實驗測試幾近隨機優越下的Epsilons

Abstract

幾近隨機優越（Almost stochastic dominance, ASD）藉由限制效用函數，提供了切實可行的評比機率分配方法。這些受到限制的效用函數集合能夠包含幾乎所有的決策者，同時排除掉那些極端而在經濟意涵上沒有意義的效用函數。我們分別藉由運用第一階(first-degree)、第二階(second-degree)、第三階(third-degree)幾近隨機優越理論，進行實驗測量加諸在理性決策者、風險趨避者、謹慎(prudent)者效用函數上的限制範圍。在此同時，也能夠找出讓幾近隨機優越得以被運用的“可接受違反區域(allowed area violation)”。實驗經過精密的設計使得當在測量高階幾近隨機優越的可接受違反區域時，低階的幾近隨機優越不會造成影響。幾近二階隨機優越的框架效應也同時被檢視。此研究不僅詳細給定出實際可以拿來運用的幾近隨機優越評比方法，同時揭露所觀察的決策者偏好以描繪在某些集合中的效用函數。

Almost stochastic dominance (ASD) offers a feasible ranking rule by imposing restrictions on utility in certain classes. This restricted set of utility accommodates most decision makers and exclude some pathological and economically irrelevant utility. We conducted experiments to measure the range of the restrictions on utility for rational, risk-averse, and prudent decision makers by applying almost first-degree, second-degree, and third-degree stochastic dominance theorem respectively. In the meantime, the “allowed area violation” with which ASD ranking rule is valid for almost all decision makers is also revealed. The experiments were carefully designed so that the effect of lower-degree ASD didn’t exist when we examined the allowed area violation for higher-degree ASD. Framing effect for almost second-degree SD was also examined. This research not only specifies the empirical ASD ranking rule, but also reveals the observed preferences to depict utility in certain classes.