Essays in Labor Economics: Empirical Methods and Evaluations of Social and Employment Programs

Abstract

This dissertation consists of three essays in labor economics, which are motivated by underappreciated but important policy rules in social and employment programs. The essays attempt to use these rules to evaluate program effects, inform policy design, and study methodological issues that arise in the process.

The first chapter, "Dynamic Opting-in Incentives in Income-tested Social Programs: Evidence from Medicaid/CHIP", examines families' responses to the continuous eligibility provision in Medicaid/CHIP and proposes a framework to evaluate the optimal eligibility recertification frequency. Conventional studies of labor supply in the presence of income-tested social programs implicitly assume that income eligibility for program participation is constantly monitored by the government. However, this is not how most of these programs operate in practice, and the time until the next eligibility recertification can be as long as a year. In particular, the Balanced Budget Act of 1997 gives states the option of insuring children in their Medicaid/CHIP program continuously for up to 12 months regardless of changes in family income. The long recertification period in effect increases the size of the benefit notch, and neoclassical labor supply models predict that agents may lower their labor supply before the application month to gain program eligibility and then increase their labor supply until the next eligibility check. I use the 2001 and 2004 panels of Survey of Income and Program Participation (SIPP) to empirically examine the income and labor supply responses of parents whose children are publicly insured. Comparing theoretical predictions and the empirical evidence points to little labor supply response. Given the absence of strategic behavior, I propose a simple framework to compute the optimal length of the continuous eligibility period relying on the mechanical properties of the income processes observed in SIPP, and derive a mapping from the recertification cost parameters to the optimal monitoring frequencies.

In the second essay, "Regression Discontinuity Design with Measurement Error in the Running Variable", I extend the regression discontinuity (RD) design to allow for the running variable to be mismeasured. The need for the method arises, for example, when a researcher tries to apply an RD design with noisy income measures from survey data to evaluate the effects of income-tested social programs using discontinuities in eligibility rules. This paper provides sufficient conditions to non-parametrically identify the true running variable distribution and RD treatment effect when the measurement error is independent of the true running variable. A simple estimation procedure is proposed based on a minimum distance formulation, and the resulting estimators are root-N consistent, asymptotically normal and efficient. Simulations show that the procedure is informative for typical sample sizes encountered in relevant empirical studies.

In the third essay, "Quasi-Experimental Identification and Estimation in the Regression Kink Design", co-authored with David Lee and David Card, we consider nonparametric identification of the average marginal effect of a continuous endogenous regressor in a generalized non-separable model when the regressor of interest is a known, deterministic, but kinked function of an observed continuous assignment variable. This design arises in many institutional settings where a policy variable of interest (such as weekly unemployment benefits) is mechanically related to an observed but potentially endogenous variable (like previous earnings). We characterize a broad class of models in which a "Regression Kink Design" (RKD) provides valid inferences for the underlying marginal effects. Importantly, this class includes cases where the assignment variable is endogenously chosen. As in a regression discontinuity design, the required identification assumption implies strong and testable predictions for the pattern of predetermined covariates around the kink point. Standard local linear regression techniques can be adapted to obtain "nonparametric" RKD estimates. We illustrate the RKD approach by examining the effect of unemployment insurance (UI) benefits on the duration of benefit claims, using rich microdata from the state of Washington. We find that a 10 percentage point increase in the UI replacement rate leads to a 1.6 week increase in the duration of insured unemployment, which is in the higher range of magnitudes found in the existing literature.