Temptation and Self-Control: Some Evidence and Applications

This paper studies the empirical relevance of temptation and self-control using household-level data from the Consumer Expenditure Survey. We construct an infinite-horizon consumption-savings model that allows, but does not require, temptation and self-control in preferences. To distinguish temptation preferences from others, we exploit individual-level heterogeneities in our data set, and we rely on an implication of the theory that a more tempted individual should be more likely to hold commitment assets. In the presence of temptation, the cross-sectional distribution of the wealth-consumption ratio, in addition to that of consumption growth, becomes a determinant of the asset-pricing kernel, and the importance of this additional pricing factor depends on the strength of temptation. The empirical estimates that we obtain provide statistical evidence supporting the presence of temptation. Based on our estimates, we explore some quantitative implications of this class of preferences for capital accumulation in a neoclassical growth model and the welfare cost of the business cycle.