Distribution-Free Structural Estimation with Nonlinear Budget Sets

I develop a structural method for evaluating labor supply in nonlinear budget sets that does not require any distributional assumptions. The model only requires that preferences are convex on the budget frontier. It can be extended to account for features such as fixed costs of work and the stigma cost of welfare participation. It can also be adapted for estimation of earnings, hours of work, and functions that depend on the labor supply distribution, including tax revenue and cumulative distribution functions. The method is applied to estimate the effects of taxes on various labor supply outcomes in the U.S. and Sweden.