THE SEASONAL STABILITY OF THE FACTOR STRUCTURES OF STOCK RE

Investigation is made of the month-by-month stability of: 1. daily returns and correlation coefficients of stock returns, 2. correlation and covariance matrices, 3. number of return-generating factors, and 4. the Arbitrage Pricing Theory (APT) relationship. Data on 340 companies are obtained from daily stock returns on the New York or American Stock Exchanges and listed on the CRSP Daily Returns File during 1973. The results demonstrate that there is a January effect and a small-firm effect in stock returns. Correlation matrices are more stable than covariance matrices, but both kinds of matrices are not stable across months and across the sample sets. The number of return-generating factors is rather stable most of the time and for most of the sample groups, but there is some relevant instability that is associated with the average correlation coefficients among stocks. The APT pricing relationship does not appear to be supported by the 2-stage process using the maximum-likelihood factor analysis.