Measuring self-control problems: A structural estimation

We perform a structural estimation of the preference parameters in a buffer-stock consumption model augmented with temptation disutility. We adopt a two-stage Method of Simulated Moments methodology to match our simulated moments with those observed in the US Survey ofConsumer Finances. To identify the parameters we use liquid and quasi-liquid (retirement) wealth holdings at different ages as target moments. We find evidence of a significantly positive degree of temptation, and estimates under the unrestricted model are more precise than estimates under the restricted model with no temptation. Overall, a model with temptation preferences explains better the observed target moments.