本文考量實際股市可能價格異象之潛在影響,並力求樣本的完整性,將研究焦點置於Fama and French(1992)研究過程,實證分析排序前beta估計期、樣本分組數、以及股票報酬衡量期間,對台灣股市beta之估計及解釋能力之影響。本文發現排序前beta的估計會受到估計期長短的影響,並改變公司的分組別,以及排序後beta之估計與分派結果。同時,排序後beta針股票報酬衡量期間亦具有敏感性。實證結果顯示引用Fama and French(1992)方法可能受到排序前beta估計期、報酬衡量期間及分組數多寡之影響。最後,本文發現月beta無法解釋股票預期報酬,但季beta、半年beta及年beta則有部分呈現顯著解釋能力,其主要集中於上市比較久的公司且分組數為36組之時。
Basing our empirical study on a sample of corporations listed on Taiwan Stock Exchange and Taiwan's OTC Securities Exchange between 1981 and 2000, we find that Fama and French(1992) approach, including estimation procedure and the explanatory power of beta, is sensitive to the length of time to estimate pre-ranking beta, the size of grouping portfolio, and the return measurement interval used in estimating post-ranking beta. In addition, the result indicates that monthly beta fails to capture the cross-section of expected returns in Taiwan. However, we have presented evidence that average returns do reflect substantial compensation for beta risk, provided that betas are measured at the quarterly or even larger intervals, for aged firm samples sorted into 36 portfolios. This finding is in support of the view that the beta is still alive in Taiwan.