The University of Michigan
555 South Forest Street
Ann Arbor, MI 48104-2531
SUBSIDIES HAVE SMALL EFFECT ON COVERAGE
Federal employees gained
the ability to pay health insurance premiums with pre-tax dollars,
with the subsidy extended to postal employees in 1994
and then to remaining employees in 2001. A study by Jonathan Gruber and Ebonya Washington, funded by the Economic Research Initiative on
(ERIU), takes advantage of this natural experiment to estimate the
effects of premium subsidies on coverage rates among federal employees.
Small Effects of Premium Subsidies
Worker responses to the subsidy suggest
an elasticity of take up with respect to premiums of roughly –0.02,
similar in scale but lower than the response found in other studies.
Large Effects of Premium Subsidies on Costs
Simulations of the effect of the policy change on the number insured show the number of newly insured persons ranging from 8,261 to 17,954. The range reflects assumptions about what share of newly insured were previously uninsured (50 percent at the low end; 100 percent at the high end.) Total costs are $693 million, of which $647 million represents the revenue loss associated with allowing employees to pay premiums with pre-tax dollars, resulting in yearly costs per newly-insured person that range from $31,000 to $83,000.
Small Effects of Premium Subsidies on Choice of Health Plan
In contrast to previous work, federal employees appear less likely to switch health plans in response to a change in the employee share of premium. Such choices lead employees to offset only about 2.5% of the additional premiums.
This paper suggests that tax credits will not be very effective at increasing coverage rates among the uninsured who work for firms that offer coverage, and may be very costly to the federal government.
An important caveat is that this work is based on an analysis of federal workers, who may differ from other workers in important ways. Earnings for federal workers in the sample were on average $45,700 in 2002 compared to an economy-wide average of $30,700. Federal workers also have relatively high baseline takeup of offered coverage (87 percent). Because of these and other differences, the responsiveness of federal workers to the subsidy may differ from that of other uninsured workers, such as low income workers. Similarly, the price responsiveness of these federal workers may differ from that of workers currently not offered coverage.
Also, the nature of the intervention studied is a subsidy generated by a tax code change rather than an explicit policy intended to expand coverage. A subsidy program that is more visible and designed to increase coverage may be more effective.
Finally, important variables such as the availability of spousal coverage and family income were imputed because of data limitations. The imputation schemes and associated measurement error may have created a downward bias in the estimates. Despite these caveats, it is important to note that these findings are consistent with a large body of previous literature, examining different populations and using different methods, that concludes that responsiveness of uninsured individuals to subsidies is low.
The primary source is administrative data on federal employees, ages 21 to 64, for 1991to 2002. Approximately one-third of the sample is postal workers. The sample is restricted to full-time, health insurance-eligible workers, resulting in a sample of 27 million observations, collapsed into state/year/income decile/sex/postal v. non-postal cells, for 24,480 observations. Data from Current Population Survey (CPS) are used to impute demographic and income variables. Data from Statistics of Income (SOI) are used to impute average amount itemized, contingent on itemizing. NBER’s TAXSIM model is used to calculate marginal tax rates.
Imputations from CPS and SOI are attached to each person under four possible states (since actual type unknown in employee data): married, non-itemizer; married, itemizer, single, non-itemizer; single, itemizer. Marginal tax rates are calculated for each person in each state, then a weighted average depending on the odds of being in each state. Individual data are collapsed into state/year/income decile/sex/postal-non-postal cells. The dependent variable (premium or take up rate) in each cell is regressed on the after-tax share of premiums paid by each employee, cell level characteristics for such variables as income and age, and fixed effects for income decline*sex, postal v. non-postal worker, state, and year. The remaining, identifying variation is across income-sex groups, postal/nonpostal groups, states, and years.
Subsidies to Employee Health Insurance Premiums and the Health Insurance Market.
Jonathan Gruber and Ebonya Washington, MIT
Conference paper presented at ERIU Research Conference, July 2003
NBER Working Paper #9567
The final version of the paper appeared as:
Gruber, Jonathan and Washington, Ebonya. “Subsidies to Employee Health Insurance Premiums and
the Health Insurance Market.” Journal of Health Economics, 2005: 24(2): 253-276.
ERIU Working Paper #15 (Adobe PDF)
Back to top
Funded by The Robert Wood Johnson Foundation, ERIU is a five-year program shedding new light on the causes and consequences of lack of coverage, and the crucial role that health insurance plays in shaping the U.S. labor market. The Foundation does not endorse the findings of this or other independent research projects.