Marina Azzimonti, Eva de Francisco, and Per Krusell
We study a dynamic version of Meltzer and Richard’s median-voter model where agents differ in initial wealth. Taxes are proportional to total income, and they are redistributed as equal lump-sum transfers. Voting takes place every period and each consumer votes for the current tax rate that maximizes his or her welfare. We characterize time-consistent (differentiable) Markov-perfect equilibria in three ways. First, by restricting the class of utility functions, we show that independently of the number of wealth types, the economy’s aggregate state can be summarized by two statistics: mean and median wealth. Second, we derive the median voter’s first-order condition and interpret it in terms of a tradeoff between distortions and net wealth transfers. Third, we illustrate the key endogenous-taxation mechanisms using 1- and 2-period versions of the model. Quantitatively, we find that the model in its baseline form cannot explain the large wealth inequality that we observe in most economies.