Health Insurance Subsidies, Coverage, and Costs

One of the most serious challenges facing the U.S. health care system is the problem of the uninsured. Forty-one million Americans, or nearly 15% of the non-elderly population, currently lack health insurance coverage. This lack of coverage has real consequences in terms of both access to care and health outcomes. A recent report by the Kaiser Family Foundation found that "the uninsured receive less preventative care, are diagnosed at more advanced disease stages, and once diagnosed, tend to receive less therapeutic care and have higher mortality rates.⁽¹⁾

One frequently proposed remedy is to subsidize the premiums that employees are charged for employer-provided health insurance. Roughly one-quarter of the uninsured have access to insurance through their own job or that of a family member but decline to take it up; in fact, most of the decline in insurance coverage over the past two decades results from decreases in the rate of take-up by employees. If these individuals are price-sensitive in their take-up decisions, offering premium subsidies may encourage them to enroll in employer-provided plans, decreasing the number of uninsured.

However, premium subsidies may be costly. As a practical matter, it is difficult to offer subsidies solely to uninsured workers. But, among those who are offered insurance, only about 7% are uninsured. So it is very costly to provide subsidies to all workers in an effort to increase coverage among that small share that is uninsured. Moreover, subsidies may encourage insured employees to choose more expensive plans by shielding them from the full cost of upgrading.

In Subsidies to Employee Health Insurance Premiums and the Health Insurance Market (NBER Working Paper 9567), Jonathan Gruber and Ebonya Washington estimate the effect of premium subsidies on the take-up and costs of employer-provided health insurance. The appeal of premium subsidies as a partial solution to the problem of the uninsured will depend on the magnitude of these behavioral responses.

Previous research on the topic has used differences in premiums across firms to examine the link between premiums and take-up. But estimates based on such differences are suspect. If, for example, workers who like insurance lobby their firms for a low employee premium share, then the observed relationship between premiums and take-up will overstate the true effect of premiums on take-up.

The authors exploit a major rule change for the Federal Employees Health Benefit Plan, which allowed postal employees to pay premiums on a pre-tax basis starting in 1994 and all other federal employees to do so starting in 2000. This rule change lowered employees' premiums substantially - for example, a middle income family could see their employee premiums fall by 45%. Moreover, the sequential timing of the change created large differences in premiums paid by federal workers depending on the year and whether they were postal employees. As these differences are likely to be unrelated to worker preferences for insurance, they provide a compelling means to estimate the true effect of premium changes.

The authors construct a unique data set of personnel records for federal employees for the years 1991-2002 using information obtained from the Postal Service and the Federal Office of Personnel Management through a Freedom of Information Act request. They also calculate the employee share of insurance premiums, accounting for whether premiums are paid on a pre-tax or post-tax basis.

The authors find that a decrease in the employee share of premiums is associated with a increase in take-up of family coverage, though the effect is modest - a 10% decrease in the employee share increases take-up by only about 0.2%. A decrease in the employee share of premiums is associated with a decrease in the take-up of individual coverage, which the authors suggest may occur because people switch from individual to family coverage as the subsidy rises.

Next, the authors explore the relationship between premiums and the cost of plans chosen by employees, and find that a decrease in employee premiums is associated with a increase in costs. On average, the selection of more expensive plans represents only 2.5% of the premium decrease, though the employer's additional costs are two to three times as large.

Finally, the authors use their results to simulate the cost of the 2000 rule change for non-postal employees, which they estimate to be between $31,000 and $83,000 per newly insured worker. Although a more targeted subsidy program would be somewhat less expensive, the authors conclude that offering premium subsidies is unlikely to be a cost-effective approach to address the problem of the uninsured.

This research was supported by the Economic Research Initiative on the Uninsured at the University of Michigan and by the Commonwealth Fund. It was summarized by Courtney Coile.

1. "Not Covering the Uninsured," Kaiser Family Foundation, June 2003.