|
Article Excerpt
I. INTRODUCTION
Nursing home care is provided in the United States by for-profit, private nonprofit, and government organizations. (1) These three organizational forms may compete directly in the same markets--with either differentiated or nondifferentiated output--or serve entirely separate markets. An understanding of which organizational forms will provide the most effective care in a given set of circumstances requires knowledge of in what ways and to what extent the nature of health care provision varies across organizational forms. In the context of a market economy in which profit-maximizing firms are the "default producers," the natural question from a welfare perspective is whether consumer surplus would decline if government and nonprofit providers were replaced with for-profit providers.
This study examines whether nonprofit and government nursing facilities expand access to nursing home care through an examination of patterns of nursing home entry in the rural United States. I employ methods due to Bresnahan and Reiss (1990) to identify markets that are unlikely to be profitable for for-profits to enter and then ask whether nonprofits and government nursing homes--as a consequence of policies that lower their costs relative to for-profits--locate in these "unprofitable markets." Equivalently, if government and nonprofit nursing homes were to exit isolated markets, could one reasonably expect that for-profits would enter to replace them? Any evidence suggesting that they would not supports the argument that government and nonprofit nursing home care providers contribute positively to social welfare by expanding access to care.
In assessing potential differences in care provided by the different organizational forms, the literature has focused largely on access and quality differentials. Two questions of interest naturally arise when studying access to care: first, is any care available in the market--that is, is there a nursing home? Second, if care is available to some in the market, is it available to all--that is, are nursing homes full, or do they discriminate at the point of admission on the basis of wealth, health status, or other observable measures? The second question has received more attention in the literature. Most studies of access to nursing home care--for example, Reschovsky (1996), Ettner (1993), and Gertler (1992)--presume the presence of a facility and ask to what extent prospective residents have access to the facility, often focusing on individuals covered by Medicaid. Relatively fewer focus on issues of access in isolated markets; Yu and Bradford (1995) is an exception. While this study limits its focus to the first question by analyzing isolated rural markets in the United States, the methods used here in principle can be extended to focus on the second question as well. In focusing on the question of access, this analysis abstracts from potential differences in quality across organizational forms, deferring to existing studies on the subject, including Grabowski and Hirth (2003), Hirth (1999), Chou (2002), Holtmann and Ullmann (1991), and Weisbrod (1988).
The question of whether nonprofits enter unprofitable geographic markets has not been previously addressed, in the nursing home industry or elsewhere. The study closest in spirit to the present one is the analysis of nonprofit and for-profit hospital location by Norton and Staiger (1994), in which the authors conclude that for-profits tend to locate in better-insured areas--that is, areas that are more likely to be more profitable--than nonprofits do. This study differs from Norton and Staiger in that it explicitly calculates market profitability for markets that for-profits have entered and projects it for markets that for-profits have not entered. This analysis also complements the work of Kapur and Weisbrod (2000) by assessing whether government nursing homes serve unprofitable markets, while their paper addressed the question of whether government nursing homes serve unprofitable patients. Yu and Bradford (1995) present a comparison of nursing home quality and access between rural and urban markets. Like Yu and Bradford, the present study is particularly concerned with addressing access issues in less populated areas, differing primarily by its emphasis on the comparison between nonprofit and for-profit institutions instead of that between rural and urban markets.
Nonprofit and government nursing homes both benefit from policies that reduce their costs relative to for-profits. One consequence of such policies is that these organizational forms may be able to operate profitably in markets where for-profits cannot. Nonprofit organizations, as defined by Section 501(c)(3) of the Internal Revenue Code, enjoy an exemption from federal and state income taxes and, frequently, from property taxes. Nonprofit organizations also are eligible to receive donations that are tax deductible (subject to certain limits) to the donor. This favorable tax treatment potentially confers a cost advantage on nonprofit firms, as the income tax exemption increases retained variable profits while the property tax exemption and availability of tax-deductible donations reduce the burden of fixed costs. Nonprofits also tend to have greater access to volunteers, which may lower labor costs.
Government-run organizations, like nonprofits, do not pay taxes. Unlike nonprofits, however, government organizations generally have access to tax revenues. Given this access and studies such as Kornai (1986) suggesting that governmental organizations are subject to relatively soft budget constraints, government nursing homes may have a particular capacity to locate in unprofitable markets. There are countervailing factors, however, that may reduce or even eliminate the theoretical cost advantage of nonprofit and government nursing homes. The primary one is a lack of residual claimant. To the extent that organizations lacking residual claimants are less efficient, any cost advantages conferred by policy may be dissipated or captured by inefficient or self-interested management. Hence, the question of whether nonprofit and government nursing homes' conduct is consistent with the presence of a cost advantage is fundamentally an empirical one.
In order to assess whether nonprofit or government nursing homes locate in unprofitable markets, one must ask upon observing a nonprofit or government facility occupying a particular market whether a for-profit would be capable of entering that same market if the incumbent were to exit. That is, is the market unprofitable or is the nonprofit/government facility there simply because it entered first? As an identification strategy, I assume that patterns of for-profit location reveal market profitability in that these types will seek to enter all markets that are profitable and will not enter any markets in which they expect to lose money. Market profitability is defined here with respect to for-profit nursing homes: a market is considered to be profitable if there exists a prospective for-profit nursing home with a cost function such that the expected discounted profits from a monopoly position in that market are positive. Empirically, I model market profitability as a function of several variables, including population, population density, and household income, with (for-profit) profits as an unobserved latent variable and whether or not a for-profit is observed in a market as the corresponding observed dichotomous variable that changes values depending on the value of profits. Based on the analysis of for-profit location, market profitability can be predicted for those markets in which nonprofit and government facilities operate. In order to address the question of whether nonprofit or government nursing homes locate in unprofitable markets, I test whether the probability that markets occupied by a nonprofit or government facility are profitable is statistically lower than the probability that markets entered by a single for-profit are profitable. The models estimated here consistently show that while government nursing homes locate in markets that are likely to be unprofitable for for-profits to enter, the markets that nonprofits enter are comparable in predicted profitability to those entered by for-profits. The empirical strategy is outlined in Section II, followed by a discussion of the data and estimation results in the third and fourth sections, respectively. In Section V, I discuss the implications of the findings for the relationship between organizational form and access to care, as well as implications for policy and limitations of this study.
II. ESTIMATION AND PREDICTION OF MARKET PROFITABILITY
Nonprofit nursing homes will locate in unprofitable markets only if their expected discounted profits in that market are nonnegative and their objectives do not preclude such location. Since government nursing homes tend to be locally owned (and therefore need to meet some rough form of profitability test as well), one might expect a similar set of conditions to apply to them. One implication of these conditions is that even if nonprofit and government facilities have a cost advantage over for-profits, they will not necessarily locate in unprofitable markets. For example, nonprofits typically may care more about providing higher quality than about providing access to isolated markets even if they could profitably do so. A second implication is that nonprofit and government facilities that cannot preserve their cost advantage--perhaps because cost savings are captured by management--will not be able to enter unprofitable markets. Thus, while a finding that nonprofit or government nursing homes enter unprofitable markets provides information on both costs and objectives, the data do not permit the identification of the separate influences of costs and objectives in the event of the opposite finding.
In order to emphasize the distinction between for-profits and other organizational forms in the discussion of the empirical strategy that follows, I refer to the set of organizational forms that are not for-profit--namely, private nonprofit and government organizations--together as "not-for-profits." Determining whether not-for-profit nursing homes locate in markets that are likely to be unprofitable requires the ability to identify those markets that are profitable to enter and those that are not. I use for-profit patterns of entry to address this issue. In the absence of barriers to entry, when one observes an empty market--that is, a market with no firms--one can reasonably conclude that this market is not profitable; otherwise, a for-profit would have entered to capture those profits. Likewise, when one observes a single for-profit operating in a market, it is reasonable to assume that that market is profitable for one firm. When one observes a single not-for-profit operating in a market, however, either of two explanations may be valid. The first is that the market is unprofitable, but there exists a not-for-profit with a cost advantage that enables it to operate profitably in that market. The second possible explanation is that not-for-profits, owing to some combination of their objectives and financial constraints, enter only profitable markets but that the observed market does not yield positive duopoly profits and the not-for-profit occupies it (instead of a for-profit) only because it entered first. This section of the paper derives a strategy for distinguishing between these two explanations.
In addressing the question of which markets have access to nursing home care and which do not, the relevant comparison is between markets that have no nursing homes (no access) and those that have at least one nursing home (access). In order to abstract from potential differences in strategic behavior across organizational forms, this study focuses on a comparison between markets that have no nursing homes with markets that have exactly one nursing home. The present value of profits of for-profit nursing home i operating in market m is a latent variable (designated by [[pi].sub.i]), which is the sum of a deterministic function and a random component drawn from a distribution G:
(1) [[pi].sub.i]([x.sub.m] [equivalent to] [pi]([x.sub.m]; [theta]) + [[epsilon].sub.i],
where [theta] is a vector of parameters to be estimated and x is a vector of market characteristics. The random component captures unobserved determinants of profit, which vary across prospective nursing homes and are assumed to be uncorrelated with market characteristics. These include, for example, differences in efficiency across firms that make a given market profitable for one firm but not for another.
In concept, a market is unprofitable if [[pi].sub.i](X) 0} > 0.5. (3)
A for-profit nursing home i locates in market m if and only if [[pi].sub.i]([X.sub.m]) > 0. While profits are not observed, whether a for-profit is located...
|
