Optimal Dynamic Taxation of Households: Insurance, Incentives and Commitment

This paper introduces households in a dynamic Mirrleesian economy and explores the resulting implications for optimal taxes. Households are modelled as a long run partnerships between two individuals. Each agent also in a long term relationship with the government, whose preferences are defined over individual utilities. Each individual is subject to an idiosyncratic privately observed skill shock. Partners do not have an informational advantage relative to the government, but they can pool income after the realization of their idiosyncratic shock. The fact that individuals are linked via the household structure introduces externalities in the relation between the government and each individual. The nature of these externalities depends on the process intra-household allocation. We examine two settings. In the first, partners have commitment. In the second, partners remain in the household only if their utility is greater than an outside option that depends on their history of shocks. We find that inter- and intratemporal distortions are greater than in the corresponding model without households in both cases, which implies that marginal asset and labor income taxes will be greater in absolute value. Instead, the properties of the distribution of individual and household consumption differ markedly in the two models. We study numerical examples to illustrate our findings.