اثر دسترسی به مراقبت های خیریه برای تقاضای فاقد بیمه برای بیمه سلامت خصوصی

The economic reasons why some people do not obtain health insurance are unclear. In this paper, I test the hypothesis that the availability of charity care to the uninsured reduces the likelihood of obtaining private coverage. I utilize variation in the availability of charity care across the different markets in the Community Tracking Study's Household Survey (CTS-HS) using an “access to care” measure of the uninsured's cost-related difficulties in obtaining medical care, to both aggregate across the various “safety net” providers and control for its potentially endogenous supply. I find evidence supporting this hypothesis for low-income people, in both the individual market and the employment-based group market. I also estimate a joint model of offer and take-up decisions for the group market sample and find that the availability of charity care reduces low-income workers’ offer rates but not their take-up rates.

Why do the uninsured in the United States fail to obtain private health insurance? There were almost 44 million people without health insurance in 2002, and various policies are currently under consideration to expand coverage. Despite the significant policy interest in the uninsured, there exists a fair amount of uncertainty about the economic determinants of whether people ineligible for public insurance purchase private insurance. What is known is that large subsidies for private health insurance premiums will likely be needed to induce a large number of the uninsured to obtain coverage ( Gruber and Levitt, 2000 and Pauly and Herring, 2001). Most policymakers focus on issues related to the magnitude of premiums and ways to reduce the net prices for insurance that people face, but an interesting underlying question is why is the willingness-to-pay for private coverage of the uninsured so low. In this paper, I argue that it is not necessarily the absolute cost of health insurance that is prohibitive for many of the uninsured; instead, it is the cost of health insurance relative to the costs associated with remaining uninsured that is important for one to consider. Various “safety net” providers supply free or subsidized care to the uninsured due to altruistic concerns, which lowers the uninsured's expected out-of-pocket expenses considerably. Rational economic actors will realize that the availability of charity care lowers the value of obtaining private health insurance coverage, and thus the relative likelihood of purchasing private coverage should decrease. I present an empirical test of this hypothesis in this paper.1 Testing this relationship between insurance coverage and the availability of charity care, however, is not clear-cut for two main reasons: there are many different safety net providers of charity care, and these providers may increase their supply of charity care in response to larger numbers of uninsured. To address these issues, I use a local-level “access to care” measure of the absence of cost-related difficulties in obtaining care reported by the area's uninsured. I argue below that such a measure both appropriately aggregates across the different safety net providers (which serve as substitutes in different areas) and is not subject to reverse causality. I examine the likelihood of obtaining private coverage in the individual market and the employment-based group market separately. Since one must be offered coverage and take up offered coverage to be insured in the employment-based group market, I estimate a simultaneous model of offer and take-up decisions for people in the group market. Section 2 of the paper reviews some theory regarding the demand for insurance and presents a simple theoretical model to illustrate my hypothesis. Section 3 of the paper details the amount of charity care available to the uninsured by examining medical expenditure data. Section 4 of the paper presents the empirical model and its results for the demand for private insurance as influenced by the availability of charity care; the beginning of Section 4 describes the access measure that I use and verifies that it well-specified, while the latter portion of Section 4 estimates various empirical models for insurance coverage. Section 5 of the paper discusses my findings.

My analysis of the availability of charity care to the uninsured using the MEPS data reveals that the uninsured on average pay out-of-pocket for about one-third of the medical care they consume. My analysis of the CTS-HS data indicates that charity care from hospitals, physicians, and community health centers is generally successful in decreasing the cost-related difficulties in obtaining care faced by the uninsured, but that variation in the availability of charity care across markets in the U.S. is considerable. I utilize this variation to test the hypothesis that the availability of such charity care to the uninsured has a negative effect on the demand for private health insurance, since it is rational for some people to forego purchasing private insurance in order to remain uninsured and utilize charity care.23 The empirical test I present in this paper is novel in its use of a unique “access to care” measure of the cost-related difficulties in obtaining care, which can both aggregate the various providers of charity care to the uninsured into a single measure and capture only exogenous variation in local altruism towards the uninsured by not being endogenously related to the local proportion of uninsured. For the entire non-elderly population (not covered by public insurance), my estimates indicate that increasing the availability of charity such that the expected out-of-pocket spending of the uninsured decreased by 10% would result in decreasing the proportion of non-elderly individuals with private health insurance by approximately 0.53%.24 This would correspond to an overall increase of about 0.9 million uninsured in the U.S.25 The estimated magnitude of this behavior is, as hypothesized, larger for low-income people than for high-income people. My point estimates indicate that the proportion of low-income people obtaining individual insurance would decrease by approximately 3.06% if expected out-of-pocket spending decreased by 10% while the proportion of low-income people obtaining employment-based group insurance would decrease by approximately 0.94% if expected out-of-pocket spending decreased by 10%. However, the difference between these individual and group market estimates is not statistically significant. Although the results are tentative due to the less-than-ideal identification of the bivariate probit model for joint offer and accept decisions, my analysis for low-income workers indicates that the effect of the availability of charity care on the demand for group insurance appears to be driven by the likelihood of these workers’ being offered group insurance rather than their likelihood of taking up offered insurance. These results have important policy considerations. President Bush and Republicans in Congress have frequently advocated a substantial increase in the number of community health centers in the U.S. in order to increase access to care for the uninsured. There has also been increased media attention given to certain hospitals’ procedures for billing low-income uninsured patients, and it is possible that policymakers will be pressured to regulate hospital uncompensated care in a way that increases the availability of charity care. My findings imply that while such policies will likely be successful in actually increasing access to care for the uninsured, there will also be the unintended consequence of increasing the number of uninsured, since some low-income people who currently obtain private insurance will likely drop their coverage to instead utilize a strengthened safety net. For this reason, it is possible that the resources spent on improving the safety net of care to the uninsured might be better spent on efforts to expand insurance coverage by increasing Medicaid and SCHIP eligibility and/or implementing refundable tax credits for private insurance. However, my results (as well as those from the Medicaid “crowd out” literature) also imply that such expansions will be rather expensive, since large subsidies will needed to lower the “net price” of private insurance for the uninsured below their reservation prices (which are relatively low due to the availability of charity care); thus, wide-scale expansions may unfortunately be politically infeasible due to their high cost. To truly evaluate the cost-effectiveness of these various proposals, we researchers need to better understand the different incremental effects on health status resulting from safety net providers, public insurance, and private insurance.