Federal Reserve Bank of San Francisco

FRBSF Economic Letter

1998-12 | April 17, 1998

More Economic Letters

Health Insurance and the U.S. Labor Market

Tom Buchmueller and Rob Valletta

Health insurance in the United States largely is employment-based: nearly 90% of Americans with private insurance are covered through employer-provided plans. Furthermore, health insurance is the largest nonwage component of total compensation, accounting for 34% of expenditures on voluntary employee benefits and 7% of total compensation (U.S. BLS 1994). Combined, these facts suggest that employer provision of health insurance is an important feature of the U.S. labor market. In this Economic Letter, we describe some unique aspects of the market for employer-provided health insurance, and we discuss their implications for several key employee decisions: job mobility, retirement, and hours worked. We conclude by discussing the policy implications of recent research findings.

The economic rationale for employer-provided health insurance

Although the earliest employer-sponsored health plans date to the 1920s, two public policies from the 1940s and 1950s firmly established the link between health insurance and the workplace. First, during World War II Congress responded to excess demand for labor by enacting limits on the extent to which employers could increase wages. Since these limits did not apply to fringe benefits, many employers began offering health insurance to attract and retain workers. Second, in 1954 the IRS created a permanent incentive for employers to substitute in-kind benefits for cash wages by declaring that fringe benefits are not taxable. Although workers ultimately pay for health insurance, either through direct premium contributions or reduced wages, this tax treatment provides a powerful incentive for workplace provision of health insurance.

Another incentive is the cost savings associated with risk pooling and minimizing “adverse selection.” In risk pooling, insurers diversify their risk by spreading it across a large number of individuals; “adverse selection” refers to the tendency for individuals with the highest expected health care costs to be the most likely to buy insurance. Given these factors, insurers can reduce costs by selling insurance to groups of people that are formed for reasons other than purchasing insurance, such as all the workers at a single large firm.

Despite these incentives for tying health insurance to employment, doing so can affect other job-related decisions by workers. In the rest of this Economic Letter we discuss the influence of employer-provided health insurance on several key labor market decisions, and recent policy initiatives aimed at reducing these behavioral side effects..

Job mobility

“Job lock” is one effect of employer-provided health insurance that has received a great deal of attention from the media, policymakers, and academic economists. Job lock is defined as a reduction in workers’ willingness to quit jobs (reduced voluntary job mobility) arising from the risk of losing health coverage. This effect refers not to reduced mobility arising from the monetary value of health insurance, but rather to any additional reductions in mobility arising from risk and information problems associated specifically with insurance markets. For example, an important source of job lock is “pre-existing condition” clauses, which insurers often invoke to deny full employment-based medical coverage to a new employee who has a medical condition at the time of job change.

Workers without pre-existing medical conditions also may experience job lock. The risk of being uninsured between jobs raises expected search costs, thereby reducing workers’ willingness to endure the unemployment that may be required for optimal job search. Moreover, employer-sponsored health insurance plans often impose probationary periods of up to several months, during which new employees are denied full health coverage, and job changers may lose credit toward deductibles and out-of-pocket expenditure limits from their previous plans. Each of these factors increases the expected cost of job change and perhaps reduces voluntary mobility.

Measuring the extent of job lock is complicated by the tendency for insurance-providing jobs to be “good jobs”–i.e., to be highly desirable for a number of reasons besides the provision of health insurance. To overcome this problem, several studies (including Buchmueller and Valletta 1996) have used a quasi-experimental design to investigate the extent of job lock. These studies compared the job change behavior of individuals who are covered only under their own plan (and therefore are subject to job lock) with that of individuals who also are covered through a spouse’s plan (and therefore are not subject to job lock). Our results suggested that job lock induced by employer-provided health insurance substantially reduces voluntary job mobility for married women; the results for married men were not statistically significant. Although a similar test can not be applied to unmarried workers, we applied an alternative test to this group and once again found more evidence of job lock for women than for men. The estimated reductions in female job mobility were large, on the order of 30–50%.

Additional evidence is provided by examining the impact of state “continuation of coverage laws,” which specify that separated employees be allowed to purchase group coverage from their former employers at a comparable rate for up to 18 months after separation. Gruber and Madrian (1997) report that such laws raise voluntary turnover rates for men with health insurance by 14%, which suggests the presence of substantial job lock in the absence of such laws. They also found that, although unemployment spells are longer in continuation of coverage states, wage gains after unemployment are larger. This suggests that the relaxation of job lock constraints enhances the quality of the job search and matching process, which is a key rationale for continuation of coverage laws.


Retirement is another type of job change decision that is affected by health insurance availability. The availability and cost of health insurance is a critical issue for retired and older workers, because both the magnitude and variability of medical expenses increase substantially with age (Gruber and Madrian 1995). As a result, the cost gap between employer-provided group insurance and equivalent individually purchased private insurance also increases with age.

Gruber and Madrian investigated the relationship between health insurance and retirement by estimating the effects of state continuation of coverage laws on the timing of retirement. The ability to purchase post-retirement health insurance at a rate comparable to that under a former employer’s plan may lead to earlier retirement. This effect is limited, however, by the duration of continuation coverage (18 months) and availability of alternatives (primarily employers’ post-retirement health plans, and Medicare at age 65).

Using data from the 1980s, Gruber and Madrian found that by acting as a “bridge to Medicare” the availability of continuation coverage substantially increases retirement probabilities for men aged 55–64; in particular, each additional year of coverage raises the probability of retirement in a given year by 30%. Their results also suggest that each year of post-retirement insurance coverage is valued at $13,600, a figure that is 3–4 times greater than the implied cost differential between employer-provided and individual private insurance. Although this figure is large, it may be explained by an inability of many retirees (particularly those with pre-existing conditions) to obtain individual insurance at all.

Labor supply

The retirement decision also can be viewed as a labor supply decision; it represents older workers’ decision to supply zero hours of labor. However, employer-provided health insurance is likely to have more general effects on the hours worked decision. Because employers seldom offer health insurance to part-time workers, employees in general must work full-time in order to acquire health insurance.

There are two key reasons why employers often limit health insurance to full-time workers. First, health insurance benefits represent a “quasi-fixed” cost that varies with the number of workers rather than the number of hours; on an hourly basis, this makes health insurance more expensive to provide to part-time workers. Second, IRS non-discrimination rules limit the ability of firms to offer different benefit packages to different groups of full-time employees but allow the differential treatment of full-time and part-time employees. Part-time employees are more likely to be low-wage workers, and for some low-wage workers health benefits are likely to be a larger fraction of total benefits than they prefer. Thus, excluding part-time workers from fringe benefit plans is one way that employers can exclude low-wage employees from their health plans without violating IRS non-discrimination rules.

The restriction of health benefits to full-time workers, combined with the high value of such benefits, suggests that some workers who prefer to work part-time may instead work full-time in order to acquire health insurance for themselves and their families. Such behavior is rare among male household heads, since the overwhelming majority of them work full-time for reasons unrelated to insurance provision. However, married women whose husbands do not have insurance may engage in such behavior on a widespread basis.

In a recent paper (Buchmueller and Valletta 1999) we tested this proposition empirically, using 1993 household survey data and a technique that accounts for both observable and unobservable differences between families defined by husbands’ health insurance status. We found evidence of frequent switching by wives from non-participation or part-time work to full-time work in order to acquire health insurance. In numerical terms, the estimates indicate a 15–36% increase in wives’ labor supply associated with a lack of insurance coverage through their husbands.

Policy implications

In the past decade or so, state and federal policymakers have passed several laws that make it easier for workers who leave insurance-providing jobs to maintain their coverage. For example, in 1996 Congress passed the Health Insurance Portability and Accountability Act (HIPAA), which builds on both COBRA (Consolidated Omnibus Reconciliation Act) and the state insurance reforms to further reduce the potential loss of insurance associated with changing jobs. In general, HIPAA enhances access to coverage by limiting insurer use of pre-existing condition clauses and screening based on genetic or health factors, and by preventing insurers from refusing coverage to individuals previously covered under a group plan who have exhausted their COBRA coverage extensions (although insurers have wide latitude in setting rates charged to individuals covered by HIPAA).

The evidence from the job lock and retirement studies cited above suggests that these incremental insurance reforms have a substantial impact on worker behavior. In particular, both job change and early retirement are increased substantially by continuation of coverage laws. The exact economic welfare effects of such changes have not yet been worked out. However, continuation of coverage policies may improve economic efficiency by reducing the distortionary impact of health insurance market characteristics on labor market decisions. For example, Gruber and Madrian’s (1997) finding that wage gains resulting from job change are larger in continuation of coverage states is consistent with this interpretation.

On the other hand, continuation of coverage laws enable older workers to remain on their employer’s plan after retirement, without any change in cost. Because health care costs increase with age, this generates economic inefficiency in the form of a subsidy from younger workers to older workers. However, a recent policy initiative proposed by the current presidential administration would enable early buy-ins to the Medicare system for individuals aged 62–64 and displaced workers aged 55 and older, and continued coverage for retirees aged 55–64 whose former employers have dropped their retiree health coverage. This policy would reduce the subsidy from the young to the old by requiring retirees to pay either Medicare premiums or a higher premium than do other COBRA participants, although it has the likely drawback of requiring transfers from general tax revenues for financing.

Finally, the labor supply analysis in Buchmueller and Valletta (1999) suggests that universal coverage reform, although it may be desirable for other reasons, may have unintended side effects. In particular, our results indicate that policies that remove the linkage between health insurance provision and full-time work are likely to reduce hours worked by married women; this may be problematic if coverage is financed through a payroll tax. The potential for such unintended side effects of policy reform suggests a need for additional research on the linkages between the markets for health insurance and labor.

Tom Buchmueller
Assistant Professor, UC Irvine

Rob Valletta
Senior Economist


Buchmueller, T.C., and R.G. Valletta. 1996. “The Effects of Employer-Provided Health Insurance on Worker Mobility.” Industrial and Labor Relations Review 49(3), pp. 439-455.

______, and ______. 1999. “The Effect of Health Insurance on Married Female Labor Supply.” Forthcoming, Journal of Human Resources.

Gruber, J., and B.C. Madrian. 1995. “Health-Insurance Availability and the Retirement Decision.” American Economic Review 85(4) pp. 938-948.

______, and ______. 1997. “Employment Separation and Health Insurance Coverage.” Journal of Public Economics 66, pp. 349-382.

U.S. Bureau of Labor Statistics. 1994. Employment Cost Indexes and Levels, 1975-94. Washington, D.C.

Opinions expressed in FRBSF Economic Letter do not necessarily reflect the views of the management of the Federal Reserve Bank of San Francisco or of the Board of Governors of the Federal Reserve System. This publication is edited by Sam Zuckerman and Anita Todd. Permission to reprint must be obtained in writing.

Please send editorial comments and requests for reprint permission to
Research Library
Attn: Research publications, MS 1140
Federal Reserve Bank of San Francisco
P.O. Box 7702
San Francisco, CA 94120