權益成本估計的不確定性將嚴重影響企業價值評估的準確性，因此本文檢視CAPM單一因子模型及Fama and French三因子模型，依1982年7月至2007年12月之台灣二十八類產業為研究樣本，以傳統的所有樣本期間迴歸（Full-period Regression）與滾動式迴歸（Rolling Regression）兩種不同估計方式，發現CAPM及Fama and French三因子模型之風險係數隨著時間改變的平均真實隱含標準差均高達0.18以上。比較兩模型在產業權益成本預測能力的比較上，Fama and French三因子模型不論預測近期或遠期的表現均優於CAPM外，預測的期間越久，兩模型預測能力差距將越小，其預測能力也越接近。在我們進一步利用兩種定價模式估計產業權益成本發現，Fama and French三因子模型明顯降低了產業權益成本的波動狀況，並且CAPM與Fama and French三因子模型使用五年滾動式迴歸模型估計的差異大於使用完整樣本期間估計的方式。有鑑於此，當我們選擇資產定價模型做為資產評價上之權益成本應用時，在預測短期間的權益成本估計上，Fama and French三因子模型以長期間的完整歷史時間估計仍為有較佳方式。然而在預測遠期的權益成本估計時，CAPM仍為不錯的估算方式。

在知識經濟時代下，無形資產的對於公司的重要性愈來愈高。有別於在工業時代下的生產重心，著重在大量的土地，機器設備...等有形的資產，在二十一世紀競爭中致勝的關鍵因素卻是那些無實體存在的知識累積，例如：研究發展的能力、員工的素質、顧客關係的維持…等;然而，會計處理對於無形資產卻是停留在歷史的取得成本，而不是現時的市場價值，更甚，有些無形資產根本無法入帳;因此，資本市場如何看待與反應公司的無形資產就是一項有趣的議題。本研究之研究目的是：依據Fama and French (1993)三因子模型，以橫斷面的分析方式，欲控制了系統風險、規模效果和淨值與市價比效果後，進一步分別探討研究發展活動與專利權對於股票報酬之影響，是否擁有投入愈多的研發活動與專利可以在股票市場獲得愈高的報酬？是否研究發展費用與專利權數對於股票報酬有遞延效果的影響？ 樣本期間從民國七十一年到民國九十三年，包含上市與上櫃公司，總共有21717筆觀察值，在研究發展活動方面，本研究採用了當期研究發展費用與依五年資本化後之研究發展費用二種替代變數，專利權方面，採用專利權數與累積專利權數二種代理變數，其實證結果發現： (1)當期研究發展費用溢酬與股票超額報酬呈現顯著的正向相關，將研究發展費用依五年資本化後，資本化後之研究發展費用溢酬仍與股票超額報酬呈現顯著的正向相關。 (2)專利權數之溢酬與股票超額報酬卻是顯著的負相關，累積專利權數之溢酬與股票超額報酬也是呈現顯著的負向關係，可能的原因是：在本研究樣本裡的大部分的專利權數量是非常集中在少部分的公司。 (3)研究發展溢酬對於超額報酬最多有三年的遞延效果，專利權溢酬對於超額報酬至少有五年遞延的效果。 (4)當期研究發展費用溢酬與資本化後之研究發展費用溢酬對於超額報酬有顯著不同的影響，二者比較下，當期研究發展費用溢酬對於股票報酬的影響程度大於資本化後之研究發展費用溢酬。可能的原因是:Fama and French三因子模型某種程度上代表著流量的概念，因此，當期研究發展費用溢酬的效果較為顯著。 (5)在專利權數之溢酬與累積專利權數之溢酬二者之間，對於超額報酬不具有顯著差異性的影響。可能的原因是:大部分的樣本都沒有專利權，因此，專利權數之溢酬與累積專利權數之溢酬沒有太大的差異。 (6)以研究發展與專利來說，二者對於超額報酬具有顯著不同差異的影響。 / Since the change in the global economy in the last decade, from manufacturing and industry-based to knowledge-based, it has created new interest in intellectual capital and increased the demand for measuring and reporting the effect on business and profitability. Nonetheless, accounting conventions based on historical cost often understate their value. Thus, from a practical point of view, how the stock market responses to the innovative activity is an interesting issue. Here, the major objective of this study is, on the basis of the three-factor model in Fama and French (1993), to investigate the relationship between innovation activities in firms and stock returns. That is, the aim in this study is to examine whether the intellectual capital, in particularly focusing on R&D and patents, has impact on stock returns. Does the market provide the premium for the value of the innovation in firms? Do the stocks with more innovation efforts worth the higher market rate of returns? Do R&D and patents have time lag effect on returns? We find that: (1) The return premiums are significantly greater for high-level of R&D than for low-level R&D. The mimicking returns both for the R&D-expense factor and capitalized-R&D factors are significantly positive related to excess stock returns. (2) Contrary to our intuition and expectation, the mimicking returns both for patent count and cumulated patent count are significantly negative associated with excess stock returns. One possible explanation is that the distribution of patented innovation is known to be extremely skewed, implying that a few patents are very valuable and many are worth almost nothing. (3) R&D-related return premiums have 3-year lag effect on excess stock returns at most. As for patent-related return premiums, it shows 5-year lag at least for excess stock returns. (4) R&D expenses have more impact on stock returns than the R&D capitalization. One possible explanation is that the “flow” concept is more suitable than the “stock” concept in the Fama and French (1993) regression of stock returns. (5) There is no difference between patent count and cumulated patent count in explaining stock returns. It is likely that, for a large proportion of the sample, they do not possess any patents. (6) When it comes to compare R&D to patents, we find that there is statistically significant difference between the two in explaining excess stock returns.