by sci- fall primarily on capital expenditure and social entific research and monitoring. For example, there is sectors. a danger of creating excessive expectations about the · There may be considerable in-kind reallocation of precision and potential uses of climatic forecasting, human and physical resources within recurrent specifically in relation to crisis-affected southern Africa. expenditure. At this time, weather forecasts can prompt governments · With a few exceptions, reallocations are typically to move to a higher level of preparedness but cannot poorly documented and cannot be easily quanti- effect significant changes in agricultural production. fied. · The conditions under which decisions on post- Risk Transfer disaster reallocations of budgetary resources are made are usually far from ideal. 1. Insurance and capital market instruments have 4. In all eight countries studied by the authors (see played a very small role to date in the ex ante transfer chapter 1), it has been impossible to ascertain actual of catastrophic risk in developing countries, despite levels of expenditure on mitigation and preparedness increasing interest in these mechanisms. or on postdisaster response. 2. There are potentially significant obstacles to greater 5. The available data on external assistance suggest coverage in a number of developing countries, in part that disasters have little impact on trends in total aid relating to affordability, attitudes toward risk, and the flows. Many donors appear to respond to disaster current structure of the insurance industry. crises by reallocating resources and advancing com- 3. The record on the use of insurance in promoting mitments within existing multiyear country programs disaster mitigation appears rather disappointing in many and budget envelopes. developing countries, with a few notable exceptions. Findings of the Study and Implications for Policy and Research 63 Policy Implications As part of this process, there should be adequate invest- ment in risk-mapping, monitoring, assessment, and dis- The findings from the study have implications for national semination, and information should be provided in an development and macroeconomic policies, public easily understood and usable form (see "Knowledge and finance, the generation and use of information, and the Information as Public Goods," below). Services sec- financing of disaster costs. A precondition for improve- tors, including financial institutions, as well as pro- ment in all these areas is good governance. ductive sectors, should be included in initiatives to promote adequate risk management. 5. Postdisaster reconstruction needs to be better National Development Policy and the Macroeconomy planned and carefully orchestrated to exploit potential organizational, technological, and other improve- Natural hazards warrant more serious consideration in ments that could be made in rebuilding an economy the formulation of national economic policies and strate- while keeping priority development objectives on track. gies. Risk assessments should be made from a broad Governments should consider preplanning possible macroeconomic standpoint, exploring areas of both sen- reconstruction and rehabilitation programs on the basis sitivity and resilience. Assessments should seek to under- of disaster scenarios and, within that, identifying crit- stand the underlying factors determining vulnerability, ical projects that should receive priority in postdisas- including the potentially complex and dynamic inter- ter funding. There is a case for exploring economic policy linkages between various influences and the scope for options through disaster scenarios that include the likely risk reduction. Vulnerability has to be assessed accord- effects of fiscal changes and monetary measures in order ing to the particular type of hazard. to develop guidelines for policymakers in responding 1. Regular assessment of hazard risk is required to to disasters (see "Public Finance," below). ensure that risk management strategies are appropriate. 6. National or economywide disaster impacts, includ- From a macroeconomic perspective, vulnerability can ing total financial losses, should be reassessed as a matter shift quickly, particularly in countries experiencing rapid of course 12 to 18 months after an event. This task might growth and socioeconomic change, including urban- be undertaken as part of an end-of-project report for a ization. recovery loan or in a paper for consideration at an annual 2. For geographically large countries, where nation- consortium or roundtable meeting. wide disasters are rare, regional analysis is potentially 7. Lessons learned from particular disasters need to more appropriate than country-level analysis for under- be understood and, where necessary, acted on. Disas- standing vulnerability and designing relevant policies. ters can induce policy changes and institutional inno- 3. Management of natural hazard risk should be inte- vations that are ultimately beneficial, not only in reducing grated into the broad development process, as part of vulnerability but also in supporting economic growth the determination of priorities, policies, and strategies. and development. This requires an assessment of vulnerability from a macroeconomic perspective that distinguishes between potential short- and longer-term impacts and takes Public Finance into account knock-on or multiplier effects. Special efforts need to be made to minimize any adverse impacts Expansion of the government budget may be appro- of disasters on priority policy areas, such as poverty priate as a disaster response where aggregate demand reduction (see box 2.2). is likely to be depressed by the impacts of the disaster. 4. Risk management necessarily involves the private This is most obviously so for a drought shock. But the sector and civil society, as well as the public sector. The domestic and international sources of funding for private sector should be encouraged and supported in additional public expenditure have to be carefully and enhancing its understanding of natural hazard risks and explicitly considered to avoid crowding out private sector adopting appropriate risk management tools. Both struc- demand, and financially destabilizing and inflationary tural and nonstructural measures may be required. pressures need to be guarded against. Where a disaster 64 Understanding the Economic and Financial Impacts of Natural Disasters causes great damage, which might quickly stimulate a disasters. Low-income countries in general are at their reconstruction boom, but disruption of economic activ- most vulnerable, financially and economically, to dis- ity is limited, a more cautious approach, largely reliant aster shocks when there are problems of governance on temporary reallocations, is indicated. and poor fiscal and monetary management. The inter- 1. A disaggregated approach should be taken in review- national community needs to be especially alert to poten- ing the public finance implications of disasters. tial disaster in such circumstances. 2. Postdisaster reallocations, disaster relief expendi- ture, and rehabilitation spending need to be closely Knowledge and Information as Public Goods monitored and reported. In particular, the consequences for income distribution of such decisions need to be As in all areas of public action, the availability of good- carefully assessed and the implications for pro-poor quality, trustworthy data is a necessary condition for policies analyzed. effective management of natural disaster risk. The gen- 3. Postdisaster reallocation of budgetary resources eration of knowledge about hazards and the sustaining should be made through a formal process, in the con- of global and regional information systems bearing on text of a careful strategic review. Four elements are risks must be international priorities. The interna- required to ensure that the process is rational and cost- tional community should also concern itself with the effective: a clearly defined and applied policy frame- adequacy of funding and the quality of complemen- work; an already established system of prioritization for tary programs for risk monitoring and dissemination investment projects; up-to-date and reliable informa- of information in both low-income and extremely hazard- tion on the current demand for and availability of vulnerable middle-income countries. Special attention resources; and a considered assessment of the broader should be given to the risk exposure of poor people macroeconomic impact of the event. and communities. On the country level, risk informa- 4. Better documentation of expenditure on all aspects tion, as a national public good, should be identified as of ex ante risk management is needed, together with a thematic issue in periodic public expenditure reviews. data on the postdisaster response and the results of 1. Natural hazard risks require regular (perhaps five macroeconomic vulnerability assessments. This infor- times a year) reappraisal because of the interacting influ- mation will help inform governments and the interna- ences of socioeconomic and environmental change on tional community about appropriate spending levels on vulnerability. mitigation and preparedness and areas that might be 2. Much more can be done to make better use of sci- underfunded (or overfunded). Relevant activities and entific information on natural hazards in setting public related expenditure should be tagged. policy at the country, regional, and international 5. Risk management concerns need to be integrated levels. This is illustrated by examples of climatic fore- with the annual budgetary process and appropriately casting, volcanic hazard monitoring, and riverine reflected in the allocation of financial resources. flood warning. 6. Where disaster-related expenditure occurs regu- 3. Some recommendations can be made for increas- larly, the use of predesignated calamity funds should ing the usefulness of climatic forecasting: be explored. · Forecasting should be focused on climatic vari- 7. Greater use of financial risk transfer instruments ability more broadly, not just drought. This requires is required to help meet the cost of larger rehabilitation more research, downscaled to zonal levels, and programs, alleviating some of the pressure on budget- intraseasonal timescales. ary resources. · More specific information should be derived from 8. Because of a combination of limited administra- real-time monitoring on the weather situation as tive capacity, the effects of the HIV/AIDS pandemic, and it evolves. This would be of value to those user exceptionally severe budgetary constraints, Sub-Saharan groups that can quickly modify their actions, such African governments may require special support in as commercial farmers and managers of food secu- planning for and dealing with the consequences of rity stocks and water resources. Findings of the Study and Implications for Policy and Research 65 · Greater agronomic-meteorological collaboration basis of funding. In some situations, more conventional would help national and international institutions administrative, publicly funded, and subsidized pro- make more effective use of forecast information grams to protect livelihoods and provide safety nets in their food security and agricultural support are still the appropriate way to tackle extreme and pos- decisions. sibly growing vulnerability. This protection is likely to · More rapid reporting of intraseasonal variability be part of the provision of basic food and health secu- would help enable faster responses to an evolv- rity by government and civil society organizations, with, ing situation. as appropriate, international support. Some specific sug- 4. Volcanic hazard assessment has a strongly regional gestions can be made for achieving a wider funding basis and national focus and so lacks an international insti- for meeting disaster costs. tutional framework comparable to that for meteoro- 1. The increased use of risk transfer mechanisms for logical issues. International assessment of the adequacy well-defined hazards, particularly in the middle-risk of regional and national monitoring systems is required, range, should be promoted. Efforts to design initiatives along with, perhaps, supervision. and instruments that would support greater uptake 5. Flood hazard warnings for major regional river should be continued. systems pose special problems. Such information is 2. The application of a combination of mechanisms not an international public good because of the poten- for different layers of loss coverage is likely to be con- tial tradeoff between the interests of stakeholders in dif- structive. Options for risk sharing should be consid- ferent parts of the river system. Some recent disasters ered as one of the potential mechanisms. have highlighted weaknesses in monitoring and hazard 3. Strategies for insurance should be used to promote warning, especially failure to include both upper and risk reduction more effectively. For example, the issuance lower riparian levels. International efforts to ensure of insurance policies for buildings and equipment or development of the knowledge base on hazards in major for business interruption should be made conditional river systems need to be sustained. Special efforts are on the implementation of specific loss reduction meas- required to strengthen regional cooperation, as well as ures and on adherence to building and land use zoning cooperation at the national level between central and codes. Premium discounts could be offered for better- provincial authorities. protected risks. Insurance providers and governments 6. Hazards that are likely to be entirely within a coun- (the latter through legislation and promulgation of land- try, especially geophysical hazards, landslides, and flash zoning and building codes) have complementary roles flooding, require separate attention because national to play. hazard information systems in developing countries are 4. Fresh initiatives to make insurance and related likely to be underfunded. instruments available to poor households and com- 7. Governments in geophysical high-risk zones should munities are needed and should be actively encouraged be encouraged to develop hazard plans that include ex and appropriately supported by governments and inter- ante risk reduction measures, preparedness, and post- national institutions. It should be recognized that the disaster reconstruction scenarios. central bank or donors may have to accept some form of contingency liability. Options for risk-pooling arrange- ments with other microcredit institutions, perhaps across Financing Future Disaster Costs borders, should be explored. Governments need an appropriate risk management strategy that includes ex ante financial planning for Good Governance future disasters over an 8­10-year time horizon, beyond the normal medium-term planning period of 3 to 5 Many of the policy recommendations in this chapter fit years. Disaster-related external assistance for immedi- within the broader framework of initiatives to promote ate relief or reconstruction cannot be assumed to be good governance. There is clear evidence that weak- additional, and the strategy should seek to extend the nesses in governance have contributed to heightened 66 Understanding the Economic and Financial Impacts of Natural Disasters vulnerability to disasters. Four aspects of good gover- An extremely worthwhile project would be to try to nance are highlighted that would contribute to disas- integrate these different approaches into an exemplary ter reduction: investigation for a highly disaster-prone country that · Efficient and accountable public sector institutions, would explore the longer-term economic and poverty including strong public expenditure management implications of hazard vulnerability. For such an inves- systems--for example, better reporting of relevant tigation to be feasible within a realistic budget and expenditure and postdisaster reallocations, and time frame, relatively good time-series data on haz- enhanced capacity to respond effectively and effi- ards, income distribution, poverty, and the economy ciently to disasters while also protecting priority areas would have to be already available. The study would of government spending need to be able to draw on substantial intellectual invest- · "Good" policymaking, which includes taking natu- ments in exploring longer-term development and poverty ral hazard risks into account in policy and strategy processes at both the microeconomic and macroeco- formulation and seeking to reduce vulnerability at nomic levels. If the relative strengths of different com- both microeconomic and macroeconomic levels putable models for exploring future disaster scenarios · Predictability of policies through long-term com- are to be compared, these models would also need to mitment to mitigation and preparedness, including be available for use. investment in and maintenance of structural miti- The implications of disasters for poverty include gation measures, sustained expenditure on scien- not only direct impacts on affected people and com- tific monitoring and forecasting, and continued munities but also effects transmitted through the impacts commitment to an optimal bundle of mechanisms of disasters on the wider economy. Further research is for transfer of financial risk needed that combines both micro-socioeconomic and · Elimination of corruption (or, more realistically, macroeconomic analyses. More specifically, investiga- curtailment of rent-seeking behavior) via, for exam- tions should include: ple, the introduction and enforcement of appropri- · Analysis of the regional or provincial impacts of nat- ate land use and building codes. ural hazards in geographically larger countries and how these impacts are felt in the national economy · Analysis of the medium- to longer-term impacts of Directions for Future Research geophysical events, extending up to several decades beyond particular events, to explore to what extent It was recognized at the outset that this study would be and in what ways such shocks may have positive investigating a complex and multifaceted subject and effects and to support efforts to ensure that any poten- would probably highlight many areas worthy of further, tial benefits are maximized separate examination outside its own scope. The study · Examination of the implications of deregulation has, indeed, identified several such areas, as outlined and globalization for levels and forms of vulnerabil- next. ity, the sectors and socioeconomic groups that require particular support, and potential opportunities that could be exploited for economic gain. Macroeconomics of Disasters Three methodological approaches are commonly used Public Finance to investigate the macroeconomic impacts of natural disasters: narrative case studies, quantitative model- The budgetary impacts of natural disasters are com- ing, and cross-sectional statistical analysis. Given their plex and multifaceted, and this study has only begun strengths and limitations, these approaches should to scratch the surface. A number of areas are worthy of be regarded as complementary rather than alternative further investigation. In many cases the findings ways of exploring the complex consequences of natu- would help support broader initiatives to promote and ral disasters. sustain good governance and to manage natural disasters Findings of the Study and Implications for Policy and Research 67 more effectively. The suggested subjects for further that advance global poverty reduction goals and devel- research are: oping country priorities. The recommended studies · Real-time studies of a government's response to a dis- include: aster, documenting how key decisions are made and · Review of the adequacy of support for research on how the crisis is handled through the recovery and monitoring of natural hazards as a requisite phase and drawing lessons for future occasions. part of every country poverty reduction strategy or This proposal is similar to the "learning support office" plan. concept developed for humanitarian operations by · Review of the patchy performance of recent efforts ALNAP (2002). to strengthen national information systems, espe- · Detailed analysis of the budgetary impacts of disas- cially through external technical cooperation. This ters in Sub-Saharan Africa to explore why the fiscal review could provide a better basis for international impacts of such disasters are more readily discernible assistance. there than in other countries, in order to support · More systematic research into the relationship between improved fiscal management of climatic variability. climatic variability and economic performance, espe- · Detailed investigation of the impact of disasters on cially the performance of the rural economy in low- autonomous and semiautonomous agencies. Improved income countries. The research should cover knowledge and understanding of such impacts, which relationships, in southern Africa and probably Sub- are not apparent in central government budgetary Saharan Africa, between crop performance and reports, are important for the establishment of finan- weather conditions--not only drought, but also erratic cially sustainable public risk management policies. rainfall and very high rainfall. Climatic change is · Analysis of risk management at provincial and local expected to bring increased rainfall to some areas, government levels to explore the balance between and the consequences are not necessarily unam- the cost of responsibilities and the availability of biguously beneficial. Such investigations are there- funds, as well as the implications for regional dis- fore needed to better inform assessments of the parities in the standards and level of provision of serv- consequences of global climatic change on both ices and infrastructure. regional and country levels. · Examination of the relationship between disasters and aid flows, exploring the extent to which post- Financing Future Disaster Costs disaster response entails additional aid resources, the distribution of resources between relief and rehabil- Insurance and capital market instruments have played itation needs, and the effectiveness and efficiency of a relatively small role to date in the transfer of risk in financial instruments such as the IMF's compensa- developing countries. There is merit in continuing to tory financing facility and the European Commis- explore the possibilities for their greater use, particu- sion's (EC's) post-Stabex, Cotonou, supplementary larly in view of the significant pressures that such dis- grants, as compared with conventional project-type asters can place on government, donor, and private relief and rehabilitation assistance. resources and, ultimately, on sustainable development. Topics for research include: · The potential scope for increased use of risk trans- Information and Disaster Reduction fer mechanisms for well-defined hazards in the middle- The benefits and costs of knowledge generation for each risk range, such as extreme tropical storms of the main types of natural hazard deserve fuller · Ways of making insurance and related instruments investigation. The object is to determine priority infor- available to the poor, with attention to the role of mation areas for support as international public goods microfinance institutions. Appendix A Dominica: Natural Disasters and Economic Development in a Small Island State The study of Dominica focuses on hazard events since Layou River Valley landslide in March­November 1997, Independence in 1978. During this period, the coun- can have substantial socioeconomic impacts, as well as try has experienced major hurricanes, tropical storms, the potential for loss of life. Although there has been and a landslide. A volcano alert has been in place since only one volcanic event in Dominica's recorded his- 1998. tory, the island is now in a period of increased seismic The Commonwealth of Dominica is a small, lower- activity. The risk of volcanic activity is relatively high, middle-income Caribbean island state with an estimated particularly in the south, where 20 percent of the pop- population of 76,000, 30 percent of whom were esti- ulation lives, and there is a related risk of earthquake. mated in 1996 to be living at or below the poverty line. Droughts, storm surges, floods, bush fires, and tsunamis There is a long-established pattern of net out-migration. have been regarded as lesser hazards, but any of these The island is of volcanic origin and is the most moun- could assume major proportions if physical conditions tainous of the eastern Caribbean islands, with dense or social activity altered levels of vulnerability. vegetation, very high rainfall, and many narrow, There is much uncertainty about natural hazard risks, deeply incised river valleys and steep ridges. Slopes of which need to be regularly reappraised. For example, 30 degrees or more are found in at least 60 percent of Hurricane Lenny caused extensive unexpected damage the island. Reflecting this topography, most of the pop- in 1999 when the storm tracked west to east, 150 ulation and infrastructure is concentrated on the coast miles from the island. No hazard warnings were issued and is vulnerable to strong winds and high seas. Dominica because the "normal" approach is from the east, and at is a member of various regional organizations created the distance involved, a hurricane would not "normally" to promote cooperation in trade and economic devel- affect Dominica. opment. Preferential trading arrangements for bananas with the United Kingdom and the European Union have Environmental Impacts of Natural Hazards and been a significant factor in the island's development. Global Climate Change Natural Hazards and Disasters The scope of this study does not extend to natural haz- since Independence in 1978 ards as part of continuing geophysical processes. The analysis does, however, explore some aspects of the envi- Dominica lies well within the Atlantic hurricane belt, and ronmental impact of natural hazards and the possible the most common and historically most significant of the effects of environmental and global climate change on natural hazards facing it are tropical storms and hurri- vulnerability, because of the economic and financial canes. Since 1978, the island has suffered the damag- implications. ing effects of seven hurricanes--three of them Category Dominica's increasingly important ecotourism sector 4 storms with sustained wind speeds in excess of 210 is closely tied to its image as the "Nature Isle of the kilometersperhour--andthreetropicalstorms.Themost Caribbean," with a wealth of flora and fauna, rich marine extreme disaster event since 1978 has been Hurricane biodiversity, an expanse of forest area, and surface David, in 1979. Landslides are common and, like the evidence of its volcanic origins. This environmental 69 70 Appendix A resource wealth is vulnerable to natural hazards. Hur- Economic Performance and Natural Hazards ricane David damaged an estimated 5 million trees, and Analysis of annual fluctuations in Dominica's agricul- studies indicate that species populations have taken 20 tural, nonagricultural, and total GDP alongside the inci- years to recover from its impact. Hurricanes and trop- dence of major disasters demonstrates the important ical storms can also accelerate erosion, damaging beaches impact of natural hazards on the island's economic and reefs. Human activity--for example, forestry man- performance since 1978 (see figure 2.1). This influ- agement practices or the excavation of deltaic sites for ence is confirmed by the results of a more formal regres- building materials--may increase the vulnerability of sion analysis. The combined impact of Hurricanes David, environmental assets to natural hazards. There are also Frederick, and Allen in 1979 and 1980 was particularly concerns that a rise in global air and sea temperatures devastating because of the already weak economy, the in the Caribbean Basin could substantially damage the scale of physical damage and disruption, and the result- marine environment and give rise to more frequent ing budgetary pressure. Real GDP plummeted by 17 and more powerful hurricanes. Environmental change percent in 1979, while agricultural sector output alone in the short term may be so small as to be barely per- fell by 32 percent in that year and by another 2.1 per- ceptible, but over a period of several decades, the cumu- cent in 1980. Hurricane David was a significant factor lative effects could be considerable. in forcing the country into a structural adjustment Enhanced environmental monitoring is needed to program (SAP) that was aimed at achieving a sustain- obtain robust baseline data and to measure the precise able rate of economic growth, lower unemployment, ecological impacts of natural hazards and climatic change better living standards, and strengthened fiscal and bal- on resources such as forests and fauna and thus pro- ance of payments positions. Relatively high rates of vide the impetus for measures to reduce vulnerability. growth between 1986 and 1988, averaging 7 percent In the meantime, according to expert advice, there are per year, reflected the success of the SAP, as well as rapid measures that need to be taken--for example, in rela- increases in banana prices and in volumes of exports tion to forestry management--to reduce the vulnera- and high levels of aid flows. But Hurricane Hugo in Sep- bility of important environmental assets to future tember 1989 destroyed about 70 percent of banana pro- hurricanes and storms. duction, and an unfavorable shift in the exchange rate reduced the value of banana exports. Overall, GDP fell year-on-year by 1.1 percent in 1989, with agricultural Macroeconomy GDP alone dropping by 14.6 percent. (Nonagricul- tural GDP rose by 4.4 percent.) The visible trade Dominica has a small, very open economy that is still deficit increased to the equivalent of 38.5 percent of heavily reliant on a single export crop, bananas, which GDP. During the 1990s, GDP growth averaged 2.4 per- represented a third of total merchandise export earn- cent a year, the weaker performance reflecting difficul- ings in 1997. Although the agricultural sector's share ties in the banana industry and adverse weather of GDP fell from 37 percent in 1977­78 to 20 percent conditions. The continued growth of GDP was attrib- in 1997­98, it remains the main productive sector and utable to compensatory expansion of the manufactur- is the principal source of livelihoods. Despite limited ing and service sectors and to the reduced share of growth in other private sector activity since the mid- agriculture in GDP. 1970s, the private sector remains small. Between 1977 and 1998, manufacturing output rose from 3.9 to 8.2 Economic Development Strategies percent of GDP, and there has been promising growth in the offshore financial services industry. Along with Among the world's countries, Dominica is one of the tourism, which by the late 1990s accounted for an most vulnerable to both natural hazards and other exter- estimated 35 percent of external earnings, manufac- nal shocks. It is vitally important that risk reduction turing and financial services have helped meet the sub- become an integral part of policies for achieving social stantial deficit in the visible external trade account. and economic stability. This is true, as well, for other Dominica: Natural Disasters and Economic Development in a Small Island State 71 similar small island developing states. Achieving risk and commercial sectors of the economy. Critical fac- reduction remains one of Dominica's biggest challenges. tors in reducing economic impacts at the sectoral level Since before Independence, Dominica's govern- include action to build disaster mitigation features into ment has emphasized economic diversification away facilities and to spread risk through insurance. The pro- from banana production within the agricultural sector tection of key infrastructure is vital. and into nonagricultural sectors, with the aim of cre- ating a more resilient economic structure. There has Agriculture, Livestock, and Fisheries also been explicit recognition, notably following Hur- ricane David, of the vulnerability of the one-crop econ- Around half the population of Dominica is still directly omy to natural hazards. In practice, however, there is dependent on agriculture. Damage to agricultural pro- little evidence that this vulnerability has been addressed. duction and markets therefore affects the welfare of the Hurricane David, in fact, furthered the shift into banana majority of the population immediately and deeply. The production: the government did not actively promote dominant crop, bananas, is especially sensitive to damage diversification through such measures as incentives, and from winds of 40 miles per hour or more, and even the bananas offered a quick, low-cost way to restore agri- fringe impacts of less severe storms can cause serious cultural production and export income. Banana pro- damage. Nevertheless, a number of factors have com- duction continued to be profitable through the 1980s, bined to reinforce the concentration on banana pro- and during this period the government did not explore duction, to the exclusion of other crops that could potential opportunities for developing the services sector significantly reduce the overall sensitivity of Dominica's and diversifying within the manufacturing sector. agricultural sector to natural hazards. Banana produc- The declining profitability of bananas in the 1990s tion levels can be restored quickly, with replanting, forced a reexamination of the composition of the agri- within 6 to 9 months following a complete loss of the cultural sector. Yet there is still no clear-cut growth strat- crop; a compulsory insurance scheme for banana pro- egy, and the risks from natural hazards are still not being ducers has returned farmers only about 20 percent of effectively taken into account in the formulation of the value of lost production; and there has been an economic strategies and policies. The same is true for assured export market. The livestock sector is small and the Caribbean region in general. This shortcoming needs reasonably resilient, although meat imports increase to be addressed; the evidence from Dominica indicates temporarily following storm damage. The fishing sector, real scope for reducing the structural vulnerability of by contrast, is highly vulnerable and is slow to recover. the economy provided that the opportunities are grasped. Although fishing is declining, it is important in a diverse The reduced importance of the agricultural sector in the economy and provides the main livelihood of many economy has helped lessen the impact of recent storms, poor families. and the burgeoning international financial services sector could play a significant role in reducing vulnerability Manufacturing, Tourism, Construction, to hazards. More information and analysis on the eco- and International Financial Services nomic and financial impacts of natural hazards would be helpful in integrating risk reduction concerns into Although the manufacturing sector grew by an average medium- and long-term economic and financial plan- 7.1 percent a year between 1977 and 1999, in 2000 ning and would thereby contribute to sustainable growth. the government still regarded it as being in an embry- onic state, heavily concentrated as it was on coconut- Sectoral Impacts based soap and detergent production. (Soap and dental products overtook bananas in 1996 as the largest The macroeconomic analysis shows that not all sectors merchandise export in value terms.) Natural disasters and subsectors of the Dominica economy have been are not seen as a major constraint on growth in the sector, equally vulnerable to natural hazards. The study explored although some individual products may be sensitive to in more detail the effects on the principal productive weather conditions. There is some evidence of effective 72 Appendix A risk management measures (e.g., higher construction development. Given the severe financial constraints, standards for the Dominica Coconut Products Ltd. jetty, construction costs were kept as low as possible. The to withstand storms). A major failure in risk manage- investment took place following more than 20 years ment is inadequate business insurance coverage. without any major hurricane impacts. Under these cir- Expansion of tourism has been a central thrust of cumstances, adequate hazard-proofing was not pro- Dominica's development strategy for the past 20 years. vided, and when Hurricane David struck, the Assessing natural hazard impacts and drawing lessons consequences were devastating. for the future for this sector are therefore particularly Following Hurricane David, there was wider inter- important. Hurricane David had a devastating impact est in reducing vulnerability by incorporating more effec- on tourism, halting its growth for five to six years. Infra- tive mitigation measures into design and construction. structure investment following the hurricane, includ- Since then, substantial, but uneven, progress has been ing measures to protect cruise ship facilities, was helpful made in reducing vulnerability in all areas of infra- in supporting rapid growth from the late 1980s on and structure and building. Although some investments have in reducing vulnerability. But underinsurance is still a been exemplary, financial constraints and an emphasis problem, especially with the occurrence of an extreme on restoring facilities to normal use as quickly as pos- event such as Hurricane David, and it is of particular sible after a disaster have meant that initial construction, concern in a sector composed entirely of locally owned repair, and restoration work has often not incorpo- sole proprietorships and partnerships. Uncertainty about rated fully effective hazard-proofing features. The hazard risks--for example, of a volcanic eruption or long-term effect has been higher overall costs and an earthquake affecting the strongly tourism-focused increased pressure on limited resources. The rehabili- businesses in the south--may make for difficulties in tation cost of major storms since 1979 is estimated at securing insurance coverage and investment funding. EC$380 million (US$140 million), at 1999 prices, equiv- The massive increase in the growth of the cruise ship alent to EC$18 million per year, with around EC$10 business suggests that tourism may now be more sen- million per year for key infrastructure. sitive to the impact of natural hazards on the wider The history of the deepwater port at Woodbridge Bay, Caribbean tourism market. which was built between 1974 and 1978, highlights the Constructionistheoneindustryinwhichdisasterbrings value-for-money case for designing new infrastructure increased activity, but there is no evidence of a general and buildings to withstand hurricane damage. An inter- postdisaster construction-led boom. This may reflect nal rate of return of 13 percent was achieved by sub- Dominica's reliance on imported building materials. stantially scaling down the original design, even though Since the mid-1990s, the government has sought to there was evidence of a high risk of storm conditions make Dominica an offshore financial center, and the potentially far more damaging than were provided for financial services sector has grown rapidly. Although in the sea defenses incorporated within the port reduction of vulnerability was not a factor in the gov- design. Rehabilitation costs following the damage caused ernment's decision to promote this sector, financial serv- by Hurricane David were equivalent to 41 percent of ices are likely to be largely unaffected even by major the original costs. Had the original structure been built disaster events. Continued expansion of the sector should to withstand a Category 4 hurricane, the initial invest- therefore bring about a further reduction in broad eco- ment costs would have been only 11 percent higher. nomic vulnerability. Because of continued financial pressure, hazard risk again seems to have been underestimated in the Infrastructure and Buildings expansion of the port. The case raises awkward ques- tions in relation to economic analysis. Between 1950 and 1978, Dominica was transformed Most of Dominica's road system runs along the narrow from an underdeveloped plantation-cum-subsistence coastal strip, very near the shore, rendering it poten- colony into an independent, middle-income economy. tially highly vulnerable to storm damage. Other key Key to this achievement was rapid infrastructure infrastructure networks--telecommunications, electric Dominica: Natural Disasters and Economic Development in a Small Island State 73 power, and water distribution--run alongside the accessible mechanisms for countering shocks in other road and are similarly vulnerable. The record of invest- sectors may now be equally important. ment in constructing sea defenses to protect this infra- Disaster-related increases in capital inflows follow- structure and in building roads to more robust standards ing hurricane years have typically overcompensated is patchy. In places, there have been some exemplary for downward pressures on the trade account and have investments that incorporate higher storm-resistance contributed to reconstruction investment. specifications, but elsewhere the road system remains highly vulnerable. The apparent slow progress in pro- viding sea defenses partly reflects the scale of invest- Investment and Domestic Consumption ment financing required and partly a lack of donor coherence. Piecemeal projects focus on particular coastal It is difficult to discern evidence of the effect of natu- sections--perhaps reflecting different donor priorities ral disasters on total investment or consumption in such as overall economic development or targeting of Dominica except in the aftermath of Hurricane David poorer geographic areas--and involve different design in 1979. Then, the scale of both losses and reconstruction and construction processes. All three key public utility funds created an opportunity to replace and update systems were devastated by Hurricane David and were much of the island's infrastructure and commercial partially disrupted by coastal damage from Hurricane and productive capital, following years of inadequate Lenny in 1999. The vulnerability of this infrastructure maintenance and limited investment. Gross domestic to natural hazards now varies, reflecting the degree of investment increased by almost 25 percent year-on-year mitigation investment that has taken place and the asso- in 1979 and by over 65 percent in 1980, but it fell by ciated practical and funding issues. over 25 percent the following year. The falloff in pri- vate investment from 1981 on was marked, suggesting that once repairs had been completed, the hurricane External Accounts may have deterred new investment. Subsequent hurri- canes have not led to any comparable infusion of recon- The links between disaster shocks and export earnings struction capital. are clear and direct; the links with imports are more Nevertheless, the World Bank attributes the relatively inferential. Dominica has typically had a real trade deficit high level of consumption volatility in the Caribbean equivalent to 12­13 percent of GDP. In some years, the region as a whole to the fact that, in the face of high deficit has been much higher, reflecting variability in vulnerability to a range of external shocks, countries export levels and postdisaster surges in imports--notably, are not spreading their risks optimally, despite having of building materials, equipment, and food. The declin- relatively well-developed financial systems. The Bank ing trend in banana export earnings since about 1989 concludes that financial and insurance markets need has reduced the sensitivity of the trade account to dis- to be developed further, through, among other means, aster shocks. Despite the loss of banana earnings in closer harmonization within the region's financial and 1995­96, total export earnings increased because of the insurance markets, strengthening of securities markets, growth in earnings from soap and detergent products pension reform, and the more efficient transfer of risks and from relatively resilient nonfactor services such as to the international market. offshore banking and tourism. The shift shows how the nature of an economy's sensitivity to natural disas- ter shocks can change rapidly. Governments and inter- national organizations need to take account of this and Financial Aspects regularly reappraise their disaster response policies. For example, arrangements for buffering the effects of nat- The study examined how natural disasters affect finan- ural disaster shocks have focused on compensating cial systems and how private and public financial insti- for loss of primary commodity export earnings. Easily tutions cope with the associated pressures. 74 Appendix A Banking and Credit Institutions for some essential products, but it is not clear whether this law has been applied. Banking and credit systems can help spread risk in disaster-prone countries. Natural disasters place pres- sure on banking and credit institutions as money is with- Insurance and Other Risk Transfer Mechanisms drawn, loan repayments are deferred or defaulted on, Dominica's insurance industry is relatively well and increased credit is sought to finance uninsured reha- developed in comparison with that in many other bilitation costs and disruptions in income flows. In developing countries. There is, however, a major prob- extreme cases, the outcome may be the collapse of part lem of underinsurance, partly reflecting the high and of the banking sector. Dominica has no national cen- volatile cost of insurance in the Caribbean region. tral bank; rather, the Eastern Caribbean Central Bank About 80 to 85 percent of gross property insurance (ECCB) conducts monetary policy for the Organisa- premiums in the region are transferred to reinsurers; tion of Eastern Caribbean States (OECS), supervises and any fluctuations in reinsurance costs, whether local, regulates commercial banks in the member states, and regional, or global in origin, are passed on directly to is lender of last resort (although it has never been those paying the premiums. Premium rates typically called on in this capacity). Dominica itself has five com- increased three- to fourfold between the 1970s and mercial banks, one national bank, a rapidly growing 2000, although they fluctuated considerably year by network of credit unions, and several nonprofit bodies year. Data on underinsurance specific to Dominica are involved in loans and credit. not available, but it is estimated that 25 to 40 percent The available data suggest that natural disasters of dwelling stock in the Caribbean region is uninsured, have had relatively little overall effect on Dominica's and much of the insured dwelling stock may be under- banking and credit sector. Nevertheless, the sector's lim- insured. Use of business interruption policies is also ited and fragmented ability to spread and transfer risk, very low, perhaps as little as 5 percent. Insurance of not only in Dominica but also in the Caribbean region public property is almost certainly limited by bud- in general, is a serious concern, and measures are getary constraints. being explored to address the problem. Within the The proliferation of insurance players in the Caribbean, OECS, some services are being initiated that will facil- some of them very small, has led to concerns about the itate risk-sharing as banks in the regions shift increas- efficiency and, more important, the safety of the insur- ingly into syndicate lending. There has also been some ance industry. New insurance legislation to be enacted discussion about the establishment of a jointly owned by OECS members aims to address some of the main lending subsidiary that could diversify across territo- concerns, through disincentives to small insurance play- ries and fund loans that are too large for individual ers, incentives to encourage amalgamation of insurance banks. In Dominica banks and credit institutions are operators, and regulations on minimum levels of share seeking to reduce risk exposure by limiting agricul- capital required for registration. The World Bank has tural lending, but this may make postdisaster recovery also recommended that companies and households in more difficult. the region establish reserve funds to finance uninsured disaster losses. In addition, for a number of years now, Inflation regional and international organizations have been The rate of inflation has remained low over the past 20 discussing the creation of some form of regional risk years and, with one exception, natural disasters have had management tool, such as an intercountry insurance- only a limited effect on it. The exception was Hurricane pooling arrangement. David, following which the food index alone was reported To date, there has been limited use of insurance as a to have increased by over 45 percent between 1978 and mechanism for promoting mitigation through differ- 1979. Dominica's reliance on imports for most con- entiated premiums to reflect lower risks on strength- struction materials may have helped constrain post- ened land and buildings. More widespread discriminatory disaster price hikes. Legislation allows prices to be fixed premium pricing would be valuable, but it would require Dominica: Natural Disasters and Economic Development in a Small Island State 75 comprehensive hazard mapping and inspection of indi- External Assistance vidual properties. External assistance accounts for a large proportion of the Dominican government's public capital spending Public Finance and for some current expenditure. Although there has been considerable short-term variability in total aid Since Independence in 1978, one of the Dominican gov- flows, no discernible trends are visible for the period ernment's objectives has been to strengthen and stabi- since Independence. Disaster-related aid is often received lize public finances by exercising tighter control on some time after the event, and to some extent, aid pro- spending and by increasing revenue. Balancing the budget vided in the aftermath of a disaster is diverted from non- has nevertheless been difficult; the high costs of pro- disaster-related projects. Apart from Hurricane David, viding an infrastructure and public services network to which was severe enough to attract widespread inter- a small and dispersed population, together with the costs national publicity, aid levels have not altered signifi- implied by Dominica's extreme vulnerability to natural cantly in response to public expenditure pressures hazards, place huge burdens on the country's limited following natural disasters. resources, and it has been difficult to balance the budget. The government's ability to increase the money supply to reduce public debts is strictly limited by the ECCB, Social Issues and Poverty which is responsible for monetary policy across the OECS. Aid and other foreign funding have contributed signif- At the household level, poverty is the single most impor- icantly to both capital and recurrent expenditure; taxes, tant factor determining vulnerability to natural hazards. for example, accounted for only around 60 percent of This vulnerability reflects housing quality (location, total revenue between 1978/79 and 1997/98. building materials used, and failure to provide adequate Aggregate budgetary figures suggest that disasters support for low-cost housing in the aftermath of disas- have had little impact on public finances except fol- ters); the primary sources of household income for poor lowing the most extreme events, such as Hurricane people; limited financial resources; and lack of access David. Until the 1990s, expenditure on mitigation and to risk-spreading mechanisms. Despite the importance preparedness also appears to have been low. In reality of issues of social vulnerability and the impacts of nat- (although this is largely unrecorded), the costs of ural hazards at the household and community levels, meeting more urgent needs in the aftermath of disas- there has been little research to date on this subject in ters have tended to displace planned investment, and Dominica. These issues need to be better understood the extent of the necessary reallocations has been exac- so that a risk management strategy can be developed erbated by the absence of any calamity reserve fund. that addresses the special vulnerability of poor house- Over time, cumulative reallocations and disaster-related holds and recognizes that poverty can be caused or exac- damage to and destruction of public infrastructure and erbated by natural disasters. other facilities have had an adverse effect on the pace Several highly vulnerable economic sectors are par- and nature of physical and economic development, ticularly important to the livelihoods and incomes of making it more difficult to attract and sustain new pro- poor households. A large number of extremely small, ductive investment. Against this background, it is essen- marginal producers are involved in the banana indus- tial that the budgetary impacts of disasters be measured try, which has given many poor households a means of more explicitly and that risk reduction concerns be built reestablishing their incomes following disasters. The into medium- and long-term economic and financial compulsory insurance scheme, however, dispropor- planning. In this way, a more rational response to risk tionately benefits the larger producers. Huckstering-- and disasters will be possible, including more soundly trade between Dominica and nearby islands--is an based reallocation decisions and better targeting of exter- important part of the informal labor market and is typ- nal assistance. ically undertaken by women, often from poor rural 76 Appendix A households. It is highly vulnerable to natural hazards, disaster management, would be more effective both in but no insurance coverage is available for huckster con- a crisis and in planning for future disasters than the signments. Fishing, which is pursued by some of the existing department-based arrangements. country's poorest households, is another extremely vul- Environmental assessment of hazards poses partic- nerable sector, but the evidence shows that little has ular problems for small developing countries. Concerns been done to recompense or rehabilitate households about the growing exposure to risks highlight the need affected by disasters. On the positive side, there is increas- for increased assessment and monitoring and for pro- ing recognition of the importance of tackling vulnera- vision of clear and accessible information about hazard bility to natural hazards as part of a poverty reduction risk. Dominica has no capacity to undertake this effort agenda for the region, and the government has deter- independently. The necessary regional approach is mined that further investigation into the nature of poverty now emerging in work being done under the auspices in Dominica is needed. of the OAS, with international funding and expertise. A particular cause for concern in Dominica's disas- ter mitigation and management arrangements is that Disaster Management the opportunities for promoting highly cost-effective hazard-proofing by setting appropriate design and con- Considerable progress has been made over the past 20 struction standards have not been grasped. Regional years in Dominica's state of preparedness for natural dis- civil engineering experts have estimated that spending asters. The 1996 National Disaster Plan is a substantial 1 percent of a structure's value on hazard-proofing meas- and detailed document that focuses on disaster pre- ures can reduce the probable maximum loss from hur- paredness and immediate postdisaster responses. Lead ricanes by, on average, one-third. Currently, however, responsibility for the plan and for disaster preparedness Dominica has no formal building code. The island's lies with the Ministry of Communications, Works, and infrastructure remains vulnerable to tropical storms; Housing. Individual government departments are respon- land use planning is weak; and there is evidence that sible for undertaking postdisaster impact assessment, as recently as 1996 new housing was built without and donors interested in supporting particular aspects adequate anchoring of roofs to walls. of relief approach the relevant ministry. The National Disaster Plan, however, largely over- looks long-term mitigation and prevention issues and Conclusions and Policy Implications measures for addressing economic impacts and pro- moting economic recovery. Overall, there does seem to A number of key conclusions emerged from the coun- be a good awareness of disaster issues, but the gener- try study: alized concern and commitment have not been trans- · Considerable uncertainty and unpredictability are lated into a coherent overall strategy for disaster reduction. attached to natural hazards. Hurricanes may not Such a strategy would need to include or use informa- follow established patterns, causing unforeseen tion that has not been available so far: comprehensive damage, and a series of disasters may strike after a and robust assessments of potential physical damage and long period of calm. Regular reappraisal of risks is of macroeconomic and social impacts; mitigation and needed. preparedness plans; implementation plans to support · The economy's vulnerability is constantly changing, affected groups; and a comprehensive rehabilitation reflecting changes both in the levels of development plan. It would require consultation with local stake- and capital investment in the island and in the holders and NGOs and cooperation with regional bodies structure and composition of economic activity. Mea- and funding agencies. sures can be taken to reduce the structural vulnera- Experience in Dominica and in the wider region sug- bility of the economy. Greater integration of hazard gests that a high-level interdepartmental task force, well risks into medium- and long-term economic and supplied with resources and given the responsibility for financial analysis and planning could substantially Dominica: Natural Disasters and Economic Development in a Small Island State 77 reduce the economy's vulnerability to natural haz- is also a need for clear and accessible public infor- ards and contribute to sustainable growth. mation. Achievement of these information objectives · In the immediate aftermath of a disaster, the public for a small island economy like Dominica will require and private sectors face choices between pursuing regional and international cooperation. rapid recovery and seeking reductions in longer-term · There are wider implications for other small island vulnerability. Similar choices arise in determining the states. The vulnerability of a small, open economy quality and standard of development investments. can alter rapidly, reflecting the significance of exter- In Dominica, with its limited public and private nal factors. In considering appropriate disaster mit- resources, and in the absence of the necessary polit- igation measures, it is important to recognize the ical impetus or financial incentives for investment physical characteristics underlying the island's in mitigation and changes in land use, the emphasis economy and society. Fostering less vulnerable has been on quick recovery and least-cost invest- areas of economic activity and strengthening key ment. But the opportunities for reducing vulnera- infrastructure will help ensure long-term sustainable bility must be seized. Improved information on the growth and minimize disaster impacts. Much more budgetary impact of disasters is needed to facilitate needs to be done in the Caribbean region as a whole the cost-effective allocation of resources and to empha- to spread and reduce hazard risks through insur- size the importance of integrating hazard concerns ance and other financial risk transfer mechanisms. into medium- and long-term planning. Donor agencies can also secure better returns to invest- · The unsatisfactory levels and forms of hazard risk ments in disaster mitigation by undertaking more information available in Dominica hinder effective joint programs and having certain agencies take the risk-averting decisionmaking. Urgent action is needed lead on particular projects, rather than working in to ensure sufficient investment in monitoring, assess- parallel, and by supporting regional solutions and ment, mapping, and dissemination activities. There reducing micromanagement of some projects. Appendix B Bangladesh: Disasters and Public Finance This appendix, which summarizes a study on natural percent of Bangladesh's population is classified as poor disasters and public finance in Bangladesh, focuses and about 34 percent as living in extreme poverty. These mainly on the period since 1980. During that time, people typically live and work in the areas most at risk five major disaster events triggered national and inter- from natural disasters. At the household level, poverty national emergency responses. The study uses available is still the single most important factor determining vul- data to examine movements around trends and to nerability to hazards. compare economic performance (forecast and actual) It is difficult to make robust, quantitative assessments before and after disasters, and it places these findings of the probable levels of risk associated with the main in the economic and social policy context of Bangladesh. natural hazards in Bangladesh. Some reasons are: It breaks new ground in that, although the immediate · The exceptional complexity of the natural environ- emergency and short-term budgetary implications of ment in the country individual natural disasters have been much studied, · The obvious limitations of historical data this is the first investigation of the recurrent longer-term · Environmental changes, some recorded and others aspects of disaster risk, which are particularly relevant so far only identified as possible consequences of for disaster-prone countries. global climatic change · Human activity in Bangladesh and the immediate region, which is possibly altering the likelihood of Natural Disasters and Hazard Risk in Bangladesh specific events as well as, more obviously, the asso- ciated effects. Bangladesh is one of the most disaster-prone countries Current understanding of the risks of extreme, cat- in the world. Most of its large, densely settled popula- astrophic events, and the problems of assessment, varies tion of 130 million people is at significant risk from widely according to the particular hazard. more than one form of natural hazard. In terms of area and number of people directly affected, impact on eco- River Flooding nomic activity, and damage or destruction of assets, the types of disaster that have been most important Four extreme flooding events have occurred over the since Independence in 1971 are exceptionally wide- past 30 years, in 1974, 1987, 1988, and 1998. Very spread riverine flooding, severe tropical cyclones and high floods in 1976 and 1984, although less severe than associated coastal storm surges, river bank erosion, the extreme events in height, maximum flow, and pro- and drought. Official estimates are that 139,000 portion of the country's area inundated, caused wide- people were killed during the 1991 cyclone and that spread suffering and loss and elicited an international 31 million were directly affected by the 1998 floods. emergency response. The implied annual risk of an Rapid-onset flash flooding, tornadoes, and landslides extreme flood is a high 10 to 20 percent. Historical are frequent causes of more localized, but intense, human data on heights and flows for the major rivers are of suffering and loss. Severe earthquakes have been rare doubtful quality and are available only for one station but are a potentially catastrophic hazard. Around 50 on the Ganges (in operation since 1934) and for one 79 80 Appendix B station on the more unstable and recently more threat- rainfall, causing moisture stress and limiting crop growth. ening Brahmaputra River (since 1956). The 1950 Assam Drought as a disaster category, however, should be earthquake upset the flow of the Brahmaputra, and the restricted to extreme, uncommon events. Furthermore, course of the Jamuna section of the river continues to agriculture may now be less sensitive to rainfall vari- move westward. Massive upper-riparian interventions ability within a wider range than previously because of have been made in neighboring countries since 1950-- the rapid spread of irrigation and the emergence in the so far, mostly for the Gangetic system--and the poten- late 1990s of irrigated boro (dry-season) rice as the main tial impacts were highlighted by the floods of 2000, crop. However, the consequences of draw-down of the the worst in 80 years in southwestern Bangladesh. water table in an exceptionally dry year might imply both a different seasonal pattern and overall intensity of impacts on agriculture, other water-intensive activ- Cyclones and Storm Surges ities, and human water supply. Finally, there are the A record of more than 100 years exists for storm tracks potential impacts of climatic change, which could lower and approximate intensities of storms in the Bay of the risk of extremely low annual rainfall patterns that Bengal. At least 14 very severe storms affected Bangladesh have historically caused agricultural drought. during that time, implying an annual risk of over 10 percent. Clusters of storms, such as the six major storms that affected Bangladesh between 1960 and 1970, can River Bank Erosion occur within the otherwise random series. So far, there Erosion of river banks is a continuous, rather than inter- is no evidence of changes in the frequency or intensity mittent, problem that can be predicted with some accu- of storms in the Bay of Bengal. If, however, mean sea racy. The high level of risk makes insurance of assets or level were to rise, that would increase the expected sever- production impracticable. The most effective mitiga- ity of impact of a cyclone and the associated storm surge. tion measure is strict land use zoning, but this is cur- Human intervention, including deforestation in the Sun- rently unenforceable. Consequently, both NGOs and, darbans and the network of often poorly maintained more recently, the government, through the National embankments, may alter the incidence of impacts. Water Management Plan of 2001, have moved toward supporting microscale protection and nonstructural Earthquakes measures. Bangladesh lies within a high-risk region. Minor tremors are common, and one of the most extreme events, an Flash Flooding 1897 earthquake that measured 8.8 on the Richter scale, had its epicenter in the nearby Shillong Plateau of the Flash flooding is an annual phenomenon, and there is Indian state of Meghalaya. Because of the inadequacies considerable scope for predicting the areas of likely inci- of the available data, local assessments provide only dence and the timing of floods. This would involve com- highly tentative risk-zoning within the country, and only bining geographic information system (GIS) data, local in map form. ground confirmation, and real-time information on rain- fall and river levels and would require international cooperation to communicate and assess information at Drought least daily. The estimation of risks and their likely impacts on the basis of historical data is likely to yield seriously inac- curate results. There have been seven reported droughts Tornadoes and Line Squalls in Bangladesh since 1971, an implied frequency of one Tornadoes and line squalls are a common, localized phe- almost every fourth year. Such repeated reports of drought nomenon, traditionally associated with the Bengali New suggest widespread problems of lower-than-average Year in April. The risks are potentially computable in Bangladesh: Disasters and Public Finance 81 terms of risk-zoning and expected frequency, making may not have reduced the vulnerability of export insurance of assets a possibility. earnings to natural disasters. For example, as the world's primary jute producer, Bangladesh was a price-setter, even in times of disaster. By contrast, as Impact of Major Disaster Shocks on Bangladesh's a ready-made garment manufacturer, it faces severe Economic Development competition. Disruption of production results in loss of export revenue and overseas markets. The level Since Independence, Bangladesh's economy has achieved of foreign exchange reserves has periodically emerged impressive rates of growth. In the late 1970s, it recov- as a critical issue, in part because of its vulnerability ered rapidly from the devastating effects of natural to natural hazards and other shocks. disasters, war, and famine that marked the years · External assistance has helped fund a major share of 1970­75. The average real annual growth rate of GDP public investment (and, by implication, of total invest- was 4.2 percent in the decade 1980­90 and increased ment) and meet the country's large trade gap, and it to 5 percent in the decade 1990­2000. Average has contributed significantly to postdisaster relief and annual growth of per capita GDP rose from an average rehabilitation. of 1.7 percent in the 1980s to 3.3 percent in the · The ratio of debt to GDP has been growing (from 1990s, reflecting both higher GDP growth and declin- 6 percent in 1973 to 47 percent in 1998) because ing population growth. At the same time, there has fiscal debts have been financed through borrow- been a change in the structural composition of the econ- ing. Any further growth in domestic borrowing-- omy: agriculture's share of GDP has declined, while the for example, as a result of disaster-related budgetary industrial and service sectors have expanded, result- difficulties--would lead to pressure for cuts in dis- ing in a sharp shift in the composition of the country's cretionary areas of recurrent spending and then, exports. A gradual process of structural adjustment and potentially, to cuts in programs for postdisaster trade liberalization, along with more disciplined rehabilitation. monetary management, resulted in maintenance of a single-digit inflation rate in the 1990s and an annual Natural Disasters and Macroeconomic Performance current account deficit of less than 2.5 percent of GDP. The reforms have also helped increase private Some highlights emerge from a simple assessment of sector development and foreign direct investment. Fiscal the sensitivity of Bangladesh's economic performance policy has not been so successful: a World Bank to major disaster, as measured in terms of fluctuations report points to large fiscal deficits, a low tax-to-GDP in agricultural, nonagricultural, and total GDP: ratio, and relatively poor-quality spending. · In the period 1965­75, there was extreme volatility Several features of Bangladesh's economy are particu- in the economy, clearly linked to catastrophic natu- larly relevant for the public finance impacts of disasters: ral disasters as one destabilizing influence. · Because of the high level of public sector involve- · Major disasters, with the notable exception of the ment in the economy, the government ultimately 1998 floods, have led to a downturn in the agricul- bears a substantial share of any disaster-related losses. tural sector's annual growth rate. · High reliance on imports for industrial develop- · The impact of disasters on the nonagricultural sector ment and production and the large proportion of appears much less significant, although the longer- government revenue generated from import earnings term impacts are not reflected in interannual fluctu- mean that the performance of imports and exports ations. If resources are diverted from productive and the availability of foreign exchange are impor- investment to disaster management, the pace and tant determinants of the amount of resources avail- nature of development will be adversely affected. able to the government. The export base remains · The sensitivity to disasters of both the agricultural narrow and therefore vulnerable to external shocks. and the nonagricultural components of GDP appears The shift from agricultural to manufacturing exports to be declining over time, suggesting greater resilience. 82 Appendix B Of course, the economic impacts do not reflect the allocations against specific central government projects, severity of disasters in terms of loss of life and human as well as the investment projects of public enterprises tragedy. funded with government and foreign aid contributions. ADP spending has increased over time, and foreign aid accounts for a declining but significant share of its expen- Impacts of Disasters on Public Finance: Budgetary diture (on average, 45 percent between 1980/81 and Process and Performance 1998/99). Over the same period, actual expenditure Public finances are highly centralized in Bangladesh; averaged 89 percent of the ADP budget, with little appar- in the mid-1990s, the central government accounted ent linkage between the incidence of disasters and the for about 97 percent of total revenues and 93 percent ratio between actual and budgeted spending. The ten- of public spending. The following analysis is restricted dency to underspend reflects factors such as project to central government operations. delay (due in turn to counterpart funding constraints), The movements of broad budgetary aggregates-- staff shortages, complex bureaucratic and procure- revenue expenditure, development spending, central ment procedures, delays in land acquisition, conflict- government revenue, and the budget deficit--in the ing conditionalities set by donors, and poor monitoring. period since 1981 suggest that disasters have had little Delays lead to cost increases and greater exposure to impact on central government finances. Revenue expen- natural hazards, since unfinished projects are more vul- diture, development spending, and central government nerable to damage. revenue have risen gradually over the period. The Over time, the composition of the ADP has changed. overall budget deficit remained fairly stable until the Investment in roads, water, and sanitation has assumed 1998 flood, when it rose markedly. Closer analysis of increased importance. When cutbacks have been nec- expenditure and sources of finance, however, gives a essary, agriculture, rural development, water resources, rather different picture. and infrastructure have generally been spared as far as possible, with the social and industrial sectors taking The budgetary process. Responsibility within the cen- the largest cuts. The fact that disasters have not forced tral government for public finance planning, alloca- a significant increase in spending on the ADP suggests tion, and monitoring is shared by a number of that they have instead led to reallocation of resources government departments. Procedures in place for track- or to delays in ongoing projects. If that is happening, ing and controlling total annual public expenditure it is important that reallocations be made in a way that (equivalent to US$3 billion) have not provided the clear minimizes their long-term developmental impact. lines of responsibility and delegation necessary to This will require: respond effectively to disaster shocks. For example, the · A clearly stated and applied policy framework linked distinction between revenue expenditure and capital to defined and achievable objectives spending is somewhat artificial; revenue projections · A sound method for prioritizing projects and expenditure forecasts have typically been overop- · Reliable information on resource availability, and full timistic; and under the existing financial reporting sys- and accurate damage assessments tems, public accounts tend to underestimate the full · Proper evaluation of postdisaster rehabilitation and effect of disasters on public finance. Changes now under reconstruction projects. way to improve monitoring of spending should facil- The basic medium-term strategic planning tool for itate more cost-effective reallocation of resources, pro- capital spending in Bangladesh has been the five-year vided that there is also a clear system for prioritizing plan. A three-year rolling plan was introduced in 1991/92. projects. ADP projects are examined for their compatibility with the strategic plans and for their cost, technical quality, Development spending. The Annual Development Plan and economic viability. But the composition of the (ADP)isthebasicinstrumentforpubliccapitalinvestment ADP may also reflect political pressures, the availabil- planning and implementation, covering expenditure ity of external assistance, and donor priorities. Heavy Bangladesh: Disasters and Public Finance 83 dependence on aid has, not surprisingly, led to less gov- government spending. Gradual budget increases have ernment control of a development vision. reflected rising nondiscretionary expenditure on pay Effective reallocation of resources depends on a clear and debt servicing, which together make up almost half ranking of projects according to level of priority. In theory, of total revenue spending. Despite budget growth, some "core" projects have been given priority under the ADP. nondiscretionary spending areas covered have been con- A formal core program of projects was introduced in sistently underfunded. Commentators have highlighted, 1983, but the practice was already informally in place in particular, low levels of expenditure on operations in the mid-1970s. In that early period, priority was and maintenance (O&M). Discretionary expenditure accorded to quick-yielding projects and to those in is by definition more vulnerable to cutbacks, including advanced stages of completion. Among sectors, agri- cuts in the wake of natural disasters. culture (in particular, labor-intensive, small-scale drainage Surprisingly, growth in the revenue budget has not schemes) and resource exploration projects were favored. been substantially higher in disaster years. Disaster- Draining and dredging of small rivers and canals took related expenditure seems to have been met within the precedence over work to control major rivers, and small- existing budget envelope by drawing on unallocated scale irrigation projects were preferred over large-scale resources (e.g., to meet needs for agricultural subsidies) ones. In reality, rather than lower-priority projects and through diversion from other uses--for example, being targeted for spending reductions, across-the-board through redeployment of government staff. The real and cuts were generally used to restrain expenditure. underlying implications of disasters for revenue No specific guidelines for prioritization were pro- spending can therefore be understood only with the vided in the formal core projects system introduced in help of more detailed knowledge of decisions taken at 1983. The criteria for selection included stage of com- the time of a disaster. pletion, production and employment potential, level of donor support, and links with other projects. In the Government revenue. Bangladesh's revenue structure is late 1980s, the World Bank found that these priorities very weak. Total revenue averaged only 6.9 percent of were being reflected in the composition of the ADP but GDP in the 1980s and 9.3 percent in the 1990s. Exter- were not being applied effectively in determining pref- nal grants and loans have supplemented revenue, increas- erential access to funds. Following the 1987 and 1988 ing the total available by an average 40 percent in the floods, available funds were again spread more thinly 1990s. Tax revenue has accounted for the largest part across existing projects rather than being allocated to of total revenue, and there is an overwhelming depen- priority projects. With the introduction of three-year dence on indirect taxes, particularly import duties. rolling plans, core projects are now designated for pri- Between 1992/93 and 1997/98, income tax accounted ority funding, and other projects are assigned rank- for only 15 percent of tax revenue. ings. Despite this, the World Bank's assessment is that The available evidence suggests that natural disas- more consistent use of objective criteria based on an ters have had little impact on aggregate tax performance. economic rationale is needed to set expenditure prior- Examination at a finer level of detail shows, however, ities between and within sectors. Only by strengthen- that adjustments have been made. In response to the ing this area of public and financial planning can the 1987 and 1988 flooding, the government increased government be confident that the development proj- some taxes to meet the additional costs of disaster aid ects actually undertaken represent the best value for the while at the same time reducing the tax burden on other public good and for longer-term growth and income. parts of the economy to stimulate growth and aid recov- Donors who play a large part in project design have a ery. Since the late 1980s, various measures, such as the responsibility for helping to secure improvements in introduction and subsequent extension of a value-added this area. tax, have been taken to enlarge the tax base and reduce exemptions. It is widely agreed that the government still Revenue budget. Since 1987/88, the revenue budget has needs to increase its tax revenue, perhaps to 13­14 per- typically accounted for more than half of total central cent of GDP. But higher reliance on taxes to fund public 84 Appendix B expenditure implies greater tax losses as a result of dis- additional pressure on the resources available for planned asters if there is no change in tax composition. This projects. The substantial pledges of foreign aid were would have consequences for alternative financing critical to the recovery programs, but notwithstanding requirements and place further pressures on expendi- the generosity of donors, there was a lag between the ture. Longer-term tax reforms in Bangladesh, therefore, commitments and the availability of funds in the ADP need to aim at reducing the overall vulnerability of tax budget. The system for reallocating resources from lower- revenue to disasters, as well as increasing revenue. to higher-priority projects was applied more effectively following the 1998 floods than in the aftermath of the Funding the government deficit. Since Independence, 1987 and 1988 floods. Expenditure on rural develop- Bangladesh has run large, but gradually declining, public ment and institutions and on transport increased as a expenditure deficits. In recent years, the costs of fund- result of the reallocations, reflecting efforts to repair and ing the deficit have been rising, with increasing rehabilitate damaged infrastructure. Reductions were reliance on higher-cost domestic debt instruments rather made in spending on health, population, and welfare, than foreign assistance. Time-series statistics show no in line with the tendency to reduce spending on social discernible relationship between disaster shocks, on the sectors in the aftermath of disasters. Given the priority one hand, and patterns of expenditure and funding accorded to social sectors in the poverty reduction strat- sources, on the other, although there is some evidence egy, such reallocations need to be carefully considered of lags in the impact of shocks on public finance. because they may be inappropriate. Two important consequences of the 1987 and 1988 floods were (a) a comprehensive review of flood policy Major Flood Disasters and Public Finance by the government of Bangladesh and the international Performance: The Floods of 1987, 1988, and 1998 community, and (b) deregulation of private agricultural investment, which quickly led to more investment in It is worth highlighting some of the public finance impacts lift irrigation and rapid growth in rice production. of the most extreme recent floods, in 1987, 1988, and A key difference between the earlier major floods and 1998, and some aspects of the government's response. the one in 1998 was that in the more recent crisis, much A comparison between forecast and actual annual more detailed information was available to assist plan- revenue and expenditure suggests that the flood crises ning, and management of the crisis was more trans- had limited effects on public finances. The govern- parent. The consequences of the 1998 floods were more ment sought to limit the budgetary impact of the floods. closely examined within Bangladesh than was the case Domestic revenue was augmented by the introduction with the floods a decade earlier, and this greater of specific measures to increase tax revenue (although understanding offered an opportunity to mitigate and there were also targeted fiscal measures to assist recov- manage more effectively the future effects of natural haz- ery), and action was taken to contain revenue spend- ards. The experience of these flood crises highlights ing. Actual figures for revenue, expenditure, and the the need for standardized guidelines to assist with dis- budget deficit were relatively close to budget. aster damage assessment and reporting. The 1998 floods The main adverse public finance impacts of the flood again exposed the uncertainties of postflood forecast- crises are seen in ADP expenditure, which was sub- ing and the need for a full reassessment 18 to 24 months stantially lower than budgeted, in part reflecting the after the disaster (box B.1). government's usual practice of relying on the ADP to bear cutbacks in total expenditure. In the case of the floods of 1987 and 1988, the limited availability of local The Food Account and Food Operations currency resources also constrained the scope of the ADP and may have led to underutilization of donor A distinctive feature of Bangladesh's public finances is funding. In all the flood years, the introduction of flood- the maintenance of a separate set of food accounts for related reconstruction and protection projects placed public sector operations. The origins lie in the post­World Bangladesh: Disasters and Public Finance 85 Box B.1 Uncertainties in postdisaster economic forecasting in Bangladesh Following the 1998 flood in Bangladesh, some adjustments were made to annual economic forecasts for the year. Before the flood, there had been a sense of optimism, at least on the part of the government, that a high rate of growth would be sustained. The government had forecast an annual GDP growth rate of 6.3 percent for 1998/99 (Bangladesh 1999b). Other goals included containment of the inflation rate at 5 percent, the rebuilding of reserves to US$2 billion, and a budget deficit held to 5.3 percent of GDP. The 1998 floods had a devastating impact on these plans. Total losses to output and infrastructure were estimated at US$2 billion (Bangladesh 1999b), equivalent to 6 percent of 1997/98 GDP, and agricultural activities were disrupted. Rehabilitation costs were estimated at US$1.5 billion (UNDP and Bangladesh 1998). The government rapidly announced a series of measures intended to address the macroeconomic implications of the floods. These included pledges of balance of payment and other forms of external assistance, as well as steps by the government to raise revenue and reduce expenditure. Revised economic forecasts, released in October­November 1998 by the World Bank (1998a) and the International Monetary Fund, that took these measures into account esti- mated a GDP growth rate of 3.3 percent, an 8 percent rate of inflation, and year-end foreign reserves of US$1.7 billion. Revised forecasts for 1998/99 released in April 1999 by the government were more optimistic, suggesting higher GDP growth (3.8 percent) and better visible trade performance than had been previously forecast. Actual economic performance was even better than expected, with a growth rate of 4.9 percent. This outcome was primarily attributable to a record postflood crop of dry-season, irrigated boro rice and a bumper wheat crop. Agricultural output expanded by 4.8 percent in real terms, reflecting producer response to losses to the aman (main-monsoon- season) rice crop and a comprehensive program of agricultural rehabilitation. The volume of foodgrain production was 5.6 percent higher than in the preceding year, compared with a preflood projection of only 2.4 percent and ini- tial postflood assessments of a 10 to 11 percent decline in output. Construction activity, boosted by the rehabilitation process, showed an annual growth rate of 8.9 percent. Recovery was less rapid in certain other sectors, particularly in the nonagricultural processing subsectors of manufacturing. Import growth was lower than had been expected in the aftermath of the flood, and foreign remittances increased more than anticipated. Export growth, however, was much lower than forecast, and foreign exchange reserves fell to US$1,525 million in fiscal year 1999, the equivalent of only 2.3 months of imports. Annual inflation stood at 8.9 percent. War II ration system and the country's historical reliance government's response to the 1998 floods because pri- on food as commodity aid. In this study, our focus was vate imports and nonmonetized distribution together on exploring (a) the extent to which food operations kept domestic prices from moving above import parity are related to disaster management and (b) whether a prices. The direct distribution of food has become the separate food account helps insulate public accounts key instrument in the government's disaster response, (especially the development component) from damag- with peaks in direct distribution following all of the ing volatility, or whether it would now be preferable to country's major disasters. incorporate the food account into the budget and As to whether a separate food account remains a help- apply conventional accounting practices. ful means of managing public food operations, our con- Our conclusion is that food operations are an impor- clusion is that there would be an advantage in properly tant, even essential, part of disaster management and incorporating food into the budget; food operations poverty alleviation in Bangladesh. This is illustrated by now represent a smaller share of public expenditure and the clear association between major natural disasters revenue, and there are already close links between and foodgrain off-take or use of public supplies, both food operations and other budgets. Furthermore, financ- for sale (monetized) and for direct distribution (non- ing of public food operations has become more com- monetized). In the case of grain sales, the link is clear- plex over time. Adopting resource-based accounting est for the 1984, 1987, and 1988 floods and the 1991 would offer greater transparency and facilitate finan- cyclone. Monetized operations played no role in the cial planning and budgetary control. 86 Appendix B External Aid represented additional funding for R&R and additional overall expenditure. The developing role and impor- External aid has been a major source of finance in tance of aid channeled through NGOs, and issues of Bangladesh, totaling US$38 billion (in real 1999/2000 coordination and of the information basis for aid action prices) between 1980/81 and 1999/2000. External loans and allocation, also merit further investigation. and grants flow directly to the government and pub- licly owned bodies, as well as through NGOs. The gov- ernment categorizes aid as project aid (which largely Public Agencies and Financial Institutions appears as approved development activities under the ADP), food aid received in kind for relief or develop- Although this study is focused on central government ment, and commodity aid (a residual category that operations, other important bodies are involved in public includes balance of payments support). During the finance operations, and natural disasters also affect them. 1970s, food and commodity aid were the largest cate- gories of external assistance, but in the 1980s, project Local Governments aid emerged as the primary form of planned aid, account- ing by 1999/2000 for 89 percent of total commitments Government in Bangladesh remains highly central- and 81 percent of total disbursements. ized. Local governments lack substantial sources of rev- Therehastypicallybeenasubstantialgapbetweenpro- enue of their own, and local investment is so reliant on jected aid and actual expenditure, reflecting factors such central funding that decisions on its allocation and pri- as procedural difficulties, changes in project staff, pro- orities have been influenced by the central authority. curement delays, and lack of local resources for counter- This, in turn, has led to unequal access to services. The part financing. In 1998/99 total aid commitments rose Fifth Five-Year Plan envisaged extensive decentraliza- by 46 percent, but disbursements increased by only 22 tion of policies, but little practical action has been taken, percent.Theavailabledatasuggestthatmanydonorshave despite the transfer of some major urban flood pre- responded to disaster crises by reallocating resources and vention responsibilities to local agencies. A fuller explo- bringingcommitmentsforward.Foreignaidcommitments ration of the implications of disasters for local have typically increased in the year of, or immediately governments would be helpful. following, a major disaster but have fallen sharply in the following year. Food aid commitments have generally State-Owned Enterprises peaked rapidly in response to major disasters. Project aid has tended to drop in disaster years, as aid is shifted to There is extensive state ownership of industrial and trad- morequicklyavailableemergencyassistance,andthiscan ing enterprises in Bangladesh. Profits from these enter- delaytheprogressofongoingprojectsandthestartofnew prises could in principle be an important source of ones.Thisaspectoftherelationshipbetweendisastersand nontax revenue. In practice, state-owned enterprises foreign aid warrants further investigation. (SOEs) have generally had persistent and mounting A large, although highly variable, amount of aid now losses, placing a continuing burden on public resource flows via NGOs rather than directly to the government. management. Disasters may have contributed to the Data suggest that, in the period immediately following poor performance of SOEs, and assessments of the full a major disaster, at least a quarter of all aid is channeled public cost of disasters need to include the impact on through NGOs, some of which operate as both funding SOEs and the knock-on effects on public finance. and operational organizations and also support smaller NGOs with funding. NGOs have, in effect, established Financial Institutions a parallel structure for international support for disas- ter relief and rehabilitation (R&R). Available evidence Despite a financial liberalization policy and increased on the aid response to the 1998 floods confirms heavy private sector activity, the public sector continues to be involvement by NGOs in R&R; most of this involvement substantially involved in the financial sector. The long Bangladesh: Disasters and Public Finance 87 history of subsidized credit, loan amnesties, and debt 1970, and 1991 was based on the construction of safe restructuring has imposed heavy costs on the budget-- areas and solidly built community shelters. Many people although there is a lack of transparency as to the pre- survived the 1991 cyclone by taking refuge in public cise level of public support to the banking system buildings. The evacuation strategy was even more suc- following disasters. A major shift in policy in response cessful when a cyclone of similar intensity struck in May to disasters took place after the 1998 flood. Although 1997; only 100 fatalities occurred. The strategy now is agricultural credit was increased significantly in 1998, to design and rehabilitate cyclone shelters for multi- it was also made clear that payments on existing loans purpose use, mainly as schools. would be deferred but no longer waived, as they had been during the floods of the 1980s. Case studies, per- Rethinking the Strategy for Mitigation haps focusing particularly on banks specializing in agri- and Preparedness culture and industry, are required to investigate the impacts of disasters on financial institutions and the les- Inherited strategies have been reviewed in the light of sons to be learned. social and economic transformation (population shifts from rural to urban areas, and the increased economic importance of industry and services compared with agri- Microfinance Institutions culture) and of a critical evaluation of the effectiveness Microfinance institutions (MFIs) were pioneered in of the strategy following the floods of the 1980s and Bangladesh in 1976. Currently, more than 1,000 1990s. The new (2001) strategy for water manage- NGOs operate microfinance programs, and there is sub- ment gives priority to protection of concentrations of stantial public sector involvement, through government people and high-value assets in urban and periurban departments, public agencies, and banks, as part of a areas against extreme floods, cyclones, and river ero- broad strategy for tackling poverty. MFIs can play a very sion. The strategy for rural areas is still being formu- constructive role in poverty reduction efforts, but they lated, but coastal areas are a priority. Overall, there is a are vulnerable to damaging liquidity problems in times fresh emphasis on a more decentralized approach, on of disaster. Further analysis is needed to explore ways the involvement of disaster-prone communities them- of improving their resilience to disasters, while main- selves, and on disaster preparedness. The strategy that taining responsible attitudes toward debt repayment seems to be emerging would be more financially real- and ensuring access to credit to those most affected by istic. Central government investment requirements are disasters. reduced, and the otherwise unbridgeable gap between O&M requirements and revenue budget resources is addressed by introducing a policy of beneficiary respon- Disaster Mitigation, in the Long and Short Term sibility and cost contributions to help meet O&M costs. These necessary radical changes pose major challenges The government of Bangladesh and the international for the government and for civil society. community have made substantial investments in dis- aster mitigation and preparedness, focused historically Land Use Controls and Building Regulation on flood protection, drainage, and irrigation. This empha- sis reflects a strategy first applied when Bangladesh's The most effective mitigation strategy against all major population and economy were almost entirely rural. The hazards, drought excepted, is strict land use zoning and objective was to protect and encourage agricultural devel- building regulation. Because of social pressures, no seri- opment and, in particular, to promote self-sufficiency ous attempts are being made to restrict the use of highly in foodgrains. A multipurpose approach was taken, with vulnerable rural areas for economic activity or human the aim of mitigating the effects of all four main disas- occupation. The need for controls in urban areas is ter types. The cyclone protection strategy for high-risk well understood. Urban planning directorates, such as coastal areas at the time of the cyclones in the 1960s, the Dhaka and Chittagong Improvement Trusts, were 88 Appendix B given the responsibility for regulating the siting and more innovative tools, such as weather derivatives method of construction of buildings. The trusts, how- and catastrophe bonds. ever, have been unable to carry out this task because of In Bangladesh it is useful to distinguish between local- massive problems of governance. Instead, the response ized hazards, which affect one or another part of the of the government has been to provide protection through country each year, and more severe, less frequent dis- structural measures, such as embankments and, in coastal asters. Costs relating to the former are met (largely on storm surge areas, cyclone shelters. Building regulation, an ad hoc basis) from the revenue budget. There is a particularly in relation to seismic hazard, has been an strong case for allocating more preassigned funds for issue of mounting concern. Serious risk was taken into this purpose. The World Bank has suggested establish- account in the design of, for example, the Jamuna Bridge ing a disaster sinking fund in the revenue budget, financed in the late 1980s, and a seismic zone map was incor- from regular appropriations and external financing. Sur- porated into the national building code in 1993. Fur- plus funds in any particular year could be used to ther investigation is needed into the hazard risk posed build up reserves. by the rapidly expanding stock of domestic and com- In years of more severe disasters, resources beyond munal buildings in the main urban areas. the means of the government are required. Financial risk transfer mechanisms could potentially play a major role in helping manage costs and spread related expen- Decisions on Rehabilitation and Reconstruction ditures over time. At present, there is little use of even conventional catastrophe insurance by either the public A potential problem in the aftermath of disasters is that or the private sector. Total annual insurance premiums in the haste to fast-track assistance, new projects are not amount to only US$0.61 per capita; natural hazard cov- properly designed or appraised. The opportunity costs erage, which is not included in basic fire and allied poli- of investments in terms of other investment forgone cies, is much less. A significant increase in hazard are not considered. It may be entirely appropriate to insurance would require structural reform of the review investment priorities, but care should be taken insurance industry to be viable. At present, the indus- to ensure that disaster-related projects do not displace try is highly regulated and is almost entirely domestic. other high-priority activities in the ADP that are more Furthermore, a full 50 percent of reinsurance has to be viable and affordable, as well as necessary for support- placed with the state-owned reinsurer. Flood or weather ing long-term growth. index­based insurance may present insuperable prob- lems in Bangladesh because of the practical difficulties in defining appropriate trigger events and determining Financing Future Disaster Costs: Insurance entitlement to compensation. Initially at least, there may and Risk-Spreading be more scope for insuring assets--agricultural machin- Historically, in Bangladesh and in other countries, there ery, commercial buildings, and so on--in urban and has been heavy reliance on donor funding to meet rural areas. disaster-related costs. Donor resources are now declin- A possibility, because of the moral hazard issue that ing globally, and countries are beginning to explore has dogged all credit programs, is to build some form alternative ways of meeting these costs. The World Bank of highly subsidized insurance into formal, already sub- has been at the forefront in helping to develop and sidized, lending. Microcredit is a possible priority area apply appropriate risk transfer mechanisms to cushion for, at least, pilot schemes. If there were even a partial against costs related to natural disasters. Mechanisms insurance element in the contract, this might encour- under consideration typically involve some form of age, when losses occur, an orderly response based on market-based insurance, entailing a large share of rein- an entitlement to refinancing, rather than a damaging, surance in order to transfer risks to the international disorderly collapse in recovery rates. market. Various insurance methods might apply, includ- Insurance coverage against losses in the event of a ing conventional catastrophe damage insurance and disaster can be developed and applied successfully only Bangladesh: Disasters and Public Finance 89 if the government and people of Bangladesh are con- · Much foreign aid has involved reallocating and vinced of its merits. Thus far, additional external assis- bringing forward commitments, rather than pro- tance appears to have been available to meet the costs viding additional funding. There have been shifts of disasters. But this study has shown that ultimately, from project aid to food and commodity aid, and disaster-related external assistance is very largely not concern that some donors have followed their own additional but displaces funding for development. project priorities rather than considering wider implications. · Newer forms of spreading the financial costs of disas- ters,including disaster-earmarked contingency reserves Main Preliminary Findings and Policy Options and market-based instruments for risk-spreading, merit further consideration. Risk-spreading could · Major disasters have had clearly demonstrable neg- safeguard more public finance resources for develop- ative macroeconomic impacts on Bangladesh. The ment and also strengthen microfinance operations, relative severity of the impacts has diminished con- which have a key role in the poverty reduction siderably since the 1970s, largely because of struc- strategy. tural change in the economy. Agricultural output is · There is a need for better assessment of natural hazard more resilient, nonagricultural sectors are more impor- risk, including that from earthquakes, and for strength- tant and are less severely affected, and trade liberal- ened regional cooperation. ization and integration of food markets are reducing price effects. · There have been improvements in disaster response by the government, donors, and civil society. Still, Research Implications some reorganization of public finance reporting (e.g., on food operations) would help clarify and perhaps · The consequences of the 1998 floods were much improve budgetary practice. more closely examined than was true of the floods a · Handling of the 1998 floods by the government of decade earlier, and this helped mitigate and manage Bangladesh and by NGOs was better than in earlier the effects. There is a need for agreed on and gener- disasters. Foreign aid was substantial and timely. ally adopted guidelines covering postdisaster and Improved assessments of losses helped. mitigation assessments. This is a task for Bangladesh. · The aggregate effects of disasters on public finance · Further economic research into natural disasters have been progressively contained. The response, would be valuable. The use of regression analysis to however, has largely involved reallocations of bud- quantify the short-term impacts of disaster shocks geted spending to relief, often by cutting expendi- on national accounting aggregates could be applied ture across the board without regard to the effects to public finance and external assistance. It would on longer-term economic growth. Development proj- also be useful to extend quantitative economic inves- ects funded through the ADP have been particularly tigations to consider the public expenditure and affected. revenue implications of disaster shocks. · There is increasing recognition of the need to reflect · The specific focus of future investigations of disas- policy objectives such as poverty reduction in the ter impacts should be on the composition of public development of strategies for disaster management expenditure to identify key areas of capital and recur- and mitigation. rent expenditure that need to be protected, as well · Investment planning and project prioritization need as areas where additional spending is needed to assist to be strengthened. Disaster mitigation should be immediate crisis management, recovery, and disas- part of project design. Increased expenditure on O&M ter mitigation. could be a cost-effective means of reducing disaster · The full cost of disaster impacts should embrace impacts. public enterprises and agencies and public finance 90 Appendix B operations. Case studies are needed to explore this strategy. The classification of types of aid also needs area more fully because a useful composite analysis to be updated. is not possible with the limited data available. · Natural hazards are extremely complex. State-of-the- · A full inventory of aid from all external donors and art scientific and technical advice is needed to make agencies and of the ways they assist following disas- sense of many issues. A review of the adequacy of ters would be a useful contribution to the integration support for research on and monitoring of hazard of disaster reduction into longer-term development vulnerability is desirable. Appendix C Malawi and Southern Africa: Climatic Variability and Economic Performance The study summarized here was undertaken between Malawi's 2002 Food Crisis: How Might Better November 2001 and July 2002. It focused on climatic Climatic Forecasting Have Helped? variability and the usefulness of climatic information in southern Africa and included a country study of The 2002 food crisis in Malawi, which was emerging Malawi. The study area was chosen because of the during the time of the study, highlights both the reported progress in seasonal climatic forecasting, the important potential gains that good climatic forecast- attention given to the El Niño phenomenon, and con- ing offers, in terms of managing the risks associated with cern about long-term climatic change. climatic variability, and the problems that need to be Malawi is a small, landlocked country in southern overcome to develop and make full, effective use of Africa with an estimated population of 10.8 million in meteorological monitoring and forecasting in the region. 2000. It is one of the poorest countries in Africa; It provides a highly relevant starting point for examin- about 65 percent of the population is below the national ing the linkages and issues at the heart of the study. poverty line, and 28 percent is in extreme poverty. The impetus for improving drought risk management Health and social indicators are among the lowest in at a regional level for southern Africa--an effort involv- Africa. Infant mortality in 2000 was 134 per 1,000 ing regional bodies, national governments, and the inter- live births, compared with an average of 92 for Sub- national community--stemmed from the droughts of Saharan Africa. Average life expectancy (now 37 at 1991/92 and 1994/95. By 1997/98, a formal process birth) is declining as a result of HIV/AIDS, which in for consensus-based, long-lead or seasonal climatic fore- 1999 affected 16 percent of the adult population and casting had emerged, managed through the Southern 31 percent of women in antenatal care. Adult literacy African Regional Climate Outlook Forum (SARCOF). is under 60 percent, and only 78 percent of children Within Malawi, the Department of Meteorological Ser- attend school. Agriculture accounted for some 40 per- vices was providing 10-day bulletins on rainfall, tem- cent of GDP in 2000; its share of GDP has been increas- perature, and sunshine for the meteorological stations ing since the early 1990s as a result of industrial under its control. Despite this progress, the meteoro- stagnation and the contraction of the public service logical input into anticipating and assessing the scale of sector. About 89 percent of the economically active the emerging crisis seems to have been limited. SARCOF's population is classified as rural. Malawi is heavily forecasts were for above-average rainfall in the region in dependent on maize, which is the main food staple and 2000­01 and for broadly average rainfall in 2001­02. in a normal year probably accounts for about three- Because the overriding concern in the region has tradi- quarters of the population's calorie consumption. Export tionally been the risk of drought, the potential negative earnings are dominated by tobacco (61 percent), tea impacts of higher-than-average rainfall were not recog- (9 percent), and sugar (8 percent). This dependence nized. Policy decisions taken in Malawi were therefore on rainfed crops makes Malawi vulnerable to variations predicated on a normal or favorable climatic outlook in rainfall and temperature, as well as to commodity and on what turned out to be an overoptimistic view of price shocks. the likely maize crop. In apparently favorable conditions, 91 92 Appendix C a poverty reduction scheme that provided all small-scale · Takes stock of the current capacity of climatic fore- subsistence farmers with a minimum package of seeds casting and of progress in research to review the range and fertilizer was halved in order to reduce public expen- of potentially useful outcomes diture. As a result, many farmers were unable to respond · Examines the institutional capacity and financing to the early onset of the main rains, and this con- issues that arise if effective use is to be made of strength- tributed to reduced production. Furthermore, on the ened forecasting ability. advice of the IMF, and with World Bank agreement, Malawi sold two-thirds of its strategic grain reserve, Climatic Variability, Agriculture, and Economic which in 2000 was near capacity, to reduce its debt. Performance in Southern Africa The decision was taken prematurely, while planting was still under way and the size of the maize crop was The droughts of 1991/92, 1994/95, and 1997/98 were uncertain. In the event, crop yields were low, with dan- all associated with El Niño events. Climatologists have gerous consequences for food security, and the govern- established a highly significant relationship between the ment had to make replacement purchases of grain that El Niño­Southern Oscillation phenomenon (ENSO) wiped out the savings from the sales. and interannual variations in rainfall in southern Africa. The 2002 crisis was the outcome of many factors, of (See note 3 in chapter 1 for an explanation of ENSO which climate was only one. But a better understand- and related events.) But it is not a simple canonical ing of agro-meteorological relationships, reliable crop relationship: not every El Niño event brings low rain- production data, and less generalized climatic fore- fall, and in some years extremely low annual rainfall is casts for informing economic and food security deci- not clearly linked to El Niño events. Much less well sions would undoubtedly have helped avoid some of understood oceanic-atmospheric interactions in the the extreme consequences of the low crop yields. Indian and South Atlantic Oceans are now recognized There was apparently little understanding of how fragile as important influences on rainfall patterns. Malawi's society and the economy had become, and there Cereal production, especially maize production, is was insufficient appreciation of the sensitivity of the central to food security in southern Africa. It is also maize and tobacco crop to weather conditions throughout highly sensitive to drought and to climatic variation theseasonandthedamagingeffectoferraticrainfalllevels. more generally. In a crisis, ensuring maize supply is In the region at large, an overconcentration on the risks likely to take priority over other trade considerations of drought was in evidence, leading to "undue confi- and in public spending decisions. So, it makes sense to dence," given the highly generalized forecasts of average look first at the impact of climate on cereal and maize or above-average rainfall. Because of financial and human production and how that affects the economies of south- resource constraints, data collected from meteorological ern Africa. South Africa is by far the largest agricultural stations within Malawi were not analyzed and inter- producer in the region, accounting for 64 percent of preted to draw out the agro-meteorological linkages or cereal production and 62 percent of maize production permit the closer monitoring of weather on a local during the period 1993­98. Cereal production in South basis throughout the growing season. This more Africa and the rest of the region is correlated, generally robust data monitoring is essential in order to assess moving in the same direction. and deal with the vulnerability of the important small- The relationship throughout the region between holder agricultural sector. volatility of production and climatic events is striking. As the 2002 crisis demonstrates, major benefits can But the pattern is more complex than one in which a be derived from strengthening climatic forecasting region- drought caused by El Niño results in low crop yields. ally and at the country level. Using evidence up to the Different sequences in drought impacts at the country late 1990s, this study: level--some of them ahead of El Niño­linked droughts-- · Reassesses the economic consequences of climatic are reflected in year-to-year changes in maize yields and variability in the light of experience such as the El in agricultural GDP. In 1997/98 the risks associated with Niño event in 1997/98 the strong El Niño event led climatologists to forecast Malawi and Southern Africa: Climatic Variability and Economic Performance 93 severe drought in southern Africa and very low crop the poorest in the population. The clearest impacts of yields. In fact, although regional crop yields were lower drought are on cereals and especially on production of than normal, the rainfall associated with oceanic activ- and trade in maize. The extreme 1991/92 drought ity in the Indian Ocean resulted in more favorable con- reduced maize production by 10 million metric tons, ditions in countries in the north of the region, and crop and it cost US$1 billion in cereal losses at import yields were higher than had been anticipated by scien- parity prices and US$500 million in the actual logisti- tists using only El Niño­based models. The conclusion cal costs of importing cereal into affected southern we draw from this is that total rainfall is a better explana- African countries. There were also severe wider impacts tory variable than ENSO for analyzing crop yield varia- on GDP and the agricultural sector of at least double tions. Obviously, there are other important factors. this magnitude over 12 months. The climatically less Nevertheless, focusing on rainfall and output makes for severe 1994/95 drought involved costs of US$1 billion better understanding of the consequences of climatic in cereal losses because of higher prices in a tighter inter- variability historically and in the future, with implica- national cereal market. The 1997/98 El Niño event tions for food security and economic policy. caused significant but less serious losses. The effects of Drought has been commonly seen as the main cli- the 2002 crisis are beyond the scope of the study, but mate issue in the region. The recent disastrous floods in the development of another El Niño has led to emer- Mozambique, however, and the role that the extremely gency cereal import costs that exceeded losses in 1997/98. high rainfall in 2000/01 played in the lead-up to the food Costs of this scale require action at the national, regional, crisis of 2002, have highlighted the risks associated with and international levels to prepare an economic strat- high rainfall. Plotting annual cereal and maize outputs egy and to coordinate aid policy. The value of climatic against the southeastern African rainfall index suggests forecasting lies in offering early evidence of enhanced that outputs plateau at rainfall levels about 15 percent risk of a major shock and in anticipating the costs and above the mean levels for 1960­90. Above that, there the scale of measures that may be needed at the national is increased probability of reduced production. A par- and regional levels. allel analysis for Zimbabwe showed a similar pattern. In Malawi, which is at the northern margin of the cli- Climatic Variability and the Malawi Economy maticregion,therewasnosignificantrelationshipbetween crop yields and either the regional rainfall index or ENSO Periods of below-average or erratic rainfall were less variables. There was, however, a link between crop yield extreme and less general in their effects in the 1970s and country-specific rainfall levels for the critical month and 1980s than in the 1990s. The droughts of 1991/92 of February, rather than for the year. Our conclusion is and 1993/94 had very severe impacts on agriculture in that climatic forecasting and early-warning systems need Malawi, and in particular on the smallholder sector, to give more attention to the likelihood of extremely high which accounts for the largest share of maize produc- rainfall events and that localized monitoring and agro- tion. Maize production declined by around 60 percent meteorological interpretation of data are needed to reflect in 1991/92, to the equivalent of only 45 percent of aver- the significant variations between and within countries age production levels in the previous five years. High in the region and to inform critical decisions. and well-distributed rainfall, combined with policies The wider economic impacts of droughts in the region for assisting smallholders, resulted in a bumper maize largely reflect multiplier and linkage effects from the crop and a record tobacco crop in 1992/93. To avoid agricultural sector and are felt in the subsequent year the producer disincentives that might result from these with a lag of 6 to 12 months. very high yields, Malawi's Agricultural Development and Marketing Corporation (ADMARC) made record purchases (over 375,000 metric tons) of maize, adding Costs of Climatic Shocks to the financial pressures on the government. But in As is the case with most natural hazard risks, the liveli- 1993/94, following low and erratic rainfall in key grow- hoods most affected when disaster strikes are those of ing areas, maize production again fell sharply. In 1994/95, 94 Appendix C when South Africa and Zimbabwe were affected by low, been a factor in Malawi's relatively unstable public poorly distributed rainfall, Malawi's agriculture finances. largely recovered. These zonal differences in the pat- tern and timing of drought impacts in 1994 and 1995 Climatic Variability in Southern Africa and the Links highlight important climatic variations both within the to Wider Climatic Processes country and regionally. In 2000/01, following excep- tionally high rainfall and widespread flooding, maize At a general level, the effects of destabilizing climatic production fell by 30 percent and tobacco was down hazards are increasingly well understood and predictable, 16 percent. but there are still important gaps in our knowledge. The The wider economic consequences of drought in a study examined climate variability and links to wider Sub-Saharan African economy like Malawi's include global processes in the light of recent research and events direct impacts on agriculture and on other productive in the study area. sectors that rely on water, such as hydroelectricity, as InthepredominantlysemiaridsouthernAfricanregion, well as indirect multiplier relationships. The overall rainfall varies significantly from year to year, with a pro- pattern of drought impacts on public finances in Malawi nounced seasonal cycle. The rainy season generally has been broadly consistent with standard expected extends from October­November to April, reaching a patterns. Drought severely reduced agricultural pro- peak between December and February. Rainfall distri- duction toward the end of one financial year, and the bution during the rainy season is variable, depending financial effects of relief and recovery assistance fol- on the interplay between tropical and midlatitude weather lowed in the next financial year. Flawed or problem- systems and convective variability. As a result of increased atic data have made it difficult to undertake in-depth temperatures and higher water evaporation rates, future sectoral or wider economic analysis of the effects of global climate change is likely to alter short-term cli- climatic shocks or to isolate the effects of drought. Nev- matic variability and change rainfall patterns, reducing ertheless, the evidence suggests that Malawi's economy water availability. Rainfall during the peak of the wet was among the most sensitive to drought shocks of any season is likely to increase, but with offsetting decreases in the region. in the drier months. Both droughts and floods may Before the 1991/92 drought, there were signs of become more likely, but uncertainty will be greater. improvement in the economy, with export revenue Fluctuations in seasonal rains are linked to regional rising and public expenditure falling. In 1991, how- sea surface temperatures and the global ENSO phe- ever, the combined effects of several factors--the refugee nomenon. The links between ENSO and the regional and trade impacts of the Mozambique conflict; increas- weather system are robust and are relatively well under- ing political difficulties within Malawi that temporar- stood. Models can predict ENSO up to a year in advance, ily halted nonrelief development aid; and the extreme and, using that information, useful predictions of south- drought in 1991/92--resulted in a near-chaotic bud- ern African rainfall can be made at lead times of up to geting situation. Public expenditure rose by 30 percent five months. During El Niño events, southeastern Africa in real terms between 1991/92 and 1994/95, and the is likely to experience a 50­60-millimeter shift toward rate of inflation jumped from 12.5 percent in 1990/91 drier conditions. During La Niña events, models show to 75 percent in 1994/95. Fiscal measures, combined above-normal rainfall in southeastern Africa for all with better agricultural performance, led to a tempo- rainy season months except February. By contrast, equa- rary stabilization in 1995/96 and 1996/97. Neverthe- torial eastern Africa is likely to experience relatively less, public finances in Malawi have continued to be wetter periods during El Niños and relatively drier phases volatile. Upward pressures on expenditure have associated with La Niñas. The severity of the impacts intensified in recent years. Foreign aid levels, on which dependsonthespecificpatternofanextremeENSOevent. development funding depends, have been influenced Climatic zones, of course, do not follow national bound- by political and governance issues, as well as by eco- aries: Malawi lies between the core zones of southeast- nomic and humanitarian considerations, and this has ern and equatorial eastern Africa, increasing the difficulties Malawi and Southern Africa: Climatic Variability and Economic Performance 95 of climate forecasting for that country. Moreover, it is in terrestrial meteorology, so this effect can be con- changes in the distribution of rain during the wet season firmed only by correlating remote-sensing and associated with El Niño events, rather than total rain- agronomic data. Excessive rainfall also disrupts infra- fall amounts, that are crucial to understanding agricul- structure and communications, with associated costs. tural impacts. These changes are complex and difficult · The emphasis on drought risks is understandable, to predict, limiting the precision of forecasts. given the devastating effect of drought. It has, how- ENSO is not the only factor that affects rainfall in ever, led to a perhaps undue concentration on and southern Africa; regional sea surface temperatures and oversimplified interpretation of the impact of El Niño topography are also important. Predictions of sea sur- events on southern Africa's climate, and to an assump- face temperatures of the Indian and Pacific Oceans are tion that if drought is not a prospect, the agricul- used to produce seasonal forecasts for South Africa, and tural season will be good. As another example, a water South Atlantic temperatures also help shape atmos- management strategy in southern Africa focused on pheric circulation. Despite advances in forecasting capa- building up capacity to ensure adequate flows in the bility, for some areas of southern Africa predictability, dry season; in 2000, emergency releases from over- or the "skill" of the forecast, may still be relatively low. full reservoirs then exacerbated downstream flood- Certainly, there are complex relationships that go beyond ing in Mozambique. the influence of El Niño that need to be taken into · Households and the national food system are oper- account in reviewing the potential and actual roles of ating within increasingly narrow margins because of climatic forecasting. socioeconomic pressures--demography, the HIV/AIDS As noted, drought has been seen as the main climatic pandemic, and economic adjustment. They are poten- hazard. This is reflected in the importance accorded to tially more fragile and more sensitive to erratic intrasea- drought management in macroeconomic policy and in sonal distribution of rainfall, which is difficult to the institutional arrangements for disaster management. predict. But more recent events, including the 2002 food crisis In summary, drought remains the most likely source in Malawi, have highlighted other important climate of food crisis and climate-related economic shock. Nev- risks: ertheless, it is now clear that the food system, the · Erratic rainfall, particularly an extended halt in livelihoods of the poor majority of the largely rural pop- rains at the critical flowering time, can considerably ulation, and the wider economy are sensitive to any reduce crop yields, even if total annual or seasonal destabilizing climatic risks. In these circumstances, the rainfall is at or near normal. Food security implica- value of well-coordinated work, supported by adequate tions are particularly serious if there is excessive resources, to improve understanding of the evolving dependence on a single crop, such as maize. Fur- weather situation and build climatic-forecasting ther investigation is needed into the extent and fre- capacity for the region is clear. quency of the phenomenon of midseason dry spells. With increasing cultivation of marginal lands, a useful climatic-forecasting product would be a probability Climatic Forecasting assessment of the likelihood of an erratic rainfall pat- tern and the risk of extended dry periods. Are extended There has been considerable progress toward better inte- dry periods at critical points in the growing season gration and strengthening of meteorological systems closely linked to below-average overall rainfall, or within the Southern African Development Commu- are there other influences on the short-term distri- nity (SADC). SARCOF now provides a formal process bution of rainfall? for consensus-based, long-lead or seasonal climatic fore- · Extremely high rainfall can also reduce crop yields casting. The SARCOF forecasts rely heavily on forecasts through flooding, or perhaps because of reduced solar from global statistical models (which partly reflect the radiation as a result of more extensive and denser behavior of ENSO), with additional details from national cloud cover. Cloud cover is not regularly monitored meteorological services. They are made seasonally, in 96 Appendix C September for October­December and for January­ · Identified priorities for further research to improve March, with the January­March forecast reassessed in forecasting ability December. SARCOF provides forecasts in three broad · Created systems for assessing climatic risk that can probability bands for below-normal, near-normal, and feed into decisionmaking. above-normal total rainfall for the relevant periods, and This study has not assembled complete costing for forecasts are shown for spatial zones with similar rain- forecasting work. The financial costs attributable to fall response. the whole forecasting effort for southern Africa are around The precision of SARCOF forecasts is still very US$5 million, spread across services and research limited, and probabilities are more difficult to assign institutions within and outside the region. These costs for zones farther away from the core areas of south- are modest compared with the economic costs imposed eastern Africa. For example, in the 2001/02 forecasts, by climatic variability in the region, which are estimated the assigned probabilities varied between 20 and 60 to be at least US$1 billion a year. Regional climatic percent, with about a 40­50 percent probability for forecasting needs to be sustained as a learning process. the most likely outcome band. The forecasts are dif- Although long-lead forecasting is still in its infancy, cli- ficult to downscale and are imprecise as to the risks matic research is making rapid progress--for example, of erratic rainfall patterns that have a critical effect on toward including the oceanic influences of the Indian crop performance. Implicitly, the focus has contin- and South Atlantic Oceans in forecasting models. An ued to be on the risk of major drought. The greater important point is that the benefits are not confined to attention now paid to forecasting and monitoring the region. The private sector, the international donor weather through the season, however, ensures that sci- community, and financial institutions are all involved entific data on a 10-daily basis are more rapidly avail- in managing the effects of climatic variability. able to inform assessment and decisions. Global climatic The ultimate test of the usefulness of information is developments are also closely watched, and assess- whether and how it is used, and with what results. A ments are quickly disseminated through the Internet. survey of potential and actual users of forecasts, under- A problem is that decisionmakers would like very clear taken as part of the study, has confirmed the value of predictions ("This climatic event will lead to this pat- forecasting. First of all, country-specific forecasts can tern of weather in the coming months"), particularly alert international and national agencies and NGOs to when food security depends on good crop-growing the need for precautionary measures to safeguard conditions. The reality is that because of the com- food security and water supplies and to reduce the cost plexity of weather patterns and impacts, forecasts often of crisis measures, which could be financially destabi- have to reflect uncertainty. For example, forecasters lizing. But the survey also highlighted problems that identified a high probability of a relatively weak El at present limit the value of forecasts. For example, Niño event toward the end of 2002, but there was great the spatial scale is often not detailed enough; there is uncertainty about what this might mean for the 2002/03 insufficient detail about the distribution of rainfall wet season. In effect, the models were saying that deci- within the wet season; information about the start and sions about an already difficult food security situa- end of the rains is needed; there needs to be sufficient tion had to be taken in circumstances of more than time to respond to forecasts; and users would like more usual uncertainty. information about the accuracy of past forecasts. At Although it is difficult to place a robust value on cli- present, only some commercial farmers are able to matic forecasting, qualitatively its usefulness is clear. respond to more specific seasonal forecasts. Small- Climatic forecasting work has: holders lack the technical options and resources to · Provided a process for scientific consensus modify significantly their choice of crop, seed variety, · Integrated and strengthened meteorological sys- or traditional planting practices. The use being made tems in the region of climatic forecasting is promising, but considerable · Established systems for closer monitoring and report- institutional strengthening and technical capacity build- ing of weather throughout the year ing, more systematic application of current scientific Malawi and Southern Africa: Climatic Variability and Economic Performance 97 knowledge, and investment in data and equipment Southern African agriculture is becoming more sensitive are still required. to climatic shocks. There is evidence of increasing volatil- ity in agricultural indicators such as maize yields and in macroeconomic performance. Factors contributing Conclusions and Recommendations to this fragility include: · Unsustainable agricultural practices. Cereal produc- The findings that emerge from the study of Malawi and tion is stagnating due to failure to follow cropping southern Africa pertain to the interplay between cli- patterns that sustain soil nutrient levels or to increase matic variability, agriculture, and economic performance; fertilizer applications to compensate for the effects the value and limitations of climatic forecasting; and of intensified land use and environmental degrada- the need for better information on climate. Recom- tion. mendations are proposed as appropriate. · Structural change in agriculture. A shift in production to smallholders has not been accompanied by suffi- ciently successful efforts to establish a viable credit Climatic Variability, Agriculture, system, provide support in the form of seeds and and Economic Performance other inputs, and create a supportive marketing struc- The agriculture and economies of southern Africa are ture for smaller producers. highly sensitive to climatic variability--more so than had · Institutional weaknesses that constrain smallholder been recognized. The 2002 food crisis underscored the agriculture and contribute to food insecurity and mal- vulnerability of the region, and especially the rural poor, nutrition. to food insecurity resulting directly from climatic · Political instability in the region. instability and shocks, and not just drought. · Foreign aid that has been influenced by political and The intense impact of droughts between 1981/82 governance issues, as well as by directly economic and the mid-1990s led to a too-narrow preoccupation and humanitarian needs. with drought at the expense of the broader problem of · The effects of HIV/AIDS on human resources, which climatic variability. But agricultural performance is are insidious but are so far largely unquantified. also sensitive to rainfall 25 percent or more above Climatic change may affect the area's hazard risk. average and to intraseasonal variations in the distribu- Although there is as yet no conclusive evidence that tion of rain. Agriculture in the region is likely to per- the region is becoming drier or is suffering more fre- form best when annual rainfall is within a 90 to 120 quent extreme climatic events, both are anticipated as percent band of long-term mean total rainfall. High rain- consequences of global climatic change. A fuller fall, as well as drought, should signal the need for understanding of the environmental and socioeconomic increased concern about regional food security. consequences of variability is needed in order to iso- Both El Niño and La Niña are important influences. An late the forms of climate change and their implications. increased risk of an extreme El Niño event should put the region on the alert for a possible drought and a Climatic Forecasting and How to Increase related food crisis, particularly in countries near the core Its Usefulness of the southeastern African climatic region. But El Niño events alone are not good predictors of agricul- There is an urgent need to reduce vulnerability to climate vari- tural performance; the floods and the poorer agricul- ability and the threat posed by climate change. Critical to tural year in 2000/01 were associated with a La Niña achieving reduction in vulnerability is improvement in event. Countries to the north of the region are more sen- the information that forms the basis for decisions at all sitive to erratic intraseasonal rainfall distribution than levels, from smallholders to national and international to relatively rare low-rainfall or drought years. Cli- bodies. The study looked at what has been achieved and matic variability at the country and subregional levels what remains to be done and asks whether climatic fore- needs closer monitoring. casting should be a priority for international aid and for 98 Appendix C the use of scarce human resources within southern Africa. More rapid reporting of variability would facilitate quicker Among the findings are the following: responses to an evolving situation. · Efforts to improve regional forecasting and provide More systematic research is needed into the relationship better frameworks for disseminating information are between crop performance and erratic and very high rainfall. continuing. · The costs of forecasting (presently estimated at around Information and Public Action US$5 million for long-term forecasting for southern Africa) are modest compared with the very high Poor-quality information was an important factor con- economic losses caused by climatic variability. Even tributing to the 2002 food crisis in Malawi and perhaps a small reduction in losses through improvements in southern Africa at large. Weaknesses in statistical data-- in public decisions and private risk management jus- meteorological, agricultural, and economic--which ham- tifies investment in strengthening forecasting. pered this study, have become more serious during an · Although there is no doubting the usefulness of extended period of budgetary near-chaos in Malawi and forecasting, there is some disappointment about what underfunding of statistical and scientific information has been achieved so far. The discovery of El Niño systems. The HIV/AIDS epidemic may also be eroding effects created unrealistic expectations about the the human resources needed for this work. power and precision of forecasting; the full extent of The availability of good-quality, trustworthy data is a the increasing sensitivity of the region's agricultural necessary condition for effective management of natural dis- economy to variability in general, rather than just to aster risks (and indeed, in all areas of public action). Strength- drought, had not been appreciated; and although ening and sustaining information systems as a public users can see the value of forecasting, their ability to good in low-income countries has to be an international respond is often limited. priority. Forecasting needs to be focused on climatic variability Much remains to be done to make better use of climatic more broadly. This requires more research, downscaled information in public policy at the country, regional, and to zonal levels and intraseasonal timescales. international levels. As soon as there is evidence of an More specific information about the evolving weather sit- enhanced risk of an extreme event, the international uation would be useful to specific groups such as water community, as well as SADC countries, must use the system managers, commercial farmers, and public insti- available information to prepare for aid policy discus- tutions and NGOs working with small farmers. sions and to develop economic strategies for the coun- Closer agronomic-meteorological collaboration is needed tries involved. The experience of the 1991/92 drought to help national and international institutions make led to greater efforts to ensure food security, assess the more effective use of forecast information in their deci- need for humanitarian aid, and prepare for wider eco- sions, including decisions on food security and agri- nomic and financial consequences, but the 2002 crisis cultural support. demonstrated that the situation is not yet satisfactory. Notes Chapter 1 Chapter 2 1. See, for example, Atkins, Mazzi, and Easter (2000) and the 7. The case studies are discussed in more detail in appendixes authors' case studies for Fiji (Benson 1997a) and Montserrat A, B, and C and are documented in full in the separate case study (Clay and others 1999). reports. 2. Despite the considerable literature on disasters in Bangladesh, 8. Other broadly comparable countries made a more rapid but that country is not exceptional. The public finance dimension similar transition, leaving Dominica the least developed of the had received little attention except for official postdisaster assess- former British Caribbean island colonies. ments of individual events such as the floods in 1998. 9. Because bananas are highly sensitive to damage from winds 3. An El Niño event is a periodic warming of the surface waters of 40 miles per hour or more, even the fringe impacts of less in the equatorial zone of the central and eastern Pacific that is severe tropical storms can cause serious damage. Smallholders linked to a reversal of the normal ocean currents within the are ill equipped to bear heavy losses because of their limited assets Pacific. There are teleconnections or associations between El and lack of access to credit. Recovery of banana plantations, how- Niño (and the obverse--a La Niña event) and abnormal weather ever, takes only 9 to 12 months, even when crops are totally worldwide (Stockdale and others 1998; Zebiak 1999). One such devastated. So, where finance for replanting is available and association, especially when an El Niño is intense or sustained, marketing channels are not disrupted, bananas are highly resilient. is an increased risk of low and erratic rainfall across much of By contrast, production of copra, Dominica's other major com- southern Africa (Clay and others 2003). mercial crop in 1979, took three to four years to recover. 4. For example, an official assessment of the costs of the 1998 10. Most of Dominica's road system runs along the narrow coastal floods in Bangladesh aggregated capital losses such as damage strip very near the shore, rendering it highly vulnerable to storm to infrastructure with rice crop losses. An assessment of Hurri- damage.Otherkeyinfrastructurenetworks--telecommunications, cane Lenny in Dominica in 1999 included costs of physical electricity, and water transmission and distribution--run along- damage and reductions in income from small-scale fisheries. side the road and are similarly vulnerable. This practice is quite general: the widely cited estimates of dis- 11. The deepwater port at Woodbridge Bay, built between 1974 aster-related "direct damage" in the Emergency Events Data- and 1978, illustrates the value-for-money case for designing new base (EM-DAT) of the Centre for Research on the Epidemiology infrastructure and buildings to withstand hurricane damage. of Disasters (CRED) include crop losses (reductions in the flow 12. Formally, hydro-meteorological hazards are determined by of agricultural output), along with damage to stocks of assets highly complex, chaotic climatic processes, as illustrated by the such as infrastructure and housing (IFRC 2002). well-known El Niño­Southern Oscillation. 5. There is no generally accepted definition of vulnerability 13. The base period for climatic analysis is 30 years (currently, beyond recognition of sensitivity and resilience as component 1961­90). A much longer run of data is required to assess aspects (Alexander 1997). climate-related risks. Two examples from case study countries 6. Although there is no accepted term covering both atmospheric illustrate the potentially costly consequences of expectations (or climatic) and hydrological hazards, World Disasters Report formed on the basis of short-term experiences. In Dominica, 2002 uses the term "hydro-meteorological" to characterize events which had not experienced a direct hit from a Category 3 or resulting from atmospheric and oceanic processes that are more hurricane for over 40 years prior to 1979, there was wide- recurrent, widespread, and dynamic, under the influence of spread underestimation of the levels of storm-proofing required global climatic change (IFRC 2002). in buildings and infrastructure. In southern Africa the decision 99 100 Notes to Chapter 2 rules established in the relatively wetter 1970s for managing 19. Case study evidence for Dominica and Montserrat indicates water levels in Lake Kariba failed to provide sufficient reserves that both capital and labor mobility are realities in the Caribbean to prevent the 1991/92 drought from disrupting electric power region. supplies to Zimbabwe (Benson and Clay 1998). 20. Preliminary investigations for Indonesia, when it was being 14. Manufacturing and exports of coconut-based soap products, considered as a possible country case for this study, similarly encouraged by a Caribbean Community and Common Market indicated that it might be difficult, because of the country's phys- (CARICOM) agreement, increased during the 1990s in Dominica. ical size, to isolate on a national scale the effect of disasters that The destruction of coconut trees and reductions in yield as a had a large provincial impact. consequence of a hurricane in 1995 had little impact on man- 21. Some of the environmental and geographic factors studied ufacturing production. By then, a substantial share of copra are climate, location (coastal or landlocked), availability of nat- was being brought in from overseas because of high domestic ural resources, agricultural productivity, and incidence of dis- prices and Dominica's very limited production. By contrast, in ease (see, for example, Diamond 1998; Gallup and Sachs 1999). Zimbabwe during the early 1990s, strong intersectoral linkages 22. For example, farmers in more hazard-prone ecosystems of transferred the impacts of the agricultural sector's increased Bangladesh, Vietnam, and other South and Southeast Asian coun- vulnerability to drought to the rest of the economy. tries have been less well placed to take advantage of higher- 15. The governments of Botswana and Namibia have had suffi- yielding but less hazard-tolerant strains of rice (Catling 1994; cient resources of their own to finance substantial relief pro- Hossain, Bose, and Chowdhury 2001). Similarly, some combi- grams, reflecting the importance of the extractive mining sectors nation of greater incidence of natural hazards and (related) higher in both countries (Drèze and Sen 1989; Thomson 1994). incidence of poverty can influence the choice of location of invest- 16. Socioeconomic change related to development can lead to ments. In Vietnam, for instance, this choice is contributing to the breakdown of traditional familial support, declines in tradi- widening regional disparities, as some more hazard-prone regions tional ways of life and the associated coping measures, and the have received disproportionately small shares of both private increased occupation of more hazardous land. These processes and public investment and external assistance (Benson 1997c). are in part associated with urbanization. Increased provision of 23. The postdisaster shift out of agriculture is explained by a infrastructure and services may also alter vulnerability and even combination of a gradual reduction in larger-scale production heighten it. For instance, construction of arterial roads from rural (because of failure to invest fully in replacement), a movement to urban areas can shift cropping patterns away from lower- of smallholders into employment in other sectors, and, follow- yielding and less marketable but more hazard-tolerant, traditional ing Hurricane David in 1979, off-island migration. Hurricane crops and toward more marketable but higher-risk crops. Kelly David resulted in the temporary exodus of almost 20,000 people, and Khan Chowdhury (2002) explore these issues for Bangladesh. equivalent to about a quarter of the predisaster (1978) popula- 17. The gradual integration and deregulation of food markets tion. Among those who left were many school-age children, in Bangladesh have helped reduce the macroeconomic effects and 20 years later, the population had still not recovered to its of major disasters. By contrast, the deliberate rundown of 1978 level. The fisheries sector in Dominica, which provides maize stocks in Malawi in 2001 proved highly costly, partly livelihoods for many poor families, contracted with each major because the private sector had failed to respond to previous disaster. Capital losses are high in every major storm, and some deregulation. The export of maize stocks by Zimbabwe before fishermen, lacking insurance, fail to replace damaged boats and the 1991/92 drought offers an uncomfortable parallel. In Fiji equipment. sugar reserves have been used to maintain export earnings and 24. Writing a decade earlier, Khan and Hossain (1989: 144) prevent loss of export markets in the aftermath of natural dis- also concluded that "inadequate infrastructural facilities consti- asters (Benson 1997a). tute a serious obstacle to the economic and social development 18. Bangladesh faces severe global competition in the export of of Bangladesh" and that "the physical infrastructure is in urgent ready-made garments. By contrast, it was the world's primary need of rehabilitation and expansion" (181). jute producer and, as such, was a price setter on the interna- 25. The continuing volcanic crisis in Montserrat provides an tional market. Disruption to the production of ready-made gar- extreme example of the long-term impacts of a disaster. The ments could result not only in the direct loss of export revenue present eruption, which began in 1995, has had a devastating but also in the loss of markets overseas. impact on the economy, with serious implications for the island's Notes to Chapters 2 and 3 101 medium- and long-term development. Most of Montserrat's admin- additional measures to counteract the budgetary impacts of the istrative, commercial, and industrial facilities have been destroyed, floods, including the imposition of a 6 percent surcharge on and the crisis has forced fundamental changes in the economic income tax; the collection of an additional surcharge of 5.1 per- structure. It has also had serious ramifications for financial cent on excise duties on certain items; deduction of a 4 percent institutions, precipitating high rates of default on outstanding relief and rehabilitation levy from dividend and interest incomes loans. The economy will not be viable in either the short or the on balances in savings and fixed-deposit accounts; and a levy medium term without large-scale subsidies from the United of 4 percent on telex and telephone bills. Expenditure on the Kingdom. At the individual level, Montserratians have faced loss Annual Development Program (ADP) was almost 20 percent of livelihoods and of other assets, including savings. Demo- lower than budgeted, despite flood-related expenditure, imply- graphic effects have also been massive; over 50 percent of the ing that spending on some projects was much lower than planned. residents have left. The country has been fragmented by mass 32. Before the eruption, Montserrat had a resident population migration and relocation, and community and household of 12,000 in an area of 100 square kilometers. It was a middle- structures have broken down (Clay and others 1999). income country, with GDP per capita of US$3,600 in 1994. In 26. In Zimbabwe, for instance, goats are often kept as a form of 1997 revenue receipts (excluding budgetary assistance in the savings to pay for secondary education. In the aftermath of the form of a special grant) totaled only 59 percent of the 1993­94 1991/92 drought, a number of households were forced to sell average, and only 35 percent of the figure for 1998. Meanwhile, their goats to sustain short-term levels of consumption, with government expenditure increased by 56 percent in real terms, implications for longer-term investment in human capital (Hicks despite a decline in population. As a consequence, Montserrat 1993). became the recipient of budgetary support for the first time since 27. This process of pauperization is well documented--for exam- 1981. According to British government procedures, if an over- ple, in drought-prone areas in the Ethiopian highlands (Dev- seas territory receives budgetary aid on a regular basis, or is likely ereux, Sharp, and Amare 2002). to do so, the finances of that territory must come under the super- vision and, in effect, control of the secretary of state for inter- national development. Accordingly, as a direct consequence of Chapter 3 the volcanic crisis, the island's government lost any semblance of financial independence (Clay and others 1999). 28. The authors have so far been unable to identify any other 33. For instance, according to a UN damage assessment under- in-depth retrospective analysis of disasters and public finance, taken in the wake of the 1998 flood in Bangladesh, US$186 apart from single-event studies. million was required for the repair of damage to roads and 29. For example, despite the 1988 floods, Bangladesh's overall highways alone, yet the 1999/2000 ADP contained no projects budget deficit for 1988/89 was actually lower than had been orig- in this sector whose titles indicated that part or all of the expen- inally projected. Total revenue was higher, both relative to bud- diture was intended to address damage resulting from the geted figures and compared with actual revenue in all earlier flood. years during the 1980s. Meanwhile, recurrent expenditure on 34. In Dominica capital investment projects relating to post- public food operations, via the Food Account, and on the rev- disaster rehabilitation and reconstruction are typically not enue budget (effectively, all recurrent expenditure) was only mar- identified as such in annual budget statements. But a large part ginally higher than budgeted, despite increased flood-related of the increase in capital expenditure between 1983/84 and expenditure. 1985/96--that is, up to seven years after Hurricane David-- 30. The block allocations, which are comparatively new, are sub- could be attributed to major road investment projects necessi- stantial and increasing. For example, in the 1998/99 budget they tated in part by the hurricane. Part of this increase in capital totaled 11.2 billion taka (Tk), or 7 percent of the total gross rev- expenditure would have been required in any case to make up enue budget. Of this, Tk 5.3 billion was allocated for unexpected for years of inadequate maintenance and low investment. expenditures. 35. The figure of 20 percent should be put in context. Actual 31. The most serious pressures on the separately administered ADP expenditure is, in any case, typically lower than planned; Food Account were not felt until 1989/90, when stocks were over the period 1980/81 to 1998/99, it averaged 89 percent of rebuilt. Meanwhile, revenue was boosted by the introduction of the budgeted amount. 102 Notes to Chapter 3 36. In the particular case of Bangladesh following the 1988 flood, of funds is permitted, again potentially contributing to a devi- the World Bank commented that "while this is a prudent approach, ation away from stated objectives in the wake of a disaster. it is not necessarily desirable in all situations, particularly when 41. Foster and Fozzard continue: "Unfortunately, such indis- (as at present) domestic demand is stagnant, the inflation rate criminate cuts ignore spending priorities and the differing com- is declining, the banking system has adequate liquidity, and there position of expenditure, particularly as regards non-discretionary are substantial donor resources available to assist in flood reha- items. Where a substantial proportion of sector expenditure is bilitation that are likely to be under-utilized because of short- dedicated to payroll, as in the case of the social sectors, cuts on ages of local funds. In this situation, increased deficit financing discretionary items are likely to be more severe than in the sector can be helpful to the economy, provided that the government is with a smaller payroll component. In all sectors, cuts will be prepared to take subsequent measures to increase revenues and directed at consumables. In some cases this will mean that staff restrict expenditures so that the economy does not overheat" continue to be paid although they lack the basic materials nec- (World Bank 1989: para. 2.19). essary to deliver services. Investment projects are another common 37. In 1988 the government of Bangladesh lifted restrictions on target of cuts in expenditure, leading to the postponement of the import and manufacture of agricultural pumping equipment, projects or the failure to meet commitments with donors for the encouraging an accelerated expansion of private investment in financing of internal contributions" (2000: 18). lift irrigation and an associated rapid growth in production, espe- 42. The ProVention Consortium is "a global coalition of gov- cially in output of boro (dry-season) rice. The benefits of this ernments, international organizations, academic institutions, the expansion were particularly noticeable in terms of agricultural private sector and civil society organizations dedicated to increas- resilience in the aftermath of the 1998 floods. ing the safety of vulnerable communities and to reducing the 38. The accounting treatment of bonds for investment, recapi- impact of disasters in developing countries" (from the ProVen- talization, and replacement of nonperforming loans is also ambigu- tion Website, www.proventionconsortium.org/index.htm; accessed ous. Foreign aid is shown as a receipt for financing the ADP, but November 5, 2003). the loan component of aid is not included in public debt 43. In Bangladesh the budgetary difficulties resulting from the (World Bank 1996). A further issue concerns the practice of cash 1988 floods made it exceptionally difficult to prepare the budget accounting, in which any internal transactions go unreported. for the fiscal year 1989/90 because of the implied additional 39. The Electricity Commission of Malawi (ESCOM) incurred financial demands and the uncertainties created by the crisis. losses as a consequence of the drought because tariffs did not keep At one point during the preparation of the 1998/99 budget, it pace with inflation. Nonpayment also peaked in the crisis year of was even suggested that there would be a deficit in the revenue 1992. The National Railways of Zimbabwe ran a deficit in 1992/93 (recurrent) budget, implying no surplus--or resources--for owing to increased operating costs as a direct consequence of the the ADP. The ADP had to be reworked again and again as pri- drought-relatedimportprogramthat,amongotherimpacts,obliged orities and estimates of the available budgetary envelope changed. it to hire wagons from South Africa and elsewhere (EC 1993). 44. As Foster and Fozzard (2000: 12) state, "The budget cycle 40. In Bangladesh the basic medium-term planning exercise is needstobenestedwithinalongertermpolicyandplanningprocess, the five-year plan. A three-year rolling plan was introduced in which provides a clear link from planning to the allocation of fiscal 1991, with the terminal year of each plan coinciding with resources...anannualbudgetistooshortatimeframeforaddress- the first year of the next. The two types of plan are used as the ing development priorities, which require sustained implementa- basis for drawing up the ADP. Projects within the ADP are tion of policies and reforms over a medium to long term period." examined in terms of their compatibility with these plans, as 45. Fozzard and others (2001) report that many OECD gov- well as their cost, technical quality, and economic viability, but ernments have already introduced a medium-term expenditure political pressures and the availability of external assistance for framework (MTEF) and that a number of developing countries particular activities also play a role. Thus, development priori- are also embarking on this process. The MTEF consists of a top- ties may already be compromised before the occurrence of a down resource envelope consistent with macroeconomic sta- disaster. Problems are exacerbated by the slow process of for- bility and explicit strategic priorities; a bottom-up estimate of malizing budgetary allocations and reallocations; postdisaster the current and medium-term costs (both recurrent and invest- expenditure is sometimes undertaken before it has been formally ment) of existing and new policies, which are also reviewed to approved. Some flexibility (varying among ministries) in the use verify their consistency with overall government priorities and Notes to Chapter 3 103 spending limits; and an iterative decisionmaking process that provision of infrastructure to support growth in agriculture, man- matches these costs with the available resources. ufacturing, and tourism. The weak infrastructure base has been 46. In 1999, 20 years after Hurricane David had inflicted severe consistently identified as a critical constraint on the pace of devel- damage, and almost a decade after the first comprehensive sea opment, limiting the country's ability to attract and sustain new defense protection plan was completed, Hurricane Lenny again productive investment. Part of this weakness relates to the con- exposed the inadequacies of sea defenses and the vulnerability tinuing vulnerability of the internal transport and communica- of the road network and other infrastructure along the coast. tion network to adverse weather systems, necessitating expenditure 47. The study demonstrates the potential benefits of structural to rehabilitate roads in the aftermath of storms. The indirect fiscal mitigation through a retrospective analysis of public and pri- impact of disasters has been an additional factor contributing vate projects in the Caribbean that have suffered damage from to limited government saving and restricting the availability of tropical storms. One project examined was the deepwater port counterpart financing. The lack of counterpart financing is in Dominica, which was constructed by the government to handle reported to have "led to the non-implementation or the defer- banana exports more efficiently and to lower the handling costs ral of important projects in the social sector, notably in hous- of imports. A year after completion of the facility, Hurricane ing, water and sewerage" (Dominica 1998: 24). David struck, causing reconstruction costs equivalent to 41 51. The strains that the Philippines' rising population places on percent of the cost of the original port. The study estimated that the country's ability to provide sufficient classrooms and other had the original facility been built to a higher standard, able to social infrastructure have been exacerbated by the damage inflicted resist Category 4 hurricanes (an option rejected on grounds of by natural hazards. Indeed, the National Physical Framework Plan, cost), investment costs would have been only about 12 percent 1993­2022, lists this damage and the consequent redirection of higher (Vermeiren, Stichter, and Wason 1998). resources (which in turn hampers the implementation of other 48. Bangladesh is arguably a notable exception, at least to the infrastructural projects), as one of six key issues and concerns in extent that considerable public resources have been invested in the development of infrastructure (Philippine NLUC 1992). flood control under the highly interventionist flood control, 52. In the Philippines the 1990 earthquake and the 1989­90 drainage, and irrigation (FCDI) strategy launched in the 1960s. drought were reported to have contributed to a 6.7 percent It is difficult, however, to estimate total spending on disaster- increase in total external debt in 1990 and a 22.4 percent increase related activities, other than in terms of the substantial propor- in debt owed to official creditors alone (Ernst & Young 1991). tion of development expenditure absorbed by the water resources An examination of the impact of the mid-1980s drought on exter- sector. Furthermore, the proportion of FCDI accounted for by nal borrowing in six Sub-Saharan African countries revealed that disaster reduction is difficult to isolate. Expenditure on mitiga- the growth rate in total debt stocks accelerated during the year tion of other hazards, most obviously earthquakes, may be far of most severe drought in five of the six (Benson and Clay 1998). too low from an economic perspective. The exception, Zimbabwe, had been pursuing a deliberate 49. There is "a significant bias toward capital expenditures, driven long-term policy of debt reduction. by governments which perceive the current coverage of services 53. The Japanese Overseas Economic Cooperation Fund (OECF) and infrastructure to be inadequate and the expansion of service extended a loan to the Philippine government for the con- networks as a priority. . . . One of the results of this capital bias is struction of rural roads in northern Luzon that were subse- toreducethefundsavailableforO&M,leadingtoinadequatefund- quently damaged by typhoons (Benson 1997b). Such loans ing of service provision and the gradual degradation of capital cannot be canceled. investmentsandthequalityofpublicservices"(Fozzardandothers 54. By 1991, Zimbabwe's economy had failed to meet certain 2001:46).Theprojectbiasofaidtowardadditionalityfavorsinvest- extended SAP targets, most notably falling 0.6 percentage point ing in new physical and human resource development rather than below the target GDP growth rate. Some commentators believed making up for inadequate recurrent spending. that the economy was unlikely to meet target performance indi- 50. The Dominica government states that "the fiscal burden [of cators in 1992, either, and by the end of 1991 the government natural disasters] has been significant necessitating the diver- was beginning to warn of adverse short-term conditions as an sion of scarce resources from programmed activities" (Dominica unavoidable part of the structural adjustment process. The sub- 2000: 4). To achieve one of its principal goals of economic diver- sequent drought certainly impeded the progress of adjust- sification, the government has placed particular emphasis on the ment, and a number of the original economic targets for 1992 104 Notes to Chapters 3 and 4 and 1993 were not achieved. In particular, the drought hampered 61. For example, there is a large gap between import and efforts to reduce the budget deficit (and thus to reduce domes- export parity prices for cereals in the landlocked countries of tic borrowing) and to restructure the civil service and parastatal southern Africa. Failure to recognize the increasing risks of an bodies. As a consequence, the expected domestic supply-side impending poor production season has resulted in decisions to response to the reform program, which was critical to its suc- run down national food security stocks of maize, necessitating cess, was partly curtailed. But the drought was retrospectively costly imports as replacement stocks. This happened in Zim- seen by the Zimbabwe government, as well as by the World babwe in 1991 and in Malawi in 2001. Liberalization of inter- Bank and the IMF, as the principal cause of the economic diffi- nal and external markets for cereals shifts part of the culties experienced in 1992, rather than as a factor exacerbating decisionmaking to the private sector. If the private sector mis- existing problems. Cynics might even argue that the drought pro- judges the situation, governments and the international com- vided a convenient scapegoat for the country's economic diffi- munity face the contingent liability of ensuring food security culties, allowing the IMF and the World Bank, which were keen and preventing famine at almost any cost. for a "success" story of adjustment, to continue to hold up Zim- 62. Even more problematic than the shelter program has been babwe as a potential triumph. Meanwhile, the drought enabled the construction and maintenance of the network of coastal the Zimbabwe government to sustain the reform process by avoid- embankments to mitigate the effects of storm surges. There can ing a large decline in public support (Benson 1998). be, for example, rivalry between rice producers, mostly small 55. John Roberts, Overseas Development Institute, personal farmers, and large-scale shrimp producers, who sometimes breach communication, February 2003. the embankments to admit saline seawater, thus jeopardizing 56. General contingency reserves are often used to meet the cost the protective function of the works. of any public sector wage increases negotiated during the year. 63. In 2001 the government of Bangladesh contributed 56 per- cent of the recurrent operational costs (totaling US$460,000) of the Cyclone Preparedness Program managed by the Bangladesh Chapter 4 Red Crescent. The International Federation of Red Cross and Red Crescent Societies covered the remainder (IFRC 2002). 57. Recent important theoretical and policy contributions to 64. Monitoring units have been installed, one in each partici- the subject include Kaul, Grunberg, and Stern (1999) and Ferroni pating country. Such a project raises problems of sustainability, and Mody (2002). The work by Cornes and Sandler (1996) is and trust funds amounting to US$50,000 have been agreed on widely regarded as the fullest statement of the theoretical for the maintenance of each unit. In the light of the damage suf- framework on which these discussions are based. fered by some units during Hurricane Lenny, further expendi- 58. The World Bank (2001) estimated total annual average aid ture will be required to improve storm resistance (information commitments during the period 1995­98 at US$5 billion for derived from www.cpacc.org). core IPG and US$11 billion for complementary activities, 65. As yet, there are no models that integrate real-time meteor- yielding a total of US$16 billion. On a narrower basis, te Velde, ological information for upstream areas in India to predict Morrissey, and Hewitt (2002) estimated total annual average IPG flood hazards in Bangladesh. Warnings depend on direct commitments during 1996­98 as US$9 billion, the equivalent assessments by dam and river engineers that go through official of 9 percent of ODA. channels within India and are then passed on to authorities in 59. The spatial scale of forecasts is often not detailed enough, lower riparian Bangladesh (Chapman and Rudra 2002). and there is insufficient detail about the distribution of rainfall 66. Since 1967, three national risk zone maps have been pro- within the wet season. Information about the start and end of duced for Bangladesh. The latest (1993) version is reproduced in the rains is needed. There also needs to be sufficient time to Benson and Clay (2002a: map 3). Local geologists acknowledge respond to forecasts. Finally, users would like more informa- the tentative nature of these assessments, which are based on the tion about the accuracy of past forecasts. inadequate available data. In addition, visual inspection suggests 60. The International Development Association (IDA) grants that the latest Bangladesh map is inconsistent with the seismic concessional assistance to low-income countries. It and the Inter- hazard map produced in India for neighboring West Bengal State. national Bank for Reconstruction and Development (IBRD) make 67. The SRU is an autonomous entity within the University of the up the World Bank. West Indies, St. Augustine Campus, Trinidad. Its core funding Notes to Chapters 4 and 5 105 is provided by Trinidad and Tobago; 20 percent of the funding 72. The World Bank is providing a contingent credit line to a comes from Barbados, and 30 percent comes from six other juris- compulsory earthquake insurance scheme established in certain dictions: Antigua, Dominica, Montserrat, St. Lucia, St. Vincent areas of Turkey in the wake of the 1999 Marmara earthquake. and the Grenadines, and St. Kitts and Nevis. Additional funds The credit line provides initial capitalization of a newly created are obtained from specific contracts such as that for providing insurance pool, ensuring its financial solvency should an earth- seismic monitoring for Netherlands overseas territories. The SRU quake occur while funds are being accumulated. has faced pressures to reallocate core staff to university teach- 73. The one apparent exception, United Insurance, has actively ing (Clay and others 1999). The islands of Martinique and Guade- promoted structural mitigation. Premium discounts of up to 40 loupe, between which Dominica is sandwiched, are part of the percent are available for retrofitted commercial properties, with French national monitoring system and are not in the same reductions of 17 to 25 percent for retrofitted domestic proper- seismic network as Dominica. The U.S. territories rely on the ties. The program, which was officially introduced only in U.S. Geological Survey. The Caribbean Disaster Emergency 1997, has already achieved impressive results. The average cost Response Agency (CDERA), an organization that supports dis- of claims on affected risks following Hurricane Jose in Antigua aster preparedness and disseminates information, is confined in 1999 was equivalent to 10 percent of the total sum insured to the former U.K. colonies and the remaining U.K. overseas but to only 4.75 percent of the sum insured in the case of retro- territories. The OAS, which supports disaster mitigation and loss fitted projects. Although the program was not intended as a mar- reduction, does not include European overseas territories. keting tool, it has generated new clients for the company in 68. H. Alderman, lead human resource economist, Africa Region, Antigua. But although United has considered making its miti- World Bank, personal communication, November 2002. gation program mandatory for all its insurees, as the sole com- pany offering such a scheme it has concluded that, on balance, Chapter 5 it would probably lose clients if it did so. 74. This coverage was estimated by an insurance industry 69. Insurance payouts can be made in as little as 24 hours after spokesperson at about 90 percent of businesses (including fac- a disaster if the adjusters can be organized quickly. tories) and 18­20 percent of households in 1996. The securing 70. The law of demand states that the lower the price of a prod- of mortgages is conditional on the acquisition of cyclone insur- uct, the greater the amount demanded. The position and shape ance, which can be obtained only on presentation of a certifi- of a consumer's demand curve--that is, the amount of a product cate confirming compliance with the 1985 National Building demanded for a given price--will depend on ability to pay, as Code. Even then, the system may break down in a softer determined by the financial resources at the consumer's command, market. As of early 1996, the Fiji insurance market had become and on willingness to pay, as determined by the utility (satisfac- highly competitive following record profits in the previous tion or need fulfillment) derived from consumption of a particu- (disaster-free) year. This attracted new entrants who were rumored larquantityofagood.Demandcurvesforallconsumersinamarket to be demanding no cyclone-proofing certification for issuing can be aggregated to produce a market demand curve showing cyclone insurance policies. the total amount consumers would like to purchase at each price. 75. Following an escalation in August 1997 of Montserrat's 71. For example, the World Bank (1998b) recommended com- continuing volcanic crisis, international insurers suddenly can- plementary actions by both the public and the private sectors celed most insurance policies on the island. Because the small to help overcome problems relating to the safety of insurance economy was a marginal part of their business, they could coverage in the Caribbean region. Companies and households exit with little impact on their overall portfolios. With the in the region should be encouraged to establish financial reserves cancellation of insurance policies, mortgaged assets immedi- to supplement insurance and cover uninsurable losses. Gov- ately assumed a zero value, resulting in the effective collapse ernments should consider the establishment of reserve funds of the country's only building society, the locally registered that could be drawn on for infrastructure repairs. The OAS has Montserrat Building Society, and this had multiplier impacts similarly recommended both incentives and requirements for throughout the economy, as well as severe societal effects (Clay the creation of financial reserves in the Caribbean region. and others 1999). References Aghion, P., and P. Howitt. 1998. Endogenous Growth Theory. Cam- ------. 1999b. "Memorandum for Bangladesh Development bridge, Mass.: MIT Press. Forum 1999­2000." Economic Relations Division, Ministry Ahmed, A. W. Nuruddin, Lutful Haque Chowdhury, and of Finance and Planning Commission, Ministry of Planning, Steven Haggblade. 2000. "History of Public Food Interven- Dhaka. tions in Bangladesh." In Raisuddin Ahmed, Steven Haggblade, ------. 2000. "Memorandum for Bangladesh Development and Tawtiq-e-Elahi Chowdhury, eds., Out of the Shadow of Forum 2000-2001." Economic Relations Division, Ministry Famine: Evolving Food Markets and Food Policy in Bangladesh. of Finance and Planning Commission, Ministry of Planning, International Food Policy Research Institute. Baltimore: Johns Dhaka. Hopkins University Press. ------. 2002. "Bangladesh: A National Strategy for Economic Albala-Bertrand, J. M. 1993. The Political Economy of Large Nat- Growth and Poverty Reduction." Economic Relations Divi- ural Disasters with Special Reference to Developing Countries. sion, Ministry of Finance, Dhaka. Oxford: Clarendon Press. Benson, Charlotte. 1994. "Book Review of The Political Economy Alexander, David. 1997. "The Study of Natural Disasters, of Large Natural Disasters with Special Reference to Developing 1977­1997." Disasters 21 (4): 284­304. Countries, by J. M. Albala-Bertrand." Disasters 18 (4): 383­86. ALNAP (Active Learning Network for Accountability and Per- ------. 1997a. "The Economic Impacts of Natural Disasters in formance in Humanitarian Action). 2002. Annual Review 2002. Fiji." ODI Working Paper 97. Overseas Development Insti- Humanitarian Action: Improving Performance through Improved tute, London. Learning. London: ALNAP, Overseas Development Institute. ------. 1997b. "The Economic Impacts of Natural Disasters in Anderson, Mary B. 1991. "Which Costs More: Prevention or the Philippines." ODI Working Paper 99. Overseas Develop- Recovery?" In Alcira Kreimer and Mohan Munasinghe, eds., ment Institute, London. "Managing Natural Disasters and the Environment: Selected ------. 1997c. "The Economic Impacts of Natural Disasters in Materials from the Colloquium on the Environment and Nat- Viet Nam." ODI Working Paper 98. Overseas Development ural Disasters Management." World Bank, Washington, D.C. Institute, London. Apthorpe, Raymond. 1984. "Agriculture and Strategies: The Lan- ------. 1998. "Drought and the Zimbabwe Economy, 1980­93." guage of Development Policy." In Edward J. Clay and Bernard In Helen O'Neill and John Toye, eds., A World without Schaffer, eds., Room for Manoeuvre. London: Heinemann. Famine? London: Macmillan. Aspinall, William P., and others. 2002. "Evidence-Based Vol- ------. 2003a. "The Economy-wide Impact of Natural Disas- canology: Applications to Eruption Crises." Draft. Department ters in Developing Countries." Draft doctoral thesis. Univer- of Earth Sciences, University of Bristol, U.K. sity of London. Atkins, J. P., S. Mazzi, and C. D. Easter. 2000. "A Commonwealth ------. 2003b. "Macroeconomic Concepts of Vulnerability: Vulnerability Index for Developing Countries: The Position Dynamics, Complexity and Public Policy." In Greg Bankoff, of Small States." Commonwealth Economic Paper Series 40. Georg Frerks, and Dorothea Hilhorst, eds., Mapping Vulnera- Commonwealth Secretariat, London. bility: Disasters, Development and People. London: Earthscan. Bangladesh. 1999a. "Annual Budget 1999­2000--Budget Speech: Benson, Charlotte, and Edward J. Clay. 1998. TheImpactofDrought Part I. Presented by Shah M. S. Kibria, Finance Minister, on on Sub-Saharan African Economies: A Preliminary Examination. June 10, 1999." Ministry of Finance, Dhaka. World Bank Technical Paper 401. Washington, D.C. 107 108 References ------. 2000. "Developing Countries and the Economic Impacts Disaster Risk Management Working Paper Series 7. World of Natural Disasters." In Alcira Kreimer and Margaret Bank, Washington, D.C. Available at http://www.provention- Arnold, eds., Managing Disaster Risk in Emerging Economies, consortium.org/files/malawi_full_doc.pdf. 11­21. Disaster Risk Management Series 2. Washington, D.C.: Cochrane, Harold C. 1994. "Disasters, Indebtedness and Fal- World Bank. tering Economic Growth." Prepared for the 9th International ------. 2001. "Dominica: Natural Disasters and Economic Devel- Seminar on Earthquake Prognostics, September 9­23, San opment in a Small Island State." Disaster Risk Management José, Costa Rica. Hazards Assessment Laboratory, Colorado Working Paper Series 2. World Bank, Washington, D.C. Avail- State University, Boulder. ableathttp://www.proventionconsortium.org/files/dominica.pdf. Cornes, Richard, and Todd Sandler. 1996. The Theory of Exter- ------. 2002a. "Bangladesh: Disasters and Public Finance." Dis- nalities, Public Goods and Club Goods. 2d ed. Cambridge, U.K.: aster Risk Management Working Paper Series 5. World Bank, Cambridge University Press. Washington, D.C. Available at http://www.proventioncon- del Ninno, C., P. A. Dorosh, L. C. Smith, and D. K. Roy. 2001. sortium.org/files/Bangladesh.pdf. "The 1998 Floods in Bangladesh: Disaster Impacts, Household ------. 2002b. "Disasters, Vulnerability and the Global Econ- Coping Strategies, and Response." IFPRI Research Report 123. omy." Paper prepared for ProVention/World Bank conference International Food Policy Research Institute, Washington, D.C. on "The Future Disaster Risk: Building Safer Cities," Decem- Denison, E. F. 1967. Why Economic Growth Rates Differ: Postwar ber 4­6, Washington, D.C. Experience in Nine Western Countries. Washington, D.C.: Brook- Briguglio, L. 1995. "Small Island Developing States and Their ings Institution. Economic Vulnerabilities." World Development 23 (9): 1615­32. Devereux, Stephen. 2002. "The Malawi Famine of 2002: Causes, Brown, Warren, and Craig F. Churchill. 1999. "Providing Insur- Consequences and Policy Lessons." Institute of Development ance to Low-Income Households. Part I: A Primer on Insur- Studies, Brighton, U.K. ance Principles and Products." Bethesda, Md.: Microenterprise Devereux, Stephen, Kay Sharp, and Yared Amare. 2002. "Des- Best Practices. titution in the North-eastern Highlands (Amhara Region)." Catling, H. David. 1994. Rice in Deep Water. London: Macmillan. Interim report for discussion at policy consultation workshops Chapman, Graham P., and Kalyan Rudra. 2002. "Society and in Bahr Dar and Addis Ababa, November 2002. Institute of Water: Bengal's Millennium Flood." Department of Geogra- Development Studies, Brighton, U.K., and SC-UK, Addis phy, Lancaster University, U.K., and Department of Geogra- Ababa. phy, Habra S.C. College of Commerce, West Bengal, India. Diamond, Jared. 1998. Guns, Germs and Steel: A Short History of Charvériat, Celine. 2000. "Natural Disasters in Latin America Everybody for the Last 13,000 Years. London: Vintage. and the Caribbean: An Overview of Risk." Research Depart- Dominica, Commonwealth of. 1998. "Dominica: Medium-Term ment Working Paper 434. Inter-American Development Bank, Economic Strategy Paper." Roseau. Washington, D.C. ------.2000. "Dominica: Medium Term Strategy 2000­2002." Clay, Edward J. 1985. "The 1974 and 1984 Floods in Bangladesh: Roseau. From Famine to Food Crisis Management." Food Policy 10 (3): Drèze, Jean, and Amartya Sen. 1989. The Political Economy of 202­06. Hunger. Oxford: Clarendon Press. Clay, Edward J., and others. 1995. "Evaluation of ODA's Response EC (European Commission). 1993. The Courier, no. 139 to the 1991­1992 Southern African Drought." Evaluation (May­June). Brussels. Report EV568. Overseas Development Administration, London. ECLAC (Economic Commission for Latin America and the Clay, Edward J., Christine Barrow, Charlotte Benson, Jim Caribbean) and IADB (Inter-American Development Bank). Dempster, Peter Kokelaar, Nita Pillai, and John Seamans. 1999. 2000. "A Matter of Development: How to Reduce Vulnera- "An Evaluation of HMG's Response to the Montserrat Volcanic bility in the Face of Natural Disasters." Paper prepared for Emergency." 2 vols. Evaluation Report EV635. Department seminar on "Confronting Natural Disasters: A Matter of Devel- for International Development, London. opment," March 25­26, New Orleans. Mexico City, and Clay, Edward J., Louise Bohn, Enrique Blanco de Armas, Singand Washington, D.C. Kabambe, and Hardwick Tchale. 2003. "Malawi and South- ECLAC (Economic Commission for Latin America and the ern Africa: Climatic Variability and Economic Performance." Caribbean) and IDNDR (United Nations International Decade References 109 for Natural Disaster Reduction). 1999. "Manual for Estimat- Disaster Management Facility Working Paper Series 2. Urban ing the Socio-economic Effects of Natural Disasters." Mexico Development Division, World Bank, Washington, D.C. and Geneva. Available at www/proventionconsortium.org/ Haacker, Marcus. 2002. "The Economic Consequences of toolkit.htm. HIV/AIDS in Southern Africa." IMF Working Paper 02/30. Ernst & Young. 1991. "Booming and Busting." In Economic Fore- International Monetary Fund, Washington, D.C. cast. Asia Pacific Economics Group, Canberra. Handmer, J., and P. Thompson. 1997. "Economic Assessment Ferroni, Marco, and Ashoka Mody, eds. 2002. International Public of Disaster Mitigation: A Summary Guide." Resource and Envi- Goods: Incentives, Measurement, and Financing. London: Kluwer ronmental Studies 13. Centre for Resource and Environmen- Academic Publishers/World Bank. tal Studies, Australian National University, Canberra. Fischer, Gunther, Mahendra Shah, and Harrij Van Velthuizen. Harriss, John. 2002. "The Case for Cross-Disciplinary Approaches 2002. "Climate Change and Agricultural Vulnerability." in International Development." World Development 30 (3): International Institute for Applied Systems Analysis, Laxenburg, 487­96. Austria. Hicks, Deborah. 1993. "An Evaluation of the Zimbabwe Drought Fischer, Stanley, and William Easterly. 1990. "The Economics Relief Programme 1992/1993: The Roles of Household Level of the Government Budget Constraint." World Bank Research Response and Decentralized Decision Making." World Food Observer 5 (2): 127­42. Programme, Harare. Foster, M., and Adrian Fozzard. 2000. "Aid and Public Expen- Hossain, Mahabub, Manik Lal Bose, and Alamgir Chowdhury. diture: A Guide." ODI Working Paper 141. Centre for Aid and 2001. "Changes in Agriculture and Economy in the Flood- Public Expenditure, Overseas Development Institute, London. Prone Environment in Bangladesh, 1988­2000: Insights from Fozzard, Adrian, M. Holmes, J. Klugman, and K. Withers. 2001. a Repeat Survey of 16 Villages." Joint IRRI-BRRI Workshop "Public Spending for Poverty Reduction." Draft. Poverty Reduc- on Flood-Prone Environments. International Rice Research tion Strategy Program, World Bank, Washington, D.C. Avail- Institute, Los Baños, Philippines. able at www.worldbank.org/poverty/strategies/sourctoc.htm. Hulme, Mike, and others. 2001. "African Climate Change: Freeman, Paul K., Leslie A. Martin, Reinhard Mechler, and 1900­2100." Climate Research 17: 145­68. Koko Warner, with Peter Hausmann. 2002. "Catastrophes and IADB (Inter-American Development Bank). 2001. "Preparatory Development: Integrating Natural Catastrophes into Devel- Group Meeting on Natural Disasters: Disaster Prevention opment Planning." Disaster Risk Management Working Paper and Risk Reduction." Working Paper. Working Group on Series 4. World Bank, Washington, D.C. Available at Financing, Inter-American Committee on Natural Disaster http://www.proventionconsortium.org/files/catastrophes_com- Reduction, Washington, D.C. plete.pdf. IDC (International Development Committee). 2003. "The Human- Freeman, Paul K., Leslie A. Martin, Joanne Linneroot-Bayer, itarian Crisis in Southern Africa. House of Commons, Ses- and Koko Warner. 2003. "National Systems for the Compre- sion 2002­3, 3rd Report, together with the Proceedings of the hensive Management of Disaster Risk: Financial Strategies International Development Committee, Evidence and Appen- for Natural Disaster Reconstruction." Presented at the 3d meet- dices." March 4. London. ing of the Natural Disasters Management Network, March IFRC (International Federation of the Red Cross and Red Cres- 6­7, Washington, D.C. Inter-American Development Bank, cent Societies). 1999. World Disasters Report 1999. Geneva. Washington, D.C. Available at http://www.iadb.org/int/DRP/ing/ ------. 2002. World Disasters Report 2002. Geneva. Red6/Docs/FreemanSistNacion03-03eng.pdf. IPCC (Intergovernmental Panel on Climate Change). 2001. Cli- Froot, K. A. 1999. "Introduction." In K. A. Froot, ed., The Financ- mate Change 2001: Synthesis Report. Cambridge, U.K.: Cam- ing of Catastrophe Risk. Chicago: University of Chicago Press. bridge University Press. Gallup, J. L., and J. D. Sachs. 1999. "Geography and Economic Islam, Nurul. 1997. Development Planning in Bangladesh: A Development." Consulting Assistance on Economic Reform Study in Political Economy. London: Hurst. II Discussion Paper 39. Harvard Institute for International Jovel, J. R. 1989. "Natural Disasters and Their Economic and Development, Cambridge, Mass. Social Impact." CEPAL Review 38 (August): 133­45. Gilbert, Roy, and Alcira Kreimer. 1999. "Learning from the World Kanbur,Ravi,ToddSandler,andKevinMorrison.1999."TheFuture Bank's Experience of Natural Disaster Related Assistance." of Development Assistance: Common Pools and International 110 References Public Goods." Policy Essay 25. Overseas Development Coun- Governance: Budget Responses 2000," June 17­18, Dhaka. cil, Washington, D.C. Centre for Policy Dialogue, Dhaka. Kaul, Inge, Isabelle Grunberg, and Marc A. Stern, eds. 1999. Royal Society. 1992. Risk: Analysis, Perception and Management. Global Public Goods: International Cooperation in the 21st Cen- London. tury. Oxford and New York: Oxford University Press. Sandler, Todd. 2002. "Financing International Public Goods." Kelly, Charles, and M. H. Khan Chowdhury. 2002. "Poverty, In Marco Ferroni and Ashoka Mody, eds., International Public Disasters and the Environment in Bangladesh: A Quantita- Goods: Incentives, Measurement, and Financing. London: Kluwer tive and Qualitative Assessment of Causal Linkages." Academic Publishers/World Bank. Bangladesh Issues Paper. U.K. Department for International Sobhan, Rehman, ed. 1998. Crisis in Governance: A Review of Development, Dhaka. Bangladesh's Development 1997. Dhaka: Centre for Policy Dia- Khan, A. R., and Mahabub Hossain. 1989. The Strategy of Devel- logue and University Press. opment in Bangladesh. Basingstoke and London: Macmillan. Solow, R. M. 1956. "A Contribution to the Theory of Economic Klugman, Jeni. 2002. "Overview." In, A Sourcebook for Poverty Growth." Quarterly Journal of Economics 70: 65­94. Reduction Strategies. Vol. 1: Core Techniques and Cross-Cutting Spencer, Hilary, Julia Slingo, and Mike Davey. Forthcoming. "Sea- Issues. Washington, D.C.: World Bank. sonal Predictability of ENSO Teleconnections: The Role of Kreimer, Alcira, Margaret Arnold, and Anne Carlin, eds. 2003. the Remote Ocean Response." Climate Dynamics. Building Safer Cities: The Future of Disaster Risk. Disaster Risk SRU (Seismic Research Unit). 2000. "Volcano Hazard Report Management Series 3. Washington, D.C.: World Bank. for Southern Dominica: Interpretation of 1998­2000 Earth- MacKellar, L., Paul Freeman, and T. Ermolieva. 1999. "Estimating quakes and Hazard Mapping Results." Seismic Research Natural Catastrophic Risk Exposure and the Benefits of Risk Unit, University of the West Indies, Trinidad. Transfer in Developing Countries." International Institute for Stewart, Frances, Valpy FitzGerald, and associates. 2001. War Applied Systems Analysis, Laxenburg, Austria. and Underdevelopment. Vol. 1: The Economic and Social Conse- Malawi. 2001. "Malawi Poverty Reduction Strategy Paper." Draft. quences of Conflict. Oxford: Oxford University Press. Lilongwe. Stockdale, T. N., D. L. T. Anderson, J. O. S. Alves, and M. A. Munich Re. 2002. "Topics Annual Review: Natural Catastrophes Balmaseda. 1998. "Global Seasonal Rainfall Forecasts Using 2002." Munich. Available at www.munichre.com/pdf/topics. a Coupled Ocean-Atmospheric Model." Nature 392 (March OAS (Organization of American States). 1996. "Insurance, Rein- 26): 370­73. surance and Catastrophe Protection in the Caribbean." Swiss Reinsurance Company. 1999. "Alternative Risk Transfer Working paper prepared in collaboration with the World Bank (ART) for Corporations: A Passing Fashion or Risk Manage- by A. F. Evans and others. Caribbean Disaster Mitigation ment for the 21st Century?" Zurich. Project, OAS, Washington, D.C. Te Velde, Dirk Willem, O. Morrissey, and A. Hewitt. 2002. "Allo- Pantelic, J., and others. 2000. "Promoting Community-Based cating Aid to International Public Goods." In Marco Ferroni Disaster Mitigation in Bangladesh." Draft. Disaster Manage- and Ashoka Mody, eds., International Public Goods: Incentives, ment Facility, World Bank, Washington, D.C. Measurement, and Financing. London: Kluwer Academic Pub- Philippine NLUC (National Land Use Committee). 1992. "National lishers/World Bank. Physical Framework Plan, 1993­2022." NLUC, National Eco- Thomson, Anne. 1994. "The Impact of Drought on Government nomic and Development Authority, Manila. Expenditure in Namibia in 1992/93." London: Overseas Devel- Quah, D. 1993. "Empirical Cross-Section Dynamics in Economic opment Institute. Growth." European Economic Review 37: 426­34. Thomson, Anne, Penny Jenden, and Edward J. Clay. 1998. "Infor- Rahman, Atiur, and others. 2000. The Budget and the Poor. mation, Risk and Disaster Preparedness: Responses to the 1997 Study commissioned by the Institute for Development Policy El Niño Event." Department for International Development, Analysis and Advocacy, Proshika, and conducted by Shamunnay. ESCOR Research Report AG1215. SOS Sahel, London. Dhaka: University Press. UNDP (United Nations Development Programme). 1999. Dis- Rahman, M. 2000. "Performance of the External Sector in FY2000: aster Reduction and Recovery Programme: Progress Report and Growth Dynamics and Emergent Concerns." Paper pre- Plan of Action. Geneva: Disaster Reduction and Recovery Pro- sented at two-day dialogue on "State of the Economy and gramme, UNDP. References 111 UNDP (United Nations Development Programme) and to the People's Republic of Bangladesh for an Emergency Flood Bangladesh. 1998. "Flood Rehabilitation Assessment." Dhaka. Recovery Project." Report P-7264-BD. Washington, D.C. UNISDR (United Nations International Strategy for Disaster Reduc- ------. 1998b. "OECS Financial Sector Review." Report 17554- tion). 2002. "Natural Disasters and Sustainable Development: LAC. Finance, Private Sector and Infrastructure Unit, Caribbean Understanding the Links between Development, Environment Country Management Unit, Latin America and the Caribbean and Natural Disasters." Background Paper 5. Geneva. Region, Washington, D.C. Vermeiren, Jan, Steven Stichter, and Alwyn T. Wason.1998. "Costs ------.2000. "Managing Catastrophic Risks Using Alternative and Benefits of Hazard Mitigation for Building and Infra- Risk Financing and Insurance Pooling Mechanisms." Discus- structure Development: A Case Study in Small Island Devel- sion draft. Finance, Private Sector and Infrastructure Depart- oping States." Caribbean Disaster Mitigation Project, ment, Caribbean Country Management Unit, Latin America Organization of American States, Washington, D.C. Available and Caribbean Region, Washington, D.C. at http://www.oas.org/en/cdmp/document/papers/tiems.htm. ------. 2001. Global Development Finance 2001: Building Coali- World Bank. 1989. "Bangladesh PER: Public Resource Manage- tions for Effective Development Finance. Washington, D.C. ment during the Fourth Five-Year Plan, FY91­95." Report ------. 2002. "An Evaluation of the Applicability of Risk Trans- 7545-BD. Asia Country Department I, Washington, D.C. fer and Funding Mechanisms for Natural Disaster Losses: Con- ------.1996. "Bangladesh Public Expenditure Review." Report cept Paper." South Asia Region, Washington, D.C. 15905-BD. Country Operations Division, Country Depart- Yunus, Md. 1998. Banker to the Poor: The Autobiography of ment 1, South Asia Department, Washington, D.C. Muhammad Yunus, Founder of the Grameen Bank. Dhaka: Uni- ------. 1998a. "Memorandum and Recommendation of the versity Press. President of the International Development Association to the Zebiak, Stephen. 1999. "El Niño and the Science of Climate Executive Directors on a Proposed Credit of SDR 146.0 Million Prediction: A Review Assessment." Consequences 5 (2): 3­15. Index Boxes, figures, and notes are indicated, respectively, by b, f, and n following the page number. accounting. See reporting and accounting of government irrigation to reduce effects, 28, 102n37 finance food policy reforms, 28, 84­85 adaptation, 6, 17, 28 garment manufacturing, 13, 100n18 Africa. See Malawi and southern Africa; Sub-Saharan Africa; information dissemination as damage reduction method, 43 specific countries flood warnings, 43, 47, 104n65 agricultural sector risk zone maps, 104n66 Bangladesh, 11­12, 34 infrastructure damage and reconstruction, 12, 38­39 Dominica, 9­10, 26, 70 insurance in insurance and, 54, 55 parametric weather insurance and defining of trigger Malawi and southern Africa, 14, 97 event, 56 See also cereal production in Malawi and southern Africa promoting risk reduction measures, 58 aid flows. See external aid land use controls, 87­88 AIDS. See HIV/AIDS microcredit programs and microfinance institutions, 13, 42, Albala-Bertrand, J. M., 22 55, 87 Antigua and hurricane insurance, 105n73 poorer regions more hazard-prone, 25 Argentina reconstruction financing, sources of, 24b2.3 postdisaster reconstruction, 25, 88 private expenditure on, 25 balance of payments and inflation, 19 public finance and impact of disasters on, 30, 82­84, 99n2, bananas 101n29 as chief crop of Dominica, 9, 18, 70, 73, 75 deficit financing, 102n36 quick recovery after disaster, 10, 99n9 expenditure and reallocation, 32, 83, 89, 101n31, 102nn40 WINCROP banana crop insurance, 10, 55, 56b5.1, 58 & 43 Bangladesh, 79­90 local governments, 86 agricultural sector, 11­12, 34 major floods and, 84, 89 building standards, 13, 48, 87­88 policy context for, 33­34 cyclones, 11, 46­47, 80 reallocation as appropriate response, 36­37 shelters, 46, 87 reporting and accounting practices, 34 damage assessment after flood, 37, 101n33 state-owned enterprises, 86 deficit of government, 84 taxation and revenue, 33, 83­84 disaster reduction strategy and, 41, 43, 87­89, 103n48 research implications for, 89­90 earthquakes, 47­48, 80 risk transfer tools and, 55 economy of, 11­13 selection as country for study, 4 fluctuations as affected by natural hazards, 12, 12f2.2, 23, soil erosion, 79­80 81­82 tornadoes, 80­81 postdisaster forecasting, 85bB.1 trade embargo violation by, effect of, 35 external aid for, 34­35, 86, 89 vulnerability to natural hazards, 11­13, 15 financial institutions, 86­87 economic growth and development's effect on, 18 financial system, 13 public finance and, 42 floods and droughts, 11­12, 17, 79­80 ranking, 20b2.1 assessment of costs of, 37, 99n4 War of Independence in, 11 flash flooding, 80 Bangladesh Bank, 55 information dissemination as damage reduction method, Bayes' rule applied to 1998 volcano situation in Dominica, 49, 43, 47 49b4.1, 51 113 114 Index Botswana risk management's priority, 60 economic reaction to natural disasters, 23, 100n15 Dhaka (capitol of Bangladesh), 12, 13 effects of drought on, 18 disaster management, 16 Brazil See also risk management poorer regions more hazard-prone, 25 Bangladesh, 103n48 regional analysis as appropriate for, 27 Dominica, 76 budgetary consequences of disasters. See public finance disaster preparedness building standards See also risk management Bangladesh, 13, 48, 87­88 Bangladesh, 87 Dominica, 11, 72­73, 76 costs, 29, 30, 37, 38 Dominica, 75 Caribbean Community and Common Market (CARICOM) for localized disasters, 41, 65 agreement, 100n14 domestic policies and impact of natural hazards, 19 Caribbean Hotel Association, 58 Dominica, 69­77 cash accounting of developing countries, 36­37 agricultural sector, 9­10, 26, 71 catastrophe bonds, 53 bananas. See bananas catastrophe insurance. See insurance building standards, 11, 72­73, 76 Central America and hazard damage, 17 catastrophe insurance required for mortgages, 53, 58 See also specific countries and types of hazards climate change, impacts of, 70 Centre for Research on the Epidemiology of Disasters (CRED), construction industry, 72 99n4 currency situation of, 11 cereal production in Malawi and southern Africa, 14, 15f2.4, damage assessment after disaster, 37 92, 97, 104n61 disaster reduction strategy and, 41, 76 climate change, 16, 51 earthquake risk, 69 disaster planning and, 41 economy of, 9­11, 70­72 effect not apparent in hazard data, 61 fluctuations as affected by natural hazards, 10, 10f2.1, effect on Dominica, 70 22, 70 climate-related hazards risk reduction and diversification of, 10­11, 18, 41, See also hydro-meteorological hazards 70­71, 100n23 compared with geophysical hazards, 17, 61 emigration as result of disaster, 26, 100n23 multi-sector effects of, 18 environmental impacts of natural hazards in, 69­70 climatic forecasting, 44, 64­65 external accounts of, 73 See also information dissemination on natural hazards external aid for, 34­35, 75 in Bangladesh, 46 financial institutions, 74 in Southern Africa, 16, 44­46, 51, 62, 95­97, 97­98 hurricanes and tropical storms, 9, 17, 47, 69 climatic variability, 16, 61, 93­95, 97 coastal protection against waves, 43, 47 Cochrane, Harold C., 24, 25 impacts of natural hazards on, 22­23, 69­70 coconut-based soap products from Dominica, 100n14 inflation, 74 conflicts and vulnerability levels, 19, 23 information dissemination improvements needed, 77 See also specific countries for wars and independence struggle infrastructure, damage and evolution of, 11, 72­73, 99n10 construction. See building standards insurance in, 53, 57, 58, 74­75 copra from Dominica, 99n9, 100n14 investment and domestic consumption, 73 corruption, elimination of, 66 landslides, 9, 16, 69 cyclones, 17 manufacturing sector, 9, 11, 71­72 See also tropical storms policy implications of study on, 76­77 Bangladesh, 11, 16, 46­47, 80 postdisaster reconstruction projects, 101n34, 103n47 shelters, 46, 87 private expenditure on, 25 warning system, 43, 46 poverty, 75­76 Fiji insurance for, 105n74 reduction strategy for, 21b2.2 Philippines, 23 public finance and impact of disasters, 30, 75 expenditure and reallocation, 32, 103n50 damage assessments, requirements for, 37, 101n33 long-term policy consequences for, 39 definitions, 5­6 taxation and revenue, 33 deforestration in Dominica, 16 selection as country for study, 4 derivatives, 55 soil erosion, 70 weather derivatives, 53 tourism industry, 9, 11, 69, 70, 72 developed vs. less developed countries volcanic activity, 9, 69 catastrophe insurance and, 54 Bayes' rule applied to 1998 situation, 49, 49b4.1, 51 likelihood of disasters and, 25 monitoring and alerts, 48­49 Index 115 vulnerability to natural hazards, 9­11, 15, 17 See also public finance economic growth and development, effect on, 18, 76 Bangladesh, 13 ranking, 20b2.1 Dominica, 11, 74 droughts. See floods and droughts Zimbabwe, 18 findings of study, 61­62 earthquakes, 17 fire insurance, 54 Bangladesh, 80 floods and droughts hazard assessment for, 47­48 Bangladesh, 11­12, 17, 79­80 Dominica's risk of, 69 damage assessment after flood, 37, 99n4 Turkey and contingent credit line for compulsory earthquake flash flooding, 80 insurance, 105n72 information dissemination as damage reduction method, Eastern Caribbean Central Bank (ECCB), 11, 57, 74, 75 43, 47 Economic Commission for Latin America and the Caribbean irrigation to reduce effects, 28, 102n37 (ECLAC), 37 climatic variability and, 16 economic losses Dominica, 69 See also macroeconomic impact of disasters; specific countries government finance and, 30­31, 63 increases between 1950s and 1990s, 3 Malawi, 14, 15, 17, 30, 92­93, 95 later reassessment of, 63 Philippines, 23 stage of development and, 18­19, 61 riverine floods, 17, 47 structure of economy as factor, 17­18, 61 warning system problems, 65 ecotourism. See tourism in Dominica Sub-Saharan Africa Ecuador's poorer regions more hazard-prone, 25 climatic forecasting in, 44­46, 51, 62, 95­97 El Niño events, 15, 44, 92­93, 95, 97, 99n3 fiscal impact of, 31b3.1, 62 El Niño-Southern Oscillation (ENSO), 44, 45, 92­93, 95, 99n12 water management and, 16 emigration as result of disaster, 26, 31, 100n23 Zimbabwe, 18, 30, 101n26 empirical analysis of impacts of disasters, 25 food aid, 35, 85, 89 further research needs, 26 food shortage. See famine environmental change, 16, 61 forecasting. See climatic forecasting disaster planning and, 41 foreign aid. See external aid due to natural hazards, 69­70 foreign exchange reserves, 29 expenditure and reallocation, 31­33, 62 future disasters, financing costs of, 53­67 Bangladesh, 32, 83, 89, 101n31, 102nn40 & 43 See also insurance; risk transfer tools improved systems needed for, 39­40, 64 creative solutions for, 57­58 policy implications for, 64 demand as issue, 55, 105n70 export markets, 19, 73 future research needs, 67 See also specific crops (e.g., bananas) potential obstacles to, 54­57, 62 external aid, 14, 23, 34­35, 62, 65 promoting risk reduction, 58­59 for Bangladesh, 86, 88, 89 future research needs, 66­67 for Dominica, 75 on Bangladesh, 89­90 expected aid and demand for risk transfer tools, 55 for southern Africa, 97 garment manufacturing in Bangladesh, 13, 100n18 geophysical hazards, 17 famine See also specific type (e.g., earthquake, volcano, etc.) Bangladesh, 11, 13, 28 compared with climate-related hazards, 17, 61 Malawi, 14, 35, 43, 91­92 defined, 6 Fiji economic vulnerability associated with, 27, 61 cyclone insurance, 105n74 information dissemination about, 47­49, 50, 62, 65 impacts of natural hazards on, 22­23, 100n17 global climate change. See climate change private expenditure on reconstruction, 25 globalization and risk management, 28, 62 public finance and impact of disaster, 30 governance expenditure and reallocation, 32 aspects of good governance, 65­66 taxation and revenue, 33 public finance and, 41­42, 66 financial institutions government borrowing. See public finance Bangladesh, 86­87 government risk premiums elevated due to disasters, 29 Dominica, 74 financial services industry hail insurance, 54 Dominica and offshore services, 9, 11, 70 health hazards. See specific illnesses (e.g., HIV/AIDS) promoting risk management, 63 hedge instruments, 53 financial system HIV/AIDS, 14­15, 19, 20b2.1, 98 116 Index Honduras infrastructure and impact of disaster, 11, 12, 38­39, 72­73, insurance for flood and storm risk, 55 99n10 reconstruction financing, sources of, 24b2.3 insurance, 53­54 human rights abuses and external aid flow, 35 affordability of, 54, 59 Hurricane Allen (1980), 9, 10 Bangladesh, 88 Hurricane Andrew (1992), 54 demand for, 55 Hurricane David (1979) determination of parametric insurance triggers, 55­57 assistance following, 35 Dominica, 74­75 banana production and, 10 effectiveness in promoting risk reduction, 58, 62, 65, 74 damage caused by, 9, 11, 69, 70 Fiji, 105n74 economic impacts of, 71, 72, 73 future research needs, 67 emigration following, 100n23 microcredit programs and, 55, 59, 88 localized hazards associated with, 17, 19 Montserrat, withdrawal from, 59, 105n75 reforms in Dominica following, 39, 70, 71 recommendations for, 65 Hurricane Frederick (1979), 9, 10 risk transfer tools and, 53­54, 59 Hurricane Hugo (1989), 9, 10, 23, 31, 70 self-insurance with high deductibles, 57 Hurricane Jose (1999), 105n73 structure of industry, 57 Hurricane Lenny (1999) international aid. See external aid assessment of costs of, 37, 99n4 International Development Association assistance to low- damage caused by, 9, 11, 43, 73, 103n46 income countries, 104n60 poverty reduction strategies following, 21b2.2 International Institute for Applied Systems Analysis (IIASA), unpredicted, 47, 69 24, 24b2.3, 25 WINCROP claims arising from, 56b5.1 International Monetary Fund (IMF) support to Hurricane Mitch (1998), 17 Dominica, 39 hurricanes, 17 international public goods (IPGs), 44, 104n58 See also specific hurricane or country International Research Institute for Climate coastal protection against waves, 43, 47 Prediction (IRI), 44 hydro-meteorological hazards, 16­17, 99n12 irrigation controls. See floods and droughts See also specific type (e.g., drought, hurricane, etc.) island economies and vulnerability, 19, 20b2.1, 77 defined, 6, 99n6 See also specific island nations (e.g., Dominica, Fiji, etc.) economic vulnerability associated with, 27, 61 short-term impacts and, 22­23 information dissemination on, 44 lagged effects of disasters, 30, 32­33 IIASA. See International Institute for Applied Systems landslides in Dominica, 9, 16, 69 Analysis La Niño events, 97 indebtedness of country and impact of natural disasters, 24 large countries. See size of country index of vulnerability, 20b2.1 less developed countries. See developed vs. less developed India's failure to warn Bangladesh of impending flood, 47, countries 104n65 lessons learned Indonesia and regional analysis, 27, 100n20 macroeconomic impact of disasters and, 26­28 inflation in Dominica, 74 policy changes based on, 63 information dissemination on natural hazards, 43­51 public finance impact of disasters and, 39­42 See also climatic forecasting; specific countries local governments in Bangladesh, 86 Bayes' rule applied to 1998 volcano situation in Dominica, localized disasters 49, 49b4.1, 51 See also specific types (e.g., landslides and floods) country-specific, 44, 45, 62 analysis on regional level appropriate for, 27, 62, 63, 88 failure in provision of, 47­49, 77 forecasting on regional level, 46, 47, 50, 65 findings and conclusions about, 49­51 poorer regions more hazard-prone, 25 future research needs, 67 predisaster allocation of funds for, 41 geophysical hazards and, 47­49, 50, 62 reserves established for, 57 international coordination of, 44, 50, 62, 65, 77 long-term impacts of disasters, 23­26 transnational river systems and, 47, 65 distinguishing from shorter-term impacts, 27, 63 public action and, 43 empirical analysis of, 25 public funding for monitoring and dissemination, 50 further research needs, 26 as a public good, 43­44, 49­50, 62, 64­65, 104n58 modeling of, 24­25 reallocation of public funds requiring current, reliable public finance and, 32, 39 information, 36­37, 62 qualitative analysis of, 25­26 setting public policy, need for, 64 further research needs, 26 study findings on, 62 study findings on, 61 Index 117 macroeconomic impact of disasters, 9­28 volcanic activity on, 17, 31, 100n25 Bangladesh, 81­82, 89 withdrawal of insurance industry from, 59, 105n75 complementary approaches to impact assessment, 26­27 moral hazard concerns about external aid, 34, 55 Dominica, 70 Mozambique findings of study on, 61 El Niño events and, 45 future research needs on, 66 flooding, 95 lessons learned on methodological issues, 26­28 failure to share information downstream, 47 long-term impacts, 23­26 Multi-Fibre Agreement (MFA), 13 national development policy and, 63 opportunities to reduce, 27­28 Namibia reallocation requirements for information on, 37 economic reaction to natural disasters, 23, 100n15 short-term impacts, 22­23 effects of drought on, 18 structure of economy and stage of development, 17­20 national disaster impacts theoretical perspectives, 21­22 information systems and needed improvements, 65 vulnerability and, 9­17 later reassessment of, 63 See also vulnerability natural disasters, defined, 5 maize, 56, 92, 93­94, 100n17, 104n61 natural hazards, 16­17 Malawi and southern Africa, 91­98 See also specific type or country cereal production, 14, 15f2.4, 92, 93, 104n61 defined, 5 climate forecasts, value of, 45­46, 51, 95­97 socioeconomic conditions and, 19 climatic variability and, 93­95 Nicaragua reconstruction financing, sources of, 24b2.3 drought, 16, 17, 30­31, 92­93, 95 Nigeria and regional analysis, 27 economy of, 13­14 nonexcludable benefits of a good, 43­44 fluctuations as affected by natural hazards, 13f2.3, 23, nonrival benefits of a good, 43­44 93­94 El Niño events and, 45 objectives of study, 3­4 external aid for, 14, 34­35 operations and maintenance (O&M). See infrastructure and famine, 14, 35, 43, 91­92 impact of disaster floods, 17, 93 Organisation of Eastern Caribbean States (OECS), 57, 74, 75 government finance and impact of disasters, 30­31 Organization of American States HIV/AIDS in, 14­15 study on infrastructure damage from disaster, 38 information and public action needs, 98 support for scientific research on disaster management, 47, parametric weather insurance and defining of trigger 76 event, 56 political instability, 14 parametric insurance, 55­57 poverty reduction strategy for, 21b2.2 Peru's poorer regions more hazard-prone, 25 public finance and impact of disasters Philippines expenditure and reallocation, 32 economic reaction to natural disasters, 23, 27 governance and, 42 private expenditure on reconstruction, 25 planning for climatic variability and, 41 public finance and impact of disaster, 30 public sector enterprises affected by disasters, 102n39 expenditure and reallocation, 32, 103n51 selection as country for study, 4­5 external borrowing, 103nn52 & 53 study recommendations for, 97­98 transport system improvements reduced due to disaster subsidized livelihood programs and safety nets for rural response, 26, 39 population, 59 planning vulnerability to natural hazards, 13­15 of disaster preparedness, 65 ranking, 20b2.1 of postdisaster reconstruction efforts, 28, 41, 63 manufacturing sector policy context for public finance decisions, 33­34 Bangladesh, 13, 100n18 policy framework for reallocation of funds, 36 Dominica, 9, 11, 71­72 policy implications of study, 63­66 meteorological hazards. See hydro-meteorological hazards postdisaster reconstruction efforts, 22, 24­25 method of investigation, 6­7 See also specific countries MFA (Multi-Fibre Agreement), 13 financial resource gaps, effect of, 24, 24b2.3 microcredit programs, 55, 59, 65 planning for, 28, 41, 63 in Bangladesh, 13, 42, 87, 88 public finance and, 29, 32­33, 40­41, 55, 101n34 Montserrat technical advances as result of, 22 economic impacts of natural hazards on, 22­23 poverty reduction emigration as result of disaster, 26 climate forecasts, value for, 45 government finance and impact of disasters, 31, 101n32 strategy paper (PRSP), 21b2.2 118 Index vulnerability to natural hazards and, 19, 21b2.2, 63 rice Dominica, 75­76 in Bangladesh, 12, 16 preferential trade agreements for bananas, 10, 11 flood-tolerant cultivars for, 16, 100n22 prioritization of investment projects for reallocation of funds, risk management 36, 102n41 assessment of hazard risk for, 63 private sector Bangladesh expenditure on reconstruction by, 24b2.3, 25 cyclone warning system, 43, 46 insurance disaster mitigation and preparedness, 87­89 demand in developing countries, 55 developing countries reserves established to cover, 57 ignoring priority of, 60 risk management and, 28, 63 planning over longer time period, 65 ProVention Consortium, 37, 102n42 Dominica's need for, 70 public finance, 29­42 economic decisions and, 17, 19 See also specific countries insurance's effect on promoting, 58, 62 borrowing necessary to meet disaster costs, 29 integrated into broad policies and strategies, 63, 66, broad impact of disasters for, 30­34 76­77 disaggregated examination of implications, 29, 64 private sector involvement in, 28, 63 disaster contingency reserves, 41, 64, 104n56 public finance and, 38­39 expenditure and reallocation, 31­33 spreading of risk, 58, 65, 89 policy implications for, 64 strategies to suit types of disasters, 27­28, 41 external aid. See external aid risk transfer tools, 53­54, 59, 62, 64 future research needs, 66­67 See also insurance governance and, 41­42, 66 Bangladesh, 88 for hazard monitoring and dissemination, 50 Dominica, 74­75 improved systems needed for, 39­40, 62, 64 future research needs, 67 implications of disasters for, 29­30, 62 recommendations for, 65 long-term policy consequences for, 39 rival benefits of a good, 43­44, 104n62 policy context for, 33­34, 63­64 riverine floods. See floods and droughts reallocation, whether appropriate course of action, 35­38 reporting and accounting practices for, 34, 102n38 SARCOF. See Southern African Regional Climate Outlook improved systems needed for, 64 Forum reallocation requirements, 36­37 secondary effects of disasters, need to study, 3 revenue and taxation, 33, 83­84 Seismic Research Unit (Trinidad), 48, 104n67 risk reduction activities and, 38­39 selection of countries for study, 4, 27 study findings on, 62 constraints on, 6­7 subsidized livelihood programs and safety nets for rural self-insurance with high deductibles, 57 population, 59 shelters from cyclones, 46 public sector enterprises affected by disasters, 34, 57, 73, 102n39 short-term impacts of disasters, 22­23, 61 See also infrastructure and impact of disaster distinguishing from longer-term impacts, 27, 63 size of country qualitative analysis of disaster impacts, 25­26 economic reaction to natural disasters and, 22­23, 27 further research needs, 26 regional approach to analysis. See localized disasters small islands. See island economies and vulnerability rainfall patterns, 45, 92­93, 94­95 slow-onset events, defined, 6 See also climatic variability small islands. See island economies and vulnerability rapid-onset events, defined, 6 social sectors and impact of public fund reallocation after reallocation. See expenditure and reallocation disaster, 36 reconstruction. See postdisaster reconstruction efforts socioeconomic conditions and impact of natural hazards, 19, recurrent disasters, 3 55, 100n16 regional disasters. See localized disasters soil erosion, 16 reinsurance, 54, 58, 59, 88 Bangladesh, 79­80 reporting and accounting of government finance, 34, 102n38 Dominica, 70 improved systems needed for, 39­40, 64 South Africa and cereal production, 92, 94 reallocation requirements, 36­37 South America. See specific countries risk reduction activities not reported, 38 Southern African Regional Climate Outlook Forum (SARCOF), reserves for disaster contingency financing, 41, 64, 104n56 45, 95­96 resilience, 61 Sub-Saharan Africa of Bangladesh, 12 See also Malawi and southern Africa; other specific countries defined, 6 (e.g., Botswana, Zimbabwe, etc.) Index 119 climatic forecasting in, 44­46, 51, 62, 95­97 comparison of case study countries and, 15 external borrowing due to drought, 103n52 defined, 5­6, 99n5 fiscal impact of drought in, 31b3.1 dynamic nature of, 9­15, 17 special support for, 64 measuring of, 20b2.1 overview of factors determining, 15­17, 61 taxation and government revenue in Bangladesh, 33, 83­84 poverty reduction and, 19, 21b2.2, 63 technical advances as result of disaster, 22 terminology, 5­6 war. See conflicts and vulnerability levels; specific country tornadoes in Bangladesh, 80­81 warning systems tourism in Dominica, 9, 11, 69, 70, 72 for cyclones and tropical storms, 43, 44, 46, 47, 69 trade for floods, 43, 47, 65, 104n65 See also export markets wave protection from hurricanes and tropical storms, 43, 47 embargo violation by Bangladesh, 35 weather derivatives, 53 external accounts of Dominica and, 73 WINCROP (Winward Islands Crop Insurance) banana crop preferential agreements for bananas, 10, 11 insurance, 10, 55, 56b5.1, 58 transnational rivers and failure to share information about, World Bank 47, 65 assistance to low-income countries, 104n60 tropical storms, 46­49 climatic forecasting in Malawi and, 45­46 See also cyclones; hurricanes on consumption volatility in Caribbean region, 73 Philippines, 23 on deficit financing in Bangladesh, 102n36 warning systems for, 44, 62, 69 parametric weather insurance investigations by, 55­56 Turkey and contingent credit line for compulsory earthquake risk transfer mechanisms encouraged by, 57, 105n71 insurance, 105n72 study of insurance industry, 57 World Meteorological Organization (WMO), 44 United Nations Development Programme (UNDP), Zambia and effects of drought, 18 development of vulnerability indicator, 20b2.1 Zimbabwe United Nations Millennium Development Goals, cereal production, 94, 104n61 20b2.2 drought goats sold to survive in, 101n26 volcanic activity, 17 government borrowing related to, 30 Bangladesh, 11 small savings system used to mitigate effects of, 18 Dominica, 9, 69 economic reaction to natural disasters, 23 monitoring and alerts, 48­49 public finance and impact of disasters for, 103n54 international monitoring needed for, 65 debt reduction policy of, 103n52 Montserrat, 17, 31, 100n25 expenditure and reallocation, 32 vulnerability taxation and revenue, 33 Other Disaster Risk Management Series Titles 1 Managing Disaster Risk in Mexico: Market Incentives for Mitigation Investment 2 Managing Disaster Risk in Emerging Economies 3 Building Safer Cities: The Future of Disaster Risk THE WORLD BANK TMxHSKIMBy356852zv,:&:':):; 1818 H Street, N.W. Washington, D.C. 20433 U.S.A. Telephone: 202-473-1000 Internet: www.worldbank.org E-mail: feedback@worldbank.org ISBN 0-8213-5685-2