臺灣博碩士論文加值系統

　　本研究得到結論如下：
　　一、本研究所採用之模型，即 Daniel , Grinblatt, Titman，和 Wermers所提出以特徵建立基準的評估模型，所得到的結果為大部份基金在選股能力、擇時能力，都不具有顯著性。
　　二、本研究採用特徵模型之所有能力指標兩兩之間的相關性為大部份基金不具有一致性，代表所有能力指標兩兩之間不能兼顧。
　　三、本研究利用複迴歸分析，得到特徵模型各個能力指標股票特徵的決策偏好考慮之檢定結果：
　　　　(一)在舊投信時期，大多能力指標偏好股票特徵之市值因素來做決策考慮。
　　　　(二)在新投信時期，大多能力指標偏好以股票特徵的帳面對市值比因素來做決策考慮。
　　四、本研究特徵模型和以CAPM為基礎模型之所有能力指標排名一致性上，都不具有一致性。
　　五、大部份基金之本研究特徵模型大多能力指標不具有持續性。

　　Since aggravating marketable weight of incorporatator is one of major tonalities of government in developing security market, the most direct mode in a short term is allowing investment trust issue new funds. And government wants to improve domestic stock market. Therefore, as more and more marketable funds are increased, the status of mutual funds is more and more important day by day .
　　Because domestic funds performance evaluation that was used by tradition CAPM model in the past, as Jensen Index Treynor Index , Sharpe Index, Treynor and Mazuy Model, Henriksson and Merton, etc., had some disadvantages, The disadvantages were benchmark was improbably effective investment portfolio, miscalculation timer performance and so on .. This Research a dopt a characteristic-based benchmark model, propounded by Daniel, Grinb latt, Titman and Wermers in 1997. The modeling is not CAPM Model but a characteristic-based benchmark model, that is built by holdings, contents of mutual funds and stock characteristic. Besides a characteristic-based benchmark model has empircal advantage.( Daniel, Grinblatt, Titman and Wermers, 1997) Therefore we can measure performance of fund correctly and objectively.
　　The conclusing of this research was fellows:
　　1.The result of a characteristic-based model adopted by this research is that most mutual funds have no significant stock selectivity and timing ability.
　　2. The correlation between all performance index of a characteristic-based model in pairs is that most mutual funds have no consistency and represents all performance index in pairs can not combine each other.
　　3.The results of stock characteristic preference in all perfornance indes of a characteristic-based model by using multi-regression are:
　　　　(1) The preference for stock characteristic of most performance inde of a characteristic-based model in the period of old funds is market value to be considerated for making a decision.
　　　　(2) The preference for stock characteristic of most performance inde of a characteristic-based model in the period of new funds is market value to be considerated for making a decision.
　　4.All performance indes rank in a characteristic-based model and CAPM-based models have no comsistency.
　　5.Most performance index of a characteristic-based model of most funds have no persistence.