Research in Finance, 11 (1993): 1-35
Wayne E. Ferson
University of Washington
Campbell R. Harvey
This paper examines linear asset pricing models using aggregate consumption data which are not seasonally adjusted. Portfolio returns exhibit more diverse and often stronger consumption correlations, and the parameters of representative-agent utility functions seem to be estimated more precisely using these data. We incorporate seasonality in the form of "taste shift' parameters and seasonally varying heteroscedasticity. The pricing relations can be rejected using a short-term bill or multiple asset returns. The data indicate that the "equity premium puzzle" is not an artifact of seasonal adjustment. The evidence includes long-term bonds as well as bills and stocks, and seems to involve not only the relation of average returns to consumption covariances but; also the patterns of variation in expected returns over time.