Three Essays in Social Insurance

This dissertation is a collection of three essays on the design of safety net programs for low-income households in the U.S. Many U.S. safety net programs involve in-kind transfers, which are used in order to both alter consumption patterns among recipients and limit take-up by ineligibles. However, in the absence of its own network of providers, the government must rely on private vendors to serve as its agents in rendering transfers, giving rise to two types of agency problems: (1) vendors may refuse to participate in government programs, leaving needy people unserved or (2) vendors may engage in fraud in order to increase their payoff from participation. A separate issue arises when government intervention in private markets causes general equilibrium effects on third parties. The first essay examines attempts to reduce vendor fraud in the Supplemental Nutrition Program for Women, Infants, and Children (WIC) using data on the staggered rollout of a fraud reduction program in Texas. Vendors were required to move to an electronic payment system, which allowed regulators to more easily verify reimbursement claims. I show that the program was effective in reducing fraud, but also that it increased vendor non-participation, leading to a reduction in WIC take-up among eligible women. I also show that the fraud reduction program increased prices paid by non-WIC shoppers by 9\%. My results indicate that the effectiveness of policies intended to alter consumption patterns among welfare recipients depend crucially on the incentives of providers and that enforcement measures interact with these incentives. The second essay, co-authored with Ilyana Kuziemko and Maya Rossin-Slater, analyzes the effects of contracting out Medicaid benefits to insurance companies on health disparities among low-income mothers and children. Increasingly in U.S. public insurance programs, the state finances competing, capitated health plans rather than using a fee-for-service (FFS) model. We study how high- and low-cost infants (blacks and Hispanics, respectively) are affected by the transition from FFS to Medicaid managed care (MMC). We find that black-Hispanic infant health disparities \emph{widen}---e.g., black mortality increases by 12\% while the Hispanic mortality \emph{decreases} by 22\%---and care worsens for blacks. Additionally, black birth rates fall. We present a model of risk-selection in which capitation incentivizes competing plans to offer better care to low-cost clients to retain them in future periods. The third essay uses a novel identification design to study the impact of the Earned Income Tax Credit (EITC) on fertility among low-income mothers. Maternal labor market time is thought to play an important role in childbearing. Therefore, wage subsidies like the Earned Income Tax Credit (EITC) may impact fertility among low-income households. Existing literature finds no effect of the EITC on completed fertility, however. I therefore consider whether the EITC affects a different fertility outcome: birth spacing. If there are economies of scale in childrearing, mothers may reduce space between births to minimize time spent out of the labor market. Close spacing is thought to be detrimental to child health and educational outcomes. To identify the effects of the EITC, I use a new regression discontinuity design (RD) in first child's birth month around the end of the year. Children born before the end of the year can be claimed as dependents on that year's tax returns, substantially increasing EITC eligibility for first time parents. My design incorporates recent evidence that first time EITC eligibility functions as an information shock for many recipients. I find that EITC receipt decreases time to second child by 3-4\%. Effects are concentrated among single mothers (19\% decrease), whereas I find no effects for married mothers or on completed fertility. My findings suggest there may be unintended negative effects of welfare-to-work policies on children in single parent households. Taken together, my findings demonstrate that the design of welfare programs --- including whether or not delivery of benefits is contracted out to private firms --- plays a crucial role in program efficacy, affecting both equity and efficiency concerns. My findings regarding the role of private vendors contrast with traditional economic research on safety net programs, which tends to focus on agency issues between the government and the low income recipients