摘要
根据最优投资组合理论,本文证明可以采用均值-方差相交的方法确定模拟资产组合作为因素模型的定价因子。为了得到稳健的实证结论,文中对GMM估计的Wald统计量进行了Block-Bootstrap模拟,同时研究了模拟过程中区组(Block)长度的选择问题。实证结果表明,本文构造的大、中、小三个规模组合可以作为中国股市的定价因子;当区组长度选定时,GMM-Wald统计量的Block-Bootstrap模拟不会随着模拟次数的变化发生显著变化,但当区组长度超过10时,它与其它区组长度对应的模拟结果会有显著差异。
This paper proves that we can use the method of mean-variance intersection to determine the mimic portfolio as the pricing factor for the factor model by the theory of the optimal portfolio. To get moderate empirical conclusion, this paper applies the Block-Bootstrap simulation to the GMM-Wald test, and explores how to choose the block at the same time. The empirical study shows that the constructed little ,middle and the large size portfolio can be the pricing factor of the Chinese stock market, and the Block-Bootstrap simulation of the GMM-Wald statistics will not change a lot as simulation frequency changes when the block length is fixed, but when the block length is more than 10, the simulation result will be obviously difference relative to the other block length.
出处
《南方经济》
北大核心
2007年第8期74-84,共11页
South China Journal of Economics