Association for Public Policy Analysis & Management

The Earned Income Tax Credit (EITC) has decreased poverty, increased income, and improved health for millions of low-income families. However, these benefits come at a high cost: $65 billion for tax payers and the federal government in 2017. Are the benefits of the EITC high enough to justify this reported cost?

Unlike most public assistance programs, EITC benefits require that recipients work. Increased employment means that these mothers pay more payroll taxes, and increased family resources -- with a propensity to consume near one -- leads to more state and local sales taxes paid. Furthermore, the EITC decreases reliance on other forms of public assistance (e.g. welfare and food stamps). These factors suggest that the overall cost of the EITC -- to tax payers and the government -- is less than generally understood.

To estimate the overall cost of the EITC, we use three decades of population tax data merged with confidential CPS ASEC data and exploit dozens of plausibly exogenous federal and state EITC policy changes. Running tax simulators (e.g. TAXSIM, Bakija model) on this data provides estimates of various types of taxes paid and EITC benefits received, at the federal, state, and local level.

We first use public CPS data and find that EITC-led changes in payroll, unemployment, and sales taxes, and in public-assistance usage completely offset the EITC's cost to government. A third of this effect is due to increased tax revenue and two-thirds is due to decreased public assistance. If anything, these results may be lower-bound estimates since work by Bruce Meyer shows that public assistance is under-reported in survey data. Marginal treatment effects appear roughly constant and suggest that state and federal governments could further expand the EITC at a fiscal cost near zero.

Next, we will follow the approach in Jones (2013) and link administrative tax data with the CPS ASEC and see how the public-data results hold up using the linked data.

The EITC has also improved health and decreased incarceration and crime, as well as improved economic outcomes for children of EITC recipients. Accounting for the social value of these benefits suggests that the EITC more than pays for itself. If this result were more salient to lawmakers, the EITC would likely be expanded.

Evaluating the static results above are more complicated over a lifetime. For example, payroll taxes are eventually paid back as social security benefits; reduced public assistance likely reduces the public assistance of the next generation; and young mothers that begin working, gain experience, see wage increases over time, and pay even more taxes. We construct a simple dynamic model to account for these factors and consider the overall cost of the EITC over a longer time horizon.