- 學位論文
利用益本比預測台灣股票之報酬
Using Earnings-Price Ratios to Predict Taiwan's Stock Returns
摘要
本論文旨在研究益本比對於台灣股票報酬之預測能力，所使用的資料為股票報酬及益本比的月資料，並由台灣21個產業類別中各取一家代表性的上市公司做為研究對象。我們首先利用Harvey 等人在2006年所提出的modified residual-based ratio test方法探討益本比是否有整合性階次的變動情形，接著利用Campbell and Yogo (2006) 的Bonferroni Q-test檢定益本比對於股票報酬的預測能力，最後，再利用Granger因果關係檢定來檢視兩變數間的領先、落後關係。 實證結果顯示，絕大多數的公司其益本比的整合階次確實會從I(0)變動至I(1)、或從I(1)變動至I(0)，而股票報酬則否。當益比本及股票報酬皆為穩定的I(0)序列時，益本比對於股票報酬是具有預測能力的；然而當益本比轉變為I(1)序列時，則股票報酬之預測迴歸式將不具對稱關係，益本比亦會失去其預測能力。此結論對於為何益本比在某段期間會有較強預測力而在某些期間則否提供了一個計量上的解釋。
並列摘要
This thesis provides evidences of predictability on Taiwan’s stock returns using earnings-price ratios. Data employed in this research are monthly stock returns and earnings-price ratios of representative companies chosen from Taiwan’s 21 industries. The modified residual-based ratio tests of Harvey et al. (2006) are first adopted to explore whether earnings-price ratio series have undergone a change in persistence. Then, the Bonferroni Q-test in Campbell and Yogo (2006) is used to examine the predictive power of earnings-price ratios. Finally, Granger causality tests are adopted to investigate lead-lag relationships between these two financial variables. It is discovered that most companies’ earnings-price ratio series have experienced a change in persistence either from I(0) to I(1) or from I(1) to I(0), while stock return series have not. Earnings-price ratios have predictive power for future stock returns when the two variables are I(0) processes. However, if earnings-price ratios become I(1) processes, the predictive regression would be unbalanced and earnings-price ratios lose their predictive power. This provides an econometric explanation for why earnings-price ratios show strong predictive power during some periods while they have weak or no predictive power during other periods for Taiwan’s stock returns.