Context: Previous studies show that nonprofit hospital spending on charity care declined in Medicaid expansion states. We test whether state community benefit regulations mitigated the decline in charity care spending.
Methods: We use a fixed effects model to evaluate the association between state regulations and nonprofit hospital community benefit spending and its subcategories as a share of total expenses in Medicaid expansion states. We obtained community benefit spending data from the Internal Revenue Service Form 990 Schedule H filings of 1,738 hospitals in 44 states and the District of Columbia from 2010 to 2017. We determine the stringency of state regulations by comparing the provisions of state and federal requirements based on regulation information compiled by the Hilltop Institute.
Findings: State minimum community benefit requirements are associated with increased community benefit and charity care spending by nonprofit hospitals in Medicaid expansion states.
Conclusions: States that imposed minimum community benefit requirements on nonprofit hospitals did not experience a decline in charity care spending after Medicaid expansion. The results suggest state minimum community benefit rules may expand the provision of community benefit and charitable care spending.
Both the federal government and states use community benefit spending as a criterion for determining the tax-exempt status of nonprofit hospitals. However, recent studies by Herring et al. (2018) and Zare, Eisenberg, and Anderson (2021) suggest most nonprofit hospitals do not provide community benefits commensurate with the financial value of their tax exemption. The Affordable Care Act (ACA) imposed numerous new regulations on nonprofit hospitals aimed at increasing the provision and quality of community benefit spending. For example, all US nonprofit hospitals are now required to prepare a community health needs assessment that identifies and plans for the public health needs of the communities they serve. Hospitals must also establish financial assistance policies that specify the conditions that must be met for patients to receive free or discounted medical care.
These new federal requirements supplanted many extant state-level regulations of tax-exempt nonprofit hospitals. Currently, 33 states and Washington, DC, impose one or more regulations related to the reporting or provision of community benefit expenses (Nelson, Tan, and Mueller 2015; Somerville, Nelson, and Mueller 2013). In some cases, the stringency of a state-level hospital regulation exceeds that of its federal counterpart. For example, federal regulations require nonprofit hospitals to engage in community benefit spending to maintain their tax-exempt status but do not specify how much hospitals must spend. Five states separately mandate specific minimum community benefit spending thresholds that nonprofit hospitals must meet to fulfill their community benefit requirement. States also impose other regulations on community benefit spending and reporting that exceed the stringency of those implemented by the federal government. We group and summarize state regulations based on a comprehensive evaluation by the Hilltop Institute at the University of Maryland, Baltimore County, in table 1.
We examine the provision of community benefit spending by nonprofit hospitals following state participation in Medicaid expansion with attention to the role of state community benefit laws. Under the ACA, states are permitted to enroll adults with incomes of up to 138 percent of the federal poverty level (FPL) in Medicaid.1 As of 2020, 36 states and Washington, DC, opted to participate in Medicaid expansion (KFF 2022). Previous studies find nonprofit hospital charity care expenses decreased following state participation in Medicaid expansion (Camilleri 2018; Dranove, Garthwaite, and Ody 2016). Garthwaite et al. (2019) show that the decline in charity care spending likely reflects a reduction in the uninsured population in those states. The goal of this study is to determine the extent to which state regulations contributed to or mitigated the decline in charity care spending following state Medicaid expansions. We also examine the relationship between state regulations and other community benefit spending subcategories.
We accomplish this by identifying state-level hospital regulations that are more stringent than their federal counterparts imposed under the ACA. We then evaluate the association of these regulations with total community benefit spending and its component subcategories using a fixed effects regression model applied to Internal Revenue Service (IRS) Form 990 Schedule H data from 2010 to 2017. The analysis shows that state minimum community benefit spending rules are associated with increased community benefit spending by nonprofit hospitals in Medicaid expansion states. About one third of the increase in community benefit spending reflects growth in charity care spending specifically.
This study contributes to the literature by demonstrating that state regulations influenced the provision of community benefit spending and charity care by nonprofit hospitals in Medicaid expansion states. A unique feature of this study is that we consider the impact of state regulations that are more stringent than federal regulations imposed under the ACA. We begin with a brief overview of the extant literature on the impact of the Medicaid expansion and state regulations on community benefit spending.
State Medicaid Expansion and Community Benefit Spending
Several studies have evaluated whether state Medicaid expansions encouraged hospitals to reallocate resources from direct patient care to other population health services. Nikpay, Buchmueller, and Levy (2015) and Valdovinos, Le, and Hsia (2015) found that uncompensated care spending declined in nonprofit hospitals located in California and Connecticut, two early Medicaid expansion states. More recent studies found a broad reduction in uncompensated care in Medicaid expansion states nationwide (Camilleri 2018; Dranove et al. 2016). Earlier studies relied on cost report data from the Centers for Medicare and Medicaid Services (CMS), which does not disaggregate many hospital expenses. For example, the CMS data do not distinguish changes in spending on charity care from patient bad debt (uncollectable payments from patients), which the IRS does not categorize as a community benefit.
More recent studies used the IRS Form 990 Schedule H, which was added to nonprofit hospitals' annual filing requirements under the ACA and contains detailed community benefit spending information by subcategory. Many studies have shown that the decline in charity care resulting from the Medicaid expansion was partially offset by increased Medicaid expenses that were not reimbursed to nonprofit hospitals (Alberti, Sutton, and Baker 2018; Goodman et al. 2020; Stoecker et al. 2020; Valdovinos et al. 2015; Young et al. 2019). Kanter et al. (2020) showed that this pattern continued to hold through 2017, the most recent year for which Schedule H data are presently widely available.
Other studies show that charity care expenses were reallocated to other categories of community benefit spending. For example, Alberti, Sutton, and Baker (2018) found that nonprofit teaching hospitals in expansion states replaced their charity care spending with increased spending on other qualified community benefit spending, such as community health improvement services and research. Research by Rubin, Singh, and Young (2015), Singh et al. (2016), and Leider et al. (2017) similarly suggests that the universal state-imposed community health needs assessment (CHNA) requirement enacted under the ACA and separately enforced by many states may have encouraged nonprofit hospitals to devote more resources to broader preventive care services and community health improvement services. However, to our knowledge, no extant study considers how existing state community benefit regulations interact with state Medicaid expansions and whether they jointly influence spending on community benefits.
State Community Benefit Regulations
State community benefit rules typically differ in structure and purpose. For example, community benefit reporting requirements serve to increase transparency in information disclosure (Singh et al. 2018). Minimum community benefit requirements and mandatory income eligibility criteria for free and discounted care are intended to ensure that nonprofit hospitals provide a sufficient level of community benefit to justify their tax-exempt status (Singh et al. 2016). Required CHNAs ensure that the provision of community benefits meets the need of the community that the hospital serves (Alberti, Sutton, and Baker 2018; Rubin et al. 2015). Many earlier studies have evaluated the impact of one or more of these state regulations on total community benefit spending or various subcategories.2
Singh et al. (2018) considered the impact of several different state regulations on direct patient care services, community health initiatives, and total community benefit spending between 2009 and 2011.3 They found that state minimum community benefit requirements encouraged nonprofit hospitals to increase expenditures on community health initiatives while decreasing spending on direct patient care services and total community benefits. Johnson et al. (2019) also found that state minimum requirements had no significant association with total community benefit spending but were associated with an increase in financial assistance in some states. Kennedy et al. (2010) and Rothbart and Yoon (2021) separately found that the introduction of state minimum benefit laws in Texas and Illinois had no impact on the provision of charitable care, on average. Overall, there is presently little evidence that minimum community benefit regulations increase the total supply of hospital community benefit spending or charity care.
Singh et al. (2018) also found that state-imposed CHNA requirements were the only state regulation that was positively associated with spending on direct patient care services, community health initiatives, and total community benefit spending. Their analysis also indicates that community benefit spending rose in all three categories when states adopted more hospital regulations overall. They speculate that this result is driven in part by state regulations functioning as a signal to nonprofit hospitals about the importance of fulfilling their community benefit spending obligation. By contrast, Fos et al. (2019) found that the federal CHNA mandate had no impact on total community benefit spending in North Carolina nonprofit hospitals, and that spending on community health programs declined after preparation of a CHNA.
Several other studies indicate that state community benefit reporting requirements increased community benefit spending. For example, Young et al. (2013) found that state-level community benefit reporting requirements were the only regulation significantly and positively associated with the amount that hospitals spent on both patient care services (e.g., charity care, unreimbursed Medicaid expenses, and subsidized health services) and community services, which includes all other community benefits. Johnson et al. (2019) found that state community benefit reporting requirements increased community benefit spending, while Lamboy-Ruiz, Cannon, and Wantanabe (2019) separately found that state community benefit reporting requirements increased charity care spending. Both studies suggest reporting requirements are sufficient to induce additional spending even in absence of other regulations, such as minimum community benefit spending requirements.
In summary, states impose a variety of regulations on nonprofit hospitals to monitor and regulate the provision of community benefits associated with their tax-exempt status. Many earlier studies suggest that state community benefit rules have the potential to expand total community benefit or charitable care spending.4 However, there is presently little consensus about which state regulations are responsible for the hospital spending responses detected by these studies. This might reflect the fact that many earlier studies considered nonprofit hospitals in a single state or evaluated the response nationally using just two or three years of data in the period immediately after adoption of the ACA. Previous research also largely ignores new federal regulations imposed on nonprofit hospitals nationwide under the ACA that rendered many state regulations redundant. This study builds on the literature by using seven years of data to examine whether stringent state regulations mitigated or amplified any of the national spending patterns that emerged in Medicaid expansion states.
Data and Methods
To conduct our analysis we use community benefit spending information reported by nonprofit hospitals on the IRS Form 990 Schedule H from 2010 to 2017. The data were compiled by RTI International (2021) in the Community Benefit Insight database, and they contain information for nonprofit hospitals in all 50 states.5 We begin by excluding multistate hospital systems that are subject to multiple state regulations but file a single Schedule H. Following Camilleri (2018), we exclude hospitals in six states that expanded Medicaid eligibility at the state level using a Section 1115 waiver. These waivers permit states to modify coverage, financing, and other aspects of the program in addition to changing eligibility requirements. Finally, we also remove a small number of hospital-year observations after a hospital's merger into a multistate hospital system. These changes reduced the total observation count from 17,684 to 12,207, representing 1,738 unique Schedule H hospitals. The specific impact of each sample restriction is summarized in table 2.
After restricting the analytical sample, we compute each hospital's total community benefit expenses as a share of total expenses. We also decompose community benefit spending into five subcategories. Charity care expenses are incurred by providing free or discounted care to low-income patients. Unreimbursed Medicaid expenses reflect the cost of services provided to Medicaid-insured patients that were not reimbursed by the state Medicaid program and are therefore losses borne by the hospital. These losses arise from providing uncovered services to Medicaid-insured patients or when the Medicaid reimbursement rate is not sufficient to cover the expense of the service. Other patient care expenses include unreimbursed expenses from other means-tested programs and subsidized health services, which are losses incurred from clinical services. Community health expenses include community health improvement activities and administrative costs arising from community benefit operations as well as cash and in-kind contributions for community benefit. Finally, we consider education and research expenses that qualify as community benefit spending.
We winsorize each of the six outcome variables at the 1 and 99 percentiles to diminish the influence of severe outliers, and we change 1,210 hospital-year observations that have a negative value for a community benefit spending subcategory to zero. We report the summary statistics for the final analytical sample with a breakdown of those data between Medicaid expansion and nonexpansion states in table 3.6Figure 1 contains a map displaying Medicaid expansion and nonexpansion states as of 2016 as well as the six Section 1115 waiver states excluded from this analysis.
Comparison of State and Federal Community Benefit Rules
We review the specific provisions of the state-level community benefit regulations to determine their stringency relative to the federal requirement. We group state regulations into six categories based on regulatory data compiled by the Hilltop Institute. Specifically, those state regulations are:
Minimum community benefit requirement
Community benefit reporting requirement
CHNA with implementation plan or strategy requirement
Financial assistance policy requirement
Financial assistance policy dissemination requirement
Limitations on charges, billing, and collections
We provide broad descriptions of each regulation and which states impose the requirement in table 1.
Many existing state regulations were effectively duplicated by national requirements enacted under the ACA. For this purpose of this study, state policies are treated as more stringent than those enacted under the ACA if their specific requirements are more expansive than the federal counterpart. For example, although the federal government applies a community benefit standard to nonprofit hospitals, it does not specify that hospitals engage in a minimum level of community benefit spending annually. On the other hand, five states require nonprofit hospitals to justify their tax-exempt status by engaging in charity care or total community benefit spending that exceeds a percentage of their net revenues or that are commensurate with the dollar value of their annual state or local tax savings. We treat these five states as possessing a more stringent community benefit regulation than the federal government and summarize their specific rules in greater detail in table A1 of the online appendix.
In 2008, the IRS introduced the Form 990 Schedule H, which inquired about tax-exempt hospitals' spending on charity care, unreimbursed government-sponsored care, and other community benefit. Hospitals were expected to file a Schedule H with their 2008 Form 990. The following year, the IRS amended Schedule H to include additional information regarding compliance with the ACA's other community benefit rules (Rosenbaum 2015). Several states impose their own community benefit reporting requirement on nonprofit hospitals that in some cases are more rigorous than the federal counterpart. For example, Minnesota requires nonprofit hospitals to report the total dollar value and the number of charity care services that are provided to patients in three categories: (1) patients with household income at or below 275 percent of the FPL; (2) patients with household income above 275 percent of the FPL; and (3) patients whose household income cannot be determined. Under the federal rule, nonprofit hospitals must only report their gross charity care expenses on the Form 990 Schedule H. Based on the more extensive requirement, we assign Minnesota to the group of states with stringent reporting requirements, given that the state requires greater information disclosure from its nonprofit hospitals than the federal government does. In addition to Minnesota, we identified four other states that impose reporting requirements on the provision of community benefits that go beyond the requirements of Schedule H. We summarize those state rules in table A2 of appendix A.
The ACA also imposed additional reporting requirements on nonprofit hospitals. For example, nonprofit hospitals are now required to prepare a CHNA every three years, with input from community stakeholders. The federal requirement also mandates that an authorized body of the hospital must adopt an implementation strategy to meet identified community health needs. Unlike the ACA's other requirements, implementation of the CHNA rule was delayed until taxable years beginning after March 23, 2012. Several states also place their own regulations on the preparation or adoption of CHNAs. For example, New York requires hospitals to report their CHNA implementation strategy and performance in meeting community health needs every year. We treat New York as possessing a strict state-level CHNA requirement, as the state rule entails more frequent and detailed reporting than the federal counterpart. We identified eight states that similarly require hospitals to prepare a CHNA annually and/or require CHNAs to contain specific information, such as measurable objectives or implementation dates. Each stringent state CHNA rule is summarized in table A3 of the online appendix.
The ACA also requires hospitals to develop a financial assistance policy (FAP) that identifies which emergency and medically necessary care qualifies for hospital financial assistance and the eligibility criteria patients must meet to receive it. This rule went into effect with implementation of the ACA on March 23, 2010, and was applicable to tax years that began after that date. We identified 10 states that impose mandatory eligibility criteria for patients to obtain financial assistance. For example, Rhode Island's FAP requires nonprofit hospitals to provide free care for patients with annual incomes at or below 200 percent of the FPL. They must also provide discounted care for patients with annual incomes between 200 and 300 percent of the FPL. By contrast, the federal FAP regulation permits hospitals to set their own eligibility criteria for financial assistance. We categorize Rhode Island and other states that impose mandatory eligibility criteria for financial assistance in the group of stringent FAP states.
In addition to the ACA's new FAP requirement, nonprofit hospitals must also meet certain FAP dissemination regulations as of implementation of the ACA on March 23, 2010. Specifically, hospitals must publicize their FAP online, post it in specific locations in the hospital facility, and make a paper copy available upon request. We identified 14 states that impose additional FAP dissemination requirements on nonprofit hospitals. Nearly all of these require hospitals to provide direct notice of the availability of financial assistance in patient billing statements or other communication. These state rules are more stringent than the federal FAP dissemination rule, as the latter requires patients to actively request that information or search for it online or in the facility. We summarize stringent state FAP rules and FAP dissemination requirements in tables A4 and A5 of the online appendix.
Finally, several states impose stringent limitations on charges, billing, and collections compared to federal rules. The ACA requires hospitals to charge patients eligible for financial assistance no more than they generally charge insured patients. Hospitals may determine their own collection policies but must make reasonable efforts to identify patients who are eligible for financial assistance before engaging in collection actions. These provisions are applicable in tax years that began after enactment of the ACA on March 23, 2010. We identified 18 states that impose additional limits on charges, billing, and/or collections above and beyond those enacted under the ACA. For example, New Jersey hospitals cannot charge uninsured state residents with family income less than 500 percent of the FPL more than 15 percent above the federal Medicare payment rates. We treat New Jersey's state limitations on charges, billing, and collections as more stringent than the federal counterpart because of the state's mandatory eligibility criteria and because Medicare payment rates are generally much lower than private insurance or gross charges. Other state rules limit collection actions and/or the amount that may be collected from indigent patients. We summarize state rules pertaining to charges, billing, and collections in table A6 of the online appendix.
We completed this classification exercise using the Hilltop Institute's Community Benefit State Law Profiles compiled by Nelson, Tan, and Mueller (2015) or the source text of the state law if necessary and comparing it to the text of the Section 501 rules described on the IRS (2021) website. One noteworthy drawback to our assessments is that they largely ignore the degree to which state requirements are more stringent than federal regulations. Specific requirements vary across states, and some state-level requirements are more extensive than others.
where yit is one of the total hospital community benefit expenses, or one of five community benefit subcategories, as a share of the total expenses of hospital i in year t; Rst is a vector of binary variables equal to 1 for each stringent community benefit regulation imposed by state s in year t and 0 otherwise; Mst is a dichotomous variable equal to 1 in years t after state s participated in the Medicaid expansion; Xit is a vector of covariates specific to hospital i in year t;θs and λt are state and year fixed effects; and μst is the idiosyncratic error term. The model estimates a vector of coefficients, β, for each of the interaction terms. These coefficients report the relationship between each state regulation and total community benefit spending or its subcategories conditional on the participation of hospital i's state in the Medicaid expansion between 2010 and 2017. We estimate the model with standard errors clustered at the state level.
Earlier studies document numerous factors that contribute to nonprofit hospitals spending on community benefits outside the Medicaid expansion. For example, geographic location (e.g., rural vs. urban) and demographic factors (e.g., the uninsured rate in the county, the number of Medicare patients, and the number of low-income patients) are shown to influence community benefit spending (Blavin 2016; Camilleri 2018; Kaufman et al. 2016; Valdovinos et al. 2015).
The covariates included in the model above reflect several potential hospital- and community-specific determinants of community benefit spending. Specifically, we include the unemployment rate of each hospital's county as well as the proportion of the population younger than age 65 that is uninsured. We also include a measure of state hospital market concentration based on the number of hospital beds in each state calculated using the Herfindahl-Hirschman Index, the total number of hospital employees, and binary variables corresponding to whether the hospital has a teaching or church affiliation, and if it serves an urban area.
We report the baseline model estimation results in table 4. Column 1 corresponds to the relationship between state regulations and total community benefit spending, while columns 2 to 6 report the change in each subcomponent of total community benefit spending. All variables are expressed as the share of total hospital expenses. The first six rows of the table report the estimated change in community benefit spending based on the interaction of the relevant state regulation and state Medicaid expansion status as identified using an interaction term between the two binary variables in the regression model. The seventh row indicates the change in hospital community benefit spending based on state Medicaid expansion status. Given that each model tests the influence of multiple state regulations on community benefit spending, we evaluate the statistical significance of each result at only the 99% or 99.9% confidence thresholds.7
The estimation results show that the charity care expenses of nonprofit hospitals declined in expansion states. Specifically, the results indicate that nonprofit hospital charity care expenses declined by 0.8 percentage point (t = −3.09; p = 0.003) following state Medicaid expansion. The results also suggest the decline in charity care spending was offset by an increase in unreimbursed Medicaid expenses, but the coefficient is not statistically significant (t = 1.88; p = 0.067). Following earlier studies, the estimation results show little net change in total community benefit spending in Medicaid expansion states.
We next turn to the analysis of the effect of state regulations on community benefit spending in expansion states. The first estimate in column 1 of table 4 shows that total community benefit spending as a share of hospital expenses increased by 2.7 percentage points (t = 4.54; p < 0.001) following the Medicaid expansion in states that impose minimum community benefit requirements. On net, community benefit spending rose by 2.5 percentage points in Medicaid expansion states with a minimum community benefit requirement.8 This change reflects a 1.0 percentage point (t = 2.71; p = 0.009) increase in charity care spending as well as a 1.4 percentage point increase in unreimbursed Medicaid expenses (t = 3.52; p = 0.001). The results indicate that the effect of state minimum community benefit regulations fully offset the 0.8 percentage point estimated decline in charity care spending in Medicaid expansion states.
Consistent with Singh et al. (2018) and Johnson et al. (2019), the estimation results also indicate that state minimum community benefit requirements are associated with a statistically insignificant 0.5 percentage point (t = −2.04, p = 0.048) decline in total community benefit spending across all states. However, this result is driven primarily by a 0.6 percentage point (t = −3.18, p = 0.003) decline in unreimbursed Medicaid expenses. The estimation results also show that state minimum community benefit requirements are associated with a statistically insignificant 0.3 percentage point (t = 2.00, p = 0.051) increase in charity care spending across all states.
Most of the other state community benefit requirements considered in the analysis reported in table 4 had no significant association with community benefit spending. One exception is states that impose stringent CHNA requirements relative to federal rules. States with stringent state-level CHNA requirements experienced a 1.7 percentage point (t = 2.79; p = 0.008) increase in community benefit spending following state Medicaid expansion. This result was driven in large part by a 0.9 percentage point (t = 2.86; p = 0.007) increase in unreimbursed Medicaid expenses, as well as a statistically insignificant 0.7 percentage point (t = 1.70; p = 0.096) increase in charity care expenses. However, the net impact of stringent state CHNA requirements is close to 0 when considered with the 1.5 percentage point decline in total community benefit spending across all states that impose stringent CHNA requirements.
The analysis also shows that states with stringent financial assistance policies experienced a 2.3 percentage point (t = −4.58; p < 0.001) decline in community benefit spending following state Medicaid expansion. This result was driven primarily by a 1.5 percentage point (t = −5.43; p < 0.001) decrease in unreimbursed Medicaid expenses. Taken together, these results do not suggest that stringent financial assistance policy requirements at the state level encourage hospitals to engage in less community benefit spending. Unreimbursed Medicaid expenses are a function of services provided to Medicaid-insured patients and the reimbursement decisions of state Medicaid programs; thus, hospitals have limited ability to control their spending in this community benefit category.
In sum, the results of this analysis suggest that hospitals substitute between charity care and unreimbursed Medicaid expenses to fulfill their community benefit spending obligations. In contrast, hospitals in expansion states with a minimum community benefit requirement supplied more charity care even as their unreimbursed Medicaid expenses increased with the population of Medicaid-insured patients. State CHNA requirements also appear to the preserve the provision of community benefit spending in expansion states, though we do not find robust evidence that they are associated with increased charity care spending.
We also estimate several alternative specifications of the regression model above. First, we reestimate the baseline model with hospital fixed effects in lieu of state fixed effects and report the estimation results in table B2 of the online appendix. These estimation results generally support those obtained using the state fixed effect model with a larger number of statistically significant results. We also execute a separate version of the model with no state or hospital fixed effects and report the results in table B3 of the online appendix. This model continues to show that state minimum community benefit requirements are associated with increased community benefit and charity care spending. However, these results do not show that stringent CHNA requirements are associated with increased community benefit spending in Medicaid expansion states, as in the baseline analysis.
Next, we estimate the state fixed effects model on three restricted samples. Table B4 of the online appendix reports the estimation results corresponding to a panel with all 2010 Schedule Hs excluded. In other words, the panel is restricted to 2011 to 2017, reflecting the tax years when most of the ACA's community benefit provisions were effective. These results are largely consistent with those reported in table 4. We also execute a separate analysis with the hospital panel restricted to 2013 to 2017 Schedule Hs and report the estimation results in table B5 of the online appendix. This subsample aligns with the delayed start date of the ACA's CHNA requirement, which applied to tax years beginning after March 23, 2012. We again find results that support those in table 4 despite the reduced sample. Both analyses show statistically significant increases in community benefit and charity care spending in Medicaid expansion states that impose minimum community benefit requirements.
A potential drawback to the fixed effect research design is the staggered implementation of Medicaid expansion across states. Goodman-Bacon (2021) shows that staggered timing of policy interventions in a two-way fixed effects model may result in biased estimates, as early adopting states effectively act as control units for late adopters. We do not believe this poses a concern with respect to state regulations, as there was very little change in state community benefit rules over the 2010 to 2017 period. There was, however, considerable variation in implementation of the Medicaid expansion across states. As a robustness check, we narrow the panel to only nonprofit hospitals in the 18 states that expanded Medicaid in 2014 as well as the 19 nonexpansion states in the analytical sample. We report those results in table B6 of the online appendix. These results are once again consistent with those of the baseline model and other robustness checks that show increased community benefit and charity care spending in expansion states that impose minimum community benefit rules. We report the estimation results of other supplemental analyses in the online appendix.9
As with most other studies on this topic, we do not explore the causal mechanisms that underlie these results. One potential explanation is that all three Medicaid expansion states with minimum community benefit rules specifically mandate a minimum level of charity care spending. This is noteworthy because hospitals in those states could not easily reduce charity care spending in response to increased unreimbursed Medicaid expenses attributable to their state's Medicaid expansion. This hypothesis is supported by the empirical results showing that charity care, unreimbursed Medicaid, and total community benefit spending all increased in expansion states with minimum charity care expense requirements. However, earlier studies by Kennedy et al. (2010) and Rothbart and Yoon (2021) suggest that there is likely significant heterogeneity in hospital responses to state minimum charity care requirements underlying our finding. Based on a detailed review of the literature, Rozier (2020: 10) reports that “internal operations related to community benefit largely remain a black box.” More research on how state and federal community benefit regulations inform financial and managerial decision-making at the hospital level is needed.
This analysis also has a few noteworthy limitations. First, we are forced to exclude multistate hospital systems, which may be subject to a variety of state community benefit regulations. We exclude hospitals in the six states that participated in the Medicaid expansion using a Section 1115 waiver. Second, the precise nature of nonprofit hospital regulations varies considerably across states. We also treat the stringency of state-level regulations as a binary variable, which obscures exactly how stringent those requirements are relative to federal rules. Finally, Schedule H data are self-reported by nonprofit hospitals. We do not know the extent to which hospitals misreport their financial information to the IRS or manipulate the underlying cost of patient services billed to Medicaid or reported as charity care.
Discussion and Conclusion
Alberti, Sutton, and Baker (2018) and Young et al. (2018, 2019) found no significant change in total community benefit spending by nonprofit hospitals following state Medicaid expansions. Instead, these studies found evidence of changes in the composition of community benefit spending in expansion states. Nonprofit hospitals in these states reported fewer charitable care expenses and more unreimbursed Medicaid spending. Our analysis of nonprofit hospital financial data reported in the IRS Schedule H supports these earlier findings through 2017. Garthwaite et al. (2019) suggest these changes are driven by the transition of the uninsured into Medicaid following state participation in the Medicaid expansion. It is not clear, however, whether the shift in community benefit spending was strictly the result of changes in the insured status of patients or whether it resulted from a strategic reduction in charity care spending by nonprofit hospitals as their unreimbursed Medicaid expenses grew. The latter possibility is an important consideration given that recent research by Herring et al. (2018) and Zare et al. (2021) questions whether nonprofit hospitals provide community benefit that matches the value of the tax savings they presently enjoy.
This study fills a gap in the community benefit literature by considering possible interactions of state regulations with hospital spending in Medicaid expansion states. We find that state minimum community benefit requirements are associated with increased community benefit and charity care spending in Medicaid expansion states. Although this result contrasts with earlier findings on state minimum community benefit rules, earlier work does not consider the potential interaction of this regulation with the Medicaid expansion. Based on these results, current and future Medicaid expansion states are advised to consider adopting a minimum community benefit requirement. We also encourage future researchers to further examine how state and federal regulations jointly influence the provision of community benefits and hospital management decisions generally.
The authors thank Nathan Dietz, the editorial team, and two anonymous reviewers for helpful comments and suggestions. The authors received no financial support for the research, authorship, and/or publication of this article and have no potential conflicts of interest to declare.
Under the ACA, all states were expected to complete Medicaid expansion by January 1, 2014. States also gained the option to expand Medicaid before that date; six states opted for early expansion (California, Colorado, Connecticut, Minnesota, New Jersey, and Washington).
A few studies have examined how hospital-specific community benefit rules affect spending; however, because of the differences in model specification and data, there is little consensus on their impact. Tahk (2014) found that hospital adoption of an income limit for free and discounted care (e.g., patients with incomes below 350 percent of the federal poverty level can be eligible for free care) was associated with an increase in total community benefit expenses. By contrast, Goodman et al. (2020) found that hospitals with restrictive financial assistance policies, such as free and discounted care being restricted to those with incomes of less than 200 percent of the FPL are not associated with total charity care expenses.
Direct patient care services comprise all spending on charity care, Medicaid shortfall, and subsidized health services. Community health initiatives comprise the aggregated amount of expenditures on community health improvement services and cash and in-kind contributions to community groups. Total community benefit spending is the combined cost of direct patient care services, community health initiatives, and spending on research and health professional education.
Other research suggests that state community benefit regulations may influence other hospitals that are not subject to those rules. Ginn and Moseley (2006) and Ginn, Shen, and Mosley (2009) found that state regulations encouraged investor-owned hospitals and nonprofit hospitals in nearby states to engage in additional community-orientated activities.
This study is exempt from institutional review board approval because it is not considered human participant research.
Table B1 of appendix B reports the summary statistics for 1,489 nonprofit hospitals that filed a Schedule H before Medicaid expansion in 2010. This also corresponds to the first year of the Community Benefit Insight panel.
These confidence thresholds were determined using the Bonferroni correction method described by Lee and Lee (2018). This method entails dividing the significance threshold (α) by the number of hypotheses tested (k) to obtain the adjusted alpha. Under six hypothesis tests, the adjusted significance threshold of a standard 95% confidence test is (α / k = 0.05 / 6 = 0.0083), which rounds to an adjusted significance threshold of p < 0.01, equivalent to the 99% confidence threshold. The adjusted significance threshold under 36 hypothesis tests is (α / k = 0.05 / 36 = 0.00139), which rounds to p < 0.001, equivalent to the 99.9% confidence level.
This result reflects the estimated 2.7 percentage point increase in community benefit spending net of the 0.2 percentage point decline across all Medicaid expansion states.
Specifically, we estimate two additional models that separately consider the influence of state Medicaid expansion status and state regulations without the interaction terms in the baseline model. Table B7 reports the estimated impact of state Medicaid expansion status on community benefit spending. Table B8 reports the impact of stringent state regulations and Medicaid expansion status without the interaction terms.