Agricultura Production Risk lProd Innovations ... - World Bank

Managing AgriculturalProduction Risk Innovations in Developing Countries

Agriculture and Rural Development Department

REPORT NO. 32727-GLB

THE WORLD BANK

Agriculture & Rural Development DepartmentWorld Bank1818 H Street, N.W.Washington, D.C. 20433http://www.worldbank.org/rural

Managing

AgriculturalProduction

Risk

32727

Pub

lic D

iscl

osur

e A

utho

rized

Pub

lic D

iscl

osur

e A

utho

rized

Pub

lic D

iscl

osur

e A

utho

rized

Pub

lic D

iscl

osur

e A

utho

rized

THE WORLD BANKAGRICULTURE AND RURAL DEVELOPMENT DEPARTMENT

Managing AgriculturalProduction Risk Innovations in Developing Countries

© 2005 The International Bank for Reconstruction and Development / The World Bank1818 H Street, NWWashington, DC 20433Telephone 202-473-1000Internet www.worldbank.org/ruralE-mail [email protected]

All rights reserved.

This volume is a product of the staff of the International Bank for Reconstruction andDevelopment/The World Bank. The findings, interpretations, and conclusions expressedin this paper do not necessarily reflect the views of the Executive Directors of The WorldBank or the governments they represent. The World Bank does not guarantee the accu-racy of the data included in this work. The boundaries, colors, denominations, and otherinformation shown on any map in this work do not imply any judgment on the part ofThe World Bank concerning the legal status of any territory or the endorsement oracceptance of such boundaries.

The material in this publication is copyrighted. Copying and/or transmitting portions or all of this work without permission may be a violation of applicable law. TheInternational Bank for Reconstruction and Development/ The World Bank encouragesdissemination of its work and will normally grant permission to reproduce portions ofthe work promptly.

For permission to photocopy or reprint any part of this work, please send a request withcomplete information to the Copyright Clearance Center, Inc., 222 Rosewood Drive,Danvers, MA 01923, USA, telephone 978-750-8400, fax 978-750-4470, http://www.copyright.com/.

All other queries on rights and licenses, including subsidiary rights, should be addressedto the Office of the Publisher, The World Bank, 1818 H Street NW, Washington, DC20433, USA, fax 202-522-2422, e-mail [email protected].

iii

A C R O N Y M S A N D A B B R E V I AT I O N S vii

P R E FA C E A N D A C K N O W L E D G M E N T S ix

E X E C U T I V E S U M M A RY xi

Introduction 1

Risk and Risk Management in Agriculture 5Informal Mechanisms 6Formal Mechanisms 8

Approaches to Agricultural Risk in Developed Countries 11Crop Insurance Programs in Developed Countries 11Why the Experience of Developed Countries is not a Good Model

for Developing Countries 14

Innovation in Managing Production Risk: Index Insurance 15Index Insurance Alternatives 15Basic Characteristics of an Index 15Relative Advantages and Disadvantages of Index Insurance 17The Trade-off Between Basis Risk and Transaction Costs 17Where Index Insurance Is Inappropriate 17

0 New Approaches to Agricultural Risk Management in Developing Countries 21

Role of Government 21Policy Objectives 23Constraints in Agricultural Risk Management 24Risk Principles: Layering and the Role of Index Insurance 25Addressing the Market Insurance Risk Layer 26Market Failure Layer 29

5

4

3

2

1

Contents

Policy Instruments 30Index Insurance as a Source of Contingent Funding

for Government Disaster Assistance and Safety Net Programs 32

From Theory to Practice: Pilot Projects for Agricultural Risk Transfer in Developing Countries 35

Nicaragua: A Seven-Year Incubation Period 36Morocco 38India: Private Sector Led Alternative Agricultural Risk

Market Development 39Ukraine 41Ethiopia: Ethiopian Insurance Corporation and Donor Led Ex Ante

Disaster Risk Management 43Malawi and SADC: Weather Risk Transfer to Strengthen Livelihoods

and Food Security 47Peru: Government Led Systemic Approach to Agricultural

Risk Management 48Mongolia: World Bank Contingent Credit for Livestock Mortality

Index Insurance 49Global Strategy: The Global Index Insurance Facility (GIIF) 51

Potential Roles for Governments and the World Bank 53Government Roles 53World Bank Roles 54

R E F E R E N C E S 59

Appendix 1: Weather Risk Management for Agriculture 63The Financial Impact of Weather 63The Weather Market 64Weather Risk and Agriculture 65Structuring a Weather Risk Management Solution 67Valuing Weather Risk 74Weather Data 79Further Reading 81References 81

Appendix 2. Case Studies of Agricultural Weather Risk Management 83Indexed-based Insurance for Farmers in Alberta, Canada:

The AFSC Case Study 83Alternative Insurance Through Weather Indices in Mexico:

The Agroasemex Case Study 85Weather Insurance for Farmers in the Developing World:

Case Studies from India and Ukraine 90Technology Application Case Studies: Grassland Index Insurance

Using Satellite Imagery 107References 108

N O T E S 111

7

6

iv Contents

Contents v

Tables2.1 Risk Management Strategies in Agriculture 8

4.1 Advantages and Disadvantages of Index Insurance 18

5.1 Risk Transfer Strategies 28

6.1 Summary of Case Studies 36

6.2 Reasons for Buying Weather Index Insurance in India 43

A2.1 Options for CHU Contracts 85

A2.2 Total Liability Factored into the Agroasemex Business Plan for Autumn-Winter 2001––2002 86

A2.3 Summary of the Methodology to Calculate the Eleven FCDD Indices 87

A2.4 Comparative Analysis Between the Observed Historical Severity Indices (indemnities/total liability) and the Estimated Severity Indices for the Crops and Risks Selected 88

A2.5 Specifications of Call Option Structures Considered by Agroasemex 89

A2.6 Estimated Commercial Premium for Weather Derivative Structures (in US$) 90

A2.7 Weather Insurance Contracts Offered to Groundnut and Castor Farmers 94

A2.8 Pilot Statistics, 2003 95

A2.9 Payout Structure Per Acre for Groundnut Weather Insurance Policy for Narayanpet Mandal, Mahahbubnagar District (2004) 96

A2.10 Payout Structure Per Acre for Castor and Groundnut Excess Rainfall WeatherInsurance Policy for Narayanpet, Mahahbubnagar 97

A2.11 Relationship Between SHR and Winter Wheat Yields During the Vegetative Growth Phase of Plant Development 101

A2.12 Relationship Between SHR and Financial Losses Associated with Winter Wheat Yield Fluctuations 102

A2.13 Correlation Coefficients for the Interannual Variability of Cumulative Rainfall, Average Temperature, and the SHR Index Measured at Five UHC Weather Stations in Kherson 103

Boxes2.1 Asset-Based Risk Management 7

5.1 Reinsurance 22

6.1 India Impact Assessment 42

7.1 Examples of Potential World Bank Investment Lending Projects to Facilitate Risk Management 57

Figures2.1 Independent Versus Correlated Risk 9

3.1 Crop Insurance Premiums and Indemnities in the United States 12

4.1 Payout Structure for a Hypothetical Rainfall Contract 16

5.1 Framework for Governmental Agricultural Risk Management Policy Formulation 23

5.2 Average April to October Rainfall for Thirteen Malawi Weather Stations 26

5.3 Histogram of Simulated SADC Drought Events 29

5.4 Government-Sponsored DOC as Risk Transfer Product Between National and International Risk Markets 30

7.1 Potential Impacts of Natural Hazards 54

A1.1 Notional Value of All Weather Contracts in US$ 65

A1.2 Percentage of Total Weather Contracts by Location (excluding CME trades) 66

A1.3 Potential End User Market by Economic Sector 2003–2004 66

A1.4 Call Option Payout Structure and Wheat Grower’s Losses 72

A1.5 Collar Payout Structure and Agrochemical Company’s Deviation from Budgeted Revenue 73

A1.6 Schematic of Historical Revenues of a Business and the Impact of Weather Hedging 78

A2.1 Relationship Between the Daily Rate of Development of Corn Minimum and Maximum Temperatures 84

A2.2 Comparative Accumulated Distribution Probability Function Based on a “Probability of Exceedence Curve” for the Historical and Modeled Results (payouts in US$) 89

A2.3 Mahahbubnagar District Groundnut Yields Versus Groundnut Rainfall Index 93

A2.4 Payout Structure of Groundnut Weather Insurance Policy Held by Farmers with Small, Medium, and Large Land Holdings 94

A2.5 Payout Structure of Groundnut Weather Insurance Policy for Narayanpet Mandal, Mahahbubnagar District, 2004 97

A2.6 An Example of the Marketing Leaflet for Groundnut (DGN), Castor (DCN), and Excess Rainfall (EN) Protection in Narayanpet Mandal, MahahbubnagarDistrict, 2004 98

A2.7 Winter Wheat Yields for Kherson Oblast, 1971–2001 100

A2.8 Cumulative Rainfall and Average Temperature for Behtery Weather Station for April 15 to June 30, 1973–2002 104

A2.9 SHR Index for Behtery Weather Station, 1973–2002 105

A2.10 Sample Contract for Behtery Weather Station 106

vi Contents

vii

ACP Africa-Caribbean-Pacific

APF Agricultural Policy Framework of Canada

APH actual production history

ARD Agriculture and Rural Development Department of the World Bank Group

BASIX Livelihood Promotion and Microfinance entity of Andhra Pradesh

BIP base insurance product

BSFL Bhartiya Samruddhi Finance Limited (part of BASIX)

CAIS Canadian Agricultural Income Stabilization

CAT catastrophe

COFIDE Corporación Financiera de Desarollo S.A. (Development Finance Corporationlocated in Lima, Peru)

CRDB Cooperative and Rural Development Bank Limited, a private commercial bank

CRMG Commodity Risk Management Group (ARD, The World Bank)

DECRG Development Economics Research Group of The World Bank

DOC disaster option for CAT risk

DPPC Disaster Prevention and Preparedness Commission (Ethiopia)

DRP disaster response product

EC/ACP European Commission/Africa-Caribbean-Pacific

EIC Ethiopia Insurance Corporation

ENESA Entidad Estatal de Seguros Agrarios, the National Agricultural InsuranceAgency of Spain

ENSO El Niño southern oscillation (sea surface temperatures)

ESDVP Environmentally Sustainable Development Vice Presidency

ESSD The World Bank Environmentally and Socially Sustainable DevelopmentAdvisory Service

FAO Food and Agriculture Organization of the United Nations

FCIP Federal Crop Insurance Program

FSE The Financial Sector Group of The World Bank

GDP gross domestic product

GIIF Global Index Insurance Facility (proposed by CRMG)

Acronyms andAbbreviations

GMO genetically modified organisms

IBLI index-based livestock insurance

ICICI A private general insurance Lombard company in India

ICRISAT International Crops Research Institute for the Semi-Arid Tropics

IFC International Finance Corporation of the World Bank Group

IFFCO-Tokio A private general insurance company in India, a joint venture between Tokio-Marine and the Indian Fertilizer Association

IFPRI International Food Policy Research Institute

IMF International Monetary Fund

INISER Instituto Nicaraguense de Seguros y Reaseguros

Nicaraguan Institute for Insurance and Reinsurance

ISMEA Istituto di Servizi per il Mercato Agricolo Alimentare (Italian Institute forServices to Agricultural Food Markets)

KBS LAB Krishna Bhima Samruddhi Local Area Bank

LIL learning and innovation loan

MAMDA Mutuelle Agricole Marocaine d’Assurance

MMPI Malawi Maize Production Index

NASFAM National Smallholders Association

NDVI normalized difference vegetation index

NGO nongovernmental organization

NMSA National Meteorological Services Agency

OECD Organization for Economic Cooperation and Development

OI Opportunity International

PI production insurance

RI reinsurance

SADC Southern African Development Community

SECO State Secretariat for Economic Affairs, Swiss Trade Commission

SENAMHI Servicio Nacional de Meteorologia e Hidrologia del Peru (NationalMeteorology and Hydrology Service of Peru)

SRA Standard Reinsurance Agreement (U.S. crop insurance)

TCDAI Technical Committee for the Development of Agriculture Insurance(Peru)

UNCTAD United Nations Conference on Trade and Development

WFP World Food Program of the United Nations

viii Acronyms and Abbreviations

ix

This document was produced by Ulrich Hess, as task manager, and byJerry Skees, Andrea Stoppa, Barry Barnett, and John Nash, usingbackground papers written by Robert Townsend; Paul Siegel; andJerry Skees, Barry Barnett, and Jason Hartell. (These papers can beviewed at the Commodity Risk Management Group (CRMG) website, www.itf-commrisk.org.) Panos Varangis led the work for thisstudy during its conceptual stage. The two appendixes are shortenedversions of contributions by CRMG authors Joanna Syroka andHector Ibarra to a forthcoming ISMEA (Istituto di Servizi per ilMercato Agricolo Alimentare) publication on innovations in agri-cultural risk management.

Although motivated by the solid and growing literature on alter-native risk management techniques, this paper is ultimately drivenby empirical results that would have been impossible to obtain with-out the development community’s support and demand for action.

At The World Bank, Karen Brooks and Richard Scobey, rural sec-tor managers in the Africa Region, supported the conceptual workand instilled a sense of realism and purpose into the ideas expressedhere. Jock Anderson and Derek Byerlee in the Agriculture and RuralDevelopment department continuously refreshed our ideas in theareas of agricultural risk management and food security risk man-agement. Kevin Cleaver and Sushma Ganguly, Sector Director andSector Manager, respectively, in the Agriculture and RuralDevelopment department, gave motivational advice and guidance.Ken Newcombe, ESDVP, encouraged this work and has become achampion of the Global Index Insurance Facility (GIIF). In his IFCdays, Cesare Calari, FSE, was an early supporter of weather riskmanagement concepts, and he continues to encourage this line ofthinking in his various capacities. Rodney Lester, senior insuranceexpert in FSE, also contributed advice and support. Xavier Gine,DECRG, helped to shape our thinking on smallholder access tofinancial services. Our colleagues in the social development andsocial protection areas—notably Harold Alderman, Will Wiseman,and Elena Galliano—helped with the crossover to the social riskmanagement realm, providing a better understanding of the needs ofvulnerable populations and the relevance of insurance techniques forsafety nets.

Development partners have continuously prompted qualityleaps forward through their particular expertise. Richard Wilcoxof the UN World Food Program (WFP) pushes the weather insur-

Preface andAcknowledgments

ance idea to new limits and has shaped ourthinking at that level. Alexander Sarris, FAO,and Lamon Rutten, UNCTAD, supported CRMGin the areas of commodity risk management.

The key concepts espoused in this paper havebeen developed in the academic community aswell. Ronald Duncan and his group at the WorldBank systematically explored index insuranceideas in the early 1990s (Priovolos and Duncan1991). Also in the early 1990s, Peter Hazell,IFPRI, and Jerry Skees analyzed the shortcom-ings of traditional crop insurance and suggestedthe weather index insurance alternative.

This ESW insists on market-based insurancetechniques, the only sustainable way to transferrisk out of agriculture. At the same time, marketgaps exist, and often markets fail the poor. CRMGand its partners—by “crowding in” the privatesector—are building the bridges necessary to spanthese gaps. None of this would have been possiblewithout the visionary thinking of leaders in theweather risk management markets. Ravi Nathan,ACE Insurance of North America, in particular, hashelped to globalize the market beyond OECDcountries thanks to creative partnership and risk-sharing structures that include marketing partnersfrom developing countries. His vision continues toinspire the market and our work. Crucial advisorson the work and ideas of the CRMG as repre-sented here are Brian Tobben and William Dick ofPartner RE; Juerg Trueb, of Swiss RE; and RickMcConnell, formerly of the Agricultural FinancialServices Corporation, Alberta. Bruce Tozer, atRabobank, and Roy Leighton, at Carlyon, haveadvised and encouraged CRMG and the Inter-national Task Force for Commodity Risk Man-agement, throughout their existence, withwisdom and passion.

The demand for systematic techniques of agri-cultural risk management in developing countriesultimately came from the people who deal withfarmers and who partly make the farmers’ riskstheir own. The vision and inspiration of NachiketMor, of ICICI Bank, India, and Vijay Mahajan, ofBASIX, India, are the real motivators behind theastounding success of weather insurance tech-niques. This paper and its proposals would beunthinkable without the ICICI Lombard andBASIX weather insurance pilots and their revela-tion that farmers understand and appreciate thetransparency and timeliness of the product.Ramesh and Vasumathi in Mahahbubnagar,Ramana and Gunaranjan in Hyderabad andMumbai, Virat Divyakirti at ICICI Lombard, andBindu Ananth, also at ICICI, were the architects ofa simple innovation that promises to change India’srural landscape. Champions for pilot projects else-where are Rachid Guessous, MAMDA, in Morocco;Ramon Serrano, INISER, in Nicaragua; andShadreck Mapfumo, OI, in Malawi.

The authors wish to acknowledge the generoussupport of the Swiss State Secretariat for EconomicAffairs, SECO. SECO has supported CRMG’s pilotprograms in innovative agricultural risk manage-ment, and major lessons from these pilots informthis report. The European Commission and, inparticular, Henny Gerner are associated with thework of CRMG and, by extension, with this ESWthrough their constructive criticism of and supportfor the idea of the Global Index Insurance Facility(GIIF).

Finally, the authors express their sincere grat-itude to World Bank reviewers Jock Andersonand Stephen Mink and to Celeste Sullivan andAnne Goes, of GlobalAgRisk, Inc., for their edi-torial assistance.

x Managing Agricultural Production Risk

xi

The creation of risk transfer markets for weather events in devel-oping and emerging economies is rapidly progressing. This docu-ment describes several sources of risk that create poverty traps forpoor households and impede the development process, focusingon low-probability, high-consequence weather risk events as theyrelate to rural households. These types of risks are highly corre-lated and require special financing and access to global markets ifthey are to be pooled, rendered diversifiable, and improved in pric-ing. Thus, a significant contribution of this paper is the introductionof index insurance, highlighting its use at the micro-, meso-, andmacrolevels for risk transfer. By using index insurance products, itis possible to organize systems that take advantage of global mar-kets to transfer the correlated risks associated with low-probability,high-consequence events out of developing countries. This docu-ment presents both a conceptual backdrop for understanding thissystem and a progress report on several World Bank efforts to assistcountries in using their limited government resources to facilitatemarket-based agricultural risk transfer when faced with naturaldisasters.

While global markets providing reinsurance for natural disastersare both large and growing, they are rarely interested in taking suchrisk from developing and emerging economies. In part, this isbecause developing countries have weak primary insurance mar-kets. Before agreeing to provide reinsurance, global reinsurersengage in due diligence investigations of primary insurers and of therisks the primary insurers wish to transfer. Compared to traditionalinsurance products, index insurance has far fewer problems withhidden information and hidden action. This reduces the reinsurers’due diligence and underwriting costs and makes accepting naturaldisaster risk from new insurance providers in developing countriesmore attractive. Nonetheless, natural disaster losses can be signifi-cant, and carefully crafted ways to finance such losses are criticalpreconditions for shifting the risk into global markets. Innovation inpooling these risks globally may also facilitate the transfer of natu-ral disaster risk from developing countries.

One global innovation currently being prepared by the WorldBank and the European Commission involves a Global IndexInsurance Facility (GIIF). The GIIF will have three functions targetedat helping insurance providers in developing countries build capac-ity: (1) supporting the technical assistance and infrastructure needed

Executive Summary

to develop index insurance based on quality data;(2) aggregating and pooling risk from differentdeveloping countries to improve pricing and risktransfer into the global reinsurance and capitalmarkets; and (3) cofinancing certain insuranceproducts on a bilateral basis from donor to develop-ing country. Importantly, the third function will beseparate from the commercial activity representedin the first and second functions. A global effortto facilitate these three functions could representa major breakthrough for those developing coun-tries exposed to extreme natural disaster risk.

Another promising realm of innovation is thedevelopment of improved technology both to mea-sure weather and to link it to farming systems toforecast crop yields. Improved and less costly sys-tems for measuring weather events in developingcountries will play a significant role in the potentialsuccess of many of the ideas presented here. Secureand accurate measurement will influence both thepricing of index insurance and the demand fromend users. Improvements in developing countriesfirst in measuring the vegetative cover using satel-lite images and then in forecasting the value of thatvegetation in terms of crop yields or grazing valuecould lead to the availability of enhanced types ofindex insurance products. Additionally, moresophisticated crop models linking weather, man-agement systems, and soil condition can be used toprovide insurance products that protect against thedominant random variable affecting production—the weather.

Transferring risk out of developing countries isimportant for a number of reasons. Natural dis-asters impede development, push households intopoverty, and drain fiscal resources. Many naturaldisasters are directly tied to extreme weather eventsthat can have devastating impacts on agriculture.Nearly three-fourths of the 1.3 billion people world-wide living on less than US$1 per day depend onagriculture for their livelihoods. In many countriesaround the world, agricultural development clearsthe way for overall economic development in thebroader economy, forging a strong link betweenweather, the livelihoods of the poor, and develop-ment. Yet, no effective ex ante solutions for deal-ing with weather risks in developing countriesexist. Rather, developing countries, the WorldBank, and the donor community are currentlyheavily exposed to natural disaster risk via expost responses such as financial bailouts, debt for-giveness, and emergency response.1 None of these

responses are optimal. They fail to provide an effec-tive safety net for the poor; they can be inequitableand untimely; and they create a dependency thathas dire consequences.

If the planning for and financing of extremeweather events were to occur ex ante, access toboth formal and informal lending should improve.As broader financial services become more acces-sible to the rural poor, newer technologies willbe used, and improvements in productivity andincomes should follow.

Farmers around the world utilize various riskcoping and risk management strategies, but manyof these strategies are inefficient. The economicdevelopment literature is full of cases illustratinghow poor, risk-averse farmers often forego poten-tially higher incomes to reduce their risk exposure.Both individual households and the larger societyincur costs for smoothing consumption acrossincome shocks. In many cases, following majorincome shocks, the poor must resort to high inter-est rate loans. Many argue that the poor cannotafford to purchase ex ante insurance protectionagainst extreme weather events, but the wide-spread use of ex post loans suggests otherwise.

The challenge remains of how to make insur-ance against extreme weather events both moreeffective and more affordable. Two major consid-erations inhibit the development of risk transfermarkets for agricultural losses caused by extremeweather events: First, organizing ex ante financingfor highly correlated losses can result in ex-tremely large financial exposure; and, second,asymmetric information problems, such as moralhazard and adverse selection, lead to high trans-action costs. The latter also makes it nearly impos-sible to provide traditional agricultural insurancefor small farmers, because the large fixed transac-tion costs greatly increase the average cost, permonetary unit, of insurance protection for small-holder agriculture. Unfortunately, there are fewsuccessful examples to consider; the heavily sub-sidized crop insurance provided by governmentsin developed countries is both costly and ques-tionable in terms of net social welfare.

Researchers frequently find that economic deci-sion makers underestimate the likelihood and/ormagnitude of low-probability, high-consequenceloss events, leading to a reduced willingness topay for insurance to protect against these events.At the same time, because insurers have littleempirical information about the likelihood and/or

xii Managing Agricultural Production Risk

Executive Summary xiii

magnitude of extreme events, they tend to addlarge extra costs to premium rates for insuranceproducts protecting against them. This diver-gence between what potential purchasers will payand what insurers will accept results in agricul-tural insurance markets that clear less thansocially optimal quantities of risk transfer.

New conceptual models are being developedto facilitate the transfer of extreme weather riskout of developing countries. This documentreports on the progress of several ongoing effortsby the Commodity Risk Management Group(CRMG) at the World Bank that have been moti-vated by these models. All of these efforts are builton the premise that index-based insurance prod-ucts can effectively address the challenges of theex ante financing of highly correlated losses andhigh transaction costs. Index insurance productspay indemnities based on an independent meas-ure highly correlated with realized losses. Unliketraditional crop insurance, which attempts tomeasure individual farm yields, index insurancemakes use of variables largely exogenous to theindividual policyholder, including area yield orweather events such as temperature or rainfall.This feature greatly reduces the need fordeductibles and copayments, since it results invery little exposure to asymmetric informationproblems, such as moral hazard and adverse selec-tion. By eliminating farm-level loss adjustment,index insurance products achieve lower transactioncosts than are possible with traditional agricul-tural insurance products.

Purchasers of index insurance products areexposed to basis risk. Since index insuranceindemnities are triggered not by farm-level lossesbut rather by the value of an independent measure(the index), a policyholder can experience a lossand yet receive no indemnity. Conversely, thepolicyholder may not experience a loss and yetnonetheless receive an indemnity. The effective-ness of index insurance as a risk managementtool depends on how positively correlated farm-level losses are with the underlying index.Importantly, since farmers have incentives to con-tinue to produce or to try to save their crops andlivestock even in the face of bad weather events,index insurance should provide for a more effi-cient allocation of resources.

Since they are standardized and transparent,index insurance products can also function as re-insurance instruments that transfer the risk of

widespread, correlated agricultural productionlosses. To the extent that institutions can be createdto aggregate and pool the low-probability, high-consequence tail risk that results from writinginsurance on these events, the divergence betweeninsurers’ willingness to accept and potential pur-chasers’ willingness to pay should decrease, caus-ing the market to clear at high quantities of risktransfer.

This paper was written to inform a broad rangeof decision makers about the progress being madein risk transfer for natural disaster risk. While thefocus here is on agriculture, many of the same con-cepts can clearly also be used for other sectorsexposed to natural disaster risk. Two basic innova-tions dominate the conceptual framework: (1) useof index-based insurance; and (2) layering risk tofacilitate risk transfer. In many cases, individualswill self-insure against the layer of risk com-posed of high-probability, low-consequencelosses. Some form of government interventionmay be required to achieve higher levels of risktransfer in the layer of risk composed of low-prob-ability, high-consequence losses. Between thesetwo extremes lies a layer of risk that, with appro-priate risk transfer and pooling structures, can betransferred using market mechanisms.

Since catastrophe risks (CAT risks) are one ofthe impediments to market development, aframework has been developed for governmentaction in the management of agricultural risk thatincludes models for government intermediationof catastrophic risk through government disasteroptions for CAT risk, or DOC. This framework pro-poses that governments buy index-based cata-strophic risk coverage in international marketsand offer them at rates lower than global marketrates to local insurers, who then pass the savingson to end users in developing countries. This sys-tem would mitigate large-loss/infrequent risksthat are usually difficult and expensive to reinsurein traditional reinsurance markets and would ulti-mately allow local insurers to cover more peopleagainst the extreme risks in an ex ante fashion.

This paper includes several case studies illustrat-ing the application of these concepts in countriesaround the world. While the specifics vary based oneach country’s needs, all of the cases involve the useof index insurance and/or the layering of risk tofacilitate risk transfer. The final chapter of this doc-ument describes potential future roles for the WorldBank in the area of agricultural risk management.

1

This document presents innovations in agricultural risk managementfor natural disaster risk, with the focus on defining practical roles forgovernments of developing countries and the World Bank in devel-oping risk management strategies.2 Recent success stories demon-strate that the World Bank can play a role in assisting countries intaking actions that effectively use limited government resources tofacilitate market-based agricultural risk transfer. This is important, asdeveloping countries, the World Bank, and the donor community arecurrently heavily exposed to natural disaster risk without the benefitof ex ante structures to finance losses. Instead, at each big drought orother natural disaster, those affected must appeal for financial sup-port, leaving them vulnerable to the mercy of ad hoc responses fromgovernment, the international financial institutions, and donors. In most developing countries, livelihoods are not insured by inter-national insurance/reinsurance providers, capital markets, or evengovernment budgets. In addition, natural disasters and price risk inagriculture also impede development of both formal and informalbanking. Without access to credit, risk-averse poor farmers are lockedin poverty, burdened with old technology, and faced with an ineffi-cient allocation of resources.

Advances in risk transfer in developed countries are leading theway to solutions to many social problems. Shiller (2003) documentsprogress and charts a course for far more innovation as the democra-tization of finance and technology spur global risk pooling. Financialand reinsurance markets in developed countries are rapidly devisingindex-based instruments that allow for the transfer of systemic risksand even of livelihood risks. Innovations in risk transfer for naturaldisasters have been well documented (Doherty 1997; Skees 1999b).The challenge is to make these innovations relevant in developingcountries and to facilitate knowledge and access.

Is the absence of formal transfers of natural disaster risk inevitablein developing countries? Clearly not; formal global markets for off-setting natural disaster risks and weather risks are widely used indeveloped countries.3 This document demonstrates how these mar-kets can be used to insure natural disaster risk in developing coun-tries. Agricultural sectors in developing countries are much moreexposed to the vagaries of weather than are those of richer countries,so this protection would be even more valuable to them.

Is it a luxury to offer insurance to poor people who lack properroads or even safe drinking water? Every government must set its

Introduction1

own priorities. Careful consideration of the bene-fits and costs of different interventions is critical.Still, the poor are forced to make production deci-sions using the objective of minimizing risk, ratherthan maximizing profits, and thus they must foregomore remunerative activities that could providemeans of escape from their poverty. An effective andtimely insurance mechanism might allow people toengage in higher risk, higher return activities with-out putting their livelihoods at risk. Spurring devel-opment via improved financial markets is importantfor developing countries.

Are there any effective precedents for agricul-tural insurance mechanisms in developing coun-tries? While these innovations are just taking hold,progress has been made with weather insurancefor farmers in India, Ukraine, Nicaragua, Malawi,Ethiopia, and Mexico. Several other experimentsare also documented in this work. Weather insuredfarmers in India say they either have a good crop—in which case it does not matter if they do not recoupthe insurance premium—or they have a monsoonfailure, in which case they receive an insurance pay-out. This payout will at least cover the farmers’ cashoutlay and perhaps provide them with enough extramoney to keep their children in school and to pre-serve assets they would otherwise be forced to liqui-date, often at greatly reduced prices. These farmerswill be likely to invest a little more in the right seedsand fertilizer at the right time. Quantifying thisimpact is difficult right now, but a large impact assessment will soon provide more informationon the effectiveness of this program. It is clear already, however, that when offered the choice,many farmers in India will pay for fully pricedweather insurance. Even farmers with access tothe government-subsidized crop insurance prod-uct choose to buy the market-priced weather in-surance product. They say they like the objectivenature of the weather index; they can check theweather station measurements themselves. Theyalso like the timely payout. Indeed, on this count,the new rainfall index insurance, which pays on atimely basis, compares favorably to the nationalcrop insurance product, which might pay only afteras much as eighteen months.

Is this insurance only suitable for large commer-cial farmers? One true advantage of weather insur-ance is that it can be targeted to small farmers, as nomonitoring is needed to verify farm-level losses.The Indian experience clearly demonstrates thatsmall farmers find value in weather insurance.

BASIX (a microfinance entity in Andhra Pradesh)estimates that all of the 427 farmers who boughtweather insurance policies in 2003 have small- tomedium-sized farms of between two and ten acres,providing an average yearly income of 15,000 to30,000 Rupees, or between US$1 and US$2 per day.Currently, many farmers buying weather insur-ance in India are repeat customers. Clearly, thesefarmers were not too poor to buy the product. Earlysurvey results demonstrate that more than half ofthose purchasing the insurance list managing riskas their primary reason. Some farmers might havechosen this new insurance option over the prospectof paying high interest to moneylenders when cashis needed after a harvest failure.

Is India’s insurance program sustainable? Withthe pilot program now in its third year and otherinsurance companies replicating and selling theproduct, BASIX has mainstreamed the weather in-surance product and automated delivery to an ex-pected 8,000 clients for the 2005 season. Countriesin sub-Saharan Africa and Latin America are start-ing their own weather insurance projects at micro-and macrolevels. Ethiopia is piloting a weather-insurance-based drought emergency response, forexample. Furthermore, weather insurance seems tobe a good business. The Indian weather insuranceprogram has emerged without the support of gov-ernment subsidies. The Commodity Risk Manage-ment Group (CRMG) of the World Bank has advisedthose who were ready to try these new approachesto agricultural risk management.

How can this process be operationalized in theWorld Bank and elsewhere? Task managers andpractitioners may want to follow this work withpotential projects, but how do they get started?This document presents ideas on how to structurea solid framework of action. Among the importantpublic goods that governments and the World Bankmight provide are, for example, weather stationsand risk financing for catastrophic protection.

Governments in drought-prone countries anddonors and relief agencies should also be aware ofother kinds of projects that use risk managementmarkets to improve the response to weather-relatedshocks. This document explores how current adhoc disaster relief mechanisms can be modified andcomplemented by a more systematic response torecurrent droughts.

When assessing proper roles for government,the first factors to consider are the economic bene-fits that can be created by risk management tools,

2 Managing Agricultural Production Risk

Introduction 3

the characteristics of the risks faced by farmers ina specific area, and the challenges associated withcreating and maintaining risk management toolssuch as insurance. In general, agricultural riskmanagement presents no “one-size-fits-all” policyrecommendation for the role of government. Mostgovernments consider at least four criteria whenconsidering alternatives for addressing agricul-tural risk management needs: (1) fiscal constraint;(2) growth; (3) market-oriented risk-transfer; and(4) social goals of reducing poverty and vulnerabil-ity in rural areas.

Chapter 2 of this document begins with anoverview of risk in agriculture, focusing on howdecision makers currently cope with and managerisk in developing countries and on the impedi-ments to developing effective risk transfer markets.High transaction costs, problems with correlatedrisk, and the classic problems of moral hazard andadverse selection clearly increase the cost of tradi-tional insurance. Chapter 3 reviews in detail the ex-periences of some developed countries withagricultural risk transfer. A clear message emergesabout the costs to governments and the inefficien-cies of these systems, supporting the need to searchfor new solutions appropriate for developing coun-tries. The stark contrast between what is possible ina developed country versus what is possible in adeveloping country further motivates a continuing

search for new solutions. Chapter 4 explores alter-nate solutions based on the concept of weatherindex insurance that covers farmers against weatherevents leading to serious agricultural losses, high-lighting the advantages of such systems for devel-oping countries. Chapter 5 brings together two coreinnovations: first, the use of index insurance to in-sure against detrimental weather events, a formwith significantly lower monitoring costs; and sec-ond, the layering of insurance products to segmentrisk more efficiently, thus allowing for transfer ofcorrelated risk. These innovations provide a richframework for introducing new approaches to risksharing and risk transfer in developing countries.Chapter 5 outlines an effective role for the WorldBank and other donors in this important domain ofnatural hazard risk management. Chapter 6 pro-vides an overview of a number of ongoing agricul-tural risk pilot programs and case studies for invarious countries. Finally, Chapter 7 makes rec-ommendations for the role of the World Bank andcountry governments in facilitating the develop-ment of innovation in agricultural risk manage-ment. Following the core chapters, the reportincludes two detailed appendixes: the first explainshow to structure and price weather index insur-ance; the second provides more background to risktransfer programs and experiences in Ukraine,Mexico, Canada, and India.

5

Agricultural risk is associated with negative outcomes stemmingfrom imperfectly predictable biological, climatic, and price variables.These variables include natural adversities (for example, pests anddiseases), climatic factors not within the control of agricultural pro-ducers, and adverse changes in both input and output prices. To setthe stage for the discussion on how to deal with risk in agriculture,we classify the different sources of that risk.4

Agriculture is often characterized by high variability of productionoutcomes, that is, by production risk. Unlike most other entrepreneurs,agricultural producers cannot predict with certainty the amount ofoutput their production process will yield, due to external factorssuch as weather, pests, and diseases. Agricultural producers can alsobe hindered by adverse events during harvesting or collecting thatmay result in production losses.

Both input and output price volatility are important sources ofmarket risk in agriculture. Prices of agricultural commodities areextremely volatile. Output price variability originates from bothendogenous and exogenous market shocks. Segmented agriculturalmarkets will be influenced mainly by local supply and demand con-ditions, while more globally integrated markets will be significantlyaffected by international production dynamics. In local markets, pricerisk is sometimes mitigated by the “natural hedge” effect, in which anincrease (decrease) in annual production tends to decrease (increase)output price (though not necessarily farmers’ revenues). In integratedmarkets, a reduction in prices is generally not correlated with localsupply conditions, and therefore price shocks may affect producersin a more significant way. Another kind of market risk arises in theprocess of delivering production to the marketplace. The inabilityto deliver perishable products to the right market at the right timecan impair producers’ efforts. The lack of infrastructure and of well-developed markets makes this a significant source of risk in manydeveloping countries.

The ways businesses finance their activities is a major concern formany economic enterprises. In this respect, agriculture has its ownpeculiarities. Many agricultural production cycles stretch over longperiods, and farmers must anticipate expenses they will only be ableto recuperate after marketing their product. This leads to potentialcash flow problems, which are often exacerbated by lack of access tocredit and the high cost of borrowing. These problems can be classi-fied as financial risk.

Risk and Risk Managementin Agriculture2

Institutional risk, that is, risk generated by un-expected changes in regulations that affect produc-ers’ activities, constitutes another important sourceof uncertainty for agricultural producers. Changesin regulations can have significant impact on theprofitability of farming activities. This is particularlytrue for import/export regimes and for dedicatedsupport schemes, but sanitary and phytosanitaryregulations too can restrict producers’ activities andimpose costs on households.

Like most other entrepreneurs, agricultural pro-ducers are responsible for all the consequences oftheir activities. Growing concern over the impact ofagriculture on the environment, however, includ-ing the introduction of genetically modified organ-isms (GMO), may cause an increase in producerliability risk. Finally, agricultural households, alongwith other economic enterprises, are exposed topersonal risks to the well-being of people who workon the farm and asset risks, including possible dam-age or theft of production equipment and assets.(See Box 2.1.)

In discussing how to design appropriate riskmanagement policies, it is useful to understandstrategies and mechanisms employed by producersto deal with risk, including the distinction betweeninformal and formal risk management mechanismsand between ex ante and ex post strategies.5 Ashighlighted in the 2000/2001 World DevelopmentReport (World Bank, 2001), informal strategiesare identified as “arrangements that involve indi-viduals or households or such groups as commu-nities or villages,” while formal arrangements are“market-based activities and publicly providedmechanisms.” The ex ante or ex post classificationfocuses on the point at which the reaction to risktakes place: ex ante responses take place before thepotential harming event; ex post responses takeplace after the event. Ex ante reactions can be furtherdivided into on-farm strategies and risk-sharingstrategies (Anderson, 2001). Table 2.1 summarizesthese classifications.

INFORMAL MECHANISMS6

Ex ante informal strategies are characterized bydiversification of income sources and choice of agri-cultural production strategy. One strategy producerscan employ is simply to avoid risk. In many cases,extreme poverty makes people very risk averse;producers facing these circumstances often avoidactivities that entail significant risk, even though

the income gains might be larger than for less riskychoices. This inability to accept and manage risk andaccumulate and retain wealth is sometimes referredto as the “the poverty trap” (World Bank 2001).

Once producers have decided to engage in farm-ing activities, the production strategy selected be-comes an important means of mitigating the risk ofcrop failure. Traditional cropping systems in manyplaces rely on crop and plot diversification. Cropdiversification and intercropping systems reducethe risk of crop failure due to adverse weatherevents, crop pests, or insect attacks. Morduch (1995)presents evidence that households whose con-sumption levels are close to subsistence (and whichare therefore highly vulnerable to income shocks)devote a larger share of land to safer, traditionalvarieties such as rice and castor than to riskier,high-yielding varieties. Morduch also finds thatnear-subsistence households diversify their plotsspatially to reduce the impact of weather shocksthat vary by location.

Apart from altering agricultural productionstrategies, households also smooth income by diver-sifying income sources, thus minimizing the effectof a negative shock to any one of them. Accordingto Walker and Ryan (1990), most rural householdsin villages of semi-arid India surveyed by the Inter-national Crops Research Institute for the Semi-AridTropics (ICRISAT) generate income from at leasttwo different sources; typically, crop income is ac-companied by some livestock or dairy income. Off-farm seasonal labor, trade, and sale of handicraftsare also common income sources. The importanceto risk management of income source diversifica-tion is emphasized by the Rosenzweig and Stark(1989), who find that households with high farmprofit volatility are more likely to have a householdmember engaged in steady wage employment.

Buffer stock accumulation of crops or liquid as-sets and the use of credit present obvious means bywhich households can smooth consumption. Limand Townsend (1998) show that currency and cropinventories function as buffers or precautionarysavings.

Crop-sharing arrangements in renting land andhiring labor can also provide an effective meansof sharing risk among individuals, thus reducingproducer risk exposure (Hazell 1992). Other risksharing mechanisms, such as community-levelrisk pooling, occur in specific communities or ex-tended households where group members transferresources among themselves to rebalance marginal


Risk and Risk Management in Agriculture 7

utilities (World Bank 2001). These arrangements,however, while effective for counterbalancing theconsequences of events affecting only some mem-bers of the community, do not work well in casesof covariate income shocks (Hazell 1992).

Typical ex post informal income-smoothingmechanisms include the sale of assets, such as landor livestock (Rosenzweig and Wolpin 1993), or thereallocation of labor resources to off-farm laboractivities. Gadgil, et al. (2002), argue that southern

Box 2.1 Asset-Based Risk Management

Siegel (2005) broadens the risk discussion into anasset-based risk management framework. This compre-hensive approach considers the dynamics of riskswithin a given context. The asset-based approach usesa “livelihood focus,” recognizing that rural householdshold a portfolio of assets that they allocate among arange of welfare generating activities and that the par-ticular livelihood activities pursued reflect explicit (orimplicit) multidimensional objectives that include eco-nomic, social, cultural, and environmental outcomes(Chambers and Conway 1992; Carney et al. 1999).The asset-based approach helps clarify why and howhouseholds manage assets and risks to “select” certainlivelihood strategies for achieving welfare outcomesgiven specific asset-context interface conditions.

The asset-based risk management approach focuseson the long-term implications of short-term decisionsabout the allocation of assets. Coping strategies usedby poor rural households can lead to the degradationor decapitalization of assets, as when, for example,trees are cut down or children are removed fromschool, and these actions can contribute to a cycle ofpoverty. Alternatively, livelihood strategies that leadto improved asset portfolios, for example, invest-ments in improved technology, training programs,and empowerment through social and political net-works, can foster a virtuous cycle of sustainablegrowth. Asset accumulation and changes in liveli-hood strategies are thus important for sustained improvements in household well-being.

Improved management of rural risk is critical toachieve rural growth and reduce poverty. It is critical,however, to move beyond a narrow risk managementfocus to a more holistic rural development approachthat focuses attention on building, enhancing, main-taining, and protecting household assets. The develop-ment of new rural risk management instruments offers

the potential to improve household livelihood options,yet their ultimate success will depend on the linkagesamong assets, context, behavior, and outcomes. Thus,the real question to be asked is what optimal risk man-agement instruments will allow households to maxi-mize their objectives in terms of expected income andvariability of income?

The relationship between assets and productivityexplains the poverty cycle and the difficulty the poorhave in improving their livelihoods. A household’sportfolio of assets influences their risk attitude andtheir ability to respond to risk. Assets also determinethe types of activities that can be undertaken. Moreproductive activities are typically associated withgreater risk, so how assets are utilized will impactproductivity as a function of both expected incomeand variability of income. At the household level,agricultural risk management instruments reduce thevariability of household incomes. The expectation isthat by reducing risk and uncertainty, households willbe able to accumulate assets and undertake moreproductive investments.

In the design of risk management instruments, it isimportant to account for the unique context pre-sented in different situations. Risk management in-struments must be tailored to specific constraints andobjectives within the country, community, andhousehold context.

In considering the potential applications of indexinsurance in developing countries, it is important toremember that index insurance is not necessarily ap-plicable or replicable for every situation. Nor should itbe inferred that index insurance is a substitute forother risk management strategies. Index insurancecan, however, provide a starting point and, ideally, aspringboard for the development of a variety of riskmanagement mechanisms.

Note: A more detailed discussion of these issues can be found in “Looking at Rural Risk Management Using an Asset-Based Approach,” abackground paper for this report by Paul Siegel. In particular, the reader is directed to Figure 1, which depicts the relationships among assets,context, behavior, and outcomes.

Source: Siegel 2005.

Indian farmers who expect poor monsoon rainscan quickly shift from 100 percent on-farm laboractivities to mainly off-farm activities. Fafchamps(1993), in his analysis of rain-fed agriculture amongWest African farmers, emphasizes the importanceof building labor flexibility into the productionstrategy.

As reported by Townsend (2005), in analyzingthe cost of risk on ex ante agricultural productionstrategies, Rosenzweig and Binswanger (1993),Morduch (1995), and Kurosaki and Fafchamps(2002) all find considerable efficiency losses associ-ated with risk mitigation, typically due to lack ofspecialization—in other words, farmers trade offincome variability with profitability.

The need to smooth consumption not only againstidiosyncratic shocks but also against correlatedshocks comes at a serious cost in terms of productionefficiency and reduced profits, thus lowering theoverall level of household consumption. A majorconsideration for innovation would be to shift cor-related risk from rural households (Skees 2003).One obvious solution would be for rural householdsto share risk with households or institutions fromareas largely uncorrelated with the local risk condi-

tions. Examples of such extra-regional risk sharingsystems are found in the literature, including, creditand transfers between distant relatives (Rosenzweig,1988; Miller and Paulson 2000); migration and mar-riages (Rosenzweig and Stark 1989); or ethnic net-works (Deaton and Grimard 1992). Although thesestudies find some degree of risk sharing and thusof insurance against weather, use of such systemsis not so widespread as to cover all households, nordo they come even close to providing a fully efficientinsurance mechanism. Most households are there-fore still left with no insurance against correlatedrisks, the main source of which is weather.

FORMAL MECHANISMSFormal risk management mechanisms can be classi-fied as publicly provided or market based (Table 2.1).Government action plays an important role in agri-cultural risk management, both ex ante and ex post.Ex ante education and services provided by agri-cultural extension help familiarize producers withthe consequences of risk and help them adoptstrategies to deal with it. Governments also reducethe impacts of risk by developing relevant infra-


Table 2.1 Risk Management Strategies in Agriculture

Formal Mechanisms

Informal Mechanisms Market Based Publicly Provided

Avoiding exposure to riskCrop diversification and intercroppingPlot diversificationDiversification of incomesourceBuffer stock accumulationof crops or liquid assetsAdoption of advancedcropping techniques (fertilization, irrigation, resistant varieties)

Crop sharingInformal risk pool

Sale of assetsReallocation of laborMutual aid

Contract marketingand futures contractsInsurance

Credit

Agricultural extensionPest management systemsInfrastructures (roads, dams,irrigation systems)

Social assistanceSocial fundsCash transfer

Source: Anderson 2001; Townsend 2005; World Bank 2001.

On-farm

Sharing riskwith others

Coping withshocks

EXPO

STEX

AN

TEST

RATE

GIE

SST

RATE

GIE

S

Risk and Risk Management in Agriculture 9

structure and by adopting social schemes and cashtransfers for relief after shocks have occurred.7

As mentioned in the section on informal mech-anisms, production and market risks are probablythose with the largest impact on agricultural pro-ducers. Various market-based risk managementsolutions have been developed to address thesesources of risk.

Price Risk Management

One way producers have traditionally managedprice variability is by entering into preharvest agree-ments that set a specific price for future delivery.These arrangements, known as forward contracts,allow producers to lock in a certain price, thus re-ducing risk but also foregoing the possible benefitsof positive price deviations. In specific markets,and for specific products, these arrangements haveevolved into futures contracts, traded on regulatedexchanges on the basis of specific trading rules andfor specific standardized products. This reducessome of the risks associated with forward contract-ing (for example, default). A further evolution inhedging opportunities for agricultural producershas been the development of price options, a priceguarantee that allows producers to benefit from afloor price while also allowing them to take advan-tage of positive price changes. With price options,agents pay a premium to purchase a contract thatgives them the right (but not the obligation) to sellfutures contracts at a specified price. Price optionsfor commodities are regularly traded on exchanges,but they can also be traded in over-the-countermarkets. Futures and options contacts can be ef-fective price risk management tools as well as im-

portant price discovery devices and market trendindicators.

For agricultural producers in developing coun-tries, access to futures and options contracts is prob-ably the exception rather than the rule. Futures andoptions markets in developed countries representimportant price discovery references for inter-national commodity markets, however, and indirectaccess to these exchange-traded instruments maybe granted through the intermediation of collectiveaction by producer groups such as farmer cooper-atives or national authorities.8 While an importantreality for some commodities, futures and optionsare not available for all agricultural products.

Production/Weather Risk Management

Insurance is another formal mechanism used inmany countries to share production risks. Insurancedoes not as efficiently manage production risk,however, as derivative markets do price risks. Pricerisk is highly spatially correlated and, as illus-trated in Figure 2.1, futures and options are ap-propriate instruments for dealing with spatiallycorrelated risks. In contrast, insurance is an ap-propriate risk management solution for indepen-dent risks. Agricultural production risks typicallylack sufficient spatial correlation to be effectivelyhedged using only exchange-traded futures or op-tions instruments. At the same time, agriculturalproduction risks are generally not perfectly spatiallyindependent; therefore, insurance markets do notwork at their best. Skees and Barnett (1999) refer tothese risks as “in-between” risks. According toAhsan, et al. (1982), “good or bad weather may havesimilar effects on all farmers in adjoining areas,”

Figure 2.1 Independent Versus Correlated Risk

Auto, life,� fire

Crop� yields

Prices,� interest rates

Perfectly� correlated�

(systemic)

Perfectly� independent

Insurance� markets

Options and futures� markets

Source: Miranda and Glauber 1997.

and, consequently, “the law of large numbers, onwhich premium and indemnity calculations arebased, breaks down.” In fact, positive spatial corre-lation in losses limits the risk reduction obtainableby pooling risks from different geographical areas.This increases the variance in indemnities paid byinsurers. In general, the more the losses are posi-tively correlated, the less efficient traditional insur-ance is as a risk-transfer mechanism. For many ideaspresented in this document, a precondition for suc-cess is a high degree of positive correlation of losses.

The lack of statistical independence is not theonly problem with providing insurance in agricul-ture. Another set of problems relates to asymmetricinformation, the situation in which the insured hasmore knowledge about his or her own risk profilethan does the insurer. Asymmetric informationcauses two problems: adverse selection and moralhazard. In the case of adverse selection, farmershave better knowledge than do the insurers aboutthe probability distribution of losses. The farmersthus occupy the privileged situation of knowingwhether or not the insurance premium accuratelyreflects the risk they face. Consequently, only farm-ers bearing greater risks will purchase the cover-age, generating an imbalance between indemnitiespaid and premiums collected. Moral hazard simi-larly affects the incentive structure of the relation-ship between insurer and insured. After enteringthe contract, the farmer’s incentive to take propercare of the crop diminishes, while the insurer haslimited effective means to monitor what may provehazardous behavior by the farmer. Insurers maythus incur greater than anticipated losses.

Agricultural insurance is often characterized byhigh administrative costs, due, in part, to the riskclassification and monitoring systems that insurersmust put in place to forestall asymmetric informa-tion problems. Other costs include acquiring thedata needed to establish accurate premium rates andconducting claims adjustments. As a percentage ofthe premium, the smaller the policy, typically, thelarger the administrative costs.

Spatially correlated risk, moral hazard, adverseselection, and high administrative costs are all im-portant reasons why agricultural insurance marketsmay fail. Cognitive failure among potential insur-ance purchasers and ambiguity loading on the partof insurance suppliers are other possible causes ofagricultural insurance market failure.9

If consumers fail to recognize and plan for low-frequency, high-consequence events, the likelihood

that an insurance market will emerge diminishes.When considering an insurance purchase, the con-sumer may have difficulty determining the valueof the contract or, more specifically, the probabil-ity and magnitude of loss relative to the premium(Kunreuther and Pauly 2001). Many decision mak-ers tend to underestimate their exposure to low-frequency, high-consequence losses. Thus, theyare unwilling to pay the full costs of an insuranceproduct that protects against these losses. Low-frequency events, even when severe, are frequentlydiscounted or ignored altogether by producers try-ing to determine the value of an insurance contract.This happens because the evaluation of probabilityassessments regarding future events is complexand often entails high search costs. Many peopleresort to various simplifying heuristics, but proba-bility estimates based on these heuristics may dif-fer greatly from the true probability distribution(Schade et al. 2002; Morgan and Henrion 1990).Evidence indicates that agricultural producers for-get extreme low-yield events. The general findingregarding subjective crop-yield distributions is thatagricultural producers tend to overestimate themean yield and underestimate the variance (Buzbyet al. 1994; Pease et al. 1993; Dismukes et al. 1989).

On the other side, insurers will typically loadpremium rates heavily for low-frequency, high-consequence events where considerable ambiguitysurrounds the actual likelihood of the event (Schadeet al. 2002; Kunreuther et al. 1995). Ambiguity isespecially serious when considering highly skewedprobability distributions with long tails, as is typicalof crop yields. Uncertainty is further compoundedwhen the historical data used to estimate probabil-ity distributions are incomplete or of poor quality,a very common problem in developing countries.Small sample size creates large measurement error,especially when the underlying probability distrib-ution is heavily skewed. Kunreuther et al. (1993)demonstrate via experimental economics that whenrisk estimates are ambiguous, loads on insurancepremiums can be 1.8 times higher than when insur-ing events with well specified probability and lossestimates.

Together, these effects create a wedge betweenthe prices that farmers are willing to pay for cata-strophic agricultural insurance and the prices thatinsurers are willing to accept. Thus, functioningprivate-sector markets may fail to materialize or,if they do materialize, they may cover only a smallportion of the overall risk exposure (Pomareda 1986).


11

To better understand agricultural risk management markets andgovernment policies to facilitate access to risk management instru-ments, it is worthwhile to analyze critically the experiences of somedeveloped countries. The experiences of the United States, Canada,and Spain are thus described for reference, but it is important to con-sider that these systems may not be replicable in or suitable for mostdeveloping countries. In addition, many developed countries haveinvolved market support and income transfer programs that extendwell beyond crop insurance. To the extent they are based on farmincome, these programs involve levels of protection against severecrop failures. The European community has extensive policies focus-ing on income protection.

CROP INSURANCE PROGRAMS INDEVELOPED COUNTRIESThis section presents overviews of agricultural risk managementprograms in three developed countries: the United States, Canada,and Spain. These countries have been able to implement substantialprograms to reduce yield and revenue risk for agricultural produc-ers. While these programs offer a variety of risk management prod-ucts for farmers, the programs require levels of government supportunfeasible for most countries.

The United States

In the United States, multiple peril yield and revenue insurance prod-ucts are offered through the Federal Crop Insurance Program (FCIP),a public/private partnership between the federal government andvarious private sector insurance companies.10 The program seeks toaddress both social welfare and economic efficiency objectives. Withregard to social welfare, private companies selling federal crop in-surance policies may not refuse to sell to any eligible farmer, regard-less of past loss history. At the same time, the program aims to beactuarially sound.

Policies are available for over one hundred commodities but in2004 just four crops—corn, soybeans, wheat, and cotton—accountedfor approximately 79 percent of the US$4 billion in total premiums.Excluding pasture, rangeland, and forage, approximately 72 percentof the national crop acreage is currently insured under the FCIP.

Approaches to Agricultural Risk in

Developed Countries

3

About 73 percent of total premiums are for revenueinsurance policies, while 25 percent are for yield in-surance policies.11

Most FCIP policies trigger indemnities at the farm(or even subfarm) level.12 Yield insurance offers arebased on a rolling four-to-ten-year average yield,known as the actual production history (APH) yield.The federal government provides farmers with abase catastrophic yield insurance policy, free of anypremium costs.13 Farmers may then choose to pur-chase, at federally subsidized prices, additionalinsurance coverage beyond the catastrophic level.This additional coverage, often called “buy-up”coverage, may be either yield or revenue insurance.Farm-level revenue insurance offers are based onthe product of the APH yield and a price index thatreflects national price movements for the particularcommodity.

For some crops and regions, defined alongcounty barriers, area yield and/or area revenuebuy-up insurance policies are offered through FCIP.On a per acre insured basis, area-level insuranceproducts tend to be less expensive than farm-levelinsurance products. Thus, in 2004, area yield andarea revenue policies accounted for 7.4 percent oftotal acreage insured but less than 3 percent of totalpremiums.

The federal government also provides a rein-surance mechanism that allows insurance compa-nies to determine (within certain bounds) whichpolicies they will retain and which they will cedeto the government. This arrangement is referred to

as the standard reinsurance agreement (SRA). TheSRA is quite complex, with both quota share rein-surance and stop losses by state and insurance pool;however, in essence, it allows the private insurancecompanies to adversely select against the govern-ment. This is considered necessary since the compa-nies do not establish premium rates or underwritingguidelines but are required to sell policies to alleligible farmers.

The federal costs associated with the U.S. pro-gram have four components:

• Federal premium subsidies range from 100 per-cent of total premium for catastrophic (CAT)policies to 38 percent of premium for buy-uppolicies at the highest coverage levels. Acrossall FCIP products and coverage levels, theaverage premium subsidy in 2004 was 59 per-cent of total premiums.

• The federal government reimburses adminis-trative and operating expenses for private in-surance companies that sell and service FCIPpolicies. This reimbursement is approximately22 percent of total premiums.

• The SRA has an embedded federal subsidywith an expected value of about 14 percentof total premiums.

• The program, by law, can be considered ac-tuarially sound at a loss ratio of 1.075. Thisimplies an additional federal subsidy of 7.5 percent of total premiums.

On average, the federal government pays approx-imately 70 percent of the total cost for the FCIP.Farmer-paid premiums account for only about 30 percent of the total cost. While the direct farmersubsidy varies by coverage level, the United Stateshas consistently passed legislation increasing thesubsidy level to farmers for crop and revenue insur-ance products. The rate of subsidy is one componentthat has influenced the growth in overall premium.Figure 3.1 clearly shows that the growth in premiumsubsidy is greater than the growth in farmer-paidpremiums. The rate of subsidy increased in 1995and 2001.

Canada14

In 2003, Canada revised its agricultural risk manage-ment programs. The “Business Risk Management”element of the new Agricultural Policy Framework(APF) is composed of two main schemes: ProductionInsurance and Income Stabilization.


Figure 3.1 Crop Insurance Premiums and Indemnities in the United States

0.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

2003200119991997199519931991

U.S

.bill

ion

dolla

rs

Crop Year

Producer-paid Premium subsidy

Source: Babcock et al. 2004.

Approaches to Agricultural Risk in Developed Countries 13

The Production Insurance (PI) scheme offersproducers a variety of multiple peril production orproduction value loss products similar to many ofthose sold in the United States. One major distinc-tion, however, is that the Canadian program is mar-keted, delivered, and serviced entirely and jointlyby federal and provincial government entities, although it is the provincial authorities who areultimately responsible for insurance provision. Thisallows provinces some leeway to tailor products tofit their regions and to offer additional products.

Production insurance plans are offered for overone hundred different crops, and provisions havebeen made to include plans covering livestock lossesas well. Crop insurance plans are available based oneither individual yields (or production value in thecase of certain items, such as stone-fruits) or areabased yields. Unlike the U.S. program, Canadianproducers are not allowed to separately insure dif-ferent parcels but rather must insure together allparcels of a given crop type. This means that lowyields on one parcel may be offset by high yields onanother parcel when determining whether or notan overall production loss has occurred. Insurancecan also be purchased for loss of quality, unseededacreage, replanting, spot loss, and emergency works.The latter coverage is a loss mitigation benefit meantto encourage producers to take actions that reducethe magnitude of crop damage caused by an in-sured peril.

Cost sharing between the federal governmentand each province for the entire insurance programis to be fixed at 60:40, respectively, by 2006. Federalsubsidies as a percentage of premium costs vary,however, from 60 percent for catastrophic losspolicies to 20 percent for low deductible produc-tion coverage. Combined, the federal and provin-cial governments cover approximately 66 percentof program costs, including administrative costs.This is roughly equivalent to the percentage of totalprogram costs borne by the federal governmentin the U.S. program. Provincial authorities are responsible for the solvency of their insurance port-folio. In Canada, the federal government competeswith private reinsurance firms in offering deficitfinancing agreements to provincial authorities.

Beginning in 2004, the Canadian AgriculturalIncome Stabilization (CAIS) scheme replaced andintegrated former income stabilization programs.CAIS is based on the producer production margin,where a margin is “allowable farm income,” includ-ing proceeds from production insurance minus“allowable (direct production) expenses.” The pro-

gram generates a payment when a producer’scurrent year production margin falls below thatproducer’s reference margin, which is based on anaverage of the program’s previous five-year mar-gins, less the highest and lowest. One importantfeature of CAIS is that producers must participatein the program with their own resources. In partic-ular, a producer is required to open a CAIS accountat a participating financial institution and depositan amount based on the level of protection chosen(coverage levels range from 70 percent to 100 per-cent of the “reference margin”). Once producersfile their income tax returns, the CAIS program ad-ministration uses the tax information to calculatethe producer’s program year production margin.If the program year margin has declined belowthe reference margin, some of the funds from theproducers’ CAIS accounts will be available forwithdrawal. Governments match the producers’withdrawals in different proportions for differentcoverage levels.

The total investment by federal and provincialgovernments for the “business risk management”programs is CAN$1.8 billion per year. In 2004, approximately CAN$600 million was provided bygovernments as insurance premium subsidies.

Spain

The Spanish agricultural insurance system isstructured around an established public/privatepartnership. On the public side is the NationalAgricultural Insurance Agency (ENESA), whichcoordinates the system and manages resources forsubsidizing insurance premiums, and the InsuranceCompensation Agency (Consorcio de Compensaciónde Seguros) that, together with private reinsurers,provides reinsurance for the agricultural insurancemarket. Local governments are involved only tothe extent that they are allowed to augment pre-mium subsidies offered at the national level. Onthe private side, insurance contracts are sold byAgroseguro, a coinsurance pool of companies thataggregates all insurance companies active in agri-culture. Farmers, insurers, and institutional rep-resentatives are all part of a general commissionhosted by ENESA that functions as the managingboard of the Spanish agricultural insurance system.

Similar to programs in the United States andCanada, Spain’s combined program offers insur-ance policies covering multiple perils. Policies areavailable for crops, livestock, and aquaculture activ-ities, with risks being pooled across the country by

Approaches to Agricultural Risk in Developed Countries 13

Agroseguro. Compared to the United States andCanada, however, farmers’ associations are moreactively involved in implementation and develop-ment of agricultural insurance. The governmenthas reserves to cover extreme losses, and, as a finalresort, the government treasury covers losses thatoccur beyond these reserves.

Total premiums for agriculture insurance poli-cies purchased reached around US$550 million( 490 million) in 2003, of which approximatelyUS$225 million ( 200 million) have been providedby the government (Burgaz 2004). The rationale be-hind subsidizing agricultural insurance is that thisoutlay serves as a disincentive for the governmentto also provide free ad hoc disaster assistance. Toreinforce the point, Spanish producers are ineligi-ble for disaster payments for perils for which in-surance is offered. For noncovered perils, ad hocdisaster payments are available, but only if the pro-ducer had already purchased agricultural insur-ance for covered perils.

WHY THE EXPERIENCE OFDEVELOPED COUNTRIES IS NOT A GOOD MODEL FORDEVELOPING COUNTRIESFor various reasons, developing countries shouldavoid adopting approaches to risk managementsimilar those adopted in developed countries.Clearly, developing countries have more limitedfiscal resources than do developed countries. Evenmore importantly, the opportunity cost of thoselimited fiscal resources may be significantly greaterthan in a developed country. Thus, it is critical for adeveloping country to consider carefully how muchrisk management support is appropriate and howto leverage limited government dollars to spur in-surance markets. In developed countries, govern-ment risk management programs are as muchabout income transfers as they are about risk man-agement. Developing countries cannot afford to

facilitate similar income transfers, given the largesegments of the population often engaged in farm-ing. Nonetheless, since a larger percentage of thepopulation in developing countries is typically in-volved in agricultural production or related in-dustries, catastrophic agricultural losses will havea much greater impact on GDP than may occur indeveloped countries.

Policymakers should also carefully consider thevarying structural characteristics of agriculture indifferent countries. In general, farms in developingcountries are significantly smaller than are farmsin countries like the United States and Canada. Fortraditional crop insurance products, smaller farmstypically imply higher administrative costs as a per-centage of total premiums. A portion of these costsare related to marketing and servicing (loss adjust-ment) insurance policies. Another portion is relatedto the lack of farm-level data and cost effectivemechanisms for controlling moral hazard.

Developing countries also have far less accessto global crop reinsurance markets than do devel-oped countries. Reinsurance contracts typicallyinvolve high transaction costs related to due dili-gence. Reinsurers must understand every aspect ofthe specific insurance products being reinsured (forexample, underwriting, contract design, rate mak-ing, and adverse selection and moral hazard con-trols). Some minimum volume of business, or theprospect for strong future business, must be presentto rationalize incurring these largely fixed transac-tion costs. For a global reinsurer to be willing toenter a market, the enabling environment must fos-ter confidence in contract enforcement and institu-tional regulations. An enabling environment is, infact, a prerequisite for effective and efficient insur-ance markets, and these components are largelymissing in developing countries. Setting rules assur-ing that premiums will be collected and that indem-nities will be paid is not a trivial undertaking. Thealternative risk management products discussed inChapter 5 are structured to overcome many of theseproblems.


15

INDEX INSURANCE ALTERNATIVES16

Given the problems with some traditional crop insurance programsin developed countries, finding new solutions to help mitigate sev-eral aspects of the problems outlined above has become critical. Indexinsurance products offer some potential in this regard (Skees et al.1999). These contingent claims contracts are less susceptible to someof the problems that plague multiple-peril farm-level crop insuranceproducts. With index insurance products, payments are based on anindependent measure highly correlated with farm-level yield orrevenue outcomes. Unlike traditional crop insurance that attemptsto measure individual farm yields or revenues, index insurancemakes use of variables exogenous to the individual policyholder—such as area-level yield or some objective weather event or measuresuch as temperature or rainfall—but have a strong correlation tofarm-level losses.

For most insurance products, a precondition for insurability is thatthe loss for each exposure unit be uncorrelated (Rejda 2001). For indexinsurance, a precondition is that risk be spatially correlated. Whenyield losses are spatially correlated, index insurance contracts can bean effective alternative to traditional farm-level crop insurance.

Index products also facilitate risk transfer into financial marketswhere investors acquire index contracts as another investment in adiversified portfolio. In fact, index contracts may offer significantdiversification benefits, since the returns generally should be un-correlated with returns from traditional debt and equity markets.

BASIC CHARACTERISTICS OF AN INDEXThe underlying index used for index insurance products must be cor-related with yield or revenue outcomes for farms across a large geo-graphic area. In addition, the index must satisfy a number of additionalproperties affecting the degree of confidence or trust that market par-ticipants have that the index is believable, reliable, and void of humanmanipulation; that is, the measurement risk for the index must be low(Ruck 1999). A suitable index required that the random variable mea-sured meet the following criteria:

• observable and easily measured;• objective;• transparent;

Innovation in ManagingProduction Risk

Index Insurance15

4

• independently verifiable;• reportable in a timely manner (Turvey 2002;

Ramamurtie 1999); and• stable and sustainable over time.

Publicly available measures of weather variablesgenerally satisfy these properties.17

For weather indexes, the units of measurementshould convey meaningful information about thestate of the weather variable during the contractperiod, and they are often shaped by the needsand conventions of market participants. Indexesare frequently cumulative measures of precipita-tion or temperature during a specified time. Insome applications, average precipitation or tem-perature measures are used instead of cumulativemeasures.

New innovations in technology, including theavailability of low-cost weather monitoring sta-tions that can be placed in many locations and sophisticated satellite imagery, will expand thenumber of areas in which weather variables canbe measured as well as of the types of measurablevariables. Measurement redundancy and auto-mated instrument calibration further increase thecredibility of an index.

The terminology used to describe features ofindex insurance contracts resembles that used for fu-tures and options contracts rather than for other in-surance contracts. Rather than referring to the pointat which payments begin as a trigger, for example,index contracts typically refer to it as a strike. Theyalso pay in increments called ticks.

Consider a contract being written to protectagainst deficient cumulative rainfall during a crop-ping season (for example, see Figure 4.1). The writerof the contract may choose to make a fixed paymentfor every one millimeter of rainfall below the strike.If an individual purchases a contract where thestrike is one hundred millimeters of rain and thelimit is fifty millimeters, the amount of payment foreach tick would be a function of how much liabil-ity is purchased. There are fifty ticks between theone hundred millimeter strike and fifty millimeterlimit. Thus, if $50,000 of liability were purchased,the payment for each one millimeter below onehundred millimeters would be equal to $50,000/(100 − 50), or $1,000.

Once the tick and the payment for each tick areknown, the indemnity payments are easy to calcu-late. A realized rainfall of ninety millimeters, for ex-ample, results in ten payment ticks of $1,000 each,for an indemnity payment of $10,000. Figure 4.1maps the payout structure for a hypothetical $50,000rainfall contract with a strike of one hundred mil-limeters and a limit of fifty millimeters.

In developed countries, index contracts that pro-tect against unfavorable weather events are nowsufficiently well developed that some standardizedcontracts are traded in exchange markets. Theseexchange-traded contracts are used primarily byfirms in the energy sector, although the range ofweather phenomena that might potentially be in-sured using index contracts appears to be limitedonly by imagination and the ability to parameterizethe event. A few examples include excess or defi-cient precipitation during different times of the year,insufficient or damaging wind, tropical weatherevents such as typhoons, various measures of airtemperature, measures of sea surface temperature,the El Niño southern oscillation (ENSO) tied to ElNiño and La Niña, and even celestial weather eventssuch as disruptive geomagnetic radiation from solarflare activity. Contracts are also designed for com-binations of weather events, such as snow and tem-perature (Dischel 2001; Ruck 1999). The potentialfor the use of index insurance products in agricul-ture is significant (Skees 2001).

A major challenge in designing an index insur-ance product is minimizing basis risk. Basis riskrefers to the potential mismatch between index-triggered payouts and actual losses. It occurs whenan insured has a loss and does not receive an in-surance payment sufficient to cover the loss (minusany deductible) or when an insured has a loss andreceives a payment that exceeds the amount of loss.


Figure 4.1 Payout Structure for a Hypothetical Rainfall Contract

$0

$10,000

$20,000

$30,000

$40,000

$50,000

$60,000

0 20 40 60 80 100 120

Rainfall in mm

Inde

mni

typa

ymen

t

Source: Skees 2003.

Innovation in Managing Production Risk 17

Since index-insurance indemnities are triggeredby exogenous random variables, such as area yieldsor weather events, an index-insurance policyholdercan experience a yield or revenue loss and not re-ceive an indemnity. The policyholder may also ex-perience no yield or revenue loss and still receivean indemnity. The effectiveness of index insuranceas a risk management tool depends on how posi-tively correlated farm yield losses are with theunderlying index. In general, the more homoge-neous the area, the lower the basis risk and the moreeffective area-yield insurance will be as a farm-levelrisk management tool. Similarly, the more closelya given weather index actually represents weatherevents on the farm, the more effective the indexwill be as a farm-level risk management tool.18

RELATIVE ADVANTAGES AND DISADVANTAGES OF INDEX INSURANCEIndex insurance can sometimes offer superior riskprotection compared to traditional farm-level,multiple- peril crop insurance. Deductibles, co-payments, or other partial payments for loss arecommonly used by farm-level, multiple-peril in-surance providers to mitigate asymmetric informa-tion problems such as adverse selection and moralhazard. Asymmetric information problems aremuch lower with index insurance because, first, aproducer has little more information than the in-surer regarding the index value, and second, indi-vidual producers are generally unable to influencethe index value. This characteristic of index insur-ance means that there is less need for deductiblesand copayments. Similarly, unlike traditional insur-ance, few restrictions need be placed on the amountof coverage an individual purchases. As long asthe individual farmer cannot influence the realizedvalue of the index, liability need not be restricted.An exception occurs when governments offer pre-mium subsidies as a percentage of total premiums.In this case, the government may want to restrictliability (and thus, premium) to limit the amountof subsidy paid to a given policyholder.

As more sophisticated systems (such as satelliteimagery) are developed to measure events causingwidespread losses, indexing major events shouldbecome straightforward and quite acceptable tointernational capital markets. Under these condi-tions, traditional reinsurers and primary providersmay begin offering insurance in countries they

would never previously have considered. New riskmanagement opportunities can develop if rele-vant, reliable, and trustworthy indexes can be con-structed. A detailed technical overview of indexinsurance is presented in Appendix 1. Key advan-tages and challenges are summarized in Table 4.1.

THE TRADE-OFF BETWEEN BASISRISK AND TRANSACTION COSTSAmong the most significant issues for any insur-ance product is the question of how much moni-toring and administration is needed to keep moralhazard and adverse selection to a minimum. Toaccomplish this goal, coinsurance and deductiblesare used so that the insured shares the risk and anymistakes in offering too generous coverage aremitigated. Considerable information is needed totailor insurance products and to minimize the basisrisk even for individual insurance contracts. In-creased information gathering and monitoring involve higher transaction costs, which convertdirectly into the higher premiums needed to coverthem. Index insurance significantly reduces thesetransaction costs and can be written with lowerdeductibles and without introducing coinsurance.When farm yields are highly correlated with theindex being used to provide insurance, offeringhigher levels of protection can result in risk trans-fer superior even to individual multiple-peril cropinsurance (Barnett et al. 2005).

The direct trade-off between basis risk and trans-action costs has implications for achieving sustain-able product designs and for outlining the role ofgovernments and markets. Chapter 5 introducesthe idea of layering risk. These concepts also greatlydepend on understanding the trade-off betweenbasis risk and transaction costs. At every level ofrisk transfer, someone must accept a certain degreeof basis risk if the products are to be both sustain-able and affordable. In short, extremely high trans-action costs must be paid for. The social cost ofhaving products with some basis risk may be signif-icantly lower than the social cost associated withthe high transaction cost entailed in attempting todesign products that have no basis risk.

WHERE INDEX INSURANCE IS INAPPROPRIATEIndex insurance contracts will not work well forall agricultural producers. Many agricultural com-

modities are grown in microclimates. Coffee growson certain mountainsides in various continents andcountries, for example, and fruits such as apples andcherries also commonly grow in areas with verylarge differences in weather patterns within only afew miles. In highly spatially heterogeneous pro-duction areas, basis risk will likely be so high as tomake index insurance problematic. Under theseconditions, index insurance will work only if it ishighly localized19 and/or can be written to protectonly against the most extreme loss events. Even inthese cases, it may be critical to tie index insuranceto lending, since loans are one method of mitigat-ing basis risk.

Overfitting the data is another concern withindex insurance. If one has a limited amount ofcrop yield data, fitting the statistical relationshipbetween the index and that limited data can becomeproblematic. Small sample sizes and fitting regres-sions within the sample can lead to complex contractdesigns that may or may not be effective hedgingmechanisms for individual farmers. Standard pro-cedures that assume linear relationships betweenthe index and realized farm-level losses may beinappropriate. While scientists are tempted to fitcomplex relationships to crop patterns, interviewswith farmers may reveal more about the types ofweather events of most concern. When designing


Table 4.1 Advantages and Disadvantages of Index Insurance

Advantages Challenges

Basis riskWithout sufficient correlation between the index and actuallosses, index insurance is not an effective risk managementtool. This is mitigated by self-insurance of smaller basis riskby the farmer; supplemental products underwritten by privateinsurers; blending index insurance and rural finance; andoffering coverage only for extreme events.

Precise actuarial modelingInsurers must understand the statistical properties of theunderlying index.

EducationUsers must be able to assess whether index insurance willprovide effective risk management.

Market sizeThe market is still in its infancy in developing countries andhas some start-up costs.

Weather cyclesActuarial soundness of the premium could be underminedby weather cycles that change the probability of the insuredevents, such as El Niño, for example.

MicroclimatesThese production conditions make rainfall or area-yield indexbased contracts difficult for frequent and localized events.

ForecastsAsymmetric information about the likelihood of an eventin the near future creates the potential for intertemporaladverse selection.

Source: Authors.

Less moral hazardThe indemnity does not depend on the individual producer’srealized yield.

Less adverse selectionThe indemnity is based on widely available information, sothere are few informational asymmetries to be exploited.

Lower administrative costsUnderwriting and inspections of individual farms are notrequired.

Standardized and transparent structureContracts can be uniformly structured.

Availability and negotiabilityStandardized and transparent, the contracts may be tradedin secondary markets.

Reinsurance functionIndex insurance can be used to transfer the risk of wide-spread correlated agricultural production losses more easily.

VersatilityIndex contracts can be easily bundled with other financialservices, facilitating basis risk management.

Innovation in Managing Production Risk 19

a weather index contract, one may be tempted tofocus on the relationship between weather eventsand a single crop. When it fails to rain for an ex-tended period of time, however, many crops willbe adversely affected. Likewise, when it rains foran extended period of time, resulting in significantcloud cover during critical photosynthesis periods,a number of crops may suffer.

Finally, when designing index insurance con-tracts, significant care must be taken to assure thatthe insured has no better information about the like-lihood and magnitude of loss than does the insurer.Farmers’ weather forecasts are quite often highly ac-curate. Potato farmers in Peru, using celestial obser-vations and other indicators in nature, are able toforecast El Niño at least as well as many climate ex-perts (Orlove et al. 2002). In 1988, an insurer offereddrought insurance in the U.S. Midwest. As the salesclosing date neared, the company noted that farm-ers were significantly increasing their purchases ofthese contracts. Rather than recognize that thesefarmers had already made a conditional forecast thatthe summer was going to be very dry, the company

extended the sales closing date and sold even morerainfall insurance contracts. The company experi-enced very high losses and was unable to meet thefull commitment of the contracts. Rainfall insurancefor agriculture in the United States suffered a sig-nificant setback. The lesson learned is that whenwriting insurance based on weather events, it iscrucial to be diligent in following and understand-ing weather forecasts and any relevant informa-tion available to farmers. Farmers have a vestedinterest in understanding the weather and climate.Insurance providers who venture into weatherindex insurance must know at least as much asfarmers do about conditional weather forecasts. Ifnot, intertemporal adverse selection will render theindex insurance product unsustainable. These issuescan be addressed; typically, the sales closing datemust be established in advance of any potential fore-casting information that would change the proba-bility of a loss beyond the norm. But beyond simplysetting a sales closing date, the insurance providermust have the discipline and the systems in placeto ensure that no policies are sold beyond that date.

21

ROLE OF GOVERNMENTShould the lack of effective private-sector agricultural insurance mar-kets in developing countries be addressed through government inter-vention? High transactions costs preclude emergence of many markets,but this does not necessarily justify government intervention.

In the case of high-frequency, low-consequence losses, govern-ment intervention is likely to distort incentives and create rent-seeking opportunities, possibly to an extent that actually reduces netsocial welfare. Farmers can employ other risk management mech-anisms to cover these losses. In fact, insurance products for high-frequency, low-consequence losses are seldom offered because thetransaction costs associated with loss adjustment renders the insur-ance cost prohibitive for most potential purchasers.

Governments may have no inherent advantage over markets intrying to facilitate the provision of individual farm-level yield or rev-enue insurance products. The private sector typically does not providethese insurance products in part because of information asymmetriesthat cause moral hazard and adverse selection problems (Mirandaand Glauber 1997); it is difficult to see how a government providerwould have any advantage in addressing this problem.

In the case of low-frequency, high-consequence loss events, how-ever, government intervention may be justified. As explained in thesection on production/weather risk management, research suggeststhat many decision makers tend to underestimate their exposure tolow-frequency, high-consequence losses, a tendency reinforced whenthe decision maker believes the government will provide assistance inthe event of a disaster. Thus, producers thinking in this way will be un-willing to pay the full costs of insurance products that protect againstthese losses. Those who do buy insurance against low-frequency, high-consequence losses often cancel the policy if they do not receive anindemnity for an extended period. Thus, it seems that to be success-ful agricultural insurance products must be constructed so that theymake indemnity payments with reasonable frequency, for example,once every seven or ten years.

On the supply side, insurers will typically load premium ratesheavily for low-frequency, high-consequence loss events where con-siderable ambiguity surrounds the actual likelihood of the event.Together, these effects create a gap between the prices farmers willpay for catastrophic agricultural insurance and the prices insurers willaccept. Thus functioning private sector markets fail to materialize, or,

New Approaches toAgricultural Risk

Management inDeveloping Countries

5

if they do materialize, they cover only a small por-tion of the overall risk exposure. This type of mar-ket failure is commonly cited as justification forgovernment interventions to facilitate provision ofproducts or services not otherwise provided (orprovided in sufficient quantity) by private markets.

Subsidies for catastrophic reinsurance (see Box5.1) are a type of government intervention thatcan facilitate the provision of insurance for low-frequency, high-consequence loss events. Hardaker,et al. (2004), provide the following arguments forsuch an approach:

1. Governments already provide disaster re-lief; providing assistance through reinsur-ance might be more efficient.

2. The financial involvement of a governmentmay address a moral hazard problem in itsbehavior: many catastrophes can either beprevented or magnified by government poli-cies or lack thereof. Government financial re-

sponsible for some losses might be an incen-tive for putting in place appropriate hazardmanagement and mitigation measures.

3. A government’s financial involvement in rein-surance may reduce political pressure to pro-vide distorting and often capricious ad hocdisaster relief.

4. Governments can potentially provide reinsur-ance more economically than can commer-cial reinsurers. A government’s advantages,including its deep credit capacity and uniqueposition as the country’s largest entity, enableit to spread risks more broadly.

If governments are to intervene in agricultural in-surance markets, the social benefits of reducing theinefficiencies brought on by risk must outweigh thesocial cost of making agricultural insurance work.This chapter presents a framework for governmentagricultural risk policy formulation that focuseson policy objectives, constraints on government


Box 5.1 Reinsurance

Reinsurance is insurance for insurers. Just like insurance,reinsurance is “fundamentally the promise to paypossible future claims against a premium today.”Insurers often hold undiversifiable or extreme risk intheir portfolios, and since they do not wish to retainall of it, they transfer some risk to reinsurance com-panies, paying the reinsurers a premium to do so.Reinsurers also advise insurers on product developmentand more complex risk-taking.

Reinsurance agreements can be proportional ornonproportional. With proportional agreements, insurers and reinsurers divide premiums and losses ina contractually defined proportion; with nonpropor-tional agreements, the insurer usually pays all lossesup to a defined amount and the reinsurer indemnifiesfor losses above that limit. Quota-share and surplusreinsurance are examples of proportional reinsuranceagreements. Excess-of-loss and stop-loss agreementsare examples of nonproportional reinsurance.

Reinsurers seek to operate across boundaries inorder to build globally diversified portfolios. Morethan 250 reinsurers in 50 countries wrote annual re-insurance premiums of approximately US$176 billionin 2003.a Nonlife reinsurance premiums accountedfor US$146 billion, or about 14 percent, of the globalnonlife primary insurance industry. Only US$25 billionof these premiums are written outside North Americaand Western Europe.b The ten largest reinsurers writeabout 54 percent of reinsurance premiums, and thetwo giants in the business, Munich RE and Swiss RE,write around US$49 billion of reinsurancepremiums.c

Securitization, an alternative to traditional reinsur-ance, transfers catastrophic risks to capital markets inthe form of financial securities. Securitization hasbeen used for exposure to natural catastrophes, suchas earthquakes and hurricanes.

Notes:a. Standard & Poor’s Global Reinsurance Highlights, 2004 Edition.b. Latin America: $US4.7 billion; Asia: $US13.8 billion; rest of the world: $US6.7 billion. For comparison, the World Bank disburses approximately $US0.5 billion per year in emergency assistance grants and loans to developing countries.c. This premium volume includes life and health reinsurance premiums.

Source: Swiss Re 2004.

New Approaches to Agricultural Risk Management in Developing Countries 23

action, risk principles, and potential policy instru-ments (Figure 5.1). The framework is then used toconsider alternative models for government inter-vention in agricultural insurance markets.

POLICY OBJECTIVESGovernments that seek to spur growth and eradi-cate poverty almost inevitably mix economic poli-cies meant to enhance efficiency and growth withsocial policies meant to address poverty and vul-nerability. Governments also often pursue equityor income redistribution objectives. Thus, govern-ment policies related to agriculture and rural areastend to pursue the following objectives:

• Growth. Economic growth in ruralareas—in particular higher agricul-tural yields and value-added process-ing as well as development of off-farmactivities—is perceived to be the bestway out of poverty in the mediumterm. While better incentives formarket players and an enabling in-frastructure are key drivers, bettermanagement of agricultural produc-tion risk is also critical for growth, asit enhances access to credit and adop-tion of new technologies.20

• Reduction of poverty and vulnerabilityin rural areas. To achieve social andequity goals, governments directlyintervene in a targeted manner, be-cause free markets do not necessarilyalleviate poverty for those in societywho cannot effectively participate inthem. Safety nets provide one tool forsuch government intervention.21

Given limited resources in developingcountries and the existence of other sectorsrequiring government attention, these objectives are typically pursued within anenvironment of binding fiscal constraints.The two objectives target different seg-ments of the rural population and differ-ent risk profiles. Growth objectives focuson increasing profitability so that less poorfarmers can continue adopting productiontechnologies even when high-frequency,low-consequence loss events occur. Povertyreduction policies seek to increase the aver-age income of poor farmers, thus decreas-

ing the volatility of their income and the likelihoodthat a risk event will wipe out hard-won asset gains.

A precondition for achieving sustainable growthand poverty reduction is an ex ante system for dis-aster risk management. Disaster risk managementcovers severe and very infrequent events affectingmostly the poor, because the poor are more vul-nerable and tend to live in marginal and more risk-exposed areas. Susceptibility to and the experienceof major natural disasters tend to trap people inpoverty, due to the lack of efficient risk manage-ment at the household level.22 Government disasterrisk policies often entail some form of monetarycompensation for victims of natural disaster. The

Figure 5.1 Framework for Governmental Agricultural RiskManagement Policy Formulation

Source: Authors.

Objectives

Agricultural and rural economic growth

Poverty reduction

n

n

Constraints

Underdeveloped financial sector

Disaster risk

Agricultural sector dominated by small farms

Government fiscal limitations

Underdeveloped regulatory framework

n

n

n

n

n

Principles

Segment independent versus correlated risk

Minimize rent seeking that creates market distortions

Diversification of risk — risk management — risk layering

Risk transfer cost optimization — reduce transaction costs

n

n

n

n

Policy Instruments

Mechanisms for transferring catastrophic risk layers

Limited government subsidies

Contingent funding for disaster relief and enhanced social safety nets

n

n

n

challenge is to deliver timely and predictable aid indisaster situations. This requires ex ante planningrather than just ex post disaster responses. This alsoimplies efforts to forestall political demands forex post, ad hoc government disaster assistance.Indeed, a credible and reliable disaster risk man-agement system can put farmers and countries ona higher growth path by making people more com-fortable with taking calculated and protected risks.

Naturally growth and poverty-reduction objec-tives overlap, but this makes it even more importantto identify clear objectives and to design effectiveand cost-efficient ways to achieve them. Mixingobjectives can lead to suboptimal outcomes. Manygovernment-facilitated crop insurance programs,for example, attempt to accomplish social welfareand economic efficiency objectives simultaneously.

CONSTRAINTS IN AGRICULTURALRISK MANAGEMENTWhen making decisions about agricultural riskmanagement programs, policymakers face a num-ber of constraints. They must consider whether thebenefits of such programs outweigh the costs andif the benefits from putting resources into risk man-agement programs are greater than the benefits of using these resources for other social needs.Governments must construct risk managementprograms that minimize distortions in resourceallocation and reduce opportunities for rent-seekingbehavior. They must take into consideration thestatus and development of financial and insuranceinstitutions within the country, any regulatory con-straints on the operations of those institutions, andthe infrastructure for enforcing contracts. Finally,policymakers must consider the dichotomy, presentin many countries, between smallholder farms andlarge farms producing for export markets.

Cost-Benefit Analyses of Agricultural RiskManagement Projects

Traditional economic analyses of projects (or othersector interventions) weigh social benefits againstsocial costs, usually in monetary terms. In theory,this procedure should make it possible to comparethe net benefits from these projects with the netbenefit of a government risk management program.Conducting such a comparison is not a trivial exer-cise, however, because numerous assumptions, not

always robust across different projects, are requiredto quantify risk management benefits. Still, it isworthwhile to compare the net benefits of govern-ment risk management programs with the net ben-efits from other projects, if only to get a sense of theorders of magnitude involved.

Fiscal Constraints

Government expenses for agricultural insuranceprograms can be quite high, a reality often maskedby how the actuarial performance is presented.Governments typically report loss ratios, or costto premium ratios, as indemnities paid divided bytotal premiums collected. This method presents twoproblems: first, due to government premium sub-sidies, farmers pay only a fraction of the total pre-mium; second, governments typically absorb mostadministrative and operating costs. When calculat-ing loss ratios for private sector insurance products,administrative costs are included in the numerator.When considering only indemnity relative to pre-miums (without noting that significant portions ofpremiums are paid by the public sector), both theU.S. and Canadian crop insurance programs have,in recent years, reported loss ratios around 1.0.These loss ratios are then cited as evidence that theprograms are actuarially sound. But when admin-istrative and operating costs are added to the nu-merator and government premium subsidies aresubtracted from the denominator, so that the lossratio is equivalent to the standard used for pri-vate sector insurance products, crop insuranceloss ratios are about 3.6 for the United States and2.9 for Canada.23 Hazell (1992) estimates similarratios for a number of government-based crop in-surance programs. His estimates for programs inthe Philippines, Japan, and Brazil, for example,show loss ratios (as defined in the private sector)exceeding 4.0.

Policymakers often suggest agricultural insur-ance programs as alternatives to free ex post dis-aster assistance. In principle, insurance programshave many advantages over ex post disaster as-sistance. Disaster assistance programs, it is oftenargued, for example, can generate perverse in-centives that increase the magnitude of losses insubsequent disaster events (Barnett 1999; Rossi et al.1982). But, in practice, agricultural insurance pro-grams have often evolved into alternate vehicles fortransferring wealth from the public sector to agri-cultural producers. Furthermore, not much evidence



indicates that agricultural insurance programs havebeen successful in forestalling free ex post govern-ment disaster assistance. In the United States, forexample, more and more costly crop insurance pro-grams have coexisted with disaster payments forwell over twenty years (Glauber 2004).

Operational Constraints: MinimizeDistortions/Rent-Seeking Opportunities

Governments should only invest public resourcesin developing agricultural insurance if the socialcosts of the inefficiencies resulting from the lack ofsuch insurance products outweigh the social costsof government intervention. Social costs includenot only the opportunity costs of public resourcesused to create and maintain the agricultural insur-ance products but also any resource allocation dis-tortions that result when farmers and rural decisionmakers respond to the incentives created by theinsurance products. This can include rent-seekingand regressive effects that benefit mostly large com-mercial farmers.

Contract Enforcement

Contract enforcement is critical to achieving effec-tive and sustainable risk management programs.It is very difficult to develop insurance contracts ifthe legal and regulatory environment does not existfor contract enforcement. Purchasers will lose trustin the program if indemnity payments are not madeon a timely basis or if they are frequently tied up inlengthy legal procedures.24 Likewise, insurers willlose trust in the program if they are forced to payindemnities for losses that the contract was not in-tended to cover.

Level of Financial Sector Development

Complex agricultural insurance programs are un-likely to be sustainable unless they are accompaniedby adequate insurance capital and expertise. Indeveloping countries, insurance sectors are oftenunderdeveloped and concentrated in very few linesof business, for example, automobile, property, andcasualty insurance. Insurance companies in devel-oping countries also tend to be based in urban areasand to shy away from doing business in rural areas,where the insurance market is characterized byhigh transaction costs and small policies.

New products will be required if agriculturalinsurance is to take root in countries with under-developed traditional insurance sectors. Insuranceproducts based on an index recognized and acceptedby international reinsurers, for example, can provideopportunities to bypass in-country insurance capac-ity constraints. If the reinsurer accepts the indexdata and settlement procedures, the insurer’s capi-tal becomes somewhat less relevant than for tradi-tional lines of insurance; this is because the reinsureris not really accepting the insurer’s underwriting riskbut only the risk inherent in the index. Experiencewith reinsurance for weather index contracts re-veals that reinsurers may even be willing to take100 percent of the risk. For operational and regu-latory reasons, however, international reinsurersprefer to deal with professionally-run companiesto source the risk.

Structure of Agricultural Sectors

Agricultural dominated by smallholders imposesclear constraints on the large scale roll-out of sophis-ticated crop insurance programs or, indeed, of anyagricultural risk management scheme. Farmerswith one hectare of land or less will never offer an attractive marketing target for insurance compa-nies. The challenge is to identify suitable aggre-gators of risk, such as microfinance institutions,banks or cooperatives, or even local authorities whocan enroll farmers in group insurance programs.Agricultural sectors need to be segmented, withdistribution channels tailor-made to specific needsand local customs.25

Regulatory Constraints

Agricultural risk transfer involves financial con-tracts that are regulated according to prudentialprinciples. Insurance companies must organize thefinancing to pay for the possibility of the worst casescenario. This constrains the type and sophistica-tion of contracts, which may also be constrained bylimitations in the regulator’s ability to understandand supervise new products.

RISK PRINCIPLESLayering and the Role of Index Insurance

Segmenting risk into different “layers” is a key riskmanagement principle. Consider, for example,Figure 5.2, which shows the probability distribution

for average April to October rainfall at thirteenweather stations in Malawi.26 Suppose that farmersstart incurring production losses whenever rainfallis less than one thousand millimeters. The domainof losses might be segregated into three risk layers,with different entities holding each layer:

• For rainfall in excess of seven hundred mil-limeters, farmers would retain the loss risk,either individually or with financial serviceproviders: the risk retention layer.

• For rainfall between five hundred and sevenhundred millimeters, the risk would be trans-ferred to an insurance company via a weatherindex insurance product: the market insur-ance layer.

• For rainfall levels below five hundred milli-meters, the risk in this example would not beinsured due to cognitive failure and ambigu-ity loading: the market failure layer.27

Farmers would absorb losses in the risk retentionlayer using self-insurance strategies such as thosedescribed in Chapter 2. Strategies for effectivelytransferring the other risk layers are describedbelow.

ADDRESSING THE MARKETINSURANCE RISK LAYERReferring again to Figure 5.2, suppose that an insur-ance provider writes a rainfall index insurance con-tract with a strike of seven hundred millimeters anda limit of five hundred millimeters. Limits are com-monly used by weather index insurance writers toavoid open-ended exposure to catastrophic weatherevents. The insured would select the amount ofinsurance (the liability) and the payment per tickwould be calculated using this formula.

Assume that a farmer has a crop with an expectedvalue of $15,000. At only five hundred millimetersof rainfall, the farmer is estimated to lose two-thirdsof the value of the crop. Thus, the farmer purchases$10,000 of liability, with a payment for each tick(each millimeter of rainfall) of fifty ($10,000 dividedby (700 − 500)). If the realized value for the rainfallindex is six hundred millimeters, for example, theindemnity will be $5,000 ((700 − 600) × $50).28

Payment Per TickLiability

Limit Strike� =

−


Figure 5.2 Average April to October Rainfall for Thirteen Malawi Weather Stations

2500200015001000500 7000

April-October rainfall (mm)

Limit Strike

Source: Authors.

X ≤ 505 1.0%

X ≤ 1776 99.0%


The limit of five hundred millimeters caps theinsurance provider’s loss exposure on the indexinsurance product. Without the limit, the contractwould be extremely expensive, since it would pro-tect against losses in the extreme lower tail of theprobability distribution. Buyers would exhibit cog-nitive failure regarding the probability of eventswith less than five hundred millimeters of rainfall,while insurance providers would load the premiumfor ambiguity regarding these same events. Thus,even if insurance was available to protect againstrainfall events of less than five hundred millimeters,few transactions would be likely, since the premiumwould exceed most buyer’s willingness to pay.

Spatial Correlation of Risk

Weather events that cause agricultural losses areoften highly spatially correlated. In the presence ofsuch spatial correlation, index insurance products,such as the rainfall index insurance described above,can be effective risk transfer mechanisms. Once therisk is transferred from the farmer to a local insur-ance provider, however, spatial correlation makes itvery difficult for the local insurance provider to gen-erate much risk reduction through pooling. Unlesssome mechanism exists for transferring the spatiallycorrelated loss risk out of the region or country, localinsurance providers will be reluctant to offer insur-ance products, even if those products protect onlyagainst losses in the market insurance layer.

Risk Transfer Strategies

At least three strategies exist for transferring riskfrom index insurance contracts: (1) direct transferof contracts into reinsurance markets; (2) packagedtransfer of independent contracts; and (3) poolingof risk and subsequent transfer of the pool tail risk.(See Table 5.1.) Under the first two strategies, nobasis risk occurs, insofar as every single contract isreinsured against payouts that exceed a definedlevel. Since no pooling occurs prior to the risk trans-fer, however, direct and packaged risk transferstrategies will likely have higher reinsurance pre-mium rates than will the transfer of pooled risks,even if the reinsurer offers portfolio-adjusted pric-ing. Under the third strategy of pooling risk priorto transfer, insurers could be exposed to some basisrisk, insofar as a pool of indexes does not perfectlyreflect the payout likelihood of each individual con-tract, and only the excess risk of the overall pool

would be reinsured. If there are opportunities to di-versify risks within the pool, however, this strategycould lead to lower reinsurance premiums relativeto either of the other two strategies, since the risk ofthe overall pool (rather than each individual con-tract) would be reinsured. The first strategy doesnot involve the government in the transfer of risk.The other two strategies may involve government,either in facilitating risk transfer (for the secondstrategy) or in pooling risk and facilitating risktransfer (for the third strategy).

Pooling of Risk

The third risk transfer strategy identified above in-volves pooling risks within the country or region.Risk pooling is based on the statistical law of largenumbers, which states that the more uncorrelatedrisks are added to a portfolio, the lower the vari-ance in the outcomes of the overall portfolio. For aninsurer, this results in lower capital needs and,therefore, lower capital costs.

Index-based insurance contracts can be pooledand transferred in a number of ways. In one method,the reinsurance contract can be based on a basketindex that is a weighted average of the indexes con-tained in the pool. A risk management programbeing considered for Malawi would have private in-surers sell rainfall-based index insurance contractsfor various weather stations around the country.The government would purchase reinsurance pro-tection and sell it to the insurers. For reinsurancecoverage, the government could use the MalawiMaize Production Index (MMPI), a weighted aver-age of weather station indexes with each station’scontribution weighted by the corresponding ex-pected maize production from that location. Themore highly spatially correlated the risks on theunderlying indexes, the better the basket indexwill perform as a reinsurance mechanism (that is,the lower the reinsurance basis risk). But, of course,the more highly spatially correlated the risks on theunderlying indexes, the less advantage there is topooling within the country as opposed to simplytransferring the underlying weather station indexesto the reinsurance market using either of the firsttwo strategies identified above.

A pool of index insurance risks can also be trans-ferred using traditional stop-loss reinsurance. In thiscase, in exchange for a reinsurance premium, thereinsurer would simply cover all losses in excess ofa predefined percentage (for example, 110 percent)

of the total premium dollars in the pool. With thistype of reinsurance (and unlike reinsurance basedon a basket index), the pool would not be exposedto basis risk. The transactions costs for the reinsurerwill be much higher compared to the basket indexbased reinsurance, however, since the reinsurerwill need to conduct due diligence on not only theunderlying indexes but also the underwriting ofthe pool. All other things being equal, higher trans-actions costs will lead reinsurers to charge higherreinsurance premiums. Despite this, if spatial di-versification opportunities are sufficiently high,pooling may reduce risk exposure to such an extentthat reinsurance premium costs are reduced.

This concept can be extended to the pooling ofmulticountry risks within a region. Weather riskcan be retained and managed internally if the areasunder management are significantly diverse in theirweather risk characteristics. This immediately sug-gests that the weather sensitivity of neighboringcountries must be taken into account when consid-ering a country’s weather-risk profile and its needfor outside reinsurance. Consider the example ofthe region of the Southern African DevelopmentCommunity (SADC; Figure 5.3). Analysis showsthat, on average, two countries in the region suffera drought each year. The distribution of droughtevents in SADC is extremely long-tailed, however,


Table 5.1 Risk Transfer Strategies

Strategy Advantages/Disadvantages Role of Government

No basis risk. Pooling occurs at rein-surer level. If spatial diversification op-portunities exist, reinsurance premiumrates will likely be higher than if riskswere pooled at insurer level (even if thereinsurer offers portfolio adjusted rein-surance premiums). Reinsurer will needto perform extensive due diligence onindex but little due diligence on insurer.

Same as above, only may pay lowerreinsurance premium rates becausebundling reduces transactions costs forthe reinsurer.

Some basis risk. If spatial diversifica-tion opportunities exist, reinsurancepremium rates will be lower than withother strategies. In the case of poolreinsurance based on traditional stop-loss coveragea transactions costs maybe higher, since the reinsurer will needto perform due diligence not only onthe index but also on the pool. In caseof reinsurance based on index insur-ance, pool due diligence is avoided,but basis risk would be higher.b

Government is not involved in facilitat-ing risk transfer.

Either government or an association ofinsurers can facilitate the bundling andtransfer of contracts to the reinsurancemarket.

Either government or an association of insurers can facilitate the risk poolingand transfer of pool tail risk to the rein-surance market.

Notes:

a. For the agricultural insurance pool proposed by the Mongolian project of the World Bank, see the case study in Chapter 6.

b. See the Agroasemex case in Appendix 2.

Source: Authors.

Direct risk transferContracts are transferred directly frominsurers to reinsurers.

Packaged risk transferContracts are bundled among compa-nies and transferred to one (syndicate)of reinsurers.

Pooling and transferContracts are pooled within the countryand/or region with only the tail risk ofthe pool transferred to reinsurers.


with the possibility of widespread drought eventsthat could potentially devastate the region.

A SADC pool of rainfall-based index insurancecontracts could be constructed, with each membercountry being charged an actuarially fair assess-ment of the risk transferred to the pool. Supposethe financial impact to the pool of four SADC coun-tries experiencing simultaneous droughts is aboutUS$80 million. The pool may wish to transfer therisk of losses beyond US$80 million to the inter-national reinsurance market. This could be donein layers with, for example, one layer of US$80 to350 million being transferred using reinsurancemechanisms.29 Losses in excess of US$350 million,as might occur with simultaneous droughts in tenSADC countries, occur with a frequency of about1 percent. Instruments such as catastrophe (CAT)bonds might be used to transfer this extreme layer.CAT bonds allow the transference of very largeexposures into financial markets and often havetenures of up to three years.

More efficient means of transferring risk implythat costs could be greatly reduced for the member

countries by transferring risk as part of a regionalstrategy rather than by transferring the risk onecountry at a time. The SADC pooling approachabove, for example, would reduce insurance costsby 22 percent for one of the countries, Malawi, dueto risk-pooling effects (Hess and Syroka 2005).Managing a pool requires a high degree of under-writing and actuarial sophistication, however.Reinsurers will conduct due diligence and will bevery reluctant to write traditional excess of lossreinsurance unless they are convinced that the poolis being managed appropriately.

MARKET FAILURE LAYERAt the catastrophic loss layer represented by marketfailure, private decision makers will likely not pur-chase adequate insurance due to cognitive failure,ambiguity loading of premiums rates, and perhaps,expectations of government or donor disaster re-lief. Some form of government intervention may berequired to facilitate adequate transfer of the risk inthis layer.

Figure 5.3 Histogram of Simulated SADC Drought Events

Source: Hess and Syroka 2005.

Freq

uenc

y(o

utof

5677

drou

ghte

vent

s)

4000

3500

3000

2500

2000

1500

1000

500

0

X ≤ $80m 69.0%

X ≤ $350m 95.5%

X ≤ $1.1b 100%

Fund Reinsurance Securitization

0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5 0.55 0.6 0.65 0.7 0.75 0.8 0.85 0.9 0.95 1 1.05 1.1 MoreSADC drought-risk exposure ($ billion)

POLICY INSTRUMENTSRisk layering provides an extremely helpful con-ceptual framework for thinking about governmentintervention in risk transfer markets. The discus-sion of the market insurance layer described situa-tions in which government packaging or pooling ofrisk could potentially reduce the transaction costsassociated with risk transfer and thus the premi-ums paid by end users. This section explores otherpossible government interventions, including gov-ernment facilitation of risk transfer in the marketfailure layer, the role of government subsidies inrisk transfer markets, and potential uses of indexinsurance instruments to finance government dis-aster relief and safety net policies.

Government Disaster Option for CAT Risk:A Policy for the Market Failure Layer30

Cognitive failure and ambiguity loading occur pri-marily with events in the extreme tail of the lossdistribution, the area previously termed the marketfailure layer. For this reason, and as a substitute forad hoc disaster relief payments, governments maydecide to cofinance risk transfer mechanisms forthese events. A government, for example, could de-

sign Disaster Option for CAT risk (DOC) index rein-surance contracts for catastrophic risks. Returningto the example in Figure 5.2, a DOC could insureagainst rainfall less than five hundred millimeterswith a payment per tick of say, $50. Primary insur-ers could then offer coverage beyond the earlierimposed limit of five hundred millimeters andtransfer the catastrophic tail risk to the governmentusing the DOC. Even if primary insurers are sellingtraditional crop insurance, they could use a DOCto transfer part of the catastrophic tail risk in theirportfolio of crop insurance policies.31 DOCs couldbe offered for a variety of strikes and settlementweather stations, as long as the coverage is for cata-strophic risk layers and can be offset in internationalweather risk markets. The government could evenoffer other DOC indexes (for example, excess rain-fall or wind speed) to reinsure other lines of insur-ance, such as property and casualty (see Figure 5.4).

The government would reinsure DOCs in inter-national reinsurance or capital markets using anyof the three risk transfer strategies described ear-lier.32 Since DOCs would address only extremecatastrophic loss events, reinsurance premium rateswould likely contain an ambiguity load. Premiumscould be subsidized to offset part of this ambiguityload so that DOC purchasers would pay something


Figure 5.4 Government-Sponsored DOC as Risk Transfer Product between National and International Risk Markets

DOC�(market failure

layer)

Market insurance

layer layer

2500200015001000500 7000

April-October rainfall (mm)

Source: Authors.

Risk retention

layer


closer to a pure premium rate.33 DOCs could betailor-made to individual insurers’ needs; for exam-ple, DOCs could be based on individual weatherstations or written as regional weighted averagebaskets of weather stations. Strikes should be setso that the DOC covers only infrequent events (forexample, events with an expected frequency ofonce every thirty years or less). This is the domainof the probability distribution over which poten-tial insurance purchasers tend to experience cog-nitive failure and insurance providers engage inambiguity loading. Primary insurers and ultimatelyinsured parties would pay a premium for this cata-strophic protection, but it would be significantly lessthan what the market would charge.

Those who reinsure DOC contracts will insist onverifying the credibility of the underlying indexes.The premium required to transfer the risk to inter-national markets would provide a baseline for set-ting DOC premium rates.

The risk-layering approach proposed here wouldinstitutionalize the social role of government in sub-sidizing extreme risk events at the local level. Pre-mium rates could be subsidized to offset ambiguityloading. Furthermore, by organizing DOC contractsat the local level, victims of isolated severe eventsthat fail to capture national policymakers’ attentioncould still receive some structured assistance.

The following list summarizes the major advan-tages of offering index-based DOCs:

• DOC contract provisions established ex anteallow for better planning than do ad hoc dis-aster payments.

• DOCs provide a structure that provides morespatial and temporal equity in governmentdisaster assistance.

• DOCs facilitate commercial insurance productdevelopment by providing a means by whichcatastrophic risk layers can be effectively trans-ferred into international markets.

• DOCs can be subsidized to address the mar-ket failure associated with ambiguity loadingand cognitive failure.

• Governments can estimate their own DOC sub-sidy cost exposure based on actuarial estimatesof the risk inherent in the index. Reinsurancecoverage adds a market check on the credibil-ity of the index and the adequacy of DOC pre-mium rates.

• While DOCs may be partially subsidized, endusers still pay part of the cost to transfer the

risk into international markets. This reducesthe potential for perverse incentives that couldencourage excessive risk taking.

Subsidies34

Governments frequently subsidize agriculturalinsurance products. These subsidies take a varietyof forms. The government may cofinance insur-ance purchasing with direct premium subsidies,reimburse primary insurers for administrative orproduct development costs, or provide reinsur-ance at below market premium rates. Regardlessof the form, government subsidies are generallydesigned to increase insurance purchasing by low-ering the premiums charged to agricultural insur-ance purchasers.

Such subsidies are extremely controversial. Theytend to benefit operators of larger farms more thanthose of smaller farms. A wide range of stakeholderscan and will engage in rent seeking once subsidiesare introduced. Subsidies are costly to maintain andare subject to close scrutiny regarding social costsversus social benefits. Many times, subsidies areprovided based on the rationalization that agricul-tural insurance markets are missing or incomplete,without careful consideration of the core reasonswhy such market limitations exist. This documenthas carefully considered why agricultural insuranceis missing or incomplete in many settings: adverseselection and moral hazard, high transaction costs,cognitive failure and ambiguity loading, and expo-sure to highly correlated loss events. Any govern-ment subsidies should be carefully targeted toaddress one of these specific sources of market fail-ure. Even then, however, the costs of addressingthat market failure may simply be too high to justifyuse of limited government resources to that end.

The rents resulting from even the most carefullytargeted subsidies can still be captured by politi-cally powerful elites. Government insurance subsi-dies may crowd out demand for private sector risktransfer instruments. The World Bank supports thedevelopment of financial institutions that operateprofitably on a commercial basis by offering prod-ucts and services that meet the needs of a wide rangeof clients, including the poor. Thus, any World Bankefforts to facilitate the provision of risk transferinstruments should be based on careful consider-ation of whether subsidies or grants can be pro-vided without distorting or inhibiting the growthof private sector financial markets.

Some types of subsidies are likely to be lessdistorting than others. Subsidies and grants forsupporting financial intermediaries and financialinfrastructure, such as technical assistance and datasystems needed to develop effective index insuranceproducts, generally create little distortion. Beyonddistortions in the markets, legitimate reasons existfor supporting infrastructure to improve market ac-cess among the rural poor. Finally, some public sup-port for product development may be justifiablebecause of the free rider problem. Innovative insur-ance products are costly to develop, yet it is difficultto recoup these costs in a competitive market. Anyfirm can simply copy and compete with the newproduct without the expense of recovering productdevelopment costs. Unfamiliarity with index insur-ance products can heighten these problems in manydeveloping countries.

Examples of subsidies for financial intermedi-aries and infrastructure include:

• Providing technical assistance to financialintermediaries to improve systems that en-hance efficiency, such as management infor-mation systems;

• Developing and introducing demand-drivenproducts on a pilot basis;

• Helping to develop or improve service deliv-ery mechanisms that enable greater outreachinto rural areas;

• Covering a portion of the cost of establishingnew branches in areas lacking financial inter-mediaries to serve the poor;

• Creating capacity within regulatory and super-visory bodies;

• Supporting the creation of industry assoc-iations;

• Developing training institutes and insuranceinformation agencies;

• Supporting data for weather stations or otherdata to be used to develop effective indexes;and

• Providing technical assistance to develop newproducts in an emerging market in develop-ing countries.

Premium Subsidies

While it is common for developed countries to cofinance premiums for farmers with direct pre-mium subsidies, these types of subsidies are par-ticularly problematic. Generally, direct premiumsubsidies reflect income enhancement objectives

as much or more than they do risk managementobjectives. Such subsidies are typically provided ona percentage basis. This clearly benefits higher riskareas relatively more than lower risk areas. Evenattempts to subsidize to levels that represent a purepremium or expected loss basis may favor higherrisk areas relatively more than lower risk areas, sincein a commercial market, premium rates for higherrisk areas would likely contain higher catastrophicloads. Thus, any attempt to introduce premiumsubsidies will likely be distorting.

In principle, if subsidies are targeted to the mar-ket failure layer, as described above, market dis-tortions should be minimal. Given the ambiguityloading and cognitive failure that occur in this layer,carefully targeted subsidies (such as cofinancingof DOCs) may even be welfare enhancing. For themarket insurance layer, however, subsidies should,in general, be avoided. Any subsidies in the marketinsurance layer should be targeted to reducinguncertainty loads in premium rates. Commercialinsurers will tend to load premium rates based onthe quantity and quality of data used to generatepure premium rates. The better (worse) the dataused to generate the pure premium rates, the lower(higher) the premium load. These loads could beoffset with cofinancing from donors. Here again,however, donors should be very clear about thelevel of these subsidies and the intent behind them.

INDEX INSURANCE AS A SOURCEOF CONTINGENT FUNDING FOR GOVERNMENT DISASTERASSISTANCE AND SAFETYNET PROGRAMSIn addition to rural economic growth, governmentsalso want to manage disaster assistance efforts moreeffectively and to combat poverty by pursuing socialand equity objectives. Rather than listing the multi-tude of social policy responses to these objectives,this document focuses on the link between fundingfor social policy tools and risk. Specifically, index in-surance is proposed as a source of contingent fund-ing for government disaster assistance and safetynet programs.

Ex Ante Disaster Risk Management

Disaster financing has generally focused on pro-viding resources for ex post relief operations to



cope with shocks rather than on making dedicatedresources available ex ante. This has often meantproviding in-kind emergency resources rather thancash resources. Additional transient needs are metthrough emergency relief operations that oftenduplicate ongoing interventions: that is, throughpublic works and assistance to the vulnerable. More-over, due to delays in declaring emergencies andmobilizing and then distributing resources, reliefoften takes significant time to arrive and, indeed,can arrive too late.

Index insurance could be used to provide con-tingent ex ante funding for emergency relief opera-tions. The relief could be distributed through normalemergency channels but would benefit from ex antefunding and timelier provision of assistance. Currentfunding for emergency activities in food-insecurecountries is based on a protracted appeals-based sys-tem that delivers food aid well after crop failures andweather shocks. By this time, the people affected mayhave already had to sell productive assets and/ormigrate. Additionally, the support that does comeis not consistent; delivered as a result of appeals toindividual donors subject to their own approvalprocesses and budget cycles, deliveries are unpre-dictable. The use of index insurance as a means ofcontingent funding for emergency assistance maymitigate some of the shortcomings of the currentsystem. Index insurance provides timely and pre-dictable payouts during emergencies; by fundingearly relief they preserve livelihoods and to someextent preempt emergencies (Skees et al. 2005; Goesand Skees 2003).

Safety Nets

Safety nets respond to the needs of the poorest andmost vulnerable by providing livelihood supportand contributing to immediate food security, oftenthrough community-driven public works schemesand transfers to vulnerable labor-poor individuals.In times of adverse climatic shocks, the number of households in need of assistance dramatically

increases, necessitating the scale-up of the safetynet. Because the emergency response capabilitiesof existing safety nets are currently limited, how-ever, they could be complemented with index-based disaster insurance.

The scaled-up safety net is limited by two factors:

• Design. Safety nets often focus on addressingchronic poverty rather than transient poverty.Although efforts have been made to scale upsafety nets in time of drought, for example, thishas proved difficult due to delays in mobiliz-ing financing and organizing activities.

• Capacity. Existing safety net operations have in-creasingly focused on implementation throughlocal government structures. This is a positivedevelopment, as it will lead to enhanced localcapacity in the long run, but capacity at thelocal level is limited, and scaling up rapidlyand effectively in times of need requires sub-stantial existing capacity.

Safety nets could be enhanced using index insur-ance. A rainfall index, for example, could be usedto automatically trigger payments to districts inwhich the drought-affected population is concen-trated, with the sums insured based on this popu-lation’s likely size. Targeting to the household levelwould then be used to determine which individu-als in the district should receive payments. Cash fi-nancing would be distributed to districts early (thatis, immediately after the weather shock and beforeharvest) to scale up existing safety nets as rainfallmeasures indicate where production shortfalls willoccur. This plan distributes cash during the criticalcoping period, several months earlier than undercurrent emergency arrangements, and before thehungry period has set in.35 This mechanism wouldnot replace emergency operations but would in-stead provide timely contingent funding to scale upexisting safety net structures. Providing assistancein the early stages of a disaster event may preemptthe need for more extensive, long-term emergencyresponses.

35

The previous chapter presented the conceptual foundations for de-veloping risk transfers. This more pragmatic chapter offers concreteexamples of the progress made in using index insurance for agricul-tural risk transfer in several developing countries. Index insurance isnot a new concept. Chakravati in India was writing about this type ofinsurance as early as 1920. Sweden and Quebec, Canada, had area-yield insurance programs beginning in the 1950s and 1970s, respec-tively. The United States introduced the Group Risk Plan in 1992(Skees et al. 1997). The concept of index insurance based on area rain-fall follows many earlier efforts with area-yield insurance.

The World Bank and other donors were involved in crop insuranceprojects in the 1970s and 1980s. These efforts were soon abandoned,however, as many of the problems with introducing multiple-perilcrop insurance in developing countries became insurmountable con-straints. Hazell (1992) emphasized the problems with traditional cropinsurance and recommended using rainfall insurance. Hazell andSkees (1998) participated in the World Bank’s first efforts to return tocrop insurance work, undertaken in Nicaragua. Skees and Miranda(1998) followed the work in Nicaragua, and this lead to the develop-ment of the Skees, Hazell, and Miranda (1999) document. In 1999, ateam of World Bank professionals and outside consultants obtaineda Development Market Place award to work in Morocco, Nicaragua,Ethiopia, and Tunisia. Many of the efforts describe in this chapter fol-low the conceptual development of that project.

As with any innovation, the adoption of this new insurance prod-uct went through various life cycle stages. Often an idea is largelyignored for decades before being slowly adopted. After the idea hasbeen tested, the replication phase begins. The overall efforts describedin this document are just entering the replication phase. Initial effortsto introduce the concepts in Nicaragua and Morocco have been slowto develop into projects. Nonetheless, these efforts and the experienceof performing feasibility studies in these countries proved invaluablein the overall adoption process.

Table 6.1 lists the chapter’s country case studies in the order inwhich they are presented. Nicaragua and Morocco are covered first,as they were the first two countries to undertake the work, followedby India and Ukraine, both countries in which weather index insurancehas been used. Ethiopia, Malawi, and the SADC appear next, pre-sented together because of the common elements in their experiences.The next listed countries, Peru and Mongolia, each demonstrate uniqueaspects. Finally, the current progress of the Global Index Insurance

From Theory to PracticePilot Projects for Agricultural RiskTransfer in Developing Countries36

6

Facility is described; this effort, much broader inscope than the individual country efforts, could sig-nificantly facilitate risk transfer for all of the preced-ing programs as well as any future activity.

NICARAGUAA Seven-Year Incubation Period

Country Context and Risk Profile

The contribution of agriculture to the NicaraguanGDP has been in decline, but it still remains a sig-

nificant economic activity. In 2003, agriculture ac-counted for nearly 18 percent of the US$4.1 billionGDP of Nicaragua, and thirty percent of popula-tion is involved in agricultural activities. The majorcommodities produced include coffee, meat, shrimp,corn, sugar, and beans. Since the 1990s, however,agriculture has had little or, often, negative growth.With its agricultural production hindered by expo-sure to drought and flood risks, Nicaragua has re-mained a net food importer of cereals and grains.

Nicaragua has provided the World Bank’s firstexperience in recent history of serious consideration


Table 6.1 Summary of Case Studies

Objectives

Initial Work Better Social andby the Disaster Poverty Conceptual Significance of

Country World Bank Status Growth Risk Mgmt Reduction Risk-Transfer Model

Pilot in 2005

No project

Three years of salesFirst sales in2005

Pilot in 2006

Pilot 2006

Pilot 2005Feasibilitystage

Pilot plannedfor 2006Pilot plannedfor 2006

Concept note

▪

▪▪▪

▪

▪

▪▪

▪

▪

▪

▪

▪

▪

▪

▪

▪

▪▪

Direct link to loans and reduction of in-terest rates when farmers purchase indexinsuranceMore efficient and effective drought riskmanagement for cereal producersLarge scaling-up and mainstreaming ofweather insurance for smallholdersRegulatory approval under traditionalinsurance legislation and piloting ofweather index insurance (first weatherinsurance contracts sold in April 2005)World Bank addressing rural risk in com-prehensive manner; weather insurancefor smallholdersWFP/WB jointly developed ex anteweather insurance based financing ofearly response to weather failure leadingto negative coping strategiesWeather insurance for groundnut farmersIntroduction of scaled-up safety nets; improved food security risk managementcomprehensivelySystematic approach to dealing with agricultural risk by governmentWorld Bank pilot project mainstreamingdesigned to learn if herders will pay acommercial rate for mortality index insur-ance; prepaid indemnity pool coupledwith a structure to completely protect thefinancial exposureReinsurance intermediation for micro-and macrolevel insurance for insurers,governments, and banks

Source: Authors.

Nicaragua

Morocco

India

Ukraine

Ethiopia,Micro

Ethiopia,Macro

MalawiSADC

Peru

Mongolia

Global IndexInsuranceFacility

1998

2000

2003

2002

2003

2003

20042004

2004

2001

From Theory to Practice 37

of rainfall insurance. Hazell and Skees providedthe first feasibility study in the spring of 1998.Subsequently, Skees and Miranda (1998) examinedthe issue in more detail and made specific recom-mendations about rainfall insurance in the majorcereal production area of northwest Nicaragua,which suffers major risks to cereal production frominsufficient or excess rainfall, concluding that therisk of both could be hedged using rainfall indexinsurance contracts sold to individual farmers.Nonetheless, Skees and Miranda also pointed tolarge hurdles blocking such an introduction in adeveloping country and offered four key recom-mendations for the development and sustainabilityof such an insurance scheme:

• Analytical work and development of human cap-ital. Extensive data analysis and modelingwould be necessary to design and price theinsurance contracts. Training Nicaraguans inthese methods would be equally importantin developing capacity within the countryfor future efforts.

• Pilot development for demonstration, education,and evaluation. In its first year, the pilot shouldstart small and target primarily learning anddemonstration. Education, marketing, andsales would be primary goals. Only three sta-tions should be used in the first year: Leon;San Antonio; and Chinandega. The marketthus delineated would be contiguous andwould cover no more area than eight hundredsquare kilometers. To obtain the most effec-tive risk management, only producers withinten kilometers of the stations should purchasethe rainfall contracts.

• Infrastructure development and pilot expansion.During year one of the pilot, investments inadditional secure weather stations should bemade to increase the density of stations withinthe original eight hundred square kilometermarket area. By year two, sales and exposureshould increase to about US$10 million.

• In-country project management and support. It isessential to have a key person in Nicaragua tomanage and support the pilot project. Thisperson should know all aspects of the projectand take an active role in every dimension ofthe project. Central goals for this individualwould be monitoring the activity and provid-ing international reinsurers with the con-fidence necessary to participate. Beyond the

pilot test area, the key person should investi-gate new regions with the potential to standon their own, with private support, in whichto inaugurate additional pilot programs; fos-tering similar activity in other regions will helpentice the international reinsurance commu-nity. Additional responsibilities for the keyperson would be facilitating an active educa-tion program and managing and deployingfunds for advertising and promotion.

Discussion of these concepts was progressing in Nicaragua’s public and private sectors whenHurricane Mitch arrived with its devastation inOctober 1998. After this event, the World Bank’stechnical assistance efforts in Nicaragua shifted todeveloping an aggregate weather index that wouldprovide disaster financing to the government dur-ing severe weather events. This work developed tothe point at which a specific set of weather stationswere indexed into a single aggregate index to pro-tect against catastrophic risk; the index was evenpriced in the global reinsurance markets. After thecontract was priced, however, the government re-jected the idea, maintaining that they did not needto purchase insurance because they could dependon the global community for assistance when majorcatastrophes occurred. Subsequent to this decision,no further activity on index insurance has been pur-sued in Nicaragua. Nevertheless, the Nicaraguanexperience provides a number of significant lessons:

• It takes time to develop innovation. The literatureon innovation emphasizes that it takes time,sometimes as much as a generation, for newideas to gain acceptance. The Nicaraguan ex-perience perfectly illustrates this observation.The original weather insurance idea was pre-sented in Nicaragua seven years ago, but newproducts deriving from those ideas are onlynow being introduced. One reason Nicaraguamay be proceeding now is because in themeantime other countries have ventured intothis domain.

• The expectation that countries will purchase cata-strophic protection presents an inherent moralhazard. The excellent work completed follow-ing Hurricane Mitch to develop a mechanismfor the government of Nicaragua to indem-nify catastrophic losses from extreme weatherevents met with a cool reception. The govern-ment was likely correct in its conclusion that

this type of protection was not needed, sincethe global community has been very respon-sive with free aid after major catastrophes.

• Linking index insurance to banking in Nicaraguais an excellent addition to ongoing work else-where around the globe. Early indications arethat Nicaragua’s banks have agreed to reduceinterest rates for production loans for farmerswho purchase the new weather index insur-ance products. Nicaragua may be the first coun-try to forge an explicit tie between interest ratesand the amount of index insurance purchased.This is an important development that shouldbe evaluated and more fully understood.

Proposed Agricultural Risk Management Structure

In November 2004, CRMG responded to INISER’sinterest in developing a local weather index insur-ance market for agriculture. CRMG provided tech-nical assistance to analyze potential markets for apilot project in 2005 and decided to concentrate ondeveloping a pilot project to secure lending for thegroundnut sector. Banks have expressed interestin internalizing some part of the risk reduction bylowering interest rates and providing financing forfarmers to pay premiums as incentives for a pro-active financial risk management approach.

Armed with prototype contracts, INISER/CRMGhas launched consultations with end users, finan-cial intermediaries, and the insurance regulator.Final contracts have been designed and priced byreinsurers, although they still await approval fromthe regulator. The pilot project is expected to beginoperations in the spring-summer of 2006.

The government of Nicaragua had adopted a“wait-and-see” strategy, based on several previousfailures to launch either traditional or weatherindex insurance for agriculture. It was not until themost recent proposal was being developed and thegovernment could clearly see the interest and par-ticipation of the international financial markets thatit opened the door for serious policy dialogue onthe issue. In particular, the government has offeredto support INISER in the implementation phasewith economic resources as well as guidance forscaling up the current pilot project. This has openedthe door to work with several productive sectors,including small farmers, in a comprehensive con-text of economic development in which insurance

becomes a useful tool for facilitating investments inthe agricultural sector.

MOROCCOCountry Context and Risk Profile

In Morocco, 47 percent of the total population andmost of the poor live in rural areas. Agricultureplays a crucial role in rural livelihoods. On average,agriculture accounts for about 17 percent of theGDP, but this percentage fluctuates, mainly due toclimatic—especially rainfall—variations. Moroccanagriculture is characterized by a dichotomy betweenthe traditional and commercial sectors. The tradi-tional sector consists of small farms in rain-fed areasinvolved predominantly in cereal, legume, and live-stock production; the commercial sector operatesmainly in irrigated areas. Farm surveys indicate thatabout 70 percent of farms are small in size (under5 hectares) and account for 23 percent of total landunder cultivation. Farms less than 20 hectares (ha)in size represent 96 percent of farms in operation.Average farm size in Morocco is 5.7 ha. Almost 90 percent of Moroccan agriculture is nonirrigated,and the dependence of most crops on adequaterainfall translates into wide variations in yieldsand production. Drought caused cereal produc-tion, for example, to fall from 9.5 million tons in1994 to 1.6 million tons in 1995.

Current Response

In 1995, the Moroccan government activated theProgramme Secheresse (Drought Program), a state-sponsored insurance program managed by the localmutual agricultural insurance company (MAMDA)to address the drought problem by implementinga yield insurance scheme. The program, revised in1999, is structured on the coverage of three revenuelevels: 1,000, 2,000, and 3,000 Moroccan Dirhams(MAD) per hectare (ha). Payments are triggered bya ministerial declaration certifying the occurrenceof drought. For the first revenue threshold, the pay-out is based on an area-yield base mechanism,while for the 2,000 and 3,000 MAD/ha levels, spe-cific farm yield assessments are required. The pro-gram proved to be popular, but it also encounteredtypical yield insurance problems, such as high costsfor supporting insurance premiums and severemanagement problems related to individual farmyield assessment (Hess et al. 2003).




Given the limitations of the Drought Program, theMoroccan government agreed to participate in aWorld Bank research project aimed at exploring thefeasibility of weather-based insurance as an alter-native to traditional yield insurance. The investi-gations led the team to conclude that a droughtinsurance program based on rainfall indexes couldhave potentially significant benefits over the currentscheme, minimizing moral hazard and adverseselection risk and promoting a more rapid, stream-lined pay-out process, in addition to increasing thepotential interest of international reinsurers andcapital markets in investing in the program. Basedon analysis of rainfall and cereal-yield data acrossthe country, the study determined that an index-based rainfall insurance product could be feasiblein Morocco. Following the feasibility study, an international team sponsored by the IFC and theItalian Technical Assistance Trust Fund assistedMAMDA in structuring the insurance coverage tobe launched as a pilot program in some cereal grow-ing regions.

Products

The product proposed was a rainfall index insur-ance contract that would indemnify cereal pro-ducers when the rainfall index in a given area fellbelow a specified threshold.

The indexes, developed by local agronomiststogether with farmers’ representatives, added im-portant insights into the relationship of rainfall toyield. They were not just cumulative measures ofrainfall but included specific weights for differentplant growth phases and a “capping” procedure totake into account the loss of water in excess of stor-age capacity and hence unavailable to contributeto plant growth. This process allowed the indexesdeveloped to reach correlation values of over 90 per-cent (Stoppa and Hess 2003), and they were greatlyappreciated by the potential end users.

Constraints

Despite the wide consensus gained by the pro-posed rainfall index contracts among governmentofficials, insurers, and producers, the implemen-tation of the planned pilot programs in Moroccodid not take place. The main reason for this failure

was that rainfall precipitation in the selected areasshowed a downward trend, and the reinsurancecompany involved in the deal made the cost of theinsurance prohibitive for producers. The experiencedeveloped through Morocco’s feasibility studyand planned implementation project, however,generated expertise that led to the realization ofother WB-facilitated deals (for example, in India)and of other independent programs (for example,in Colombia).

INDIAPrivate Sector Led Alternative Agricultural Risk Market Development


A 1991 household survey addressing rural accessto finance in India revealed that barely one-sixthof rural households had loans from formal ruralfinance institutions and that only 35 to 37 percentof the actual credit needs of the rural poor werebeing met through these formal channels (Hess2003). A survey based on the Economic Census of1998 (Hess 2003) shows that Indian formal financialintermediaries reportedly met only 2.5 percent ofthe credit needs of the unorganized sector throughcommercial lending programs.37

Current Response

Farmers, then as now, responded to the lack of for-mal financial services by turning to moneylenders;reducing farming inputs; overcapitalizing and in-ternalizing risk; and/or by overdiversifying theiractivities, leading to suboptimal asset allocation.Smallholders cannot risk investing in fixed capitalor concentrating on the most profitable activitiesand crops, because they cannot leverage the start-up capital and they face catastrophic risks, such asdrought, that could wipe out their livelihoods atany time. The challenge for banks is to innovatelow-cost ways to reach farmers and help them bettermanage risk.


An initial study explored the feasibility of weatherinsurance for Indian farmers to determine if it wouldbe possible to extend the reach of financial ser-vices to the rural sector by reducing exposure to

weather risk (Hess, 2003). The study identified sev-eral potential project partners. In response to thisstudy, CRMG, in collaboration with the Hyderabad-based microfinance institution BASIX and theMumbai-based insurance company ICICI Lombard,a subsidiary of ICICI Bank, initiated a project tolaunch the first weather insurance initiative everundertaken in India: a small weather insurancepilot program for groundnut and castor farmers inthe Andhra Pradesh district of Mahahbubnagar.

The insurance contracts were designed by ICICILombard, with technical support from CRMG andin consultation with BASIX, to protect farmers fromdrought during the groundnut growing season.The products were marketed and sold in the fourvillages selected by the extension officers of KrishnaBhima Samruddhi Local Area Bank (KBS LAB)38

using workshops and meetings with the BASIXborrowers. In total, 230 farmers (154 groundnutand 76 castor farmers) bought the insurance forkhariff, the monsoon season from June to September,2003. Most purchasers fell into the small farmercategory, with less than 2.5 acres of landholding.The entire portfolio of weather insurance contractssold by BASIX was insured by ICICI Lombard,with reinsurance from one of the leading inter-national reinsurance companies.

ICICI Lombard was also involved in anotherproject in khariff 2003 in Aligarh, Uttar Pradesh,where 1,500 soya farmers bought protection againstexcessive rainfall. ICICI Lombard filed all the nec-essary forms and terms of insurance with the Indianinsurance regulator, registering their productsbefore the programs were launched.

A second pilot program was launched in khariff2004 and introduced significant changes to the 2003design following farmer feedback from the pilotprogram, with technical assistance from CRMG.The program was extended to four new weatherstation locations in two additional districts in AndhraPradesh: Khammam and Anantapur. The weatherinsurance contracts were offered to both BASIXborrowers and nonborrowers and marketed andsold through KBS LAB in the Khammam andMahahbubnagar districts and through BhartiyaSamruddhi Finance Ltd. (BSFL)39 in the Anantapurdistrict using village meetings, farmer workshops,and feedback sessions during the month leading upto the groundnut and castor growing season. Newcontracts were also offered for cotton farmers in theKhammam district and an excess rainfall productfor harvest was offered to all castor and groundnut

farmers. In total, over 400 farmers bought insurancethrough BASIX in 2004, and a further 320 groundnutfarmers, members of a the Velugu self-help grouporganization in the Anantapur district, bought in-surance directly from ICICI Lombard. Severalfarmers were repeat customers from the 2003 pilot.In contrast to 2003, ICICI Lombard did not seekreinsurance for the BASIX farmer/weather insur-ance portfolio in 2004.

In 2004, a number of other transactions also tookplace within the Indian private sector in responseto the 2003 pilot program initiated by CRMG. In2004, BASIX themselves bought a crop lendingportfolio insurance policy based on weather in-dexes. For the first time, BASIX used this protectionto cover their own risk and passed neither the costnor the benefits to their farmers. The protectionallowed BASIX to keep lending to drought-proneareas by mitigating default risk through the insur-ance policy claims in extreme drought years. BASIXbought a policy, insured by ICICI Lombard withstructuring support from CRMG and reinsured intothe international weather market, covering threebusiness locations.

During 2004, not only did BASIX expand theirweather insurance program, a number of other in-stitutions, including the originator ICICI Lombard,began expanding the market for weather insur-ance in India. In 2004, IFFCO-Tokio, a joint ventureinsurance company, launched weather insurancecontracts similar to the 2003 contracts, sellingover 3000 policies to farmers throughout India. In conjunction with ICICI Lombard, the govern-ment of Rajasthan launched a weather insuranceprogram for orange farmers, insuring 783 orangefarmers from insufficient rainfall in khariff 2004;they also covered 1036 coriander farmers in rabi(the October to March growing season) 2004. TheNational Agricultural Insurance Company (NAIC),responsible for the government-sponsored area-yield indexed crop insurance scheme, also launcheda pilot weather insurance scheme for twenty dis-tricts throughout the country in 2004, reachingnearly 13,000 farmers; the scheme was even men-tioned in the Indian government’s budget for thefinancial year 2004–2005. It is estimated that nearly20,000 farmers bought weather insurance through-out India in 2004.

In 2005, BASIX/ICICI Lombard further improvedits weather insurance product and automated under-writing and claims settlements. In 2005, BASIX soldarea-specific weather insurance products in all of



its fifty branches, finally selling 7,685 policies to6,703 customers in thirty-six locations in six Indianstates. In addition, ICICI Lombard scaled up itsagricultural weather insurance sales and expandedinto other sectors, while NAIC and IFCCO-Tokiostepped up their efforts to sell weather insuranceproducts and to develop better products for farm-ers. New insurance providers such as HDFC Chubbalso entered the market. It is estimated that during2005, 250,000 farmer bought weather insurancethroughout the country. In partnership with ICICILombard, over seventy new automated weatherstations were installed by private company Delhi-based National Collateral Management ServicesLimited, on which weather insurance contracts werewritten for the 2005 monsoon season. The companyplans to scale-up their installations throughout thecountry with more insurance-provider partnersin 2006.

Monitoring will be an important element of thenew pilot programs. Ultimately, it will be necessaryto learn not only if farmers are buying these prod-ucts but how the purchases are changing their be-havior and the lending behavior of local financialinstitutions. Box 6.1 describes the initial steps beingtaken to monitor the Indian weather insuranceproducts. An early result of monitoring efforts—learning why farmers purchase the insurance—isreported in Table 6.2.

UKRAINECountry Context and Risk Profile

Rural financial institutions in Ukraine increasinglyuse future harvests as collateral, since farm equip-ment is generally antiquated and of limited value.These lenders also tend to require harvest insur-ance to hedge against crop losses.40 The major banksactive in agricultural lending, such as Aval (witha total of 4600 loans and 30 percent market share),do not lend on the basis of uninsured collateral, soto obtain credit a farmer must have a proper insur-ance policy written by a preapproved insurer. Toprovide for the lending insurance needs of farmers,most banks set up their own insurance companies.Most farmers do not yet understand the particularnature of weather index insurance, but they arefamiliar with weather risk and would like to haveprotection against multiple natural perils.

Crop risk is diverse throughout Ukraine. Crop-yield data for five major crops (maize, sunflowers,

sugar beets, wheat, and barley) in all twenty-fiveoblasts in the 1970 to 2001 period show a substan-tial geographic spread of the agricultural valuesconcentrated in central and southern Ukraine. Thecorrelation of crop yields between eastern Ukraineand the southern region near Odessa is nearly zero,facilitating risk pooling and in-country retention ofa large share of natural risks.

Current Response

In this market, the types of insurance policies cur-rently offered are input cost insurance, generallylinked to agricultural credit collateral requirementsand limited to very low insured sums, and harvestinsurance, covering hail, storm, excessive precipi-tation, frost, and fire risk. Drought insurance is of-fered by only a few companies and is not generallycovered. Two crop insurance pools, one composedof five companies and the other of sixteen, werefounded in 2003 as part of attempts to provide moresecure crop insurance to Ukrainian farmers. The in-surance companies agreed to pool their agricul-tural risks to improve their risk-bearing capacityand to obtain access to international reinsurancemarkets. Nevertheless, crop insurance policy saleswere very limited (around eighty for both pools).Market participants cited the following reasons forthe low uptake: inability to pay for the policy, un-clear loss adjustment and underwriting procedures,mistrust of insurance companies, and insufficientinformation available to farmers. Moreover, by pro-viding ad hoc disaster assistance to farmers in 2003and 2004, the government of Ukraine (GoU) low-ered incentives for farmers to pay for commercialinsurance premiums. According to recent marketinformation, by the end of 2004, the biggest agricul-tural insurance pool had shrunk to six companies.

Policy Objectives

The GoU has experimented with compulsory cropinsurance and is now establishing a crop insurancesubsidization scheme. The regulator has approvedweather index insurance as an insurance product,and a few weather insurance policies were sold tofarmers in the first pilot sales season of 2005.

A feasibility study by CRMG presents a risk man-agement framework and considers several optionsfor government intervention in the sector. An in-vestment phase would consist of the acquisition andinstallation of automated weather stations, includ-

Box 6.1 India Impact Assessment

CRMG and DECRG designed a baseline surveya thatwas implemented by the International Crop ResearchInstitute (ICRISAT). The survey was conducted to studythe introduction of the rainfall insurance products designed by ICICI Lombard and marketed throughBASIX. The main objectives were to assess, first, thetake-up rate, that is, the factors influencing the deci-sion to purchase the insurance product, and, second,the impact of the insurance product in the treated villages as compared to the control villages. A samplewas drawn from Hindupur, Anantapur district, andNarayanpet, Mahahbubnagar district, of 1,052 farm-ing households, including 267 buyers, 186 nonbuyerswho attended a marketing meeting, and 299 nonatten-dees in the treated villages. In addition, 300 farminghouseholds were interviewed in control villages.

Anantapur and Mahahbubnagar are characterizedby low and uncertain rainfall, low levels of irrigation,and shallow and infertile soils. Anantapur has virtuallya groundnut monoculture, while Mahahbubnagar hascastor bean, groundnut, sorghum, pigeon pea, maize,cotton, paddy, and finger millet crops. Crop failure isvery frequent in these districts, mostly triggered bydroughts. Indeed, 80 percent of farmers considereddrought their main risk. In a drought year, farmers canlose about 25 percent of income. Drought affects mostvillagers at the same time, rendering informal insur-ance networks useless. Instead, in bad years, farmerssell livestock or their few assets and migrate to urbanareas or other states. In addition, they borrow fromformal and informal rural financial institutions. Theunion and state governments offer employment gener-ation schemes, watershed development programs, andother welfare schemes to stem migration and assuagethe misery of the people.

The rainfall insurance product was explained byBASIX and ICICI in village meetings. Most people whoheard about the meeting decided to attend; of those,35 percent attended because they trusted BASIX andanother 35 percent because friends and neighbors attended. Only 27 percent of the buyers purchased theinsurance during the marketing meeting, because theproduct was new and meeting attendees lacked therequisite funds. Meeting participants well understood

the crop to which the rainfall insurance was linkedand the premium and payouts, but not the trigger levels. In fact, insurance trigger levels are expressed in millimeters of cumulative rainfall, but most farmersdo not understand the concept of a millimeter. Mostfarmers determine when to sow by analyzing themoisture in the ground, and, indeed, only 10 percentwere able to make an estimate in millimeters of theminimum accumulated rainfall required to sow.

Nonetheless, take-up was high. Buyers said theypurchased the insurance for security reasons (exposureto rain, large cultivation of castor or groundnut, etc.)and because they were advised to do so by others. Yetinitially, many buyers thought of the insurance policyas a gamble. They put money at risk in the hope ofmaking a profit if the accumulated rainfall was belowa certain threshold. To support this claim, we find thatrisk-loving people are more likely to buy the policy aswell as those that believe that the monsoon rains willstart later, for whom the gamble has favorable odds.In addition, buyers are generally more educated, farmmore land (total and irrigated), have more savings atthe time of purchase, and are more likely to trust theinsurance product and BASIX, as compared to non-buyers. At the time of the survey, most farmers intreatment villages reported that they would like to pur-chase the insurance for the next khariff (main mon-soon) in June 2005. In addition, 14 percent of poorerfarmers said they would like to open savings accountsin November to save for the premium. Again, whenasked why they would like to buy the insurance in2005 (see Table 6.2), 60 percent cited security reasons,but a full 30 percent cited the experience of a payoutin 2004.

This willingness to purchase the policy as a resultof a payout is particularly telling in the context of theintroduction of a new product. Farmers may be uncer-tain that BASIX will honor its promise and thus may de-cide to wait and see and not change behavior. Indeed,the preliminary analysis conducted suggest that whilethere are no differences in input usage or area devotedto cash crops for farmers that do not trust Basix or theproduct, it does seem that trust in BASIX allows buyersto use the insurance policy as a hedging instrument.

Note:a. Financed by Swiss Trade Commission, SECO.

Source: This information is based on preliminary findings by economist Xavier Giné (DECRG, World Bank), working in collaboration withDon Larson (DECRG, World Bank), Robert Townsend, professor at the University of Chicago, and James Vickery, an economist at the FederalReserve Bank of New York.



ing analysis of the density of the network requiredto cover Ukraine’s weather exposure and designof an adequate maintenance program to ensure thequality of observations across time.

In addition, the GoU could consider a backstopfacility for weather risk insurance retention.Ukrainian insurance companies would need inter-national reinsurance for insuring against systemicrisks. A risk pool “facility” in Ukraine would allowfor the underwriting of agricultural reinsurancebased on preestablished guidelines to retain asmuch risk inside the country as possible. This poolwould then reinsure itself through a GoU fund.Extreme or catastrophic risk would be reinsuredon the international reinsurance market based ontransparent and competitive premium ratemakingprinciples; that is, once the pool and the GoU fundare depleted, international reinsurers would paythe remaining claims. Aggregation and layeringof risk would help interest reinsurers in reinsuringrisk in Ukraine, causing them to price risk compet-itively. Individual insurance companies sometimesface insurmountable difficulties even accessinginternational reinsurance markets, let alone obtain-ing competitive prices. The combination of intro-ducing a transparent index insurance product and

an efficient and well-regulated risk pool can over-come this market failure. Risk layers representingrelatively frequent (but mild) adverse events wouldbe insured by the GoU risk fund. Intermediate risklayers (for example, events occurring once in twentyyears to once in one hundred years) could be trans-ferred to the GoU Backstop Facility. The catastrophicrisk layer (the once in one hundred year event)could be transferred to international reinsurancemarkets.

ETHIOPIAEthiopian Insurance Corporation and Donor Led Ex Ante Disaster Risk Management


Ethiopia is one of the poorest and least developedcountries in the world, ranking 169 of the 175 coun-tries in the Human Development Index. More than85 percent of the population make their living inthe agricultural sector, which accounts for 39 per-cent of Ethiopia’s GDP (2002/2003) and 78 per-cent of foreign earnings. In Ethiopia, agricultureis predominantly rain-fed, and more than 95 per-cent of its output comes from subsistence and

Table 6.2 Reasons for Buying Weather Index Insurance in India

Khariff 2004 Khariff 2005

Reasons for Buying Insurancea Freq. % Freq. %

Security/risk reduction 144 54.8 181 53.2Could not afford to lose harvest income 25 9.5 11 3.2Low premium 19 7.2 1 0.3Advice from progressive farmers 18 6.8 0 n/aOther trusted farmers bought insurance 7 6.5 5 1.5Advice from village officials 10 3.8 1 .3High payout 10 3.8 10 .9Concentration on castor crop 7 2.7 4 1.2Product was well explained 5 1.9 0 n/aConcentration on groundnut crop 4 1.5 0 n/aLuck 4 1.5 5 1.5Paid out for previous year 0 n/a 107 31.5Advice from BUA members 0 n/a 11 3.2

TOTAL 263 100 340 100

Note:

a. The categories listed were created from open-ended survey responses to the question, “Why did you buy the insurance product for the lastkhariff?” The same categories may not apply for both years.

Source: ICRISAT survey, courtesy Xavier Gine.

smallholder farmers. The staple diet for the majorityof Ethiopians consists of coarse grains, includingmaize, teff (a cereal grain), and sorghum. Productionof coarse grains is valued at around US$380 millionand cereals at US$585 million.

At the household level, adverse weather patterns,primarily lack of rain, are detrimental to yields andoutputs and result in significant income losses andnegative impacts on farmers’ livelihoods. Ethiopiafaces highly variable rainfall and suffers from bothnational and regional droughts that can have ex-treme impacts on farmers who utilize traditionalagricultural practices with little irrigation and whorely on the country’s thirty-five million head of live-stock. This rainfall variability, in addition to limit-ing the ability and motivation of farmers to investin agricultural technology and yield-increasingassets, reduces overall production, which can de-crease both household consumption and income.At the national level, average grain production inthe country is 8.9 million metric tons (MT) and issubject to recurrent drought. The Ethiopian min-istry of agriculture has indicated that the level ofproduction is too low to feed the whole populationeven in good rainfall years.

Current Response

With 10 percent of the population of seventy-twomillion requiring food aid assistance each year,food insecurity is a chronic issue. Emergency re-sponses have been frequent if not constant, ac-counting for an annual average of 870,000 MT offood aid between 1994 and 2003. In 2003, a recordthirteen million Ethiopians required emergency as-sistance as a result of drought and the consequentfailed harvest in 2002. These emergency responseshave saved millions of lives in the short term, butdestitution has worsened, assets have eroded, andvulnerability has increased. The uninsured loss ofincome and assets caused by natural disasters, pri-marily droughts, in developing countries such asEthiopia threatens the lives and livelihoods of vul-nerable populations. Insurance is a critical require-ment for development, as uninsured losses lockentire populations in vicious cycles of deepeningdestitution. It is estimated that in sub-SaharanAfrica approximately 120 million people are at riskfrom natural disasters and that, for these popula-tions, humanitarian aid provides the only insur-ance protecting their lives and livelihoods. Buthumanitarian aid is often too unreliable, unpre-

dictable, and untimely to provide an effective in-surance function.

In 2003, in part to address this issue, the gov-ernment of Ethiopia (GoE), donors, United Nationsagencies, and nongovernmental organizations(NGOs), launched the New Coalition for FoodSecurity with the goal of achieving food securityfor the part of the Ethiopian population categorizedas “chronically food-insecure” and to improve sig-nificantly the food security for the additional tenmillion people vulnerable to becoming so in the nextfive years. To achieve these goals, starting in January2005, the organizations began working throughthe government to introduce a productive safetynet for five to six million people. The safety net isnot meant to serve as an emergency activity butto change the vulnerability and risk profile of thechronically food-insecure. Responses to chronicand to emergency food shortages began to be ad-dressed through different channels: the former,essentially a development activity, fell to the pro-ductive safety net program coordinated by the FoodSecurity Coordination Bureau, and the latter, a re-sponse mechanism to unpredictable humanitarianneeds, to the Disaster Prevention and PreparednessCommission (DPPC). Accordingly, those house-holds not covered by the safety net program but stillconsidered in need of government relief assistancewill fall under the emergency program throughearly warning and annual needs assessments.

Proposed Agricultural Risk Management Structures

To address the current situation in Ethiopia, two agricultural risk management structures are currently being considered, one at the farmer ormicrolevel and the other at the government ormacrolevel.

Microlevel Weather Insurance

The state-owned Ethiopia Insurance Corporation(EIC) plans to launch a small pilot weather insur-ance program for wheat and pepper farmers insouthern Ethiopia in the wereda (district) of Alaba,SNNPR. The EIC has previously experimentedwith agricultural insurance for farmers, but it metwith little success. The EIC is keen to explore newpotential products to address the risks of larger,commercial farmers. A pilot program, for whichit receives technical support from CRMG, is due tostart in April 2006. Part of the work includes the



demand assessment and participatory design ofcontracts with Alaba farmers

Macrolevel Ex Ante Funding of EmergencyRelief Operations

The World Bank and the United Nations WorldFood Programme (WFP) have launched a pilot toinvestigate the feasibility of index-based weatherinsurance as a reliable, timely, and cost-effectiveway of funding emergency operations in Ethiopia.The intention is to address the extreme emergencydrought situations that put pressure on donorbudgets and GoE strategic grain and cash reserves.The pilot is designed to serve vulnerable popula-tions who are neither food-insecure nor included inthe country’s new safety net program but who are“at risk” to income and asset losses and consump-tion shocks resulting from the more severe naturaldisasters. It is estimated that at least a further 35 per-cent of the population, above those consideredchronically food-insecure and covered by the safetynet, is at risk from hunger in the event of an extremedrought such as that in 1984. A traditional food aidresponse to a catastrophic drought in today’s priceswould be estimated to cost about US$1.6 billionfor all beneficiaries, chronic and nonchronic.41 Inpreparing for a future drought, rather than relyon traditional funding approaches based on pro-tracted appeals to international donors, the insur-ance approach focuses on transferring the risk to thereinsurance and capital markets. Such a mechanismwill ensure predictable and timely availability offunds with which the DPPC can launch emergencyrelief operations and appropriate interventions inthe event of a well-defined rainfall deficit at harvesttime. Some of the benefits of this type of insurance-based emergency funding include objective pay-outs, timely delivery, and funding in cash. In thecase of Ethiopia, the insurance approach wouldallow intervention four to six months earlier thandoes the traditional appeals-based system.

Policy Objectives

Both proposed agricultural risk management struc-tures are in line with the GoE current poverty reduc-tion strategy, which focuses on (1) agricultural-led,rural-based growth, recognizing the importance ofimproving the environment for exports, privatesector growth, and rural finance; and, linked to this,(2) food security. Clearly the microlevel weather in-surance initiatives are complementary to the govern-ment’s primary focus on agricultural development.

The poverty reduction strategy is character-ized by strong country ownership and focuses ona broad-based participatory process. In particular,the GoE favors a gradual shift from food assistance,assistance in kind, toward financial assistance thatcould be used to purchase food from the domesticmarket. The New Coalition for Food Security at-tests to the government’s ambitious poverty reduc-tion strategy: the main features of the safety net aremultiannual funding, transition toward cash-basedprogramming, scaled-up public/community works,linkages with broader food security programs, har-monized budgeting, and monitoring and evalua-tion. The Food Security Coordination Bureau hasbeen created, under the Ministry of Agriculture andRural Development, to coordinate all food securityprogramming, including the safety net. Targetingthe nonchronically hungry but food-insecure orvulnerable populations, an index-based weatherinsurance approach for Ethiopia aiming to providecontingency cash funding for responses to severeand catastrophic drought clearly aligns with thegovernment’s strategy and complements the safetynet initiative.

The objective of the macrolevel pilot project is tocontribute to an ex-ante risk-management systemto protect the livelihoods of Ethiopians vulnerableto severe and catastrophic weather risks. The pilotwill use a weather derivative to demonstrate thefeasibility of establishing contingency funding foran effective aid response in the event of contractu-ally specified severe and catastrophic shortfalls inprecipitation. WFP will put in place a small hedgefor Ethiopia’s 2006 agricultural season from Marchto October 2006, demonstrating the possibility ofindexing and transferring the weather risks ofleast-developed countries and facilitating price dis-covery for Ethiopian drought risk in internationalfinancial markets. In effect, in the pilot stage of theinitiative, the WFP will be the counterparty to a com-mercial transaction with the international risk mar-ket. Donors will pay for the premium associatedwith this risk transfer. Ideally, however, the ulti-mate aim of the initiative would be for the GoE totake responsibility for the risk management pro-gram as part of its overall long-term poverty reduc-tion strategy.

Constraints

Two major constraints might, in the short term, limitthe proposed risk management frameworks. Thefirst involves the weather-observing network and

the weather data available in Ethiopia. The NationalMeteorological Services Agency (NMSA) is respon-sible for a network of over five hundred weather sta-tions and rain gauges throughout Ethiopia. Not allof these weather stations, however, offer reportingcapabilities or historical data of a quality sufficientto transfer risk to the international markets or evento perform an actuarial analysis of the weatherrisks involved. Furthermore, given the large sizeand challenging topography of the country, thespatial distribution of the network is inadequate toprotect the entire country from weather risk. Theseissues will hamper both micro- and macrolevel ef-forts. On the microlevel, initially, only farmers wholive near good weather stations will benefit fromthe availability of weather insurance. Furthermore,the EIC may find it difficult to secure reinsurancefor this risk until the quality and security of theNMSA network improves. On the macrolevel scale,the weather protection can only be designed usingweather stations that adhere to the strict quality re-quirements of the international weather market.This will naturally limit the scope of the project inits first years.

The second constraint, more relevant for themacrolevel weather-risk transfer, involves fiscalissues: namely, the ability of the government ofEthiopia eventually to take over the ex ante fund-ing of the emergency relief operations program andto take responsibility for the premium paymentsnecessary to establish and maintain this fundingmechanism.

Products and Risk Transfer Structure

Both micro- and macrolevel proposals focus onindex-based weather risk management solutions.

At the microlevel, the EIC will market and sellweather insurance contracts to kebeles (small groupsof farmers) and/or farming cooperatives to protecttheir farmer members from the financial costs associ-ated with crop failure as a result of adverse weather.The products will be similar in concept to the prod-ucts offered to farmers in India (see Appendix 2), butit will be sold at the group rather than individuallevel in line with farmer preferences identified dur-ing discussions and focus groups in Alaba. The EICwill then seek international reinsurance for theirportfolio of weather risk.

At the macrolevel, lack of rainfall is the domi-nant, immediate cause triggering emergency relief

operations in Ethiopia. It is therefore an appropri-ate proxy for representing economic loss due todrought and also a simple, objective basis for indexinsurance. The appropriate index must be based ona weighted average, or “basket,” of as many stationsas possible to capture the macrolevel nature of therisk the GoE faces. The government may be able tocope with small, localized droughts by transport-ing food supplies from other regions of the countryand by sourcing government budget reserves.Retaining such risks will most probably be a morecost-effective solution than would seeking insur-ance, and Ethiopia should be able to take advan-tage of any natural diversification of the countryto reduce its insurance costs. In situations wheredrought severely affects a single region or affectsseveral regions or the entire nation, however, thegovernment may find this reallocation of resourcesunmanageable, making it appropriate to utilize thebasket-based insurance product to fund the ex-pected emergency relief operations in a predictableand timely manner. The basket approach also re-duces the risk of reliance on one weather stationand the associated issues of moral hazard and basisrisk. On this note, including more stations in the bas-ket not only provides better national coverage and,hence, enhances the representation of the index, italso increases the placement potential of the struc-ture in the international reinsurance markets. In2006, the index to be piloted is based on a basketof 26 weather stations distributed throughout theagricultural producing areas of the country.

In the pilot stage of the program, the WFP willbe the counterparty to any commercial transactionwith the international risk market and donors willpay for the premium associated with this risk trans-fer. In the event of an extreme and catastrophicdrought, however, any payment triggered by theinsurance would be made available to the GoEDPPC. This would allow the early provision of re-sources to the GoE and thus to the beneficiaries toensure appropriate consumption smoothing and toavoid distressed sales of assets, a vital outcome ifthe intervention is to play an effective and protec-tive role. With the availability of cash, the interven-tion can also be used to fund activities other thanfood aid that have already been established in otherparts of the country, such as cash-transfers, food-for-work, or cash-for-work schemes. Ultimately, thelong-term objective of these insurance plans wouldbe for the GoE to go directly to the market and take



responsibility for the program rather than having itcontinue to operate through the intermediary WFP.

MALAWI AND SADCWeather Risk Transfer to StrengthenLivelihoods and Food Security42


Malawi is dominated by smallholder agriculture,with farmers cultivating mostly maize, the staplefood. Maize is very weather sensitive and requiresa series of inputs. The economy and farm liveli-hoods are affected by rainfall risk (and resultingfood insecurity), soil depletion, lack of credit, andlimited access to inputs. Malawi suffers seriouscapacity constraints because it is ravaged by povertyand AIDS. Very few people have the energy andskills to build financial service programs.

Current Response

Malawi once had a paternalistic state culture. Therole of the state in agricultural marketing (mainlytobacco but also maize) is still strong. Prices are notfree, and smallholder incentives are distorted dueto food aid and sales of subsidized maize by the statemarketing board. The state and donors respond torecurrent drought-induced food crises by ad hocdisaster relief programs.

Proposed Agricultural Risk Management Structures

At the micro- or farm-level, weather-based indexinsurance allows for more stable income streamsand could thus protect peoples’ livelihoods and im-prove their access to finance. An insurance productcan be based on a crop production index constructedfrom weather data recorded at the airport weatherstation in Lilongwe (Malawi’s capital). Analysisand simulations conducted for the Lilongwe areaindicate that the match between potential insurancepayouts and farm-yield losses would be adequate.All that is needed is for demand to be aggregated atproduct distribution channels such as the NationalSmallholders Association (NASFAM). Rural finan-cial institutions could finance the insurance premi-ums and lower interest rates to borrowers, since thefinancial institutions stand to benefit from reduceddefault risk.

At the intermediary level, banks can packageloans and weather insurance into a single product,a weather-indexed crop production loan. Farmerswould enter into higher interest rate loan agree-ments that include weather insurance premiumsthat the bank would then pay to the insurer. In caseof a severe drought impacting crop yields, theborrower would pay only a fraction of the usualloan due and would thus be less likely to default,strengthening the bank’s portfolio and risk profile.Historical simulations in Malawi of such productsfrom maize demonstrated that the years of reducedloan payments coincided with the drought years inwhich farmers suffered from much lower yields,mainly 1992 and 1994. Recently, CRMG partneredwith Opportunity International (OI) to developweather insurance products to secure credit forgroundnut farmers. Nearly 1000 policies weresold in October 2005 for the 2005/2006 groundnutgrowing season.

At the macrolevel, a specific nationwide maizeproduction index for the entire country could formthe basis of an index-based insurance policy or op-erate as an objective trigger to a contingent creditline for the government in the event of food emer-gencies that put pressure on government budgets.Applying the Lilongwe maize farmer index ap-proach to the macrolevel situation, a Malawi MaizeProduction Index (MMPI) can be defined as theweighted average of farmer maize indexes mea-sured at weather stations located throughout thecountry, with each station’s contribution weightedby the corresponding average or expected maizeproduction in that location. Given the objective na-ture of the MMPI and the quality of weather datafrom the Malawi Meteorological Office, such astructure could be placed in the weather risk rein-surance market. Analysis shows that Malawi couldneed up to US$70 million per year to financiallycompensate the government in case of an extremefood emergency. Given the size of this figure, sucha transaction would be treated on a stand-alonebasis, with an estimated premium of approximatelythree times the expected loss for the reinsurer. Inthis case, the expected loss—given forty years ofhistorical rainfall data and assuming the govern-ment retains the cost associated with deviationsin maize production up to 25 percent away fromnormal—would be US$2.32 million, implying apremium of US$6.96 million or an insurance rateof 10 percent for such a product.

The weather index/drought risk managementapproach suggested for Malawi could be extendedto a regional level to include all members of SADCat some future point. Weather risk can be retainedand managed internally if the areas under manage-ment are significantly diverse in their weather riskcharacteristics. This immediately suggests that theweather sensitivity of neighboring countries, theSADC members, must be taken into account whenconsidering Malawi’s weather risk profile and itsneed for outside insurance. Analysis of the SADCregion shows that, on average, two countries sufferdrought each year. The distribution of droughtevents in SADC is extremely long-tailed, however,with the possibility of widespread drought eventsthat could potentially devastate the region. Thisindicates that the most efficient way to layer andthus manage the risk is as follows:

• SADC Fund: The size of the SADC fund couldbe set at US$80 million, the average financialimpact of four average droughts in the region,with each member contributing its share ac-cording to an actuarially fair assessment ofthe expected claim of each country.

• Reinsurance and/or contingent credit lines: SADC-wide events incurring a financial loss of, say,US$80 million to $350 million could be trans-ferred to the weather-risk reinsurance/profes-sional investor market. Alternatively, in suchsituations, the SADC members could have access to a World Bank contingent credit line.

• Securitization: The final and extreme layer ofrisk, such as drought in ten countries, occur-ring 1 percent of the time, could be securi-tized and issued as a CAT bond (investorslose the principal if the event occurs in ex-change for a higher coupon) in the capitalmarkets. The advantage of capital marketsfor this risk transfer is the immense financialcapacity of these markets and also the longertenure of CAT bonds: up to three years andpossibly longer.

A more efficient means of transferring risk impliesthat costs could be greatly reduced for the membercountries by transferring risk as part of a regionalstrategy rather than by transferring that risk onecountry at a time. The SADC fund approach out-lined above, for example, would reduce insurancecosts by 22 percent for Malawi due to risk poolingeffects.

PERUGovernment Led Systemic Approach toAgricultural Risk Management


Peru is currently negotiating a Free Trade Agreementwith the United States. Agriculture, because of itslack of competitiveness, is one of the most vulner-able sectors when an economy is opened. In thiscontext, the Peru’s Ministry of Agriculture (MA)is preparing a multidimensional strategy involv-ing extension services to farmers and innovativefinancial schemes, with the private sector partici-pating to facilitate access to better technology andnew markets. Because of farmers’ lack of bankablecollateral, the MA intends to facilitate the emer-gence of a sustainable private agriculture insur-ance market.

Current Response

Two major efforts in the last decade have attemptedto introduce agriculture insurance in Peru, but theresults were disastrous. Lack of technical knowl-edge and exposure to catastrophic events likeEl Niño generated big losses in the industry. Fromthe consumers’ perspective, these schemes werenot transparent and lack of education translatedinto dissatisfaction about the scope and use of thesefinancial instruments. Currently, crop insurance orsimilar instruments are not available to farmers.


The government of Peru (GoP) created a specialcommission in 2003 to draft a strategic plan forthe implementation of an agriculture insurancescheme. The treasury ministry, agriculture depart-ment, insurance regulator, private and developmentbank representatives, farm unions, and insurancerepresentatives participated in the discussions andrecommendations for the strategic work plan. A spe-cific body designed for that purpose is the TechnicalCommittee for the Development of AgricultureInsurance (TCDAI), which was created by ministe-rial resolution in September 2004 and is housed inthe agriculture ministry. The TCDAI is currentlyworking on several technical studies related to thedesign and implementation of agriculture insur-ance in Peru.



Policy Objectives

The main objectives of the GoP are (1) to maintainthe prudent fiscal, monetary, and exchange ratepolicies essential to attract investment and promotecontinued growth; and (2) to complement growthwith direct interventions that address inequalityand poverty, focusing on excluded groups: indige-nous people, Afro-Peruvians, and at-risk groupssuch as youths and single mothers (Peru, 2004–06).

Constraints

In addition to fiscal constraints, Peru’s agriculturalsector is divided into two: a group of powerful export-oriented, high-value agricultural produc-ers concentrated in twelve valleys along the coastand a group of smallholder agricultural producersoccupying the sierra (highlands) and selva (jungle)areas.

Products

The technical committee, assisted by CRMG, pro-posed a four-part work plan:

1. Design of prototype index contracts: The feasibil-ity of these contracts is tested for several cropsin the three main agricultural areas of Peru(coastal, sierra, and selva). The contract de-sign requires weather data from the Peruvianweather service (SENAMHI), acquisition ofwhich is a priority for the work plan.

2. Demand assessment: This activity will aim atgauging the demand for weather insurance bytype of producer and will include participatorydesign sessions addressing questions such aswhat types of contracts to develop and for whatperiods. This activity will include trainingpotential end users (farmers) regarding indexinsurance basics (for example, types of in-demnities, how indemnities and premiumsare calculated, and how contracts are settled).

3. Delivery model design: Based on a mapping ofrural financial intermediation in Peru, thisactivity will evaluate segmented deliverymodels to be used for real distribution chan-nels to farmers with small- and medium-sized farms with viable production potential.Prototype contracts by institution and clientsegment will be used in working with poten-tial intermediaries.

4. Regulatory review: The purpose of this activ-ity is to develop a strategic work plan withthe insurance regulator to prepare the neces-sary technical documentation for the indexinsurance product to be approved under theguidelines of property insurance.

The TCDAI has defined the following crops andareas of interest for the feasibility study:

Rice—San MartínMango—PiuraYellow maize—LimaPotato—HuanucoCoffee—CuzcoCotton (Tangis)—IcaCotton (Pima)—PiuraAsparagus—Lima

Risk-Transfer Structure

The GoP seeks to enhance risk-taking capacity inthe country generally by facilitating special risktransfer arrangements with insurance companiesin Peru, particularly those wishing to launch agri-cultural insurance. Specifically, the GoP wishes toset up a US$50 million fund, managed by the lead-ing second-tier bank (COFIDE), to take agriculturalrisk. In addition, the technical committee plans todevelop for insurers index-based products directlytransferable into international risk markets.

MONGOLIAWorld Bank Contingent Credit for LivestockMortality Index Insurance43


The economy of the Mongolian countryside isbased on herding: agriculture contributes nearlyone-third of the national GDP, and herding ac-counts for over 80 percent of agriculture. Animalsprovide sustenance, income, and wealth, protect-ing nearly half the residents of Mongolia. Shocksto the well-being of animals have devastating im-plications for the rural poor and for the overallMongolian economy. Major shocks are commonas Mongolia has a harsh climate, and animals areherded with limited shelter. From 2000 to 2002,eleven million animals perished due to harsh winters(dzuds). The government of Mongolia has struggledwith the obvious question of how to address thisproblem.

The Mongolian government requested specificassistance in coping with extreme livestock losses.Given the nature of highly correlated death ratesfor animals in Mongolia, an index-based livestockinsurance (IBLI) product was proposed and inMay 2005, and the World Bank approved a loanto Mongolia to finance the Index-Based LivestockInsurance Project. This project will support a three-season pilot program in three Mongolian states andincludes a contingent debt facility to serve as amechanism for protecting against extreme lossesduring the pilot. The major objective of the pilotprogram is to determine the viability of IBLI inMongolia, including testing herders’ willingness topay for an IBLI product. The index would pay in-demnities based on adult mortality rates by speciesand by soum (province). By law, Mongolia per-forms a census of animals each year. Elaborate sys-tems are in place to assure the quality of the data.The proposed pilot involves three distinct layers ofrisk: (1) self-retention by the herder; (2) a base in-surance product (BIP) for mortality rates in a cer-tain range; and (3) a disaster response product(DRP) for livestock losses beyond the layer coveredby the insurer.

An index-based insurance program was recom-mended because of significant concerns about themoral hazard, adverse selection, and extreme mon-itoring costs associated with any individual live-stock insurance program in the vast open spaces ofMongolia. Weather index insurance was consid-ered; however, it was determined that the weatherevents contributing to livestock deaths were toocomplex to develop this alternative. The projectwill support continued research to strengthen themortality index by incorporating other indexes, forexample, the Normalized Difference VegetationIndex (NDVI), as a means of establishing a moresecure index for paying losses.

While it is believed that the index-insuranceproduct can be effectively underwritten, signif-icant financial exposure for a nascent insurancemarket with extremely limited access to globalrisk-shifting markets remains among the largestchallenges. Given concerns about financing ex-treme losses, the pilot design involves a syndicatepooling arrangement for companies. Pooling riskamong the insurance companies offers some op-portunity to reduce the exposure for any individ-ual insurer. In the short term, the government ofMongolia will offer a 105 percent stop-loss on thepooled risk of the insurance companies. Herderpremiums go directly into a prepaid indemnity

pool. Insurers must replace the reinsurance costand the exposure above 100 percent for the pre-paid indemnity pool.

In the syndicated pooling arrangement, partici-pants share underwriting gains and losses based onthe share of herder premium they bring into thepool. Each insurer also pays reinsurance costs con-sistent with the book of business they bring into thepool. This gives the reinsurance pool the benefits ofthe pooling arrangement and provides the oppor-tunity to build reserves for the overall activity. Thereinsurance pool pays for the first layer of lossesbeyond the 105 percent stop-loss. Once the re-insurance pool is exhausted, the government ofMongolia can call upon the contingent debt to payfor any remaining losses.

A major advantage of having a prepaid indem-nity pool is that all other lines of the insurance busi-ness are protected from the extreme losses that canoccur from writing a highly correlated agriculturalrisk policy. In the long-term vision, the syndicatewill be well positioned to find risk-sharing partnersin the global community quickly, as the poolingarrangement is both risky and profitable. Reinsurersmight be willing to provide capital and enter quota-share arrangements on that risk. To the extent thatthe risks within the pool are standardized, using thesame measures and procedures, one can also envi-sion this mechanism as a means to securitize the risk.Finally, the design also offers the opportunity totransition the system to the market once it is learnedwhether herders find the BIP an acceptable productand demonstrate a willingness to pay.

The first challenge to the risk transfer structure isthe uncertainty of the livestock mortality index-based on an annual government census of all ani-mals in the country. Several systems are in place tomonitor potential problems during the pilot, for ex-ample, the movement of animals across soum bor-ders. From the perspective of the reinsurer, even thegovernment could have the incentive to tamper withthe data if this data determines the level of reinsur-ance claims. The project seeks to establish systems toverify losses using third-party audits. A second chal-lenge is the sustainability of the proposed poolingmechanism that determines reinsurance premiumsfor each participating insurer using advanced mod-eling procedures. Human capital within the countrymust be developed to perform these duties. Poolingmechanisms generally tend to fail because of collec-tive action problems and high transaction cost. Thechallenge in Mongolia will be to move the poolingmechanism to a private sector entity by the comple-



tion of the pilot; otherwise, if left to the governmentto maintain, the system will likely be unsustainable.

GLOBAL STRATEGYThe Global Index Insurance Facility (GIIF)

Background

The economic growth prospects of developing coun-tries are negatively impacted by external shocks,which create both short- and long-term physicaland financial distress. The lack of coherent andtimely response to shocks, coupled with indirectimpacts on growth and investment, compound thecost of direct physical damage. Uninsured enter-prises do not develop their full earnings potentialbecause they engage in low-risk/low-return ac-tivities to minimize downside risks. Generally,too much capital goes into nonremunerated self-insurance. OECD countries, on the other hand,tend to be better equipped to manage shocks sincethey have larger diversified economies that canwithstand such events and because private assetsare insured. Demand for risk management instru-ments is often frustrated by market gaps and entrybarriers. International reinsurers, for example, require substantial minimum risk amounts: “Thegreatest challenge is not to find capacity, but to finda large enough portfolio to make it worth under-writing” (Tobben 2005).

The GIIF seeks to close the gap between the developing country’s demand for insurance againstsevere shocks at public and private levels and the index insurance markets. The World BankCommodity Risk Management Group (CRMG)already addresses the knowledge gap throughtechnical assistance and the demonstration effectsof pilot transactions, but credit and market gapswill limit its ability to scale up. GIIF would lowerthe entry barrier for international risk transfer bypooling smaller transactions, thereby helping toscale up risk transfer from developing countries.

Present

The European Commission allocated a total of25 million for a commodity risk management

facility and submitted the concept to the CouncilWorking Group of Member States as part of the“conditional billion” package, the final tranche of the Ninth EDF/2003 to 2007. CRMG is putting together a proposal for a Global Index Insurance

Facility (GIIF) that would intermediate weather,disaster, and price risk (all index-based) amongdeveloping country-based primary insurers, gov-ernments, banks, and organized markets. CRMGis in intense dialogue with market makers as tothe risk-taking capabilities of the GIIF, with a focustoward “crowding-in” rather than “crowding-out”the private sector. The facility would consist of a

100m capital investment in a risk-taking entitythat would underwrite global weather, disaster,and price risks in developing and, in particular,the African-Caribbean-Pacific (ACP) countries. Themain objective of the facility would be to achievereturns on equity and build a diverse portfolio ofrisk from developing countries not previouslytransferred to the capital and insurance markets,thereby leveraging private risk transfer. The maindevelopment objective would be to alleviate povertyby facilitating effective disaster insurance and riskreduction, allowing countries and enterprises toprofitably invest resources rather than waste themwith inefficient self-insurance. The GIIF would fur-ther facilitate risk transfer by absorbing transactioncosts for developing country clients through cofi-nancing of premiums, funded separately by EC/ACP funds, and through reinvestment of dividendsby public sponsors.

Types of Risks Underwritten by the GIIF

The GIIF would provide cover for disaster, weather,and price risks by underwriting index-based insur-ance contracts. Index insurance also allows verytimely automatic settlements, which is crucial foreffective disaster response. Price risk managementcontracts will be based on liquid exchange-tradedinstruments, set at market prices. All indexes mustbe objective, transparent, published, and sustain-able; price indexes must be liquid. The GIIF wouldregularly publish insurable indexes.

Exit Strategy

The GIIF seeks to catalyze a commercial marketfor index-based insurance products in develop-ing countries by “crowding in” the private sector.Following GIIF’s start-up phase, it is expected thatthe market for developing country risk will be suf-ficiently developed and competitive to offer riskmanagement products to end-user countries andclients at a reasonable cost. This period could varyfrom seven to ten years.

53

Agricultural producers and other rural residents are often exposed toa variety of biological, geological, and climatic factors that can nega-tively affect household income and/or wealth, as well as tremendousvariability in output and/or input prices. Given this environment,risk-averse individuals often make investment decisions that reducerisk exposure but also reduce the potential for income gains andwealth accumulation. Thus, risk contributes to the “poverty trap”experienced by rural people in many developing countries.

For a variety of reasons (discussed in Chapter 2), markets for trans-ferring these risks are typically either very limited or nonexistent.This “market failure” has stimulated a number of policy responses.Many developed countries have highly subsidized, farm-level agri-cultural insurance programs. Critics argue that, in addition to beingvery expensive, these programs stimulate rent-seeking activity, arehighly inefficient, and may actually increase risk exposure by encour-aging agricultural production in high-risk environments (Chapter 3).Given fiscal constraints in most developing countries, highly subsi-dized, farm-level agricultural insurance programs are not a realisticpolicy option.

Index-based insurance products have been proposed as an alterna-tive risk-transfer mechanism for rural areas in developing countries.While not a panacea for all risk problems, index-based insuranceproducts may prove to be valuable instruments for transferring thefinancial impacts of low-frequency, high-consequence systemic risksout of rural areas (Chapter 4). For a variety of reasons, however, gov-ernment intervention may be required to generate socially optimalquantities of risk transfer. Governments must carefully considerthe extent and nature of any intervention in markets for index-basedinsurance products (Chapter 5). These efforts can be facilitated byWorld Bank policy advice, lending instruments, and monitoring andevaluation systems (see World Bank 2004; 2005b). This chapter setsout policy and operational implications for governments and subse-quently for the World Bank operational agenda.

GOVERNMENT ROLESRisks in rural areas must be managed at the macro-, meso-, and micro-levels. Governments need to (1) understand the country’s rural risk pro-file; (2) quantify the impact of this risk on the economy and revenues;(3) design a rural risk management framework; and (4) implementrisk reduction and risk transfer.44

Potential Roles forGovernments and the

World Bank

7

Identify the Risk Profile for Private andPublic Assets and Business Flows

A natural risk assessment identifies the types ofrisks that affect major private and public assets andeconomic activities in rural areas.45 This assessmentdistinguishes between micro- and macrolevel riskand considers both geographical and seasonal vari-ations. Identification of risks at the microlevel istypically based on household surveys as well asspecific risk surveys. The objective is to understandthe types of risks that affect households and the na-ture of those risks. At the macrolevel, the assess-ment would consider the aggregate economic effectof household risk with a particular focus on gov-ernment budget exposure.

Quantify Risk Impacts at All Levels

Once the major risks have been identified, govern-ments need to quantify the potential impact ofthose risks. What is the magnitude of potential

physical and indirect losses for different types ofassets and economic activities? As represented inFigure 7.1, a variety of indirect business flow lossesoften compound the direct physical losses causedby natural hazards.

Design a Rural Risk Management Framework

Government intervention in risk transfer marketsmust be based on a careful analysis of marketshortcomings and a clear statement of how gov-ernment involvement will address those short-comings (Chapter 5). A well-designed rural riskmanagement framework clearly delineates publicand private roles in the ex ante world of risk re-duction and risk financing and also in the ex postworld of emergency response. This framework takescountry-specific objectives and constraints into ac-count instead of replicating developed country his-torical models (Chapter 3). The objective is to learnfrom these historical examples and then to applythat understanding to country-specific efforts thatincorporate new and innovative risk transfer in-struments (Chapter 4). To plan appropriately, pri-vate decision makers need to know where and howgovernment would intervene at different risk levels.Where a credible and reliable insurance cover is inplace, for example, agricultural enterprises mightintensify production.

Implement a Risk Management Strategy

To be successful, a well-conceived risk managementstrategy must be supported by a credible govern-ment commitment that is sufficiently funded overthe long term. While appropriate government roleswill vary to reflect country-specific circumstances,one strategy might be government intermediationof index-based risk management products madeavailable in international capital and reinsurancemarkets and government creation of infrastruc-ture to support the development and implemen-tation of new private risk management products.

WORLD BANK ROLESThe World Bank can engage in a number of activi-ties that, in coordination with governments, maylead to increased risk-transfer opportunities foragricultural producers and other rural residents indeveloping countries. In general, these activities

54 Managing Agricultural Production Risk54 Managing Agricultural Production Risk

Figure 7.1 Potential Impacts of Natural Hazards

Natural Hazard Risk in the Rural Sector

Direct Losses

Fixed Capital

Public Assets • Government Buildings • Public Infrastructure

Private Assets • Industrial Infrastructure • Residential Infrastructure

Inventories • Staple Foods/Other Crops • Inputs

Indirect Losses

Flow Losses

Loss of Tax Revenue and�Tax Base

Business Interruption Reallocation of Investments

Long Term Consequences Scarce data

(Usually not recorded)

Short Term Consequences Humanitarian Crisis

(Attracts International Attention)

Source: Authors.

Potential Roles for Governments and the World Bank 55

include educational efforts, incorporating risk man-agement into holistic rural development strategies,investment lending operations designed to encour-age the development of risk transfer markets, exante coordination of donor responses to natural dis-asters, and monitoring and evaluation of the per-formance of index insurance instruments.

Building Global Knowledge of the IndexApproach to Agricultural Risk Management

The World Bank is uniquely placed to reach govern-ments and decision makers on all continents. TheWorld Bank, in general, and ARD (the Agricultureand Rural Development department), in particular,can facilitate technology transfer across continents.This economic and sector work of ARD will be dis-seminated outside the World Bank: in fiscal year2006. CRMG is planning Global Distance Learningevents that will have a component on agriculturalrisk management concepts and also two workshopsin two different regions, possibly in connection withweather insurance pilot project launches. Insidethe World Bank, information sharing will takeplace mainly through “brown bag” lunches andworkshops.

Incorporating Risk Management Strategiesinto Rural Development StrategyFormulation and Development PolicyLending Programs

While the World Bank and the IMF have a long his-tory of assisting governments in dismantling un-sustainable mechanisms for managing price risk,this often took place in the absence of alternativerisk management tools or a clear risk managementagenda for deregulated markets. This gap has con-tributed to a breakdown in marketing arrangementsand credit channels, so that these efforts have some-times not produced the projected results (Kherallahet al. 2002). While the task will be neither quick noreasy, the importance of addressing issues of collat-eral policies and institutional development as inte-gral to reform is now widely understood.

While the index-based risk management toolsdiscussed here are not a cure-all, they can help creditinstitutions, producer organizations, and (in somecases) producers to manage production risk directly;by doing so they can help reconnect farmers to out-put and credit markets. In assisting policymakersin the design of a country’s reform programs, the

World Bank should routinely consider how to facil-itate the development of risk management instru-ments and should be prepared to support thisprocess through policy advice and, in some cases,lending operations. Often, this may require reform-ing collateral, macroeconomic, or regulatory policies.Risk management instruments using internationalmarkets, for example, cannot operate properly whileexchange controls are in place. Often, local regula-tions affecting insurance or financial markets alsomust be revised.

Because government or World Bank involve-ment in any risk management program may requiretrade-offs with other means of enhancing rural de-velopment and reducing vulnerability (for example,irrigation, infrastructure, and so on), the programshould be embedded in an overall rural develop-ment strategy so any trade-offs can be carefullyweighed. This will also allow formation of linkageswith other rural development objectives (for exam-ple, rural finance). The overall rural developmentstrategy should take a holistic approach to riskmanagement, recognizing that diversification of in-come sources (remittances, off-farm employment,and others) is often an important means of reduc-ing rural vulnerability. In addition to formal riskmanagement markets, the strategy should considerwhat reforms are needed to encourage income diversification and to allow farmers a full rangeof choices in a functioning marketplace. This mayinclude, for example, market liberalization andprivatization; investments in transportation, com-munication, and market infrastructure; legal rightsguaranteeing market access (especially for womenand ethnic minorities); provision of market infor-mation; and measures to better integrate rural andnonrural labor markets (see Siegel 2005; Lanjouwand Feder 2001; Lloyd-Ellis 1999; and Mead andLiedholm 1998). Attention should also be dedicatedto safety nets designed to minimize the need to liq-uidate productive assets in times of emergency andto be scaled up quickly and efficiently at need (seeJorgensen and Van Domelen 1999; Jutting 1999;and Morduch 1999).

Creating Investment Lending Operationsthat Encourage Risk Management

At the macrolevel, a number of World Bank instru-ments (and those of other donors) exist or are beingexplored that can cushion the fiscal and balance ofpayments adjustments required when countries

face shocks from natural disasters or internationalprice movements of major commodity exports orimports. These include automatic mechanisms toadjust debt service—or even to augment financing—in response to exogenous shocks. (For a full discus-sion, see World Bank 2004; 2005b.)

At the mesolevel, risk management tools can beused to improve the functioning of government so-cial safety net programs, either at central or decen-tralized levels. Index-based insurance instruments,for example, could be used to provide ex antecontingent funding that would allow safety netprograms to expand when they are most needed,without the delays and uncertainties caused by re-liance on budgeting processes or on external aid.Likewise, use of index-based insurance by individ-ual farmers, associations, processors, or rural financeinstitutions would reduce their degree of uncer-tainty and facilitate primary producers’ access tocredit and input markets.

In addition to policy advice, the primary WorldBank tool now being used to support developmentof risk management markets, investment lendingprojects may also be useful in some cases. Examplescan be found in World Bank-facilitated price riskmanagement efforts. In Turkey, for example, a com-modity market development learning and innova-tion loan (LIL) had the objective of first supportingthe development of physical commodity markets,which in the long term could evolve into a domes-tic platform for trading futures contracts. The pro-ject financed the upgrading of testing laboratories,warehouse facilities, and regional market infra-structure, and it provided technical assistance toenhance and harmonize grades and standards forsome commodities, upgrade the warehouse receiptssystem, and improve the operations of the commod-ity market regulatory authority. While there Turkeystill has no domestic futures trading, progress hasbeen made toward the more limited objectives ofestablishing better linkages between producersand buyers and of encouraging forward contract-ing for spot delivery, providing another means ofreducing price risk. In addition, the project has fa-cilitated more efficient price discovery: the pricesfor cotton and wheat determined on two exchangesparticipating in the project now serve as the offi-cial record of domestic market prices for those twocommodities.

Another project being explored focuses on theestablishment of a regional system of weather in-surance in southern Africa (see Box 7.1).

The target of the project, as currently conceptu-alized, would be individual farmers, but a projectlike this could be targeted at the mesolevel as well.Pooling risk at the subregional level (a complexclimate system) can reduce financing requirementsby taking advantage of scale. The subregion as awhole is more attractive to international insur-ance markets (due to risk-spreading) than wouldbe individual countries. Other direct benefits in-clude the faster spread of ideas and the more ef-fective development of capacity made possible bycross-country collaboration and the presence ofpreexisting regional institutions ready to supportproject implementation.

Donor Coordination

Like farmers, governments may suffer from a formof moral hazard. Donor response to catastrophescan reduce the interest of the developing countrygovernment in using markets to shift natural disas-ter risk, as was the case in Nicaragua following theoverwhelming donor response following HurricaneMitch. Donor responses, however, cannot be pre-dicted with certainty and often are not timely.Furthermore, the international community mayoverlook localized disasters, which may devastate acommunity despite having limited impact beyondit. A better solution would be to take advantage ofthese donations in a more structured and ex antefashion. Donors could, for example, contribute to aninsurance pool for the country or region. The WorldBank—particularly the teams in countries espe-cially prone to disasters—can play a leading role inthis through the consultative group process.

A special case of aid in response to disaster isfood aid following a serious drought. Here, theneed for an improved approach is particularlyacute, as in-kind assistance often has counterpro-ductive effects in undermining development oflocal production and marketing channels. Also,aid given ex post in response to droughts is oftenlate in arriving, forcing starving victims to liqui-date productive assets, thus perpetuating a cycleof poverty. Use of an index-based instrument tofund emergency food aid holds the promise of amuch more rapid response, since payment wouldbe triggered by weather events far in advance ofthe actual food shortages, and of far less disruptionof local markets, since food aid agency payoutswould be made in cash that would be used to pro-


Potential Roles for Governments and the World Bank 57

cure food locally, to the extent possible, or to paybeneficiaries directly. The World Bank is collabo-rating with the World Food Program and otherdonors to pilot such an approach.

Monitoring and Evaluation of Transactions

The work on index insurance in developing coun-tries is still in an early stage, and its developmentimpact is not yet proven. A number of assumptionsabout the value of these instruments, their utility at

the farm level, and their development impacts needto be evaluated. CRMG has launched a first base-line study with DECRG (Research Department ofthe World Bank). Generally, utility at the farm levelcan be gauged by the level of take-up of unsubsi-dized and unbundled products and, particularly,the level of repeat buying. Panel studies will revealthe actual impact of these products. Indicators are the level of inputs used and the diversificationof farm activities, particularly the share of cashcrops in the overall portfolio. Another important

Box 7.1 Examples of Potential World Bank Investment Lending Projects to Facilitate Risk Management

Global level

Global Index Insurance Facility: The facility wouldconsist of a capital investment in a risk-taking entitythat would underwrite global weather, disaster, andprice risks in developing countries. The main devel-opment objective would be to absorb costs for initialtransactions for developing country clients throughcofinancing of premiums, funded both separately andthrough reinvestment of dividends by public spon-sors. The main commercial objective of the facilitywould be to generate a modest return to its share-holders through active management of a diverse port-folio of developing country risk not previouslytransferred to the capital and insurance markets. Thefacility would perform several commercial functionsproviding benefits to developing countries.

National level

Infrastructure: Fallback stations, new weather stations,maintenance of weather stations, communicationsequipment for weather services, contract with datavetting services (such as the U.K. Met Office), set-upof weather databases (online), and the cleaning andenhancing of weather data.

Regulatory assessment: Review of legislation,drafting of new regulations, general policy frame-work review, and country-specific policy frameworkreview (including recommendations on subsidy levels, national weather risk funds, basis risk matchingfunds, and so on).

International market/pilot transactions: Travel to international reinsurance market contacts,

technical assistance from international experts (including CRMG), and premium cost-sharing funds.These premium support funds would compensate for the extra premium costs that international andnational insurers add in the infancy stages of theproduct and as a result of data uncertainty. Thesepremium support funds would be phased out as volumes increased and as the extra costs for pre-miums declined.

Knowledge transfer: Travel costs, expertise, designof methodologies and tools to quantify risk exposure,underwriting guidelines, manuals, operational systemdevelopment, and study tours.

Financial backing of risk-taking entities: Governmentmediation of catastrophic risk between internationalrisk insurance markets and insurers or other risk takersin the country; governments could either set up sepa-rate risk-taking vehicles or enter into contingent creditagreements with the World Bank to lower annual premium costs.

Regional level

Financial contribution to a regional index insurancefund: Pooling systemic risk at the regional level, signif-icantly lowering premium costs and warranting set-upof a regional risk fund that would insure its membersaccording to sound actuarial rates before it lays off riskin international markets.

Climate prediction and forecasting technologies:Can be cost effectively rolled out only at a regionallevel that achieves economies of scale and enforcescollaboration.

Source: Authors.

linkage will be to gauge whether index insuranceproducts improve access to credit or improve theterms of credit for small farmers in developingcountries. Both the Indian and the Mongolian pilotproject have very explicit monitoring and evalua-tion components that will attempt to gauge theseactivities.

As with any innovation, index insurance prod-ucts for agricultural production risk will go throughsome significant changes in the next few years. It islikely that we will learn that they work under some

circumstances and not under others. Mistakes willbe made. Learning from those mistakes will requirecareful evaluation and subsequent adjustments. Atthis stage, the key value added from index insur-ance products appears to be the opportunity forstructured ex ante financing of catastrophic risk tiedto highly correlated losses resulting from weatherrisk in agriculture. Such risk cannot be pooled at thelocal level, and the special structures introduced inthis ESW give hope that they can be shifted intoglobal markets.


59

Ahsan, S. M., A. A. G. Ali, and N. J. Kurian. 1982. “Towards a Theory of Agricultural Insurance.” American Journal ofAgricultural Economics 3: 520–29.

Anderson, J. R. 2001. “Risk Management in Rural Development:A Review.” Rural Development Strategy BackgroundPaper 7, Rural Development Department, The World Bank,Washington, D.C.

Babcock, B., C. Hart, and A. Stoppa. 2004. “La gestione del rischio nell’agricoltura Usa.” In La Gestione Del Rischio inAgricoltura: Strumenti E Politiche, ed. F. Filippis. Rome,Italy: Coldiretti.

Barnett, B. J. 1999. “U.S. Government Natural DisasterAssistance: Historical Analysis and a Proposal for the Future.”Disasters: The Journal of Disaster Studies, Policy & Management23: 139–55.

Barnett, Barry J., J. Roy Black, Yingyao Hu, and Jerry. R. Skees.2005. “Is Area-Yield Insurance Competitive with Farm-YieldInsurance?” Journal of Agricultural and Resource Economics30 (2): 285–301.

Burgaz, F. J. 2004. “Il sistema delle assicurazioni agricole com-binate in Spagna.” In La Gestione Del Rischio in Agricoltura:Strumenti E Politiche, ed. F. Filippis. Rome, Italy: Coldiretti.

Buzby, J. C., P. L. Kenkel, J. R. Skees, J. W. Pease, and F. J.Benson. 1994. “A Comparison of Subjective and HistoricalYield Distributions with Implications for Multiple-Peril CropInsurance.” Agricultural Finance Review 54: 15–23.

Byerlee, D. 2005. Agricultural Growth for the Poor: An Agenda forDevelopment. Directions in Development Series. Washington,DC: World Bank.

Carney, D., M. Drinkwater, T. Rusinow, K. Neefjes, S. Wanamali,and N. Singh. 1999. “Livelihood Approaches Compared: ABrief Comparison of the Livelihoods Approaches of DFID,CARE, Oxfam, and UNDP.” Department for InternationalDevelopment (DFID), London.

Chambers, R., and G. Conway. 1992. “Sustainable RuralLivelihoods: Practical Concepts for the 21st Century.” IDSDiscussion Paper 296, Institute for Development Studies,Brighton, U.K.

Deaton, A., and F. Grimard. 1992. “Demand Analysis and TaxReform in Pakistan.” The World Bank Living Standards andMeasurement Survey Working Paper, No. 85, The WorldBank, Washington, D.C.

Dercon, S., ed. 2005. “Insurance against Poverty.” UNU-WIDERStudies in Development Economics. Oxford: Oxford UniversityPress.

———. 2002. “Income Risk, Coping Strategies, and Safety Nets.”The World Bank Economic Research Observer 17 (Fall).

Dischel, R. 2001. “Double Trouble: Hedging.” Weather Risk:A Special Report. Risk Magazine and Energy and Power RiskManagement August: 24–26.

Dismukes, R., G. Allen, and B. J. Morzuch. 1989. “Participationin Multiple-Peril Crop Insurance: Risk Assessments and Risk

Preferences of Cranberry Growers.” Northeastern Journal ofAgricultural and Resource Economics 18 (October): 109–17.

Doherty, N. A. 1997. “Financial Innovation in the Managementof Catastrophic Risk.” Paper presented at ASTIN/AFIRConference, Australia, August.

Fafchamps, M. 1993. “Sequential Labor Decisions underUncertainty: An Estimable Household Model of West AfricanFarmers.” Econometrica 61 (September): 1173–97.

Gadgil, S., P. R. Seshagiri Rao, K. Narahari Rao. 2002. “Use of Climate Information for Farm-Level Decision Making:Rainfed Groundnut in Southern India.” Agricultural Systems74: 431–57.

Glauber, J. W. 2004. “Crop Insurance Reconsidered.” AmericanJournal of Agricultural Economics 86 (December): 1179–95.

Goes, A., and J. R. Skees. 2003. “Financing Natural Disaster RiskUsing Charity Contributions and Ex Ante Index Insurance.”Paper presented at the annual meetings of the AmericanAgricultural Economics Association, July 27–30, Montreal,Canada.

Hardaker, J. B., R. B. M. Huirne, J. R. Anderson, and G. Lien.2004. Coping with Risk in Agriculture, 2nd ed. Wallingford,Oxon, U.K.: CABI Publishing.

Harwood, J., R. Heifner, K. Coble, J. Perry, and A. Somwaru.1999. “Managing Risk in Farming: Concepts, Research, andAnalysis.” Economic Research Service (ERS), U.S. Departmentof Agriculture.

Hazell, P. B. R. 1992. “The Appropriate Role of AgriculturalInsurance in Developing Countries.” Journal of InternationalDevelopment 4: 567–81.

Hazell, P. B. R., and J. R. Skees. 1998. “Potential for RainfallInsurance in Nicaragua.” Report to the World Bank, March.

Hess, U. 2003. “Innovative Financial Services for Rural India:Monsoon-Indexed Lending and Insurance for Smallholders.”Agriculture and Rural Development Working Paper 9, TheWorld Bank, Washington, D.C. www.itf-commrisk.org.

Hess, U., and J. Syroka. 2005. “Weather-Based Insurance inSouthern Africa: The Case of Malawi.” Agriculture and RuralDevelopment (ARD) Discussion Paper 13, The World Bank,Washington, D.C.

Hess, U., A. Stoppa, K. Richter. 2003. “Weather Risk Managementfor Agriculture and Agribusiness in Developing Countries.”In Climate Risk and the Weather Market: Financial RiskManagement with Weather Hedges, ed, R. Dischel. London: RiskBooks.

Jorgensen, S. L., and J. Van Domelen. 1999. “Helping the PoorManage Risk Better: The Role of Social Funds.” Paper pre-sented at the IADB Conference on Social Protection andPoverty, February.

Jutting, J. 1999. “Strengthening Social Security Systems in RuralAreas of Developing Countries.” ZEF Discussion Paper onDevelopment Policy. Center for Development Research,Bonn, Germany.

References

Kherallah, M., C. Delgado, E. Gabre-Madhin, N. Minot, andM. Johnson. 2002. Reforming Agricultural Markets in Africa.Baltimore: Johns Hopkins University, Press for the Inter-national Food Policy Research Institute.

Kunreuther, H., R. M. Hogarth, and J. Meszaros. 1993. “InsurerAmbiguity and Market Failure.” Journal of Risk and Uncertainty7: 71–87.

Kunreuther, H., and M. Pauly. 2001. “Ignoring Disaster, Don’tSweat Big Stuff.” Wharton Risk Center Working Paper 01-16-HK, Wharton Business School, University of Pennsylvania.

Kunreuther, H., J. Meszaros, R. M. Hogarth, and M. Sprance.1995. “Ambiguity and Underwriter Decision Processes.”Journal of Economic Behavior and Organization 26: 337–52.

Kurosaki, T., and M. Fafchamps. 2002. “Insurance MarketEfficiency and Crop Choices in Pakistan.” Journal ofDevelopment Economics 67 (April): 419–53.

Lanjouw, P., and G. Feder. 2001. “Rural Non-Farm Activitiesand Rural Development: From Experience toward Strategy.”Rural Strategy Discussion Paper 4, The World Bank,Washington, D.C.

Lim, Y., and R. Townsend. 1998. “General Equilibrium Modelsof Financial Systems: Theory and Measurement in VillageEconomies.” Review of Economic Dynamics 1: 59–118.

Little, P. D., A. P. Castro, M. P. Stone, W. Negatu, and T. Mogues.2004. “Moving in Place, Drought and Poverty Dynamics inSouth Wollo, Ethiopia.” BASIS CRSP Policy Conference:Combating Persistent Poverty in Africa, November 15–16,Four Points Sheraton Hotel, Washington, D.C.

Lloyd-Ellis, H. 1999. “Endogenous Technological Change andWage Inequality.” American Economic Review 89: 47–77.

Mahul, O., and J. R. Skees. 2005. “Managing AgriculturalCatastrophic Risk at the Country Level: The Case of LivestockMortality Risk in Mongolia.” Working Paper, The WorldBank, Washington, D.C.

Marshall, J. F. 1989. Futures and Options Contracting: Theory andPractice. Cincinnati, OH: South-Western Publishing Company.

Mead, D. C., and C. Liedholm. 1998. “The Dynamics of Microand Small Enterprises in Developing Countries.” WorldDevelopment 26: 61–74.

Miller, D., and A. Paulson. 2000. “Informal Insurance and MoralHazard: Gambling and Remittances in Thailand.” Contributedpaper No. 1463, Econometric Society World Congress.http://econpapers.repec.org/paper/ecmwc2000/1463.htm.

Miranda, M. J., and J. W. Glauber. 1997. “Systemic Risk,Reinsurance, and the Failure of Crop Insurance Markets.”American Journal of Agricultural Economics 79 (February): 206–15.

Morduch, J. 1999. “The Microfinance Promise.” Journal of EconomicLiterature 37: 1569–1614.

———. 1995. “Income Smoothing and Consumption Smoothing.”Journal of Economic Perspectives 3: 103–14.

Morgan, M., and M. Henrion. 1990. Uncertainty: A Guide toDealing with Uncertainty in Quantitative Risk and PolicyAnalysis. Cambridge: Cambridge University Press.

Orlove, B., J. Chiang, and M. Cane. 2002. “Ethnoclimatology inthe Andes: A Cross-Disciplinary Study Uncovers a ScientificBasis for the Scheme Andean Potato Farmers TraditionallyUse to Predict the Coming Rains.” American Scientist 5: 428–35.

Pease, J., E. Wade, J. R. Skees, and C. M. Shrestha. 1993.“Comparisons between Subjective and Statistical Forecasts ofCrop Yields.” Review of Agricultural Economics 15: 339–50.

Peru, Government of. Hoja de Ruta (Road Map) 2004–06.Pikor, G., and A. Wile. 2004. “L’esperienza canadese nella ges-

tione del rischio in agricoltura.” In La Gestione Del Rischio inAgricoltura: Strumenti E Politiche, ed. F. Filippis. Rome, Italy:Coldiretti.

Pomareda, C. 1986. “An Evaluation of the Impact of CreditInsurance on Bank Performance in Panama?” In Crop Insurance

for Agricultural Development: Issues and Experience, eds. P. B. R.Hazell, C. Pomareda, and A. Valdes. Baltimore: Johns HopkinsUniversity Press.

Priovolos, T., and R. C. Duncan, eds. 1991. Commodity RiskManagement and Finance. Washington, D.C.: Published for theWorld Bank by Oxford University Press.

Ramamurtie, S. 1999. “Weather Derivatives and Hedging WeatherRisks.” In Insurance and Weather Derivatives: From Exotic Optionsto Exotic Underlyings, ed. H. Geman. London: Risk Books.

Rejda, G. E. 2001. Principles of Risk Management and Insurance,7th ed. Boston: Addison Wesley Longman.

Rosenzweig, M. 1988. “Risk, Implicit Contracts and the Familyin Rural Areas of Low-Income Countries.” Economic Journal98 (December): 1148–70.

Rosenzweig, M., and H. P. Binswanger. 1993. “Wealth, WeatherRisk and the Composition and Profitability of AgriculturalInvestments.” Economic Journal 103 (January): 56–78.

Rosenzweig, M., and O. Stark. 1989. “Consumption Smoothing,Migration, and Marriage: Evidence from Rural India.” Journalof Political Economy 97: 905–26.

Rosenzweig, M., and K. Wolpin. 1993. “Credit Market Constraints,Consumption Smoothing and the Accumulation of DurableProduction Assets in Low-Income Countries: Investmentsin Bullocks in India.” Journal of Political Economy 2: 223–44.

Rossi, P., J. Wright, and E. Weber-Burdin. 1982. Natural Hazardsand Public Choice: The State and Local Politics of Hazard Mitigation.New York: Academic Press.

Ruck, T. 1999. “Hedging Precipitation Risk.” In Insurance andWeather Derivatives: From Exotic Options to Exotic Underlyings,ed. H. Geman. London: Risk Books.

Schade, C., H. Kunreuther, and K. Kaas. 2002. “Low-ProbabilityInsurance Decisions: The Role of Concern.” Wharton RiskCenter Working Paper No. 02-10-HK, Wharton BusinessSchool, University of Pennsylvania.

Shiller, R. J. 2003. The New Financial Order: Risk in the 21stCentury. Princeton: Princeton University Press.

Siegel, P. B. 2005. “Looking at Rural Risk Management Using anAsset-Based Approach.” Paper prepared for the CommodityRisk Management Group, Agricultural and Rural Develop-ment Department, ESW, The World Bank, Washington, D.C.

Skees, J. R. 2003. “Risk Management Challenges in RuralFinancial Markets: Blending Risk Management Innovationswith Rural Finance.” Thematic papers presented at theUSAID Conference: Paving the Way Forward for RuralFinance: An International Conference on Best Practices, June2–4, Washington, D.C.

———. 2001. “The Bad Harvest: More Crop Insurance Reform:A Good Idea Gone Awry.” Regulation: The CATO Review ofBusiness and Government 24: 16–21.

———. 1999b. “Opportunities for Improved Efficiency in RiskSharing Using Capital Markets.” American Journal ofAgricultural Economics 81: 1228–33.

———. 1999a. “Agricultural Risk Management or IncomeEnhancement?” Regulation: The CATO Review of Business andGovernment 22: 35–43.

Skees, J. R., and B. J. Barnett. 1999. “Conceptual and PracticalConsiderations for Sharing Catastrophic/Systemic Risks.”Review of Agricultural Economics 21: 424–41.

Skees, J. R., B. J. Barnett, and J. Hartell. 2005. “Innovations inGovernment Responses to Catastrophic Risk Sharing forAgriculture in Developing Countries.” Paper prepared forthe Commodity Risk Management Group, Agricultural andRural Development Department, ESW, The World Bank,Washington, D.C.

Skees, J. R., J. R. Black, and B. J. Barnett. 1997. “Designing andRating an Area-Yield Crop Insurance Contract.” AmericanJournal of Agricultural Economics 79: 430–38.


References 61

Skees, J. R., and U. Hess. 2003. “Evaluating India’s Crop FailurePolicy: Focus on the Indian Crop Insurance Program.” Paperdelivered to the South Asia Region of the World Bank,November.

Skees, J. R., P. B. R. Hazell, and M. Miranda. 1999. “NewApproaches to Public/Private Crop-Yield Insurance.” EPTDDiscussion Paper No. 55, International Food Policy ResearchInstitute, Washington, D.C. www.cgiar.org/ifrpi.

Skees, J. R., and M. J. Miranda. 1998. “Rainfall Risk Contracts forNicaragua.” Report to the World Bank, Washington, D.C.

Skees, J. R., P. Varangis, D. Larson, and P. Siegel. 2005. “CanFinancial Markets Be Tapped to Help Poor People Cope withWeather Risks?” In Insurance against Poverty, ed. S. Dercon.Oxford: UNU-WIDER Studies in Development Economics,Oxford University Press.

Stoppa, A., and U. Hess. 2003. “Design and Use of WeatherDerivatives in Agricultural Policies: The Case of Rainfall-Index Insurance in Morocco.” Paper prepared for theInternational Conference on Agricultural Policy Reform andthe WTO: Where Are We Heading? June 23–26, Capri, Italy.

Swiss Re. 2004. “Understanding Reinsurance: How ReinsurersCreate Value and Manage Risk.” Technical Publishing Series,Swiss Reinsurance Company, Zurich, Switzerland. www.swissre.com.

Tobben, B. (Partner Re). 2005. “Weather Risk Markets forAgriculture in Central America.” Paper presented at the

World Bank CRMG and Inter-American Development Bankworkshop on Innovative Agricultural Insurance for CentralAmerica, May 9–11, Antigua.

Townsend, R. 2005. “Weather Insurance in Semi-Arid India.”Paper prepared for the Commodity Risk ManagementGroup, Agricultural and Rural Development Department,ESW, The World Bank, Washington, D.C.

Turvey, C. G. 2002. “Insuring Heat Related Risks in Agri-culture with Degree-Day Weather Derivatives.” Paper pre-sented at the AAEA Annual Conference, July 28–31, LongBeach, CA.

Walker, T. S., and J. G. Ryan. 1990. Village and HouseholdEconomies in India’s Semi-Arid Tropics. Baltimore: The JohnsHopkins University Press.

World Bank. 2005b. “Managing the Debt Risk of ExogenousShocks in Low-Income Countries.” The World Bank,Washington, D.C.

———. 2005a. “Rural Finance Approach.” ESW report preparedby the Agricultural and Rural Development Department, TheWorld Bank, Washington, D.C.

———. 2004. “Exogenous Shocks in Low-Income Countries:Policy Issues and the Role of the World Bank.” Report pre-pared for technical briefing of the Board, ARD, PREM, andFRM, No. 9, March 9, The World Bank.

———. 2001. World Development Report 2000/2001: AttackingPoverty. Washington, D.C.: The World Bank.

63

The emerging weather risk market offers new riskmanagement tools and opportunities for agricul-ture. The aim of this appendix is to illustrate howan end user in the agricultural industry could usea market-based solution to mitigate the financialimpact of weather on its business operations. Theappendix draws information from the wealth ofliterature written on the subject of weather riskmanagement to provide the reader with a step-by-step guide to how weather risk management in-struments could be developed for and used in theagricultural sector. After discussing the financialimpact of weather on agriculture, this Appendix ex-plores the key steps required to structure a weatherrisk management solution, from identifying therisk to execution. Also discussed are the pricing ofweather risk management instruments, with a briefoverview of how the weather market approachesand values weather risk and the implications for theend user. Finally, the Appendix treats the prerequi-sites for weather risk management instruments: theweather data used to construct weather indexes andsettle contracts and the data cleaning and analy-sis necessary when pricing and structuring a poten-tial transaction. Selected references suggest furtherreading on weather risk management.

THE FINANCIAL IMPACT OF WEATHERWeather risk impacts individuals, corporations,and governments with varying degrees of frequency,severity, and cost. Around the world, people face thevagaries of the weather on a daily basis. The mediacontinually reports catastrophic weather events—floods, hurricanes, and droughts—that impact indi-viduals’ property, health, and lives. Consequently,governments are also financially exposed to weatherrisk. They are called upon to provide direct finan-cial, nutritional, and housing support to their citi-

zens in the event of weather-related disasters andmust increase spending for rehabilitation and re-construction of infrastructure and assets as a resultof damage incurred. Moreover, the economy of acountry is also at risk to weather through businessinterruption, supply shocks, diversion of domesticinvestment from productive activities to mitigationof the disasters’ impacts and, for some countries, areduction in foreign investment in the aftermath ofan extreme weather-related event. While often sucheffects are reversible and short-term, the impact onthe economy of a poor country can be significantand long lasting. Between 1997 and 2001, the averagedamage per natural disaster in low-income countrieswas 5.8 percent of GDP (IMF, 2003). Evidence fromsixteen Caribbean countries shows, for example, thatone percentage point of GDP in direct damage fromnatural disasters can reduce GDP growth by half apercentage point in the same year (Auffret 2003).Furthermore, the humanitarian cost of weather-related disasters is also greater in the developingworld: approximately 80 percent of all fatalitiesdue to weather disasters between 1980 and 2003 oc-curred in the “uninsured world,” comprising pre-dominantly low-income countries (Loster 2004).

Even noncatastrophic weather events have a fi-nancial impact. The U.S. Department of Commerceestimates that nearly one-third of the U.S. econ-omy, or US$1 trillion (U.S. Congress 1999) is mod-ulated by the weather, and that up to 70 percent ofall U.S. companies are weather sensitive. Weatherrisk can impact a business through its overall prof-itability or simply through the success or failure ofan initiative as a consequence of the weather. Likegovernments, businesses can face both demand-and supply-driven weather risks. Energy compa-nies, for example, can be exposed to demand-drivenweather risk. In the event of a warmer than averagewinter, for instance, gas companies, in particularthose dealing with domestic customers, face a po-tential drop in gas sales as customers use less gas

Appendix 1Weather Risk Management

for Agriculture46

than expected to heat their homes. Therefore, evenif the company has adhered to prudent price riskmanagement practices by protecting their salesmargin from fluctuations in the gas supply price,weather-driven demand fluctuations can lead toa drop in sales volume below expected levels thatsignificantly affects budgeted revenues. A supply-side example of weather risk can be found in theconstruction industry. Because building materi-als have specific weather requirements, cold andwet weather conditions can impact constructionprogress; concrete, for example, cannot be pouredin wet or below-freezing conditions. Contractorsmust assume this supply-driven weather risk,which can significantly delay a construction projectand result in hefty penalties if the project is notcompleted on schedule.

Weather has traditionally been the scapegoat inbusiness for poor financial performance (Clemmons2002). Annual reports, financial statements, andpress releases frequently contain declarations suchas, “[c]ooling degree days were 21 percent belowlast year’s quarter and 16 percent below normal.The effects of milder weather compared with lastyear had a negative impact on [earnings before in-terest and taxes] of about $35 million for the quarter”(Duke Energy 2003); “4 cents per share [decline]for lower gas deliveries due to warmer weather in the fourth quarter of 2003” (Energy East 2004);and “Europe’s performance continued to be im-pacted by unfavorable summer weather with vol-ume down 12 percent in the third quarter andyear-to-date volume down 6.5 percent” (Coca-Cola2004). Given such examples, it is not surprising thatthe financial community has begun to seek prac-tical solutions to controlling the financial impactof weather. Centrica Plc, for example, one of thelargest domestic gas suppliers in Great Britain, isone of a number of utilities that has chosen to man-age its weather risk in order to “protect the com-pany against variability in earnings of its gas retailbusiness due to abnormal winter temperatures inthe UK” (Ulrich 2002), and it has been doing so since1998. London-based Corney and Barrow Wine BarsLimited deploys several weather hedges to providefinancial protection against cool summers resultingin poor customer patronage: “After the exceptionalsummer of 2003 Corney and Barrow was keen tosecure protection against the possibility of the re-verse experience [in 2004]” (XL Trading 2004). Withthe emergence of a market for weather risk man-agement products, a business can now be protected

from such ancillary risks that create unpredictableearnings streams. Just as interest rate and currencyrisks are currently managed through market-basedsolutions, weather risks that increase business un-certainty can now be neutralized, allowing a com-pany to focus on its core business and to protectearnings per share forecasts and growth.

THE WEATHER MARKETIn 1997, a formal weather risk market was born inthe United States through the first open market de-rivative transaction indexed to weather. Motivatedby the deregulation of the energy industry, whichled to the break-up of regulated monopolies in elec-tricity and gas supply, the nascent weather marketresponded to energy companies’ need to increaseoperational efficiency, competitiveness, and share-holder value. In 1996, the Kansas-based energy com-pany, Aquila, entered into a transaction with NewYork-based Consolidated Edison that combined tem-perature and energy indicators, protecting the lat-ter against a cool August that would reduce powersales. The first publicized transaction in 1997, how-ever, was between energy companies Koch Energyand Enron. Additional deals soon followed, withother energy market participants wanting protec-tion against risks, primarily temperature, associatedwith volumetric fluctuations in energy.

In 2001, the Weather Risk Management Asso-ciation (WRMA)—the industry body—commis-sioned PricewaterhouseCoopers (PWC) to conducta survey of weather risk contracts executed amongWRMA members and survey respondents fromOctober 1997 to March 2001 and since then on anannual basis. Since 1997, the survey has shown thatover US$20 billion has been transacted throughthe weather risk market47: the market has grown toaround US$4.6 billion outstanding risk for the yearApril 2003 to March 2004 (PWC 2003; 2004; seeFigure A1.1), although some believe this to be anunderestimate.48 Active trading occurs in U.S.European, and Japanese cities (Figure A1.2); mostnotable among the few transactions occurring out-side these three main trading hubs are agriculturaltransactions in Mexico, India, and South Africa. Themarket has also evolved to include nonenergy appli-cations. Survey respondents, when asked to list re-quests received from potential end users of weatherrisk management products, identified end users inthe retail, agriculture, transport, and leisure andentertainment industries (Figure A1.3), although


Appendix 1. Weather Risk Management for Agriculture 65

energy still contributes approximately 56 percentof the potential weather risk management end usermarket. As a result of this expansion, the markethas also broadened its product offering to includetransactions on nontemperature indexes49 such asrainfall, wind, and snow.

Today, the key market participants include(re)insurers, investment banks, and energy compa-nies. (Re)insurers and investment banks provideweather risk management products to end usercustomers—such as Corney and Barrow Wine BarsLimited and Centrica Plc—and form the primarymarket; all three participate in a secondary mar-ket in which players transfer weather risk amongthemselves through over-the-counter (OTC) finan-cial transactions and exchange-based derivative con-tracts on the Chicago Mercantile Exchange (CME)50

to diversify and hedge their portfolios.Weather risk management is also being intro-

duced to the developing world through the work oforganizations such as the World Bank CommodityRisk Management Group (CRMG) and the UnitedNations World Food Program (WFP). The WorldBank was involved in the first index-based weatherrisk management program—in India in June 2003—and it is currently working on several projectsaround the world. The small pilot program waslaunched by Hyderabad-based microfinance insti-tution BASIX and the Indian insurance companyICICI Lombard, in conjunction with CRMG, when230 groundnut farmers in Andhra Pradesh boughtweather insurance to protect against low monsoonrainfall (Hess 2003). Currently the WFP, in con-junction with the World Bank, is investigating thefeasibility of weather-based insurance as a reliable,timely, and cost-effective way of funding emer-gency operations in countries such as Ethiopia (TheEconomist 2004). Work is also underway to see ifdeveloping country governments in southern Africacan benefit from weather risk management productsand strategies (Hess and Syroka 2005). The globalweather-risk market is particularly interested inthese types of transactions, as they provide muchsought after diversification to their books throughnew locations and risks.

WEATHER RISK ANDAGRICULTUREOne of the most obvious applications of weatherrisk management products, weather insurance orweather derivatives is in agriculture and farming.

Indeed 13 percent (PWC 2004) of the end user re-quests in the weather market are now focused on theagricultural sector (Figure A1.3). Weather affectsmany aspects of the agricultural supply and demandchain. From the supply side, weather risk manage-ment can help control both production or yield riskand quality risk.

Technology plays a key role in production riskin farming. The introduction of new crop varietiesand production techniques offers the potential forimproved efficiency; however, agriculture is alsooften affected by many uncontrollable events re-lated to weather—including excessive or insuffi-cient rainfall, hail, extreme temperatures, insects,and diseases—that can severely impact yields andproduction levels. Countless examples can be givenon the impact of cold temperatures on deciduousfruit (Guaranteed Weather 2005b), deficit rainfallon wheat (Stoppa and Hess 2003), excess rainfallon potato yields (Meuwissen et al. 2000), and even

Figure A1.1 Notional Value of All Weather Contracts in US$

0

500

1,000

1,500

2,000

2,500

3,000

3,500

4,000

4,500

5,000

1998/1999

1999/2000

Source: Author’s figures, using PricewaterhouseCoopers industry survey data from 2003 and 2004.

2000/2001

2001/2002

2002/2003

2003/2004

Weather trading year

Not

iona

lval

ue($

US

mill

ion)

CME contractsNon-CME contracts

temperature stress on cattle and thus dairy produc-tion (Guaranteed Weather 2005a). In 2003, 59 per-cent of Ukraine’s winter grain crop was destroyeddue to winterkill temperatures (USDA 2003) and40 to 50 percent of northeastern England’s oil rape-seed crop was lost due to excessive rain at harvestin August 2004 (BBC 2004). The costs associatedwith drops in expected or budgeted production dueto such uncontrollable factors can have a signifi-cant impact on a producer’s revenues and contrac-tual obligations. A producer may seek protectionagainst adverse weather conditions affecting cropyield. Weather can also impact the quality, if not theabsolute production levels, of a crop (GuaranteedWeather 2005c).

On the demand side, weather also affects relatedagricultural products through the use of pesticides,fertilizers, and herbicides. Agricultural chemicalproducers, for example, can use weather risk man-agement instruments to hedge against the costsassociated with fluctuations in the demand forchemicals by farm operators. The cotton boll weevil,for example, which costs cotton producers in the


Figure A1.2 Percentage of Total Weather Contracts by Location (excluding CME trades)

0

10

20

30

40

50

60

70

80

90

100

1998/1999

Source: Author’s figures, using Pricewaterhousecoopers industry survey data from 2003 and 2004.

1999/2000 2000/2001 2001/2002 2002/2003 2003/2004 Weather trading year

Perc

enta

geof

tota

lcon

trac

ts(E

xc.C

ME

cont

ract

s)

OtherEuropeAsiaNA EastNA Mid-WestNA SouthNA West

Figure A1.3 Potential End User Market by Economic Sector 2003–2004

Source: Author’s figures, using PricewaterhouseCoopers industry survey data from 2004.

Energy56%

Other11%

Transportation4%

Construction7%

Retail9%

Agriculture13%


United States US$300 million per year,51 is a weathersensitive pest; its numbers vary from year to yearlargely due to the severity of the winter. In ex-tremely cold winters, weevil numbers drop sig-nificantly, directly affecting the net earnings of anagrochemical company. Chemical producers couldhedge their earnings volatility caused by fluctua-tions in pesticide sales by purchasing a weather riskmanagement instrument specifically indexed to thephenology of the pests their products target.

Index-based weather insurance is a relativelynew product, and the use of weather risk manage-ment products in the agricultural sector is still in itsinfancy, with very few publicized transactions inthe United States and Europe. A number of agri-cultural transactions have occurred outside of themain weather market trading hubs, however, mostnotably in Canada (Ontario—maize; Alberta—forage), Argentina (Sancor—dairy), South Africa(Gensec Bank—apple cooperative freeze cover),and India (ICICI Lombard—groundnut, cotton,coriander, and orange). Given weather is one ofthe biggest risks faced by farmers, weather-indexedrisk management products have been suggested asa potential alternative to the traditional crop insur-ance programs for smallholder farmers in the emerg-ing markets.

STRUCTURING A WEATHER RISKMANAGEMENT SOLUTIONDeveloping a successful weather risk managementand transfer program for agriculture involves fouressential steps:

• Identifying significant exposure of an agricul-tural grower/producer to weather;

• Quantifying the impact of adverse weather ontheir revenues;

• Structuring a contract that pays out whenadverse weather occurs; and

• Executing the contract in optimal form to trans-fer the risk to the international weather market.

Each of the steps is outlined in the following foursubsections, and they are fully explored in the casestudies in the next appendix.

Identifying the Risk

Identifying weather risk for an agricultural groweror producer involves three steps: identifying the re-

gions at risk from weather and the weather stationsthat reflect that risk; identifying the time periodduring which risk is prevalent; and identifyingthe weather index providing the best proxy for theweather exposure. This last step is the most criticalin designing an index-based weather risk manage-ment strategy. Rather than measuring the actualimpact on crop yields—or related fluctuations indemand, supply, or profitability—the index acts asa proxy for the loss experienced due to weather andis constructed from actual observations of weatherat one or more specific weather stations.

Location and Duration

All weather contracts are based on the actual ob-servations of weather variables at one or more spe-cific weather stations. Transactions can be based onobservations from a single station or a basket ofseveral stations or on a weighted combination ofreadings from multiple stations. (More informationon the weather station and data requirements forweather risk management instruments appearsbelow.) If an individual farmer is interested in pur-chasing weather protection for his particular crop,the index-based weather contract must be writtenon the weather station nearest the farmer’s land toprovide the best possible coverage for the farmerclient. A larger grower, with several production re-gions, may be more interested in purchasing aweather contract based on several weather stationsto reflect the weather conditions in all areas cov-ered by the business. The grower’s risk manage-ment strategy can be either to purchase a weathercontract on each of the identified weather stationsor to purchase a single contract on a weighted aver-age of several stations, with the weightings chosento reflect the importance of the different stations tothe overall weather exposure of the business. Theapproach chosen depends on the risk preferencesand risk retention appetite of the grower, althoughweighting is generally the cheaper and more effi-cient approach. Retaining localized risks will mostprobably be a more cost-effective solution thanwould transferring them to a third-party, while stillproviding protection in situations where adverseweather affects several regions and involves theoverall production portfolio of a producer. The lat-ter approach will also reduce the risk of reliance onone weather station and hence the associated issueof basis risk,52 covered below.

All contracts have a defined start and end dateto limit the period over which the underlying index

is calculated. This calculation period describes theeffective dates of the risk protection period duringwhich relevant weather parameters are measuredat the specified weather stations. For agriculturalend users, the duration of the weather contracts willbe determined by the specific requirements of theirbusiness. Contract duration is flexibility to addressindividual end-user business exposures; contractscan be weekly, monthly, seasonal, and even multi-annual. Final settlement of the weather contractstypically occurs up to forty days after the end ofthe calculation period, once the collected weatherdata have been cross-checked and quality con-trolled by the relevant data-collecting body, usu-ally the National Meteorological Service.53

Underlying Indexes

A weather index can be constructed using any com-bination of measurable weather variables and anynumber of weather stations that best represent therisk of the agricultural end user. Common variablesinclude temperature and rainfall, although transac-tions on snowfall, wind, sunshine hours, river flow,relative humidity, and storm/hurricane locationand strength are also possible and are becomingmore frequent. Unlike energy indexes, in which therelationship between energy demand and weather ismore transparent and is linked primarily to temper-ature, weather indexes for agriculture demonstratemore complex, albeit still quantifiable, relationshipsbetween crop yields or pesticide use.

The normal process for designing an index-basedweather insurance contract for an agriculturalgrower, for example, involves identifying a mea-surable weather index strongly correlated to cropyield rather than measuring the yield itself. Aftergathering the weather data, an index can be de-signed by (1) looking at how the weather variableshave or have not influenced yield over time; (2) dis-cussing key weather factors with experts, such asagrometeorologists and farmers; and/or (3) refer-ring to crop growth models using weather vari-ables as inputs for yield estimates or phenologymodels illustrating how weather variations relateto pest development. A good index must accountfor the susceptibility of crops to weather factorsduring different stages of development, the biolog-ical and physiological characteristics of the crop,and the properties of the soil. If a sufficient degreeof correlation is established between the weatherindex and crop yield or quality, a farmer or an agri-cultural producer can insure his production or qual-

ity risk by purchasing a contract that pays if a spec-ified undesirable weather event occurs or a spec-ified desirable weather fails to occur. The indexpossibilities are limitless and flexible to match theexposure of the agricultural grower or producer, aslong as the underlying data are of sufficient qual-ity. A few examples of weather indexes for specificagricultural exposures appear below. Although theexamples are based on temperature and precipita-tion, the principles apply to all weather parametersrecorded by ground-based meteorological weatherstations. More examples are given in the case studiesin Appendix 2.

Example 1: Growing Degree Days

Growing Degree Days (GDDs) is a common indexused in the agricultural sector, similar to HDDs andCDDs in the energy sector. GDDs are a measure-ment of the growth and development of plants(both crops and weeds) and insects during a grow-ing season. Organisms that cannot internally regu-late their own temperature are dependent on thetemperature of the environment to which they areexposed. Development of an organism does notoccur unless the temperature is above a minimumthreshold value, known as the base temperature,and a certain amount of heat is required for devel-opment to move from one stage to the next. Thebase temperature varies for different organisms andis determined through research and scientific con-siderations. A GDD is calculated by the followingequation:

where L is the baseline temperature and Taverage is thedaily mean temperature, defined as the average ofthe daily maximum (Tmax) and minimum (Tmin) tem-peratures. If this average is greater than the thresh-old temperature L, the GDD accumulated for thatday is the threshold temperature minus the dailyaverage temperature. If the daily average tempera-ture is less than the base temperature, then the GDDfor that day is zero. Adding the GDD values of con-secutive days gives the accumulated GDDs over a specific period. Accumulated GDDs are a goodproxy for establishing the development stages ofa crop, weed, or insect and can give an indicationas to the development and maturity of a crop or theproper scheduling of pesticide or herbicide appli-

Daily GDD T L

T T T

average

average

� = −( )= −

max

max

0, ;

mmin( ) 2 1( )



cations. Measuring the amount of heat accumu-lated over time provides a physiological time scalethat is biologically more accurate than are calendardays (Neild and Newman 2005), and specific or-ganisms, pest or plant, need different accumulatedGDDs to reach different stages of development. Bycomparing accumulated GDD totals with those ofprevious years, it can be seen if a normal amount ofheat energy has been made available to a crop. Ingeneral, assuming adequate moisture supplies areavailable, the total GDDs received by the end of thegrowing season are often related to crop yield, andtherefore GDDs can be a good index for crop pro-duction. The cumulative temperature index can beused to establish a relationship between GDDs andproduction and thus ultimately with a producer’srevenues.

Example 2: Event-based Indexes

Crop damage can also be the result of specific orcritical temperature events that can be detrimen-tal to yield or quality. Freezing conditions, for in-stance, were reported to have caused more thanUS$600 million in damage to the U.S. citrus crop ina single week of December 1998, with US$300 mil-lion occurring in Tulare County, California, alone(Guaranteed Weather 2005b). Critical temperaturescausing crop damage may vary depending on thelength of time that temperatures remain belowfreezing as well as on the variety, health, and devel-opment stage of a plant. Preventative and proactivemeasures can often be taken to protect crops fromsuch events, but these may have limited impact orbecome more difficult for crops that are farmed inlarge areas, such as cereals and grains.

Winter wheat yields at harvest, for example, de-pend to a great extent on how well the plants sur-vive the winter hibernation period. In the territoryof Kherson, in Ukraine, winter wheat crops havebeen known to die when air and therefore soil tem-peratures fell below a critical level for one day orlonger. These winterkill events cause damage anddeath of the plants’ tillering node: “[with little or nosnow, plants begin to die when] the daily minimumair temperature drops below −16 deg C; [a crop canbe completely lost if this happens for] four daysin a row or in the minimum temperature dropsbelow −21 deg C” (Adamenko 2004). Snow coverconsiderably improves conditions of winter wheathibernation, as the difference between air and soiltemperature increases from 0.5 to 1.1°C per cen-timeter of snow cover. Snow cover on the territory

of Kherson is often unstable, hence complete winterwheat crop failure due to winterkill is a potentialrisk in the southern steppe zone of Ukraine; the cropusually dies in years with no snow cover or whenthe stable snow cover appears late in winter, as itdid in 2003. A winterkill index, based on days whenthe daily minimum temperature is less than −16°C,could therefore be used by a farmer to obtain pro-tection against such crop failure risk. A farmer couldenter into a contract with the recovery of the fullvalue of the crop, as expected under normal weatherconditions, if the recorded daily minimum air tem-perature is less than −16°C for four or more con-secutive days at any time during the winter periodfrom November to March.

Example 3: Deficit Rainfall and Drought

Meteorological drought is usually defined in termsof deviation of precipitation from normal levelsand duration of a region’s dry periods. Agriculturaldrought refers to situations in which soil moisturecontent no longer meets crop growing needs in anarea due to insufficient rainfall. Crops, particu-larly rain-fed crops, often have a minimum overallthreshold of cumulative rainfall necessary for suc-cessful and healthy plant development. Dry beans,for example, can consume up to 368 mm of waterduring the growing season, depending on plant va-riety, soils, climate, and weather conditions (Efetha2002). For dry-land corn farming, 450 to 500 mm ormore of rainfall during the growing season is re-quired for high yields (Neild and Newman 2005).These water requirements must be met by naturalrainfall, stored soil moisture from precipitationprior to the growing season, or supplemental irri-gation. Therefore, a deficit of rainfall below theselevels, in the absence of irrigation, can cause plantmoisture stress that affects development and re-duces yields. A simple cumulative rainfall indexcan be developed to suit a grower’s specific insur-ance requirements with regard to such decreases inrainfall and yield. Looking at historical yield data,for example, can establish an empirical relationshipbetween seasonal cumulative rainfall and yield. Thedistribution of rainfall during the growing seasonor at specific stages of a plant’s development isoften more important than total rainfall, however,and customized indexes must be developed to cap-ture this risk (Stoppa and Hess 2003). Such indexesmay also include other weather parameters, suchas temperature and relative humidity. Crop growthmodels or historical yield data can be used to infer

the empirical relationship between rainfall amountsand yield/quality for specific soil and crop types.

Quantifying the Risk

Once the index has been identified, it must be cali-brated to capture the financial impact of the speci-fied weather exposure as measured by the index.Two approaches are possible at this stage: identify-ing the financial exposure per unit of the definedindex, and/or establishing the limit, the total finan-cial protection, required per risk period, that is, themaximum payout necessary in a worst-case sce-nario. The approach chosen depends on the natureof the underlying index and weather event. If theweather exposure is event driven, for example,such as a Category 5 hurricane hitting a particularlocation or a cold winterkill event destroying anentire wheat crop, the latter approach is more ap-propriate. If the weather exposure is of a cumula-tive nature, such as drought or Growing DegreeDays, the former approach should be chosen. Takinginto consideration the maximum protection requiredper risk period can also inform the financial expo-sure per unit index.

Unit Exposure

After developing weather indexes to capture theimpact of adverse weather conditions on a specificcrop’s yield, it is straightforward to calculate thefinancial impact of these events for producers. Indesigning the index, expert scientific agrometeo-rological assessments, either in conjunction withcrop model output or with verification using his-torical yields, have been employed to construct anunderlying index that best proxies the weather sen-sitivity of the crop in question. Having identifiedthe index, the crop yield, Y, or volume, V, variabil-ity per unit of the defined index, I, can be defined,as follows:

where, a(I) is some function of I that relates theindex to the yield Y, and H is the planting area of thecrop. In order to calibrate an appropriate weathercontract, the variation in crop yield must now beconverted into a financial equivalent that mirrorsthe producer’s exposure. This can be done, for ex-ample, by considering a producer’s production andinput costs per hectare planted or by considering hisexpected revenue from the sale of the crop at har-

∆ ∆ ∆Y V H a I I= = ( ) ( )2

vest. Producers with fixed-price delivery contractsor those using price risk management instrumentsto protect themselves from market fluctuations in theprice of their crop at harvest time know the financialvalue of each kilogram or metric ton they produceand hence can quantify the financial cost of a short-fall in production. If a grain producer, for example,knows he will receive $X per metric ton of crop, thefollowing relationship must hold for his change inrevenue:

A good weather hedge must offset the negative ∆ Revenue fluctuation in the event of a drop in yieldfrom budgeted levels if a producer is to protect hisearnings. In order to perfectly replicate his position,the farmer could enter into a weather contract withthe following incremental payout P per unit index:

Therefore, his overall position would be:

Producers may have contractual obligations to de-liver a predefined amount of their farmed productto a buyer at harvest time, with associated penaltiesif these obligations are not met. In such a situation,it would be straightforward to quantify and struc-ture a hedging product to protect producers fromthese contractual costs in the event of weather-related shortfalls in production.

The Limit

Most weather contracts have a limit, which corre-sponds to the maximum financial payout or recov-ery from the contract in a worst-case scenario, suchas a complete crop failure. The maximum payoutcan be set by either considering the value-at-risk forthe producer in the event of a total crop failure orby looking at historical index, production, andsales data to find the worse-case scenario histori-cally in order to establish a limit. Alternatively, aproducer may simply want to insure his produc-tion and input costs in order to recover these out-lays if the crop fails. If a producer’s productioncosts are $Z per hectare farmed, $Z will therefore

∆ ∆

∆

Revenue + AP = − × ×

+ × × ( ) × =

X Y H

X H a I I 0 5( )

∆ ∆P X H a I I= × × ( ) × ( )4

∆Revenue = × −( ) ×

=

X Actual Yield Expected Yield H�

XX V X H a I I× = ± × × ( ) ×∆ ∆ ( )3



correspond to the maximum payout, the limit ofthe weather contract, for each hectare the producerwishes to insure. The unit exposure P will thereforebe as follows:

Structuring the Product

Structure Type

Once the index has been identified and calibrated,the next step is to structure a contract that payswhen the specified adverse weather occurs, thusperforming a hedging or risk-smoothing functionfor the agricultural grower or producer. Derivativeand insurance products form the mainstay of theweather risk management market. While the twoinstruments feature different regulatory, account-ing, tax, and legal issues, the risk transfer charac-teristics and benefits are often the same. One of thedrivers of market growth has been the flexibilitybetween both instruments and the possibility of tai-loring risk management solutions to a client’s needs(Corbally and Dang 2002). A risk management prod-uct can be either of the following:

• A traditional insurance-style product, that is,risk transfer that results in downside protec-tion in exchange for a premium; for example,a call or put option structure. Or,

• A risk-exchange derivative-based product, thatis, a product based on giving away upside ingood years or seasons to finance downside pro-tection; for example, a collar or swap structure.

Call and Put Options

A call option gives the buyer of the option the right,but not the obligation, to buy the underlying indexat a predefined level at the maturity, or end date, ofthe contract.54 In exchange for this right, the buyerpays a premium to the seller. Similarly, a put op-tion gives the buyer the right, but not the obliga-tion, to sell the underlying index at a predefinedlevel at contract maturity; in exchange for this right,the buyer of the option pays a premium to the seller.Every option contract and, in general, most weathercontracts are defined by a set of standard specifica-tions including:

• The reference index, I, and weather station(s):complete specification of the index and dataused to construct it;

∆ ∆

∆

P = −( ) ×

= ( )Y Expected Yield Z

a I I Expected Yield(( ) × Z for, ( )∆Y < 0 6

• The term, T: the risk protection period of thecontract, including the start and end date ofthe contract;

• A strike, K: also known as an attachmentlevel, the level at which the weather protec-tion begins;

• The payout rate, X: the financial compensationper unit index deviation above (call) or below(put) the strike at maturity, defined as the unitexposure in the previous section; and

• The limit, M: the maximum payout per riskprotection period.

The payout, Pcall, of a call option can be definedusing the following equation:

The payout, Pput, of a put option can be defined asfollows:

The type of option purchased depends on the riskprofile of the buyer. Assume, for example, a winterwheat grower loses 4 percent of his expected yieldevery day that the maximum daily temperaturerises above 30°C in the months of May and June,incurring a cost per day per hectare of 16. Thegrower has 10,000 hectares of wheat under cultiva-tion and is prepared to accept yield losses due toheat stress of up to 480,000, but he wants protec-tion for any losses in excess of that amount. In thiscase, the grower may consider purchasing a call op-tion, either in derivative or insurance form, withthe following specifications:

Reference Weather Station (RWS): Growerstown, ID No. 12345Index: Daily Tmax > 30 C, measured

at RWSCalculation Period: 1 May 2005 to 30 June 2005

(inclusive)Call Strike: 3 eventsPayout Rate: €160,000 per event above the

strikeLimit: €1,600,000

To secure such protection the grower must pay apremium, but he is allowed to recover 160,000 foreach day in May and June that the daily maximumtemperature exceeds 30°C in excess of the strikelevel. Figure A1.4 illustrates the impact of such ahedging strategy on the revenues of the grower: by

P K I X Mput = −( ) ×[ ]min max 0 8, , ( )

P I K X Mcall = −( ) ×( )min max 0 7, , ( )

purchasing the call option, his downside exposureis now limited to 480,000, unless the number ofheat events exceeds an unprecedented 13 duringthe calculation period. Modifications can obviouslybe made to this simplified example to better repli-cate the exposure of the grower; a more sophisti-cated product may be based, for instance, on anindex that considers only consecutive days of ex-cessive temperature, includes relative humidity, orestablishes a nonlinear payout rate that increasescompensation as the number of heat events duringthe calculation period increases. Alternatively, thegrower many want to purchase a digital call option,an all-or-nothing structure that will pay the growera lump sum, rather than incremental payouts, if theheat stress reaches a critical level at which most ofthe crop will be lost. Similarly, an end user buyinga put option would protect himself from eventswhen the index drops below the strike level.

Collars and Swaps

A business may be averse to paying an upfront pre-mium for risk protection. An alternative is a con-tract in which the business receives downsideprotection in return for sacrificing upside revenue

if the weather is beneficial for the business. Inessence, the business can forego a portion of profitto offset the cost of reduced revenues by selling aput option and then buying a call option from theprovider, or vice versa. A collar, therefore, com-bines both a call and put option, but it does not in-volve an exchange of premium from the end userto the provider. A collar is a means by which twoparties can exchange risk; hence, collars may oftenbe structured with asymmetric call and put optionsto make the risk exchange of equal value to bothparties. This approach may not be applicable to allweather risk management problems in agriculture.Furthermore, businesses may be averse to givingup profits in a good year. A very simple example ofa possible application can be found by consideringa local agrochemical company whose sales of a par-ticular pesticide vary depending on the number ofpest growing degree days (PGDDs) recorded intheir sales region during the winter. When therecorded PGDDs are high, pest attack incidents in-crease, and pesticide sales increase accordingly.When PGDDs are low, demand for pesticides dropsand sales are low. The company has quantified thisrisk and finds that, on average, it loses or gains$12,000 per PGDD from budgeted revenues if theaccumulated PGDDs are below or above the 1700PGDDs expected in the region’s normal winter. Thecompany may be interested in a collar agreementbecause, not only is it costless to enter into, it alsoreduces the company’s weather related revenuevolatility. In this case, the company may considerpurchasing a collar with the following specifications:

Reference Weather Station (RWS): Growerstown, ID No. 12345Index: Cumulative PGDDs measured

at RWSCalculation Period: 1 November 2005 to 31 March

2006 (inclusive)Call Strike: 1800 PGDDsPut Strike: 1600 PGDDsPayout Rate: $12,000 per PGDD above/

below strikesLimit: $2,400,000

The historical distribution of November to MarchPGDDs in Growerstown is found to be symmetricaround the 1700 PGDD average with a standarddeviation of 100 PGDDs; hence the call and put op-tions have strikes equidistant of the average to cre-ate a zero-cost collar. Figure A1.5 illustrates theimpact of such a hedging strategy on the revenuesof the company: the collar reduces a potential two


Figure A1.4 Call Option Payout Structure and WheatGrower’s Losses

–3,000,000

–2,500,000

–2,000,000

–1,500,000

–1,000,000

–500,000

0

500,000

1,000,000

1,500,000

2,000,000

0 2

Source: Authors.

4 6 8 10 12 14

Number of heat events(daily Tmax > 30° C) in May-June

Gro

wer

loss

esan

dw

eath

erco

ntra

ctpa

yout

(Exc

ludi

ngpr

emiu

mco

st)(

Euro

s)

Grower losses due to heat events

Net losses of grower with weather hedge

Call optionpayout structure


standard deviation fluctuation in revenues for thecompany from +/− $2,400,000 to +/− $1,200,000.

A swap is a contract in which a buyer makes apayment to the seller when a weather index risesabove a predefined strike level and entitles the buyerto receive a payment from the seller when the indexfalls below the same level. Essentially, a swap is aput and a call option with the same strike, paymentrate, and limit, which, like a collar, is costless toenter. In the example above, rather than using a col-lar contract, the local agrochemical company could“sell” a swap contract to a provider with a strike of1700 PGDDs and a payout rate of $12,000 per PGDD.This would ensure that the business achieves nomore or less than its budgeted revenue. Swaps arederivative OTC contracts that are commonly tradedin the secondary derivative weather risk market;they are rarely used outside the energy industry,however, as they do not always offer the best corre-lation to the underlying risk. Swaps are only avail-able in derivative form (Raspe 2002).

Exotic Structures

In theory, a weather risk management solution cantake any form or combination of options, swaps,triggers, and indexes. Possible exotic combinationsinclude knock-in or knock-out options, which grantthe buyer a standard call or put option if a partic-ular knock-in or knock-out threshold is breached,either on the same or a different index (for example,a heat stress call option for wheat that is only trig-gered if precipitation during the same calculationperiod drops below a critical level); compound op-tions, known as “an option on an option,” that grantthe buyer the right to purchase an underlying op-tion at some future date (for example, a multiyearstructure that gives the buyer an option to buy anoption on the weather conditions for the next grow-ing season at the end of the current season); andstructures with a variable start date depending onthe timing of a pre-specified event (such a structuremay be appropriate for crops with variable plant-ing dates that can be associated with cumulativerainfall or growing degree day totals).

Reference indexes may also include nonweathervariables. Temperature contingent commodity calloptions, for example, may give a purchaser the rightbut not the obligation to buy an underlying com-modity at a prespecified price and volume only ifcertain temperature, that is, growing conditions,have been met. Such exotic structures could poten-

tially provide total revenue insurance for agricul-tural producers whose revenues depend on boththe price at which they sell their produce and thevolume they produce. Such contracts exist and aretraded in the OTC energy derivatives markets.

Risk Retention and Premium

It is clear that an important aspect to consider whenstructuring an index-based solution is the retentionof risk by the party seeking protection. This meansdefining the index trigger level at which the weatherprotection begins. The strike determines the insuredparty’s level of risk retention and is the key to pric-ing and success in transferring the risk. A strikevery close to the mean of the index indicates a lowlevel of risk retention by the end user and a highprobability that the contract will pay out. This im-plicitly means a large premium, as well as the pos-sibility of inspiring little interest in the weathermarket if the location or nature of the risk is outsidethe main liquid trading hubs or variables. A strikefarther away from the mean reduces the probabil-ity of a payout and hence the premium of the con-tract, as the entity is retaining the more frequent,

Figure A1.5 Collar Payout Structure and AgrochemicalCompany’s Deviation from Budgeted Revenue

–5,000,000

–4,000,000

–3,000,000

–2,000,000

–1,000,000

0

1,000,000

2,000,000

3,000,000

4,000,000

5,000,000

1350 1450 1550 1650 1750 1850 1950 2050Cumulative PGDDs from November-March

Dev

iatio

nfr

omco

mpa

ny’s

budg

eted

reve

nue

and

wea

ther

cont

ract

payo

ut(E

xclu

ding

cont

ract

prem

ium

)($)

Deviation from budgeted revenue

Collar payoutstructure

Deviation from budgetedrevenue with weather hedge

Source: Authors.

near-the-mean risk internally and transferring lessto the market. The level of risk retention will dependon the risk appetite and business imperatives of theend user and the sensitivity to the premium associ-ated with entering into a contract. To reduce thepremium payment, for instance, the wheat growerin the call option example above could increase thestrike for heat stress events. By retaining more risk,all things being equal, the producer would reducethe premium of the contract. Alternatively, thegrower could reduce the payment rate to partially,instead of fully, hedge his exposure. Premium pay-ment terms must be defined before entering aweather contract, and an overview of how suchcontracts are priced by weather market providersappears in the following section.

Execution

The Market Providers

The main providers of risk capacity, product struc-turing, and/or pricing for end-user customers in thecurrent weather risk market can be categorized intothree main groups:

• Insurance and reinsurance companies thatview noncatastrophic weather insurance as anatural extension of their traditional businessand given analysis capabilities. Examples include ACE, AXA, Munich Re, Partner Re,Swiss Re, Tokio Marine and Fire Insurance,and XL Capital. Most of these entities also offerderivative products and, although some maychoose to retain the risk by dealing in a largeamount of diversified end-user business, sev-eral are among the most active portfolio man-agers in the secondary market, using financialderivatives contracts to manage their weatherrisk portfolios, including both high- and low-frequency risk.

• Banks that structure weather risk solutionsto fit the needs of their clients. Examples in-clude ABN AMRO, Calyon, Deutsche Bank,Goldman Sachs, Merrill Lynch, and Rabobank.Banks have a large potential client base forweather derivative products and may findmany marketing and cross-selling opportuni-ties in many different sectors of business. Banksgenerally do not have as much risk capacity asdo the (re)insurers; they often pass the posi-tions of their end-user customers to other mar-ket providers or actively hedge positions in the

secondary OTC and exchange-traded deriva-tives market.

• Specialized hybrid companies or funds. Theseinclude organizations such as Coriolis Capital(formerly Société Générale) and GuaranteedWeather Trading Ltd., which were establishedspecifically to trade and invest in weather risk.Such hybrid entities deal in weather deriva-tives and reinsurance and offer weather risksolution products to customers.

The energy companies responsible for the birth of themarketplace—Enron, Aquila, Southern Company,and Entergy Koch (now Merrill Lynch)—are nolonger active in the weather market. Although themarket is still predominantly driven by energy re-lated weather risk, with energy companies andseveral banks hedging their energy portfolios withweather derivatives, the major source of secondarymarket liquidity is now driven by the three pre-dominant types of counter-party outlined above,through the hedging of end-user deals or the takingof speculative positions.

Regulatory Issues

Depending on the jurisdiction, weather risk man-agement products can be classified as financial (de-rivative), insurance, or gaming contracts. Dependingon their classification, these contracts are subject tospecific tax and accounting treatments, which canrender one form more optimal than another for anend user’s purposes and business. Interested partiesare strongly advised to contact their local financialservices authority, insurance regulator, or a profes-sional specializing in insurance law to find out howweather contracts are treated in their jurisdictionand the legal and financial implications associatedwith each (Raspe 2002).

VALUING WEATHER RISKPricing Overview

The premium of an index-based weather contract isdetermined actuarially by conducting a rigorousanalysis of the historical weather to reveal the sta-tistical properties and distribution of the definedweather index and, therefore, the payouts of the in-surance or derivative contract. Such an analysis in-cludes (1) cleaning and quality control of the data,that is, using statistical methods to in-fill missingdata and/or to account for significant changes, ifany, as a result of instrumentation or station loca-tion changes; (2) checking the cleaned data for sig-



nificant trends and detrending to current levels ifappropriate (this is particularly pertinent for tem-perature data, which, in general, exhibit a strongwarming trend in the Northern Hemisphere); and(3) performing a statistical analysis on the cleanedand detrended data and/or a Monte Carlo simula-tion, using a model calibrated by the data, to deter-mine the distribution of the defined weather indexand the subsequent payouts of the contract. By de-termining the frequency and severity of weatherevents specified by the index, an appropriate pre-mium can be calculated.

It should be noted that the premium charged byproviders in the weather market may depend onseveral factors, not all as objective as the underlyingstatistical analysis of the weather data. Institutionscharge different risk margins, or discounts, overthe expected value or fair price to potential buyers;these choices are driven by the risk appetite, businessimperatives, and operational costs of the provider(Henderson et al. 2002). An overview of pricing isgiven in this section, and the implications of thepremium charged for the end user will also be dis-cussed. The data issues associated with points 1 and2 above will be covered later in this Appendix.

Expected Loss and Risk Margin

To illustrate the pricing process, an index-basedweather contract is structured as a call option (seeabove). The payout, P, of the contract is determinedby the following equation:

where K is the strike, I is the index measured duringthe calculation period, X is the payout rate per unitindex, and M is the limit of the contract. To calculatethe premium for the contract, one must determinethe following parameters:

• The expected loss of the contract, E(P), that is,the average or expected payout of the structureeach year;

• The standard deviation of the payouts of thecontract, σ(P), that is, a measure of the vari-ability of the contract payouts; and

• The xth-percentile of the payouts, that is, ameasure of the value-at-risk (VaR) of the con-tract for the seller, VaRX(P). The 99 percentVaR, for example, represents the economicloss for the provider that is expected to beexceeded, with 1 percent probability, at theend of the calculation period of the contract.

P I K X M= −( ) ×[ ]min max 0 9, , ( )

These three parameters quantify the expected (a) and variable or risky (b, c) payouts of the con-tract and must be determined from the historicalweather data, either by using the historical indexvalues from the available cleaned and detrendeddataset or by using the data to calibrate a MonteCarlo simulation model to generate thousands ofpossible realizations of I in order to fill out thedistribution of payouts and to determine betterestimates of E(P), σ(P), and VaRX(P). A completedescription of the various methods for determiningthese payout statistics are beyond the scope of thisappendix, but an overview of possible approachesappears in the following subsection. It is clear,however, that E(P), σ(P), and VaR99(P) will varywith the strike, payout rate, and limit.

Having established values for the expected andvariable payout parameters, the price of a contractis then determined by the risk preferences of the(re)insurance company or financial institution pro-viding the risk protection: that is, by how they mea-sure the cost of risk with respect to return for thepurposes of pricing, risk management, and capitalallocation (Henderson et al. 2002). As a result, thisaspect of the risk pricing process is the most sub-jective, as it is largely driven by the institutionalconstraints and risk appetite of the provider. It is clear, however, that the provider will chargeE(P) plus an additional risk margin for taking theweather risk from the end user, that is,

There are many methods for measuring risk andhence for determining a risk taker’s risk margin.Two examples of simple methods that have beensuggested (Henderson et al. 2002) for the weathermarket are the Sharpe Ratio and the Return on VaR;both measure expected excess return in terms ofsome measure of risk and hence determine the “costof risk” for the contract seller.

Return on Var 99% Premium� ( ) = − ( )[ ]( )

,β E P

VaR P99 −− ( )[ ]= ( ) +

( ) − ( )[ ]

E P

E P

VaR P E P

Premium

β 99 12( )

Sharp Premium

Premium

� Ratio E P P

E P

, α σ= − ( )[ ] ( )= ( )) + ( )ασ P ( )11

Premium Margin= ( ) +E P Risk � ( )10

The Sharpe Ratio uses standard deviation as theunderlying measure of risk; therefore α representsthe “cost of standard deviation” as determined bythe seller’s risk preferences. One of the benefits ofrelating risk to the standard deviation of payouts isthat it constitutes an easy parameter for estimating;however, it is a symmetric measure of risk captur-ing the mean width of the payout distribution, and,for traditional risk exchange products, the payoutdistribution is often not symmetric but has a longtail. The Return on VaR method uses VaR(99%) asthe underlying measure of risk and therefore β rep-resents the “cost of VaR.” Value-at-Risk (VaR) is aterm that has become widely used by insurers, cor-porate treasurers, and financial institutions to sum-marize the total risk of portfolios. The advantage ofVaR99 is that it is computed from the loss side of thepayout distribution, where loss is defined with re-spect to the expected payout E(P), and thereforecaptures the potential financial loss to the seller.Using the Return on VaR method is more appropri-ate for pricing structures that protect against low-frequency/high-severity risk, which have highlyasymmetric payout distributions. VaR99 is a harderparameter to estimate, however, particularly forstrike levels set far away from the mean, and it isusually established through Monte Carlo simula-tion. The worst-case recorded historically can oftenbe used as a crosscheck for VaR. In both methodsoutlined above, α and β quantify the risk loadingappropriate for the risk preferences of the provider.

It is also worth noting that weather market par-ticipants can often enter into financial derivativescontracts to manage their weather risk portfoliosand actively hedge positions in the secondary OTCand exchange-traded derivatives market. This isparticularly true if the end-user risk is in a locationincluded in or positively correlated to locationscommonly traded in the market. Moreover, even ifa market provider chooses to retain the risk inter-nally, a new potential contract may look attractivein comparison to the overall portfolio of the risktaker; that is, it may be a contract that, like hedg-ing, will reduce the relative σ and VaR99 parame-ters and the overall risk position of the portfolioand hence reduce or increase the premium whilemaintaining the same cost of risk α. A reasonableestimate for α and β, given prices in the weathermarket, are α = 15–30% and β = 5–10%.

Approaches to Pricing Weather Risk

In order to price a weather contract, given the over-view above, the parameters that quantify the ex-

pected (E(P)) and variable (VaR99(P), σ(P)) payoutsof the contract must be determined. This sectionbriefly outlines three possible approaches, repre-senting varying degrees of difficulty and effort,commonly used by weather market participants. Ingeneral, providers may use several or all of thesemethods to crosscheck results and compute a con-tract price.

Historical Burn Analysis

Historical Burn Analysis (HBA) is the simplestmethod of weather contract pricing. It involvestaking historical values of the index, which maybe based on raw, cleaned, and possibly detrendedweather data, and applying the contract in questionto them. Assuming the data used to calculate thehistorical indexes are of good quality for the riskanalysis, HBA can give a useful and intuitive firstindication of the mean and range of possible pay-outs of a weather contract from which parameterssuch as E(P) and σ(P) can be calculated. The methodis simple and can easily be done in a spreadsheet.The disadvantage of HBA is that it gives a limitedview of possible index outcomes: it may not capturethe possible extremes, and it may be overly influ-enced by individual years in the historical dataset.Estimates of parameters such as VaR99(P) thereforebecome very difficult, although the largest historicalvalue is always a good reality check when consider-ing the possible variability of payouts. Additionally,the confidence level that can be attached to averagesand standard deviation calculated from historicaldata is limited by the number of years of data avail-able. The standard error in the average decreasesas the number of years included in the average in-creases, however; although weather stations oftenhave thirty to forty years of historical data, the rep-resentative nature of older data for today’s weatherand climate should also be questioned (see below).

Historical Distribution Analysis

Much can be gained from understanding the statis-tical properties of the underlying index. If indexvalues are calculated using historical meteorologi-cal data, then looking at the distribution of theseindex values and ascertaining the probability dis-tribution function of the index will provide a betterestimate of the parameters necessary to specifythat function and, therefore, the expected and vari-able payouts of the contract. Historical DistributionAnalysis (HDA) involves determining the probabil-ity distribution that best fits the historical (possibly



detrended) index data. The process is very much oneof trial and error, and various standard tests andgoodness-of-fit statistics, each with strengths andweaknesses, can be used to pick the best distributionfrom a potential selection; these include Quantile-Quantile plots, calculation of moments, and statisti-cal tests such as chi-squared, Kolmgorov-Smirnov,Anderson-Darling, root-mean squared error, andmaximum likelihood methods. By determining thedistribution and therefore the parameters necessaryto define it, such as the mean and standard devia-tion, the E(P) and σ(P) VaR99(P) can be calculatedeither by simulation from the distribution (seebelow) or analytically, depending on the type ofdistribution chosen and the underlying complexityof the contract to be priced. Closed form solutionscan be derived for call and put options using dif-ferent underlying distributions (Jewson et al. 2005),such as the Normal distribution, kernel density, andGamma distribution. Although the HDA method ismore accurate than HBA for computing expectedand variable payouts (Jewson 2004a), and often sim-pler due to the availability of analytical formulas, itassumes the underlying distribution is a correctrepresentation of the data. Fitting and putting toomuch emphasis on a distribution that does not cap-ture the higher moments of variability, for exam-ple, can lead to an underestimate of variability and,therefore, the premium.

Monte Carlo Simulation

Once a distribution is identified to represent anindex, constraints associated with the length of thehistorical data record are no longer valid, and thou-sands of realizations of the index can be simulated,to estimate the contract statistics to any arbitrarydegree of statistical accuracy, using the distribu-tion to make Monte-Carlo simulations. The IndexSimulation (IS) method is commonly used for pric-ing weather contracts. Index values can be simulatedstatistically by drawing samples from the chosendistribution to generate large numbers (years) ofartificial index values. The weather contract struc-ture is applied to each of these values to create arange of payout outcomes that can be used to cal-culate the price of the contract. The IS method isparticularly good for cumulative contracts, suchas GDDs, or for contracts that depend on severalweather variables where the correlation betweenthese variables can be included in the simulationprocess. An additional advantage of the IS and HDAmethods is that weather forecasts can be incorpo-

rated in the pricing process though the E(P) andpossibly σ(P) terms by their dependence on E(I)and σ(I). The weather market actively follows fore-cast information and will modify its estimates ofE(I) and σ(I) based on historical information if nec-essary (Jewson 2004b). Complex daily simulationmethods can also be used. Building models that cor-rectly capture the physical relationships betweenmany meteorological variables at many sites at adaily resolution poses significant scientific, mathe-matical, and programming challenges (Brody et al.2002), however, and should be required only forpath-dependent contracts or nonlinear structuresthat depend on several variables or critical dailyvalues.

End-User Perspective

On receiving a price quotation for a weather riskmanagement solution from a market provider, anagricultural grower or producer must decide if,given the price, such a solution is the best strategyfor the business to manage its weather risk. Someof the advantages and disadvantages for end usersof using a market-based risk management tool arehighlighted below. A grower can take many tech-nical and practical measures to make crops moreresilient to the vagaries of the weather; examplesinclude better irrigation systems, new strains ofseed, or new farming technologies. Likewise, anagricultural product sales company, for example,may choose to diversify into other products to re-duce their overall exposure to a particular weatherevent. Although such strategies will not be coveredin this appendix, end users should consider therelative cost and efficiency of choosing such ap-proaches over an insurance or derivative weatherbased-solution. Ideally, the end user should focuson the most cost-efficient and effective means fordealing with weather risk by determining the opti-mal interaction of risk retention, risk transfer, andpotential operational strategies to create a compre-hensive risk management solution.

Revenue Volatility and Value-at-risk

From an agricultural end user’s perspective, thecost of E(P) is essentially already embedded in thebusiness: it is the average annual cost (loss) ofweather inherent in running the business in ques-tion, be it farming a crop in a particular region orselling a specific agrochemical product. In otherwords, without protection, the grower or producer

can expect to lose this amount on average eachyear. Therefore, the premium the grower or pro-ducer ultimately pays for a weather risk manage-ment product is only the risk margin charged by theprovider over the expected loss. This is illustratedby the schematic below (Figure A1.6). By purchas-ing a tailored weather hedge, an end user receivesa reduction of revenue volatility due to weather,but at a cost: the risk margin. Reducing the volatil-ity at an appropriate cost, however, increases thereturn per unit risk, or the quality of earnings ofthe end user.

Obviously, the end user must also consider theefficacy of the weather hedge and decide whetherthe risk management contract offers adequate pro-

tection, particularly in a worst-case scenario, forhis business. This can, to a certain extent, be quanti-fied with historical information. The relevant ques-tion the end user should consider is whether thepayout from a risk management contract based ona weather index effectively reduces the end user’svalue-at-risk (VaR); in other words, the end usershould determine whether the contract reducesthe potential for economic loss with a given prob-ability within a given time horizon (Hull 2000). Agrower or producer’s VaR is an effective measureof the overall vulnerability of the business to exter-nal shocks, be it price movements or fluctuations insupply and demand for his product. Weather pro-tection that limits a business’s potential downside


Figure A1.6 Schematic of Historical Revenues of a Business and the Impact of Weather Hedging

Hedged expected revenuewith weather protection

Unhedged expected revenuewithout weather protection

Reve

nue

Time

Unhedged Value-at-Riskwithout Weather Protection

Hedged Value-at-Riskwith Weather Protection

The risk margin Assuming a contract is priced by actuarialmethods, if the annual premiumwas equalto the expected loss of the contract, then,on average, the payout of the contractwould equal the premium over time andthe unhedged and hedge expectedrevenue would be the same.

Source: Authors.


revenue exposure reduces the end user’s overallVaR. Minimizing VaR also has the associated cost—the risk margin—but it raises the question as towhether a business could withstand extreme sys-tematic shocks and their ramifications without pro-tection, limiting losses in catastrophic years.

The birth of the weather market has created anopportunity for businesses to attain protection ontheir income statements from the impact of noncat-astrophic weather variations. Previously, traditionalinsurance products dealt primarily with losses affecting the balance sheet by protecting physicalassets from damage due to catastrophic weather.A business that protects its revenues and, as a re-sult, has a less volatile revenue stream may benefitby receiving, for example, a lower cost of debt oran increased access to credit and, for public com-panies, potentially improved stock valuations orstronger credit ratings (Malinow 2002). Eliminatingthe uncertainty associated with noncatastrophicweather-related risk allows an operation to con-centrate on its core business and focus on control-lable targets and growth. These benefits associatedwith reducing revenue volatility and VaR, in rela-tion to the effective cost of hedging, are considera-tions for the end user. Just like the weather marketproviders, end users must also decide how theyvalue risk in relation to return in the context of theirbusiness. It must define how much risk it is willingto hold and the budgeted cost at which it is willingto do so.

Basis Risk

A major concern with index-based weather riskmanagement products is basis risk: the potential mis-match between contract payouts and the actual lossexperienced. On considering weather-index insur-ance as a product for growers, Skees (2003) writes,“[t]he effectiveness of index insurance as a risk man-agement tool depends on how positively correlatedfarm yield losses are with the underlying area yieldor weather index.” As with the regulatory concernsregarding the definition of insurance (describedabove), this statement relates to the question ofwhether insurance based on a weather index cansubstitute for a traditional crop insurance policy andindemnify the grower for his losses.

Basis risk is a concern with all weather variables,but it is particularly important for rainfall, whichexhibits a high degree of spatial and temporal vari-ability. The weather station on which a weathercontract is based, for example, may not experience

the same rainfall patterns or totals during the calcu-lation period as do the locations an end user wishesto protect. For this reason, weather market providersdo not offer contracts based on hail; hail is a highlylocalized meteorological phenomenon, and althoughit can be indexed to an observing weather station,such indexing may not be an effective risk man-agement strategy for an end user. Although histor-ically an index and losses may correlate strongly—showing that an index could be used as an under-lying trigger to indemnify losses in an insurancecontract (see above)—a good correlation is not aguarantee that the underlying contract payout willmatch the actual loss experienced. Basis risk, there-fore, which can often be minimized by effective orintuitive structuring and by using local stations(Hess and Syroka 2005), is always an issue whendealing with an index-based risk management solu-tion. A potential basis risk outcome can be quanti-fied by using historical data; however, the key pointto consider, as outlined above, is the efficacy of thehedge and the effective reduction in the insuredparty’s overall operational VaR (Hess 2003).

WEATHER DATAData Requirements

In order to implement a successful weather riskmanagement program, the data used to constructthe underlying weather indexes must adhere tostrict quality requirements, including reliable andtrustworthy on-going daily collection and report-ing procedures; daily quality control and cleaning;an independent source of data for verification, forexample, GTS (Global Telecommunication System)weather stations; and a long, clean, and internallyconsistent historical record permitting proper actu-arial analysis of the weather risks involved (at leastthirty years of daily data are ideal).

The premium associated with weather risk man-agement strategies is based on a sound actuarialanalysis of the underlying risk. The commercial risktaker will charge a premium reflecting the givenprobability and severity of specific weather events;hence the quality of historical and on-going weatherdata is paramount. Nearly all weather contractsare written on data collected from official NationalWeather Service (NWS) weather stations; ideally,these will be automated stations reporting daily tothe World Meteorological Organization (WMO)GTS providing data in the internationally recog-nized standard format that then undergo standard

WMO-established quality control procedures. Endusers without access to weather data satisfying theabove criteria, or living in areas in which the spatialcoverage of a NWS weather station network maynot be sufficient to fully represent their weather riskprofiles, may not able to benefit from weather riskmanagement solutions.

All contracts traded in the active secondary OTCderivative market are based on climatic weatherdata collected and published by the NWS of thecountry in question. Historical climate data, anddaily up-dates, can be purchased from each NWS,a list of which can be found on the WMO website(www.meteo.org/wmo). In the United States, forexample, the primary source of weather data is theNational Climatic Data Center. In Great Britain,weather data can be purchased from Weather-xchange (www.weatherxchange.com), a joint ven-ture with the U.K. Met Office set up to support theEuropean weather derivatives market. Weather-xchange provides quality-controlled historical cli-mate and SYNOP datasets across Great Britain andhas distribution rights to data from several NWSorganizations across Europe, including those ofGermany, Italy, France, Netherlands, Austria, andSpain. Data can also be purchased from privatedata vendors, such as Risk Management Solutions/EarthSat (www.rms.com, www.earthsat.com) andApplied Insurance Research (AIR; www.air.com).Private vendors often offer additional value-addedservices such as data cleaning and adjusting (seebelow).

Cleaning and Adjusted Data

Despite the NWS quality control procedures, datafrom some meteorological observing stations maystill have missing and erroneous values. Stationsmay also have undergone instrumentation and/orstation location changes, which can introduce sys-tematic changes to a historical dataset. A stationmoved from a rural to an urban location, for ex-ample, may now be in a location several degreeswarmer than before, creating an artificial jump inthe station’s historical temperature record. Recordsof station or instrumentation changes are usuallykept by the NWS for each weather station. For datato be usable for pricing weather risk managementproducts, the raw data must be cleaned to correctfor errors and missing values and checked and per-haps adjusted for nonclimatic inhomogeneities thatcould make the historical data unrepresentative of

current values. The methods of cleaning and ad-justing data often involve statistical procedures be-yond the scope of this appendix. An awareness ofthe possible need for cleaning and adjusting data isrecommended, however, and the approaches usedare briefly outlined below. Cleaned and adjusteddatasets can also be purchased from private ven-dors with proprietary data estimation models, suchas RMS and AIR.

Detrending Data

Meteorological data often contain trends that arisedue to natural climate variability, urban heating ef-fects, or the impact of global warming. Regardlessof the cause, in some circumstances it may be use-ful to be able to remove such trends from the data.Such a procedure is known as detrending. The aimof detrending data for pricing weather risk is to ob-tain better estimates or forecasts of E(I), σ(I), andVaRX(I) based on historical data. Warming trends,for instance, can significantly impact the definingparameters of the underlying data. By failing to ac-count for such trends, E(I) may be significantly un-derestimated and σ(I) overestimated, which canlead to mispricing of contracts that settle based onfuture data. Many different mathematical methodsexist for detrending data, each based on a differentset of assumptions.

In essence, the aim of detrending is to statisticallymodel the underlying process by decomposing adataset into a deterministic trend and a stochasticnoise term around the trend:

where, D(t) is the process represented by the dataset,Y(t) is the deterministic and therefore predictablecomponent, ε(t) is a normally distributed noise com-ponent with a mean of zero, and standard devia-tion σ and t is unit time. Determining how much ofthe historical data variability is attributed to Y(t)gives an indication of how well a particular modelrepresents the underlying data. The method and ap-proach chosen for detrending data can be highlysubjective, and the decision to detrend or not shouldbe informed by some underlying criteria (Jewsonand Penzer 2004). Choosing a detrending methodthat is better than another at predicting future datavalues—or even not detrending at all—is prefer-able to using a method that increases uncertaintyin predicting future values. The performance of

D t Y t t t N( ) = ( ) + ( ) ( ) ( )ε ε σ, ~ , ( )0 132



different methods can be compared by consider-ing characteristics of the distribution of errors inthe predictions they make. By using the historicaldata to back-test various detrending methods andapproaches, estimates of the uncertainty aroundthe trend can be found and can inform the errorassociated with a particular method for estimatingfuture values.

Identifying trends and their cause is itself a sub-jective process, however, and care should always betaken to check the sensitivity of detrending results tothe underlying method used. Crosschecking severaldetrending methods and approaches and visuallysense-checking the data are always recommended.The weather market often uses the ten-year averageof an index as a quick first-guess estimate for E(I).The simplest and most commonly used methodfor detrending data is polynomial detrending. Theaim of this method is to fit a polynomial functionof time to a meteorological dataset, usually a first-order polynomial trend—a linear trend—that fits astraight line through a set of data points (Weisstein2002). Examples of other detrending techniquesinclude the moving average (Henderson et al. 2002),LOESS (Cleveland 1979), and low-pass filter (VonStorch and Zwiers 1999) methods.

FURTHER READINGInformation for this appendix was taken from thewealth of sources available on the subject of weatherrisk management. These works are strongly recom-mended to interested readers for further informationand discussion. An excellent in-depth introductionto the weather market can be found in Banks (2002),Weather Risk Management: Markets, Products, andApplications. Information on general weather riskand weather risk management issues can be foundin Dischel (2002), Climate Risk and the WeatherMarket, and Geman (1999), Insurance and WeatherDerivatives. The Social Science Research Networkat http://www.ssrn.org offers a large quantity ofpapers and articles on many aspects of weatherderivatives, including analytical pricing methods,simulation models, detrending methods, and theuse of forecasts. The papers written by Dr. StephenJewson are particularly recommended and can befound on http://www.stephenjewson.com. The recent publication, Weather Derivative Valuation: TheMeteorological, Statistical, Financial and MathematicalFoundations (Jewson et al. 2005), contains an excel-lent summary of this work and the theoretical foun-

dations for pricing weather risk. Another goodsource of articles and information on weather deriv-atives and the market is the Artemis website athttp://www.artemis.bm and the industry bodywebsite, the Weather Risk Management Association,http://www.wrma.org. The Guaranteed Weatherwebsite at http://www.guaranteedweather.com/casestudies.php includes a wealth of case studies andweather risk management examples. Information onweather risk management in the developing worldcan be found at http://www.itf-commrisk.org.

REFERENCESAdamenko, T. 2004. “Agroclimatic Conditions and Assessment

of Weather Risks for Growing Winter Wheat in KhersonOblast.” The World Bank Commodity Risk ManagementGroup (CRMG) and International Finance CorporationPartnership Enterprise Projects (IFC-PEP). Unpublished Reportfrom the Ukrainian Hydrometeorological Centre, Kiev, July.

Auffret, P. 2003. “High Consumption Volatility: The Impact ofNatural Disasters.” World Bank Working Paper 2962, TheWorld Bank, Washington, D.C.

Banks, E., ed. 2002. Weather Risk Management: Markets, Productsand Applications. New York: Palgrave Macmillan.

BBC News. 2004. “Farmers Abandon Rotting Harvest.” 21 August.http://news.bbc.co.uk/1/hi/uk/3585638.stm.

Brody, D. C., J. Syroka, and M. Zervos. 2002. “Dynamical Pricingof Weather Derivatives.” Quantitative Finance 2 (June).

Clemmons, L. 2002. “Introduction to Weather Risk Management.”In Weather Risk Management: Markets, Products and Application,ed. E. Banks. New York: Palgrave Macmillan.

Cleveland, W. S. 1979. “Robust Locally Weighted Regressionand Smoothing Scatterplots.” Journal of the American StatisticalAssociation 74 (1979): 829–36.

Coca-Cola Enterprises Incorporated. 2004. “Coca-Cola Enter-prises, Inc. Reports Third-Quarter 2004 Results.” NewsRelease, 28 October. http://ir.cokecce.com/releaseDetail.cfm?ReleaseID=146706.

Corbally, M., and P. Dang. 2002. “Risk Products.” In Weather RiskManagement: Markets, Products and Applications, ed. E. Banks.New York: Palgrave Macmillan.

Dischel, R., ed. 2002. Climate Risk and the Weather Market.London: Risk Books.

Duke Energy. 2003. “Duke Energy Reports Third Quarter 2003 Results.” News Release, 30 October. http://www.duke-energy.com/news/releases/2003/Oct/2003103001f.asp.

The Economist. 2004. “Hedging against the Horsemen.” 11 December.

Efetha, A. 2002. “Irrigation Management of Dry Beans.”Agriculture, Food and Rural Development, Government ofAlberta, June. http://www1.agric.gov.ab.ca/$department/deptdocs.nsf/all/irr4426?opendocument.

Energy East Corporation. 2004. “Energy East CorporationAnnounces 2003 Financial Results.” News Release, 30 January2004. http://www.corporate-ir.net/ireye/ir_site.zhtml?ticker=EAS&script=410&layout=-6&item_id=490266.

Geman, H., ed. 1999. Insurance and Weather Derivatives: FromExotic Options to Exotic Underlyings. London: Risk Books.

Guaranteed Weather. 2005c. “Brewery Barley Risk Management.”Agriculture Weather Risk Resources Case Study. Acquisitiondate: April 2005c. http://www.guaranteedweather.com/page.php?content_id=26.

———. “Freeze Risk to Citrus Crops.” 2005b. AgricultureWeather Risk Resources Case Study. Acquisition date:April 2005b. http://www.guaranteedweather.com/page.php?content_id=27.

———. “The Effects of Temperature Stress on Dairy Production.”2005a. Agriculture Weather Risk Resources Case Study.Acquisition date: April 2005a. http://www.guaranteed-weather.com/page.php?content_id=25.

Henderson, R., Y. Li, and N. Sinha. 2002. “Finance andAccounting Treatment.” In Weather Risk Management: Markets,Products and Applications, ed. E. Banks. New York: PalgraveMacmillan.

Hess, U. 2003. “Innovative Financial Services for Rural India.”Agriculture and Rural Development (ARD) Working Paper9, The World Bank, Washington, D.C.

Hess, U., and J. Syroka. 2005. “Weather-Based Insurance inSouthern Africa: The Case of Malawi.” Agriculture and RuralDevelopment (ARD) Discussion Paper 13, The World Bank,Washington, D.C.

Hull, J. 2000. Options, Futures and Other Derivatives, 4th ed. UpperSaddle River: Prentice-Hall International.

International Monetary Fund (IMF). 2003. “Fund Assistance forCountries Facing Exogenous Shocks.” Prepared by the PolicyDevelopment and Review Department in consultation withArea Finance and Fiscal Affairs, 8 August.

Jewson, S. 2004a. “Comparing the Potential Accuracy of Burn andIndex Modelling for Weather Option Valuation.” WorkingPaper, Social Science Research Network Electronic PaperCollection, 10 January. http://ssrn.com/abstract=486342.

———. 2004b. “Making Use of the Information in EnsembleWeather Forecasts: Comparing the End to End and FullStatistical Modelling Approaches.” Online e-print archives,Cornell University Library, 20 September. http://arxiv.org/PS_cache/physics/pdf/0409/0409097.pdf.

Jewson, S., A. Brix, and C. Ziehmann. 2005. Weather Deriva-tive Valuation: The Meteorological, Statistical, Financial andMathematical Foundations. Cambridge: Cambridge UniversityPress.

Jewson, S., and J. Penzer. 2004. “Weather Derivative Pricingand a Preliminary Investigation into a Decision Rule forDetrending.” Working Paper, Social Science ResearchNetwork Electronic Paper Collection, 11 November. http://ssrn.com/abstract=618590.

Loster, T., Munich Re. 2004. “Risking Cost of Natural Disastersand Their Impacts on Insurance.” Paper presented at theProVention Consortium International Conference, October,Zurich, Switzerland.

Malinow, M. 2002. “Market Participants: End Users.” In WeatherRisk Management: Markets, Products and Applications, ed. E. Banks. New York: Palgrave Macmillan.

Meuwissen, M. P. M., M. A. P. M. van Asseldonk, and R. B. M.Huirne. 2000. “The Feasibility of a Derivative for the PotatoProcessing Industry in the Netherlands.” Report based on apresented paper at the Meeting of the Southern Association

of Economics and Risk Management in Agriculture, March23–25, Gulf Shores, Alabama. http://www.guaranteed-weather.com/page.php?content_id=29.

Neild, R. E., and J. E. Newman. 2005. “Growing SeasonCharacteristics and Requirements of the Corn Belt.” AgronomyExtension, National Corn Handbook, Report NCH-40, PurdueUniversity. Acquisition date: April 2005. http://www.ces.purdue.edu/extmedia/NCH/NCH-40.html.

PricewaterhouseCoopers (PWC). 2004. Annual Weather RiskManagement Association (WRMA) membership survey on weather data, National Economic Consulting Group,PricewaterhouseCoopers, Washington, D.C.http://www.wrma.org/about.html.

———. 2003. Annual Weather Risk Management Association(WRMA) membership survey on weather data, NationalEconomic Consulting Group, PricewaterhouseCoopers,Washington, D.C. http://www.wrma.org/about.html.

Raspe, A. 2002. “Legal and Regulatory Issues.” In Weather RiskManagement: Markets, Products and Applications, ed. E. Banks.New York: Palgrave Macmillan.

Skees, J. R. 2003. “Risk Management Challenges in RuralFinancial Markets: Blending Risk Management Innovationswith Rural Insurance.” Paper presented at Paving the WayForward for Rural Finance: An International Conference onBest Practices, June 2–4, Washington, D.C.

Stoppa, A., and U. Hess. 2003. “Design and Use of WeatherDerivatives in Agricultural Policies: The Case of RainfallIndex Insurance in Morocco.” International Conference onAgricultural Policy Reform and the World Trade Organization:Where Are We Heading, June 23–26, Capri, Italy. www.eco-stat.unical.it/2003agtradeconf/Contributedpapers/StoppaandHess.PDF.

Ulrich, J., 2002. “Managing Director Centrica Energy RiskManagement Group.” Press Release, November. http://www.artemis.bm/html/press_releases/extpress58.htm.

U.S. Congress. 1999. William Daley, Commerce Secretary, re-marks to Congress in 1998, quoted in M. Golden and E. Silliere,“Weather derivatives are becoming a popular hedge.” WallStreet Journal, 2 February 1999.

U.S. Department of Agriculture (USDA). 2003. “Ukraine:Extensive Damage to Winter Wheat.” Online report,Production Estimates and Crop Assessment Division(PECAD), Foreign Agricultural Service, 23 May. http://www.fas.usda.gov/pecad2/highlights/2003/05/Ukraine_Trip_Report/.

Von Storch, H., and F. W. Zwiers. 1999. Statistical Analysis inClimate Research. Cambridge: Cambridge University Press.

Weisstein, E. W. 2002. “Least Squares Fitting.” From MathWorld—A Wolfram Web Resource, 16 September.http://mathworld.wolfram.com/LeastSquaresFitting.html.

XL Trading Partners. 2004. “Corney and Barrow HedgesWeather Exposure with XL Trading Partners Ltd.” Press re-lease, 26 May. http://www.artemis.bm/html/press_releases/extpress124.htm.


83

This appendix presents four case studies—fromCanada, Mexico, India, and Ukraine—showing thesuccessful application for agricultural end users ofweather risk management insurance and derivativeproducts. The first section of this appendix focuseson the Agriculture Financial Services Corporation(AFSC), the Canadian financial crown corporationof Alberta that has been offering Growing DegreeDay products to maize farmers in the province since2000. The second section covers Agroasemex, theMexican agricultural reinsurance company that hasbeen using weather derivatives to manage agricul-tural portfolio risk since 2001. The third section pre-sents two case studies from the recent work of theWorld Bank Commodity Risk Management Groupin developing agricultural weather risk markets inIndia and Ukraine. The Technology ApplicationCase Studies described at the end of this appendixbriefly outlines the principles of the AFSC programto insure grassland for pasture on an index basisusing satellite imagery and the grassland insuranceprogram in Spain.

INDEXED-BASED INSURANCE FORFARMERS IN ALBERTA, CANADAThe AFSC Case Study

Corn Heat Unit Insurance

The Corn Heat Unit insurance program is a weatherindex-based insurance product offered by the AFSCto protect farmers against the financial impact of neg-ative variations in yield for irrigated grain and silagecorn. The contract is designed to insure against lackof Corn Heat Units (CHU) over the growing sea-son. It has been offered on a pilot basis since 2000and was planned to last until 2005. The program isscheduled for a thorough evaluation to assess itsimpact over the next year. The index has been de-signed to indemnify the policyholder against an an-

nual CHU below Threshold Corn Heat Unit (TCHU)level at the specified weather station. The CHU indexfalls into the Growing Degree Day category, dis-cussed briefly in Appendix 1, and represents the en-ergy available for the development of corn. Given thesmall window for agricultural production in Canada,the availability of sufficient solar energy is vital forthe development of this crop. The CHU is estimatedfrom daily maximum and minimum temperature,beginning on May 15 each year. The Celsius-basedformula used to calculate daily CHUs is defined asfollows (Brown and Bootsma, 1993):

where Tmin and Tmax are the daily minimum andmaximum temperatures, respectively.

The daily CHU values are calculated from thesetemperatures. The daytime relationship involvingTmax, uses 10°C as the base temperature (if Tmaxis less than 10, its value is set at 10) and 30°C as theoptimum temperature, as warm-season crops donot develop when daytime temperatures fall below10°C and develop at a maximum rate at around30°C. The nighttime relationship involving Tminuses 4.4°C as the base temperature below whichdaily crop development stops. (If Tmin is less than4.4, its value is set at 4.4.) The CHU value is calcu-lated by taking into account the functional relation-ship between daytime and nighttime temperaturesand the daily rate of crop development, as shown inFigure A2.1. The nighttime relationship is a straightline (Equation 2), while the daytime relationship ap-pears as a curve that records greater CHUs at 30°Cthan at higher or lower temperatures (Equation 3).The accumulation of CHU stops on the first day on

Y T Tmax max max= × −[ ] − × −[ ]3 33 10 0 0 084 10 32. . . ( )

Y Tmin min= × −[ ]9 5 4 4 2. ( )

CHU Y Y= × + ×0 5 0 5 1. . ( )min max

Appendix 2Case Studies of AgriculturalWeather Risk Management

which a minimum temperature of minus two de-grees Celsius or less is recorded, after 700 CHU havebeen accumulated. This means the accumulationcontinues until the first killing frost hits the crop.An early frost setback is also built into the AFSCcalculation.55

The weather data for settlement of the contractsare provided by the federal and provincial weatherstations and compiled by the Irrigation Branch ofthe Alberta Government. Contract end users canselect a weather station for the settlement from thefederal and provincial stations available, choosingthe station that best represents the temperatures ontheir farms. Weather stations used for CHU insur-ance are divided into three groups based on simi-lar historical heat accumulations. Weather stationswithin each group have similar threshold options,premium rates, and loss payment functions.

Coverage is available in $25 Canadian Dollar(CD) increments with a minimum of CD$100 peracre for both grain and silage corn and a limit ofCD$225 and CD$300, respectively. Farmers canbuy the insurance product until April 30 of the yearto be covered for that year’s growing season. When

buying the insurance policy, farmers must elect thedollar coverage per acre, select the weather stationfor settlement purposes, and indicate if they prefera hail endorsement to the contract or the variableprice benefit.

The farmer must insure all the seeded acres ofeligible corn and must insure a minimum of fiveacres for each crop: grain and silage crops are con-sidered separate for the purposes of referring to aspecific insurance contract. Only producers grow-ing grain or silage corn on irrigated land in AFSCdesignated areas are eligible to buy a CHU insur-ance contract. The farmer must complete seedingby May 31 and must declare the final number ofseeded acres and a legal description for the locationof each crop no later than June 1. The insurable cropshall be grown within the risk area boundaries asdetermined solely by AFSC. Furthermore, the AFSCis responsible for controlling the use of these con-tracts to ensure that they are used only for insur-ance purposes. For control and product evaluationpurposes, the farmer is required to present a har-vested production report, stating the production ofall insured crops, no later than fifteen days after com-pletion of the harvest but no later than December 15of each calendar year.

The premium payable under the CHU contractis due upon receipt of the contract by the farmer. Atable of premium rates and payment rates for grainand silage corn is made available to the farmer andindicates the base premium rate and the percentagepayment triggered, depending on the heat unitlevel recorded at the station chosen.56 The formulato calculate the indemnity for each insurable cropis given by the following equation:

If a farmer chose to insure one hundred acres at$225 per acre, for example, and the accumulatedCHU payment rate was 30 percent of the expectedlevel, a claim of $6,750 dollars would result. Themaximum indemnity payable is 100 percent of theDollar Coverage per Acre (including the additionaldollar coverage if the Variable Price Benefit is acti-vated) multiplied by the number of insured acres.

Producers can choose between two CHU in-surance deductibles or threshold options (High

Indemnity Dollar Coverage per Acre

Payment

=

× RRate Number of Insured Acres×


Figure A2.1 Relationship Between the Daily Rate ofDevelopment of Corn Minimum and Maximum Temperatures

0

5

10

15

20

25

30

35

0 5 10 15 20 25 30 35 40 45

Minimum and maximum temperature (degrees Celsius)

Rat

eof

crop

deve

lopm

ent

Nighttime minimum temperature relationship

Daytime maximum temperature relationship

Source: Brown and Bootsma 1993.

Appendix 2. Case Studies of Agricultural Weather Risk Management 85

and Low “Trigger”); see Table A2.1. Paymentsbegin sooner under the high threshold option, sothis choice has a higher cost than the low thresh-old option.

Claims are based on accumulated CHUs calcu-lated using the temperature data recorded at theselected weather station. CHUs accumulated beforethe killing frost are compared to the thresholdchosen by the producer at the weather station. Ifthe annual CHUs are less than the chosen thresh-old, the insurance program starts to make paymentsaccording to a predetermined table. The further theannual CHUs are below the threshold, the greaterthe insurance payment.

The main peril for producers is lack of heat dur-ing the growing season, but this insurance planalso includes a provision for late spring frost. Alate spring frost can set back corn plant growth andaffect production. To trigger this provision, a tem-perature of less than zero degrees Celsius must berecorded on or after June 1 and prior to the record-ing of 700 CHUs at the weather station. If boththese conditions are met, 50 CHUs will be deductedfrom the accumulated total CHUs at the end of theyear for the first day and an additional 15 CHUs willbe deducted for every other day between June 1 andthe day the frost in question occurred.

It is important to point out that the CHU con-tract with the hail endorsement is designed to pro-tect corn against two major perils: lack of heat andhail. The grain and silage corn farmers are also eli-gible for traditional crop insurance contracts basedon individual records; nevertheless, the premiumsare lower for the CHU contract because of AFSC’sreduced transaction costs. It should also be notedthat the premiums paid by the farmers for the CHUcontract are subsidized by approximately 55 per-cent, so the farmer pays only 45 percent of the costof the contract. The subsidy is 40 percent for the hailendorsement. The federal and provincial govern-ments coshare the financial burden of the program,and they subsidize all AFSC’s administration costs.

ALTERNATIVE INSURANCETHROUGH WEATHER INDICES IN MEXICOThe Agroasemex Case Study

Agroasemex is a Mexican government-owned re-insurance company operating exclusively in agri-

cultural insurance. Agroasemex relies heavily onthe traditional reinsurance market to protect itsagricultural portfolio from inordinate losses. As aresult of a 70 percent increase in the retrocessionrates of 2001, Agroasemex’s search for new alter-natives led it to analyze the comparative efficiencyof the weather derivatives market. The purpose ofthis case study is to present the background, design,and guiding principles behind the weather deriva-tive structure ultimately created for use as a hedgefor the Agroasemex agricultural portfolio. It is worthnoting that the institution’s weather derivativetransaction in 2001 was the first of its kind in thedeveloping world. This simplified case study willoutline the approach and thought processes behindthe structuring of the Agroasemex weather risktransfer program.

Designing a Weather Risk Transfer Solutionfor the Agroasemex Agricultural Portfolio

Selection of Risks

There are two agricultural production cycles inMexico: spring-summer and autumn-winter. Theformer is primarily a rain-fed production cycle, whilethe latter is generally irrigated. The Agroasemexweather risk transfer program was specifically de-signed for the autumn-winter cycle of 2001 to 2002.The main weather risks for agriculture during thiscycle were potentially large negative deviations intemperature and excess rainfall. For some areas,where irrigation was not used, lack of rainfall wasalso an important risk. The percentages of crops dis-tributed in five states were included in the weatherrisk transfer program.

Table A2.1 Options for CHU Contracts

Deductible or Trigger (Annual CHU)

Station Long-Term Low HighGrouping Normal Option* Option**

A 2,505 2,260 2,380B 2,387 2,160 2,280C 2,332 2,100 2,220

*Approximately 90 percent of long-term CHU normal. ** Approximately 95 percent of long-term CHU normal.

Source: AFSC.

Table A2.2 Total Liability Factored into the AgroasemexBusiness Plan for Autumn-Winter 2001–2002 (the basis of the design of the weather derivative contract)

Total Liability State Crop (US$ Million)

Nayarit Tobacco 22.4000Sinaloa Beans 0.1917Sinaloa Chickpeas 0.4600Tamaulipas Sorghum 1.8200Sinaloa—Sonora Maize 2.0190

The crops and weather risks were selected giventheir relative importance in the portfolio, the con-sistency of the numerical analysis between negativedeviations in the agricultural portfolio and the pro-tection provided by the proposed weather deriva-tive structure, and the availability of consistent andhigh-quality historical weather data. Based on theoriginal risk profile and business plan report for theautumn-winter cycle of 2001–2002, the total liabilityfor the crops and risks selected for the weather risktransfer program are shown in Table A2.2.

The total expected traditional reinsurance pre-mium for the entire Agroasemex portfolio was esti-mated to be US$1,917,422. The subset in Table A2.1represents approximately 10 percent of the risk inthe entire portfolio for 2001–2002.

Transforming Weather Indices into theExpected Indemnities of the AgroasemexAgricultural Portfolio

The following method was used to establish therelationship between weather indices and the ex-pected indemnities of the Agroasemex agriculturalportfolio. First, a severity index was created for eachcrop in the portfolio in order to understand, at theportfolio level, how important this crop risk wouldbe when a given weather phenomenon, as capturedby an index, occurred. A very simple severity index(SI) is defined as follows:57

SIIndemnities

Total Liability

t

ti

=

( )

=

4

19991 92 1992 93 1999 2000, ... ;Autumn-Winter� Cyclees

i = Crop

Once the severity index was calculated for each crop,the next step was to find a mathematical relationshipbetween the SI and the weather index most rele-vant to the crop. Agroasemex performed linearleast square regressions for each crop severity indexto establish the SI–weather-index relationship:

where

and

where FCDD (Factores Climaticos Dañinos Diarios)—damage degree days or periods—that represent theindex that captures the critical weather risk of eachcrop in the portfolio outlined in Table A2.3 (seebelow); εt is a normally distributed noise term; andthe estimators for the linear gradient and intercept,m1 and m0, were calculated using a least squaresregression method.

The gradient estimator for m1, in particular, isvery important, as it establishes the relationshipbetween the individual severity indices and therelevant weather indices. Once all the linear re-gressions for each crop are performed and all the linear estimators are calculated, the expectedindemnities (in monetary terms) for each severityindex, given a certain weather index (FCDD) andtotal liability, can be calculated as follows:

FCDDs: The Weather Indices

The FCDD terms for each crop in the preceding sec-tion represents the weather index or indices thatbest capture the weather risk for that crop. If weare analyzing the exposure of beans to low tem-peratures, for example, the FCDD index could bedefined as the number of days that the daily mini-mum temperature drops below a specified dailythreshold during the growing season. To constructthe appropriate weather indices for the Agroasemexportfolio, the relevant weather historical informa-tion was collected: five Mexican weather stationson the Pacific Ocean coast were chosen to representthe western area of the country (Sonora, Sinaloa, andNayarit), while two U.S. airport stations (McAllen

Indemnities Total Liability FCDD mt t t( ) = ( ) × ×� 1 7(( )

x FCDDt t= ,

yIndemnities

Total Liabilityt

t

=

( )�

; 6

y m m xt t t= + + ( )0 1 5ε


Source: Authors.


and Brownsville) were used to represent the north-eastern area (Tamaulipas).

It is important to note that even though eachseverity index, as defined above, is a seasonal ag-gregate, the types of risks relevant for an agricul-tural portfolio of crops can occur over very shortperiods of time; for example, crop damage due tofrost can occur in just one day. Therefore the selec-tion of the individual weather indices for each cropwas based on two criteria: first, and primarily, onthe agronomical surveys and experience of the tech-nical personnel of Agroasemex, and second, on thestrength of the mathematical relationship obtainedwhen comparing the available data on indemnitiesfor the crop in question, with the weather index(Equation 4)—this was done both on a daily basis(data on indemnities were available in daily reso-lution) and on a seasonal basis.

To understand how each individual FCDD wasestimated, consider the example for the weatherindex chosen for tobacco in Nayarit: DDD-12. Lowtemperature is the greatest risk for tobacco crops inNayarit; when the daily minimum temperature

drops below 12°C, the expected tobacco yields willbe below average. Hence 12°C is the minimumtemperature threshold level for tobacco crop dam-age: DDD-12 represents Damage Degree Days witha 12°C threshold. The DDD-12 index is defined asfollows:

where the DDD-12 summation is over each day in the growing period of tobacco: November 1 toMarch 31 of the following year. Daily minimumtemperature, Tmin, is measured at a single weatherstation, Capomal, in Santiago Ixcuintla, Nayarit.The data are aggregated at a seasonal level. TheDDD-12 estimation is consistent with the El Niño,as the worst year recorded of cold temperaturesaffecting the tobacco-producing area.

In total, eleven independent FCDDs were de-signed to represent the exposure of the crops andrisks selected. The FCDD calculation methodologiesusing daily weather data are presented in Table A2.3for all crops in the portfolio.

DDD T- -12 0 12 8= ( ) ( )∑ max , min

Table A2.3 Summary of the Methodology to Calculate the Eleven FCDD Indices

Weather FCDD Calculation Methodology State Crop FCDD Station (in mm and deg Celsius) Calc. Period

Tobacco

Beans

ChickpeasSorghum

Maize

DDD-12EMNF

EMMA

DDD-5DDD-3EMF

MAX-5

EMGMAXPS

DDD-5

DDD-3

Capomal1 Capomal2 La Concha1 Capomal2 La ConchaSanalonaSanalona1 Sanalona2 El Fuerte3 Jaina1 Sanalona2 El Fuerte3 Jaina

Sanalona1 Brownsville2 McAllen

Sanalona

Sanalona

DDD-12 = Sum Daily [max (0, 12 − Tmin)]EMNF = Sum Daily [Rainfall Station 1] +Sum Daily [Rainfall Station 2]EMNF = Sum Daily [Rainfall Station 1] +Sum Daily [Rainfall Station 2]DDD-5 = Sum Daily [max (0, 5 − Tmin)]DDD-3 = Sum Daily [max (0, 3 − Tmin)]EMF = Sum Daily [Rainfall Station 1] +Sum Daily [Rainfall Station 2] + Sum Daily[Rainfall Station 3]MAX-5 = max (MP − 200, 0);MP = max (Sum 5-day D3) − max rainfallfor a consecutive period of 5 days, whereD3 = Daily Rainfall Station 1 + DailyRainfall Station 2 + Daily Rainfall Station 3EMG = Sum [max (Daily Rainfall − 55, 0)]PS = Sum [max (250 − CMP1, 0)] + 2 ∗Sum [max (250 − CMP2, 0)];CMP1 = Monthly Cum. Rainfall Station 1CMP2 = Monthly Cum. Rainfall Station 2DDD-5 = max [D5 − 22, 0];D5 = Sum Daily [max (0, 5 − Tmin)]DDD-3 = Sum Daily [max(0, 3 − Tmin)]

Dec 1–Mar 31Nov 1–Feb 28

Mar 1–Apr 30

Oct 1–Apr 30Dec 1–Dec 31Nov 1–Mar 31

Nov 1–Mar 31

Nov 1–Apr 15Oct 1–May 31

Oct 1–Apr 30

Dec 1–Dec 31

Source: Authors.

Nayarit

Sinaloa

Tamau-lipas

SinaloaSonora

The mathematical relationship between eachFCDD index and the indemnities for the corre-sponding crop in the Agroasemex portfolio wereestablished using equations 4 through 6, defining ameans of converting FCDD indices into expectedindemnities in monetary terms. By combining thisinformation, the basket of all the expected indemnityindices was used to replicate the overall weatherexposure of the agricultural portfolio. This “com-bined index”—essentially the sum of all the ex-pected crop indemnity indices—was used as anunderlying proxy and therefore hedge for theweather exposure of a portfolio. A derivative struc-ture based on this combined index, such as a calloption, is therefore conceptually the same as a stop-loss reinsurance strategy for the portfolio, as weatheris the greatest risk to Agroasemex.

Historical Back-Testing

The strength of the approach outlined above—toestablish a basket of indices that best captures theweather exposure of the Agroasemex agriculturalportfolio—was back-tested by using annual histor-ical indemnity and total liability information fromthe Agroasemex direct insurance operations from1990 to 2001. The historical portfolio indemnityrecords were compared to the estimated indemni-ties, given the total liability observed for that year

and using the FCDD-indemnity relationships established in Table A2.3.

The values of the severity index for each cropwere calculated using both the historical and themodeled data for comparison. The results showedthat the combined weather index established forthe Agroasemex portfolio had an acceptable pre-dictive power, mainly because it captured the largehistorical deviations in the portfolio (Table A2.4).

The results demonstrate that the combinedweather index model explains about 93 percent ofthe variability demonstrated by the empirical data.

Valuation of the Weather DerivativeStructure and the Agroasemex Transaction

Monte Carlo simulation, as described in Appendix 1,was used to generate an estimate of the distributionof the possible results of the combined weatherindex and therefore the maximum liability of theAgroasemex portfolio (see Figure A2.2).58 The greenline in Figure A2.2 is constructed using only histor-ical information, while the darker, smoother line isestablished from the stochastic Monte Carlo simula-tion analysis of the underlying weather variables. Itis clear that the historical payout of the Agroasemexportfolio has never exceeded US$1.65 million, whilethe simulation analysis generates more extremeresults, exceeding the US$2.5 million level.


Table A2.4 Comparative Analysis Between the Observed Historical Severity Indices (indemnities/total liability) and the Estimated Severity Indices for the Crops and Risks Selected

Tobacco Beans Chickpeas Sorghum Maize Total

Obs. Est. Obs. Est. Obs. Est. Obs. Est. Obs. Est. Obs. Est.

0.052 0.060 0.000 0.086 0 0 0.038 0.0570.017 0.019 0.020 0.018 0.003 0 0.085 0.093 0 0 0.021 0.0230.004 0.009 0.027 0.043 0 0.015 0.133 0.122 0 0 0.016 0.0210.007 0.002 0.109 0.113 0.043 0.042 0.233 0.178 0 0 0.037 0.0330.009 0.006 0.059 0.047 0 0 0.131 0.158 0 0 0.019 0.0180.067 0.068 0.164 0.178 0.117 0.126 0.000 0.004 0.017 0.017 0.067 0.0690.052 0.046 0.403 0.407 0.117 0.104 0.010 0.071 0.142 0.142 0.109 0.1030.008 0.006 0.167 0.140 0 0 0.084 0.061 0.011 0.011 0.033 0.0270.007 0.006 0.099 0.115 0 0 0.064 0.175 0.003 0.003 0.010 0.014

r = 0.985 r = 0.968 R = 0.988 r = 0.702 r = 0.999 r = 0.970r2 = 0.971 r2 = 0.936 R2 = 0.976 R2 = 0.492 r2 = 0.999 R2 = 0.939

Note: Figures are in decimals.

Source: Authors.


Figure A2.2 Comparative Accumulated Distribution Probability Function Based on a “Probability of Exceedence Curve” for the Historical and Modeled Results (payouts in US$)

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

0 500,000 1,000,000 1,500,000 2,000,000 2,500,000 3,000,000Result ($)

Prob

abili

tyof

exce

eden

ce

Stochasticprobability ofattachment

Stochasticprobability ofexhaustion

Structure A: 32.2% 1.4%Structure B: 26.4% 1.4%Structure C: 21.8% 1.4%Structure D: 17.7% 1.4%

Source: Authors.

Table A2.5 Specifications of Call Option Structures Considered by Agroasemex

Structure A B C D

Strike Price (US$) 1,000,000 1,100,000 1,200,000 1,300,000Payout Limit (US$) 1,200,000 1,100,000 1,000,000 900,000

Source: Authors.

The original analysis performed by Agroasemexfocused on four possible call option derivative struc-tures, which varied in the strike price and limit ofpayout that could be used as an alternative to a tra-ditional stop-loss reinsurance contract to managethe portfolio risk (Table A2.5).

The historical results and the stochastic analysisfor the actuarial fair value of risk for each call op-tion structure (average and standard deviation)are summarized in Table A2.5. In addition to theactuarial fair value of risk, the market the pre-mium charged for risk management solutionscombined the expected or fair value of the risk—the pure risk premium—with an additional riskmargin. Considering market standards at the time,59

the following risk loadings above the expectedvalue were considered:

• Loading Based on Standard Deviation:60 Mar-ket standards 20 to 40 percent. An intermedi-

ate loading of 30 percent was considered byAgroasemex.

• Loading Based on the Uncertainty due to Gapsin the Historical Weather Data: When missingdata exceed 1 percent of data points, marketplayers usually design a sensitivity analysisto estimate the impact of using alternativein-filling methods (see Appendix 1) and chargefor the uncertainty that arises as a result ofsuch gaps in the historical record. No estab-lished method exists for calculating this uncer-tainty loading in the market, which generallydepends on the risk appetite of the individualweather risk taker.

• Loading for Administrative Expenses: A mar-gin of 15 percent was added.

The weather stations used for the project in Mexicowere carefully selected. Nevertheless, missingdata ranged from 2.70 percent to 9.20 percent. The

weather data gaps were in-filled by Risk Manage-ment Solutions (RMS) on a monthly basis, based ondata collected from neighboring weather stations.In order to quantify the sensitivity and robustnessof the in-filling method, instead of filling gaps withdata inferred from the most correlated stations, thegaps were also in-filled with the most extreme ob-servations from a sample of stations that had accept-able correlations to the station with the missing datapoints, both for temperature and rainfall. The uncer-tainty loading due to missing data was estimated tobe 50 percent of the resulting change in the averagepayout, as a result of this sensitivity analysis, plus50 percent of the change in the standard deviationobserved. The results were aggregated to completethe analysis; Table A2.6 shows the estimated com-mercial premium (expected value plus risk margin)calculated for the four weather derivative structures.

Despite the risk loading, Agroasemex eventu-ally bought structure D from the market. The mainmotivations for this choice were the following:

• The transaction included the donation of threeautomated weather stations, worth approxi-mately US$36,000, as fallback stations. Takingthis cost into account, the ratio of the com-mercial price of the derivative to the pure riskpremium was the lowest for structure D: 1.57vs. 1.62 for the nearest structure.

• To establish credibility and brand recognitionfor future weather transactions.

• To set a market reference for the risk margin,so that future, larger deals could be negotiatedunder more narrow risk margins.

Developments Since 2001

After devising the initial weather derivative transac-tion presented above, Agroasemex devoted its insti-tutional efforts and experience to developing a localweather risk market. These activities included a thor-ough review of the weather data; further improve-ments to the weather observation infrastructure, inconjunction with the Mexican National WeatherService; and training and education for potential endusers within Mexico. The greatest interest generatedby the 2001 transaction was from the Mexican gov-ernment regarding their catastrophic weather expo-sure: since 2001, Agroasemex has sold weather indexinsurance to three Mexican states to cover the states’catastrophic exposure related to agriculture. In turn,Agroasemex has bought protection for this risk, on aquota share basis, in the international weather deriv-atives market. The three transactions together havean approximate notional value of US$15 million,with several other states in the coverage pipeline.There are unofficial reports that the internationalmarket has also closed several transactions with theprivate industry in Mexico as a result of this firstweather derivative transaction.

WEATHER INSURANCE FORFARMERS IN THE DEVELOPING WORLDCase Studies from India and Ukraine

The Commodity Risk Management Group (CRMG)at the World Bank started working on pilot weather


Table A2.6 Estimated Commercial Premium for Weather Derivative Structures (in US$)

Analysis and Statistics Structure A Structure B Structure C Structure D

Last Ten-Year HBAPure Risk Premium 181,447 151,447 121,447 91,447Standard Deviation Loading 83,372 69,669 55,987 42,34715% Margin 46,733 39,020 31,312 23,611Full Price 311,552 232,229 186,622 141,157

Simulation AnalysisPure Risk Premium 133,460 104,291 80,252 60,528Standard Deviation Loading 80,241 70,226 60,638 51,634Data Uncertainty Loading 31,750 27,584 23,693 20,13615% Margin 43,315 30,797 24,863 19,793Full Price 288,766 232,898 189,447 152,091

Source: Authors.


risk management projects in 2003. The CRMG wasinvolved in its first index-based weather risk man-agement contract in India in June 2003. Since then,the number of projects has grown. CRMG is cur-rently working on pilot projects for smallholders inIndia as well as projects in Peru, Nicaragua, Ethiopia,Thailand, Kenya, Malawi, and Ukraine. Providersin the global weather risk market are extremely in-terested in such new transactions both to diversifytheir weather portfolios through new locations andrisks and to offer opportunities for business growthand expansion.

Two case studies will illustrate some of the CRMGwork in this new area. The first case study examinesthe developing weather market in India, particu-larly the recent work of the Mumbai-based insur-ance company ICICI Lombard General InsuranceCompany Ltd. and the Hyderabad-based micro-finance institution BASIX in making weather in-surance available to smallholder farmers in AndhraPradesh. This case study provides an example of therole of insurance in access to finance for farmersexposed to weather risk. The second case study focuses on the 2005 weather insurance pilot pro-gram for winter wheat farmers in the southernoblast of Kherson in Ukraine.

Weather Insurance for Agriculture in India

In 1991, a household survey in India addressingrural access to finance revealed that barely one-sixthof rural households had loans from formal ruralfinance institutions and that only 35 to 37 percentof the actual credit needs of the rural poor werebeing met through these formal channels (Hess2003). These findings implied that over a half of allrural household debt was to informal sources, suchas moneylenders charging annual interest ratesranging from 40 to 120 percent. A survey based onthe Economic Census of 1998 (Hess 2003) showedthat India’s formal financial intermediaries report-edly met only 2.5 percent of the credit needs of theunorganized sector through commercial lendingprograms.61

In this context, the CRMG, in collaborationwith the Hyderabad-based microfinance institu-tion BASIX and the Indian insurance companyICICI Lombard, a subsidiary of ICICI Bank, initi-ated a project to explore the feasibility of weatherinsurance for Indian farmers and to determine if,by reducing exposure to weather risk, it would bepossible to extend the reach of financial services tothe rural sector.

BASIX: Weather Insurance for Groundnut and Castor Farmers

Established in 1996, BASIX has since emerged asone of India’s leading microfinance institutions. Ithas systematically addressed the issues of risk mit-igation and cost reduction with the twin aims of at-tracting investment from the mainstream capitalmarkets while maintaining and expanding its lend-ing in rural areas, including lending for agriculturein drought-prone regions (Hubka forthcoming).BASIX is the umbrella name used to denote a groupof companies focused on the provision of micro-credit and investment services as well as on im-proving the livelihoods of its clients and borrowers.To date, BASIX has approximately 150,000 borrow-ers and 8,600 savers in 7,800 villages in ten Indianstates, disbursing US$37 million in loans since 1996;currently 49 percent of loans are for nonfarm activi-ties (Hubka forthcoming). Its goal is to affect onemillion livelihoods by 2010: 500,000 directly throughfinancial services and another 500,000 through indirect means. BASIX thinks of itself not as amicrofinance institution but as “a new generationlivelihood promotion institution,”62 implying thatcredit alone is not the solution to the problems ofrural areas.

BASIX manages its risk at two levels: first, itmanages its own, institutional-level risk throughcustomer selection and lending practices and part-nerships with other institutions; and second, it helpsits borrowing customers to reduce their risk (Hubkaforthcoming). By helping customers to mitigate andmanage their own risk, and hence the risk of default-ing on their loans, BASIX in turn protects the qualityof its own portfolio. In 2003, in order to further ex-tend the risk management offerings it provides itsclients, BASIX joined forces with ICICI Lombard,and with technical support from CRMG, they de-signed, developed, and piloted a weather insur-ance product for farmers with small and mediumholdings in Andhra Pradesh.

BASIX recognized that, in many areas, farmers’yields depend critically on rainfall and that its loandefault rates were highly correlated to drought.Furthermore, BASIX found that the losses sustainedby individual farmers from below average rainfallwere on account of several factors, not direct impactson yields alone (KBS LAB 2004). In addition toweather-related yield loss affecting an individualfarmer’s ability to meet credit repayments—withcredit default disrupting the next season’s loan dis-bursal and hence the farmer’s agricultural cycle—

the systematic nature of drought leads to area-wideproduction drops, resulting in local price inflationand harder credit terms for the next growing seasonfor all producers.

The government-sponsored area-yield indexedcrop insurance scheme offered by the NationalAgricultural Insurance Company (NAIC) is com-pulsory for all crop-loan borrowers using Indianbanks and the only crop insurance option availableto BASIX customers. BASIX, as have others (Hessand Skees 2003), found, however, a number of in-efficiencies in the federal program in relation todrought. In particular, they noted that the NAICprogram only led to recovery in extreme situations,that is, following district drought declarations bythe state government, which were often the resultof political maneuvering rather than objective cri-teria. Furthermore, in the NAIC program, recoverywas based on minimum crop prices and in generaloccurred two to three years after the failed harvest.By comparison, index-based weather insurance offered the potential of a transparent, objective, andtimely settlement processes for economic losses as-sociated with noncatastrophic weather risk, withrecovery based on fair market price estimates. Withthe requirements of farmers in rain-sensitive re-gions in mind, BASIX considered these to be com-pelling reasons to launch a pilot weather insuranceprogram.

First Pilot Program: 2003

The initial pilot launched by BASIX and ICICILombard was based in the Mahahbubnagar dis-trict of Andhra Pradesh, with an objective of sell-ing weather insurance policies to two hundredgroundnut and castor farmers through KrishnaBhima Samruddhi Local Area Bank (KBS LAB), aBASIX subsidiary licensed by the Reserve Bank ofIndia providing microcredit and savings servicesin three districts.63 The farmers selected for theinitial pilot were members of a Bore Well Users’Association (BUA)64 in four BUA villages in theMahahbubnagar district: Kodur, Pamireddypally,Utkoor, and Ippalapaddy. In 1999, for example,the BUA in Pamireddypally received an agricul-tural loan from BASIX. With a 100 percent repay-ment rate, and therefore a good BASIX credit historyand standing, they were planning to borrow a furtheramount for the financial year 2003–2004. Based onthis strong customer relationship, BASIX launchedthe weather insurance pilot in Pamireddypally andthe other three villages. In particular, by linking the

new insurance pilot to farmers who had accessedfinance, BASIX would form a base from which theycould begin to understand the interaction betweensuch a product, credit repayment, and, ultimately,their crop-loan portfolio default rates.

The Weather Insurance Contract Design

Groundnut is the primary rain-fed crop grown in the Mahahbubnagar district during the June toSeptember monsoon, or khariff, season, followedby castor. While most of the cultivation of ground-nut and castor is during the khariff, crops are alsocultivated in the winter, or rabi, growing season, inpockets of irrigated land. The economics of culti-vating groundnut and castor per acre during thekhariff and rabi seasons were established throughinteractions with the BUA members in feedbacksessions and workshops organized by KBS LABand ICICI Lombard, with additional informationand crosschecking from the local agricultural uni-versity in Hyderabad. Total input costs for ground-nut were estimated at Rs 6,500 (khariff) and Rs 6,000 (rabi), and for castor at Rs 3,000 (khariff)and Rs 3,100 (rabi).

The aim of the 2003 pilot program was to designweather insurance contracts to insure farmers’ inputand production costs. The initial weather insur-ance contracts designed for the castor and ground-nut farmers were based on a weighted rainfallindex of rainfall collected and recorded at theIndian Meteorological Department (IMD) officialdistrict weather station in the district capital town,Mahahbubnagar. High-yield rainfall correlationswere measured for khariff crops in the area; never-theless agronomic information was used to enhanceand strengthen the yield-rainfall relationship forthe contract structures. In the case of groundnut, forexample, the most critical periods—when ground-nut is most vulnerable to low rainfall and thereforewater stress—are the emergence periods immedi-ately after sowing and the flowering and pod-filling phase two to three months after emergence(Narahari Rao et al. 2000). On the basis of farmer in-terviews, agrometeorological studies (Gadgil et al.2002), local yield information, and models such as theUnited Nations Food and Agriculture Organization(FAO) water satisfaction index (UNFAO 2005), agroundnut-specific rainfall index was developed.The index was defined as a weighted sum of cu-mulative rainfall during the period from May 11to October 17, the average calendar dates for thegroundnut growing season. Individual weights



were assigned to consecutive ten-day periods ofthe growing season, so the index gave more weight to the critical periods during the crop’s evolutionwhen groundnut is most vulnerable to rainfallvariability. Furthermore, a ten-day cap on rainfallof two hundred millimeters was introduced to theindex because excessive rain does not contribute toplant growth. The individual weights were deter-mined by groundnut water requirements, as ad-vised by local agrometeorologists, that maximizedcorrelation between district groundnut yields andthe rainfall index (Figure A2.3) but defined homoge-nous rainfall periods, making the contract under-standable and more marketable to the farmers andless susceptible to basis risk (see Appendix 1). Moreinformation on the index construction can be foundin Hess (2003).

The average or reference weighted index valuefor groundnut and castor at the Mahahbubnagar

weather station were determined to be 653mmand 439mm, respectively. These reference-weightedindex values represent the expected growing con-ditions that produce satisfactory yields for farmersof these crops in the region. The weather insurancecontracts were designed so that payouts started at95 percent of this reference level. Farmers partic-ipating in the program received a payment if theindex fell below the predetermined threshold, indicating that the insured should be granted anindemnity to cover lost production and inputcosts as a result of lower than expected yields. Theinitial pilot limited how much insurance a farmercould purchase by offering three different fixedcontracts depending on the size holding of thefarmer wanting to buy the insurance (Table A2.7).The payout schedule as a function of index forsmall, medium, and large farmers is given in Figure A2.4.

Figure A2.3 Mahahbubnagar District Groundnut Yields Versus Groundnut Rainfall Index

01971 1973 1975 1977 1979 1981 1983 1985 1987 1989 1991 1993 1995 1997 1999 2001

100

200

300

400

500

600

700

800

900

1000

Year

Gro

undn

utyi

eld

(kg/

hect

are)

0

200

400

600

800

1000

1200

Gro

undn

utra

infa

llin

dex*

(mm

)

Mahabubnagar district groundnut yields

Rainfall index measured atMahabubnagar weather station

Correlation: 45% 1971–2001, 69% 1997–2001

Source: District yield data are from the government of Andrha Pradesh, Bureau of Statistics and Economics in Hyderabad. Rainfall data from1994–1996 are missing.

The Marketing and Sales Campaign

The products were marketed and sold by KBS LABextension officers to the four villages through work-shops and meetings with the BUA members. Thesales period ended on April 30, 2003. In total, 230farmers bought the insurance: 154 groundnut farm-ers and 76 castor farmers, most having small land

holdings. Of the 154 groundnut farmers, 102 werewomen who belonged to Velugu (light) self-helpgroups. Velugu works with four hundred thou-sand poor women organized into self-help groupsin Andhra Pradesh. Funded by the World Bank,Velugu is implemented by the Society for Elim-ination of Rural Poverty (SERP), an autonomoussociety set up by the government of Andhra Pradeshto fulfill its poverty alleviation objectives. Thewomen were keen to purchase protection againstthe vagaries of the monsoon, as all their householdsand most of their fellow villagers grew groundnut.These fellow villagers were the primary customersof the women in the self-help groups; thereforethese women felt the impacts of a poor monsoonseason additionally through drops in sales and pur-chases of their services and hence wanted to protectthemselves also.

The entire portfolio of weather insurance con-tracts sold by BASIX was insured by ICICI Lombard,with reinsurance through one of the leading inter-national reinsurance companies. ICICI Lombardfiled all the necessary forms and terms of insur-ance with the Indian insurance regulator, regis-tering their products before the program waslaunched.

At the end of the contract term, the final valuesof the weighted indices at Mahahbubnagar weatherstation were calculated by multiplying the cumula-tive rainfall totals in each ten-day period fromMay 11 to October 17, 2003, by the specific weightassigned to that period. The weighted rainfall in-dices for groundnut and castor were calculated to be516mm and 490mm, respectively, for khariff 2003,triggering a payout for groundnut farmers and no


Table A2.7 Weather Insurance Contracts Offered to Groundnut and Castor Farmers

Category Premium (Rs) Farmer Eligibility Sum Insured (Rs)

GroundnutSmall 450 < 2.5 acres land holding 14,000Medium 600 2.5–5 acres land holding 20,000Large 900 > 5 acre land holding 30,000

CastorSmall 255 < 2.5 acre land holding 8,000Medium 395 > 2.5 acre land holding 18,000

Source: Authors.

Figure A2.4 Payout Structure of Groundnut WeatherInsurance Policy Held by Farmers with Small, Medium, and Large Land Holdings

0

5000

10000

15000

20000

25000

30000

0 10 20 30 40 50 60 70 80 90 100

Payo

utof

polic

y(e

xclu

ding

prem

ium

)(R

s.)

Percentage of reference groundnut rainfall index

Small farmers

Medium farmers

Large farmers

Source: Authors.


payout for castor farmers. Groundnut farmers withsmall, medium, and large holdings recovered Rs 320,Rs 400, and Rs 480, respectively, within two weeksof the end date of the contract, after the rainfall datawere collected and crosschecked by the IMD (seeTable A2.8).

Farmer Feedback

The overall farmer feedback from the first pilotwas positive; the farmers welcomed the new prod-uct and appreciated the objective nature of theweather insurance contracts and the timely pay-ment of claims. In particular, groundnut farmersreceived a timely recovery from the policies theypurchased, even though the Mahahbubnagar dis-trict was not declared a drought area by the gov-ernment of Andhra Pradesh in 2003 and, as aresult, no payments were made from the govern-ment’s crop insurance program. The followingpositive aspects of the pilot, as reported by KBSLAB from feedback sessions with the BUA mem-bers in Pamireddypally in January 2004, includedthe following:

• Farmers had the opportunity to reflect onrainfall shortages and the economic lossesassociated with them and to learn about theconcept and process of rainfall insurance;

• Farmers were happy that they could buy rain-fall insurance to protect themselves from themost critical risk to their farming operations;

• The product was introduced through KBSLAB, a credible source of services and facili-ties for the farmers; and

• Claims were paid in a timely manner.

Some shortfalls were perceived in the product de-sign, however; in particular, the farmers expectedthat more weight would be given to the initial sow-ing period of groundnut. Moisture stress at sowingwas associated with the greatest financial risk forfarmers, as the farmers invest most of their pro-duction costs at sowing time. If the plants do notgerminate and survive the establishment period,the entire crop will be lost along with the invest-ment costs, and the farmer will have to resow, incurring further input and production expendi-tures. In 2003, for example, the groundnut farmersexpected a greater payout than the amount recov-ered, as the rains during sowing were delayedand not optimal. The farmers felt the index didnot properly reflect that most of the investment in

the crop was made at the beginning of the growingseason; they believed more emphasis should havebeen given to this phase. Other shortfalls, as re-ported by KBS LAB after feedback sessions withthe BUA members in January 2004, included thefollowing:

• Rainfall data were collected at Mahahbubnagarweather station, but the farmers felt the sta-tion did not represent the rainfall of their vil-lage well.

• Claim calculation criteria were not clearlycommunicated to the farmers during the salesand marketing campaign; in particular, thefarmers were more comfortable with indexingclaims in millimeters rather than in percentilepoints, and the farmers did not understandthe nonlinear payout function of the insur-ance contract and were expecting a linear re-lationship between the rainfall index and theclaim amount. In 2003, for example, a 22 per-cent shortfall occurred in the rainfall index;hence the farmers expected Rs 2,800 as theclaim amount: 22 percent of the Rs 14,000 suminsured for small-hold farmers.

• Farmers felt that the product should offerphase-wise payouts for each growing phase,subject to the maximum limits, so that it wouldbe clear how the weights and therefore payoutsrelated to each growing stage. The farmersalso requested that in the future the insurancecompany send a progress report on the rain-fall for each of the crop phases in order to

Table A2.8 Pilot Statistics, 2003

Statistic Groundnut Castor Total

Total number of 154 76 230farmers insuredAggregate value 2,250,000 858,000 3,108,000of insurance (Rs)Aggregate premium 71,700 22,880 94,580paid (Rs)Aggregate amount 50,417 0 50,417of claims (Rs)Net Incurred Claim 70.3 0 53.3to Net Premium Earned (%)

Source: KBS LAB.

facilitate a better understanding within thefarming community.

• Farmers noted that excess rainfall at harvestcould result in severe crop losses and requestedthat protection against the risk of excess rain-fall be offered under the weather insuranceproduct.

Second Pilot Program: 2004

The second pilot program in khariff 2004 intro-duced significant changes to the 2003 design. Theprogram was extended to four new weather refer-ence station locations in two additional districts inAndhra Pradesh: Khammam and Anantapur. Theweather insurance contracts were offered to bothBASIX borrowers and nonborrowers and were mar-keted and sold through KBS LAB in Khammamand Mahahbubnagar districts and through BhartiyaSamruddhi Finance Ltd. (BSFL)65 in Anantapur dis-trict at village meetings, farmer workshops, andfeedback sessions in the month leading up to thegroundnut and castor growing season. A portion ofthe weather insurance contracts were written onlocal rain gauges monitored by the government ofAndhra Pradesh, rather than on the district IMD sta-tions. Because 60 percent of agriculture in AndhraPradesh is rain-fed, the government of AndreaPradesh maintains a network of 1,108 rain gaugesthroughout the state. This monitoring is done at thesmallest administrative unit in the state, known asa mandal, which is a grouping of approximately fif-teen villages. In Andhra Pradesh there are forty tofifty mandals in each district, and each mandal hasone rain gauge: 232 of the rain gauges are owned bythe IMD, and all conform to World MeteorologicalOrganization specifications. Records begin in 1956,and historical data can be purchased from theGovernment Bureau of Statistics and Economics inHyderabad. The second pilot used these rain gauges,

and, as a result, in general all rain gauges were tenkilometers away from the faming villages involvedin the scheme. This limited the basis risk to farmers,because the gauges were closer to their actual farms,but made it more difficult and indeed impossible tofind international reinsurance for the final portfolioof weather insurance contracts sold by BASIX andinsured by ICICI Lombard. In 2004, therefore, ICICILombard chose to keep the risk itself without inter-national reinsurance support.

The biggest difference in 2004, however, was thedesign of the weather insurance contracts. In light ofthe farmer feedback from khariff 2003, the droughtprotection products for 2004 were structured by di-viding the groundnut and castor growing seasonsinto three phases each, corresponding to the plants’three critical growing periods: (1) establishment andvegetative growth, (2) flowering and pod formation,and (3) pod filling and maturity. With a departurefrom the weighted index design, the new contractsspecified a cumulative rainfall trigger for each ofthe three phases, with an individual payout rateand limit for each phase. The groundnut droughtinsurance policy offered to farmers in Narayanpetmandal in Mahahbubnagar district, for example,appears in Table A2.9.

Trigger levels and payout rates were determinedin consultation with local agrometeorologists andfarmers and with reference to local yield data asin 2003. Premiums and threshold levels vary byweather station, depending on the risk profile ofeach individual location. This simplified design wasintroduced to give clarity to the recovery processby clearly associating each critical growth phasewith an individual deficit rainfall protection struc-ture. If the rainfall deficit reached the lower limitin each phase, the total payout limit for that phasewould be triggered to indemnify farmers for thesevere corresponding crop losses associated with


Table A2.9 Payout Structure Per Acre for Groundnut Weather Insurance Policy for Narayanpet Mandal, Mahahbubnagar District (2004)

Phase Dates Strike (mm) Limit (mm) Payout Rate (Rs) Limit (Rs)

Establishment and Vegetative Growth June 10–July 14 75 20 15 3,000Flowering and Pod Formation July 15–August 28 110 40 10 2,000Pod Filling and Maturity August 29–October 2 75 10 5 1,000

Source: Authors.


the lack of rainfall. Figure A2.5 shows the contractpayout structure. In a further departure from the2003 pilot, the contracts were designed to be soldper acre.

A farmer could buy as many acres of protectionas he wished, provided he actually cultivated thatmany acres of the crop to be insured. The premiumassociated with the product in Table A2.9 is Rs 250per acre insured, for a sum insured of Rs 6,000 peracre. New contracts were also offered for cottonfarmers in Khammam district, and an excess rain-fall product for harvest was offered to all castor andgroundnut farmers with the structure shown inTable A2.10.

In total, over 400 farmers bought insurancethrough BASIX in 2004, and a further 320 ground-nut farmers, members of a Velugu self-help grouporganization in Anantapur district, bought insur-ance directly from ICICI Lombard. Several farmerswere repeat customers from the 2003 pilot. In con-trast to 2003, ICICI Lombard did not seek re-insurance for the BASIX farmer weather insuranceportfolio in 2004. As in 2003, all contracts were set-tled promptly, within thirty days of the end of thecalculation period. An example of the marketingleaflet developed by KBS LAB and ICICI Lombarddetailing the weather insurance contracts for cas-tor, groundnut, and excess rainfall for Narayanpetmandal is shown in Figure A2.6. For example, inkhariff 2004, the rainfall in Narayanpet mandalwas not good for groundnut farmers. The rainfallrecorded at the local mandal rain gauge measured12mm for Phase 1 and 84.2mm for Phase 2; rain-fall during Phase 3 was above average, at 112mm.Farmers who bought this policy received a payoutof Rs 3,258 per acre insured on September 22, 2004.

In autumn 2004, CRMG commissioned a base-line survey to be conducted for the World Bank bythe International Crops Research Institute for theSemi-Arid Tropics (ICRISAT) in Hyderabad to as-certain the overall farmer feedback for the first twoyears of weather insurance. The survey, involvingone thousand farmers, some of whom have beeninvolved in both pilot programs, will be used asbase from which the impact, efficiency, and accept-ability of the weather insurance concept can bemeasured. The results provide strong guidelinesand direction for future weather insurance pro-grams in India, particularly regarding the issues ofscalability and sustainability. The results also in-dicate how these new products function in theoverall rural finance framework, with particular

emphasis on access to credit and credit repaymentby farmers.

The Future for BASIX Weather Insurance

In 2004, a number of other transactions also tookplace within the Indian private sector in responseto the 2003 pilot. In 2004, BASIX bought a crop-loan

Figure A2.5 Payout Structure of Groundnut WeatherInsurance Policy for Narayanpet Mandal,Mahahbubnagar District, 2004

0

500

1000

1500

2000

2500

3500

3000

0 20 40 60 80 100 120 140

Cumulative phase rainfall (mm)

Payo

utpe

rph

ase

(Rs.

)

Source: Authors.

Phase 1

Phase 2

Phase 3

Table A2.10 Payout Structure Per Acre for Castorand Groundnut Excess Rainfall WeatherInsurance Policy for Narayanpet,Mahahbubnagar

Dates September 1–October 10Rainy Day Index Daily rainfall greater than

or equal to 10 mmPremium Rs200 per acre insuredLimit Rs6,000 per acre insured

Excess Rainfall Payout Structure

Number of Consecutive Rainy Days Claim Amount (Rs)

4 1,5005 1,5006 3,000≥ 7 6,000

Source: Authors.

portfolio insurance policy based on weather indices.For the first time, BASIX used this protection tocover its own risk and passed neither the cost nor thebenefits to its client farmers. The protection allowedBASIX to keep lending to drought-prone areas bymitigating default risk through the insurance policyclaims in extreme drought years. BASIX bought apolicy to cover three business locations, which wasinsured by ICICI Lombard and then reinsured intothe international weather market.

In 2005, BASIX scaled-up the weather insuranceprogram for farmers, extending the projects to allof their branches in seven Indian states for khariff2005, with a sales target of ten thousand policies.BASIX sold 7,685 policies to 6,703 customers inthirty-six locations in six Indian states during the2005 monsoon season. The new policies featured a

dynamic contract start date determined by a rain-fall trigger and minimum and maximum limits tothe rainfall counted (for example, rainfall belowtwo millimeters per day is not counted). In addi-tion, BASIX simplified and largely automated theunderwriting process, which is why BASIX couldroll out weather insurance to every branch. Intensetraining sessions with loan officers, who becameliterally one-stop-shop full customer service agents,allowed BASIX to service a large array of rainfallinsurance products. At the same time, the policiesbecame more general “monsoon failure” policies,meaning they were area-specific rather than crop-specific products, targeting general livelihood lossesof farmers that have diversified agricultural port-folios at risk to weather, rather than losses associ-ated with yield variations of a specific crop. For the


Figure A2.6 An Example of the Marketing Leaflet for Groundnut (DGN), Castor (DCN), and Excess Rainfall (EN) Protection in Narayanpet Mandal, Mahahbubnagar District, 2004

Source: ICICI Lombard/BASIX.


first time BAISX also worked with another insur-ance provider, NAIC, as well as ICICI Lombard, tosell weather insurance policies in some locations. In2005, over seventy new automated weather stationswere installed throughout India, by private com-pany Delhi-based National Collateral ManagementServices Limited (NCMSL) in partnership with ICICILombard, on which weather insurance contractswere written, including many BASIX contracts. Byestablishing stations closer to the farmers, BASIXhad more reliable automatic stations as settlementbases for their contracts and more accurate productsfor their farmers. NCMSL plans to scale-up their in-stallations throughout the country with more insur-ance provider partners in 2006, which will benefitend users like BASIX in subsequent seasons.

BASIX is also interested in making the insuranceavailable to landless laborers and self-help groupwomen in its operating regions, whose livelihoodsalso suffer from the vagaries of the monsoon. In2004, three hundred women bought a weather in-surance policy from ICICI Lombard directly, travel-ing by train to Hyderabad.

BASIX’s ultimate goal is to offer weather-indexedloans to their borrowers. BASIX can package a loanand a weather insurance contract (Hess 2003), basedon the drought indices described above, for exam-ple, into one product, such as a weather-indexedgroundnut production loan. The farmer would enterinto a loan agreement with a higher interest rate thataccounts for the weather insurance premium thatBASIX would pay to the insurer. In return, in theevent of a drought as defined by the index, the farmerwill not repay all the dues. In the event of a moder-ate drought, instead of paying the loan principal andinterest, the farmer would repay the principle only;in the event of a severe drought, he would only needto repay part of the principle.

During 2004 and 2005, not only did BASIX ex-pand their weather insurance program, a numberof other institutions, including the originator, ICICILombard, began expanding the market for weatherinsurance in India. IFCCO-Tokio, a joint venture in-surance company, launched weather insurance con-tracts similar to the 2003 contracts in 2004, sellingover three thousand policies to farmers throughoutIndia in 2004 and over sixteen thousand in 2005. Inconjunction with ICICI Lombard, the governmentof Rajasthan launched a weather insurance pro-gram for farmers for the 2004 growing seasons, in-suring 783 orange farmers from insufficient rainfallin khariff 2004 and 1036 coriander farmers in rabi

2004; this was scaled up to include more cropsand farmers in 2005. The NAIC, responsible forthe government-sponsored area-yield indexed cropinsurance scheme, also launched a pilot weatherinsurance scheme for twenty districts throughoutthe country in 2004, reaching nearly 13,000 farm-ers; the scheme was even mentioned in the gov-ernment of India budget for the financial year 2004to 2005. In 2005, NAIC sold weather insurance toapproximately 125,000 farmers throughout India. Inthe same year, ICICI Lombard scaled up its agri-cultural weather insurance sales, reaching ap-proximately 100,000 farmers, and expanded intoother economic sectors. New insurance providerssuch as HDFC Chubb also entered the market in2005. In total it is estimated that during kharrif 2005250,000 farmers bought weather insurance through-out the country. Given this strong level of interestand the potential size of the end user market, agri-culture weather risk management in India is setto grow (Divyakirti 2004).

Weather Insurance for Agriculture in Ukraine66

Ukraine is one of the biggest grain and oilseed pro-ducers in the world and the agricultural sector is ofgreat importance for the national economy: agri-culture accounts for 14 percent of the country’sGDP.67 For their production, Ukrainian farmers facemultiple perils, such as drought, excess rain, andfrost, which make their incomes unpredictable andlimit their access to credit.

Empirical evidence demonstrates that the largestrisk to crop production in the Kherson oblast(province) is weather, namely drought in springand summer and low temperatures in winter.Traditional multiple-peril products offered by localinsurance companies somewhat addressed winterrisks, but drought coverage was excluded from theinsurance products available to farmers. In addi-tion, the insurance companies did not have the pro-fessional staff with agricultural expertise nor theinfrastructure necessary to offer comprehensiveagricultural insurance products. Consequently thefarmers did not trust the insurance companies andthe policies offered. High administrative costs andasymmetry of information further compoundedthese problems, rendering the agricultural insurancesystem in the country ineffective.

In 2001, the CRMG introduced the concept ofindex-based weather insurance to Ukraine in col-

laboration with IFC-PEP. The concept of weatherinsurance appeared particularly feasible in Ukrainebecause of a widespread system of 187 weather sta-tions, eight in Kherson, and the excellent quality ofdata. After extensive consultations with the farm-ers, local authorities, and agricultural scientists, IFC-PEP decided to investigate the feasibility of weatherinsurance in the southern oblast of Kherson. In orderto reach the acceptable volume of contract sales, IFC-PEP decided that the weather pilot project shouldconcentrate on regional farmers’ most importantcrops susceptible to weather risk. Potential cropsincluded winter wheat, spring barley, sunflower,and corn. Of these, winter wheat has the biggestplanted area and considerable value at risk: 1.5 to2 million tons is produced in the oblast annuallywith an approximate crop value of US$200 million,and, in addition, most of this crop is cultivatedwithout irrigation. Furthermore, financial institu-tions in the oblast had recently started to acceptstanding crops of grain as security for agriculturalloans, despite concerns over lack of sufficient insur-ance protection.

With this basis in 2004, the CRMG together withIFC-PEP Agribusiness Development Project agreedto run a small pilot project for the Kherson oblast inspring 2005.

The Kherson Oblast

A cursory glance at winter wheat yield data for theKherson oblast shows a significant interannual vari-ability in yield in the region (Figure A2.7), which re-flects the agroclimatic risk inherent to the oblast.Formal interviews with winter wheat farmers in theregion indicated the greatest perceived risks wererelated to weather.

Designing the Index

Historical yield data for Kherson are unreliable (notreported accurately) for the purposes of index con-struction, as the data does not faithfully representthe actual production in the rayons (subregions) ofthe oblast. In order to design an effective weatherrisk management instrument, key weather factorshad to be discussed with experts, such as agrome-teorologists and farmers, and crop models usingweather variables as inputs for yield estimates hadto be developed. To this end, a report (Adamenko2004) was commissioned by the CRMG and ICF-PEPfrom the Ukrainian Hydrometeorological Center(UHC) in Kiev to assess the agroclimatic conditionsand weather risks for growing winter wheat in theKherson oblast. In the absence of reliable yield data,expert assessment and the results from the reportbased on the UHC oblast-specific crop model wereused as the basis for constructing an appropriateweather index for winter wheat in Kherson.

Identified Weather Risks

According to the UHC report (Adamenko 2004) themost significant weather risks for growing winterwheat in the Kherson oblast are (1) winterkill dur-ing the crop’s hibernation period from December toMarch, and (2) moisture stress during the vegeta-tive growth period from mid-April to June.

Winter wheat yields at harvest depend to a greatextent on how well the plants survive the winter andthe hibernation period. In the territory of Kherson,the primary cause of winter wheat winter crop deathis one day or more of air temperature and, therefore,soil temperature below the critical level. These win-terkill events cause damage and death of the plants’tillering node. Snow cover considerably improvesconditions for winter wheat hibernation, as the dif-ference between air and soil temperature increaseby 0.5 to 1.1°C for each centimeter of snow cover.The crop usually dies in years without snow coveror when the stable snow cover appears late in win-ter, as it did in 2003.


Figure A2.7 Winter Wheat Yields for Kherson Oblast,1971–2001

10

15

20

25

30

35

40

45

1971 1976 1981 1986 1991 1996 2001Harvest year

Win

ter

whe

atyi

eld

(qui

ntal

spe

rhe

ctar

e)

Source: Hess et al. 2005.


Low moisture is the other main limiting factorfor high winter wheat yields in the Kherson oblast.In fact, lack of moisture in the soil and air duringthe vegetative growth period is the main cause oflow winter wheat yields. In particular, all five rayonsof the oblast are subject to frequent droughts; theprobability of a severe and medium drought (de-fined subsequently) during the vegetative period inthe region is 15 to 20 percent and 40 to 50 percent,respectively. The first critical period in which winterwheat yield formation is highly susceptible to mois-ture stress is the phase from leaf-tube formation toearing. Due to the climatic conditions of the region,this period lasts from April 15 to May 25. The waterrequirements for winter wheat during this stage,when compared to the climatic conditions for thisperiod for the oblast, are estimated by the UHC tobe 80 percent of the optimum. During the most re-cent years, in 50 percent of cases the moisture con-ditions during this period were close to optimum(1998, 1999, 2001), while in the other 50 percent ofcases they were insufficient (2000, 2002, and 2003).The second critical period for winter wheat is thephase from earing to milk ripeness, which is thekernel formation stage; this lasts, on average, fromMay 22 to June 14, but it can extend later into June.Lack of moisture during this period directly de-creases the number of kernels in a wheat ear andleads to excessive drying of the kernels. The waterrequirements for winter wheat during this stage,when compared to the climatic conditions for thisperiod for the oblast, are estimated by the UHC tobe 90 percent of the optimum.

The Selyaninov Hydrothermal Ratio Index (SHRI)68

The previous findings indicate the need to includedrought risk in a meaningful insurance product.An example of a product that has been suggestedfor Kherson oblast is outlined in this section. Agri-cultural drought can take two forms: air droughtand soil drought. Air drought describes conditionsin which precipitation is low and high air tempera-ture persists against a background of low relativeair humidity. This leads to unfavorable conditionsfor plant vegetation and drastically reduces cropyields. Soil drought describes the excessive drynessof soil, resulting in a scarce supply of moisture avail-able for crop growth and development. Air drought,characterized by a long rainless period, high air tem-perature, and low air humidity, is often describedusing the Selyaninov Hydrothermal Ratio (SHR).For the vegetative growth period for winter wheat

in Kherson, April 15 to June 30, the SHR is definedas follows:

It holds for periods when daily average tempera-tures are consistently above +10°C. This period, onaverage, begins on April 15 in the Kherson oblast.The SHR does not always serve as a reliable crite-rion of agricultural drought because it does not ac-count for soil moisture, but because soil dryness,unlike rainfall and average temperature, is gener-ally not an observed variable, the SHR is the only ob-jective indicator that can be used to capture droughtrisk during the vegetative period. Conditions forobtaining the best harvest are when the SHR isbetween 1.0 and 1.4. When the SHR is greaterthan or equal to 1.6, plant yields will be depressedby excessive moisture. When the SHR is less thanor equal to 0.6, plants are depressed by droughtconditions. In general, the isoline SHR = 0.5 coin-cides with regions of semidesert climate condi-tions. Results from the UHC crop model (Adamenko2004) that suggest the impact on yields of SHR dur-ing the vegetative growth stage between April 15and June 30 are defined in Table A2.11.

The SHR can therefore be used as an index tomonitor the impact of air drought on winter wheatcrop yields.

Quantifying the Impact of Weather

There are two possible levels for weather insuranceprotection that can identify the appropriate limit for

SHR Daily

Average

April June=

×

∑ -Rainfall

15

0 1. DDailyApril June

�-

Temperature15∑( )

Table A2.11 Relationship Between SHR and WinterWheat Yields During the Vegetative Growth Phase of Plant Development

SHR Description Yield Loss (%)

1.6 Excessive humidity 30+1.3–1.6 Damp —1.2–1.0 Sufficient humidity —0.9–0.7 Dry —< 0.7 Drought conditions —0.5–0.6 Medium drought 200.4–0.5 Severe drought 20–50< 0.4 Extreme drought 50+


a weather insurance contract: Production costs andexpected revenue. The former, in general, is moreappropriate for catastrophic weather risks early inthe growing season, such as winterkill, when thefarmer has an opportunity to resow another cropfor summer harvest if the winter wheat crop iscompletely destroyed. The latter is, in general,more appropriate for weather risks later in thegrowing season, when there is no opportunity forresowing, yet conditions, such as an April to Junedrought, can cause yield to vary significantly fromthe expected levels. The choice of a factor, how-ever, depends on the preferences of the farmer.Informal interviews with farmers in the oblast in-dicate that farmers are less concerned with win-terkill risk than with drought risk, even though itcan potentially cause complete damage, because ofthe potential to resow.

Winter wheat farmers spend a maximum of(Ukrainian Hryvna) UAH 1000 per hectare on pro-duction and inputs costs during the crop’s entiregrowing season. The limit of a mid-April to Junedrought insurance contract to cover productionand input costs should therefore be set at UAH1000 per hectare insured. In the event of total cropfailure as a result of a very extreme drought, forexample, say a SHR < 0.15 event, the farmer wouldbe indemnified for UAH 1000 per hectare insuredto compensate for the loss of the investment. Thepayout rate of the insurance contract can be deter-mined from the information in the UHC report andis summarized in Table A2.12.

Calculating the limit and payout rate for a con-tract to protect farmer revenue is a little more diffi-cult, as harvest-time commodity prices are notknown in advance when the insurance is purchased.Furthermore, commodity prices also often vary inresponse to extreme production shocks, and it isoften difficult to quantify the production (weather)price correlation. Estimates for the harvest-timeprice can be made, however; for example, the pre-vious year’s harvest-price or the five-year averageof the September price from the local commoditiesexchange could be used as a best estimate, or thegovernment minimum support price could be usedas a lower boundary for the selling price.

Structuring a Weather Insurance Contract

The Sum Insured

In order to ensure that the insurance product hassome relationship with the true risk exposure of thefarmer, the limit of the insurance contract is nego-tiable with the farmer; however, it cannot exceed amaximum estimated by the potential insured lossto the farmer, as outlined in above. In the design ofthe contract, an upper limit on the risk volume perclient will be set at the total area of the crop plantedmultiplied by the expected selling price, determinedas mentioned above by the previous year’s sellingprice according to records, the five-year average, orthe government’s minimum support price.

Contract Specifications

As outlined in Appendix 1, in addition to definingthe index, the buyer/seller information (names,crop, and hectarage insured), limit and tick-size, anindex-based weather insurance contract must alsoinclude the location (weather station of reference),the calculation period, the strike or deductible, andthe premium. In the case of Ukraine, to provide thebest possible coverage for the farmer client, index-based insurance contracts must be written on theUHC weather station nearest to the farmer’s land.Indeed, the extent of the UHC weather observingnetwork may be a limiting factor for the applicabil-ity of this type of insurance in regions that do nothave a UHC station. The correlation coefficientsfor the interannual variation in cumulative rain-fall, cumulative average temperature, and SHR forApril 15 to June 30 from 1973 to 2002 for five weatherstations in the oblast are given in Table A2.13.


Table A2.12 Relationship Between SHR and FinancialLosses Associated with Winter WheatYield Fluctuations

SHR Payout per Hectare

0.6–0.51 UAH 200 (20% loss)0.5–0.46 UAH 300 (30% loss)0.45–0.41 UAH 400 (40% loss)0.4–0.36 UAH 500 (50% loss)0.35–0.31 UAH 600 (60% loss)0.3–0.26 UAH 700 (70% loss)0.25–0.21 UAH 800 (80% loss)0.2–0.16 UAH 900 (90% loss)< 0.15 UAH 1000 (100% loss)



A very loose rule-of-thumb is that farmers liv-ing within a thirty kilometer radius of the weatherstations may purchase weather insurance indexedto that station. Temperature exhibits less spatialvariability than does rainfall. The benefit of theSHR index is that, by combining cumulative rain-fall with temperature, the spatial variability of theindex, in comparison to indexes of cumulativerainfall alone, is slightly reduced. In this example,the calculation period for the SHR drought insur-ance contract is April 15 to June 30 to cover the

leaf-tubing to kernel formation growth period ofwinter wheat. Final settlement of the weather in-surance contracts typically would occur up toforty-five days after the end of the calculation pe-riod, once the collected weather data have beencross-checked and quality controlled by the UHC.The strike would be set at a predefined SHR levelappropriate to the weather station under consid-eration. A pricing example for winter wheatdrought risk is given below for Behtery weatherstation.

Table A2.13 Correlation Coefficients for the Interannual Variability of Cumulative Rainfall, Average Temperature, and the SHR Index Measured at Five UHC Weather Stations in Kherson Oblast

Station Name Behtery Genichesk Kherson N Kahowka N Sirogozy Station Location

April 15–June 30 Cumulative Rainfall Correlation Coefficients (1973–2002)

Behtery 1 46′15″ N32′18″ E

Genichesk 0.72 1 46′10″ N34′49″ E

Kherson 0.74 0.59 1 46′38″ N32′34″ E

N Kahowka 0.70 0.41 0.65 1 46′49″ N33′29″ E

N Sirogozy 0.35 0.54 0.39 0.50 1 46′51″ N34′24″ E

April 15–June 30 Cumulative Temperature Correlation Coefficients (1973–2002)

Behtery 1 46′15″ N32′18″ E

Genichesk 0.93 1 46′10″ N34′49″ E

Kherson 0.98 0.93 1 46′38″ N32′34″ E

N Kahowka 0.98 0.95 0.99 1 46′49″ N33′29″ E

N Sirogozy 0.95 0.95 0.98 0.98 1 46′51″ N34′24″ E

April 15–June 30 SHR Correlation Coefficients (1973–2002)

Behtery 1 46′15″ N32′18″ E

Genichesk 0.72 1 46′10″ N34′49″ E

Kherson 0.74 0.59 1 46′38″ N32′34″ E

N Kahowka 0.74 0.44 0.68 1 46′49″ N33′29″ E

N Sirogozy 0.38 0.58 0.42 0.50 1 46′51″ N34′24″ E



Figure A2.8 Cumulative Rainfall and Average Temperature for Behtery Weather Station for April 15 to June 30, 1973–2002

0

20

40

60

80

100

120

140

160

180

200

1970 1975 1980 1985 1990 1995 2000 2005Year

Cum

ulat

ive

rain

fall

(mm

)

0

200

400

600

800

1000

1200

1400

1600

Cum

ulat

ive

daily

aver

age

tem

pera

ture

(deg

C)

Cumulative rainfall

Raw cumulative temperature

Detrended cumulative temperature

Source: Authors.

Example: Pricing Drought Risk as Measuredby the SHR Index

In Behtery, droughts of varying intensity happenquite frequently. Although irrigation is partiallyused by farmers in this area, farmers have ex-pressed interest in products that protect against ex-treme drought. Figure A2.8 shows the cumulativeaverage temperature and cumulative daily rainfallmeasured at the Behtery station from 15 April to 30June 1973 to 2002. The temperature data exhibitstrong trends, hence the data must be detrended tomake the historical data consistent with recentwarmer conditions that may make severe droughtevents more frequent in Behtery now than thirtyyears ago. The weather data from the UHC are ofhigh quality and do not need to be cleaned or qual-ity controlled prior to analysis. The data are de-trended by fitting and removing a best-fit leastmean square linear trend to the cumulative averagetemperature totals for April 15 to June 30 (seeAppendix 1). Figure A2.9 shows the correspondingSHR index: medium droughts (SHR < 0.6) have oc-curred nine times in the past thirty years and severedroughts (SHR < 0.4) twice. The driest conditionsoccurred in 1996, with SHR = 0.21.

The payout of a SHR index insurance contract atBehtery is determined by the following equation:

where K is the strike, SHR is the SHR index mea-sured during the calculation period, X is the payoutrate, determined by the structure of the contract,and M is the limit of the contract. A reasonable es-timate for the risk loading factors α, β, given pricesin the weather market, are α = 25% and β = 5%. Bysimply taking the thirty years of payouts in FigureA2.9, the payout statistics for a weather insurancecontract with a strike level of SHR = 0.4 can be cal-culated as follows: E(SHR) = UAH 70, σ(SHR) =UAH 220 and VaR97(SHR) UAH 800. A first-orderestimate of an appropriate premium to charge afarmer for an insurance contract with a strike levelof SHR = 0.4 at Behtery Weather Station, therefore, isbetween UAH 110 and 125 per hectare for a sum in-sured of UAH 1000.69 (See Figure A2.10 for the termsof an example of a prototype contract for Behtery.)

The 2005 Pilot in KhersonAccording to Ukrainian legislation, in order to beable to introduce a new product, such as index-

Payout K SHR X M= −( ) ×( )min max 0, ,


based weather insurance, to the market, the par-ticipating company (or companies) must designand register the rules of insurance with the stateregulatory body. Although the law on insurance—the leading document regulating the insuranceindustry—does not specifically reference “index”insurance, other legislative documents introduceindex-based products in relation to agricultural ap-plications; for example, relating to agricultural in-surance and state finance support of the agriculturalsector. As a result, there was no direct legislativebarrier prohibiting the use of index-based productsin Ukraine. In April 2005, the regulator agreed toregister rules of insurance that permit the develop-ment of different types of index-based insuranceproducts for agribusiness applications.

The insurance company partner, Kiev-basedCredo Classic, working with IFC-PEP and CRMG,submitted the necessary package of documents to the regulator in Kiev. This included draftingand registering the rules of insurance for index-based weather insurance products with the regu-lating body. The rules of insurance were acceptedat the beginning of April 2005, clearing the wayfor the first weather insurance pilot in Ukraine.The regulator confirmed that, given the nature ofthe product, the insurer is not required to carry outfield checks and loss adjustments, despite the po-

tential of basis risk. The regulator further statedthat the insured area must not be greater than theseeded area and, for the purpose of this product,a farmer’s report declaring the seeded area shouldbe sufficient proof of the maximum possible areafor insurance.

The weather insurance contract designs and mar-keting materials for the proposed pilot program inKherson were finalized following receipt of StateRegulator approval of the rules of weather indexinsurance for agricultural applications. Using feed-back and workshop sessions, IFC-PEP worked withthe insurance partner in Kherson oblast to targetgroups—including farmers, agribusinesses, andfinancial institutions—who could benefit from thenew insurance products. Only two weather insur-ance contracts protecting against drought were soldduring the brief marketing period, primarily due tothe timing of the pilot and late regulatory approval.The protection period for the first pilot finished inJuly 2005. The results of the small first pilot havebeen communicated to the public to raise awarenessabout index insurance and the pilot experience: theconcept and methodologies developed have beenmade publicly available. Presently, the insurancecompany leading the pilot in Kherson is alreadyproviding consultations to other markets playersin Ukraine on designing index-based products in-

Figure A2.9 SHR Index for Behtery Weather Station, 1973–2002


0197374

100

200

300

400

500

600

700

800

900

Harvest year

Payo

utof

exam

ple

SHR

insu

ranc

est

ruct

ure

0.00

0.20

0.40

0.60

0.80

1.00

1.20

1.40

1.60

SHR

Inde

x

Payout in UAH per hectare

SHR Index

7675 77 78 79 8180 82 83 84 8685 87 88 89 9190 92 93 94 9695 97 98 99 0100 2002


Figure A2.10 Sample Contract for Behtery Weather Station

Farmer Z

1 Wheat Street, Behtery, Kherson, UA

ABC Insurance Company

100 Hectares

April 15, 2005 to June 30, 2005 (inclusive)

Behtery Weather Station

SHR = Index 1 / ( Index 2 * Scaling Factor)

Where:

Index 1 = Cumulative Capped Daily Rainfall measured during the Calculation Period at Location. Measuring Unit: mm

Index 2 = Cumulative Daily Average Temperature measured during the Calculation Period at Location. Measuring Unit: Degrees Celsius

Scaling Factor = 0.1

Capped Daily Rainfall = min (50, Daily Rainfall Total)

Measuring Unit: mm

0.4

UAH 1000 per Hectare Insured

1. If the Index SHR is greater than the Strike K no payment is made.

2. If the Index SHR is less than or equal to the Strike K the Buyer receives a payout X per hectare insured from the Seller according to the following Settlement Calculation:

If 0.36 < max (K – SHR, 0) < 0.41, X = UAH 500

If 0.31 < max (K – SHR, 0) < 0.36, X = UAH 600

If 0.26 < max (K – SHR, 0) < 0.31, X = UAH 700

If 0.21 < max (K – SHR, 0) < 0.26, X = UAH 800

If 0.16 < max (K – SHR, 0) < 0.21, X = UAH 900

If max (K – SHR, 0) < 0.16, X = UAH 1000

The maximum payment that can be made from the Seller to the Buyer is UAH 100,000.

The Buyer will pay the Seller a premium of UAH 12,000 for the weather protection outlined above.

Ukrainian Hydrometeorological Centre, Kiev

Within 45 days of the end of the Calculation Period.

Buyer

Seller

Hectares of Winter Wheat Insured

Calculation Period

Location Behtery

Index, SHR

Capped Daily Rainfall

Strike, K

Maximum Payout, M

Settlement Calculation

Maximum Settlement

Premium

Settlement Data

Settlement Date



house and drafting the insurance rules for these newproducts. There are also plans to scale up weather in-surance activities to cover more crops and regionsin 2006.

TECHNOLOGY APPLICATIONCASE STUDIESGrassland Index Insurance Using Satellite Imagery

In recent times, the availability of new technology,such as satellite imagery, has sparked the introduc-tion of new initiatives to insure grasslands. The mostcommon technical justifications for the adoption ofsatellite imagery (SI), as the principle of area-yieldinsurance, are the following: (1) SI can measure pas-ture health and growth and represents a multiple-peril insurance approach; (2) SI can economicallyreduce the size of the area on which pasture growthand potential insurance payments are based, therebyreducing basis risk as compared to other approaches(that is, the cage clipping alternative); and (3) SI canassess pasture conditions throughout the growingseason and thereby lends itself to “intra-seasonalcoverage options.” This section will discuss the useof satellite imagery in creating useful indices to in-sure grassland following a parametric and objectiveprocedure and will describe relevant experiences inCanada and Spain, the two countries that have madethe most effective use of this kind of parametricinsurance.

Use of the Normalized Difference VegetationIndex (NDVI) for Insurance Purposes

One of the satellite networks with more informationavailable for these purposes comes from the NOAAsatellite. The NOAA satellite has blue, green, red,infrared, and thermal sensors and takes one imageper day for every square kilometer of the earth’s sur-face. The NDVI is a type of vegetative index basedon the relationship between red light and near-infrared light. Healthy vegetation absorbs the redlight from the sun and uses it for photosynthesiswhile reflecting near-infrared light from the sun.The formula used to calculate the NDVI is given by:

where NIR is near-infrared light and Red is red light.The more red light is absorbed by the plants, the

NDVI NIR NIR= −( ) +( )"Red" "Red"

smaller the amount of red light is, in turn, reflectedby the plant and recorded by the satellite, thereforethe larger the NDVI value.

Another important input for the use of NDVIas index insurance is the design of an appropriatemask. A mask is simply a set of geo-referenced information identifying specific land features thatcan be laid over the satellite imagery information.The overlaying of this information allows some ofthe satellite imagery to be extracted from the infor-mation file prior to making production assessments.

Grassland Insurance in Alberta (AFSC operated)70

In 2001, Alberta launched a pilot project usingsatellite imagery to define a historical “benchmark”production and assess annual pasture production.The pilot was limited to a geographical area of theprovince where pasture is the predominant landcover. An NDVI, scaled appropriately to reflectnative pasture production, was calculated for eachtownship in the pilot area. Insured farmers receivedpayments according to a predetermined paymentschedule when the annual township NDVI fellbelow the historical benchmark NDVI for thetownship. The program was expanded slightly in2002 to the portion of the province in which thesquare kilometer resolution (pixel image) of theNOAA satellite system was considered practicalfor pasture.71

The mask used for the project selects only infor-mation known to be at least 85 percent native or im-proved pasture at a quarter section level (160 acres).In the pilot area, where satellite imagery insur-ance operated, a significant percentage of land, 80to 90 percent, is native pasture. Areas of crop irriga-tion and some bush land also need to be extracted,or they significantly influence the program outcome.If a quarter section of land has irrigation, it is re-moved from the program dataset.

The process for calculating a township NDVIincluded the use of daily images to estimate theNDVI for each square kilometer section and scaledto identify variations in pasture observations to gen-erate a pasture vegetative index (PVI). All weekly“pixel image” PVI values within a township areaveraged to get the weekly township PVI value.While ample data existed to calculate the PVI, littleaccurate “in-field” pasture information was avail-able to judge whether the PVI actually correlatedto pasture growth. In the past, however, AFSC had

operated a cage clipping system that allowed it toobtain production estimates. The availability of in-formation allowed pursuit of a statistical procedureto assess the efficiency of the index indicator to re-flect the variations in volume of grassland, basi-cally by comparing historical PVI values to pastureproduction trends over time, and to confirm anycorrelation with farmers.

The pasture production data were available forcorrelation comparisons from 1991 to 1999 from thecage clippings at designated and consistent sites.In addition, AFSC personnel compared satellite im-agery to trends in precipitation measured at selectEnvironment Canada weather stations. Correlationresults, however, were not good (approximatelyr = 0.65). Through a series of client meetings, AFSCasked farmers to identify their two best and twoworst pasture production years in the last fifteen-year period. Since a PVI value could be calculatedfor each township from 1987 to 2000, farmers couldsee whether the extreme PVI values compared totheir recollections of historical pasture productiontrends. Production shortfalls due to drought andcool early season temperatures appeared to be iden-tified in the historical PVI values. Geographical dif-ferences among township PVI values correspondedto the anecdotal production perceptions of farmerssurveyed.

To augment the information acquired by satelliteimagery, AFSC developed research plots through-out the pilot pasture area to measure rainfall andthe growth of pasture under cages and to notechanging pasture conditions over the growing sea-son throughout the pilot area (thirty in total). Thecorrelations were improved substantially throughthis process.

Pasture insurance is sold in the spring of eachyear, but farmers must make their purchasing de-cisions by the end of February. Farmers must in-sure all the acres of pasture within the samecategory—native, improved, or bush pasture—buta lower than normal PVI value in one township isnot offset by a higher than normal PVI in another.Coverage and premium are expressed in dollarsand derived by multiplying the pounds of pastureproduction expected in each forage risk area, asdetermined by AFSC, by 80 percent of one of thefour price options available to the farmer. The pre-mium rate for the 2003 native pasture insuranceprogram was 21 percent (60 percent is subsidizedby the government).

Grassland Insurance in Spain

The parametric insurance scheme in Spain was engineered mainly to cover farmers from droughtsaffecting the pasture areas. The index utilized isalso the NDVI (estimated from NOAA images).The product has been offered since 2001 for all thefarms performing extensive livestock production,specifically cattle, sheep, horses, and goats, and isdesigned to cover the farmers experiencing morethan thirty dry days (defined as based on the aver-age historical information on pasture).

In contrast to the previous case study, the insur-able index is based only on pure imagery, that is, noverification with actual yields was performed. Theindex is therefore constructed using a historical evo-lution of the pixels to create a curve, and the indem-nity is defined when the actual observations in aparticular year are located below the average curve,based on eighteen years of data.

Also in contrast to the weekly NDVI values, thisscheme is based on a ten-day period NDVI index.A Maximum Value Composite Index (MVCI) is es-timated for each ten-day period to eliminate the ef-fect of clouds. The reference curves built from theMVCI are smoothed using different algorithms andare defined as beginning on the first ten-day periodof October and finalized on the last ten-day periodof September of the next calendar year. Wheneverinformation is not available for a particular period,a linear interpolation method is used to fill the miss-ing gaps.

The mask in this scheme is based on the CorineLand Cover (CLC-90), which is used to discrimi-nate between areas with and without grasslandproduction. The deductible is calculated from theten-day period and is defined as the historic aver-age MVCI for each area, minus 1.25 standard devi-ations from the average MVCI. The second item ofthe deductible is related to the amount of ten-dayperiods below the individual deductible for eachtime window. The time deductible is three periodsbelow the reference threshold for every ten-day pe-riod, which is equivalent to thirty days with dryvegetative indicators.

REFERENCESAdamenko, T. 2004. “Agroclimatic Conditions and Assessment of

Weather Risks for Growing Winter Wheat in Kherson Oblast.”The World Bank Commodity Risk Management Group



(CRMG) and International Finance Corporation PartnershipEnterprise Projects (IFC-PEP), unpublished report from theUkrainian Hydrometeorological Centre, Kiev, July.

Agriculture Financial Services Corporation (AFSC). 2005.“Canada-Alberta Insurance Programs for 2005 Annual Crops.”Promotional and informational brochure published by AFSC.www.afsc.ca.

Brown, D. M., and A. Bootsma. 1993. “Crop Heat Units for Cornand Other Warm Season Crops in Ontario.” Agriculture andRural Division, Ministry of Food and Agriculture FactsheetAgdex #: 111/31, Government of Ontario, Canada, October.http://www.gov.on.ca/OMAFRA/english/crops/facts/93-119.htm.

Divyakirti, V. 2004. “Saving for a Rainy Day.” EnvironmentalFinance, October.

Gadgil, S., P. R. Seshagiri Rao, and K. Narahari Rao. 2002. “Useof Climate Information for Farm-Level Decision Making:Rainfed Groundnut in Southern India.” Agricultural Systems74: 431–57.

Hess, U. 2003. “Innovative Financial Services for Rural India:Monsoon-Indexed Lending and Insurance for Smallholders.”Agriculture & Rural Development Working Paper 9, TheWorld Bank.

Hess, U., and J. R. Skees. 2003. “Evaluating India’s Crop FailurePolicy: Focus on the Indian Crop Insurance Program.” Paperdelivered to the South Asia Region of the World Bank,November.

Hess, U., J. R. Skees, H. Ibarra, J. Syroka, and R. Shynkarenko.2005. “Ukraine, Initial Feasibility Study of Developing WeatherIndex Insurance, Crop Disaster Assistance in Ukraine.” WorldBank Working Paper.

Hubka, A. Forthcoming. “BASIX Case Study.” Innovations inRural Finance, Commodity Risk Management Group, TheWorld Bank.

KBS LAB. 2004. “Weather (Rainfall) Insurance.” Visual pre-sentation prepared by corporate managers of KBS Bank, 25 January.

Narahari Rao, K., S. Gadgil, P. R. Seshagiri Rao, and K. Savithri.2000. “Tailoring Strategies to Rainfall Variability: The Choiceof the Sowing Window.” Current Science 78: 1216–30.

United Nations Food and Agriculture Organization (FAO). 2005.“Crop Water Management: Crop Water Information.” Onlineinformation, The Land and Water Development Division(AGL) Water Resources, Development and ManagementService (AGLW), Water Management and Irrigation SystemsGroup, acquisition date: May 2005. http://www.fao.org/ag/agl/aglw/cropwater/cwinform.stm.

111

EXECUTIVE SUMMARY1. The ex ante or ex post classification focuses on when the

reaction to risk takes place: prior to the occurrence of the potential harmful event (ex ante) or after the event has occurred (ex post).

CHAPTER 12. While the focus of this document is on natural disaster risks,

the World Bank is also heavily involved in assisting thetransfer of commodity price risk for certain commodities.CRMG will produce a separate document on lessons learnedin the price risk management area in 2006.

3. Given the combination of price risk and weather risk man-agement transfer, farmers with storage can reduce risk andimprove income by storing commodities and bargaining forhigher prices.

CHAPTER 24. For similar classifications, see Hardaker et al. 2004; and

Harwood et al. 1999.5. For other classifications, see Hazell 1992; World Bank

2001; Anderson 2001; Dercon 2002; Townsend 2005; Siegel2005.

6. This section is based on Townsend 2005.7. See Dercon 2002. See also World Bank 2001 for a discussion

of the role of safety nets in risk management in developingcountries.

8. Examples are the Tanzanian coffee and cotton hedging activities of a major cooperative and CRDB Bank Ltd., theleading private agricultural bank in the New York coffeeand cotton futures markets.

9. See the Skees, Barnett, and Hartell (2005) background paperfor more discussion of “cognitive failure” and “ambiguousloading.”

CHAPTER 310. For more detailed reviews of the U.S. program, see Glauber

2004; Skees 1999a; and Skees 2001.11. The remaining 2 percent of the premiums pays for a variety

of other insurance products.12. Under certain conditions, policyholders can choose to divide

farms into separately insured smaller units.13. The catastrophic policy only covers yield losses in excess

of 50 percent of the APH yield at a rate of indemnity only60 percent of the expected market price.

14. Information in this section is based on Pikor and Wile 2004.

CHAPTER 415. This section is based on the background paper by Skees et al.

2005. Appendix 2 offers additional technical details.16. This paper does not address the responses to the price risk

management needs of developing countries, as CRMG ispreparing a separate analysis and evaluation (possibly inan ESW) of its ongoing transaction support and capacity-building work in this area.

17. By contrast, area-yield indexes in developing countriesoften are not measured in a reliable and timely manner.

18. Basis risk also exists with traditional farm-level, multiple-peril crop yield insurance. Typically, a very small samplesize is used to develop estimates of the central tendency in farm-level yields (for example, four to ten years in theUnited States). Given simple statistics about the error ofsmall sample estimates, it can easily be demonstrated thatthese procedures sometimes generate large mistakes whenestimating expected farm-level yield. This makes it possiblefor farmers to receive insurance payments when yield losseshave not occurred and to fail to receive payments whenpayable losses have occurred. Thus, basis risk occurs notonly in index insurance but also in farm-level yield insur-ance. Another type of basis risk results from the estimate ofrealized yield. Even with careful farm-level loss adjustmentprocedures, it is impossible to avoid errors in estimating thetrue realized yield. These errors can also result in under- andoverpayments. Longer series of data are generally availablefor area-level yields or weather events than for farm-levelyields. Because of this, the square-root of n rule suggeststhere will be less measurement error for index insuranceproducts than for farm-yield insurance products when esti-mating the central tendency. If the standard deviation of therandom variable used for the index is lower than the stan-dard deviation of farm-level yields (as would be the case ifthe index is based on area-level yields), the index insurancewill have even less measurement error relative to a farm-level insurance product.

19. Temperature, for example, can be measured with fieldlodged temperature gauges that automatically transmitdata to a central server.

CHAPTER 520. Byerlee (2005) distinguishes between growth strategies for

irrigated high potential systems and areas with limited mar-ket access in marginal dry lands. Strategies for these twovery different types of agricultural systems put differentemphases on agricultural policy options of intensification,diversification, increasing farm size, enhancing off-farm ac-tivities, or encouraging exit from agricultural activities.

Notes

21. Dercon (2005) also cites the importance of macroeconomicstability and better functioning asset markets because theyincrease the usefulness of self-insurance. In addition, “Betteraccess to alternative economic activities and increased income-earning opportunities could strengthen income-based strategies. Public safety nets could be a useful alterna-tive, although initiatives to develop such programs shouldtake into account their effect on existing risk-coping strate-gies. Strengthening self-insurance through group-basedsavings, for example, is an alternative that remains insuffi-ciently explored” (161).

22. Little, et al. (2004), describe how disastrous droughts inEthiopia were the key external factor that “pushed vulnera-ble households into poverty out of which many had not re-covered by 2003,” three years after the major drought event.Moreover, “the occurrence of periodic droughts tends towipe out asset gains that poor households attain” (15–17).

23. These estimates are from Skees, et al. (2005). U.S. Summaryof Business data were used for the U.S. estimate, and datafrom Pikor and Wile (2004) were used for the Canadian estimate.

24. Timely payment of claims was one of the key reasons forthe success of the Indian weather insurance pilot programs.See Appendix 2 for the Indian experience with weather insurance.

25. CRMG, for example, conducts participatory sessions withfarmers to identify contract and delivery model designs.In Ethiopia, smallholders designate kebeles (local electedleaders of around six hundred farmers) to collect insur-ance premiums for group insurance. In one Malawian vil-lage, residents wanted local leaders to contract weatherinsurance that covered the smaller farmers under the pro-grams of the smallholder farmers’ association. In India,microfinance institutions function as trusted intermedi-aries for small farmers. In some places, cooperatives havegained the trust necessary to deliver insurance products tofarmers.

26. This probability distribution was developed using proce-dures that smooth historical data. In reality, few observationshave been made below the five hundred millimeter level.

27. To be clear, the threshold where cognitive behavior beginsis unknown. In this example, five hundred is used for il-lustration purposes only. If the value were known withcertainty, it would also be relatively easy to develop an an-alytical solution for the optimal subsidy level.

28. A more detailed discussion of index insurance is found inAppendix 1.

29. International donors could also reinsure this layer througha contingent credit.

30. This section draws on an idea formulated in Skees and Hess(2003) proposing a “standing disaster insurance program.”

31. See the Agroasemex case study in Appendix 2 for details onreinsuring an agricultural insurance portfolio with a weatherindex contract.

32. DOC contracts would most likely be reinsured using director packaged transfers of the underlying indexes. Poolingprior to transfer is likely to offer only minimal benefits, sincein-country spatial diversification opportunities are gener-ally limited for catastrophic layers.

33. See Appendix 1 for details on pricing methodologies.34. This section borrows from World Bank (2005a).35. The weather shock insurance safety net concept has been

launched by a Malawian government official, PatrickKabambe, and is more broadly based on the work on covari-ate shock insurance in Africa by the World Bank CRMG-SocialProtection unit (Harold Alderman and Will Wiseman) andthe CRMG-Southern Africa rural sector unit (Rick Scobey).

CHAPTER 636. More details on several of these case studies as well as ad-

ditional country examples are presented in Appendix 2.37. The unorganized sector corresponds to India’s informal or

submerged economy, small-scale nonregistered businesses,for example, particularly in rural areas.

38. This is a BASIX subsidiary and a Reserve Bank of India li-censed bank providing microcredit and savings services inthree districts.

39. BSFL is another BASIX subsidiary company. Launched in1998, BSFL is the “flagship” company of the group and is aReserve Bank of India registered nonbank financial com-pany engaged in microcredit and retailing insurance and theprovision of technical assistance.

40. A 2002 IFC survey of agricultural enterprise participantsin Ukraine reveals that the failure of farmers to repay creditwas often due to low sale prices, limited product demand,lack of market information, and high interest rates. Only 12 percent of respondents cited bad harvests as the reasonfor farmers’ inability to repay their debts. In the years beforethe survey was taken, farmers experienced marketing prob-lems for grains and good harvests. Crop failures due to frostand drought in the 2002 to 2003 season may have signifi-cantly altered farmers’ perceptions.

41. The estimate of $1.6 billion was determined by assumingWFP costs for the 1999–2000 drought, in which the WFP wasresponsible for 45 percent of the total food aid deliveries ap-pealed for by the DPPC. Using that cost estimate to deter-mine 100 percent of the cost of the drought in 1999–2000,then multiplying this cost by the magnitude of the 1984drought (assumed to be the worst case scenario), the totalcost today of a 1984 drought was estimated to be $1.6 billion.

42. See Hess and Syroka (2005) for more details on Malawi andthe SADC region.

43. Skees provided some of the background for this section; seealso Mahul and Skees (2005).

CHAPTER 744. See Appendix 1 for a four-step design of a risk management

plan at the microlevel.45. For information on this topic, see the World Bank Hazard

Risk Management Unit Web site.

APPENDIX 146. This appendix abridges a chapter in a forthcoming Istituto

di Servizi per il Mercato Agricolo Alimentare (ISMEA) pub-lication on innovations in agricultural risk management.

47. The last PWC Survey was published in June 2004; this fig-ure therefore includes transactions up to March 2004. A newPWC survey is expected in June 2005.

48. In the publication Energy Risk, survey respondents estimatedthat the market was worth around 45 percent more in 2004.The WRMA survey relies on figures from nineteen compa-nies, all members of the Washington-based organization.Some large weather trading operations, such as DeutscheBank and Calyon, are not WRMA members, however, mak-ing the true size of the market difficult to determine.

49. Most energy-related weather transactions are based on tem-perature indexes, such as Heating Degree Days (HDDs) andCooling Degree Days (CDDs), designed to correspond tofluctuations in demand for gas (heating) and power (cool-ing, that is, air conditioning).


Notes 113

50. In 1999, the Chicago Mercantile Exchange (CME) began list-ing and trading standard weather futures and options con-tracts on temperature indexes. They now list twenty-twolocations in the United States, Europe, and Japan.

51. National Cotton Council of America, http://www.cotton.org/.

52. Basis risk is a potential mismatch between insured party’sactual loss and the weather contract payment.

53. More information on the weather station and data require-ments and providers appears below.

54. To be precise, this definition describes a European Option,an option that can only be exercised at the end of its life, thatis, at maturity. In general, this is the most appropriate typeof option on an underlying weather index. Other types ofoptions include American Options, an option that can be ex-ercised at any time during its life; Bermudan Options, an op-tion that can be exercised on specific dates during its life;and Asian Options, an option with a payout function thatdepends on the average value of the underlying index dur-ing a specified period.

APPENDIX 255. Specific information on this is not available for public dis-

closure.56. The actual premium and payment rates are not available for

public disclosure and are omitted from this paper. Since thelack of heat units affects the end use of grain corn more thatit does silage corn, the table of premium and payment ratesdiffers for the two types of crop.

57. Besides working as a severity index, this mathematical rela-tionship is a percentage relationship, allowing the compari-son of figures from different years without concern for thescale of the measurement or inflation rates. It also helps elim-inate variations in the total sum insured on a yearly basis.

58. The weather information for the Mexican transaction wasreviewed directly by Risk Management Solutions (RMS;www.rms.com) which determined that no significant trends,particularly in the temperature data, occurred in the infor-mation used to construct the weather derivative structure.

Therefore, the following pricing exercise does not includeany “detrending” procedures such as those described inAppendix 1.

59. This information was provided by RMS, who worked withAgroasemex on the initial project.

60. The Sharpe Ratio method is presented in Appendix 1.61. The unorganized sector in India corresponds to the informal

or submerged economy, such as small-scale nonregisteredbusinesses, found particularly in the rural areas.

62. www.basixindia.com.63. BASIX Annual Report 2003–04.64. The BUA is a project of the Andhra Pradesh Government; it

subsidizes 85 percent of the cost of community bore wellsdug for irrigation of lands belonging to multiple villagehouseholds. The remaining 15 percent of the bore well costis met by the individual BUA members, in proportion to theland they irrigate.

65. BSFL is another BASIX subsidiary company. Launched in1998, BSFL is the “flagship” company of the group and is registered with the Reserve Bank of India as a nonbankfinancial company engaged in microcredit and retailing insurance and the provision of technical assistance. Source:www.basixindia.com.

66. This section is from Hess et al. 2005.67. As of 2003. The source of this information is the World

Development Indicators database, August 2004.68. Information on SHR is from Adamenko 2004.69. See Appendix 1 for details regarding the pricing of weather

insurance contracts.70. The information for this section is from AFSC 2005.71. The NOAA satellite system was used because historical

satellite images were readily available. To be effective, how-ever, any nonpastureland had to be excluded from the satel-lite images. With the square kilometer resolution of thesatellite image, pastureland outside the pilot area is situatedin smaller land parcels and within other crop and forestedland. Moving beyond the pilot area, with this resolution,would dictate the exclusion of many pixels that do not meetthe minimum pasture content criteria. Without a minimumnumber of pixel images, the sample size for a township pro-duction estimate is not credible.

Agricultura Production Risk lProd Innovations ... - World Bank

Documents

Transcript of Agricultura Production Risk lProd Innovations ... - World Bank