Jerry R. Skees and Barry J. Barnett Experience to date indicates that it is extremely difficult, without massive government subsidies, to insure farm level crop yields from losses caused by any number of natural perils. Those who seek effective, agricultural risk management tools, offered with little or no government subsidy, need to understand the underlying problems with farm level, multiple peril crop insurance. This chapter begins by discussing those problems. The following section presents an alternative form of insurance that makes payments based not on measures of individual farm yields, but rather on either area yields or some weather event like temperature or rainfall. This alternative form of insurance is often referred to as “index” insurance, since payments are triggered by realizations of a pre specified index measure rather than by realized farm yields. Index insurance holds significant promise for a number of reasons. In some situations, index insurance offers superior risk protection when compared to traditional multiple peril crop insurance that pays indemnities based on individual farm yields. Second, index insurance provides an effective policy alternative for governments seeking to protect the agricultural production sector from widespread, positively correlated, crop yield losses (e.g., drought). Finally, when index insurance is used to shift the risk of widespread crop losses to financial and reinsurance markets, the residual idiosyncratic risk often has characteristics that make it easier for local insurance markets to absorb.

For full version of the paper with graphs and formulars please download Word file (170 kb)

Requirements for Multiple Risk Crop Insurance

Successful insurance programs require that the insurer have adequate information about the nature of the risks being insured. This has proven to be extremely difficult for farm‑level yield insurance. Farmers will always know more about their potential crop yields than any insurer. This asymmetric information is the major problem with insuring farm yields. If an insurer cannot properly classify risk, then it is impossible to provide sustainable insurance. Those who know they have been favourably classified will buy the insurance; those who have not been favourably classified will not buy. This phenomenon, known as “adverse selection,” initiates a cycle of losses (Goodwin and Smith; Ahsan, Ali, and Kurian; Skees and Reed; Quiggin, Karagiannis and Stanton). The insurer will typically respond with “across the board” premium rate increases. But this only exacerbates the problem, as only the most risky individuals will continue to purchase the insurance. The problem can only be corrected if the insurer can acquire better information to properly classify and assign premium rates to potential insureds.

Insurers must also be able to monitor policyholder behaviour. Moral hazard occurs when insured individuals change their behaviour in a way that increases the potential likelihood or magnitude of a loss. In crop yield insurance, moral hazard occurs when, as a result of having purchased insurance, farmers reduce fertilizer or pesticide use or simply become more lax in their management. At the extreme, moral hazard becomes fraud where policyholders actually attempt to create a loss. Again, the problem is asymmetric information. Unless the insurer can adequately monitor these changes in behaviour and penalize policyholders accordingly, the resulting increase in losses will cause premium rates to increase to the point where it becomes too expensive for all but those engaged in these practices.

Insurers must also be able to identify the cause of loss and assess the magnitude of loss without relying on information provided by the insured. For automobile or fire insurance the insurer can generally identify whether or not a covered loss event has occurred and the magnitude of any resulting loss. For multiple‑peril crop yield insurance this is not always the case. It is not always easy to tell whether a loss occurred due to some covered natural loss event or due to poor management. Nor is it easy to measure the magnitude of loss without relying on yield information provided by the farmer.

Another requirement for traditional insurance products is that the loss events be independent, or at least not highly positively correlated. This characteristic allows the “law of large numbers” to generate a narrow confidence interval around the expected loss for insurer’s portfolio of insurance products. If risks are highly positively correlated (what some refer to as systemic risk) the law of large numbers is not relevant and the solvency of the insurer can be threatened by extremely large losses due to a single event. For multiple‑peril crop insurance, losses due to perils such as drought, freeze, or excess moisture, are typically highly positively correlated across exposure units.

When considering these requirements, it is useful to compare multiple‑peril crop insurance with hail insurance. For well over 100 years, the private sector has sold crop hail insurance with no government involvement. Why has hail insurance succeeded without government involvement when multi‑peril crop insurance has not? There are at least four reasons: 1) farmers have no better information than the insurer regarding the likelihood of a hailstorm; 2) farmers cannot, by changing their behaviour, increase the likelihood of a hailstorm or the magnitude of damage from a hailstorm; 3) insurers can generally tell whether or not a loss was caused by hail and accurately estimate the damage without relying on information provided by the farmer; and, 4) hail risk is largely independent across exposure units.

Actuarial Performance of the Crop Insurance Programs

Performance of publicly supported multiple peril crop insurance has been poor when all costs are considered. If companies were private, the premiums collected would have to exceed the administrative cost and the indemnities paid out. Hazell quantifies the condition for sustainable insurance as follows:

                          (A + I )/ P < 1

         where       A = average administrative costs

                          I = average indemnities paid

                 P = average premiums paid

Given this ratio, Hazell finds that in every case the value exceeds 2 (Table 1). This means that the support from government is at least 50%. However, there are cases where farmers are clearly paying only pennies on a dollar of the real cost of the crop insurance program. A ratio of 4 means that the farmer pays 25 cents for every 1 dollar of total costs. Skees (2001) reports a ratio of 4 for the current US crop insurance program and Mishra reports that India’s I/P ratio increased to 6.1 for the period 1985‑94.

Table 1 has only one case where the loss ratio of indemnities over premiums approaches 1 –Japan. In this case, the administrative costs needed to achieve this lost ratio are quite unbelievable – over 4 and ½ times higher than the farmer premium. It seems a very high price to pay to obtain ‘actuarially sound’ crop insurance. The other strategy in reaching this goal is to make the premium subsidy high enough that there is no adverse selection – even the low risk farmers soon learn that crop insurance is a good buy. Once lower risk farmers are in the risk pool, the actuarial performance can improve if one is measuring the unsubsidized premium against the loss experience. Obviously this is a pure numbers game and reflects little about the true performance of the program. This is what the US has done in recent years (Skees 2001).

Table 1. Financial Performance of Crop Insurance Programs in Seven Countries

Country	Period	I/P	A/P	(A+I)/P
Brazil	75‑81	4.29	0.28	4.57
Costa Rica	70‑89	2.26	0.54	2.80
India	85‑89	5.11	n.a.	n.a.
Japan	47‑77	1.48	1.17	2.60
	85‑89	0.99	3.57	4.56
Mexico	80‑89	3.18	0.47	3.65
Philippines	81‑89	3.94	1.80	5.74
USA	80‑89	1.87	0.55	2.42

Source: Hazell.

With such poor performance one must ask if it is even possible to run an individual multiple peril crop insurance program that is self‑sustaining. Consider the information required to deliver and monitor this program. The insurer must know the following for every individual insured unit:

Insurance yield: Estimating the expected yield for an insurance unit is a daunting task. For the US federal crop insurance program, insurance yields are based on a simple average of the most recent 4‑10 years of realized yields on the insurance unit. Farmers can establish an initial insurance yield with as little as four years of yield records (there are significant penalties if farmers cannot provide at least four years of yield records). As the farmer builds toward 10 years of yield records, realized yield in a given year is incorporated into calculation of insurance yield in subsequent years. When the farmer has built 10 years of yield records, the insurance yield is calculated as a rolling average of the most recent 10 years of realized yields. This is a rather crude method for estimating the central tendency in yields. Due to sampling error, insurance yields can either underestimate or overestimate the true central tendency depending on the random weather events over the most recent 4‑10 years. The effect of sampling error is further compounded by the fact that for most multiple‑peril crop insurance programs, insurance yields are also the primary (if not the only) mechanism for relative yield risk classification. Thus, the mechanism for establishing insurance yields can lead to adverse selection where only those farmers who believe they are getting a fair or better offer will chose to participate. Farmers who think the insurance yield is too low will not participate. Also, since farmers provide the yield records on which insurance yields are based, there are opportunities for fraud.

Loss adjustment: It is complicated and expensive to measure realized yields so that payable losses can be determined. Most farmers do not like the idea of having someone come to their farm to estimate the realized yield. Nor is loss estimation a precise science. As implied by the word “estimate,” measurement errors are common. Additional investment in personnel and training is required to minimize measurement errors. When losses are widespread, a very large workforce of trained individuals is needed. In the US, farmers are often allowed to self‑report realized yields. Spot checks are conducted with penalties for filing false reports, yet there are opportunities for farmers to receive unwarranted payments.

Gross premium rate: For most insurance products, premium rate calculation is based on historical loss experience. However, calculating crop yield insurance premium rates is more complex. One would ideally like to know the yield distribution for each individual farm. That is, one would like to know all of the possible yield outcomes and the probability of occurrence for each of those outcomes. But as indicated above, most crop yield insurance programs have difficulty estimating even the central tendency in yields. Estimating factors that influence the higher moments of the yield distribution is much more problematic. Further, simply knowing the yield distribution for a well‑classified group of farmers may not be enough. Extra losses (beyond those represented by the yield distributions) can occur due to moral hazard.

The US government has made significant investments in attempting to address these and other informational challenges inherent in farm‑level crop yield insurance. While improvements have been made, the federal crop insurance program still suffers from problems related to inadequate or asymmetrically distributed information. Many of the more obvious and inexpensive improvements in information gathering and monitoring systems have already been made. Needed additional improvements will likely come at much higher marginal cost. That cost will be borne by taxpayers and/or policyholders. If the cost is passed on to policyholders, many will decide that the insurance is too expensive and opt out of the program.

The Index Insurance Alternative

Index insurance makes payments based not on shortfalls in farm yields, but rather on measures of an index that is assumed to proxy farm yields. We will consider two types of index insurance products: those that are based on area yields where the area is some unit of geographical aggregation larger than the farm, and those that are based on weather events.

Various area yield insurance products have been offered in Quebec, Sweden, India, and, since 1993, in the US (Miranda; Mishra; Skees, Black, and Barnett). Ontario, Canada currently offers an index insurance instrument based on rainfall. The Canadians are also experimenting with other index insurance plans. Alberta corn growers can use a temperature‑based index to insure against yield losses in corn. Alberta is also using an index based on satellite imagery to insure against pasture losses. Our discussion of index insurance focuses on the US Group Risk Plan (GRP) area yield insurance product.

The information needed to run an index insurance program is much less than what is needed for a farm yield insurance program. One needs sufficient data to establish the expected value of the index and a reliable and trusted system to establish realized values. This is critical. In India the area yield insurance program has had bad actuarial experience due to poor systems for estimating area yields (Mishra). If reliable estimates of area yields can be provided, there is no need for any farm‑level information. For example, area yield insurance indemnities are based on estimates of official measurements of realized area yields relative to expected area yields. Areas are typically defined along political boundaries (e.g., counties in the US) for which historical yield databases already exist.

The logic for using index insurance is relatively simple – there is no asymmetric information (Skees and Barnett). Farmers likely have no better information than the insurer regarding the likelihood of area yield shortfalls or unusual weather events, thus, there is no adverse selection. Farmers cannot, by changing their behaviour, increase the likelihood of an area yield shortfall (if areas are defined at large enough levels of aggregation) or an unusual weather event, thus there is no moral hazard. All of the information needed for loss adjustment is available from public sources. It is easy to tell whether or not a loss has occurred and accurately measure the indemnity, without having to rely on any information provided by the policyholder. All of these factors make it much less expensive for the insurer to provide index insurance than multiple‑peril crop insurance. Thus, the cost of index insurance can be significantly lower than the cost of multiple‑peril crop insurance. Also, since adverse selection and moral hazard are not problems, there is no need for deductibles.

Of course, one could easily adapt this contract design to any number of other indexes such as aggregate rainfall measured over a stated period at a specific weather station or the number of days with temperatures above or below a specified level. The contract design used in GRP is sometimes called a “proportional contract” because the loss is measured as a percentage of the trigger. Proportional contracts contain an interesting feature called a “disappearing deductible.” As the realized index approaches zero, the indemnity approaches 100 per cent of liability, regardless of the coverage chosen.

An alternative design has been proposed for rainfall index insurance (Martin, Barnett, and Coble).[1]

Here Limit is a parameter selected by the policyholder and bounded by 0 <  Limit < Index Trigger. The choice of Limit determines how fast the maximum indemnity is paid. By their selection of Limit, policyholders can attempt to better match indemnities with expected losses over the domain of potential realized values for the index. For example, suppose that losses would occur when realized aggregate rainfall is less than 100 mm measured over a given time period at a given weather station. Further suppose that realized rainfall less than or equal to 50 mm would cause a complete loss. The policyholder would select an Index Trigger of 100 mm and a Limit of 50 mm. If realized rainfall is less than or equal to 50 mm the Indemnity would be equal to the full Liability.

One can easily see that the GRP contract is simply a specific case of this more general contract design with Limit set equal to zero. At the other extreme, the closer Limit is set to Index Trigger, the more the contract resembles a “zero‑one” contract where Indemnity equals zero or the full Liability depending on whether or not the Realized Index < Index Trigger.

Interest in Index Insurance

In the US, GRP has been controversial for a variety of reasons. Obviously, is it quite different from traditional insurance, and this raises legitimate concerns from the insurance industry. Traditional insurers find it difficult to understand and accept an insurance product where indemnities are not based on farm‑level yield losses. Farmer interest has also been mixed. Not surprisingly, most GRP policies seem to be sold in areas where crop insurance sales agents are most familiar with GRP. In 2000, about 5.6 million acres were insured under GRP. That is relatively small percentage of the total insured acreage in the US. It is very difficult for GRP to compete with subsidized farm‑level insurance that is now being offered at coverage levels of up to 85 per cent.

The Ontario rainfall insurance product was fully subscribed in the first year (2000) that it was introduced. However, this is a limited pilot test of only 150 farmers and the product was introduced following a major drought. By 2001, 235 farmers had purchased about USD 5.5 million in liability with large payments of USD 1.9 million[2].

For many developing countries, rainfall index insurance merits consideration (Hazell; Skees, Hazell, and Miranda; Varangis, Skees, and Barnett). While basis risk may generally be lower with area yield index insurance, there are several reasons why rainfall index insurance may be preferable in a developing country context. First, while it is not common to find statistics on area crop yields in developing countries, many countries have government meteorological agencies that have collected data on rainfall over long periods of time. Second, it is less costly to set up a system to measure rainfall for specific locations than to develop a reliable yield estimation procedure for small geographical areas. Finally, there is a strong correlation between rainfall and crop performance. Drought and excess rainfall are both a major source of risk for crop losses in many regions. Drought causes low yields and excess rainfall can cause either low yields or serious losses of yield and quality during harvest (Martin, Barnett, Coble). For irrigated farms, a drought can also cause increased irrigation costs.

The World Bank Group is now studying the feasibility of rainfall index insurance in a number of countries. The International Finance Corporation (IFC) of the World Bank Group may take a financial interest in making rainfall insurance offers in developing countries. The IFC is interested in supporting these innovations so that developing countries can participate in emerging weather markets. A specially funded project was also awarded to a working group within the World Bank. This project has investigated the feasibility of developing weather‑based index insurance for four countries: Nicaragua, Morocco, Ethiopia and Tunisia. Since that project began, several of the professionals involved have begun similar investigations in Mexico and Argentina at the request of those governments. The governments of Turkey, Brazil, India and Mongolia have made similar requests. There is clearly a growing international interest in weather insurance.

Index Contracts for Disaster Relief

              As an alternative to a farm-level insurance program, developing country governments could institute an index based disaster program to offer protection against catastrophic events. An index based disaster program could provide reinsurance to financial institutions or could be offered to individuals. Having a mechanism in place to handle the risk of catastrophic losses would open the door for the development of risk management mechanisms in the private sector.

              Organizations involved in disaster relief may also have an interest in index contracts.  An index based relief fund would overcome existing inefficiencies associated with time delays and misallocation of aid by providing a mechanism for quick and equitable disbursement of aid. Access to monetary resources rather than material goods should also improve the flexibility and effectiveness of relief services.

              Alternatively, a relief organization could write index-based insurance contracts to micro-finance institutions or cooperatives to protect these groups from catastrophic losses resulting from natural disasters. Community groups often can provide formal or informal risk sharing among individuals for idiosyncratic risks, but may have insufficient resources to adequately manage losses when the entire group experiences a natural disaster.  The index contact would provide cash compensation to these groups when a disaster occurs. The recipients could then make their own decisions about how to use and distribute the money among the group or community as they can make the best assessment about their needs. There are economic advantages to circulating cash in the economy relative to the adverse effects associated with the dumping of food aid and household items.

Basis Risk

The phrase “basis risk” is most commonly heard in reference to commodity futures markets. In that context, “basis” is the difference between the futures market price for the commodity and the cash market price in a given location. Basis risk is variation over time in the relationship between the local cash price and the futures price. Consider a US example where farmers, in a specific locale choose to forward price their corn using the Chicago Board of Trade (CBOT) December futures contract. By selling December futures contracts, the farmers “lock in” a price at harvest that is conditional on an anticipated relationship between the futures market price and the local cash price. For instance, they may anticipate that when they harvest and sell their crop in November, the local cash price will be 20 cents per bushel lower than the November price on the December CBOT contract. If, however, local cash prices are much lower than expected relative to the CBOT, say, 35 cents per bushel below CBOT, the farmers do not get the price risk protection that they had hoped for. Their actual realized price, from the combined cash market and futures market activities, is 15 cents per bushel less than had been expected. Conversely, the local cash price may be much higher than expected relative to the CBOT price. For instance, the local cash price may be only 5 cents per bushel lower than the CBOT price. In this case, the farmers’ actual realized price, from the combined cash market and futures market activities, is 15 cents per bushel more than had been expected.

Basis risk is a common phenomenon in futures markets. While futures contracts can still be effective price risk management tools for farmers, the existence of basis risk implies that farmers will not always receive the anticipated price. Sometimes it will be higher. Sometimes it will be lower. Because of basis risk, forward pricing in the futures market does not eliminate all exposure to price risk.

Basis risk also occurs in insurance. It occurs when an insured has a loss and does not receive an insurance payment sufficient to cover the loss (minus any deductible). It also occurs when an insured has a loss and receives a payment that exceeds the amount of loss.

Since indemnities are triggered by area yield shortfalls or weather events, an index insurance policyholder can experience a yield loss and not receive an indemnity. The policyholder may also not experience a farm yield loss and yet, receive an indemnity. The effectiveness of index insurance as a risk management tool depends on how positively correlated farm yield losses are with the underlying area yield or weather index. In general, the more homogeneous the area, the lower the basis risk and the more effective area yield insurance will be as a farm yield risk management tool. Similarly, the more that a given weather index actually represents weather events on the farm, the more effective the index will be as farm yield risk management tool.

Recently, the academic literature on crop insurance has focused on basis risk that will naturally be part of any index insurance program. But there has been little discussion of the basis risk inherent in farm‑level insurance. To illustrate how basis risk is possible for farm‑level multiple‑peril insurance programs, one need only consider the major underwriting mechanism used in the US to establish the insurable yields. Recall that in the US, the insurance yield (a measure of central tendency) is based on a simple 4‑10 year average of historical yield data for the insurance unit.

While crop yields are probably not normally distributed, the implications of this statistical formula would still hold for most reasonable assumptions of crop yield distributions. Namely, the higher (lower) the standard deviation of the true distribution, the higher (lower) will be the error in using an average as an estimate of central tendency. The higher (lower) the sample size, the lower (higher) will be the error in using an average as an estimate of central tendency.

Consider the error in using an average to estimate the central tendency of crop yields with a sample size of only 4 to 10 years of farm yield data. For simplicity, we assume a corn farm where yield is normally distributed with a mean of 100 bushels per acre. We consider values for s of 25, 35, and 45 bushels per acre. Figure 1 presents the standard error of the estimate for different values of s and n. Clearly, the higher the variability in yield, measured by s, the higher the error in using a simple average as an estimate of central tendency. However, it is also striking how much higher the error is when using 4 years of data rather than 10 years.

If the standard deviation is 35 bushels per acre (which is a reasonable value for the US), using only 4 years of data to estimate the insurance yield will result in a standard error of 17 bushels per acre. Thus, while two thirds of the APH yields would be between 83 and 117 bushels per acre, there is a 33 per cent chance that the calculated insurance yield will be less than 83 or more than 117 bushels per acre. Now consider a situation where, because of the error in using a simple average as an estimate of central tendency, the insurance yield is calculated as 120 bushels per acre when the true central tendency is only 100 bushels per acre. If the farmer selects an 85 per cent coverage level (15 per cent deductible) the trigger yield will be 102 bushels per acre, which is higher than the expected yield. While the farmer has been charged a premium rate based on a coverage level of 85 per cent, in effect, the farmer has been given a coverage level over 100 per cent. Due to the estimation error, this farmer could receive an insurance payment when the realized yield is at, or even slightly above, the central tendency.

Alternatively, if the insurance yield is estimated at 80 bushels per acre, 85 per cent coverage will generate a trigger yield of 68 bushels per acre. While the farmer has been charged a premium rate based on a coverage level of 85 per cent, in effect, the farmer has been given a coverage level of only 68 per cent. If central tendency were estimated accurately, a yield loss in excess of 15 bushels per acre would trigger an insurance payment. Because of the estimation error, this farmer must have a yield loss in excess of 32 bushels per acre to receive an insurance payment.

Because of the error in estimating central tendency, it is possible for farmers to receive insurance payments when yield losses have not occurred. It is also possible for farmers to not receive payments when payable losses have occurred. Thus, basis risk occurs not only in index insurance but also in farm‑level yield insurance.

Another type of basis risk results from the estimate of realized yield. Even with careful farm‑level loss adjustment procedures, it is impossible to avoid errors in estimating the true realized yield. These errors can also result in under‑ and over‑payments. Between the two sources of error, measuring expected yields and measuring realized yields, farm‑level crop insurance programs also have significant basis risk.

Longer series of data are generally available for area yields or weather events than for farm yields. The standard deviation of area yields is also lower than that of farm yields. Since n is higher and s is lower, the square root of n rule suggests that there will be less measurement error for area yield insurance than for farm yield insurance in estimating both the central tendency and the realization. Long series of weather data are also available, but it is not necessarily true that the standard deviation of weather measures will be less than that of farm yields.

Managing Basis Risk in Index Insurance Contracts

Basis risk in index contracts can be dealt with through a variety of different design mechanisms. Of extreme importance is to recognize the role that well constructed index insurance contracts can play in removing much of the correlated yield risk. Once the ‘big risk’ is removed with effective use of index insurance contracts, any number of possibilities exists for removing the basis risk. Local groups and institutions should be able to pool independent risks and use index contracts to hedge against correlated losses from major events such as drought or earthquakes. Using index contracts to transfer the risk of correlated losses, private companies may find they can develop effective multiple peril crop insurance that pays for losses not covered by government facilitated index insurance.

Skees (2003) and Mahul (2002) press a bit harder in recognizing that index insurance contracts open the way for more effective pooling of independent risk via mutual insurance and/or rural finance entities. Weather index insurance could be sold through banks, farm cooperatives, input suppliers and micro‑finance organizations, as well as being sold directly to farmers. Banks and rural finance institutions could purchase such insurance to protect their portfolios against defaults caused by severe weather events. Rural finance entities aggregate and pool risk. With index insurance contracts one can take advantage of such entities to become the means of mitigating basis risk via loans to farmers who have a loss and do not receive a payment from the index insurance. Similarly, input suppliers could be the purchasers of such insurance. Once financial institutions are able to shift correlated risk out of local areas with index insurance contracts, they would be in a better position to expand credit to farmers, at perhaps improved terms.

Summary of the Relative Advantages and Disadvantages of Index Insurance

Index contracts offer numerous advantages over more traditional forms of farm‑level, multiple‑peril crop insurance. These advantages include:

1.         No moral hazard: Moral hazard arises with traditional insurance when insured parties can alter their behaviour so as to increase the potential likelihood or magnitude of a loss. This is not possible with index insurance because the indemnity does not depend on the individual producer’s realized yield.

2.         No adverse selection: Adverse selection is misclassification problem caused by asymmetric information. If the potential insured has better information than the insurer about the potential likelihood or magnitude of a loss, the potential insured can use that information to self‑select whether or not to purchase insurance. Those who are misclassified to their advantage will choose to purchase the insurance. Those who are misclassified to their disadvantage, will not. With index insurance products, insurers do not classify individual policyholders’ exposures to risk. Further, the index is based on widely available information. So there are no informational asymmetries to be exploited. It is true that some will find index insurance products more attractive than others. However, unlike individualized insurance products, such self‑selection will not affect the actuarial soundness of index insurance products.

3.         Low administrative costs: Unlike farm‑level, multiple‑peril, crop insurance policies, index insurance products do not require costly on‑farm inspections or claims adjustments. Nor is there a need to track individual farm yields or financial losses. Indemnities are paid solely on the realized value of the underlying index as measured by government agencies or other third parties.

4.         Standardized and transparent structure: Index insurance policies can be sold in various denominations as simple certificates with a structure that is uniform across underlying indexes. The terms of the contracts would therefore be relatively easy for purchasers to understand.

5.         Availability and negotiability: Since they are standardized and transparent, index insurance policies can easily be traded in secondary markets. Such markets would create liquidity and allow the policies to flow to where they are most highly valued. Individuals could buy or sell policies as the realization of the underlying index begins to unfold. Moreover, the contracts could be made available to a wide variety of parties, including farmers, agricultural lenders, traders, processors, input suppliers, shopkeepers, consumers, and agricultural workers.

6.         Reinsurance function: Index insurance can be used to transfer the risk of widespread, correlated, agricultural production losses. Thus, it can be used as a mechanism to reinsure insurance company portfolios of farm‑level insurance policies. Index insurance instruments allow farm‑level insurers to transfer their exposure to undiversifiable, correlated, loss risk while retaining the residual risk that is idiosyncratic and diversifiable (Black, Barnett, Hu).

There are also challenges that must be addressed if index insurance markets are to be successful.

1.         Basis Risk: It is possible for index insurance policyholders to experience a loss and yet not receive an indemnity. Likewise, they may receive an indemnity when they have not experienced a loss. The frequency of these occurrences depends on the extent to which the insured’s losses are positively correlated with the index. Without sufficient correlation, basis risk becomes too severe, and index insurance is not an effective risk management tool. Careful design of index insurance policy parameters (coverage period, trigger, measurement site, etc.) can help reduce basis risk.

2.         Security and dissemination of measurements: The viability of index insurance depends critically on the underlying index being objectively and accurately measured. The index measurements must then be made widely available in a timely manner. Whether provided by governments or other third party sources, index measurements must be widely disseminated and secure from tampering.

3.         Precise actuarial modelling: Insurers will not sell index insurance products unless they can understand the statistical properties of the underlying index. This requires both sufficient historical data on the index and actuarial models that use these data to predict the likelihood of various index measures.

4.         Education: Index insurance policies are typically much simpler than traditional farm‑level insurance policies. However, since the policies are significantly different than traditional insurance policies, some education is generally required to help potential users assess whether or not index insurance instruments can provide them with effective risk management. Insurers and/or government agencies can help by providing training strategies and materials not only for farmers but also for other potential users such as bankers and agribusinesses.

5.         Marketing: A marketing plan must be developed that addresses how, when, and where index insurance policies are to be sold. Also, the government and other involved institutions, must consider whether to allow secondary markets in index insurance instruments and, if so, how to facilitate and regulate those markets.

6.         Reinsurance: In most transition economies, insurance companies do not have the financial resources to offer index insurance without adequate and affordable reinsurance. Effective arrangements must therefore be forged between local insurers, international reinsurers, local governments, and possibly international development organizations.

Conclusion

Index insurance is a different approach to insuring crop yields. A precondition for such insurance to work is that many farmers in the same location must be subjected to the same risk. When this is the case, index insurance has the potential to offer affordable and effective insurance for a large number of farmers. Such insurance requires a different way of thinking. It is possible to offer such contracts to anyone at risk when there is an area wide crop failure. Furthermore, unlike traditional insurance, there is no reason to place the same limits on the amount of liability an individual purchases.

As more sophisticated systems to measure events that cause widespread problems are developed (such as satellite imagery) it is possible that indexing major events will be more straight forward and accepted by the international capital markets. Under these conditions, it may become quite possible to offer insurance in countries where traditional reinsurers and primary providers would have previously never considered. Insurance is about trust. If the system to index a major event is reliable and trustworthy, there are truly new opportunities in the world to offer a wide array of index insurance products.

BIBLIOGRAPHY

Ahsan S.M., A.A.G. Ali and  N.J. Kurian. 1982. Towards a theory of agricultural insurance. American Journal of Agricultural Economics 64 (3): 520-529.

Bantwal, V.J. and H.C. Kunreuther. 2000. A cat bond premium puzzle? The Journal of Psychology and Financial Markets 1(1):76‑91.

Black, J.R, B.J. Barnett, and Y. Hu. 1999. Cooperatives and capital markets: the case of Minnesota‑Dakota sugar cooperatives. American Journal of Agricultural Economics 81(5):1240‑46.

Doherty, N.A. 1997. Financial innovation in the management of catastrophic risk. Paper presented at the ASTIN/AFIR conference, Cairns, Queensland, August 10‑15.

Goodwin, B.K. and V.H. Smith. 1995. The Economics of Crop Insurance and Disaster Aid. The AEI Press, Washington, D.C.

Hazell, P.B.R. 1992. The appropriate role of agricultural insurance in developing countries. Journal of International Development 4(6):567‑81.

Lapan, H. and G. Moschini. 1994. Futures hedging under price, basis, and production risk. American Journal of Agricultural Economics 76(3):465‑77.

Lapan, H., G. Moschini, and S.D. Hanson. 1991. Production, hedging, and speculative decisions with options and futures markets. American Journal of Agricultural Economics 73(1):66‑74.

Mahul, O. 2002. Coping with Catastrophic Risk: The Role of (Non‑) Participating Contracts.” Working paper.

Markowitz, Harry M. 1952. Portfolio selection. Journal of Finance VII: 77‑91.

Martin, S.W., B.J. Barnett, and K.H. Coble. 2001. Developing and pricing precipitation insurance. Journal of Agricultural and Resource Economics 26(1):261‑74.

Meyers, R.J. and S.R. Thompson. 1989. Generalized optimal hedge ratio estimation. American Journal of Agricultural Economics 71(4):858‑868.

Miranda, M.J. and J.W. Glauber. 1997. Systemic risk, reinsurance, and the failure of crop insurance markets. American Journal of Agricultural Economics 79(1):206‑15.

Mishra, Pramod K. Agricultural Risk, Insurance and Income: A Study of the Impact and Design of India’s Comprehensive Crop Insurance Scheme. Brookfield: Avebury Press, 1996.

Quiggin, J., G. Karagiannis and J. Stanton. 1994. Crop insurance and crop production: an empirical study of moral hazard and adverse selection. In Economics of Agricultural Crop Insurance, edited by D.L. Hueth and W.H. Furtan, pp. 253‑72. Norwell MA: Kluwer Academic Publishers.

Skees, J.R. 1999. Opportunities for improved efficiency in risk sharing using capital markets. American Journal of Agricultural Economics 81(5):1228‑33.

Skees, J.R. 2000. “A Role for Capital Markets in Natural Disasters: A Piece of the Food Security Puzzle.” Food Policy 25: 365‑378.

Skees, J.R. 2001 “The Bad Harvest.” Regulation: The CATO Review of Business and Government. 24: 16‑21.

Skees, J.R. and B.J. Barnett. 1999. Conceptual and practical considerations for sharing catastrophic/systemic risks. Review of Agricultural Economics 21(2):424‑41.

Skees, J.R., J.R. Black, and B.J. Barnett. 1997. Designing and rating an area yield crop insurance contract. American Journal of Agricultural Economics 79(2):430‑38.

Skees, J., P. Hazell, and M. Miranda. 1999. New approaches to public/private crop yield insurance. Unpublished working paper, The World Bank, Washington, DC.

Skees, Jerry R. 2003. ‘Risk Management Challenges in Rural Financial Markets: Blending Risk Management Innovations with Rural Finance’. Thematic Paper Presented at Paving the Way Forward for Rural Finance: An International Conference on Best Practices. June 2 – 4, 2003 Washington, DC.

Skees, J.R. and M.R. Reed. 1986. Rate‑making and farm‑level crop insurance: implications for adverse selection. American Journal of Agricultural Economics 68(3):653‑59.

Varangis, P., J.R. Skees, and B. J. Barnett. 2002. Weather indexes for developing countries. In Climate Risk and the Weather Market: Financial Risk Management with Weather Hedges, R. S. Dischel, ed., London: Risk Books.

[1].                 The presentation here is for index insurance that would protect against losses due to insufficient rainfall. Martin, Barnett, and Coble, present an analogous index insurance that would protect against losses due to excessive rainfall.

[2].                 Personal email communication with Mr. Paul Cudmore of Agricorp, Canada, October 23, 2001.

Workshop on Rural Finance and Credit Infrastructure in China, 13‑14 October 2003, Paris, France

Skees is the H.B. Price professor of agricultural economics at the University of Kentucky. He is also President of GlobalAgRisk, Inc. Barnett is associate professor of agricultural and applied economics at the University of Georgia. All views, interpretations, recommendations, and conclusions expressed in this paper are those of the author(s) and not necessarily those of the supporting or collaborating institutions. A version of this paper will also be published in an Australian book on crop insurance (forthcoming fall, 2003). Correspondence should be addressed to Skees at jskees@globalagrisk.com