This paper investigates the stability of (a) covariance matrix and correlation matrix of stock returns, (b) the number of factors extracted, (c) degree of relationship of APT pricing among 13 different sample groups (three types of sample groups) across monthe in Korean Stock Market.
These three types of sample groups are random samples, industry-based samples and market-value samples, containing twenty firms respectively.
There are 2160 daily returns from 1.3 1980 to 12.31 1988. We use an maximum-likelihood factor analysis and asymptotic principal component analysis to estimate the pervasive factors influencing asset returns and regress factor lodings extracted in first stage on the stock returns.
The empirical results shows that mean stock returns January are higer than those of months as well as that of the entire year. Also the number of factors extracted by maximum-likelihood factor analysis and asymptotic principal component analysis are stable across months and sample groups.
In conculusion, the number of factors are stable across months and sample groups. APT pricing relationship are particulary valid in January month. Some empirical method will be improved by future research.