Dissertation Title:  "The Evolving Dynamics of Health Insurance under State and Federal Reform"

This dissertation consists of three papers on the implications of dynamic changes in health insurance and income for state and federal health care reform. The first paper provides guidance on the optimal design of a health insurance subsidy program under the Patient Protection and Affordable Care Act (ACA). The primary theoretical contribution is a model showing the impact of allocation error, administrative costs and program design on social welfare. In the empirical application, I embed this welfare analysis within a microsimulation model benchmarked against Congressional Budget Office estimates. Overall, I find that given the institutional design of the ACA, the policy that best targets subsidies and optimizes social welfare is one that uses a lagged measure of income based on federal tax returns, but that allows some applicants to have their initial eligibility status redetermined based on current income over a six month period.

The second paper provides the first dynamic look at changes in uninsured spell durations in the wake of Massachusetts' 2006 health care reforms. Using longitudinal data from the Survey of Income and Program Participation (SIPP), I utilize semi-parametric and non-parametric survival methods to estimate changes in uninsurance spell durations. I find evidence that insurance gains in Massachusetts after 2007 were shared broadly among the short- and long-term uninsured. Non-parametric Kaplan-Meier estimates reveal a 3-month decline in the median uninsurance spell durations and a decline of 12 months at the 75th percentile. The results suggest that under health insurance reforms that include an individual mandate, uninsurance becomes largely an acute event.

The third paper provides methodological guidance on the use of household surveys to study insurance transitions. Using data from two household surveys, I non-parametrically estimate the distribution of uninsurance spells in the United States between 2001 and 2007. Using the Medical Expenditure Panel Survey (MEPS) I find median spell durations of 13 months among adults, compared to 8 months in the SIPP. I also show that efforts to reduce seam bias in the SIPP result in higher estimated spell durations over a time period when estimated MEPS spell durations were constant. The results of this study serve a cautionary tale for researchers interested in insurance dynamics, since estimated changes may be the result of survey adjustments and not changes in the population.