Economic forces and seasonality in secirity returns

This article finds strong seasonal behavior in the innovations for three Canadian macroeconomic variables (industrial production, unexpected inflation and GDP). An APT model is estimated as a restricted nonlinear multivariate regression system using seven macroeconomic variables, various residual market factor (RMF) proxies, and the returns on fifty size related portfolios of equities that traded on the Toronto Stock Exchange (TSE). As in Chen, Roll and Ross (1986), five macrofactors (lagged industrial production, lagged GDP, term structure, unexpected inflation, and risk premium) have significantly priced risk premia. The risk premia are insignificant for RMF based on two value weighted indices, and marginally significant (0.10 level) for the RMF based on an equally weighted index, which is somewhat consistent with McElroy and Burmeister (1988) and Brown and Otsuki (1989). The significance of the RMF risk premium appears not to be robust to whether portfolios or individual securities are used in the estimations. The significance of the estimated risk premia for the macrofactors also appear not to be robust to the number of portfolios (equations) used in the estimations. Unlike the risk premia estimates for the RMF, those for the other macrofactors are generally unaffected by the inclusion of a January dummy. This implies that the January seasonal remains a market phenomenon that requires further study.