| Summary: | 碩士 === 東海大學 === 財務金融學系 === 98 === Markowitz's portfolio theory : the mean - variance model has been widely used in financial portfolio managements for years. In this paper, we construct the efficient frontier by taking ETFs as an example to explore performances based on different risk averse investors、Different adjustment periods and several risk- estimating models , from January 1, 2005 ~ Dec. 31, 2009. In addition to using standard deviation as proxy variable of risk, we estimate conditional variance by GJR GARCH model(Glosten, 1993)，Constant Correlation Coefficient model (Bollerslev, 1990)and Dynamic Correlation Coefficient model(Engle, 2002),and further including macro economic variables. Empirical results indicate that (1) dynamic model of the correlation coefficient (DCC and DCC+ model) asset allocation adjustments in the period of one month, the average returns and the coefficient of variation are better than the other models, in other asset allocation adjustments, the model showed more consistent between Results (2) the more investor risk aversion tendency, in addition to compensation and risk reduction, the coefficient of variation down with validated risk and return trade-off relationship; (3) units to bear the risk in the pursuit of the highest returns, risk lovers choose the best each year to adjust the portfolio, while risk averse investors may choose to adjust the frequency of every three months have a better performance (4) restrictions on the conditions, the model and the risk of property investors, the performance is insignificant, The adjustment period to continue to adjust every month, the most well paid and can create for the high coefficient of variation
|
|---|
