Empirical tests of consumption-based asset pricing models using household-level consumption data

In this dissertation I examine empirically several implications of modern consumption-based asset pricing models for the individual consumption behavior and for its relationship with asset returns. I use the data from the Consumer Expenditures Survey to construct individual consumption.;In the first essay I consider the relationship between cross-sectional dispersion of consumption growth rates and asset returns. I find that in the context of conditional CCAPM this "idiosyncratic risk" factor is priced and helps explain the cross-section of expected returns on the Fama-French 25 portfolios sorted on size and book-to-market ratios. This finding supports recent theories that cross-sectional heterogeneity of consumers plays a role in determining asset risk premiums.;In the second essay I test implications of habit formation preferences for individual consumption. I do not find evidence which would be consistent with habit formation in consumer preferences. This finding is very stable and robust across model specifications, methods of estimation and choice of instrumental variables. This contrasts with typically affirmative conclusions of studies examining habit formation using aggregate consumption data. My results indirectly support some recently proposed models which imply habit-like behavior at the aggregate level without directly assuming habit formation utility for the individual consumers.;In the final chapter of the dissertation I consider further tests of the conditional CCAPM model studied in chapter 1 in order to address critical issues raised in recent research by Lewelen Nagel and Shanken (2007) concerning linear factor models tests on 25 Fama-French portfolios. I find that model fit deteriorates when 25 portfolios are considered in combination with additional industry-sorted portfolios. Therefore the critiques of Lewelen et al. do apply in the context of the present model as well as in many other asset pricing models which appeared in the literature and are considered here for comparison. I further find that, consistent with the results of chapter 1, the idiosyncratic consumption risk factor remains priced factor and that the fit of the model estimated separately on industry portfolios is better than on the original set of 25 portfolios.