The definition of risk used by the HarvestChoice project does not frame risk as inherently bad. After all, farmers and other individuals often voluntarily subject themselves to risks that can be avoided. That is, some individuals may view risk as undesirable, others may not care, while still others may view it as desirable. Individuals who find risk undesirable are commonly referred to as risk averse. Individuals who do not care are commonly referred to as risk neutral. Individuals who find risk desirable are often referred to as risk preferring, loving, or attracted.
Whether risk is good or bad can depend on an individual’s personal preferences. It is also important to note that an individual’s risk attitudes can vary depending on the nature of the risk (e.g., a farmer may be risk averse when it comes to making production decisions, but risk preferring when it comes to playing games of chance).
Why do farmers voluntarily accept risk?
Many individuals voluntarily engage in risky activities when they could be avoided. This is particularly true of farmers. For example, many U.S. corn farmers do not irrigate their crop even though they face the risk of inadequate rainfall or drought. One possible explanation for this type of behavior is that some individuals like risk. However, this explanation is contrary to what most researchers find regarding individual risk preferences. A more satisfactory explanation recognizes that most risky decisions are characterized by an important risk and reward tradeoff.
In the case of irrigation, many farmers choose not to irrigate because irrigation is costly. Therefore, while irrigation may reduce the variability of yields and may even increase average yields, it can also decrease average net income by more than a farmer’s risk premium. Most decisions individuals and farmers make are characterized by analogous risk and reward tradeoffs.
How do farmers make risky choices?
A variety of paradigms have been proposed to explain how individuals make risky choices. The predominant theory used by economists is the expected utility hypothesis (EUH). Given a set of alternatives (e.g., which crop to plant or how much fertilizer to apply), EUH asserts that individuals will choose the alternative with the highest expected utility. The distinct components that make up expected utility are 1) the possible outcomes, 2) the value or utility of possible outcomes, and 3) the probability of possible outcomes. Specifically, expected utility is the sum over all possible outcomes of the value of an outcome, weighted by its probability (see Dillon 1971 for a more complete, yet still accessible treatment).
While EUH has been the predominant theory used by economists to explain the risky choices, it is not without critics. These critics have proposed a variety of alternative paradigms, of which only a few have been applied to evaluating the risk attitudes and risky choices of poor farmers in the developing world.
One set of alternatives to EUH that emphasize the importance of subsistence requirements are referred to as “safety first” models. For example, Roy's safety first model hypothesizes that an individual chooses to minimize the likelihood of income falling below some threshold or subsistence level (Roy 1952). Alternatively, Kataoka’s safety first model hypothesizes that individuals maximize income subject to some low probability of income falling below subsistence (Kataoka 1963). While these models have been around for more than 50 years and are particularly compelling in the context of poor farmers, their application has been limited.
Another alternative to EUH that emerged in the late 1970’s is prospect theory (Kahneman and Tversky 1979). Prospect theory and the more recent cumulative prospect theory (Tversky and Kahneman 1992) emerged as alternatives to EUH due to several regularities that were observed in psychological and economic experiments on risky choice that could not be explained by EUH. These regularities include (i) risk averse behavior over likely gains, (ii) risk preferring behavior over unlikely gains, (iii) risk averse behavior over unlikely losses, and (iv) risk preferring behavior over likely losses (Tversky and Kahneman 1992). To accommodate such behavior, prospect theory and the subsequent cumulative prospect theory modify EUH in several important ways. First, individuals are assumed to have a reference outcome; in the context of poor farmers in the developing world, this might be subsistence income. Outcomes are assumed to be framed in relation to this reference point such that outcomes above it reflected gains, while outcomes below it reflected losses. Second, individuals are assumed to have risk-averse attitudes toward gains, but risk-preferring attitudes toward losses. Third, individuals are assumed to exhibit loss aversion, meaning that they are more sensitive to losses than they are to gains.
Finally, individuals are assumed to overweight the likelihood of low and underweight the likelihood of high probability outcomes when summing the weighted value of all possible outcomes. While it is clear that safety first models will not supplant EUH as the predominant theory for interpreting the risky choices of farmers in the developing world, recent and successful applications of prospect and cumulative prospect theory suggest that the predominance of EUH may be diminishing.
Do farmers dislike risk?
The estimation of the risk attitudes of poor farmers has been of continuing interest in the development literature. Two different approaches have generally guided attempts to estimate risk attitudes. First, many researchers have had farmers participate in risky choice experiments. In these experiments, farmers are asked to make choices over a variety of binary lotteries.
Alternatively, researches have used detailed farm household survey data to infer farmers' risk attitudes based on their production decisions and specific assumptions regarding how farmers make risky choices. For example, expected utility theory predicts that more risk averse farmers will produce less of a marketable crop if the price of the crop is characterized by risk. Therefore, the theory predicts that differences in how much farmers choose to produce can be used to compare their risk attitudes.
Regardless of which approach is used, the vast majority of studies have found that farmers in the developing world are risk averse, though the degree of risk aversion varies and is often systematically related to observable individual circumstances or characteristics. Risk aversion is typically found to be negatively related to age, education, risk pooling, luck, household size and liquidity. It is typically found to be positively related to the size of the gamble, proportion of children in the household, age of the household head, and perceived vulnerability.
There is emerging evidence that this risk aversion is sensitive to how risk is framed, as hypothesized by prospect and cumulative prospect theory. This emerging evidence suggests that farmers may in fact be risk preferring when risk is framed in terms of losses rather than gains (e.g., by using fertilizer crop yield losses can be reduced by 10 percent versus by using fertilizer crop yields can be increased by 10 percent). It also suggests that these farmers are loss averse (i.e. are more sensitive to losses than gains), ambiguity averse (i.e. avoid risk when the probability of chance outcomes is unknown), and overweight the probability of unlikely outcomes, while underweighting the probability of likely outcomes.
How do a farmer's production decisions affect risk?
Farmers make a variety of decisions that influence the risks they face. For example, irrigation can reduce yield losses in the event of drought. While fertilizer can boost yields, it can also increase yield variability. Similarly, while improved hybrid crops can boost yields, it can also result in more variable yields when compared to locally adapted crop varieties.
The notion that some production choices can decrease yield variability, while others can increase variability was first formalized by Just and Pope (1978). Since the effect of production decisions on yield variability is theoretically indeterminate, a key contribution of this article was the Just-Pope production function. The Just-Pope production function provides a practical way for empirically determining how input choices like seed variety or fertilization affect the mean and variance of crop yields. The extensive use of the Just-Pope production function by agricultural economists is a testament to its utility. The insights of Just and Pope were further developed by Pope and Kramer (1979) resulting in the taxonomical classification of input choices as risk increasing, risk decreasing, or risk neutral.
The insights of both articles have been widely applied in the context of developing farmers' input choices and the effect of these choices on risk. The most widely explored input has been fertilizer, which has also provided the most consistent results. Increased use of fertilizer tends to increase yield variability, though whether fertilizer is risk-increasing or risk-decreasing is less clear because fertilizer also increases average yields and production costs. The effect of labor and crop variety selection on yield variability and risk has also been relatively widely explored, though unlike fertilizer, the results vary widely across crops and locations.
Dillon, J.L. 1971. An Expository Review of Bernoullian Decision Theory in Agriculture: Is Utility Futility? Review of Marketing and Agricultural Economics 39(1):3-80.
Just, R.E. and R.D. Pope. 1978. Stochastic Specification of Production Functions and Economic Implications. Journal of Econometrics 7:67–86.
Kahneman, D. and A. Tversky. 1979. Prospect Theory: An Analysis of Decision under Risk. Econometrica 47:263-91.
Kataoka, S. 1963. A Stochastic Programming Model. Econometrica 31:181-96.
Pope, R.D. and R.A. Kramer. 1979. Production Uncertainty and Factor Demands for the Competitive Firm. Southern Economic Journal 46(2):489-501.
Roy, A. D. 1952. Safety-first and the Holding of Assets. Econometrica 20(3):431–449.
Tversky, A. and D. Kahneman. 1992. Advances in Prospect-Theory - Cumulative Representation of Uncertainty. Journal of Risk and Uncertainty 5(4):297-323.