Abstract

Do nonprofit hospitals provide enough community benefits to justify their tax exemptions? States have sought to enhance nonprofit hospitals' accountability and oversight through regulation, including requirements to report community benefits, conduct community health needs assessments, provide minimum levels of community benefits, and adhere to minimum income eligibility standards for charity care. However, little research has assessed these regulations' impact on community benefits. Using 2009–11 Internal Revenue Service data on community benefit spending for more than eighteen hundred hospitals and the Hilltop Institute's data on community benefit regulation, we investigated the relationship between these four types of regulation and the level and types of hospital-provided community benefits. Our multivariate regression analyses showed that only community health needs assessments were consistently associated with greater community benefit spending. The results for reporting and minimum spending requirements were mixed, while minimum income eligibility standards for charity care were unrelated to community benefit spending. State adoption of multiple types of regulation was consistently associated with higher levels of hospital-provided community benefits, possibly because regulatory intensity conveys a strong signal to the hospital community that more spending is expected. This study can inform efforts to design regulations that will encourage hospitals to provide community benefits consistent with policy makers' goals.

Nearly 60 percent of the approximately five thousand hospitals in the United States are nonprofit (American Hospital Association 2014), most of which are exempt from paying federal, state, and local taxes because of their status as charitable entities. The value of the tax exemption for these hospitals was estimated to total $24 billion in 2011 and has likely grown since (Rosenbaum et al. 2015). About half of this amount represents exemptions from federal income taxes, while the remainder is from state and local tax exemptions, including income, property, and sales taxes (US Government Accountability Office 2005; Rosenbaum et al. 2015). Hospitals typically fulfill their charitable obligations through the provision of community benefits, which consist of services and activities provided on a fully or partially subsidized basis (Young et al. 2013).

A long-standing controversy concerns whether nonprofit hospitals provide sufficient community benefits to justify their tax exemptions (Rosenbaum and Margulies 2012). Many states have responded to this controversy through regulatory initiatives designed to enhance accountability and oversight of nonprofit hospitals regarding their provision of community benefits. Currently, more than half of all states require nonprofit hospitals to make some sort of disclosure of their community benefit activities (Nelson et al. 2013). These states typically require hospitals to file annual reports that provide detailed information about their community benefit activities, including the costs incurred in providing them. Some of these states also impose additional requirements that nonprofit hospitals must meet to remain tax exempt. Additional requirements are typically more directive in terms of how much or what types of benefits must be provided in return for tax exemptions. One state (Oklahoma) is an exception in this respect because it did not have a community benefit reporting requirement during the study period but did impose additional requirements for hospitals in return for a tax exemption.

Despite the proliferation of state-level regulatory initiatives, only a small number of studies have assessed their impact on nonprofit hospitals' provision of community benefits. Ginn and Moseley (2006) reported that hospitals in states that have passed some type of community benefit regulation engaged in more community-oriented activities (e.g., having a long-term plan for improving the health of the community) than did hospitals in other states, but they did not examine how much hospitals actually spent on community benefits. Young et al. (2013) did examine hospital community benefit spending and found that such spending was greater among hospitals located in states that had adopted community benefit reporting requirements. Their study, however, did not consider any other forms of community benefit regulations. Kennedy et al. (2010) specifically examined the impact of one type of community benefit requirement, Texas's adoption of a minimum spending requirement for charity care. They concluded that the requirement did not lead to an overall increase in hospitals' provision of charity care because some hospitals that had previously spent above the required minimum level decreased their spending following the adoption of the requirement. More recently, Tahk (2014) examined the relationship between state requirements for a minimum level of hospital community benefit spending and actual levels of such spending. Study results did not point to any association between the minimum spending requirements and hospital spending.

In this study, we examined the relationship between four common types of state community benefit regulations and hospital spending on community benefits: (1) community benefit reporting, (2) community health needs assessments (CHNAs), (3) minimum levels of community benefits, and (4) minimum income eligibility standards for charity care. Understanding the effectiveness of these four types of state regulatory initiatives is particularly important in light of efforts at the federal level to exact greater accountability from nonprofit hospitals regarding their provision of community benefits. Since 2009, all nonprofit hospitals with a federal tax exemption have been required to report their community benefit spending on the revised Schedule H of the Internal Revenue Service (IRS) Form 990. Moreover, under the Affordable Care Act, Congress imposed additional requirements for nonprofit hospitals to maintain their tax-exempt status. Among these is the requirement that hospitals complete a CHNA every three years and develop an implementation strategy to address identified needs (Rosenbaum and Margulies 2012).

This study extends previous research on this topic in several ways. First, we estimated the individual and cumulative relationships between the four types of state community benefit regulations and nonprofit hospitals' spending on community benefits. Previous research largely focused on the relationship between reporting requirements and spending on community benefits. The variation among states in the adoption of the four types of community benefit regulations offers an opportunity to study the relationship between different combinations of community benefit regulations and hospitals' community benefit spending.

Second, we estimated the impact of state regulations on both the level and the pattern of hospitals' community benefit spending as reported on IRS Form 990 Schedule H. To our knowledge, our study is the first study to use community benefit spending data from Schedule H to analyze the link between state regulations and spending patterns. Community benefits as defined by the IRS take many different forms, most of which can be categorized either as spending on direct patient care services (including charity care, Medicaid shortfalls, and other subsidized health services) or as spending on broader community health initiatives. Although we know from previous research that over 80 percent of what hospitals spend on community benefits goes to direct patient care activities (Young et al. 2013; Tahk 2014), policy makers have been keenly interested in seeing hospitals devote more resources to community health initiatives that promote prevention and wellness (Rosenbaum et al. 2015). However, even when community benefit regulations are aimed specifically at direct patient care activities, spillover effects might occur for other types of community benefits if hospitals view such regulations as a broader call to expand the community benefits they provide.

Third, our research sought to address concerns about potential endogeneity bias in our ordinary least squares (OLS) regression analysis with an instrumental variables (IV) approach. All states that adopted one or more of the community benefit regulations examined in this study adopted them before 2009. Uniform national data on hospital community benefit spending, however, were not available until 2009 when the IRS first required nonprofit hospitals to complete the revised IRS Form 990 Schedule H (Young et al. 2013). This lack of historical spending data hinders study designs based on a comparison of hospital spending before and after regulatory adoption. While cross-sectional variation in state adoption of community benefit regulations allows for the estimation of the relationship between regulation and spending, it is possible that unobservable factors contributed both to a state's decision to adopt and retain such regulations and to hospitals' decisions regarding their community benefit spending.

Background

States differ in their standards for granting tax-exempt status to nonprofit hospitals. Some states grant tax-exempt status to any nonprofit hospital that has been deemed by the IRS to be in compliance with the federal standard. The federal standard generally requires that nonprofit hospitals engage in activities that benefit their communities in return for exemption from federal income tax (Colombo, Griffith, and King 2011). Other states have created their own standards that nonprofit hospitals must meet to remain exempt from state and local taxes. States with community benefit regulations generally fall into one of two groups (Rubin, Singh, and Young 2015). The first group includes states that have adopted only a community benefit reporting requirement. The second group includes states that require nonprofit hospitals to fulfill one or more additional requirements beyond reporting their community benefit activities, including, for instance, the requirement to conduct periodic CHNAs or to provide minimum levels of community benefits in return for tax exemption. Only one state included in our study deviated from this rule: during the study period, Oklahoma did not have a community benefit reporting requirement but did impose additional requirements for hospitals in return for a tax exemption.

The primary purpose of a community benefit reporting requirement is to increase accountability and transparency regarding hospitals' provision of benefits to their communities. States with a community benefit reporting requirement typically require nonprofit hospitals to submit annual reports to government agencies that detail their community benefit activities, including their expenditures on these activities. Some states make the reports available to the public, either via a publicly accessible website or upon request. Public availability of hospitals' reports, in particular, can help increase transparency and inform public debate regarding the role that nonprofit hospitals should and do play in their communities (Rubin, Singh, and Young 2015). Increased transparency can also promote ongoing scrutiny of the adequacy of nonprofit hospitals' community benefit activities and, as a result, provide incentives for hospitals to provide higher levels of community benefits.

While reporting requirements promote transparency and potentially create incentives for hospitals to provide higher levels of community benefits than they otherwise would, they do not require that hospitals provide a specific level or type of community benefit. Some states have thus chosen to go beyond reporting requirements by adopting additional regulatory requirements. Three common types of additional state-level regulations are the requirement to conduct periodic CHNAs, the requirement to spend a minimum level on community benefits, and the requirement to adhere to minimum income eligibility standards for charity care. Conceptually, each of these regulations has the potential to influence hospital community benefit spending.

CHNAs provide hospitals with detailed information about community needs that hospitals may seek to address through community benefit spending. For example, if a needs assessment points to obesity as an important problem for the community served, the hospital may decide to invest in educational initiatives for the community focusing on principles of good nutrition and exercise. Hospitals may do so either by redirecting some of their current community benefit dollars toward these activities (i.e., by changing the pattern of their community benefit spending) or by spending additional funds (i.e., by changing the level of their community benefit spending). However, CHNA requirements alone do not require hospitals to make any changes to their community benefit spending.

Minimum community benefit spending requirements aim to ensure that nonprofit hospitals provide a certain minimum level of community benefit in return for tax exemption. Depending on the state, this requirement may be formulated as a percentage of a hospital's operating budget or may be based on the value of its state and/or local tax exemptions. Minimum spending requirements typically aim to increase overall spending by hospitals on a defined set of community benefits. However, such requirements can also result in a “race to the bottom” in which some hospitals that have historically provided relatively high levels of community benefits reduce their contributions to the level required to maintain tax exemption. The previously noted study by Kennedy et al. (2010) suggests such a result in Texas following its adoption of a minimum spending requirement for charity care. Furthermore, minimum spending requirements may result in spending trade-offs whereby some hospitals respond by increasing their spending on the community benefits that are the explicit focus of the regulation while decreasing their spending on other benefits.

Minimum income eligibility standards for charity care require hospitals to adhere to minimum financial criteria in determining whether patients are eligible for charity care. For example, a state may mandate that charity care be made available to all individuals with incomes below 150 percent of the federal poverty level. This type of regulation has the potential to raise hospital spending on charity care by expanding the number of patients that are eligible for such care. Similar to minimum community benefit spending requirements, however, minimum income eligibility standards for charity care may result in a “race to the bottom” if hospitals with more generous charity care policies lower their eligibility threshold to the statewide minimum.

Methods

To examine the relationship between state-level community benefit regulation and nonprofit hospitals' provision of community benefits, we estimated pooled cross-sectional regression models that accounted for institutional, market, and community characteristics that potentially influence hospitals' community benefit expenditures (Young et al. 2013; Tahk 2014). We first estimated OLS regressions with hospital community benefit spending as the dependent variable and each of the four types of regulatory requirements for community benefits as separate independent variables. The regression analysis had the following general form: 
CBspendit = α + β1*stateCBlawsit + β2* Xit + β3*Tt + εit

where CBspendit represents hospital i's spending in year t on one of three categories of community benefits discussed in more detail below; stateCBlawsit represents a vector of indicators characterizing the community benefit regulations applicable to nonprofit hospitals in hospital i's state in year t; Xit represents a vector of institutional, community, and market-level control variables for hospital i in year t; Tt represents a vector of year dummy variables; and ɛit represents the error term for hospital i in year t.

In addition, we estimated OLS regressions with hospital community benefit spending as the dependent variable and a summary index of the comprehensiveness of a state's community benefit laws as the key independent variable. This allowed us to test the cumulative relationship between state community benefit regulations and hospitals' community benefit spending. As a sensitivity analysis, we reestimated all regressions using a random effects model to account for the nesting of hospitals within states.

Moreover, we conducted robustness tests of our OLS results using an instrumental variables (IV) approach to account for potential endogeneity bias. For an IV approach to be valid, the selected IV needs to be correlated with the independent variable of primary interest but not with the error term in the regression equation. The IV cannot directly impact the dependent variable. For the purpose of this study, an appropriate instrument must be correlated with state adoption of community benefit regulations but not with the error term in the community benefit spending equation. The instruments chosen for our robustness tests were based on the number of state health insurance mandates adopted by each state. A health insurance mandate is a requirement that an insurance company or health plan cover specific services, providers, or patient populations. The number of health insurance mandates reflects a state's propensity for government involvement in the health arena. The propensity for government involvement also tends to influence the number of community benefit regulations, so these two variables were hypothesized to be correlated.

One might be concerned, however, about whether an instrument based on the number of state health insurance mandates meets the second criterion. In theory, health insurance mandates could have two (potentially offsetting) effects on hospitals' community benefit spending. On the one hand, requiring private insurers to cover a greater array of services may reduce the need for hospitals to provide these services as part of their community benefits. On the other hand, requiring private insurers to cover more services may make health insurance less affordable and thus contribute to an increase in the number of uninsured. Higher levels of uninsured individuals in a community may lead to greater demand for charity care and other types of community benefits.

Despite these concerns, an instrument based on the number of state health insurance mandates may nevertheless satisfy the second criterion. First, many health insurance mandates have little connection to the types of community benefits hospitals typically provide. There is little reason to believe that many types of mandates on private insurers, such as mandates to cover autism treatment, in vitro fertilizations, or prescription drugs, would directly influence hospitals' provision of community benefits. Second, the potential impact of a greater number of state health insurance mandates on the likelihood that individuals within a community are uninsured can be accounted for in a regression equation by including a control variable for the percentage of uninsured individuals in the community the hospital serves. Because all of our regressions included as a control variable the percentage of uninsured persons in the local community, our instrument should be uncorrelated with the error term.

Data and Sample

Data for this study came from several sources. Data on hospital community benefit spending were obtained from hospitals' IRS Form 990 Schedule H, which contains detailed spending information for the following expenditure categories: financial assistance to indigent patients (i.e., charity care), payment shortfalls (i.e., reimbursement below costs) for patients enrolled in Medicaid and other means-tested public insurance programs, subsidized health services, research, health professions training, community health improvement activities, and cash and in-kind contributions to community groups for community benefit activities. Data on state community benefit legislation were obtained from the Community Benefit State Law Profiles compiled and published by the Hilltop Institute (www.hilltopinstitute.org/hcbp_cbl.cfm). For each state, the Hilltop Institute's profiles provide information about whether a state has adopted specific community benefit laws, regulations, and tax exemptions for hospitals. Data on the number of state health insurance benefit mandates were obtained from the Council for Affordable Health Insurance (Bunce and Wieske 2009). The council compiles and publishes detailed information on the number and types of health insurance mandates by state and year. Additional data on hospital-level institutional and market characteristics came from the American Hospital Association's Annual Survey and the Center for Medicare and Medicaid Services. Data on community characteristics came from the Area Health Resource File from the Health Resources and Services Administration.

The study sample comprised general nonprofit hospitals that were in operation at some point between 2009 and 2011 and reported community benefits on Schedule H at the individual hospital level. The unit of analysis was the hospital year. The final study sample comprised 5,384 hospital-year observations for which we had complete data. Specifically, we had complete data for 1,825 hospitals in year 2009, 1,790 hospitals in year 2010, and 1,769 hospitals in year 2011. We used American Hospital Association survey data to compare the study sample to the general population of nonprofit hospitals on several institutional characteristics (e.g., number of beds, system membership, urban/rural location). We found the two groups of hospitals to be largely comparable on these characteristics, including system membership, for which the study sample only slightly underrepresented the general population of nonprofits. This is because some hospitals belong to hospital systems that have received IRS approval to submit a consolidated Form 990 for all hospitals in the system and thus do not report their own Schedule H data to the IRS. However, many hospital systems do not submit consolidated reports, so their member hospitals report hospital-level data for Schedule H, and these hospitals are included in our study sample.

Measures

Dependent Variables

Using data from IRS 990 Schedule H, we constructed three distinct measures of hospital community benefit spending: spending on total community benefits, spending on direct patient care services, and spending on community health initiatives. Spending on total community benefits was specified as a hospital's net cost of providing all community benefits as defined by the IRS in Form 990 Schedule H, divided by the hospital's total operating expenditures. Spending on total community benefits is the sum of spending on direct patient care services, community health initiatives, and research and health professional education. Spending on direct patient care services was specified as the sum of a hospital's net cost of financial assistance to indigent patients (charity care), the unreimbursed costs of services provided to patients covered under Medicaid and other means-tested government programs (hereafter “Medicaid payment shortfalls”), and the net cost of subsidized health services, divided by the hospital's total operating expenditures. Spending on community health initiatives included the net expenditures for programs and services that benefit the community more broadly, including community health improvement programs and cash and in-kind contributions to community groups, divided by the hospital's total operating expenditures. Spending on research and health professional education is the sum of research spending and health professional education spending.

Independent Variables

The independent variables of interest were four state-level community benefit regulations. Using the Hilltop Institute's Community Benefit State Law Profiles, we determined whether or not, for each year included in our study, a state required hospitals to (1) report community benefits, (2) conduct community health needs assessments, (3) provide minimum levels of community benefits, or (4) adhere to minimum income eligibility standards for charity care. For the purpose of this study we defined each type of community benefit regulation as follows: (1) A state was considered to have a community benefit reporting requirement if the Hilltop Institute's profiles indicated that the state required hospitals to report the community benefits they provide. This included both regular (generally annual) reporting requirements (irrespective of whether the requirement exempted certain hospitals such as small hospitals based on number of beds) and reporting requirements for the purpose of securing a certificate of need. (2) A state was considered to have a CHNA requirement if the Hilltop Institute's profiles indicated that the state required hospitals to conduct regular CHNAs. (3) A state was considered to have a minimum community benefit expenditure requirement if the Hilltop Institute's profiles indicated that the state specified a minimum level of community benefits that a nonprofit hospital must provide to remain tax exempt. (4) A state was considered to have a minimum income eligibility level for charity care if the Hilltop Institute's profiles listed a defined minimum income eligibility level for charity care as part of the state's regulations for financial assistance policies.

For the first set of regression models, each of the four community benefit regulations was accounted for by an indicator variable. The indicator variable took on a value of 1 if the regulation had been adopted in a given year by the state in which a hospital was located and 0 otherwise. For the second set of regression models, we constructed a summary index, defined as the number of regulations present in a given year in the state in which a hospital was located. The summary index ranged from 0 (no regulations present in a state in a given year) to 4 (all four types of regulations present in a state in a given year).

Instrumental Variables

The IVs employed in this study were based on the number of health insurance mandates that each state had adopted in each year of the study period. Most state-mandated health insurance benefits are for specific services, such as for colorectal cancer screening or in vitro fertilization. Some are for services provided by a particular kind of provider, such as a nurse midwife or podiatrist. A few benefit mandates require coverage for particular individuals, such as adopted children or domestic partners. We specified two separate IVs: one based on the total number of health insurance mandates adopted, and the other based on a subset of mandates that related specifically to services that would ordinarily occur in ambulatory, nonemergency settings and so were potentially less likely to be provided at reduced or no cost to the uninsured or as part of community benefit initiatives (see appendix table A1). Such mandates typically comprised approximately 20 percent of a state's total health insurance mandates.

Control Variables

We included a number of institutional, community, and market characteristics as control variables in all regression models. Institutional control variables included the number of beds, system affiliation, network affiliation, case mix index, wage index, teaching status, whether or not the hospital was contract managed, church affiliation, whether or not the hospital was a sole community provider, and indicators of hospital profitability. Data for all of these variables came from the American Hospital Association's annual survey with the exception of case mix index and wage index, for which we obtained data from the Center for Medicare and Medicaid Services, and indicators of hospital profitability, for which we obtained data from IRS Form 990. Community and market characteristics included per capita income in the local community, market competition (measured in terms of the Herfindahl-Hirschman Index), the percentage of residents without health insurance, the percentage of hospital beds in the local community controlled by for-profit and public hospitals, urban/rural location, and geographic region (defined as Northeast, Midwest, South, or West). Consistent with previous studies, all community and market characteristics (with the exception of geographic region) were defined at the level of the county in which a hospital was located (Young et al. 2013; Singh et al. 2015). Data for community and market characteristics came from the American Hospital Association's annual survey with the exception of per capita income in the local community and the percentage of residents without health insurance, for which we obtained data from the Health Resources and Services Administration's Area Health Resource File.

Results

Descriptive Results

As presented in table 1, states varied considerably regarding which, if any, community benefit regulatory requirements they had adopted. In 2011, the most recent year included in this study, twenty-nine states and the District of Columbia had a community benefit reporting requirement, ten states had a CHNA requirement, four states had a minimum community benefit spending requirement, and nine states and the District of Columbia had a minimum income eligibility standard for charity care. Fifteen states and the District of Columbia had more than one type of community benefit regulation, while twenty states had no state-level community benefit requirements for nonprofit hospitals. Of the 1,769 hospitals in our study sample for fiscal year 2011, 536 (30.3 percent) were located in states with no community benefit regulations, 566 (32 percent) were located in states with one type of community benefit regulation, and 667 (37.7 percent) were located in states with more than one type of community benefit regulation.

Hospitals' community benefit spending was largely consistent over the three-year study period (table 2). In the most recent year (2011), hospitals spent an average of 7.6 percent of total operating expenses on community benefit activities. Of these expenditures, approximately 86 percent (6.5 percent of total operating expenses) went toward direct patient care services. Another 6 percent (0.47 percent of total operating expenses) was spent on community health initiatives. The remainder went toward spending on research and health professional education.

Community benefit spending varied widely among hospitals (table 2). In 2011, hospitals in the bottom quartile spent ≤4.2 percent of total operating expenditures, while hospitals in the top quartile spent ≥9.9 percent of total operating expenditures on all community benefit activities combined. Likewise, spending on direct patient care services ranged from ≤3.6 percent of total operating expenditures for hospitals in the bottom quartile to ≥8.4 percent in the top quartile. Community health initiatives spending ranged from ≤0.05 percent of total operating expenditures in the bottom quartile to ≥0.54 percent in the top quartile.

In all three years, hospitals in states with no community benefit regulations spent less, on average, on both total community benefits and direct patient care services (table 2). In 2011, average total spending by hospitals in states with no community benefit regulation amounted to 6.7 percent of total operating expenses. In comparison, hospitals in states with a community benefit reporting requirement spent 7.6 percent, and hospitals in states with a community benefit reporting requirement and at least one additional regulation spent 8.2 percent. Similarly, average spending on direct patient care services in 2011 was lowest among hospitals in states with no community benefit regulation and highest among hospitals in states with a community benefit reporting requirement and at least one additional regulatory requirement. Hospitals' spending on community health initiatives, on the other hand, did not follow the same pattern. Mean spending on community health initiatives did not differ significantly across hospitals in states with no community benefit regulation, with a reporting requirement only, and with more than one type of community benefit requirement.

Ordinary Least Squares Regression Results

Table 3 presents results from the OLS regressions. In model 1, only one of the four state-level regulatory requirements examined—the requirement to conduct periodic CHNAs—was significantly associated with greater hospital spending for all three categories of community benefit. For state-level reporting requirements, the results were more mixed. Reporting requirements were associated with greater spending on total community benefits and direct patient care services. Additional analysis (data not shown) showed that this relationship was positive and significant for charity care and subsidized health services, whereas reporting was not significantly associated with Medicaid payment shortfalls. However, reporting requirements were negatively associated with hospital spending on community health initiatives.

We also observed mixed results for state-level minimum community benefit spending requirements, which were associated with greater hospital spending on community health initiatives but lower spending on direct patient care services and total community benefits. Additional analysis of this finding (data not shown) revealed that hospital spending on charity care and subsidized health services was significantly lower in states with minimum community spending requirements, while Medicaid payment shortfalls were unrelated to minimum spending requirements. Finally, state-level minimum income eligibility standards for charity care were not significantly associated with hospital spending on any of the three community benefit categories.

Table 3 also presents the results for model 2, which analyzed whether the comprehensiveness of states' community benefit regulation was associated with hospital community benefit spending. Our summary index was positively and significantly associated with hospital spending for community benefits in all three categories, indicating that hospitals located in states with more regulation spent more on community benefits.

In both models, several control variables were associated with hospital spending for one or more categories of community benefits. Larger hospitals spent more on total community benefits and direct patient care services, as did teaching hospitals and hospitals located in communities with high rates of uninsurance. Higher case mix scores, indicating a hospital treats more severely ill patients, were associated with lower spending on total community benefits and direct patient care services. Hospitals located in communities with higher per capita income spent less on total community benefits and direct patient care services. With respect to hospital spending on community health initiatives, our results showed that hospitals that participated in a network and hospitals with higher profit margins spent more on such initiatives while hospitals with a religious affiliation spent less. Geographic patterns were also observed: hospitals located in the West generally had higher levels of spending on community benefits compared to hospitals located in other parts of the country.

Instrumental Variables Regression Results

Table 4 presents results for the IV regressions. Specifically, we used the IV procedure to test the robustness of the results presented in the second set of OLS regressions shown in table 3. Irrespective of the instrument used, the IV regression results were consistent with our OLS findings. More comprehensive community benefit regulation was associated with greater hospital spending for all three categories of community benefits. Compared to the OLS results, the magnitude of the coefficients in the IV regressions was somewhat larger. Since Hausman tests indicated that our OLS estimates might be subject to endogeneity bias, the results from the IV analysis suggest that OLS may understate the strength of the relationship between state-level community benefit requirements and hospital spending on community benefits.

With respect to the validity of the instruments, both IVs—number of health insurance mandates and the subset of mandates that pertain to services less likely to be provided by hospitals as community benefits—were strongly positively correlated with the number of community benefit regulations in each state (0.48, p < 0.01, and 0.37, p < 0.01, respectively). Our first-stage regression results (appendix table A2) showed that the two IVs captured a statistically significant part of the variation in hospital community benefit spending. The F-statistics for the first-stage models (as reported in table 4) indicated that each of the instruments was individually statistically significant and that the correlations ran in the predicted directions. The magnitudes of the first-stage F-statistics, as well as the R2 (as reported in table 4), indicated that our instruments were good predictors of the endogenous regressors. Moreover, appendix table A3 shows that many of institutional, community, and market characteristics included as control variables in our analyses did not differ significantly between hospitals located in states with below the median number of mandates and hospitals located in states with above the median number of health insurance mandates.

Discussion

Our investigation of state-level community benefit regulations suggests that three of the four regulations examined are associated with the level and pattern of hospital community benefit spending. However, the nature of these associations appears to depend on both the type of regulation and the type of community benefit examined. Of the four types of regulation studied, only CHNA requirements were consistently associated with greater spending on community benefits. By conducting these assessments, hospitals are in a position to learn more about the needs of the community they serve, thereby gaining valuable information for planning future community benefit expenditures.

Study results also point to the possibility that some regulations lead hospitals to make trade-offs in their community benefit spending such that spending increases for some types of benefits are offset at least in part by spending decreases for other types of benefits. This appeared true for both community benefit reporting and minimum community benefit spending requirements. Hospitals may respond to these types of regulation by altering the composition of their community benefit spending to favor spending on patient care services, such as charity care, over community health initiatives. Such results should give policy makers pause, particularly given the emphasis that both state and federal policy makers are placing on hospital initiatives to promote the health status of communities (Hanlon and Giles 2012; Stoto 2013).

We did not find a relationship between minimum income eligibility standards for charity care and hospitals' spending on community benefits. This finding may reflect that the adoption of such a minimum standard by a state results in hospitals that were previously above the standard lowering their spending while hospitals that were previously below the standard increase their spending. At least in states without a minimum standard, hospitals vary markedly in terms of their financial assistance policies; some hospitals have income eligibility thresholds for charity care that are more generous than the minimum standards that some states have adopted (Fuse Brown 2015; Dranove, Garthwaite, and Ody 2015). Of course, lax enforcement of minimum standards by the relevant states may also explain in part why we did not observe a relationship between these standards and hospital community benefit spending. In general, little is known about whether and how states enforce community benefit regulations.

Our study results also point to a relationship between community benefit spending and state adoption of multiple types of community benefit regulations. Hospitals located in states with multiple regulations had higher levels of spending for all three community benefit categories. Given the mixed findings for the individual types of community benefit regulation that we observed in this study, the pathway by which more comprehensive regulation influences community benefit spending is not clear. It may be that the individual positive effects from some types of regulations are sufficient to produce an overall positive effect for a combined set of regulations. Also, the adoption of multiple regulations may have a symbolic effect by sending a strong message to the hospital community that community benefits are taken seriously by the state for the purpose of granting tax exemptions. More comprehensive regulation might also encourage hospitals to critically evaluate their community benefit activities on a regular basis and make adjustments as needed. Thus, combining multiple types of regulation might have certain synergistic effects.

To our knowledge, our study is one of the first to demonstrate a relationship between state-level community benefit regulation and hospital community benefit spending. The possible impact of such regulations on hospital spending is not inconsequential. For example, based on the results of our OLS model, the adoption of a CHNA requirement potentially increases hospital community spending on average by 1 percent of a hospital's operating budget. For a hospital at the mean operating budget of approximately $200 million, this translates into an increase of $2 million in community benefit spending.

At the same time, the results of our study should be considered with several caveats. First, as noted, we lacked the data necessary to determine whether there were any baseline differences in community benefit spending between states that eventually adopted regulations and states that did not. Most community benefit regulations were adopted before 2009, when the IRS began to require federally tax-exempt hospitals to report community benefit spending. Although the similarities between the results of the OLS regression analysis and the IV analysis are reassuring, the absence of baseline data is a methodological limitation for any national study of the relationship between state-level community benefit regulation and hospital spending on community benefits.

Second, the number of states that have adopted regulatory requirements other than reporting is small. Further, the pattern of state adoption of these regulations has been such that a small number of states have the same combination of regulations. As a result, some study results may have been driven by the experience of only a few states, so the generalizability of the results is not clear.

Third, the community benefit information used in this study was limited to expenditure data reported by hospitals on IRS Form 990 Schedule H. The expenditures reported for community health initiatives, in particular, are comparatively small. In addition, for many hospitals, collecting data on such activities is a new activity. The community health initiative spending reported for the first years of the new reporting requirements likely contains errors, which may have affected our results. Moreover, the IRS does not currently require hospitals to provide more detailed information on the actual types of community benefit services and programs provided. As a result, the nature of the specific activities undertaken, the magnitude of their impact, and the identity of those who benefited from them all are unknown.

Fourth, while our study examined the differential effects of four types of community benefit regulations, for each individual type of regulation some variation exists among states in the scope of the regulation, the level of enforcement, and the sanctions for noncompliance. Moreover, some hospitals operate in states in which government agencies and hospital associations have established community benefit guidelines concerning hospital community benefit spending; such guidelines may influence hospital behavior, even if they do not carry the force of law (Ginn and Moseley 2006). Because such interstate differences may be relevant to hospital spending on community benefits, future research should consider extending our study by refining the indicators of state-level requirements used here.

Finally, the relationship between state-level community benefit regulations and hospital community benefit spending may have been influenced by recent federal requirements for nonprofit hospitals, specifically the new federal-level reporting requirement of IRS Form 990 Schedule H. This new requirement was in effect for all tax-exempt hospitals during this study's time frame, which may have affected the results. During the first few years of the new IRS requirement, however, Schedule H data were not well publicized or easy to obtain by community groups or other interested parties, so the schedule's transparency effects were likely still muted during our study period (Rubin, Singh, and Young 2015). Also, since 2012, the Affordable Care Act has required all federally tax-exempt hospitals to conduct a CHNA every three years. However, because our study focuses on the years 2009 to 2011, this provision likely did not have much of an impact on our results. It remains to be seen how the federal CHNA requirement will impact hospitals' longer-term community benefit spending.

Our study does raise important questions about the value of some forms of community benefit regulation that states have adopted. Of course, the regulations examined in this study are not the only policy options available to states interested in increasing hospitals' spending on community benefits. Several states have taken the aggressive step of revoking the property tax exemption of hospitals for failing to function as a charitable institution within general criteria for defining such institutions (e.g., Provena Covenant Med. Ctr. v. Dep't of Revenue, 925 N.E.2d 1131 [Ill. 2010]; AHS Hosp. Corp. v. Town of Morristown, 28 N.J. Tax 456 [2015]). In addition, more innovative approaches might also come under consideration. One approach proposed by Dranove, Garthwaite, and Ody (2015) addresses geographic mismatches between the supply and demand for charity care through state-level adoption of a community benefit cap-and-trade policy, which would allow hospitals to buy and sell credits to meet state-imposed standards for the provision of community benefits.

Conclusion

State-level community benefit regulation appears to have some role in influencing hospital spending on community benefits, but it may not always yield the effects policy makers and community leaders intend. Requirements to ensure hospitals regularly assess the need for community benefits may be more effective to meet policy goals than requirements that stipulate the provision of specific levels or types of benefits. At the same time, state adoption of multiple types of community benefit regulation may have the largest impact on hospital community benefit spending possibly because such intensive regulatory initiatives convey a strong signal to the hospital community that more spending is expected.

Appendix

References

References
American Hospital Association
.
2014
. “
Fast Facts on US Hospitals
.” www.aha.org/research/rc/stat-studies/fast-facts.shtml (accessed June 1, 2016).
Bunce
Victoria Craig
, and
Wieske
J. P.
2009
. “
Health Insurance Mandates in the States 2009
.” www2.cbia.com/ieb/ag/CostOfCare/RisingCosts/CAHI_HealthInsuranceMandates2009.pdf (accessed June 17, 2016).
Colombo
John D.
,
Griffith
Gerald M.
, and
King
James R.
2011
.
Charity Care for Nonprofit Hospitals: A Legal and Administrative Guide
.
New York
:
Aspen
.
Dranove
David
,
Garthwaite
Craig
, and
Ody
Christopher A.
2015
. “
A Floor-and-Trade Proposal to Improve the Delivery of Charity-Care Services by U.S. Nonprofit Hospitals
.” Hamilton Project,
October
. www.hamiltonproject.org/assets/files/floor_and_trade_improve_delivery_of_charity_care_services_dranove.pdf.
Fuse Brown
Erin C.
2015
. “
IRS Rules Will Not Stop Unfair Hospital Billing and Collection Practices
.”
American Medical Association Journal of Ethics
17
, no.
8
:
763
69
.
Ginn
Gregory O.
, and
Moseley
Charles B.
2006
. “
The Impact of State Community Benefit Laws on the Community Orientation and Health Promotion Services of Hospitals
.”
Journal of Health Politics, Policy and Law
31
, no.
2
:
321
44
.
Hanlon
Carrie
, and
Giles
Brittany
.
2012
. “
State Policymakers' Guide for Advancing Health Equity through Health Reform Implementation
.” Briefing.
Washington, DC
:
National Academy for State Health Policy
.
Kennedy
Frances A.
,
Burney
Laurie L.
,
Troyer
Jennifer L.
, and
Stroup
J. Caleb
2010
. “
Do Non-profit Hospitals Provide More Charity Care When Faced with a Mandatory Minimum Standard? Evidence from Texas
.”
Journal of Accounting and Public Policy
29
, no.
3
:
242
58
.
Nelson
Gayle D.
,
Somerville
Martha H.
,
Mueller
Carl H.
, and
Boddie-Willis
Cynthia L.
2013
. “
Hospital Community Benefits after the ACA: Policy Implications of the State Law Landscape
.” Issue brief.
Baltimore, MD
:
Hilltop Institute
.
Rosenbaum
Sara
, and
Margulies
Ross
.
2012
. “
Tax-Exempt Hospitals and the Patient Protection and Affordable Care Act: Implications for Public Health Policy and Practice
.”
Public Health Report
126
:
283
86
.
Rosenbaum
Sara
,
Kindig
David A.
,
Bao
J.
,
Byrnes
M. K.
, and
O'Laughlin
Colin
2015
. “
The Value of the Nonprofit Hospital Tax Exemption Was $24.6 Billion in 2011
.”
Health Affairs
24
, no.
7
:
1225
32
.
Rubin
Daniel B.
,
Singh
Simone R.
, and
Young
Gary J.
2015
. “
Tax-Exempt Hospitals and Community Benefit: New Directions in Policy and Practice
.”
Annual Review of Public Health
36
:
545
57
.
Singh
Simone R.
,
Young
Gary J.
,
Lee
Shoou-Yih
,
Song
Paula H.
, and
Alexander
Jeffrey A.
2015
. “
Analysis of Hospitals' Community Benefit Expenditures' Alignment with Community Health Needs: Evidence from a National Investigation of Tax-Exempt Hospitals
.”
American Journal of Public Health
105
, no.
5
:
914
21
.
Stoto
Michael A.
2013
. “
Population Health in the Affordable Care Act Era
.” Issue brief.
Washington, DC
:
Academy Health
.
Tahk
Susannah C.
2014
. “
Tax-Exempt Hospitals and Their Communities
.”
Columbia Journal of Tax Law
6
, no.
1
:
33
85
.
US Government Accountability Office
.
2005
.
Nonprofit, For-Profit, and Government Hospitals: Uncompensated Care and Other Benefits
.
May
26
.
Washington, DC
:
US Government Accountability Office
. www.gao.gov/products/GAO-05-743T.
Young
Gary J.
,
Chou
Chia-Hung
,
Alexander
Jeffrey
,
Lee
Shoou-Yih D.
, and
Raver
Eli
.
2013
. “
Provision of Community Benefits by Tax-Exempt U.S. Hospitals
.”
New England Journal of Medicine
368
, no.
16
:
1519
27
.