Statement before the U. S. Senate Agriculture, Nutrition and Forestry Committee

Keith H. Coble

Department of Agricultural Economics

Mississippi State University

March 10, 1999

Mr. Chairman and Members of the Committee, I appreciate the opportunity to address you today. My testimony will center on issues related to the provision of insurance and safety nets to agricultural producers in the United States. It is certainly appropriate to hold hearings such as these, given the substantial attention that these issues have received in recent months. My testimony will attempt to bring to bear what we have learned from our past experience with insurance programs, and to address some current risk management issues.

Agricultural risk policy is complex, as is the risk management decision environment for producers. Because of the production lags associated with crop and livestock production, producers are exposed to considerable risk from unexpected changes in input and output prices or weather events which affect yields. Agriculture also tends to be fairly capital intensive with considerable investment in items such as land and machinery. Producers are confronted with risks not only in making short-term production and marketing decisions, but also with longer term investments decisions such as purchasing land or equipment.

Any discussion of government risk policy must be framed in the context of our current federal farm policy. The attention that is now being given to crop insurance certainly is intensified by the fact that the 1996 Farm Bill eliminated the long-standing deficiency payment program. That program had for many years provided substantial payments to producers, provided target prices that were often well above market price levels, and enhanced farm incomes. Current production flexibility contract payments, while enhancing income, are not designed to counter downturns in agricultural prices or incomes.

Much of the attention being focused on crop insurance programs is so directed because these remain the primary government programs providing protection from agricultural sector risks. We also know that at the same time deficiency payments were eliminated there was a significant step taken with crop insurance programs by piloting the first of several revenue insurance designs. This has led many to suggest that we can protect against other agricultural risks besides crop yields through an insurance mechanism. And so, our discussions today largely center around the possibility that these insurance designs may be broadened to incorporate new risks or risks for other commodities.

The Origins of Current Crop Insurance Policy

It is also useful to place our current discussions in the context of where we have been with crop insurance programs in the past decade. Let me briefly summarize changes in these programs since the late 1980's. During this period, there was significant dissatisfaction with crop insurance because the program had two weaknesses. First, the 1980s were an era of significant actuarial unsoundness and the program had failed to achieve established actuarial targets. At the same time, even with subsidies, a relatively small percent of crop producers in the United States were willing to pay the price to purchase crop insurance. As a result, beginning in the late 1980's, and carrying forward into the early 1990's, we saw a succession of ad hoc disaster bills authorized, which were costly to the federal government, but perceived to be needed because crop insurance had not been widely adopted. The General Accounting Office documented widespread abuse of these disaster programs, and they were perceived to be inequitable because they were conditioned on sufficiently widespread losses to attract political notice and justify action by the Congress.

This was the context in which the 1994 Crop Insurance Reform Act was enacted. This Act had multiple objectives. The first was to create a situation where the provision of ad hoc disaster payments was not undermining insurance demand. Secondly, efforts were taken to improve the actuarial soundness of the crop insurance program. With these objectives in mind, catastrophic coverage crop insurance was created, which gave producers a roughly equivalent level of protection as that afforded by the disaster programs. Subsidies were increased on higher levels of coverage as well. For example, the subsidy on a 65 percent yield coverage policy was increased from 30 percent to 41.7 percent.

Since 1995, it appears that the crop insurance program has, in some regions and some commodities, achieved an actuarially sound and stable situation. And it has done so with reasonably high levels of participation. Crop insurance liability in 1998 was just less than $28 billion, which is slightly more than double the liability in 1994. Loss experience is more difficult to access because multiple years are needed to assess outcomes under various random events. Comparing buy-up coverage during the 1980s versus the 1990s, the aggregate loss ratio (indemnities divided by total premium) has fallen from 1.50 to 1.08. As I will discuss in a few moments, this is not the case in every crop and every region.

At the time of the 1994 reform, the changes brought about by the 1996 farm bill had not been envisioned. Thus, we are now looking at the crop insurance program from a very different perspective. That is why we are asking the question, "Can we provide a stronger safety net for agricultural producers which mitigates losses from events beyond the traditional crop insurance protection?" I think this underlies suggestions to protect multiple year losses, expand coverage to livestock producers, that we increase the subsidies of the program, and many of the other suggestions which are being brought out in this discussion.

Crop Insurance Subsidies and Participation

One common suggestion is that we increase the federal subsidy for the crop insurance programs. This would make insurance more attractive, it would encourage producers who are not currently participating in the program to do so, and it would encourage producers who currently participate at low levels of coverage to buy-up to a higher level of coverage and provide themselves greater risk protection. This is an area where past economic analysis can provide some insights. A number of studies have addressed this issue.

In the survey of crop insurance literature that Tom Knight and I wrote a few years ago, we summarized the findings regarding crop insurance participation (Knight and Coble). In a rare case of near unanimity among economists, the synthesis of existing literature strongly suggests that the demand for crop insurance is inelastic. What this implies is that as the price of crop insurance is reduced by 1%, there is less than a 1% increase in crop insurance participation. Some estimates indicate highly inelastic demand, approaching a -0.1. Some of the more recent and disaggregated analyses suggest an elasticity that probably does not go much above -0.6.

Using a more elastic point in that range, a -0.5, would suggest that reducing the price of crop insurance by 20 percent would probably increase participation in the neighborhood of 10 percent. While this is not a trivial increase, the response to subsidies is not terribly dramatic. I would suggest that there are a number of reasons for this. An obvious reason is that there are producers who are in a position where they essentially can and do insure themselves. They do this through a number of mechanisms including off-farm investments, borrowing, off-farm income and employment. Secondly, it is fairly obvious that where rates are perceived to be extremely high, then even with substantial subsidies, the producer will perceive crop insurance to be costly. Therefore, increasing participation will be best achieved by both designing a program that is actuarially sound and increasing subsidies. Finally, disaster assistance as passed last year and in every year from 1988-94 obviously has a substitution effect on crop insurance participation.

How one designs a subsidy structure, in large part, will depend on what objective is desired. Figure 1 shows the current subsidy structure as it varies across subsidy levels. The subsidy is 100% for catastrophic coverage and declines in percentage terms until the 65/100 (65% yield coverage and 100% of price election) coverage level where an additional subsidy commences. The 65/100 policy is 41.7% subsidized. The percentage subsidy for higher coverages declines as premiums increase, but subsidy levels are fixed in absolute value.

It is fairly obvious that there are tradeoffs in how a subsidy schedule is designed. If one desires to bring more producers into the program, then enhancing the fully subsidized catastrophic coverage would provide additional incentives to those who currently do not insure. Alternatively, increasing subsidies for higher coverages is more consistent with a desire to see producers already in the program obtain greater protection. However, it is unclear what percentage of catastrophic coverage participants are willing to pay for crop insurance unless they are fully subsidized. Even subsidies of 50-60% on buy-up coverages imply a large marginal cost to the catastrophic coverage participant considering moving to higher coverage.

Is Actuarial Soundness a Worthwhile Objective?

Now, let me turn to the issue of whether actuarial soundness is a worthy objective or an unnecessary constraint which limits the ability of insurance programs to assist farmers. Actuarial soundness has been an issue with crop insurance programs for many years. The Risk Management Agency has legislatively-mandated targets for actuarial soundness. One might assume that actuarial soundness targets are simply cost control. I am not denying that it can be used in that way, but our system of subsidies allows the mitigation of this effect. However, there is another significant aspect of actuarial soundness that needs to be recognized. I believe a strong case can be made that actuarial soundness is largely a matter of equity. That is, an actuarially unsound program will not be sustainable, because it will be widely recognized as arbitrarily unfair to good managers.

Let me explain why I say this. The best example that I can give comes from the situation that currently exists with crop insurance in the state where I now work, Mississippi. If you will look at the crop insurance program experience in that state, what you will see are two apparently contradictory facts. First, crop insurance programs in the Mid-south have not been actuarially sound. Figure 2 illustrates the particular situation with soybeans. Historically, indemnities paid out far exceed total premiums. At the same time, we see a program that is widely disliked and where buy-up participation is extremely low. Figure 3, shows the percentage of buy-up soybean crop insurance coverage in the Mid-south during 1998. You can see that in all these Southern states, soybean buy-up participation was well below U.S. averages.

How does one reconcile that insurance has been profitable for those who insure, but that most producers are unwilling to insure? My sense is that it is because the program is actuarially unfair. The outcome is that a few producers are able to gain a benefit from the program, while their neighbors are priced out of the insurance market. This occurs because the past loss experience of all producers in a county is used to set rates. It is my assumption that there are two dimensions of this problem. The first is that a program that is poorly underwritten and poorly designed will be subject to abuse by those inclined to do so. This creates a 'moral backlash' among nonparticipants. Further, a program that is not based on sufficiently strong actuarial data will produce inaccurate rates and unintentionally price people out of the market. This is simply a consequence of the difficult challenge of setting insurance rates on farm level yields, and the same applies to a revenue design. In this instance, good managers are receiving little or no benefit from the program. This is a challenge that exists with most of our current set of insurance designs.

Given these problems with the current designs, let me make a comment regarding possible new designs for insurance. It is very easy to suggest alternative coverages, alternative guarantees, or more generous benefits to producers. But it is very difficult to design superior insurance mechanisms that have generous benefits but do not exacerbate the problems that I have just described. Further, it is rather naive to think that increasing the subsidies will solve the problem. Increasing the subsidy to an actuarially unsound program will increase the benefits to all producers. However, high risk producers will receive a disproportionate share of any across-the-board subsidy increase. While the producers who are currently not being protected at all would receive some marginal benefit, again, their neighbors, who are receiving the majority of benefits, will have their benefits enhanced to a greater degree.

The interrelationship of Government Programs and Private Risk Markets

Another issue that I suggest should be considered as the debate regarding agricultural safety nets continues is that of what potential influence government programs may have on existing or potential private risk management markets. In the United States we have a number of privately provided risk transfer markets and mechanisms that agricultural producers are able to use. The evidence suggests that since the 1996 Farm Bill, producers have availed themselves to these markets to a greater degree (Harwood, et al). I am speaking of markets, such as commodity futures and options markets, and forward pricing contracts. Particularly, as one considers moving beyond providing yield risk protection, there is a very real potential to limit the demand for other risk management mechanisms, such as forward contracting and futures and production contracts. This suggests that there may be unintended consequences of providing a better safety net to agricultural producers. It is becoming clearer that these effects may be direct or indirect, but they can be recognized.

For example, the work that I have done in the past with Richard Heifner shows that insurance can have an effect on the demand for forward pricing and futures contracts. Further, this effect can be either positive or negative. Our results are illustrated in figure 4 where we examine the optimal quantity hedged given the purchase of four alternative insurance designs (Multiple Peril Yield Insurance -MPCI, Market Value Protection - MVP, Revenue Insurance - RI, and Crop Revenue Coverage - CRC). The figure shows the change in quantity hedged as increasing insurance coverage is purchased for representative farms in geographically diverse regions. Generally, if the insurance program is maintained at a sufficiently low level of coverage, the effect will be minimal. However, our findings generally suggest that as coverage goes up, then the influence becomes more significant. For example, we found that yield insurance tends to be complementary to forward pricing, while revenue insurance in some forms strongly substitutes for forward pricing mechanisms. However, I would note that the specific form of the insurance policy may make a substantial difference in the outcome.

Consideration of New Insurance Designs

Many suggestions are being brought forward as alternatives or additions to the currently available product list. Yet, not every risk is inherently insurable. That is, some risks are more readily insured than others. Over time, insurance experts have identified at least six ideal conditions for a risk to be considered insurable [Rejda, pp. 23, 24].

1. Determinable and measurable loss: It must be possible to clearly determine when a loss has occurred and the magnitude of the loss.

2. Large number of roughly homogeneous, independent exposure units: Insurance works by pooling large numbers of independent exposure units so that the statistical law of large numbers can provide an accurate prediction of expected future losses.

3. Accidental and unintentional loss: Losses should be paid only on "acts of nature" or other seemingly random occurrences.

4. No risk of catastrophic losses: If losses are positively correlated across exposure units (i.e., the risk is systematic or not independent) the statistical law of large numbers does not hold. A catastrophic event may cause huge losses for the insurer.

5. Calculable chance of loss: To develop a premium rate, the insurer must be able to accurately estimate both expected frequency and severity of loss.

6. Economically feasible premiums: Potential purchasers must consider the insurance premiums to be affordable.

In reality, most insurance products deviate, at least slightly, from these ideal conditions. Very few risks are totally independent as required by condition number two. Estimating the expected frequency and severity of loss, as required by condition number five, is harder for some lines of insurance (e.g., crop insurance or earthquake insurance) than for others (say, automobile or life insurance). And, contrary to condition number three, most lines of insurance are susceptible to at least some fraudulent, intentional losses. Yet insuring risks that stray too far from these ideal conditions will likely result in difficulties and unintended consequences.

Certainly new insurance designs are possible. However, all these products contain the same basic elements. To illustrate, consider the design of yield insurance. The structure of yield insurance indemnities may be written as in equation 1. An indemnity is paid only when the actual yield falls below the covered yield. When a loss occurs the producer receives the yield shortfall valued at the price guarantee.

Equation 2 is mathematically equivalent to equation 1, but more clearly shows the two main components of an insurance policy. In equation 2, the insurance protection (or liability from the perspective of the insurer) is the dollar amount covered by the insurance policy. When considering alternative insurance designs, the focus is often on the level of protection. But the triggering mechanism is the defining characteristic of an insurance policy. The trigger defines when an indemnifiable loss has occurred and the magnitude of the indemnity relative to the protection (Barnett and Coble).

It is relevant to consider how the triggering mechanisms differ for alternative insurance products. If one were to replace yield with revenue (price × yield) in the trigger component of equation 2, the result would be the trigger for revenue insurance products. Likewise, by substituting county yield for farm yield one creates the trigger for the Group Risk Plan (GRP).

Premium rates are conditioned on expectations of the frequency of loss and magnitude of loss. When developing insurance products critical questions are: 1) Is the nature of the loss event such that an objective observer can accurately identify whether the triggering criteria have been met? and, 2) Can a knowledgeable and objective observer estimate the true magnitude of loss? If the triggering mechanism is vague or unmeasurable, the first condition for insurability above is violated and there is little potential for a workable insurance product.

Our current insurance programs are largely based on an individual's yield. This is intended to protect producers against their own losses regardless of whether the cause was highly individualized or widespread. However, many of the problems with traditional crop insurance programs relate to the fact that it is very difficult to set appropriate rates because risk differs from farm to farm. This heterogeneity exists not only across regions, but may be observed even within small areas, such as a county, because of differences in soil type, production practices, slope of the soil, management practices. All can potentially influence the likelihood of loss.

Further, because the risk insured is partially under the control of the insured individual, insurance can provide incentives for producers not to use best management practices or intentionally defraud the insurer. I have investigated these incentives and they are complex, but they are real (Coble et al.). Producers can easily be placed in a situation where grain is more valuable going out the back of a combine rather than into the grain bin. This drives up the cost of the program and makes rating difficult. Further, I believe producers do not like being confronted with such choices. At a time when market prices are relatively low there is significant desire to insure prices at levels above those offered by the market. This is understandable, however, it is also one of the clearest ways to induce this kind of behavior. For example, consider a producer with a poor cotton crop and an insurance guarantee of $0.70 while market prices have fallen to $0.58. In this case every pound of cotton not produced is worth $0.12 more than a pound produced.

Let us be very clear these problems are a significant component of why our current program has a relatively high delivery cost. To correctly rate an individually-based program requires the paperwork of individual level yield histories from producers and large actuarial research staffs on the part of the insurer. To limit abuse, crop insurance programs require adjusters evaluating every loss that occurs on every unit. Thus, the transaction costs of this program are extremely high. As we consider new insurance designs that cover risk at other levels of aggregation or other commodities, these issues still loom and in fact may be exacerbated in many cases. We also know some straightforward ways to mitigate most of these problems. For example, utilizing "area" mechanism such as the GRP insurance design precludes much of overhead cost associated setting rates and preventing abuse.

Summary

Insurance can be a useful form of risk protection for agricultural producers. Today's attention is focused on crop insurance as a policy tool. However, insurance is particularly suited to some risks and less suited for others. As our current crop insurance programs are now under scrutiny there are opportunities to improve these programs. Crop insurance participation can be increased by modifications of the subsidy structure. However, the inelasticity of crop insurance demand suggests that significant increases in the subsidy will be required to significantly raise participation levels. Further participation increases may be obtained by coupling subsidy structure changes with improved actuarial work and underwriting which makes the program more equitable for those good producers who are priced out of the market.

I would argue that crop insurance is a poor vehicle for income enhancement because it has the potential to not only influence acreage decisions, but also create the perverse situation where crops are worth more lost than produced. This suggests that efforts should be directed toward developing workable, cost effective, and actuarially fair risk management tools. However, this will not come easy. Significant effort and creativity will be needed to develop and implement such changes. I am doubtful about there being a panacea waiting in the wings. Rather difficult analysis and sound decisions will be required to achieve programs which are workable and useful to the good producer desiring truly effective risk management tools.

Defining the proper public role is also difficult because of interactions between public and private institutions. Two areas come to mind. First, the current relationship between the USDA and the private crop insurance companies is highly complex and lacking in transparency. We are only beginning to have models which can analyze the implications of changing the Standard Reinsurance Agreement. However, the current agreement is complex in large part because of the multiple objectives that are incorporated into the agreement. It would seem that appropriate policy must balance incentives for creativity and competition with care in protecting the public good. There are opportunities for great private sector innovation. Allowing this innovation should require reconsideration of the appropriate Federal-private sector reinsurance relationship. Secondly, private risk management markets do exist and may be competitive with government provision of agricultural risk protection. The market failure argument for public intervention is particular weak in such instances. Ideally, government policy and private risk management tools could complement one another to enhance the risk management options available to producers.

Finally, I would conclude by noting that while our attention is currently directed toward insurance as a vehicle for risk management, it is not the only mechanism by which risk may be mitigated. For example, long-period price downturns and the risks of agricultural investment are probably much more adapted to mitigation through other mechanism. Ultimately, sustainable farm policy will need to accommodate the particular nature of these risks.

This concludes my testimony. I will be happy to answer questions.

References

Barnett, B.J. and K.H. Coble. Understanding Crop Insurance Principles: A Primer for Farm Leaders. Mississippi Agricultural and Forestry Experiment Station Bulletin.

Coble, K.H., R. Heifner. "The Effect of Crop or Revenue Insurance on Optimal Hedging." Selected Paper presented at NCR-134 Conference, Chicago, April 1998.

Coble, K.H., T.O. Knight, R.D. Pope, and J.R. Williams. "An Expected Indemnity Based Approach to the Measurement of Moral Hazard in Crop Insurance." American Journal of Agricultural Economics 79(February 1997):216-226.

Harwood, J.L., R. Heifner, K. H Coble, J. Perry, and A. Somwaru. "Managing Risk in Farming: Concepts, Research, and Analysis." USDA/ERS Agricultural Economic Report 774. March 1999.

Knight, T.O., and K.H. Coble. "Survey of U.S. Multiple Peril Crop Insurance Literature Since 1980." Review of Agr. Econ. 19(1997):128-156.

Rejda, George E. 1995. Principles of Risk Management and Insurance. New York: Harper Collins College Publishers.