This paper studies wealth inequality through the lens of costly self-control, modeled as Temptation Preferences, which introduce a novel term in the consumption–saving problem that acts as an effective discount factor. Relative to standard frameworks with fixed time preferences, temptation provides a structural, behaviorally grounded account of heterogeneity in discount rates and the positive association between patience and wealth, matching several empirical regularities. A stylized setup yields two mechanisms shaping intertemporal choice: the current resources channel (the effective discount factor increases with available resources) and future income channel (under standard calibration, it decreases with expected income). Embedding this mechanism in an otherwise standard OLG model, the current resources channel dominates, generating a discount-factor gap between richer and poorer agents. This in turn enables a parsimonious temptation model to match the observed wealth distribution more closely, outperforming a rational benchmark. It also shapes the distributional impact of taxation: relative to the rational baseline, wealth taxation is more effective than income taxation at reducing wealth inequality in the presence of temptation.
