Optimal linear and two-bracket income taxes with idiosyncratic earnings risk ¬リニ

a r t i c l e i n f o This paper quantitatively characterizes optimal linear and two-bracket income taxes. We consider a dynamic-stochastic-general-equilibrium model in which tax design involves redistributing income for both equity and social insurance. Substantive findings include: (i) a significant fraction of agents supply zero labor or hold zero assets at the optimum; (ii) neglecting tax distortion imposed on either of labor–leisure and consumption–saving decisions will lead to the prescription of tax codes that deviate substantially from the optimum; and (iii) the optimal two-bracket tax schedule will turn from regressivity to progressivity in the marginal tax rate once the volatility of idiosyncratic shocks becomes sufficiently large. The last finding is consistent with the results in Apps et al. (forthcoming), and it also reconciles the contradictory results regarding the optimal two-bracket tax schedule between Slemrod et al. 1. Introduction Mirrlees (1974) pioneered the study of optimal income taxation in a setting where ex ante identical agents face idiosyncratic shocks to their earnings, but relevant insurance markets are missing. This missing-market setting invites a role for income tax to serve as a partial substitute to absorb income fluctuations and share the idiosyncratic risk across agents. The motive for redistributive taxation here is not for equity per se, but rather for social insurance. Varian (1980) took up the issue addressed by Mirrlees (1974) with the emphasis that a large portion of income differences between agents is attributable to pure luck rather than innate ability. Unlike Mirrlees's static framework where agents make a choice between labor and leisure, Varian considered a dynamic framework where agents make a choice between current and future consumption. The Mirrlees–Varian model of optimal income taxation is one of the pioneering works in the moral hazard class of the principal-agent problem , in which the key tradeoff involved is between inducing incentives and providing insurance have elaborated on Mirrlees–Varian's original idea in a variety of directions. Our paper contributes to this line of the optimal taxation literature mainly on the front that the tax design problem in our model involves " correcting " income distribution across agents for equity as well as providing social insurance to buffer against agents' idiosyncratic risk. 1 Under plausible assumptions, Mirrlees (1971) found that the optimal non-linear income tax is approximately linear. In contrast to Mirrlees (1974), this 1971 seminal work belongs to the adverse selection class of the principal-agent …