風險與報酬的關係一直是個爭議不斷的議題。1979年Kahneman與Tversky提出展望理論(prospect theory),認為個人是基於參考點的位置不同,而會有不同的風險態度。此後許多學者利用展望理論來對風險與報酬的關係進行實證。然而,這些文獻在實證的設計上,分組的標準與樣本的觀察期間是採同一段期間,這一點並沒有符合展望理論的精神。本文則修正此一缺點,並以1974到1999年日本的上市公司為主要研究對象。此外,在過去文獻對展望理論的基本三個假設之外,本文增加了從橫斷面的角度所推論出的研究假設。 實證結果發現,若採用過去文獻的實證做法,則結果與過去文獻一致,皆是相當支持展望理論。但是採用更符合展望理論精神的作法時,結果並不顯著,且與假設相反。而在橫斷面的報酬變異部分,原本是假設低於產業中位數的公司因會採行風險愛好的態度,因此未來的累積報酬橫斷面變異數將會大於高於產業中位數的公司。但是實證結果則是發現,在中短期(本文為五年)有符合研究假設,但長期(本文為十年)則與研究假設相反。 Generally asset-pricing theories assert a positive risk-return trade-off relationship, which implies that firms are risk-averse. However, Bowman (1980), documents a negative, instead of a positive, relationship between risk and return based on accounting data on firms from 85 U.S. industries. Several studies have shown that the risk-return paradox can be explained based on Kahneman and Tversky’s (1979) prospect theory. Prospect theory argues that individuals use target or reference points in evaluating risky choices. In this article, to conform to the spirit of the prospect theory, I will reexamine the risk-return relationship by running regressions for firms below and above the target level based on returns over a past ranking sample period. I found that the prospect theory is not as strong as the traditional literature has shown.