An investigation of the measurement of individual risk attitudes

Graduation date: 1986Two direct elicitation of utility (D.E.U.) techniques\ud were used to estimate risk attitudes of a group of agricultural\ud producers. The two elicitation techniques used in the\ud study were 1) an error-in-response model using a modified\ud Ramsey method, and 2) stochastic dominance with respect to a\ud function (SDF). The primary objective of the study was to\ud determine whether the two elicitation techniques yield consistent\ud estimates of risk attitudes. A second major objective\ud of the study was to provide additional information\ud about the distribution of risk attitudes among agricultural\ud producers.\ud The study confirmed the results of other research\ud efforts that the majority of risk attitude parameters of\ud agricultural producers lie within the range -.0001 and .001\ud with income measured in dollars [King and Robison, 1980]. The\ud study also supports previous research results which indicate\ud that a significant portion of decision makers exhibit risk\ud preferring behavior, at least over some ranges of incomes.\ud The error-in-response model classified 38.1% of the respondents\ud as risk preferring, 47.6% as risk neutral, and 14.3% as\ud risk averse. With only one exception, the SDF technique\ud elicited risk preferring attitudes for every respondent over\ud some range of income values.\ud Individual and aggregate tests for decreasing\ud (increasing) absolute risk aversion were conducted. No\ud respondents were found to exhibit increasing or decreasing\ud absolute risk aversion. The statistical comparison of the\ud two elicitation techniques was inconclusive. A paired t-test\ud failed to reject the null hypothesis of no difference in the\ud estimated risk attitudes. However, the correlation between\ud the two measures was virtually zero (-.046) suggesting that\ud the two measures of risk attitudes are not closely related.\ud The two elicitation techniques were also compared on\ud other grounds. Both elicitation techniques are designed to\ud prevent certainty bias that has plagued other D.E.U.\ud methods. The SDF technique is found to be superior in overcoming\ud possible interviewer bias. Neither technique is\ud superior in coping with probability bias.\ud The SDF technique is easier to implement but the error-in-\ud response questionnaire is easier to formulate. The error-in-\ud response model results in a specific estimate of the\ud respondent's risk attitude when the negative exponential\ud utility function is used. Based on the comparisons made in\ud the study, the SDF procedure is considered to be superior to\ud the error-in-response model for eliciting risk attitudes