Target-Adjusted Utility Functions and Expected-Utility Paradoxes
提出目标调整效用模型，用线性概率偏好解释阿莱悖论等期望效用理论的反常现象，并利用实验数据验证其解释力优于累积前景理论。
Experimental tests of expected-utility theory (EU) have accumulated empirical observations in which the predictions of EU are systematically violated. The cumulative prospect theory (CPT) explains violations such as the Allais paradoxes and fourfold pattern of risk attitudes as resulting from nonlinear probability transformations. Here we show that the classical paradoxes for decisions under risk can be explained with preferences that are linear in probabilities for any choice set and that maximize an expected-utility function with respect to an endogenous target return. We introduce the maximin payoff as a plausible and even natural target return from a choice set and show that the resulting target-adjusted utility (TAU) model explains additional empirical observations such as the scale dependence of the Allais paradox that cannot be explained by standard specifications of CPT. Further, using data from three prominent laboratory experiments, we find that TAU is effective in explaining observed behaviors. This paper was accepted by James Smith, decision analysis.