Using index-based risk transfer products to facilitate micro lending in Peru and Vietnam.

This article focuses on two technical assistance projects that use aggregate index-based risk transfer products (IBRTPs) to transfer significant natural disaster risks affecting agricultural production over wide regions. In Peru, the issue is El Nino events that cause flooding in the northern regions, and in Vietnam, early flooding in the Mekong Delta. These and other countries constantly face major natural disasters. When considering how the costs of natural disasters are internalized, it is important to examine institutional arrangements, particularly in the financial sector.

A common problem for banks, microfinance entities, and other rural lenders with geographically limited loan portfolios is the threat of correlated risks that hinder the development of rural financial services and in particular, the provision of credit. The presence of correlated risk poses a dual problem for lenders: (a) a disaster event implies the potential for much higher default rates among agricultural clients; and (b) additional liquidity problems as clients simultaneously draw down savings and increase demand for borrowing to cope with the disaster (Skees and Barnett 2006).

Countries respond to natural disaster risk in different ways, but the costs are always present, and regardless of the institutional and market arrangements, are always internalized in one form or other. In addition to presenting the details of IBRTP design, this article shows how two countries at nearly opposite ends of the institutional spectrum with respect to natural catastrophe response have internalized natural disaster risk into agricultural lending systems and how, in both cases, an insurance mechanism to transfer these risks could lead to improvements.

Financial institutions failing to cope with natural disaster risks represent a common story the world over, underscoring the importance of making the costs of natural disaster risk more explicit and of considering new ways to mix market mechanisms and government initiatives to transfer this risk. An approach that places greater emphasis on ex ante rules and procedures than on ex post responses imposes discipline, transparency, and efficiency to influence the broader financial markets of insurance, credit, and savings. IBRTPs that focus on removing the big risks of correlated events for financial institutions can clear the way for the development of more robust and resilient rural financial markets.

Barnett and Mahul (2007) explain the background, applicability, and relative advantages of index-based risk transfer approaches. Index-based risk transfer uses a proxy measurement to pay for significant economic loss. For example, if it is known that extreme rainfall is highly correlated with production losses, then these measures can be used as a proxy for the loss. This design mitigates problems associated with the usual public-sector response to catastrophic risk and to credit constraints in developed countries, namely traditional forms of agricultural insurance and ad hoc disaster aid.

The first section of this article follows the link between the benefits of risk transfer and rural lending on economic growth and then considers the impact of, and a possible solution to, correlated weather risk on credit market functioning. The experiences with catastrophic weather risk in selected locations of Peru and Vietnam are then reviewed, and the IBRTPs specifically customized to address the risk for lending activity in each country are compared. We conclude by offering lessons and extensions regarding the use of IBRTPs.

Complete Financial Markets Hasten Economic Growth

There is increasing evidence that access to advanced financial services by rural people is linked to accelerated economic development. When farmers have access to a variety of financial services, they are better able to make productivity-enhancing investments, to accept the risk of greater specialization and technology adoption, and to accumulate wealth in formal savings vehicles that can be reinvested in the local economy (Hardaker, Huirne, and Anderson 2004; Skees and Barnett 2006; US-AID 2006a). However, some risks are too large to be easily mitigated or self-insured at the farm/household level. When insurance services are available to transfer such non-diversifiable risks, farmers are able to avoid other more costly risk management strategies and have recovery resources available should the insured event occur without the need to liquidate productive assets for consumption smoothing (Dercon 2005; Zimmerman and Carter 2003). The presence of insurance services also facilitates the more accurate consideration and pricing of risk, which increases the efficiency of credit markets and resource allocation between different investment decisions (Trivelli et al. 2006; USAID 2006a). When both insurance and banking services develop together, access to credit at favorable rates and economic growth are generally higher. However, insurance is often missing in the market, particularly for catastrophic weather events. Correlated risks are considered uninsurable in the strict sense in part because traditional risk pooling is ineffective (Anderson 1976; Harwood et al. 1999).

In a market-based economy, the inability to effectively hedge against correlated weather risk generates less activity in the credit market among both lenders and borrowers. Lenders respond by curtailing the volume of credit available, offering credit at less favorable terms, and sometimes by withdrawing credit services entirely following a catastrophic weather event (USAID 2006a). This response can be pronounced when creditors lack traditional forms of collateral, as is common in developing countries. Demand for credit simultaneously falls as the cost of borrowing increases and from fear of the consequences of loan default (Varangis, Hess, and Bryla 2003; Trivelli et al. 2006). When insurance markets are lacking, credit markets fail to work effectively. This restricts economic growth and increases the vulnerability of the poor to poverty traps (Dercon 2005; Skees and Barnett 2006).

The development of IBTRPs can remove a portion of the correlated risk that hinders development of financial services by offering banks, microfinance entities and other credit providers with the means to transfer the cost of catastrophic weather shocks that induce loan default and liquidity problems (Miranda and Vedenov 2001; Varangis, Hess, and Bryla 2003; Skees and Barnett 2006). A risk transfer mechanism for correlated weather risks that carefully blends markets with some degree of government effort to facilitate market development can remove a major constraint to rural financial markets (Skees and Barnett 1999; Skees and Hartell 2006; Skees, Hartell, and Hao 2006).

Case Examples: Flood Risk Transfer in Peru and Vietnam

While the source of risk is different, natural disasters in Peru and Vietnam have similar characteristics and impacts--the correlated risk of flooding creates problems with default among borrowers. The IBRTPs that we propose to manage these risks are similar. Each is designed to deal with the major correlated risks as a first step to improving banking and insurance services.

However, Peru and Vietnam each have very different political and economic institutions that internalize these risks in different ways and generate distinct experiences. Table 1 summarizes the key characteristics of the natural disaster risk, institutional response, and proposed risk transfer instrument for Peru and Vietnam. In Peru, loan defaults following catastrophic flood risk are internalized by the market economy into interest rates and severe credit rationing. In the Socialist Republic of Vietnam, the response has been to share the risk by charging roughly the same interest rate to all borrowers from the State bank, while forgiving and rescheduling loan defaults that were created by natural disasters.

Flood insurance is unavailable in developing countries for many good reasons. In addition to difficulties of managing correlated risk, loss adjustment for small farms is prohibitively expensive, and moral hazard and adverse selection are pervasive. Each of these problems poses a significant barrier to developing insurance markets that offer products for individuals. To address these constraints, the approaches taken in Peru and Vietnam focus on transferring the correlated risk through natural aggregators first. Banks and MFIs pool the risk exposure of their clients, but are unable to effectively diversify this risk. By indemnifying lenders at the outset to remove the large correlated risks, constraints to further financial market development are loosened. Focusing on existing risk aggregators is likely more feasible than attempting to create an explicit structure to provide the benefits of risk transfer directly to individual farm households. Nonetheless, effective internal policies to deal with default risk remain a challenge for agricultural lenders.

Peru: Growing the Market

The overall objective of the project in Peru was to provide microfinance institutions (MFIs) access to innovations in weather index insurance. The initiative was to contribute to the expansion and sustainability of rural finance in Peru by providing new index insurance instruments that would reduce both the portfolio risk for the MFIs and the risk to individual farm loans. In both cases the focus was on reducing exposure to correlated risk.

After performing a risk assessment from a wide range of weather data in Peru, it became clear that El Nino events are the major source of catastrophic risk. The last two severe El Nino events (1983 and 1997) devastated a number of regions in Peru with above average rainfall and massive flooding. In the northern coastal department of Piura, a typically arid region, a strong El Nino can bring rainfall that is more than 40 times the average.

The risk from this type of event has affected the availability of agricultural credit. The volume of lending by MFIs in Peru grew 350% between 1998 and 2003 (de Janvry et al. 2003). However, during the same period, agricultural lending by the same MFIs had virtually no change in volume, and agricultural lending currently accounts for only about 10% of all MFI lending. At the extreme, one MFI has nearly stopped agricultural lending altogether--the predicted response from lending institutions when there is a correlated weather risk that is likely to increase default rates. Default rates for all MFIs operating in Piura prior to the 1998 El Nino were around 8% whereas after the event, default rates increased to about 18%. While not all of this increase can be attributed to El Nino, much of it was driven by the major disruptions created by El Nino flooding.

It was determined that the ENSO 1 + 2 index, which measures Pacific sea surface temperature immediately off the coast of Peru, was highly correlated with excess rainfall and flooding in Piura. Abnormal increases in sea surface temperature typically indicate an El Nino weather pattern. More detailed statistical work on these relationships can be found in Khalil et al. (2007).

Based on feedback from a major re-insurer, it became clear that writing a direct insurance contract on the ENSO 1 + 2 was both relatively easy to do and would provide for significant risk transfer for catastrophic events. For index insurance contracts, re-insurers are most receptive to a secure measure that has a long time series. ENSO indexes are measured, maintained, and published by the U.S. National Oceanographic and Atmospheric Administration. Daily records that use very similar technology exist for about 50 years. One basic contract would simply make linear payments when the ENSO 1 + 2 index during the January-March flooding period exceeds 2, with a maximum payout when the index reaches 3. In years of recent El Nino events, 1982-83 and 1997-98, the ENSO 1 + 2 index exceeded 2.5. Using an example, an MFI with a $100 million portfolio and an estimate that a severe El Nino could cause an increase in default rates of 10 percentage points, could purchase an ENSO 1 + 2 insurance contract with a maximum payout of $10 million. If the index value were 2.6, the payment rate would be 60% of the value insured, or $6 million. The Peruvian regulator approved ENSO insurance in the summer of 2006, based on the potential of this instrument to enable the transfer of a major catastrophic risk that has plagued Peru for centuries.

Vietnam: Leading the Market

In contrast to the experience in Peru, Vietnam faces a very different institutional environment but also one that is in flux. The Socialist Republic is currently transitioning many of its state-owned banking and financial services toward greater market orientation in a process referred to as "equitization," which includes the modernization of insurance regulation and other changes to improve conditions for general insurability and risk transfer.

There has been little success in offering agricultural insurance in Vietnam. For example, the state-owned insurance company BaoViet suffered actuarial losses in excess of Vietnamese Dong (VND) 5 billon (VND 16,000 = US$1) on insurance based on rice-yield loss offered over the 1993-97 time period, resulting in an aggregate loss ratio of 110%. A subsequent attempt in 1997-98 to offer traditional multiple peril crop insurance, principally for rice, was also terminated due to massive losses (Dufhues, Lemke, and Fischer 2005). At present, there is essentially no agricultural insurance activity in Vietnam.

The state agricultural bank, the Vietnam Bank for Agriculture and Rural Development (VBARD), has become the de facto risk aggregator and agricultural insurer through its lending practices. Characteristics of VBARD lending include: (a) nearly flat interest rates that are charged throughout the country, thus VBARD pools risk nationally; and (b) in the event of a natural disaster that impacts loan repayment ability, it performs a loss assessment to determine if loans should be rescheduled or forgiven. In the past the government periodically recapitalized the bank for loan forgiveness, but this practice has recently been discontinued. Loan rescheduling and some amount of commune-level loss adjustment continue to occur if the borrower qualifies. And like an insurer, VBARD now maintains local reserves to protect against losses created when debt is postponed. These recent regulatory changes with the movement towards commercialization are serving as an important catalyst for VBARD and other lenders to consider innovative options to protect their lending portfolio from natural disaster risk.

The work in Vietnam has focused on early flooding that impacts the second rice crop prior to harvest. While the Mekong Delta floods every year, Vietnam has made massive investments in infrastructure to manage the normal extent of the flood in its early stages to extend the growing season. River flow normally starts to increase in June, but significant river flows usually do not begin until mid- to late July, with the annual flood levels peaking in October. The problem occurs when flows increase significantly during the later part of June and early July, when farmers still have their summer-autumn rice crop in the fields. In the project area, approximately 200,000 hectares of summer-autumn rice valued at over US$150 million are vulnerable to this risk.

While flooding in the Mekong Delta is a function of multiple conditions, hydrological modeling performed by the Southern Institute of Water Resources Planning (2007) reveals that water coming across the Cambodian border is the dominant factor influencing flooding. Daily water level data obtained from Tan Chau hydrological station from 1977-2004 were used to examine levels exceeding 250 cm for dates between June 20 and July 10. At 250 cm, downstream flooding becomes a serious problem, while the June 20 to July 10 period coincides with greatest harvest activity. When this measure is used, 4 out of 27 years had excess water levels (greater than one-half meter) during this period. This threshold represents roughly a 1-in-7-year chance, which should be an acceptable level of frequency for most insurance providers.

Using this information, an aggregate IBRTP can be constructed that indexes the level of water at the Tan Chau station and provides indemnity payments directly to rural lenders. This targets the economic costs of flooding to creditors and is intended to cover business costs associated with default risk and restructuring the loan portfolio. Furthermore, a certain percentage of loans will never be repaid even after restructuring. The significant question is how these credit risks should be evaluated and to what extent the bank would want to establish reserves for these costs versus the purchase of an insurance product. The optimal strategy is likely a blend of these two mechanisms.

The most straightforward IBRTP for excess water levels during this time frame would be a linear payout. The payment rate for a contract that begins paying indemnities for water levels above 250 cm with maximum payout at 350 cm, would be 1% of liability for every 1 cm of water above 250. Any risk aggregator who purchased the contract would evaluate the amount of liability they require relative to the risk exposure of their portfolio. The liability selected would drive both the premium and the indemnity payments. For example, if the liability was VND 4 billion and the premium rate was 10%, the purchaser of the index insurance contract would pay VND 400 million (0.10 x 4 billion). If the water level reached 275 cm during the critical period, the indemnity payment would be 25% or VND 1 billion.

Conclusions and Extensions Emerging from Peru and Vietnam

The country case studies of Peru and Vietnam provide two very different institutional and market maturity examples illustrating how it may be possible to construct aggregate 1BRTPs that blend banking and insurance to remove a lender's correlated portfolio risk. However, such aggregate risk transfer is only a beginning step in the process of developing efficient credit markets in developing countries.

One issue involves the challenge of providing a direct linkage between the aggregate IBRTP and financing and delivering some level of efficient insurance product to protect both smallholder borrowers and the lender from defaults. While the initial focus is on removing the big constraints for the lender, there is some question as to whether this alone will provide the desired level of "trickle down" benefit to drive an active credit market (Trivelli et al. 2006). Allowing lenders to pass on the benefits of the large indemnity payment from the IBRTP may be the most efficient way to deal with the large transaction costs associated with providing more complete financial services to small households. This does not mean that the lender would be underwriting the risks, but rather that the lender would link the benefits of the aggregate index payments to the small loans. Given special considerations from regulators about micro-insurance, this arrangement could evolve into more sophisticated products that would allow payments based on risk zones for group lending and group indemnity.

One promising area of research is the use of more advanced technology to make estimates of local losses, which would reduce the transaction costs of extending insurance products at a disaggregate level, while also controlling for moral hazard and adverse selection problems. For example, there is ongoing work in risk-zone modeling to develop accurate and reliable maps for the specific timing of catastrophic flooding events (Southern Institute for Water Resources Planning 2007). Technological innovations in satellite imagery may also offer potential in providing rapid and reliable methods for evaluating water inundation on small parcels. The institutional innovation associated with index-based insurance products can be a motivation for investing in technological advances that improve estimation of losses at the local level.

Finally, mitigation remains a critical component when designing any risk management strategy within a country. It is important to remember, however, that mitigation also comes at a real economic cost, and that overconfidence with engineering solutions can obscure serious residual risks. For example, Vietnam has invested extensively in dyke and canal systems to control flooding. Yet, experience and initial hydrological modeling both indicate that these systems can be overwhelmed. In Peru, there have been discussions about the potential for the construction of flood control structures, but these are estimated to be enormously expensive. The economic trade-off between engineered mitigation solutions relative to financial solutions for extreme catastrophic risk remains an important area of research.

References

Anderson, D.R. 1976. "All Risks Rating within a Catastrophe Insurance System." Journal of Risk and Insurance 43:629-51.

ADB (Asian Development Bank). 2007. Developing Agricultural Insurance in Vietnam. Phase I Final Report TA 4480 VIE prepared for the ADB by World Perspectives, Inc. and Global-AgRisk, Inc., March.

Barnett, B.J., and O. Mahul. 2007. "Weather Index Insurance for Agriculture and Rural Areas in Lower-Income Countries." American Journal of Agricultural Economics 89(5): 1241-47.

de Janvry, A., E. Sadoulet, C. McIntosh, B. Wydick, J. Luoto, G. Gordillo, G. Schuetz, M. Valdivia, J. Bauchet, C.E. Herrera, and R. Kormos. 2003. "Credit Bureaus and the Rural Microfinance Sector: Peru, Guatemala, and Bolivia: Assessing the Impact of Credit Referencing on the Poor." Publication of the BASIS Collaborative Research Support Program (CRSP), University of Wisconsin, Madison, December.

Dercon, S., ed. 2005. "Risk, Insurance, and Poverty: A Review." In Insurance Against Poverty, pp. 9-37. Oxford: Oxford University Press.

Dufhues, T., U. Lemke, and I. Fischer. 2005. "Constraints and Potential of Livestock Insurance Schemes: A Case Study from Vietnam." Discussion paper No. 05/2004, Department of Agricultural Development Theory and Policy, and Institute of Animal Production in the Tropics and Subtropics, Animal Husbandry and Breeding Section, University of Hohenheim, July.

Hardaker, J.B., R.B.M. Huirne, and J.R. Anderson. 2004. Coping with Risk in Agriculture, 2nd ed. New York: CABI Publishing.

Harwood, J., R. Heifner, K. Coble, J. Perry, and A. Somwaru. 1999. "Managing Risk in Farming: Concepts, Research, and Analysis." Agricultural Economics Report AER774, Market and Trade Economic Division and Resource Economics Division, Economic Research Service (ERS), U.S. Department of Agriculture, March.

Khalil, A.E, H.H. Kwon, U. Lall, M.J. Miranda, and J.R. Skees." EI Nino-Southern Oscillation-Based Index Insurance for Floods: Statistical Risk Analyses and Application to Peru." Water Resource Research 43(10), Manuscript 2006WR005281, 49 pages, October 17, 2007.

Miranda, M.J., and D. Vedenov. 2001. "Innovations in Agricultural and Natural Insurance." American Journal of Agricultural Economics 83:650-55.

Skees, J.R., and B.J. Barnett. 1999. "Conceptual and Practical Considerations for Sharing Catastrophic/Systemic Risks." Review of Agricultural Economics 21:424-41.

--.2006. "Enhancing Micro Finance Using Index-Based Risk Transfer Products." Agricultural Finance Review 66:235-50.

Skees, J.R., and J. Hartell. 2006. "Innovations in Risk Transfer Markets in Agriculture for Natural Hazards." Risk Management in Agriculture for Natural Hazards. Rome: ISMEA.

Skees, J.R., J. Hartell, and J. Hao. 2006. "Weather and Index-Based Insurance for Developing Countries: Experience and Possibilities." In A. Sarris and D. Hallam, eds. Agricultural Commodity Markets and Trade: New Approaches to Analyzing Market Structure and Instability, pp. 256-81. Northampton MA: Edward Elgar Publishing, Ltd.

Southern Institute of Water Resources Planning. 2007. "Hydrological Risk Zone Mapping in Dong Thap Province." Technical report TAR 38561, Prepared for Asian Development Bank Project ADB-TA 4480 VIE, Ho Chi Minh City, Vietnam.

Trivelli, C., M.R. Carter, E Galarza, A. Tarazona, and J. Yancari. 2006. "Can Insurance Unlock Agricultural Credit and Promote Economic Growth?" BASIS Brief No. 46, Basis Collaborative Research Support Program (CRSP), University of Wisconsin, Madison, May.

USAID. 2006a. Assessment on How Strengthening the Insurance Industry in Developing Countries Contributes to Economic Growth. Final report prepared by Chemonics International, Inc., and the International Insurance Foundation, Contract No.: EEM-E-00-05-00006-00, Task Order No. 01, February.

--. 2006b. "Hedging Weather Risk for Micro-finance Institutions in Peru." Comprehensive report prepared by GlobalAgRisk, Inc., USAID/DAI Prime Contract LAG-I-00-98-002600 BASIS Task Order 8, Rural Finance Market Development, November.

Varangis, P., U. Hess, and E. Bryla. 2003. "Innovative Approaches for Managing Agricultural Risks." In N. Scott, ed. Agribusiness and Commodity Risk: Strategies and Management, pp. 83-102. London: Risk Books, Incisive RWG, Ltd.

Zimmerman, EJ., and M.R. Carter. 2003. "Asset Smoothing, Consumption Smoothing, and the Reproduction in Inequality under Risk and Subsistence Constraints." Journal of Development Economics 71:233-60.

Jerry R. Skees is the H.B. Price Professor of Risk and Policy in the Department of Agricultural Economics at the University of Kentucky, and president of GlobalAgRisk, Inc. Jason Hartell is Ph.D. Candidate in the Department of Agricultural Economics at the University of Kentucky, and Anne G. Murphy is Vice President of GlobalAgRisk, Inc.

GlobalAgRisk, Inc. has been leading work in Peru under US-AID/DAI Prime Contract LAG-I-00-98-0026-00 BASIS Task Order 8, Rural Finance Market Development; and in Vietnam under contract with World Perspectives, Inc. for ADB TA-4480. The authors gratefully acknowledge the editorial assistance of Celeste Sullivan. This article is published under the University of Kentucky Agricultural Experiment Station Number 07-04-082.

This article was presented in a principal paper session at the AAEA annual meeting (Portland, OR, July 2007). The articles in these sessions are not subjected to the journal's standard refereeing process. Table 1. Key Characteristics of the Proposed Index-Based Risk Transfer Products in Peru and Vietnam

Peru Vietnam Natural disaster Severe flooding Extreme and early event during El Nino arrival of annual

event flood Event impact on Crop destruction, Interrupts harvest of

agriculture erosion summer-autumn rice

crop Event onset Excess rainfall Excess river flow

measurement Index for correlated Sea surface tempe- River depth at Tan

risk transfer rature, ENSO 1 + 2 Chau Station Index measurement January-March June 20 to July 10

period Producer response to Loan default Loan default

disaster Institutional res- Credit market Debt forgiveness; now

ponse (creditors) withdrawal debt restructuring Institutional moti- Indemnify lending Prepare for

vation for catas- portfolio to enable commercialization

trophic risk a return to rural by protecting

transfer market lending portfolio Impact of most 1998-Massive 2000--Extreme early

recent event flooding destroys flooding ~50% area

crops, infra- affected, harvest

structure, trade. losses, destroys

Ag lending infrastructure,

portfolio falls 10% interrupts trade,

loan defaults Source: USAID 2006b; ADB 2007.