Evaluating the C-CAPM and the Equity Premium Puzzle at Short and Long Horizons: A Markovian Bootstrap Approach.

We investigate the C-CAPM and the equity premium puzzle using asset returns and consumption data from the US and Denmark. In contrast to previous studies the investigation is carried out with both short and long investment horizons. In addition, we introduce a Markovian bootstrap approach to calculate the finite-sample distributions of the various test statistics used. Among other things, our approach allows testing the hypothesis of no pricing errors based on the Hansen and Jagannathan (1997) specification error measure, which is not possible based on asymptotic approximations. The analysis shows that there are large differences between US and Danish asset markets, and also to some extent differences depending on the length of the investment horizon. In addition, with a long investment horizon, the asymptotic and bootstrap estimated distributions often differ markedly. We also find that althrough the Hansen-Jagannathan specification error measure points to large pricing errors, when we account properly (using the bootstrap approach) for sampling error, there is no evidence against the C-CAPM with CRRA utility in neither the US nor Danish data. However, the bootstrap based tests of a zero-mean risk-adjusted equity premium clearly reject the model on US data.