پوشش بیمه سلامت و جستجوی خانگی

Health insurance in the United States is typically acquired through an employer-sponsored program. Often employees offered employer-provided health insurance have the option to extend coverage to their spouse and dependents. We investigate the implications of the “publicness” of health insurance coverage for the labor market careers of spouses. The theoretical innovations in the paper are to extend the standard partial–partial equilibrium labor market search model to a multiple searcher setting with the inclusion of multi-attribute job offers, with some of the attributes treated as public goods within the household. The model is estimated using data from the Survey of Income and Program Participation (SIPP) using a Method of Simulated Moments (MSM) estimator. We demonstrate how previous estimates of the marginal willingness to pay (MWP) for health insurance based on cross-sectional linear regression estimators may be seriously biased due to the presence of dynamic selection effects and misspecification of the decision-making unit.

Health insurance in the United States is typically acquired through an employer-sponsored program. Even though health insurance can be purchased through private markets, the cost is considered prohibitive in comparison with the effective cost of purchasing health insurance though an employer. There are many possible reasons for this difference, such as tax subsidies to firms who offer such insurance to their employees, risk-pooling among a large group of relatively healthy individuals (i.e., individuals employed at a given firm), or sharing of a cost (health insurance) that improves the quality of the employment match to both sides of the contract (e.g., Dey and Flinn (2005)). Another empirical regularity regarding health insurance purchase and coverage is that in households in which both husbands and wives work health insurance is often only purchased (through their employer) by one of the spouses. Apparently this reflects the fact that health insurance is largely a public (household) good in that most employers who offer health insurance to their employees also include the option to cover spouses and dependent children. In this research our goal is to investigate the implications of the “publicness” of health insurance coverage for the labor market careers of spouses and the cross-sectional distribution of wages and health coverage statuses of spouses. We use a relatively innovative household search framework to address this question. A large empirical literature exists on the relationship between health insurance coverage and wage and employment outcomes, though most of it is formulated at the individual level; reasonably comprehensive surveys can be found in Gruber and Madrian (2001) and Currie and Madrian (1999). The research objective in these studies is almost invariably the estimation of a distribution of marginal willingness to pay (MWP) parameters characterizing the population, and the framework is that of compensating differentials. When a formal modeling framework is developed, it is a variant of a static labor supply model, with reference made to household rather than individual choice on rare occasions. This is a questionable choice given the great deal of concern in this literature with assessing the impact of employer-provided health care coverage on job mobility. Dey (2001) and Dey and Flinn (2005) take the position that to analyze mobility behavior requires a dynamic model with labor market frictions, which led them to employ a search framework with both unemployed and on-the-job search. Estimates from the equilibrium matching-bargaining model in Dey and Flinn (2005) led them to conclude that the productive inefficiencies resulting from the employer-provided health insurance system were not large. The conclusions drawn from all of these empirical studies may be questioned due to their focus on individual rather than household behavior.1 A few attempts have been made to look at the impact of the health insurance coverage of a spouse on the other’s employment probability. For example, Wellington and Cobb-Clark (2000) estimate that having an employed husband with a job covered by health insurance reduces a wife’s probability of employment by 20%. However, their econometric model does not allow for simultaneity in these decisions, labor market frictions, and does not condition on the husband’s wage rate. To understand the distribution of health insurance and wages across spouses and households, it is necessary to formulate a more appropriate framework for the analysis. To simplify the modeling and estimation problem, and to promote comparability with previous analyses, we adopt a very simple specification of household behavior. We assume the existence of a (instantaneous) household utility function in which consumption and health insurance coverage are additively separable. The subutility function associated with consumption is a quasi-concave function of (instantaneous) household income, and the instantaneous payoff if at least one of the spouses has employer-provided health insurance is ξξ. Learning the parameter ξξ, are something analogous to it, seems to be the goal of most empirical attempts to estimate the marginal willingness to pay (MWP) for health insurance coverage. Two important contributions make clear the perils of attempting to infer tastes from cross-sectional relationships generated by dynamic choices among jobs offering different combinations of utility-yielding characteristics. Hwang et al. (1998) make the point using the equilibrium search framework of Burdett and Mortensen (1998), and Gronberg and Reed (1994) provide an empirical example by estimating a MWP parameter within a compensating differentials model using job duration data from the National Longitudinal Survey of Youth 1979. The point of both of these studies is to illustrate how the cross-sectional relationship between wages and job characteristics is determined by the primitive parameters characterizing the search equilibrium. The cross-sectional “trade-off” between wages and health insurance coverage, for example, is an extremely complicated function of ξξ and the parameters characterizing the labor market environments of the spouses. In general, the only way to consistently estimate ξξ is to simultaneously estimate all model parameters, a path that we follow in this paper. The contributions of this paper with respect to those mentioned in the previous paragraph are (1) the extension to a multiple agent setting in which job attributes have a public goods aspect and (2) estimation of the behavioral model. We provide a lengthy discussion regarding the challenges of estimating a multiple agent model in continuous time given the discreteness of the data to which we have access. We use the method of simulated moments (MSM) in conjunction with data from the Survey of Income and Program Participation (SIPP) to estimate the model parameters. We find evidence that utility is a concave function of instantaneous income and that there is a positive valuation of health insurance coverage by the household. We show that this estimate is sensitive to the specification of the instantaneous utility function, as is to be expected. Our estimates of the preference parameter ξξ vary widely depending on the moments included in the implementation of the MSM estimator. Under our preferred specification, which includes “cross moments”, (which are functions of the labor market outcomes of both spouses and not only of one), we find a high estimated value of ξξ, indicating the possibility of important welfare gains in the population if the health insurance coverage rate can be significantly increased. In a section directed to policy makers, we discuss how our estimation results cast doubt on previous attempts to infer the MWP using linear regression approaches with cross-sectional data. Using our data and estimates, we demonstrate how the difference in mean earnings between those with and without employer-provided health insurance bears a complicated relationship to ξξ and the other parameters describing the labor market environment. We also show that, even with a constant valuation of health insurance coverage in the population, cross-sectional estimates of the MWP will appear to be heterogeneous. In our framework, this phenomenon arises through the omission of relevant state variables (the labor market status of the spouse) that vary in the population in the steady state. These results point to the necessity of looking at the valuation of the health insurance coverage in the household context. The plan of the paper is as follows. In Section 2 we develop the model of household search using a household utility function approach. Section 3 contains an analysis of the implications of the model under various specifications of the household utility function. Section 4 includes a discussion of the data source and presents some descriptive statistics. In Section 5 we develop the econometric model and discuss why cross-sectional, regression-based estimates of the MWP bear little relation to the true value of that function. Section 6 contains a discussion of the estimates of model parameters, and in Section 7 we indicate the manner in which our modeling approach and results may inform policy makers in this field. A brief conclusion is provided in Section 8.

In this paper we have taken a first step in rectifying the neglect of spousal search in a dynamic model of household behavior. The framework within which we have worked is highly stylized, and most importantly omits strategic interactions between spouses as well as the presence of capital markets. It shares the problem of most theoretical and empirical studies of health insurance by not specifying the underlying behavioral and technological motives for its demand by the household. In terms of other theoretical contributions, we consider the case of multi-attribute offers. While there are a number of other job characteristics besides health insurance provision that could be included in this model, health insurance is a particularly important characteristic in the household search framework given its quasi-publicness. We illustrate the manner in which the publicness of health insurance coverage and a nonconstant marginal utility of household consumption combine to link the labor market decisions of the spouses. This interdependence in decision-making and payoffs in a dynamic context makes attempts to compute the willingness to pay for health insurance from cross-sectional wage–health insurance relationships misguided. Given the model structure, we can compute willingness to pay functions given the state variables defining the household at a given moment in time. Needless to say, there is in general no single constant View the MathML sourceMWP parameter that arises. We estimate a valuation of health insurance by the household, that seems somewhat reasonable. This gives us some hope that this type of analysis, conducted in a more ambitious equilibrium framework, may eventually prove useful in stimulating and guiding the policy debate on the health insurance coverage question.