資產價格泡沫使得其市場價值脫離了基本面，股票價格與基本面在投機的1990年代後期呈現著顯著分歧。許多研究發現在1990年代以前股票報酬是可以被預測的，本文想要探討股票價格泡沫是否會影響盈餘預測股票報酬。Phillips et al. (2011)所提出來的往前反覆迴歸方法可以偵測到資產價格泡沫起始及結束點。Goyal & Welch (2003)、Lettau & Ludvigson (2005)與Ang & Bekaert (2007)雖提出在1990年代股價報酬無法被預測，但是他們的1990年代樣本區間並不一致，本文採用Phillips et al. (2011) 可以準確的指出1990年代的S&P股價泡沫區間。本文進一步發現股價泡沫確實會影響盈餘預測股票報酬，盈餘變數只有在非泡沫時期才能預測股價報酬。投資者了解 到泡沫的出現與否，確實會影響到基本面對股價報酬的預測能力，可以調整其投資策略從事資產配置。

Asset price bubble means that the market value diverges from fundamental value and stock market price diverged significantly from the fundamental during the speculative period of the late 1990s. The majority of studies establishing strong evidence of the predictability of stock returns use data from before or up to the early 1990s. I hypothesized that bubbles will affect predictability of stock returns through earnings. The methodology presented by Phillips et al. (2011) is not only an ex ante econometric methodology but also one of the first attempts to date the origin and conclusion of a bubble period. This study clearly identifies the beginning and ending of the 1990s S&P bubble period. Goyal & Welch (2003), Lettau & Ludvigson (2005), and Ang & Bekaert (2007) all argued that stock returns could not be predicted when the sample includes the 1990s; however, their 1990s sample periods were not consistent and they did not indicate the beginning and ending of the 1990s stock bubble period. I present evidence that stock price bubbles affect the predictability of stock returns through earnings, and that this predictability only exists in the periods in which no bubbles are present, the prebubble and post-bubble periods. The results are helpful for investors seeking to identify stock bubble periods, realizing the influence and consequence of stock bubbles, and performing their assets allocations.