The substantial evidence research of the Taiwanese stock market ：The analysis by the D-CAPM

[[abstract]]過去30幾年來，實務界及學術界對於系統風險的衡量有極多的論述，且多數的討論均架構在CAPM的理論基礎之上，本研究所採用之論述是由Estrada(2002)年發表的系統風險預測的模型：稱之為D-CAPM或是downside CAPM，此模型是以CAPM為基本理論架構，針對CAPM之中立投資行為假說MVB(mean-variance behavior)提出不同的修正模型，新模型認為MSB (mean-semi-variance behavior) 的投資行為的假說更能實際反射投資人所呈現之投資行為。 Estrada(2002)實證研究發現，D-CAPM在新興市場EMs(emerging markets)的解釋能力遠高於已開發市場DMs (developed markets)，而台灣證券交易市場歷經MSCI兩次調高台灣之投資限制因子(LIF)之後，已成為MSCI新興市場指數中權重最高的國家，因此利用台灣證券交易所之交易資料作為實證將具有較高的代表性，研究結果發現台灣證券交易所的交易資料亦顯示D-CAPM並無法大幅改善決投資預期報酬率的預估，但當市場資訊較為透明，資訊不對稱的情況愈不明顯，CAPM及D-CAPM確實較能發揮預測的功能。[[abstract]]In the past more than 30 years, the practical and academic fields have numerous treatises to the measurement of the system risk, and most of the discussions are based on the theory of CAPM. .The treatise of this study adopts the model of system risk prediction announced by Estrada in 2002: Called D-CAPM or the downside CAPM. This model takes CAPM as the basic, foundation and puts forward the different correction model to CAPM’s MVB (mean-variance behavior), the new model suggests that the hypothesis of investing behavior of the MSB (the mean-semi-variance behavior) mostly reflects the investor’s actual investing behavior. The Estrada (2002) substantial evidence research finds that the hermeneutic ability of the D-CAPM in the emerging markets (EMs) is far higher than developed markets (DMs). After been keyed up twice by MSCI on Taiwan’s Limited Investability Factor (LIF), Taiwan stock trading market has became the heavy weighted market within the emerging markets, so makes use of the transaction data of the Taiwan Stock Exchange will have the substantial representation. The research finds the D-CAPM can not significantly improve the prediction of estimated return of investment based on the transaction data of the Taiwan Stock Exchange, but when the market information is more transparent and the condition of asymmetric information is more unclear, the CAPM and D-CAPM can actually provide predictive functions