Project abstract for group charivv

Optimal Fiscal Policy in a Model With Uninsurable Idiosyncratic Shocks

How should the government design an optimal tax system in the presence of heterogeneity across individuals and uninsurable idiosyncratic risk in the economy? Should the capital income be taxed more heavily than labor income? How much redistribution and insurance should the tax system provide? What is the optimal transition between the current US tax system and the optimal one? These questions have been in the center of public policy debate for a long time, however the answers are still partial and incomplete. This project addresses them by solving the optimal taxation problem in the workhorse quantitative macroeconomic model with heterogeneous agents and uninsurable idiosyncratic shocks. The researchers compute the optimal path of taxes along the transition between steady states of the model. The initial steady state is calibrated to replicate several aspects of the US economy, particularly the tax code and the distribution of wealth and earnings. The final steady state is endogenously determined by the path of taxes. In the baseline experiment the government is restricted to use three instruments: capital and labor income taxes, and lump-sum transfers. The reason why the previous literature did not achieve the level of generality these researchers do is because the computation of the transition in our environment is challenging which makes optimizing over such transitions a formidable task. Using some of the new advances in computational procedures and the parallel computing techniques they are able to overcome these difficulties. Solving this problem requires a maximization of a function (welfare over transition) of the large number of variables (levels of taxes over transition). The function is expansive to evaluate in terms of computational time. The single main experiment requires around 60,000 evaluations, requiring the power of parallel computing.