一般用於衡量異常報酬的變數，如本益比(P/E ratio)、規模(size)、及公司市場價格與帳面價格比(M/B ratio)…等常被用來分析且資料取得容易。傳統模型去觀察可能對報酬有影響的變數去捕捉異常報酬，而本研究亦在探究規模與報酬間的關係，但不同於傳統方法(Fama(1995))，Fama檢驗異常報酬是按規模大小分成三組，兩兩做比較，比較組間是否有差異，如果有差異即存在異常報酬，本研究延續其研究依規模大小分組，觀察是否存在異常報酬的現象，然而不同的地方，在於吾人引進隨機效果模型，視各組具有個別效果。由此，可檢驗異常報酬的存在，且更進一步可以檢驗是否具內生性,即可透過操作影響。實證部份以民國92年1月至民國93年12月底的台灣上市電子公司為樣本資料，按規模大小分組去捕捉個別公司是否存在異常報酬，接著比較CAPM及APT兩模型是否同時捕捉到規模效應的內外生性，並驗證出台灣上市的電子產業在CAPM模型個別效果為內生變數，個別效果受 值影響；APT模型成功捕捉到不同規模公司下有不同的異常報酬。 This study applies the random effects panel data model on Taiwan electronic industry, in which we investigate the CAPM and APT models that follow Fama (1993) to categorize the Beta coefficients by firm size. Through the individual effects that correspond to the firm sizes we can test the exogeneity by Hausman test where the insight of the anomaly effects can be revealed. This is different from the conventional methods that usually investigate the anomaly effects by the P/E ratio, or firm sizes, or M/B ratio, etc. In the empirical study, the day data between June 2003 and December 2004 are available for this research that includes 390 companies and 498 periods. In the result analysis, we find the individual effect is an endogenous variable in CAPM that indicates the anomaly effects can be maneuvered by the market operation, and from the APT model we can see the different size has distinguish anomaly return.
