`رقابت 'در میان کارفرمایان ارائه بیمه سلامت

Most employees contribute towards the cost of employer-sponsored insurance, despite tax laws that favor zero contributions. Contribution levels vary markedly across firms, and the average contribution (as a percentage of the premium) has increased over time. We offer a novel explanation for these facts: employers raise contribution levels to encourage their employees to obtain coverage from their spouses' employer. We develop a model to show how the employee contribution required by a given firm depends on characteristics of the firm and its work force, and find empirical support for many of the model's predictions.

Since the 1950s, most Americans under 65 have obtained health insurance through their employers (Employee Benefit Retirement Institute, 1995). While it is commonplace today for workers and employers to contribute jointly towards the cost of these plans, there is tremendous variation in these contributions across firms.1 While some firms continue to pay the entire cost of insurance, others require contributions towards individual coverage of US$500 or more, and contributions towards family coverage of US$2500 or more. There has also been an increase in employee contributions over time. As recently as 1980, the majority of employers bore the entire cost of health insurance (Employee Benefit Retirement Institute, 1995). We argue that these patterns of health insurance contributions may be driven, at least in part, by the presence of two-career couples and employer competition to be the `employer not chosen' for insurance. We present a model in which competitive forces lead employers to raise contributions to encourage their workers to switch plans and obtain insurance from their spouses' employers instead.2 We show that the equilibrium employee contribution is sensitive to several critical factors, including the cost of insurance, the percentage of two-career couples, the income tax rate, and the heterogeneity of plans and employee preferences. Using data from the Robert Wood Johnson Foundation Employer Health Insurance Survey, we find empirical support for several key predictions of our model. Our analysis is motivated by a fundamental difference that exists between health insurance and other employee benefits. Namely, family health insurance benefits obtained by different members of a household generally substitute for one another, whereas pensions, life insurance, and other benefits obtained by different members of a household generally augment one another. The implication is that firms can reduce their health insurance costs without necessarily reducing the well-being of their employees by encouraging them to obtain coverage from their spouses' employers. Although we have heard numerous anecdotes supporting our explanation for why employers require employees to make contributions towards health insurance, we have seen no discussion of it in the literature. Morrisey et al. (1994) note that small employers require larger contributions than do large employers but offer no explanation for either the imposition of contributions or the differential. They also report that insurance offered by small employers is less generous on other dimensions besides contribution levels. While our model focuses on contributions, the intuitions that we draw should apply to these other dimensions as well. That is, firms may reduce the generosity of insurance so as to encourage their employees to select their spouse's plans. The savings would offset the foregone tax benefits of providing more generous insurance to those workers who do not switch plans. In the theoretical part of our paper, some employees have low demand for insurance from their own employer because they have the option of obtaining insurance through their spouse's employer. Firms select contribution levels to trade off the tax advantages of employer contributions and the savings from shifting coverage onto other employers. As in the classic economic model of how tax considerations distort individuals' decisions to purchase health insurance, employers may increase contributions to discourage insurance purchases by employees with low demand for insurance.3 Our model differs from this standard model, however, because employer-specific demand is endogenous and depends on the contracts offered in equilibrium. Several important results are obtained. First, employee contributions are lower when fewer employees have working spouses. Second, we show that firms with low insurance premia require disproportionately lower contributions in equilibrium and identify a potential source of welfare loss associated with this divergence in insurance offerings. We are able to test some of the predictions of our theoretical model with data drawn from the Robert Wood Johnson Foundation Employer Health Insurance Survey, conducted by Rand and Westat in late 1993 and early 1994. First, we find that larger firms generally require smaller contributions, especially when measured in percentage terms. This is consistent with our theoretical result that firms that have higher costs of insurance tend to have disproportionately higher contributions. Second, we find that firms with more female workers have higher contributions, firms with more male workers over age 55 have lower contributions, and firms with more covered part time workers have higher contributions. These findings are consistent with our prediction that contributions should be higher in firms with a higher the percentage of working spouses. Simulations of our model show that even small increases in the percentage of dual-worker households can help to account for the rapid increase in contributions that have occurred over the past few decades. There are other factors that have contributed to the growth of employee contributions over time, however, and we do not pretend that our explanation is the main one. One factor is the growth of benefit plans that enable employees to contribute towards premia with before-tax dollars. Between 1988 and 1993, the percentage of medium to large employers offering flexible spending accounts that might enable employees to make contributions with before-tax dollars increased from 13% to 53%. Another explanation may be that as employers offer more insurance options to their employees, they are requiring contributions towards more expensive plans to try to shift employees into the less expensive ones. Although this explanation is consistent with the rise in contributions to individual plans, it does not explain why an increasing percentage of firms require their employees to make non-trivial contributions to even their least expensive plans. Levy (1998) examines the same question. She hypotheses that firms use employee contributions to distinguish between workers with different demands for health insurance, in order to provide optimal compensation when workers are mobile and recruiting workers is costly. Like us, she argues that firms will increase contribution levels so as to discourage workers with low demand for health insurance from purchasing it. In her model, the firm then rewards workers with higher wages so as to discourage them from changing jobs. In our model, firms offer symmetrical wage/benefit packages so that workers are indifferent between jobs. Levy finds some empirical support for her hypothesis, but it does not appear to fully explain variation in contributions in the cross-section or over time. Section 2introduces the basic model of firms with identical insurance costs and derives the basic results. Section 3extends this model to consider heterogeneous firms with different costs of insurance. Section 4provides some empirical support for the model. Section 5offers concluding remarks.

The lay view is that employers raise contributions when `insurance becomes too costly.' In an important sense this is consistent with our model. Firms would like their employees to have insurance. But they would prefer it if they obtained their insurance through their spouse's employer, especially as the price of insurance increases. We have analyzed the competitive interaction between firms as they trade off the tax benefits of generous insurance coverage against the direct savings from shifting coverage onto spouses. The outcome of this interaction is highly sensitive to a number of parameters, including tax rates, spousal work decisions, and interfirm differences in premia. The predictions of our model are consistent with observed patterns in the cross-section. As predicted by the model, firms that have higher costs of insurance tend to have disproportionately higher contributions. Firms whose employees are less likely to have working spouses have lower contributions, which is also consistent with our model. Although we develop the model in terms of premia and contribution levels for a given insurance plan, our findings may generalize to the choice of plan and decision to offer a plan. For example, our findings suggest that if small firms face slightly higher premia, they may disproportionately offer poor or no insurance benefits as a way to encourage workers to select their spouse's employer's plan. This is consistent with the findings of Morrisey et al. (1994) and Cantor et al. (1995) that show that small firms offer less generous insurance than do large firms, and the finding of Liebowitz and Chernew (1992) that states small premia differentials may cause a large percentage of small firms to drop coverage. Our model also identifies several factors that may have contributed towards the rapid increase in contributions over the past two decades. Our model suggests that contributions should increase when: 1. The price of insurance is high relative to the idiosyncratic differences in taste for insurance; 2. The tax rate is low; and 3. The fraction of workers with working spouses is high The sharp rise in insurance premia during the past two decades suggest that (1) is more likely to hold today than during the 1970s. At the same time, the marginal income tax rate is lower than during the 1970s.19 Finally, the fraction of workers with working spouses has steadily risen.20 All of these trends are consistent with rising contribution.