The United States spent approximately $1 trillion on income security and redistributive tax programs in 2004. Is this social safety net desirable from the perspective of maximizing social welfare? Can the structure of the social safety net be improved to provide the same level of benefits while reducing costs in terms of aggregate economic efficiency? This project aims to identify ways to improve the design of large-scale government programs. The project is organized around three broad ideas:
I. Liquidity. Existing studies have assumed that behavioral responses to government policies -- such as the increase in unemployment durations caused by unemployment benefits -- are caused by moral hazard (a substitution effect), rather than a liquidity (income) effect. However, when individuals have limited savings, as do many agents experiencing shocks such as job loss or illness, they may change their behavior simply because they have more cash-on-hand (liquidity). This project develops methods to estimate the importance of liquidity vs. moral hazard. These estimates are used to quantify the value of various government programs. For example, the method is applied to calculate the value of providing health insurance by diagnostic condition, in order to precisely characterize which types of health insurance coverage should be prioritized in government programs such as Medicare and Medicaid. II. Commitments. Many households have "consumption commitments" such as housing that are difficult to adjust when shocks occur. Existing analyses of optimal policy neglect such commitments by assuming that households can adjust consumption of all goods when shocks occur. This project shows that commitments amplify risk aversion over moderate-sized risks. Commitments therefore increase the value of programs that insure temporary shocks relative to programs that provide long-term redistribution through taxation or welfare support. Hence, the optimal design of the social safety net may have very different characteristics when commitments are taken into account. III. Salience. Another benchmark assumption in economic models of government policy is that all agents are fully informed of and optimize with respect to the incentives created by government policies. This project develops methods to test this important assumption. In collaboration with H&R Block, we study the effects of providing simple information on the Earned Income Tax Credit (EITC) to tax filers in 2007. We conduct an experiment involving 40,000 H&R Block clients and track the amount they earn over two years. The goal of the EITC is to encourage individuals to work more; our objective is to test whether making the incentives provided by the EITC clear to individuals (e.g. $4 for every extra $10 earned) helps achieve this objective.
The research has direct applicability to a broad range of policy issues. The project assists policy makers in assessing the implications of policy changes such as social security privatization, consumption tax reforms, or the introduction of unemployment and health insurance savings accounts. The project also helps identify new ways to implement policies that maximize economic welfare. For example, incorporating the EITC directly into weekly paychecks may make the incentives more salient while providing more liquidity. The project could also shed light on the best methods to simplify the tax code, an issue emphasized as a priority by the President's Tax Reform Panel.
This project explored three broad questions relevant for the design of tax and social insurance programs. First, in a paper with John Friedman, Soren Leth-Petersen, Torben Heien Nielsen, and Tore Olsen, we utilize administrative tax records on 45 million Danish households to study saving responses to changes in tax subsidies and automatic savings contributions. We find that price subsidies are less effective than automatic contributions in increasing savings rates for three reasons. First, 85% of individuals are passive individuals who save more when induced to do so by an automatic contribution but do not respond at all to price subsidies. Second, individuals who respond do so primarily by shifting savings across accounts rather than raising the total amount they save. Third, the active savers who respond to price subsidies tend to be those who are planning and saving for retirement already. These results call into question whether subsidies for retirement accounts and reductions in capital income taxation are the best way to increase savings rates. In a second project (with John Friedman and Emmanuel Saez), we estimate the impacts of the Earned Income Tax Credit on labor supply using areas with little knowledge about the policy as counterfactuals. Individuals in high-knowledge areas change wage earnings sharply to obtain larger EITC refunds relative to those in low-knowledge areas. These responses come primarily from intensive-margin earnings increases in the phase-in region, showing that the EITC is effective in increasing labor supply. With Emmanuel Saez, I also conducted a randomized experiment with 43,000 EITC recipients at H&R Block. We test whether providing information about the EITC affects recipients’ earnings decisions. Tax preparers at H&R Block gave half of their clients simple, personalized information about the EITC schedule. We find that providing information about tax incentives does not systematically affect earnings on average. However, tax preparers may influence their clients’ earnings decisions by providing advice about how to respond to tax incentives. Finally, in a project with Nathan Hendren, Patrick Kline and Emmanuel Saez, we used administrative records on the income of more than 40 million children and their parents to describe intergenerational mobility in the U.S. First, we characterize the joint distribution of parent and child income at the national level. The conditional expectation of child income given parent income is linear in percentile ranks. On average a 10 percentile increase in parent income is associated with a 3.4 percentile increase in a child’s income. Second, intergenerational mobility varies substantially across areas within the U.S. For example, the probability that a child reaches the top quintile of the national income distribution starting from a family in the bottom quintile is 4.4 % in Charlotte but 12.9 % in San Jose. Third, we find that upward mobility is correlated less residential segregation, less income inequality, better primary schools, greater social capital and greater family stability. Together, these studies show that tax policies can have meaningful impacts on economic outcomes, but careful consideration of the impacts of each program is needed to determine how best to design policy.
