Final Report Summary - OPTITAX (Dynamic Optimal Taxation with Human Capital Formation)
The goal of my research was to investigate optimal tax policies and efficient allocations in an environment where people can increase their skills by investing in human capital. Since human capital investments are made over time, this calls for a dynamic optimal taxation framework. I study dynamic optimal taxation through the lens of dynamic mechanism design approach, which places no a priori restrictions on available tax instruments. The tax instruments themselves, and their properties, are jointly a part of the solution to the problem. The key trade-off in the policy design is that the government would like to insure people against human capital and productivity shocks or redistribute resources across people, but doing so decreases incentives to supply labor and accumulate human capital.
Understanding the optimal policy design in dynamic economies is a complex problem, both conceptually and computationally. One needs to think about how to best provide incentives for labor supply and human capital accumulation, both across agents and over time, without sacrificing insurance and redistributional considerations too much. In my first paper, "Pareto Efficient Income Taxation with Learning by Doing", I investigate the dynamics of income taxes in an environment where people can increase their future human capital by working more today (learning by working). The most important findings relate to the dynamic of optimal marginal income taxes over age. I show that the marginal income taxes should be decreasing with age. The main reason is, that labor supply elasticity is low at younger agents, because people understand, that there are future benefits from working today, and this effect naturally decreases at older ages.
There are two additional aspects of human capital investments that are missing in this paper. First, consistently with the empirical literature, human capital investments are risky. Second, in order to increase their future human capital, people must spend both their time and money. In a paper "Optimal Taxation with Risky Human Capital", co-authored with Julian Neira, at the University of Exeter, we examine their importance for the optimal tax design. We provide a case for U-shaped marginal income taxes for high ability agents. Taxes are high if people receive high shocks because of redistributional concerns. But taxes are also high if high ability people receive low shocks, and this is in order to provide them with incentives to accumulate human capital. For low ability people the incentives to accumulate human capital are less important, and the U-shaped pattern disappears. Since one can only differentiate low and high ability agents on the basis of past incomes, the paper also provides a strong case for history dependence in income taxation. We also provide a case for positive schooling subsidy, although the schooling subsidies are smaller than in the previous literature.
Quantitative importance of history dependence in income taxation is investigated in my paper "Quantifying the Welfare Gains from History Dependent Income Taxation". I develop an alternative framework that puts restrictions on the type of tax functions that the government can use, but allows for history dependence, a key feature of the dynamic optimal taxation literature. The framework is not yet adapted for endogenous skill formation, but it has the potential to build a bridge between complex mechanism design models and models with simple interpretations, and practical implications. The fact that there are restrictions on the functional forms used is a downside of this approach; however, there are significant benefits in increased tractability of the problem. The model is thus able to provide a very sharp characterization of the optimal history dependence under realistic assumptions about the underlying wage process, that includes a permanent component, a persistent component, and a transitory component. All of them are standard in the empirical literature, but current mechanism design models cannot quantitatively handle environments with multiple sources of shocks. There is a number of quantitatively relevant findings: the welfare gains from history dependence are significant, although, to capture most of the gains, one needs to include a long history of past incomes; the optimal taxes are more progressive with respect to the current income, but regressive with respect to past incomes; the permanent, persistent and transitory components give conflicting predictions about the nature of history dependence, and the optimum is a compromise between the three forces. The analysis can also be extended to environments with dynastic considerations, and I do so in a separate paper "History Dependent Income Taxation in a Dynastic Framework".
Those findings contribute to a broader literature that studies optimal government policies. The research also emphasizes tractability and ability to make practical recommendations. The existing literature is only starting to incorporate human capital consideration into the design of optimal tax policies, and its policy relevant findings have a potential to shape the tax reforms, undertaken in many countries. In this particular case, the findings about the intertemporal and cross-sectional profile of marginal income taxes, as well as about the importance of history dependence are especially relevant for the design of public policies.
Understanding the optimal policy design in dynamic economies is a complex problem, both conceptually and computationally. One needs to think about how to best provide incentives for labor supply and human capital accumulation, both across agents and over time, without sacrificing insurance and redistributional considerations too much. In my first paper, "Pareto Efficient Income Taxation with Learning by Doing", I investigate the dynamics of income taxes in an environment where people can increase their future human capital by working more today (learning by working). The most important findings relate to the dynamic of optimal marginal income taxes over age. I show that the marginal income taxes should be decreasing with age. The main reason is, that labor supply elasticity is low at younger agents, because people understand, that there are future benefits from working today, and this effect naturally decreases at older ages.
There are two additional aspects of human capital investments that are missing in this paper. First, consistently with the empirical literature, human capital investments are risky. Second, in order to increase their future human capital, people must spend both their time and money. In a paper "Optimal Taxation with Risky Human Capital", co-authored with Julian Neira, at the University of Exeter, we examine their importance for the optimal tax design. We provide a case for U-shaped marginal income taxes for high ability agents. Taxes are high if people receive high shocks because of redistributional concerns. But taxes are also high if high ability people receive low shocks, and this is in order to provide them with incentives to accumulate human capital. For low ability people the incentives to accumulate human capital are less important, and the U-shaped pattern disappears. Since one can only differentiate low and high ability agents on the basis of past incomes, the paper also provides a strong case for history dependence in income taxation. We also provide a case for positive schooling subsidy, although the schooling subsidies are smaller than in the previous literature.
Quantitative importance of history dependence in income taxation is investigated in my paper "Quantifying the Welfare Gains from History Dependent Income Taxation". I develop an alternative framework that puts restrictions on the type of tax functions that the government can use, but allows for history dependence, a key feature of the dynamic optimal taxation literature. The framework is not yet adapted for endogenous skill formation, but it has the potential to build a bridge between complex mechanism design models and models with simple interpretations, and practical implications. The fact that there are restrictions on the functional forms used is a downside of this approach; however, there are significant benefits in increased tractability of the problem. The model is thus able to provide a very sharp characterization of the optimal history dependence under realistic assumptions about the underlying wage process, that includes a permanent component, a persistent component, and a transitory component. All of them are standard in the empirical literature, but current mechanism design models cannot quantitatively handle environments with multiple sources of shocks. There is a number of quantitatively relevant findings: the welfare gains from history dependence are significant, although, to capture most of the gains, one needs to include a long history of past incomes; the optimal taxes are more progressive with respect to the current income, but regressive with respect to past incomes; the permanent, persistent and transitory components give conflicting predictions about the nature of history dependence, and the optimum is a compromise between the three forces. The analysis can also be extended to environments with dynastic considerations, and I do so in a separate paper "History Dependent Income Taxation in a Dynastic Framework".
Those findings contribute to a broader literature that studies optimal government policies. The research also emphasizes tractability and ability to make practical recommendations. The existing literature is only starting to incorporate human capital consideration into the design of optimal tax policies, and its policy relevant findings have a potential to shape the tax reforms, undertaken in many countries. In this particular case, the findings about the intertemporal and cross-sectional profile of marginal income taxes, as well as about the importance of history dependence are especially relevant for the design of public policies.