بیمه درمانی و نتایج بازار کار : تصمیم گیری مشترک در خانواده

Most Americans obtain access to health insurance through an employer. In this paper, we ask how the link between health insurance and employment affects labor market choices such as whether to work full-time. To understand the effect of the incentives embedded in the employer-based insurance system, we study the joint decision-making of husbands and wives that determines the household's access to health insurance. We estimate the effect on a wife's (husband's) labor market outcomes of husband's (wife's) health insurance, allowing the health insurance of both spouses to be endogenous. Obtaining unbiased estimates of such effects is complicated by the likelihood that positive assortative mating creates correlations between a couple's characteristics and the possibility that there are important unobservable household income effects. Our innovation is to measure these biases by examining a second fringe benefit, paid sick leave, in addition to health insurance. We find that, as predicted, spouse's insurance has statistically significant negative effects on being offered own employer insurance as well as on the probability of working full-time with health insurance.

In the United States, approximately 64% of non-elderly Americans obtain their health insurance through an employer (www.statehealthfacts.kff.org, 2004). Employer group insurance is generally less costly, is often simultaneously more generous than private insurance obtained in the non-group market, and enjoys a tax advantage relative to insurance not purchased through an employer. However, these advantages may well be offset by some less obvious welfare losses resulting from the link between access to health insurance and employment if labor market decisions are influenced by the access to health insurance that some jobs convey. In this study, we ask how an individual's labor market choices are affected by the link between employment and health insurance. One way of gaining access to employer insurance when it is not available through an individual's own job, or when he or she is not employed, is through a working spouse. In an earlier study we found that having two earners in a household substantially mitigates the negative effect on access to health insurance of workers in part-time jobs, workers in small establishments, and self-employed workers (Abraham and Royalty (2005)). In our sample, 43% of wives whose husbands were not offered employer insurance were offered insurance. Among wives without own employer insurance, 65% had husbands who were eligible for insurance at the workplace. Knowing that a large proportion of married adults who do not have their own access to coverage have access through a working spouse does not tell us, however, whether this is merely fortuitous or whether these couples have sorted themselves systematically into such arrangements.1 Knowing the extent to which labor market outcomes depend on each partner's access to coverage is critical to answering many current policy questions. For example, if couples sort themselves into jobs with and without health insurance, it will affect the demand for insurance by workers and therefore the likely effectiveness of policies designed to increase insurance coverage by encouraging employers to offer insurance in jobs where they have not done so historically. Understanding these joint decisions will also help us to identify how much health insurance drives other labor market decisions when access to health insurance depends on the choice made. If workers are locked into full-time jobs or choose to join the labor force because of employer health insurance, our system of employer-based insurance may produce large welfare losses. We investigate how one spouse having access to employer insurance influences the probability that the other spouse also will be offered employer insurance. We then look at the effect of having access to health insurance through a spouse on decisions about hours of work. The innovation of the paper is to allow the health insurance of both spouses to be endogenous, controlling for bias due to assortative mating or income effects by looking at a second fringe benefit, paid sick leave. This allows us to examine how much behavior changes as married couples take advantage of the flexibility provided by being eligible for one another's health insurance and, conversely, by how much labor market outcomes may be distorted by having health insurance tied to employment when no other access to insurance is available.

In this study, we have investigated how much behavior changes as married couples take advantage of the flexibility provided by being eligible for one another's employer health insurance and, conversely, by how much labor market outcomes may be distorted by having health insurance tied to employment when no other access to insurance is available. One key challenge in examining the joint decision-making of husbands and wives is finding valid instruments for spouse's health insurance. Though a worker's age and education are good predictors of having an insurance offer, using these measures in an instrumental variables (IV) framework may still lead to biased estimates if positive assortative mating or unmeasured income effects are present. To estimate the direction and size of the bias due to assortative mating and income effects, we look at a second fringe benefit, paid sick leave, in addition to health insurance. Under certain assumptions, using the sick leave estimates we can identify the bias due to assortative mating and income effects and then net out these effects to obtain the behavioral effect of spouse's insurance on own insurance. As hypothesized, we find that spouse's insurance has statistically significant negative effects on being offered own employer insurance. In all cases, we see that a greater probability that one has access to insurance through a spouse reduces the probability of labor market outcomes associated with employer insurance such as full-time work. Or, when addressing the possibility of labor market distortions caused by our employer health insurance system, perhaps we should state the result differently: a lower probability of access to health insurance through a spouse increases the probability of particular labor market outcomes, including having one's own access to employer insurance and working full-time. The relationship is significant for both men and women. Another pattern that holds consistently across models is the gender difference in the bias term estimated in the sick leave equation. In every case, the bias for women is negative while the bias for men is positive. We interpret this to mean that the bias due to income effects outweighs any positive assortative mating effect for women while the opposite is true for men. This can be explained if women's labor market decisions are more sensitive to household income than men's. Knowing the extent to which labor market outcomes depend on each partner's access to coverage can provide important insights for assessing the potential effectiveness of policies designed to increase access to coverage, either through employer incentives to offer insurance in jobs where typically they have not done so or through premium subsidies to employees. If, as we find, married households actively sort into jobs with and without health insurance knowing that only one source of employer coverage is needed to ensure access, then part-time workers who become eligible for coverage through a policy intervention, may not exhibit strong demand for insurance. On the other hand, our results imply that other access to insurance is likely to increase part-time work. Therefore, if insurance became more widely available in part-time jobs, some workers currently working in full-time jobs for the sake of health insurance may switch jobs. These workers would be expected to have a higher demand for insurance than workers in those jobs who are already eligible for an alternative source of insurance. Our results also imply some welfare losses due to the link between health insurance and employment. We find that individuals are more likely to work in jobs that offer insurance and to work full-time when the probability of having insurance through a spouse is lower. If employer insurance were offered more widely or if insurance were available through non-employer sources on similar terms, individuals would make some different choices about their employment.