Testing Consumption Optimality Using Aggregate Data

This paper tests the optimality of consumption decisions at the aggregate level, taking into account popular deviations from the canonical constant-relative-risk-aversion (CRRA) utility function model—rule of thumb and habit. First, we provide extensive empirical evidence of the inappropriateness of linearization and testing strategies using Euler equations for consumption—a drawback for standard rule-of-thumb tests. Second, we propose a novel approach to testing for consumption optimality in this context: nonlinear estimation coupled with return aggregation, where rule-of-thumb behavior and habit are special cases of an all-encompassing model. We estimated 48 Euler equations using GMM. At the 5% level, we only rejected optimality twice out of 48 times. Moreover, out of 24 regressions, we found the rule-of-thumb parameter to be statistically significant only twice. Hence, lack of optimality in consumption decisions represent the exception, not the rule. Finally, we found the habit parameter to be statistically significant on four occasions out of 24.