The asset pricing literature has calibrated models with external habits and documented that these models are successful at generating a large set of stylized facts about asset prices. In this dissertation, I re-consider this evidence by estimating the preference specification by Campbell and Cochrane (1999) using GMM and cross-sectional regression in three different dimensions. First I assume complete insurance among all the individuals and estimate the model using aggregate consumption data. Second, I estimate the model using lower frequency data and compare it with other factor models to explain the cross-section of stock returns. This is motivated by some recent empirical evidence that suggests that consumption-based asset pricing models have found remarkable success to explain the cross-section of stock returns using lower frequency data. Finally I acknowledge that not all the households are actually trading and holding stocks, thus I use household-level data to perform estimations in other market settings. For the latter I assume market incompleteness among stockholders along with limited participation. |