Labor Supply Response to Income Cutoffs of Health Insurance in the Massachusetts Reform

In the U.S. health care reform created by the Affordable Care Act (ACA), low- and middle-income households, who are eligible to purchase health insurance plans in the Exchanges, will receive tax credits as subsidies to pay for premiums. The subsidized program not only will have direct impact on expanding insurance coverage, but also may create distortion in labor supply markets as well as other incentive effects. Previous literature has provided numerous evidences on behavioral response to programs that generate non-linear budget sets for consumers' consumption, such as Medicaid, Income, and Social Security. The population response also has policy significance in projecting the number of eligible enrollees, as well as the size of government funding for the subsidy.

In this paper, I model and then estimate the behavioral response of income manipulation to a subsidized insurance program in the Massachusetts health reform. The subsidy has a piece-wise structure that several income cutoffs sort eligible participants into different income tiers. Households who fall in higher tiers receive fewer subsidies and pay more out-of-pocket premiums. Therefore those with income slightly above the cutoffs have strong incentives to lower their incomes in order to be eligible for greater subsidies.

Income manipulation is modeled by the Regression Discontinuity (RD) approach developed by McCrary (2007). Using data from the American Community Survey (ACS), I find clear evidence suggesting a causal effect of the subsidy schedule on households’ income manipulation: there are income discontinuities around and only around the cutoffs generated by the subsidy schedule. Among the four cutoffs created by the subsidy schedule, the results show discontinuity around two cutoffs: 150 percent of Federal Poverty Level (FPL), and 300 percent of the FPL. The former is the cutoff between plans with zero and non-zero out-of-pocket premiums, and the latter between plans with the largest subsidy difference. The results indicate that the population responds actively when they face choices between free and non-free plans, or when the benefit difference is large enough. By exploring the effects in subgroups, I find the manipulation around the cutoff of 150 percent is contributed by the self-employed, and the income manipulation around the cutoff of 300 percent of FPL is concentrated among wage-workers.

I then construct a model based on the method introduced by Saez (2010) to estimate labor supply elasticity with respect to wage rate, using the evidence of income discontinuity. The labor supply elasticity with respect to wage rate is around 0.01. The standard deviation of income is around 0.1% in terms of FPL, which implies the standard deviation of income is $10 for a single person with income about $10,000. Both estimates are very small, and one possible reason is that the income manipulation larges comes from households with income right above the cutoffs, but not those with relatively high income. I also estimate the welfare impact of the behavioral response due to the subsidized program in the Massachusetts reform. I find that the welfare loss, measured as the change of income, is about $2.78 million.