Benjamin Sommers*

Benjamin Sommers*

Deputy Assistant Secretary for Planning and Evaluation, Office of Health Policy, US Department of Health and Human Services
on leave from Harvard T.H. Chan School of Public Health
*Harvard PhD Program in Health Policy Alumnus & Faculty Member

Dissertation Title: "The Dynamics of Public and Private Health Insurance in the United States"

The dissertation considers the dynamics of three types of health insurance coverage – employer-provided plans, Medicaid, and the Children’s Health Insurance Program (CHIP) – and explores how the pressures of premium growth, swelling ranks of the uninsured, and budgetary shortfalls have affected health coverage for non-elderly Americans over the past decade.

The first chapter focuses on private insurance, presenting an economic model of how employers respond to rapidly increasing premiums. The model attempts to bridge the gap between economists’ view of employer-provided health insurance – which holds that workers bear the full cost of insurance through reduced wages – and the view held by non-economists that insurance costs are a burden on firms. The paper also presents empirical tests that offer support for several of the model’s predictions.

The second chapter explores the retention rates of children in Medicaid and CHIP from 1998-2001, using the Current Population Survey (CPS) March Supplement. I find that 28% of children in Medicaid or CHIP at a given point in time have disenrolled within 12 months. Then, I classify the disenrollment into three categories – loss of eligibility, acquisition of other health insurance, and “drop-out” (children who leave despite continuing eligibility and no other insurance). Overall, I estimate that 1 in 8 children drop out of Medicaid and CHIP each year and become uninsured, despite ongoing eligibility. Multivariate analysis suggests that the clinical encounter plays a key role in keeping children enrolled in public insurance.

The third chapter compares retention in CHIP and Medicaid from 2001-2004, using the CPS’s new CHIP-specific survey item. The results indicate that CHIP has a relative risk of 1.25 for drop-out, compared to Medicaid. Multivariate logistic regression reveals that children with parents on Medicaid are much less likely to drop-out than children whose parents are not enrolled. The causality of this association is supported by instrumental variables regression. This suggests that policies covering children and adults separately are less effective at expanding insurance coverage. I also find that drop-out is 30% lower among children in states with a single combined public insurance program, rather than separate CHIP and Medicaid programs. Overall, I conclude that the United States should move towards more coordinated public insurance programs that target families for coverage rather than individuals.


Graduation Year


Dissertation Committee Member