以MIDAS模型探讨亚洲股票市场的期望报酬与风险之关联

Using the MIDAS Model to Investigate the Relationship between Expected Return and Risk in Asian Equity Markets

摘要: 本文使用混和数据抽样(MIDAS, mixed data sampling)模型，探讨亚洲主要股票市场的期望报酬与风险之间的抵换关系，经由观察每日市场报酬的波动，给予适当且弹性的权重，来估计每月的条件变异数，并与使用每月报酬数据做为条件变异数基础的ARCH模型相比较，检定何种模型较能够解释期望报酬与风险的正向关系。本文实证结果得到，使用MIDAS模型有较多股票市场能得到正向的结果，表示能有效衡量期望报酬与风险之间的关系。然而，就模型配适度而言，ARCH模型对市场报酬的预测误差少于MIDAS模型，故MIDAS模型的使用仍有所限制。

Abstract: We used the MIDAS (mixed data sampling) model to investigate the risk and expected return trade-off in major Asian equity markets. The optimal and flexible weighted of lagged daily squared market returns were adapted as the variance estimator to predict the monthly variance. In terms of the ARCH model that adapts the monthly squared returns as the variance estimator, we examined which model has greater statistical power to explain the positive relationship between risk and expected return. Our results proved that the MIDAS estimator had the ability to find more markets that had the positive and significant relationships. However, the pricing errors of the ARCH model were less than the MIDAS model, when we considered the fitness of the models. This evidence suggested that the MIDAS model cannot outperform the ARCH model in predicting market return.