We provide a new rationalization of why social security may improve welfare. We consider a setting where the introduction of funded social security cannot improve welfare for a fully rational agent. We introduce error-prone individuals who make stochastic savings decisions according to the consistent-mistakes model. The expected utility of error-prone agents is lower than rational decision-makers even if, on average, they save the same. Furthermore, error-prone individuals save less for retirement in a multi-period setting than their rational counterparts. Social security limits the scope of mistakes agents make in their savings decisions and may generate substantial welfare gains.

We investigate the design of optimal transfer policies in developing countries, characterized by substantial inter-regional economic disparities, information asymmetry between central and sub-national governments, and externalities arising from public goods. To address these challenges, we construct a principal-agent model that incorporates asymmetric information between the two tiers of government and considers externalities. Local regions are different in income levels, preference for public goods, or both. We conduct numerical simulations based on the county-level fiscal data of China during 2016-2019. Our finding reveals that the optimal marginal transfer curves under both uni-dimensional and bi-dimensional heterogeneities scenarios are considerably lower than the prevailing one. The optimal transfer undertakes the function of the Pigouvian tax to correct externalities. Moreover, transitioning from the current transfer system to the optimal one yields a substantial welfare improvement, equalling a per capita consumption increase ranging from 3.02% to 4.11%

This paper analyzes the interactions between redistribution and unemployment insurance policies and their implications for the optimal design of tax-benefit systems. In a setting where individuals with different earnings abilities are exposed to unemployment risk on the labor market, I characterize the optimal income tax schedule and the optimal unemployment benefit schedule in terms of empirically estimable sufficient statistics. I provide a Pareto-efficiency condition for tax-benefit systems that implies a tight link between optimal redistribution and optimal unemployment insurance: the steeper the profile of income taxes is, the flatter the profile of unemployment benefits should be, and vice versa. Optimal replacement rates are therefore monotonically decreasing with earnings, from 1 at the bottom of the earnings distribution to 0 at the top, and redistribution through unemployment benefits is efficient. Empirical applications show that these interactions between redistribution and unemployment insurance have important quantitative implications.