超额收益的可预测性与资产配置——基于中国股票市场数据的研究

Excess Return's Predictability and Asset Allocation: Evidence from China's Stock Market

本文首先通过在三个限制变量(M1月环比增长率、CPI月环比增长率和银行间市场回购利率)之中选取变量来构建预测模型,检验中国股票市场的可预测性在统计上是否显著;之后将交易成本和各种交易限制纳入考虑,通过最大化投资者效用,基于不同的预测模型进行实时环境下的资产组合配置,并考察统计显著性和投资绩效之间的关系。研究表明,中国股市超额收益与M1月环比增长率之间具有稳定而显著的线性关系。在短期内,统计显著性与优化的投资绩效之间并非完全相关,而长期内,依据M1月环比增长率构建预测模型的投资策略可能带来良好的绩效。

In this paper,we construct forecasting models based on three variables( M1 growth rate,CPI growth rate,and interbank market repo rate),and test the statistical significance of the predictability of the excess returns of China's stock market. Then,we make asset allocation based on the forecasting models by maximizing investor's expected utility function,and examine the relation between statistical significance and portfolio performance. We conclude that the excess return could be partially predicted by M1 growth rate. The empirical results also show that statistical significance doesn't lead to improved performance in short period,while it is a good choice in the long run to make asset allocation according to M1 growth rate in roll-over method.