ArticlesVol. 6, No. 1

Tax-Exempt Hospitals and Their Communities

Susannah Camic Tahk

Assistant Professor of Law, University of Wisconsin Law School
Citation:6 Colum. J. Tax L. 33


Hospitals in the U.S. have long been able to obtain exemption from federal income tax because they meet the requirement known as the standard of “community benefit.” Yet lawmakers and scholars know virtually nothing about the actual workings of tax-exempt hospitals, or about whether, how, and to what extent they deliver benefits to their communities. Within the last five years, however, IRS tax return forms have started asking hospitals to quantify these benefits, as well as to give detailed information about their financial practices with respect to their patients. These new questions coincide with new requirements for tax-exempt hospitals put in place as part of the 2010 Affordable Care Act. The new tax return data offer a first-time opportunity to evaluate the workings of tax-exempt hospitals from the perspective of both the traditional requirements for tax-exempt hospitals and the 2010 healthcare reforms of the Affordable Care Act.

This Article analyzes data from all tax-exempt hospitals in the U.S. in 2012 to show that tax-exempt hospitals differ widely in their provision of community benefits (and financial practices). In particular, these activities vary systematically in relation to their different notions of “community” and the characteristics of the communities where the hospitals are located. This evidence demonstrates that tax-exempt hospitals seem to be responding to the specific needs of their own communities when allocating their resources among different community-benefit activities. The data show, in addition, that while tax-exempt hospitals are generally adopting the financial policies that Congress and the IRS are requesting, hospital financial aid policies also vary by community. These findings raise several fundamental questions for lawmakers and tax policy scholars in the era of the Affordable Care Act. In particular, the findings suggest that lawmakers need to grapple seriously with how they allow tax-exempt hospitals to define their communities. For example, is it appropriate for tax-exempt hospitals merely to benefit a narrowly defined community or should they operate in terms of a broader understanding of community? In light of the new data presented, this Article considers these questions and outlines several alternatives to the “community benefit” standard to address them.



I. Introduction

At the present time, half of U.S. hospitals are exempt from federal income tax. To merit that exemption, hospitals must, under current tax law, benefit their communities. This tax exemption for hospitals is worth approximately $12 billion a year and allows hospitals to raise $5.3 billion in tax-deductible contributions annually.[1] As a result, the exemption plays a key role in providing health care in the U.S. However, the legal framework that allows hospitals to earn their tax exemption, known as the “community benefit” standard, has long been controversial. Its critics have decried the standard as overly vague, and they have argued that it does not distinguish between tax-exempt hospitals and their for-profit counterparts.

However, between 2008 and 2010, Congress and the IRS, for the first time in decades, revisited the legal framework for tax-exempt hospitals and began instituting changes. The Patient Protection and Affordable Care Act, better known as the health care reform bill (the “Affordable Care Act,” or the “ACA”), Pub. L. 111-148, 124 Stat. (2010), has now put some of these changes into place. In doing so, the ACA has catapulted institutions that have never before been in the national limelight into the center of political debate. Suddenly, tax-exempt hospitals have become a focal point of federal lawmaking efforts.

Having lived so much of its previous history on the periphery, however, the tax-exempt hospital sector is, for the most part, a virtual black box. As political leaders, legislators, pundits, and tax scholars debate the future provision of health care, they have consequently been left to guesswork about tax-exempt hospitals.

Proceeding in the absence of adequate data about the tax-exempt hospital sector, Congress and the IRS have, nevertheless, made dramatic changes to the rules that tax-exempt hospitals must follow. Already in 2008, the IRS began to compel tax-exempt hospitals to provide concrete data on the ways in which they (purportedly) worked to benefit their communities. On this new IRS “Schedule H,” which tax-exempt hospitals are now required to fill out each year as part of the tax filing process, hospitals must now quantify their specific community-enhancing projects. In addition, they must provide detailed data on a variety of different practices that govern how they interface with their communities. For instance, hospitals must now describe to the IRS each year what types of financial aid they make available and how they attempt to collect debts from patients who do not pay bills in full.

Furthermore, as part of the Affordable Care Act, Congress has set forth an additional set of requirements that both mandates and prevents certain activities on the part of tax-exempt hospitals. These new rules govern how tax-exempt hospitals may bill patients for services, offer financial aid, collect debts, and solicit information from communities about their needs. In the past several years, the IRS promulgated draft regulations under these new rules, which the agency plans to finalize this year.

These changes present a first-time opportunity to analyze the workings of tax-exempt hospitals on the basis of comprehensive empirical evidence. American hospitals have been exempt from federal income tax since the tax’s beginnings more than a century ago, but the effects of this costly exemption have remained unknown. However, the new Schedule H data finally provide a long-overdue way of examining what hospitals are actually doing to merit their exemption. Answers given by hospital administrators to the Schedule H questions enable us to see how well the traditional and revised legal frameworks are working and how they are affecting different types of hospitals around the country. They enable us to ask what hospitals are doing to benefit their communities. Are hospitals in fact responding to the needs of their communities? What kinds of financial aid policies have hospitals enacted? How generous are they? How do hospitals bill their patients and collect on those bills?

The data analysis presented in this Article takes a first step toward answering these questions. The analysis is based on Schedule H data for all tax-exempt hospitals in the U.S. in 2012. Using these data, the Article shows that the ways in which hospitals satisfy the requirements of the current law varies significantly by community. The community benefit standard appears to mean different things to hospitals in different communities. Some hospitals focus on providing community benefits as tax law has traditionally defined them: free or discounted care including Medicaid shortfalls, subsidized clinics and other direct-health interventions, education for aspiring health professionals, and medical research. Hospitals that provide more community benefits tend to be large and located in more densely populated communities with populations living just above the poverty line. Other hospitals emphasize what the IRS calls “community building” activities. These activities center on what public health scholarship calls the “social determinants of health.” Among them are improved housing, economic conditions, environmental factors, and jobs. Hospitals that focus on community building also tend to be large, but are located in communities where residents are more likely to have private insurance.

In addition to uncovering community patterns in the services that hospitals provide, the Schedule H data also highlight differences in the types of community-related practices in which hospitals engage. Across all communities, hospitals are now uniformly adopting the financial aid and debt collection practices that Congress and the IRS will soon be requiring. However, the specifics of those practices also vary by community. In particular, the availability of financial aid for care diverges across communities. In densely populated communities, where insurance rates and incomes are high, often in states that themselves have relatively stringent laws for tax-exempt hospitals, hospitals are more likely to offer free or discounted care to patients at higher income levels. The data indicate further that the types of debt collection practices that have long worried lawmakers are also associated with certain community traits. In particular, small hospitals in predominantly white communities are more likely to use what the IRS views as “extraordinary” debt collection actions.

Taken together, these findings have major implications for lawmakers seeking to evaluate the legal framework for tax-exempt hospitals in terms of both its traditional and new ACA components. In particular, this research suggests that, at the current time, hospitals are responding primarily to the needs of their immediate communities. Under the current legal standard, which gives hospitals broad latitude to define and decide how to improve their own communities, hospitals are behaving in accordance with the law. The tax policy question then becomes whether is it appropriate merely to ask hospitals to enhance their own communities as they see them. What effect does the current standard for tax exemption have on poor and disadvantaged communities? Should the federal tax law impose a broader standard of community on all hospitals, requiring them to take into account the needs of those outside of that community? This Article weighs several proposals that might either expand the legal definition of community or require hospitals to move beyond it.

Readers may be surprised that questions such as whether hospitals bill needy patients for their medical care are matters that fall within the scope of federal tax law. The fact that the IRS is the agency responsible for deciding whether a hospital can place a lien on a patient’s house may seem counterintuitive. However, using tax law to regulate social policy matters such as free health care and debt collection constitute a major part of a larger recent trend toward using tax law to conduct social policy.[2] In particular, Congress is increasingly relying on the tax code to fight poverty and to meet the needs of poor and near-poor individuals. Setting rules about how hospitals relate to disadvantaged patients and communities is one significant example of this phenomenon.

This Article proceeds in six parts. Part II details the current legal framework for tax-exempt hospitals, explaining what its requirements are, how these developed, and how they have been modified under the ACA. Part III reviews the literature on this subject. It then describes the data and methods on which the empirical analysis in this Article is based. Part IV presents the results of the data analysis, and Part V discusses these. Part VI considers the tax lawmaking implications of the data analysis, and Part VII concludes.

II. Legal Framework for Tax-Exempt Hospitals

The current legal framework for tax-exempt organizations has three components as it pertains to hospitals. First, tax-exempt hospitals must follow the general IRS requirements for tax-exempt organizations. Second, tax-exempt hospitals must comply with a series of hospital-specific rules that have developed over the past hundred years, particularly the controversial community benefit standard. Third, tax-exempt hospitals must follow the new requirements set forth in the Affordable Care Act and the regulations promulgated thereunder. This Part briefly describes each of these components and discusses some of the common criticisms of how the law treats tax-exempt hospitals.

A. General Requirements for Tax-Exempt Organizations

Organizations “organized and operated” for certain purposes may be exempt from federal income tax.[3] The most high-profile group of exempt organizations, and the group into which most tax-exempt hospitals fall, consists of “charitable, religious, educational, and scientific entities,” otherwise known as “charitable organizations,” “charities,” or “§ 501(c)(3) organizations.”[4]

Organizations “organized and operated” for one of these purposes receive two major tax benefits. First, these charities do not generally have to pay federal income tax on their net income. Second, individuals and corporations may deduct, within certain limits, donations to these charities.[5] When discussing “tax-exempt organizations,” this Article will hereafter be referring to charities.

A charity must be “organized and operated” exclusively for purposes that are “religious, charitable, scientific” in nature or include “testing for public safety, literary, or educational purposes, or to foster national or international amateur sports competition . . . , or for the prevention of cruelty to children or animals” as amplified in subsequent regulations.[6] Notably, this list does not mention health care. The hospital-specific rules described in Part I.b are what make § 501(c)(3) applicable to hospitals. The text of § 501(c)(3) also includes the “organizational” and “operational” tests for charitable status,[7] which govern the operating documents of a charity as well as its daily operations.[8]

In addition to the organizational and operational tests, a charity must observe a complete ban on “private inurement.”[9] This rule prevents a charity from distributing any of its assets as profits to shareholders.[10] The charity must also comply with the ban on “private benefit.”[11] To do that, in addition to serving its above-described charitable purpose, the organization must “serve[] a public rather than a private interest.”[12] This means that the organization must benefit the broader public, rather than any particular individual or narrowly defined small group. Two additional requirements for charities set limits on political activities and mandate compliance with the “public policy” doctrine, which stands for the general rule that charities may not operate in ways that run contrary to “public policy.”[13] Finally, a charity’s activities may not become overly “commercial.”[14]

Hospitals exempt from tax under § 501(c)(3) must comply with all of these general rules for organizations that are exempt under that provision. In addition, as the next two subparts will discuss, tax-exempt hospitals must follow a series of hospital-specific rules.

B. Traditional Hospital-Specific Requirements for Tax-Exemption

1. Nature of Traditional Hospital-Specific Requirements

Hospitals have been able to qualify for tax exemption according to longstanding historical practice, later set forth explicitly in IRS guidance. The following subpart will briefly trace the development of the rules that make tax-exempt status available to certain hospitals.

The federal income tax law that Congress passed in 1894 allowed certain “charitable” organizations to be exempt from tax. This initial definition emphasized charitable expenses that were incurred for various categories of poor people. At the time, hospitals, particularly larger hospitals, which had their roots in almshouses, did in fact serve as refuges for the poor.[15] As a result, after the 1894 statute and its successor, the 1913 income tax statute, passed, standard IRS practice treated hospitals as charitable and consequently as eligible for tax exemption.[16]

However, as the twentieth century progressed, hospitals broadened their patient pool to include more non-indigent patients and came to rely more heavily on paying patients.[17] To address the problem of whether hospitals could still qualify for tax exemption, in 1953 the IRS issued administrative guidance, Rev. Rul. 56-185. In this ruling, the IRS held that hospitals could be exempt from tax if they met several criteria.[18] Most notably, hospitals had to be “operated to the extent of [their] financial ability for those not able to pay for the services rendered and not exclusively for those who are able and expected to pay.”[19] The ruling clarified that, although they could charge patients for services if the patients were able to pay, hospitals would have to use the revenues so earned to defray operating expenses while not denying care to individuals unable to pay.[20]

In the 1960s, however, Congress passed two programs designed to pay directly for health care for the needy: Medicare and Medicaid.[21] IRS staff believed that because Medicare and Medicaid would now cover the health care costs of the poor, hospitals should no longer have to do so.[22] One staffer later told researchers that “officials at ‘other agencies’ convinced him that hospitals would only care for the poor if they participated in Medicare and Medicaid.”[23] As a result, he “concluded that existing tax law, with its requirement of free or below-cost care, was obsolete.”[24]

Responding to these changes in the health care landscape, the IRS in 1969 issued a consequential new ruling, Rev. Rul. 69-545,[25] in which, the IRS adopted what has come to be known as the community benefit standard.[26] Under this standard, hospitals could qualify for tax exemption even if they did not offer care to patients unable to pay. In fact, Rev. Rul. 69-545 explicitly “remove[d]” the earlier ruling’s “requirements relating to caring for patients without charge or at rates below cost.”[27] Instead, the 1969 ruling stated that, to merit exemption, hospitals must provide services beneficial to their communities. The ruling went on to provide an illustrative and non-exhaustive list of factors that might distinguish tax-exempt hospitals from their for-profit counterparts. These distinguishing factors included: (1) a community board, (2) an emergency room available even to patients unable to pay, (3) a medical staff open to all doctors and (4) a willingness to treat recipients of Medicare and Medicaid.[28]

From that time to the present, Rev. Rul. 69-545 and its community benefit standard have defined the basic legal requirements for tax-exempt hospitals, as they do to this day. The IRS clarified the 1969 ruling somewhat in 1983, holding that tax-exempt hospitals need not provide emergency care to qualify for tax-exempt status.[29] Since then, the IRS has issued some additional guidance that provides insight into the agency’s understanding of the community benefit standard. Perhaps most surprisingly, in 2001 the IRS issued nonprecedential internal guidance for field agents auditing tax-exempt hospitals that sheds further light on the agency’s views about the community benefit standard.[30] The guidance suggested that, Rev. Rul. 69-545 notwithstanding, the IRS believes that the community benefit standard does obligate tax-exempt hospitals to provide a certain amount of care for the poor.[31] Further, in another internal document, a tax policy update issued in 2002, the IRS indicated that absent an emergency room, a hospital policy regarding charity care is a “highly significant factor” in determining whether the hospital meets the community benefit standard, and that even without a charity care policy, exempt hospitals should provide some free or discounted care.[32] Although the 2002 document is not binding legal authority, it does suggest that, when evaluating whether hospitals qualify for exemption, the IRS might in practice examine hospitals’ practices regarding patients who are unable to pay.

2. Criticisms of Traditional Requirements for Tax-Exempt Hospitals

The IRS’s longstanding approach to regulating tax-exempt hospitals, as embodied in the community benefit standard, has been controversial since the IRS first set it forth in Rev. Rul. 69-545. While the criticisms have touched on a variety of issues, several interlinked themes have emerged. In particular, critics have alleged that the community benefit standard is overly vague and that it does not impose sufficient affirmative duties on tax-exempt hospitals. The community benefit standard, many observers have argued, does not adequately distinguish tax-exempt hospitals from their for-profit counterparts. Relatedly, the community benefit standard does not differentiate between tax-exempt hospitals that provide significant financial aid to patients and tax-exempt hospitals that offer very little, or perhaps none at all. As a result, the standard allows hospitals to diverge substantially in terms of how they treat patients who are unable to pay. In recent decades, critics of the community benefit standard have also contended that it permits financial policies that are not appropriate for tax-exempt hospitals. The following Part 1.b.ii will briefly summarize some of these common criticisms.

One frequent attack on the community benefit standard centers on its vagueness, or the fact that it simply does not provide hospitals with enough guidance about what they can and cannot do to stay exempt. To take just a few examples, health law professor Mary Crossley has written that “the vagueness of the existing federal community benefit standard and its historically lax enforcement mean that we do not really know what or how much beneficial conduct flows from the tax exemption and its forgone revenue, or whether that conduct is closely related to improving access and health outcomes for the uninsured or other groups.”[33] Similarly, health policy analysts Corey Davis, Jessica Curtis, and Anna Dunbar-Hester have written that, lacking “clear or consistent laws governing the requirements for achieving and maintaining nonprofit status, hospitals have largely been left to determine for themselves what activities qualify as community benefit.”[34] Lawyer Cecilia Jardon McGregor has also argued that “[t]he lack of specific criteria has been identified as a major problem concerning tax exemption for non-profit health care organizations.”[35]

In addition to the vagueness critique, many commentators on the community benefit standard have argued that it does not sufficiently distinguish between tax-exempt hospitals and their for-profit counterparts. In practice, these critics say, most for-profit hospitals could satisfy the community benefit standard just as easily as a tax-exempt hospital. Most famously, testifying before the Ways & Means Committee in 2005, then-IRS Commissioner Mark Everson stated, “What we have seen since 1969 has been a convergence of practices between the for-profit and nonprofit hospital sectors, rendering it increasingly difficult to differentiate for-profit from not-for-profit health care providers.”[36]    Making the same point while arguing that the IRS should replace the community benefit standard with a legal standard focused on increasing access to health care, legal scholar and expert on tax-exempt hospitals John Colombo has observed that tax-exempt hospitals generally charge for providing health care to nearby communities, which “is exactly what for-profit hospitals and other providers do.”[37] Law and public health professor Jessica Berg has similarly contended that, like tax-exempt hospitals, “[f]or-profit hospitals also provide charity care, assume some bad debt, and may have shortfalls in compensation from government programs; thus, there are serious questions about whether these categories function as an appropriate gauge of community benefit to justify tax-exempt status.”[38]

Related to the failure of the community benefit standard to distinguish between tax-exempt and for-profit hospitals is the standard’s inability, according to some critics, to differentiate between those hospitals that supply substantial assistance for financially troubled patients and those hospitals that do not. In 2009, the IRS itself surveyed 544 tax-exempt hospitals about the community benefits they provided and, in its report, raised this issue.[39] The IRS found that “[t]here was considerable diversity in the demographics, activities, and financial resources among the respondent hospitals.”[40] Further, the IRS identified a small subgroup of tax-exempt hospitals that seemed to be supplying most of the free or discounted care and other types of community benefits, observing that “[u]ncompensated care and aggregate community benefit expenditures were unevenly distributed among hospitals and concentrated in a relatively small group.”[41] Along similar lines, in 2008 the General Accounting Office issued a report showing that different hospitals measure their community benefits in substantially different ways, thereby producing substantially different results.[42]

Echoing the findings of the IRS and GAO reports, Professor Berg has also observed that tax-exempt hospitals account for free and discounted care through procedures that vary significantly in how generous they are to patients unable to pay.[43] Along similar lines, writing in the Temple Law Review, health care lawyer Leah Snyder Batchis described a series of (unsuccessful) lawsuits against tax-exempt hospitals regarding their refusal to give free or discounted care to uninsured patients.[44] In these lawsuits, the plaintiffs argued that these hospitals, while complying with language of Rev. Rul. 69-545, actually violated the more general requirement that tax-exempt organizations serve the public interest.[45]

Another common critique of the community benefit standard alleges that it allows hospitals to engage in financial practices that are inappropriate for tax-exempt organizations. These practices primarily include charging inflated rates to uninsured patients and then aggressively attempting to collect those patients’ debts. This critique emerged from a series of articles in the Wall Street Journal in 2004.[46] These articles documented how tax-exempt hospitals were often charging uninsured patients rates off of a maximum-price “chargemaster” price schedule.[47] Government and private insurance companies would negotiate substantial discounts off of those rates for covered patients.[48] However, uninsured patients, unable to negotiate discounts with hospitals, would receive bills for the entire amounts.[49] The Wall Street Journal series further documented how, when the patients were not able to pay, the tax-exempt hospitals would move to collect those debts, often with the help of private collection agencies, sometimes using practices such as garnishing wages, placing liens on houses or cars or arresting patients.[50]

Following those articles, both the Senate Finance Committee and the House Ways and Means Committee held hearings to probe these problems.[51] Much of the testimony at these hearings was highly critical of tax-exempt hospitals for their billing practices. For example, the executive director of a Virginia legal services organization told stories of clients who had received inflated bills even from tax-exempt hospitals that had financial aid policies, but failed to make patients aware of those policies.[52]

In response to these and similar criticisms, lawmakers have proposed several changes to the legal framework for tax-exempt hospitals. In particular, legislators have developed several proposals. First, in the early 1990s, in connection with a series of hearings and a GAO report very similar to the 2008 IRS report, two members of Congress introduced legislation to tighten and make more specific the rules for tax-exempt hospitals. Representative Edward Roybal’s plan would have mandated that tax-exempt hospitals maintain “open door” policies, and spend 50% of the value of their tax exemptions on unreimbursed charity care and 35% on unspecified “community benefits.”[53] Representative Brian Donnelly’s bill would have obligated tax-exempt hospitals to provide uncompensated care of at least 5% of their annual gross revenues.[54] In the alternative, tax-exempt hospitals could maintain exemptions by serving as the only hospitals in their communities, taking certain percentages of patients on Medicare or Medicaid, or devoting 10% of their gross revenues to “qualified services to the community.”[55]

The first decade of this century also saw two legislative proposals regarding tax-exempt hospitals. In 2006, Representative Bill Thomas introduced a bill that would have required tax-exempt hospitals to charge no more than $25 per medically necessary visit to patients with annual household incomes up to 100% of the federal poverty line. Additionally, tax-exempt hospitals would have been unable to charge patients whose household incomes were between 100% and 200% of the federal poverty line more than health insurers’ average rate for care.[56] Then, in 2007, Senator Chuck Grassley released a “discussion draft” of potential legislative proposals that he was considering—a draft that would have large ramifications. The draft included a series of new rules for tax-exempt hospitals, among them, changes to hospitals’ financial policies and a mandatory charity care minimum equal to 5% of revenues.[57]

Of these various proposals to reform the legal framework for tax-exempt hospitals, the Roybal, Donnelly and Thomas plans never moved beyond draft stage. The Grassley discussion draft, however, while never actually becoming law, set the stage for the new rules for tax-exempt hospitals contained in the Affordable Care Act. As subpart II.C will now describe, the law that eventually passed adopted some form of Senator Grassley’s recommendations both about financial aid policies and about “community health needs assessments”—a consequential idea that was new to the discussion. Notably, the ACA did not include any quantitative thresholds for charity care or community benefits.

C. Additional Hospital-Specific Requirements for Tax-Exemption under the Affordable Care Act

In 2010, as part of the ACA, Congress passed new legislation governing the behavior of tax-exempt hospitals. These rules followed in the steps of Senator Grassley’s discussion draft. The new rules concerned hospitals’ financial policies and methods of assessing their communities’ needs. The following subpart II.C.1 will briefly describe the ACA’s framework regarding tax-exempt hospitals, including the related draft regulations that the IRS and the Treasury Department have subsequently promulgated.

1. Community Health Needs Assessments

First, the Affordable Care Act required tax-exempt hospitals to conduct, every three years, a “community health needs assessment” (a “CHNA”).[58] Under these new rules, the hospital must also adopt an implementation strategy to meet the community health needs identified through the CHNA.[59] The legislation specifies that the CHNA must take into account input from persons who represent the broad interests of the community “served by the hospital facility” including those with special public health expertise.[60] The hospital must publicize the CHNA “widely.”[61] Hospitals that fail to meet this requirement must pay a $50,000 excise tax.[62]

In April of 2013, proposed regulations came out regarding the community health needs assessment requirement.[63] (The IRS and the Treasury Department plan to publish final regulations by the end of 2014.)[64] Because some version of the proposed regulations will soon go into effect, they merit special attention for the purposes of this Article. The proposed regulations address a number of issues emerging from the new statute. Of particular importance, the IRS and the Treasury Department considered how a hospital should identify its “community” for the purposes of assessing these communities’ needs. Specifically, the proposed regulations

provide a hospital facility with the flexibility to take into account all of the relevant facts and circumstances in defining the community it serves, including the geographic area served by the hospital facility, target populations served (for example, children, women, or the aged), and principal functions (for example, focus on a particular specialty area or targeted disease).[65]

The earlier draft regulations had noted that, “the Treasury Department and the IRS would expect a hospital facility’s community to be defined geographically but that, in some cases, the definition might also take into account target populations served or specialized functions.”[66] Consequently, the earlier regulations had requested “comments” on whether the IRS and Treasury “should define the geographic community of a hospital facility as the Metropolitan Statistical Area (MSA) or Micropolitan Statistical Area (µSA) in which the facility is located or . . . the county in which the facility is located.”[67] However, the later proposed regulations explained that many of the comments the IRS and Treasury had received supported a “facts-and-circumstances approach” to defining community and “recommended against a definition based on specified geographic boundaries.”[68] Advocates of this more flexible approach observed that, “each hospital facility is in the best position to determine its community.”[69] Politically defined boundaries such as MSAs or counties might not, according to the comments, accurately represent the group the hospital serves.[70]

On the other hand, the proposed regulations do express concerns, apparently shared in some comments to the earlier draft, about “ensuring that hospital facilities assess and address the needs of medically underserved, low-income, and minority populations in the areas they serve.”[71] In response, the proposed regulations specify that a hospital facility may not “define its community in a way that excludes medically underserved, low-income, or minority populations who are part of its patient populations, live in geographic areas in which its patient populations reside or . . . otherwise should be included” based on the hospital’s selected definition of community.[72] Hospital facilities can only exclude these groups from their preferred definition of community if “they are not part of the hospital facility’s target populations or affected by its principal functions.”[73] The proposed regulations define “medically underserved populations” as those “experiencing health disparities or at risk of not receiving adequate medical care as a result of being uninsured or underinsured or due to geographic, language, financial, or other barriers.”[74]

The proposed regulations go on to give more detail about what the CHNAs and associated required implementation strategies must involve. The regulations also provide substantial information about how hospitals must document their CHNAs and implementation plans.[75] For example, hospitals must “identify the organizations that provided input into the CHNA and summarize the nature and extent of that input,” including a description of “the medically underserved, low-income, or minority populations being represented by the organizations or individuals providing input.”[76]

2. Financial Policies

The Affordable Care Act also contains provisions relating to the financial policies of tax-exempt hospitals. Under new Sec. 501(r)(4), tax-exempt hospitals must establish specific types of written financial assistance policies and written policies relating to emergency medical care.[77] Additionally, the ACA’s new rules for tax-exempt hospitals and their financial policies limit these hospitals’ ability to charge uninsured patients at inflated rates. In particular, under the ACA, a tax-exempt hospital may not charge to the bills of financial-aid-eligible patients amounts for medically necessary care that are greater than the “amounts generally billed” (“AGB”) to insured patients.[78] Further, tax-exempt hospitals may not bill at chargemaster rates.[79] Then, when a tax-exempt hospital moves to obtain payment from patients, the hospital must “make reasonable efforts” to determine whether an individual is eligible for financial aid before engaging in “extraordinary collection actions” against the individual.[80]

In 2012, the IRS and the Treasury Department issued proposed regulations under these statutory provisions.[81] Again, the IRS has since announced its intent to finalize these regulations by the end of 2014.[82] The proposed regulations deal with a number of issues arising under the new statutory language about financial policies. For one, the proposed regulations clarify that neither they nor the statute restrict the substance of tax-exempt hospitals’ financial assistance policies. No law “mandate[s] any particular eligibility criteria” for assistance.[83] The proposed regulations do set forth the steps a tax-exempt hospital has to take to publicize and implement its financial assistance policy.[84] These regulations also take up the issue of how tax-exempt hospitals must calculate charges for financial-aid-eligible patients.[85]

Turning to the problems of bill payment, the proposed regulations cover in detail permissible debt collection activities under the ACA requirements. Grappling with a controversial issue, the Treasury Department and the IRS have set forth practices that constitute “extraordinary collection actions.”[86] These include any “actions taken by a hospital facility against an individual related to obtaining payment of a bill . . . that require a legal or judicial process.”[87] The proposed regulations list as examples placing a lien on an individual’s property, foreclosing on an individual’s real property, attaching or seizing a bank account or other personal property, commencing a civil action, causing an arrest, subjecting an individual to a writ of body attachment, and garnishing wages.[88] Extraordinary collection actions also include reporting a debt to a credit agency and selling a debt to a third party, even though neither of these things may require a legal or judicial process.[89] However, merely turning a debt over to a collection agency without actually selling the debt does not count as an extraordinary collection action.[90] Neither does refusing care to a patient because the patient has previously failed to pay a bill in full.[91] The regulations clarify in great detail what “reasonable efforts” the hospital has to take to determine financial aid eligibility that tax-exempt hospitals have to take before they can engage in one of the extraordinary collection actions.[92]

This new ACA framework for tax-exempt hospitals diverges in several important respects both from the law that came before it and from the reform proposals of the 1990s and 2000s. By providing such specific rules governing the minutiae of the practices of tax-exempt hospitals, the ACA requirements are a far cry from the pre-2010 world in which tax-exempt hospitals had near-complete freedom to decide how they would comply with the laws for tax exemption. On the other hand, the ACA, unlike any of the previous reform proposals, does not obligate tax-exempt hospitals to provide any set amount of charity care or community benefit. Instead, the ACA legislation and its regulations are almost entirely procedural. Hospitals must take carefully orchestrated steps to solicit community feedback, and they must follow many prescribed steps in dealing with patients who may need free or discounted care. However, the law does not include much by way of substantive requirements.

For instance, with regard to the CHNAs, the new law leaves it up to hospitals if and how they will address whatever needs the CHNAs identify. To take an extreme example, an asset-rich tax-exempt hospital could conduct a CHNA by soliciting feedback at one community meeting. At that meeting, perhaps a social-service agency might explain that 80% of local residents live below the poverty line and have medical debts from that hospital that they are unable to pay. The hospital would have to report in its CHNA that the social-service agency gave feedback as part of the CHNA process, and that report arguably would have to reveal what the feedback was. However, then, in the CHNA or implementation strategy, the hospital could write that, in its judgment, it was financially unable to provide any free care or reduce any outstanding debts. The hospital would have fulfilled its obligation under the CHNA rules.

Similarly, under the new rules, a tax-exempt hospital could have a financial aid policy that says, “We do not offer free or discounted care.” The hospital could then still bill a poor patient at chargemaster rates. If the patient did not pay, the hospital would have to make sure he or she is ineligible for financial aid, which would presumably be an easy decision under a policy that says no one is eligible for free or discounted care. After making and documenting that determination, the hospital could foreclose on the patient’s house, again having fully complied with the new law.

That is not to say that most tax-exempt hospitals, or even any tax-exempt hospital, would behave in this way. In designing the Affordable Care Act, members of Congress may have believed, perhaps correctly, that requiring certain procedural steps makes hospitals more likely to decide on their own to implement programs that respond effectively to the needs of poor communities, to adopt generous financial aid policies, and to refrain from extraordinary collection actions.

However, because Congress did not choose to enact substantive community benefit requirements, the broad question remains: after the 2010 reforms, what are hospitals doing to benefit their communities? More narrowly, are hospitals in fact responding to the needs of their communities? What kinds of financial aid policies have hospitals enacted? Are any of these policies associated with increased levels of free or discounted care? Are hospitals regularly engaging in extraordinary collection actions? Part III attempts to address these previously unanswered questions using a comprehensive set of new empirical data. Part III lays the groundwork for this analysis by elaborating on these questions, examining previous scholarship on the topic, and describing the data and methods of the study.

III. Questions, Literature, Data, and Methods

Part III will discuss the data and methods that are the basis of this Article’s empirical analysis of tax-exempt hospitals. Part III.A will describe the specific questions that this analysis addresses. Part III.B will briefly examine previous scholarship relating to these questions. Part III.C will describe the dataset the Article uses, and part III.D will explain the statistical models employed to analyze the data.

A. Questions Asked

This Article uses newly available data from Schedule H to IRS Form 990, the tax-exempt organization’s annual tax return, to assess, in light of the ongoing debate over the ACA, what types of community benefits tax-exempt hospitals are currently providing and what kinds of financial aid and debt- collection policies they have adopted. The Schedule H data provide a timely opportunity to address these questions. Schedule H is a schedule that all tax-exempt hospitals must file as part of their annual obligation to file tax returns. It asks hospitals for detailed information regarding the financial assistance and other community benefits they provide; their “community-building activities”; their bad debt and collections policies; and their individual facilities.[93]

The IRS first required Schedule H in 2008.[94] Before 2008, hospitals did not have to report at the federal level any information about their community benefits or financial policies.[95] Hospitals merely filled out the same tax return as any other tax-exempt organization. As a result, before 2008, no comprehensive data was available about how and to what extent tax-exempt hospitals were meeting the community benefit standard or what financial policies they might have in place. Even the IRS itself, when it set out to study the problem of tax-exempt hospitals in the mid-2000s had to rely on a survey sent out only to a subset of hospitals. The survey method, in addition to the problem of capturing only a fraction of the tax-exempt hospital sector, did not allow direct comparisons among hospitals, because the hospitals lacked clear definitions of terms such as “community benefit.” In the Instructions to the Schedule H, however, the IRS implemented standard definitions for the different figures hospitals now have to report. As a result, Schedule H for the first time offers a complete and comprehensive look at the tax-exempt hospital sector, its community benefits and its financial practices.

This Article uses a large Schedule H dataset from 2012 to answer two broad questions emerging from the traditional and new legal requirements for tax-exempt hospitals. The first deals with community benefits, the second with hospital policies, and practices. First, and most basically: how much community benefit are tax-exempt hospitals now providing? Specifically: (i) which types of benefits are most common? (ii) do these benefits differ according to the characteristics of hospitals and of the communities where the hospitals are located? and (iii) do benefits vary according to types of hospital policies or practices? Second, in what financial policies are tax-exempt hospitals actually engaging? Within that, are certain types of policies associated with certain community or hospital characteristics?

These questions arise directly from the traditional and new ACA requirements for tax-exempt hospitals. As discussed above, the traditional requirements adopt a broad notion of community benefit. This raises the question: given latitude to select the type and amount of community-related endeavors in which they engage, how much community work will hospitals choose to do? Which particular activities will they select?

Now the ACA has grafted the criterion of the concept of community responsiveness onto this traditional notion of community benefit. The CHNA provisions described above suggest that, in deciding what to do for their communities, hospitals should be weighing their communities’ particular needs. The idea that a hospital should be responsive to its community raises the question: are certain community characteristics associated with the ways in which hospitals choose to relate to their communities? For instance, will a hospital in a rural community select a different package of community activities than a hospital in an urban community?

The new requirements also supplement the traditional community benefit standard by emphasizing hospitals’ financial policies. However, at the time Congress passed the ACA, legislators had no comprehensive data on how common the regulated practices were among hospitals. As a result, the question remains: among these controversial practices and policies, which types are actually pervasive? Then, in terms of these financial questions and the community responsiveness criterion, how do hospital policies and practices vary by community characteristics?

B. Relationship to Previous Scholarship

To date, questions about what tax-exempt hospitals are doing to benefit their communities have produced few answers—and even fewer points of agreement among researchers. Perhaps because no comprehensive data was available prior to the Schedule H, legal scholarship itself has not generally approached the topic of tax-exempt hospitals from an empirical perspective. This Article is the first of which I am aware that assesses either the traditional or the new requirements for tax-exempt hospitals using data analysis of the tax-exempt hospital sector.

Insofar as studies have explored questions about tax-exempt hospitals and community benefits empirically, that scholarship has come not from the legal academy, but from the field of public health research. Even here, however, only three studies of which I am aware have tapped at all into the Schedule H data. Among these, the primary study, which appeared in the New England Journal of Medicine in 2013, offered suggestive preliminary observations about some of Schedule H’s community-benefit estimates.[96] Specifically, in this study, public health professor Gary Young and his collaborators found that, in fiscal year 2009, tax-exempt hospitals spent 7.5% of their operating expenses on “community benefits,” as defined on the Schedule H.[97] If hospitals had been allowed to add bad debt to this calculation, that figure would have risen to 11%.[98] Of these expenditures, more than 85% went to charity care and “other patient care services.”[99] Of the remaining community benefit expenditures, hospitals spent about 5% on community health improvements.[100] Hospitals in the top decile for spending on community benefits devoted approximately 20% of operating expenses to community benefits, while hospitals in the bottom decile spent approximately 1%.[101]

Further, Professor Young and his collaborators found that hospitals that provided one type of community benefit were not more likely to provide another kind of benefit.[102] They also found that those hospitals that covered more of (what they called) “patient care” expenses tended to be in states that had community benefit reporting regimes.[103] In addition, hospitals that supplied more of (what the authors called) “community service” expenses tended to be teaching hospitals that were also the sole hospitals in their communities.[104] Hospitals in the West provided more community benefits generally.[105] Aside from these results, however, Young and his collaborators were unable—using county-level demographic data—to find “any pattern of differences between hospitals that provided a relatively high level of community benefits and those that provided a relatively low level.”[106]

A second study based on Schedule H data examined whether community benefits vary by state.[107] In this case, health policy scholars Erik Bakken and David Kindig found significant differences across states.[108] Specifically, tax-exempt hospitals in Wyoming, Colorado, and Vermont spent most on community benefits, devoting more than 11% of hospital resources to them. North Dakota hospitals had the lowest state average at 3.76%.[109] Turning to per capita figures, the authors calculated the national average community benefit at $119 per person annually, but with a range from $30 per capita in Alabama to $335 per capita in Vermont.[110]

The third study using Schedule H data focused only on California tax-exempt hospitals. This research found that, in 2009, aggregate community benefit expenses amounted to 11.5% of hospitals’ total operating expenses.[111] “Uncompensated care” made up 53.7% of the total. Hospitals varied widely in their community benefit totals. The lowest quartile in terms of community benefit expenditures spent less than 7% of these on community benefit, whereas hospitals in the top quartile spent 16% or more. Charity care ranged from 0% to 6.3% of operating expenses. This study also provided descriptive statistics for each of the other categories of community-related activities on the Schedule H.

Before the Schedule H data became available, a handful of other scholars also examined tax-exempt hospitals and community benefits. Using a variety of state-level rather than national-level datasets, however, their studies produced an assortment of discrepant findings. In 2009, for instance, health policy researchers Brad Gray and Mark Schlesinger used 2001 data from Maryland to get a picture of hospital community-related activity in at least one state.[112] The Gray and Schlesinger study found that reported community benefit spending increased after Maryland implemented a community benefit reporting requirement.[113] The Maryland data also showed that “the amount and forms of community benefit activities var[ied] widely among hospitals.”[114] Also using the Maryland data, public health professor Simone Rauscher Singh found that nonprofit hospitals do not make tradeoffs among different types of community benefits.[115]

Taking a more normative stance, health policy scholars Gloria Bazzoli, Jan P. Clement, and Hui-Min Hseih used data from California and Florida to contend that hospitals were failing to provide “adequate” community benefits (except insofar as the researchers counted bad debt and Medicare shortfalls toward their totals).[116] However, in earlier work, Bazzoli and her team found that tax-exempt hospitals did in fact provide more community benefits than did their for-profit counterparts.[117] Pharmacology professor Amy Davidoff and her collaborators arrived at the same conclusion,[118] as did the Congressional Budget Office[119] and public health expert Kenneth Thorpe and his co-authors.[120]

In contrast to these findings, economist Helen Schneider—defining “adequate” community benefit using the sum of for-profit hospital uncompensated care and the federal and state income taxes they paid—found that tax-exempt hospitals were not providing enough in terms of community benefit.[121] Along similar lines, public health researchers Michael Morrisey, Gerald Wedig, and Mahmud Hassan argued that 20% to 40% of nonprofit hospitals provided insufficient uncompensated care relative to a for-profit benchmark.[122] Human ecologist Sean Nicholson and his collaborators reached consistent conclusions,[123] as did economists Edward Norton and Douglas Staiger,[124] as well as health researcher Janet Sutton and her team.[125]

Other scholars assessed tax-exempt hospitals’ community benefits by evaluating whether tax-exempt hospitals provide access to services that for-profit hospitals do not. In a series of papers, legal scholar and health economist Jill Horwitz found that nonprofit hospitals are particularly likely to provide less profitable health services, including many, like mental health services, that communities may desperately need.[126] In another study, business school professor Regina Herzlinger and economist William Krasker found that, in terms of the scope of hospital services, the number of emergency room visits and participation in health professions education, nonprofit hospitals were not generally different from for-profit hospitals.[127] In a similar study, however, health policy scholars Barbara Arrington and Cynthia Haddock reached the opposite result.[128]

A few studies have attempted to use state-level or other limited datasets to determine why some tax-exempt hospitals provide more community benefits than others. For example, health administration scholars Alva O. Ferdinand, Josué Patien Epané, and Nir Menachemi relied on American Hospital Association survey data to argue that religious hospitals engage in a significantly higher number of community-benefiting activities than other hospital ownership types.[129] This team also found that all hospitals increased their average amount of community benefits during times of economic growth, and that organizational size, teaching facilities, and urban location were all associated with higher levels of community benefits.[130] Another public health team found, also using American Hospital Association survey data, that hospitals with “community health” and “community-based quality” orientations, as identified through features like a mission statement discussing community issues, were more likely to provide community benefits than hospitals without these orientations.[131] A different study found that community orientation itself varied by hospital characteristics and tended to be more significant in hospitals that are large, part of a network of hospitals, dependent on managed care, and located in communities with diffuse community activities.[132]

In a study with significant policy implications, business school professor Frances Kennedy and her collaborators employed data from Texas to explore the impact of a 1993 Texas law that required tax-exempt hospitals to expend a fixed level of net revenue (generally 4%) on charity care to find that the law had actually led to a decrease in the total amount of charity care provided.[133] Relatedly, using American Hospital Association survey data, a team of public health researchers considered states that had passed laws governing tax-exempt hospital community benefits.[134] This study found that, on average, nonprofit hospitals in the ten states with some type of community benefit law reported significantly more community health orientation activities than nonprofit hospitals in the forty other states.[135] In addition, for-profit hospitals in the ten states with laws/guidelines reported significantly more community health orientation activities than did the investor-owned hospitals in the forty other states.[136] The same researchers examined similar issues in 2009 with slightly different results.[137]

Several other studies have looked at particular aspects of the notion of community benefit. For instance, pediatrics professor Peter Szilagyi and his collaborators examined community engagement by academic health centers and also proposed a formal framework for evaluating this practice.[138] Medical school professor Lloyd Michener and two teams of researchers also assessed ways in which academic health centers can respond to community needs.[139] Behavioral and community health scientist Jessica Burke and her collaborators considered a range of hospital community engagement projects that had been proposed in print and found that hospitals had formally evaluated very few of them.[140] Public health scholar Jeffrey Alexander and his team looked at the efforts of hospitals to be accountable to their communities and found that freestanding hospitals were more likely to achieve accountability through governance structures, whereas system-affiliated hospitals preferred to do so through active boards.[141]

C. Dataset and Variables

To put our knowledge of tax-exempt hospitals on more solid empirical foundations, this Article uses newly available Schedule H data to answer questions about tax-exempt hospitals, the benefits they provide, and their financial practices.[142] Consequently, the primary source of data consisted of IRS Form 990 and its attached Schedule H for tax year 2012. I focused on 2012 because it was the most recent year for which complete data was available. I obtained these data from GuideStar, an organization that collects, digitizes, and sells information on the entire population of U.S. tax-exempt organizations’ Forms 990 and attached schedules. Because the data consist of the entire population, sampling issues did not arise.

The dataset I received from GuideStar included Schedule H data from 2636 tax-exempt hospitals nationwide. Upon inspection, however, I saw that some of these materials actually pertained to tax years other than 2012, so when I removed those observations, I had a total of 2158 Forms 990 and their Schedules H. To correct for human error, for the primary quantitative variables in which I was interested, I reviewed the hospitals’ entries to determine whether entries that were supposed to be the sums or quotients of other entries actually were. This process revealed a handful of what I believed to be data entry errors, in which case I pulled the individual Form 990 or Schedule H in question and, where appropriate, filled in what should have been the correct answer. I also looked for similar data-entry errors with regard to variables that were not sums or quotients. Occasionally, fixing what seemed to be an unreasonable answer to a question required either pulling the original Form 990 or Schedule H (to fill in the correct answer) or finding other relevant documentation from the hospital in question—for instance, its financial aid policy—to supply a sensible response to the question.

I then merged the hospital IRS filings with the data from the 2008-2012 five-year American Community Survey from the U.S. Census Bureau.[143] Past studies of tax-exempt hospitals have used data reported for broad geographic regions, such as counties, to measure the characteristics of the communities where hospitals are located. However, the patient base at many hospitals, particularly in urban areas, draws more heavily from neighborhoods immediately surrounding the hospitals than it does from entire counties. For example, in Chicago, the University of Chicago Hospital, located on the city’s historically impoverished South Side, serves a community distinct from the one served by Northwestern Hospital—North Shore, located across the metro area from the South Side in the wealthy suburb of Evanston. In order to capture intra-urban community differences such as these, I matched each hospital with the specific census tracts that fell within five miles. For the purposes of this study, those census tracts in the five-mile radius constituted the hospital’s community.

I then created composite demographic variables for each of these communities using the American Community Survey data. Specifically, for each community, I obtained a measure of the percent of its population that is Black and Hispanic (“Percent Black” and “Percent Hispanic”), the percent of its population living below 100% of the poverty line (“Percent below 100 FPL”), the percent of its population living between 100-149% of the poverty line (“Percent 100-149 FPL”), the average age of its residents (“Age”), its population density (“Population density”), the percent of its population with private insurance (“Percent privately insured”) and the percent of its population with public insurance (“Percent publicly insured”). I then matched these demographic data with their hospitals.

Following this, I merged the hospital and American Community Survey data with a dataset available from the Hilltop Institute at the University of Maryland-Baltimore County, which contains information about the laws in each state governing nonprofit hospitals and their community benefit.[144] This dataset included the following variables. “Unconditional requirement” measures whether the state in which the hospital is located requires all hospitals in the state to provide some community benefits, broadly defined.[145] “Conditional requirement” measures whether the state in question made its tax or regulatory benefit conditional on whether the hospital provided community benefits.[146] “Mandatory minimum” measures whether the state required a hospital to provide a quantifiable amount of community benefit each year.[147]

Next, I collected a series of variables from the Form 990 and the Schedule H. I used these to measure the institutional characteristics of hospitals, as well as to obtain the community-activity and financial-policy information. The institutional-level variables I collected were as follows: “Gross receipts” (Header, line G), which measured each hospital’s gross receipts, “profitability” (Part I, line 19) which measured each hospital’s revenues less expenses, and “donations” (Part VIII, sum of lines 1a, 1b, 1c, 1f, and 1g on Form 990), which measured each hospital’s amount received in private donations.[148]

Then, I collected variables that would measure each hospital’s community activities. Here, I used every such variable available on the Schedule H. As a result, all the line numbers from this point forward refer to lines on the Schedule H. The first group were the charity care expenses, expressed as a percent of each hospital’s total expenses: “Uncompensated care” (Part I, line 7a(f)), or financial assistance provided at cost,[149] “Medicaid costs” (Part I, line 7b(f)),[150] and “Other costs” (Part I, line 7c(f)), or the costs of other government health programs such as the State Children’s Health Insurance Program.

The next group of variables consisted of the other “community benefit expenses,” again each expressed as a percent of the hospital’s total expenses. The first of these “other community benefit” variables was “community health improvements” (Part I, line 7e(f)), which included the costs of programs for the purpose of improving community health or “achiev[ing] a community benefit objective.”[151] Examples include scholarships for community members and nurse education.[152] The next variable in the group was “health professions education” (Part I, line 7f(f)), which included costs of degree programs for health professionals.[153] The third variable of this set was “subsidized health services” (Part I, line 7g(f)), which included the costs of clinical programs that meet a community need and operate at a loss.[154] The fourth was “research” (Part I, line 7h(f)) which included the costs of research that enhances public knowledge.[155] The fifth was “hospital contributions”(Part I, line 7i(f)), which included the costs of donations the hospital made to other programs.[156] The sixth variable in the group was “total benefits” (Part I, line 7k(f)), the sum of all of these “other community benefits” variables plus the three charity care variables.

The next variables measuring community benefit were the “community building” variables. Each of these measured the extent to which the hospital in question was incurring costs to address social, rather than medical, determinants of health. Again, each was expressed as a percent of the hospital’s total expenses. The IRS views “community building” as technically separate from “community benefit,” although, colloquially, community building would seem to be a type of community benefit. As a result of the IRS’s view, community building activities are separate on the Schedule H from the community benefit activities and do not count toward the community benefit total.

To give a sense of what counts as community building, the first community building variable, “physical improvements” (Part II, line 1(f)), included costs to provide and rehabilitate housing for vulnerable populations.[157] The second, “economic development” (Part II, line 2(f)), included costs of helping small businesses in vulnerable neighborhoods and creating job opportunities in areas of need.[158] The third, “community support” (Part II, line 3(f)), included the costs of childcare, mentoring, violence prevention, and public health emergency activities.[159] The fourth, “environmental improvements” (Part II, line 4(f)), included the costs of addressing environmental hazards that affect the local community.[160] The fifth, “leadership development and training for community members” (Part II, line 5(f)), included the costs of training community members in conflict resolution, civic, cultural, or language skills.[161] The sixth, “coalition building” (Part II, line 6(f)), included the costs of working with community partners on community health issues.[162] The seventh, “community advocacy” (Part II, line 7(f)), included the costs of supporting policies to advance public health. The eighth, “workforce development” (Part II, line 8(f)), included the costs of recruiting health professionals to underserved agencies.[163] The ninth, “other community building” (Part II, line 9(f)), included the costs of any other activities in which the hospital engaged that might protect its community’s health and safety.[164] The final, “total community building” (Part II, line 10(f)), was an aggregate of all of the “community building” variables.[165]

Finally, I collected variables assessing whether hospitals had certain financial policies in place. Here, I used the answers to every policy question on the Schedule H. These policies pertained, among other things, to whether hospitals had adopted financial aid and emergency care policies, what debt collection policies hospitals had in place, and the thresholds at each hospital above which free and discounted care were available.

D. Statistical Models Used

To begin, I used descriptive statistics to determine, as a general matter, how much in total community benefits, as well as in total community building costs, tax-exempt hospitals were providing. I also used descriptive statistics to describe the eligibility levels for free and discounted care at each hospital. I then used a series of ordinary least squares (“OLS”) multiple-regression models to estimate which community and hospital characteristics were associated with those factors.[166] An OLS model estimates the relationship between different variables in a data set. To illustrate: an OLS model could help interpret the relationship between lung cancer (the “dependent” variable) and smoking (the “independent” variable) in the following manner.

With data about how many people have died from lung cancer in a given year and how many cigarettes that population consumed, an OLS model can provide an estimate of the number of deaths associated with each cigarette smoked.[167] The number of deaths associated with each cigarette smoked would be called the coefficient that the statistical model calculates. If the coefficient for “cigarettes smoked” were, say, 3, that would mean that every cigarette smoked was associated with an additional three deaths. Coefficients in OLS models can also be negative. If the coefficient for “cigarettes smoked” were -3, that would suggest some very healthy cigarettes at work. Specifically, that coefficient would mean that, for the group of smokers under study, every cigarette smoked was associated with 3 fewer deaths.

An OLS model can also disentangle the relationships among multiple factors. To return to the cigarette example, with additional data about how many cheeseburgers the population consumed, the OLS model can show an estimate of the number of deaths associated with each cigarette smoked and with each cheeseburger eaten. In this situation, the statistical model calculates a coefficient for each variable analyzed. If the coefficient that the model generated for “cigarettes smoked” was 3 and the coefficient that the model generated for “cheeseburgers eaten” was 1, that would mean that every cigarette smoked was associated with another three deaths, and every cheeseburger eaten was associated with one additional death.

The OLS coefficients provide a best guess as to the relationship between the variables. The coefficients themselves do not tell us anything about the certainty of that relationship or the effect of chance. To analyze whether mere chance might have produced the relationship, we turn to the concept of statistical significance. If a coefficient is statistically significant, then it is unlikely that there is no actual relationship between the variables. Returning to the smoking example, if the coefficient of 3 is statistically significant, the likelihood is small that mere chance produced that value. In the tables I present, asterisks denote statistically significant coefficients.

In this study, I used OLS models to determine which hospital and community characteristics (the independent variables) were associated with which levels of community building and community benefit activities (the dependent variables). Next, I examined hospitals’ current financial policies using descriptive statistics and a technique called principal components analysis, followed by a series of additional OLS models to determine which hospital and community characteristics (the independent variables) were driving variations in debt collection policies (the dependent variables).

The results of these models are described in part IV below.

IV. Results

To begin with the central question, I examined the total community benefit provided by each hospital. The results are shown in the histogram in Figure 1.

Figure 1: Community Benefit Spending Totals by Frequency

Figure 1

Next, I performed the same analysis on total community building activities. The results are shown in Figure 2.This histogram shows substantial variation in the amount of total community benefit (Schedule H, Part I, line 7(k)(f)) that tax-exempt hospitals provide, ranging from 0% to 100% of total expenses. However, most hospitals spend between 5% and 10% of total expenses on community benefits. As described above, these totals include charity care costs (the costs of free and uncompensated care) along with the costs of direct community health interventions, health professions education, and medical research. They do not include bad debt or any of the community building activities.

Figure 2: Community Building Spending Totals by Frequency

Figure 2

This histogram shows the variation in the amount of total community building costs (Schedule H, Part II, line 10(f)) provided by tax-exempt hospitals. This histogram makes evident how much less hospitals spend on community building than on community benefit. Even though the community building totals include a number of different activities, most hospitals spend between 0% and 1% of total expenses on community building, with the vast majority spending less than 0.1%.

The data in Figures 1 and 2 raise the further question of how spending in these two large categories—community benefit and community building—is divided among different subcategories of spending. In other words, what percentage of hospital spending goes toward each of the activities in the larger categories?

Table 1: Breakdown of Community Benefit and Community Building Spending

Statistic Mean St. Dev. Min Median Max
Uncompensated care 2.58% 4.83% 0.00% 1.77% 100.00%
Medicaid costs 3.29% 3.68% 0.00% 2.55% 61.58%
Costs of other gov. programs 0.27% 1.47% 0.00% 0.00% 31.69%
Total charity care 6.01% 6.33% 0.00% 5.04% 100.00%
Community health improvements 0.40% 1.75% 0.00% 0.15% 76.43%
Health professions education 0.61% 1.25% 0.00% 0.07% 10.62%
Subsidized health services 1.16% 2.73% 0.00% 0.00% 42.28%
Research 0.22% 1.59% 0.00% 0.00% 43.74%
Hospital contributions 0.21% 0.96% 0.00% 0.03% 24.17%
Total other benefits 2.60% 4.10% 0.00% 1.23% 76.43%
Total community benefits 8.58% 7.48% 0.00% 7.45% 100.00%
Physical improvements 0.02% 0.27% 0.00% 0.00% 7.70%
Economic development 0.01% 0.20% 0.00% 0.00% 7.32%
Community support 0.03% 0.24% 0.00% 0.00% 10.31%
Environmental improvements 0.001% 0.01% 0.00% 0.00% 0.41%
Community leadership 0.002% 0.02% 0.00% 0.00% 0.75%
Coalition building 0.01% 0.11% 0.00% 0.00% 4.72%
Community advocacy 0.02% 0.16% 0.00% 0.00% 4.54%
Workforce development 0.03% 0.17% 0.00% 0.00% 4.18%
Total community building 0.11% 0.48% 0.00% 0.002% 10.31%

Table 1 presents the means, standard deviations, medians, minimum values, and maximum values for each of the community benefit and community building variables. As was perhaps evident from the histograms in Figures 1 and 2, the total charity care and total community benefit figures vary substantially, with the top performers providing charity care equal to 100% of total expenses.

As the data indicate, the median amount of spending for charity care is 5.04%, while the mean is 6.01%. These figures just exceed the minimum threshold that Senator Grassley’s discussion draft legislation would have required had it become law. Adding in the other community benefit variables raises the mean value to 8.58% of total expenses and the median to 7.45% of total expenses.

Of the non-charity-care community benefit variables, community health improvements, health professions education, and hospital contributions have median values above 0%. That indicates that most hospitals engage in some amount of each of these activities. However, while research and subsidized health services have maximum values of 42.28% and 43.74%, revealing that at least one hospital (perhaps a small handful) is spending substantial amounts on these two endeavors, the median value for both of these variables is 0%. Most hospitals are spending nothing on research or subsidized health services.

Turning to the community building figures, the mean amount spent by a tax-exempt hospital on community building equals 0.11% and the median is 0.0002%. As the histograms show, hospitals are spending substantially less on community building than they are on community benefit activities. No single community building activity has a median value above 0%. The community building activity with the highest mean value is workforce development, but this value is merely 0.03%. The community building activities with the highest maximum values are physical improvements and community support. At least one hospital spent 7.7% of its total expenses on physical improvements and at least one spent 10.31% of its total expenses on community support.

Next, I attempted to determine what was driving the variation in spending on community benefit and community building. I observed that more spending on community benefit was not correlated with more spending on community benefit.[168] Only six hospitals were in the top hundred for both types of spending. In other words, hospitals that spent more on community building were not especially likely to spend more on community benefit and vice versa. That raised two questions. Were any hospital or community characteristics associated with more spending on community benefit? Were any such characteristics associated with more spending on community building? To answer these questions, I ran the above-described OLS models. The first set of models—Models 1, 2, and 3—used total community benefit spending as the dependent variable. Total community benefit was the aggregate of all community benefit variables, so hospitals with more community hospitals should be the hospitals identified in the factor analysis as spending more on each of the individual variables. The results of the OLS model are given in Table 2.

Table 2: Total Community Benefit Spending According to Hospital and Community Characteristics
Model 1 Model 2 Model 3
Unconditional requirement 0.107 0.090 0.102
(0.068) (0.071) (0.070)
Conditional requirement -0.052 -0.024 -0.023
(0.064) (0.067) (0.067)
Mandatory minimum 0.056 0.005 -0.003
(0.081) (0.084) (0.085)
Mandated level 0.010 -0.030 -0.045
(0.052) (0.052) (0.052)
Gross receipts (logged) 0.132* 0.078* 0.076*
(0.017) (0.020) (0.020)
Profitability (millions) -0.0001 -0.0002 -0.0002
(0.0005) (0.0004) (0.0004)
Number of facilities (logged) -0.054 -0.033 -0.034
(0.050) (0.053) (0.053)
Donations (millions) 0.002* 0.003* 0.003*
(0.001) (0.001) (0.001)
Percent publicly insured -0.243 -0.894 -0.784
(0.567) (0.615) (0.614)
Percent privately insured -0.214 -0.440 -0.376
(0.581) (0.618) (0.617)
Population density (thousands per sq. mi.) 0.017* 0.017* 0.017*
(0.006) (0.005) (0.006)
Age -0.005 -0.012 -0.012
(0.008) (0.008) (0.008)
Percent Black -0.206 -0.280 -0.264
(0.189) (0.200) (0.200)
Percent Hispanic -0.126 -0.300 -0.278
(0.247) (0.262) (0.261)
Percent below 100 FPL 0.166 -0.151 -0.118
(0.698) (0.744) (0.746)
Percent 100-149 FPL 2.726* 2.281 2.406
(1.148) (1.231) (1.236)
Free care eligibility 0.081*
Discounted care eligibility 0.043
Intercept 0.412 2.218* 2.136*
(0.707) (0.778) (0.780)
Observations 2,158 1,755 1,752
R2 0.066 0.060 0.058
Note: * p<0.05

These results indicate which hospital and community characteristics were associated with higher spending on community benefit. The table shows three models. Model 1 does not include any of the variables measuring hospital policies. After running Model 1, I also wanted to assess the extent to which hospital policies such as debt collection practices or the content of financial-assistance policies might be associated with higher levels of spending on community benefit. The hospital-policy variables tend to be highly correlated with each other (i.e., a hospital that makes its financial aid policy widely available is more likely to have a policy authorizing debt collection actions), so I included only one of them in a regression model at a time. None of these policies came close to a significant relationship to total community benefits except for two: free care eligibility, the variable that indicates the percentage above the federal poverty line under which free care was available (i.e., free care was available for patients with incomes below 200% of the poverty line); and discounted care eligibility, the variable that indicates the percentage above the federal poverty line under which discounted care was available (i.e., discounted care was available for patients with incomes below 400% of the poverty line). For this reason, Table 2 only shows the results of the models that included free care eligibility and discounted care eligibility. Model 2 includes free care eligibility, and Model 3 includes discounted care eligibility.[169]

Model 1, which did not include either of the policy variables, indicates that four specific hospital or community characteristics were significantly associated with higher spending on community benefits: gross receipts, donations, population density, and percentage of the population with incomes in between 100% and 149% of the federal poverty line. For example, gross receipts had a coefficient of 0.132, which meant that, for the mean hospital, a 10% increase in gross receipts was associated with a 0.008% increase in expenses devoted to total community benefits.[170] Overall, these data showed that hospitals in the category providing more community benefits tended to be larger hospitals in more urban communities with residents living just above the poverty line.

In Models 2 and 3, gross receipts, donations, and population density were still significantly and positively associated with the percentage of expenses devoted to total community benefit. Model 2 showed that free care eligibility was also positively and significantly associated with total community benefit. That means that hospitals that provide free care at higher income levels are more likely to devote more of their expenses to community benefits. In this model, percent 100-149 FPL was positively associated with community benefit spending, but slightly below the statistical significance threshold.[171] That raises the possibility that some or all of the effect of having more community members in that income range on community benefit spending may not be independent of the free care policy. In other words, having more community members in that income range may directly cause hospitals to spend more on community benefit. Having more community members in that income range may also cause hospitals to adopt financial aid policies with higher thresholds. This result suggests that both of these effects are probably present to some extent.

Model 3 shows that a higher threshold for the discounted care policy was associated with a greater percent of expenses devoted to community benefit. However, the effect was not quite significant. Again, in Model 3, percent 100-149 FPL had a positive but not quite significant effect on community benefit spending. Again, this suggests a potential indirect effect. Perhaps part of the reason that the existence of community members in that income range is associated with higher spending on community benefit is because having more members in that income range makes hospitals more likely to adopt generous discounted-care policies. Those generous discounted-care policies may in turn be associated with an increase in community benefit spending.

Next, I examined which hospital and community characteristics were associated with the hospitals that devoted larger amounts to community building activities. Total community building aggregated all of the individual community benefit variables, so I used that as the dependent variable in an OLS model. Table 3 presents the results.

Table 3: Total Community Building Spending according to Hospital and Community Characteristics
Model 4
Unconditional requirement 0.036*
Conditional requirement -0.035*
Mandatory minimum -0.016
Mandated level -0.001
Gross receipts (logged) 0.020*
Profitability (millions) -0.0001
Number of facilities (logged) -0.019
Donations (millions) -0.00004
Percent publicly insured 0.017
Percent privately insured 0.335*
Population density (thousands per sq. mi.) -0.001
Age 0.001
Percent Black -0.061
Percent Hispanic 0.021
Percent below 100 FPL -0.004
Percent 100-149 FPL -0.339(0.297)
Intercept -0.306
Observations 2,158
R2 0.027
Note: * p<0.05

These results show that higher spending on community building is significantly associated with being in a state with an unconditional community benefit requirement, with higher gross receipts, and with a higher rate of privately insured individuals. Being in a state with a conditional community benefit requirement is associated with lower spending on community building. In short, the data suggest that hospitals in the group spending more on community building tended to be larger and in communities whose residents had private insurance.

Next, I turned to the question of hospitals’ financial policies. To begin, I examined hospitals’ free and discounted care policies. I started with this issue for two reasons. First, the above analysis of community benefit expenditures suggested that higher free and discounted care thresholds were associated with higher spending on community benefit. Given this, determining which hospitals are likely to have higher thresholds offers a further way to understand the variations in community benefit spending across hospitals. Second, the new ACA regulations are an effort to require hospitals to standardize most of the financial policies they report on the Schedule H. In this transition period, not all hospitals have adopted the ACA policies and regulations yet, but, presumably, almost all hospitals eventually will. However, these policies and regulations do not govern the free and discounted care thresholds. While both the Senator Thomas and Senator Grassley proposals would have required hospitals to provide free or discounted care to patients with incomes below certain percentages of the poverty line, the ACA legislation allows hospitals the freedom to set the thresholds as they choose. For this reason, hospital practice regarding the thresholds is allowed to vary widely, and vary widely it does. Further, even as the ACA regulations take effect, the thresholds may continue to vary in the years to come. That raises the question of why such wide variation exists.

The histograms in Figures 4 and 5 below show this variation in free and discounted care eligibility.

Figure 4: Hospital Free Care Eligibility Levels by Frequency

Figure 4

Figure 5: Hospital Discounted Care Eligibility Levels by Frequency

Figure 5

These histograms show wide variation in eligibility levels for free and discounted care across hospitals. The majority of hospitals provide free care for patients with incomes between 100-200% of the poverty line. However, some hospitals require incomes below 100% of the poverty line for free care, and a very small handful allow patients with incomes at 600% of the poverty line to qualify for free care.

The discounted care thresholds vary even more substantially. While the largest number of hospitals give discounts on care to patients with incomes up to 400% of the poverty line, significant numbers of hospitals use thresholds around 200% and 300% as well. A few hospitals offer discounts at up to 1000% of the poverty level. To determine what was driving this variation, I ran OLS models in which the dependent variables were the thresholds for free and discounted care, and the independent variables were community and hospital characteristics. Table 4 presents the results.

Table 4: Free and Discounted Care Thresholds according to Community and Hospital Characteristics 
Free care eligibility Discounted care eligibility
Model 7 Model 8
Unconditional requirement 0.233* 0.206*
(0.044) (0.072)
Conditional requirement -0.047 -0.052
(0.042) (0.069)
Mandatory minimum 0.183* 0.495*
(0.052) (0.086)
Mandated level -0.044 0.248*
(0.032) (0.054)
Gross receipts (logged) 0.066* 0.153*
(0.012) (0.020)
Profitability (millions) 0.001* 0.0004
(0.0003) (0.0005)
Number of facilities (logged) -0.021 -0.035
(0.033) (0.055)
Donations (millions) 0.002* 0.00003
(0.001) (0.001)
Percent publicly insured 1.051* -0.476
(0.381) (0.632)
Percent privately insured 0.764* 0.139
(0.383) (0.635)
Population density (thousands per sq. mi.) 0.011* 0.014*
(0.003) (0.006)
Age -0.001 -0.008
(0.005) (0.009)
Percent Black 0.516* 0.511*
(0.124) (0.205)
Percent Hispanic 0.777* 1.017*
(0.162) (0.268)
Percent below 100 FPL -1.008* -2.748*
(0.462) (0.765)
Percent 100-149 FPL -0.909 -4.611*
(0.765) (1.266)
Intercept -0.273 1.483
(0.483) (0.801)
R2 0.140 0.197
Note: * p<0.05

Turning to Model 8, the results show that being in a state with an unconditional community benefit requirement, as well as with a mandatory minimum level of care, is significantly associated with a higher discounted care threshold. Being in a state with rules about free and discounted care thresholds is also significantly associated with a higher discounted care threshold. Gross receipts, population density, percentage Black and percentage Hispanic are also significantly associated with higher discounted care thresholds, and percentage below the poverty line is significantly associated with lower discounted care thresholds.The results for Model 7 show that a number of hospital and community characteristics are significantly associated with having higher thresholds for free care. Specifically, a state-level unconditional community benefit requirement, a state-level mandatory minimum level of community benefit, gross receipts, profitability, donations, percentage publicly insured, percentage privately insured, population density, percentage Black, and percentage Hispanic all correspond to a higher threshold for free care. Percentage below the poverty line is negatively associated with a higher free care threshold. For instance, in terms of effect sizes, being in a state with an unconditional community benefit requirement is associated with a 23.3% increase in the free care threshold. Being in a state with a mandatory minimum amount of community benefit is associated with an 18.3% increase in the free care threshold. Overall, the hospitals that provide free care above higher income thresholds tend to be larger and more profitable and tend to receive higher levels of private donations. These hospitals are likely to be located in states with their own community benefit thresholds, as well as in urban areas with higher percentages of Black and Hispanic community members. However, these hospitals are not necessarily those dealing with notably disadvantaged populations. In fact, these hospitals are more likely to be located in communities where residents have insurance and are above the poverty line.

Next, I examined the other financial policy variables on which the Schedule H gathers data. First, I considered whether hospitals are in fact adopting these policies, many of which are explicitly required by the proposed regulations. The others are policies that the IRS presumably seeks to encourage by asking about on the form. Table 5 shows the extent to which hospitals are enacting these policies.

Table 5: Hospital Financial Policies by Percent of Hospitals Adopted
Policy Percent
Did the organization have a financial aid policy? 98.02%
If yes, was it a written policy? 97.36%
Did the organization budget amounts for free or discounted care provided under the financial assistance policy? 90.31%
Did its expenses exceed budget? 55.51%
Was the organization unable to provide some care as result? 0.88%
Have a financial aid policy? 82.56%
Used federal poverty guidelines to determine eligibility for providing free care? 79.74%
Used FPG to determine eligibility for providing discounted care? 74.80%
Explained the basis for calculating amounts charged to patients? 78.02%
Income level? 74.58%
Asset level? 50.18%
Medical indigence? 49.34%
Insurance status? 45.33%
Uninsured discount? 45.55%
Medicaid/Medicare? 42.91%
State regulation? 30.13%
Other? 11.63%
Explained the method for applying for financial assistance? 81.67%
Did the organization prepare a community benefit report during the tax year? 77.71%
If yes, did the organization make it available to the public? 73.61%
Have a billing/collections policy? 77.80%
Did the hospital facility have a policy in place requiring emergency medical assistance regardless of eligibility for financial assistance? 79.21%
Did the hospital calculate charges for financial-aid-eligible patients using one of these methods?
Lowest negotiated commercial insurance rate? 5.07%
Average of three lowest negotiated commercial insurance rates? 9.12%
Medicare rates? 9.60%
Other? 10.22%
Did the hospital charge any financial-aid-eligible patients more than amounts generally billed? 3.35%
Did the hospital charge any financial-aid-eligible patients any amounts equal to their gross charges? 13.44%

These numbers show that most of the policies that the IRS is either regulating directly or seeking to encourage are actually becoming widespread among tax-exempt hospitals. Over 98% of tax-exempt hospitals, for example, have financial aid policies and about 97% of these policies are in writing. Approximately 82% of hospitals have an emergency care policy stating that they do not discriminate on the basis of financial aid eligibility. Only about 3% of hospitals charged aid-eligible patients above amounts generally billed (although thirteen of them billed patients using gross charges).

The only policies exhibiting substantial variation are policies where, under the new ACA regulations, tax-exempt hospitals get a choice as to the content of a policy. For example, the regulations do not specify whether hospitals should calculate aid eligibility using income, assets, medical indigence, insurance status, Medicare/Medicaid, or some other item. On the Schedule H itself, the IRS lists all of these options and indicates no preference among them. Perhaps as a result, while income is the most common means of determining eligibility (74.58% of hospitals use it), hospitals are fairly evenly split among the other options. Similarly, while most hospitals appear not to have yet selected which of the prescribed methods they will use to charge aid-eligible patients, no clear favorite option has emerged. About 5% of hospitals use the lowest negotiated commercial rate, about 9% use an average of the three lowest negotiated commercial rates, about 9% use the Medicare rate, and about 10% use an unspecified other method.

After examining at these descriptive statistics, I ran a few preliminary models to determine whether any hospital or community characteristics seemed to be driving what variation does exist here. However, no clear patterns emerged. Given that the policies listed in Table 7 showed little variation, this was not a surprising result. Perhaps more surprisingly, I did not observe any relationship between any of the policy types and a hospital’s levels of community benefit and community building.

After looking at the policy variables generally, I focused more closely on the debt collection variables. I chose to look more carefully at debt collection policies because both Congress and numerous commentators had emphasized them so heavily in the decade leading up to the ACA. In terms of debt collection practices, most hospitals in this study did not have policies explicitly authorizing them to engage in extraordinary collection actions against patients whose aid eligibility had not yet been determined. Similarly, most hospitals did not carry out any extraordinary collection actions against patients without first checking to see if the patients were eligible for aid. However, about 200 hospitals checked “yes” to having some such policies in place or having engaged in one of these actions.

I then carried out a principal components analysis to combine all of the different variables relating to debt collection into a single measure of debt collection practice. Then, to determine whether these debt-collecting hospitals displayed any particular hospital-level or community characteristics, I ran another OLS model, the results of which are shown in Table 6. The dependent variable here is “debt collecting,” which is a measure, derived from the principal components analysis, of the extent to which the hospital belonged to this debt-collecting group.

Table 6: Extraordinary Collection Actions according to Hospital and Community Characteristics
Model 9
Unconditional requirement -0.003
Conditional requirement 0.135*
Mandatory minimum -0.140*
Mandated level 0.144*
Gross receipts (logged) -0.025*
Profitability (millions) -0.0001
Number of facilities (logged) 0.023
Donations (millions) -0.0004
Percent publicly insured 0.242
Percent privately insured 0.996*
Population density (thousands per sq. mi.) -0.002
Age -0.006
Percent Black -0.445*
Percent Hispanic -0.353*
Percent below 100 FPL 0.530
Percent 100-149 FPL -0.477
Intercept 0.193
R2 0.037
Note: * p<0.05

These results show that conditional requirement and percent privately insured were both positively associated with whether a hospital was a debt-collecting hospital. Mandatory minimum community benefit, gross receipts, percent Black and percent Hispanic were all negatively associated with debt-collecting behavior. In other words, being in a state with a conditional community benefit requirement was associated with a 13.5% increase in a hospital’s degree of debt-collecting behavior, while a 1% increase in the percent of the population with private insurance was associated with a 1% increase in the hospital’s degree of debt-collecting behavior.

V. Discussion

Part V analyzes the results described in Part IV. To do so, it first considers the results regarding community benefit and community building activities; it then examines the results regarding hospital financial policies.

Turning first to the topic of community benefit and community building activities, the 2012 Schedule H data considered here show that tax-exempt hospitals spend an average of 8.5% of total expenses on community benefits, with a median of 7.45%. These results are in line with the only previous study to report a community benefit average using all of the Schedule H data. That earlier study, which used 2009 data, found an average community benefit of 7.5%.[172] The higher figure for 2012 may represent a modest but genuine increase in the amount of hospital community benefit activity during the three intervening years. During this time, the passage of the ACA increased scrutiny of tax-exempt hospitals, so they may have increased their community benefit activity as a response. Additionally, during this period, hospitals became accustomed to reporting their community benefit activities. The requirement to explain their community benefits spending to the IRS each year may have made hospitals more eager to provide benefits that they could list on the form. The researchers who looked at Maryland community benefit data before and after Maryland enacted community benefit reporting requirements identified this effect, so seeing it at the federal level would not be surprising.[173]

In addition, the data here showed that the median amount of total charity care provided by hospitals was equal to 5.04% of their expenditures, with a mean of 6.01%. These figures just exceed the minimum threshold that Senator Grassley’s 2009 discussion draft legislation would have required had it become law. The data did show a fair amount of variation in community benefit spending, which in part derived from variation in the charity care measures. This means that Senator Grassley’s bill likely would have forced some hospitals to increase their charity care to reach the 5% minimum. However, as in Texas after the passage of its mandatory minimum community benefit law, other hospitals might have dropped their charity care percentages to conform to the statutory mean. Whether these effects would have balanced themselves out or would have led to a net increase or decrease in charity care remains unknown. As a result, it is difficult to compare the current 6.01% mean to the mean that would have been mandated had the Grassley discussion draft become law.

The data here also revealed substantial variation in hospitals’ spending on community benefit versus community building activities. In particular, the data suggested that one group of hospitals was especially likely to concentrate its spending on one or more community benefits, while a different group of hospitals was likely to concentrate on one or more community building activities. The hospitals in the two different groups were in two different types of surrounding communities—and different communities have different needs. The data suggested that in struggling urban areas, hospitals focus more on providing community benefits, but in richer areas hospitals may turn their focus elsewhere.

That is to say, the hospitals providing more community benefits were located in particular types of communities. Their communities were more densely populated and had more residents living close to, but not below, the poverty line—i.e., more “near poor” residents who fell within 100-149% of the poverty line. Additionally, the hospitals themselves tended to be larger.

In some ways, these findings were not surprising. For one, they echoed the conventional wisdom, sometimes cited in passing in the Congressional hearings, that smaller and more rural hospitals are less likely to engage in substantial amounts of community benefit activities as traditionally defined.[174] Similarly, the 2009 IRS report found that hospitals providing “critical access” to health care in rural communities supplied less charity care than other hospitals.[175]

However, no literature of which I am aware has previously identified a link between community benefit spending and percent of “near-poor” residents in the surrounding community. The current study introduced two important modifications that may have helped uncover this connection. First, this study used a more sensitive measure of “community” than other studies have used. As described earlier, past studies generally defined community on the basis of county-level data. By replacing this measure with one based on immediate census tracts, this study may have highlighted previously ignored ways in which hospitals respond to the needs of communities in close proximity. In particular, some hospitals have explained publicly that they define their communities using roughly five-mile radii.[176] This suggests that using census tract units, rather than counties, to obtain population data corresponds to how at least some hospitals view their own communities.

Second, while many prior studies assessed the effect on hospital spending of overall community income, or of community poverty rates, no study included in its analysis the possible effect on spending of the presence of the “near-poor” group. However, this group may be important to understanding the dynamics of community benefit spending. In the U.S., prior to the ACA’s individual mandate, the poorest individuals received medical coverage through Medicaid or other public programs. However, workers in lower-wage jobs who exceeded the eligibility thresholds for Medicaid may not have had any insurance at all. As a result, those in this near-poor group may be the patients most in need of charity care, and hospital spending may be a response to this need. In addition, “near-poor” Americans may have been particularly likely in 2012 to be under-insured. Under-insured patients also present substantial charity care needs. Elizabeth Warren’s work on individual bankruptcy suggests that a great deal of medical debt arises among individuals who have some health insurance.[177] Insofar as communities contain an under-insured near-poor group, local hospitals may respond by increasing their provision of free or discounted care.

Turning now to the community building data, my analysis here suggests that hospitals that perform well with regard to community building are also larger, but that they are located in communities with particularly high private insurance rates and in states that have unconditional community benefit requirements. Overall, however, very few hospitals are spending substantial amounts on community building. As noted, the mean total amount spent on community building equals 0.11%, with 0.0002% as the median. Many hospitals engage in no community building activities.

One likely reason that larger hospitals in areas with more privately- insured individuals provide most of these community building activities is that these hospitals may simply not have the charity care needs of larger hospitals in other communities. However, in the current political environment, larger hospitals in communities with more of privately-insured individuals may still feel pressure to get involved in their communities somehow, especially in ways that are not as expensive as providing free care to patients who are not experiencing grave financial distress. The pressures that hospitals feel to participate in their communities may be most acute in states that themselves have community benefit requirements. The states associated with higher community building rates are states where hospitals must provide some community benefit, but there is no minimum amount. States with these general requirements often define community benefit broadly so that it would encompass what the IRS itself calls community building. These community building activities, which so far tend not to consume much by way of hospital resources, would probably not be an effective way of meeting a minimum standard. However, community building could offer a promising approach for hospitals that, by reason of shifting industry norms or state regulation, need to display a nonzero degree of community engagement in a community that does not need much charity care.

This Article’s other main finding with regard to community benefit and community building relates to the role of free and discounted care. Unsurprisingly, hospitals that provide free and discounted care up to higher thresholds seem to be devoting more resources to charity care. The data do not reveal the direction of the causal effect, however. Perhaps some hospitals deliberately set out to be particularly charitable and develop policies to further this goal. These altruistically oriented hospitals might set their free and discounted care thresholds high so as to offer free and discounted care to as many patients as possible in service of this intentional charitable mission. Or, perhaps hospitals that provide free and discounted care to individuals with incomes farther above the poverty line simply wind up seeing more patients who are eligible for financial aid. Simply by virtue of following their policies, these hospitals end up having to spend more on charity care.

Regardless of the direction of the causal arrow, the data showed no connection between high thresholds for free and discounted care and the number of poor or near-poor residents in the surrounding communities. In fact, poverty rates were negatively (and significantly) associated with high free care thresholds. One reason for this may be that hospitals in poor communities, notwithstanding many patients qualifying for Medicaid, have pressing financial needs that arise from practicing in a community with few resources. As a result, these hospitals cannot afford to provide free and discounted care to patients at, say, 500% of the poverty line. It is also possible that in high-poverty communities, most hospital patients are close to the poverty line and, as a result, the hospitals see no reason to establish free care policies for patients at income levels they rarely encounter.

The data on free and discounted care policies lead into another set of this Article’s findings: those concerning hospital policies. The Article identified several instances of association between hospital and community characteristics, on the one hand, and high thresholds for free and discounted care, on the other hand. Certain state-law variables, notably an unconditional community benefit requirement and a state-level mandatory minimum level of community benefit, are associated with higher thresholds. By itself, this finding is not surprising. Hospitals in states that regulate community benefit may have needed to establish particularly generous free and discounted care policies either to meet “industry norms” within the state or to comply with an actual mandatory minimum amount. These findings may make particular sense in light of the fact that public and private insurance rates are also associated with higher thresholds for free care. If a hospital’s patients are mostly insured, then to meet a mandatory minimum threshold the hospital may have to increase the pool of patients eligible for free care. Larger hospitals in more urban and in more diverse communities are also more likely to have higher thresholds for free and discounted care—a pattern that fits with the view (cited earlier) about small rural hospitals. Some hospitals may believe they are serving their communities best by simply providing access to health care to an area that would not otherwise have it. These hospitals may not view providing free or discounted care as an important part of their mission.

The data presented here on hospital policies further show that hospitals in all types of communities are adopting the practices that Congress, the IRS, and the Treasury Department have requested in terms of billing and financial aid. Sections 501(r)(4)-(6) of the Internal Revenue Code, the statutory provisions that govern financial policies, do not go into effect until the taxable year beginning on or after the date when the IRS and Treasury publish their proposed regulations as final or temporary regulations.[178] The proposed regulations do specify that taxpayers may rely on them until final or temporary regulations are issued,[179] but the proposed regulations do not carry legal authority.[180] As a result, tax-exempt hospitals are not currently required to have policies such as a written financial aid policy or a nondiscriminatory emergency care policy. Nonetheless, most hospitals across all community types are already adopting the policies that the Schedule H asks about, and most of which Sec. 501(r) and its regulations will eventually mandate. With this mandate looming, hospitals may have decided to get an early start for fear that saying “no” now to one or more of the IRS’s policy questions may serve as an audit flag for the IRS or may send troublesome signals to stakeholders. Or hospitals may simply be preparing for the date in the near future when the IRS and the Treasury department do issue the final regulations. Probably hospitals are doing a bit of both. Additionally, many hospitals may have had some of these policies in place long before Schedule H began asking about them. Regardless, to the extent that these financial policies were not present before Schedule H, they certainly are now.

Perhaps because the policies are now so pervasive across all community types, I did not find much by way of a relationship between financial policies and hospital and community characteristics. Nor did I see any correspondence between having the financial policies in place and spending more on either the community benefit or community building activities. I found this latter non-relationship somewhat surprising. I had anticipated that hospitals slow to install compliant financial practices might also be reluctant to expend resources on their communities. However, I found no evidence that this is the case. My prior image of a small gaggle of evil hospitals greedily hoarding their revenues while gleefully putting liens on patient homes before checking their financial aid eligibility did not find support in my data.

The final finding that emerged from this Article’s analysis concerned debt collection practices. The data show that a small group of hospitals is currently authorized to carry out, and does carry out, most of the extraordinary collection actions against patients who may be eligible for financial aid. The only characteristics that these hospitals share are their small size, the lack of racial diversity in their surrounding communities, and their reliance on patients who have private insurance. Because engaging in extraordinary collection actions without reviewing financial aid eligibility will soon be illegal, the group of hospitals that is still doing this may not merit particular attention. On the other hand, as noted earlier, many extraordinary collection actions will still be acceptable under the new ACA requirements. The hospitals that are currently using extraordinary collection actions to pursue patients who may be eligible for aid may be the same hospitals that will find ways to use extraordinary collection actions in ways that Congress has yet to prohibit. For this reason, the hospitals in this group may warrant future attention from lawmakers and scholars.

VI. Evaluation of Traditional and New Affordable Care Act Requirements

The Schedule H data analyzed in Part V is of special interest because it provides a previously untapped opportunity to evaluate the legal requirements for tax-exempt hospitals on the basis of comprehensive empirical evidence about tax-exempt hospitals, their community benefits, and their financial policies. Health policy scholars Gray and Schlesinger lauded the arrival of Schedule H by saying that, with its appearance on the scene, “debate will begin anew about what should be expected of nonprofit hospitals and charitable organizations more generally.”[181] The Schedule H data can also enable Congress, the IRS, and the Treasury Department to move from guesswork to fact as lawmakers from these bodies attempt to evaluate and refine the new requirement they have put in place with the Affordable Care Act.

The data analysis in this Article raises a major question that now confronts lawmakers seeking to evaluate the rules of tax-exempt hospitals: how should the law governing tax-exempt hospitals define the pivotal concept of “community”? The results of this study suggest that how much and what kind of community benefit hospitals provide are issues that depend heavily on the way in which hospitals delineate their communities. The proposed ACA regulations give hospitals broad leeway to define their own communities. As described earlier, the regulations “provide a hospital facility with the flexibility to take into account all of the relevant facts and circumstances in defining the community it serves, including the geographic area served by the hospital facility, target populations served (for example, children, women, or the aged), and principal functions (for example, focus on a particular specialty area or targeted disease).”[182] The one proviso is that a hospital may not “define its community in a way that excludes medically underserved, low-income, or minority populations who are part of its patient populations, live in geographic areas in which its patient populations reside or . . . otherwise should be included” based on the hospital’s selected definition of community.[183]

Under the ACA requirements, every hospital will, for the first time, need to explicitly consider who makes up the community being served. Not only this, but hospitals will then have to answer this question publicly in their CHNAs. Hospitals that have already conducted and publicized CHNAs have taken a variety of approaches to the question. Some have defined communities in terms of their counties[184] or as lying within a few miles of their zip codes.[185] Some hospitals stipulate the proximate neighborhoods[186] or the municipalities served,[187] and still others incorporate both geographic and demographic markers (i.e., the city of Chicago plus children elsewhere in the state of Illinois, “members of the entertainment industry working or residing in Southern California”).[188]

While different hospitals will inevitably adopt broader or narrower definitions of community in their CHNAs, the data here suggest that hospitals are at least partially responsive to their immediately proximate communities. Hospitals in urban areas with more residents living just above the poverty line might provide more free care. Hospitals in areas with fewer free care needs may be more likely interface with their communities by building physical improvements or enhancing their environments. Hospitals in communities where people live below the poverty line may prefer financial aid policies targeted toward patients on the lower end of the income spectrum. These alternatives that this Article has identified are perhaps just a few of the ways in which hospitals react to their communities when deciding such issues as how much to spend on community benefit versus community building, and what activities to fund.

The fact that hospitals respond to their communities raises the fundamental question for lawmakers: is it acceptable that hospitals in areas with fewer needs may provide fewer community benefits? For example, is it objectionable that hospitals in areas with high rates of private insurance provide limited free care? Should those hospitals have to compensate in some way for the fact that their communities have low free care needs? Or should a hospital in an area with high insurance coverage be able to satisfy the community benefit standard by simply devoting 0.004% of its annual expenditures to leadership training for local youth? What if a hospital really is located right in the middle of a wealthy community? Under current law, that hospital has no clear obligation to do anything for disadvantaged groups.

These questions go to the core of the longstanding “community benefit” standard. By its own terms, the standard requires hospitals to serve the interest of “the community.” The contrast drawn in the original IRS revenue ruling was between “the community” and private interests. However, “the community” can still be a relatively narrow group. It might be a wealthy suburb or a group of affluent patients who can afford plastic surgery. In a stratified society like the twenty-first century United States, communities often consist largely of individuals who are similar in terms of social class, income, race, or education. To return to an example mentioned earlier in the paper, Evanston, Illinois, is a substantially different community with very different needs than areas of Chicago’s South Side.

This is not to suggest that hospitals in resource-rich communities are doing something wrong. After all, even if these hospitals are providing very little in terms of the expensive community benefits like free care, they may be responding sensitively and thoughtfully to the needs of their own communities. Literature on tax-exempt hospitals has sometimes seemed to assume that hospitals providing low levels of community benefits are behaving inappropriately. After all, these hospitals are receiving valuable tax benefits, and seemingly doing nothing in return. However, these hospitals may be doing exactly what the community benefit standard asked them to do: benefit their immediate communities.

In fact, the data here raise the possibility that hospitals in resource-rich areas have a civic orientation. They just may have fewer needs to meet, which brings their community benefit numbers down. Most notably, the data here suggest that hospitals in areas with low poverty rates and high levels of insurance coverage are likely to have particularly generous financial aid policies with high thresholds. However, a hospital in a wealthy area that offers free care up to 300% of the poverty line may still end up providing much less free care than a hospital in a lower-income area which makes free care available only for individuals who fall below the poverty line. Further, hospitals in resource-rich areas may be more likely to provide novel community enhancements that do not cost as much as free care. If a hospital’s clientele consists of patients with good insurance coverage, that hospital might actually have to experiment with community building projects like environmental cleanup or youth training. However, unless a hospital substantially reorients itself away from providing health care and toward, say, cleaning up lakes, these community building activities are likely to be less expensive than the endeavor of providing free or discounted care in an area where many patients cannot pay their hospital bills.

Yet, there remains something troubling about the fact that the community benefit standard itself imposes substantially different obligations on different hospitals. For one, this violates “vertical equity,” a longstanding tenet of tax policy. Stated in broad form, vertical equity holds that tax law should treat differently situated taxpayers differently.[189] Vertical equity is the reason many tax scholars believe that individuals with high incomes should pay high tax rates and individuals with low incomes should pay low tax rates.[190] Under the community benefit standard, however, tax-exempt hospitals that provide substantial community benefit to their needier communities take the same tax exemption as hospitals that supply very little to their more advantaged communities. That disparity violates broad notions of vertical equity.

In addition, the community benefit standard, especially as envisioned under the new ACA regulations, imposes what are arguably greater burdens on hospitals in needier communities than on their counterparts in resource-rich communities. When a hospital in a poorer community carefully considers its community’s needs and how to respond to them, it is likely to realize that tackling these needs calls for substantial resource outlays. For example, when a hospital in a working-class suburban community conducts its CHNA, it may discover that the main health problem facing its community is the need for discounted care. That may be an expensive problem to address, but the community benefit standard is potentially asking the hospital to do just that. Doing so might be a substantial burden on the hospital.

On the other hand, when a hospital in a small, wealthy rural enclave carries out its CHNA, it might find that the main health problem facing its community is frequency of drunken skiing accidents. That hospital might be conscientious and seek to respond compassionately to its community needs and to engage in best practices with regard to the community benefit standard. However, even a strict interpretation of the community benefit standard would merely require that hospital to do the best it can to educate the community about the dangers of drunk skiing. That is likely not a substantial burden on the hospital.

Furthermore, the community benefit standard treats as equivalent community benefit factors that mean something different in different contexts. However, it is not clear that the law should assign the same value to all of them. Take again the hypothetical hospital in a wealthy community with high insurance coverage. Even if that hospital has a very generous financial aid policy, and even if its charity care and community benefit numbers are high, that hospital may still be in a position to discount care for families at 1000% of the poverty line. That hospital will receive the same tax benefit as a hospital in a lower-income area that uses its charitable dollars providing free care to Medicaid beneficiaries and people just barely above the poverty line. Is it appropriate for tax law to treat discounted health care for people with household incomes of over $200,000 a year the same way as it treats discounted health care for people living below the poverty line?

This latter problem provides a vivid instance of a more general problem that arises when the federal government uses tax law to conduct anti-poverty policy. As I have discussed in earlier work, U.S. tax policy that seeks to address the needs of the poor is often ineffective in reaching the poorest individuals, even as it provides substantial benefits to the middle class or even the upper class.[191] The community benefit standard for tax-exempt hospitals addresses the needs of the poor because it asks hospitals to respond to the needs of the communities. The community benefit standard asks hospitals in needy communities to benefit those communities, but it also subsidizes through tax exemptions substantial benefits that have nothing to do with disadvantaged communities. Even the ACA regulations, as discussed above, allow room for hospitals to engage in extraordinary collection actions against poor and almost-poor debtors. While insurance coverage rates may rise after the ACA’s individual mandate becomes law, many poor and near-poor individuals will still have insurance that covers only part of what could be large medical bills.

The new regulations, while taking important steps forward, do not address these problems with the proper definition of community under the community benefit standard. For the first time in the history of the community benefit standard, the new framework does set forth a definition, but it is not one that fixes the problems identified in this Article. If anything, the CHNA component of the new regulations exacerbates these problems. The CHNA rules convey to hospitals that meeting the community benefit standard involves observing and responding to the needs of the hospital’s own community via the CHNA. However, as discussed, the CHNA regulations allow hospitals to delineate their communities as they wish. The CHNA only requires that, once a hospital chooses its community and records its needs, the hospital put in place a plan to address those needs. The hypothetical hospital that identified and responded to drunken skiing concerns is as compliant with the CHNA rules as the hospital that identifies and responded to free care needs. By requiring hospitals to respond to the needs that the CHNA uncovers, the CHNA framework arguably places a higher burden on hospitals that uncover greater needs than on hospitals that uncover insubstantial ones.

Unfortunately, there is no policy solution to these problems with the community benefit standard that does not introduce problems of its own. Even so, lawmakers have several alternatives available to them in the world of the ACA. Here, I will discuss a few of them. One, legislators or regulators from Treasury and IRS could specifically define community—for the purpose either of the community benefit standard generally or of the CHNA rules—to include disadvantaged populations. Right now, the CHNA rules tell tax-exempt hospitals that they cannot exclude disadvantaged populations that “should” otherwise fall within their communities. However, some hospitals do not have many disadvantaged groups that reasonably fall within their community borders. Under these circumstances, the federal regulations could give hospitals an affirmative obligation to draw their community boundaries so as to include some relevant disadvantaged group(s). To do this, the regulations could explicitly encourage hospitals in wealthier communities to form partnerships with hospitals in low-income areas and work with these lower-income hospitals to meet the needs of their communities.

This approach has some advantages. It asks hospital decision-makers to think broadly about community and to consider linkages between communities that might be reasonably geographically proximate yet have access to very different resources. Such an approach might serve to channel resources to serious problems that need solving, while allowing hospitals with resources the flexibility to determine how best to use those resources to help the disadvantaged.

However, this approach also has disadvantages. It would level the playing field among hospitals somewhat but not completely. If, for example, a hospital in wealthy McLean, Virginia, satisfies this new obligation by sending volunteers to vaccination day in relatively less well-off Prince George’s County, Maryland, the hospital in Prince George’s County is still going to have to respond to needs far beyond anything that the McLean hospital ever has to face. In addition, this approach would force resource-rich hospitals to expand capacity in ways that go beyond their traditional missions. A Florida facility that normally serves wealthy senior citizens with private insurance exclusively might have to stretch outside its standard activities to find an endeavor benefitting a disadvantaged group. Furthermore, some hospitals might be genuinely unable to find a disadvantaged group anywhere within a reasonably proximate geographical area. On the other hand, the IRS and Treasury could consider softer forms of this approach that would not place undue burdens on hospitals in resource-rich communities. Regulators could gently prod rather than require. For example, the regulations could simply encourage hospitals to define their communities broadly and in ways that include disadvantaged groups. Or, the Schedule H could include a question about outreach to disadvantaged groups.

In a somewhat different version of this approach, Treasury and the IRS could impose a particular broad definition of community on tax-exempt hospitals. For example, if a hospital in affluent Westchester, New York, had to take as its community the entire New York City metropolitan area, this hospital would at least have to formulate a plan to address the health problems of Staten Island and the South Bronx. This approach has the advantage of simplicity, while again requiring some hospitals in resource-rich communities to use their resources in service of pressing needs elsewhere. Furthermore, it would place less pressure on individual hospitals to identify nearby disadvantaged communities if those communities really do not exist.

However, on the negative side, the IRS and Treasury previously considered defining community broadly, but (as described above) they rejected that idea when faced with comments that expressed a preference for a “facts-and-circumstances approach.”[192] The commenters “recommended against a definition based on specified geographic boundaries,” and argued that “each hospital facility is in the best position to determine its [own] community.”[193] In addition, adopting a definition of community, such as the immediate county or metropolitan area, would do more to level the playing field among hospitals in the same urban area than among hospitals that are in geographically remote locations. But, as reported above, the data indicate that urban hospitals are the ones already providing substantial community benefits.

Yet another form of this approach would build off of Professor Colombo’s work on an “access” standard for tax exemption. Colombo proposes replacing the community benefit standard with the requirement that, in exchange for tax exemption, hospitals must provide increased “access” to health care. Hospitals could qualify for tax exemption only if they offer “access to services for previously-underserved populations or provide . . . specific services to the majority population that otherwise are not provided by the private sector.”[194] The access standard has the advantage of shedding entirely the problem of defining community. In addition, under the access standard, hospitals whose communities are not disadvantaged could pursue a different obligation: to provide some service that the market does not otherwise offer. Hospitals could probably do this without having either to expand their capacity dramatically or to identify some remote disadvantaged group to help.

On the other hand, Congress, the Treasury and the IRS have by now been actively considering the problems with the community benefit standard for more than two decades. For reasons I can only speculate about, however, none of these bodies seems to have the political will to overturn the community benefit standard and to replace it with something else. None of the four major legislative proposals—neither that of Senator Grassley nor those of Representatives Roybal, Donnelly, and Thomas—even contemplated removing the community benefit standard. Additionally, when Congress eventually did pass the ACA legislation on tax-exempt hospitals, that legislation did not touch the community benefit standard. For that reason, Professor Colombo’s idea of putting in place a new and different standard may not be a realistic option at this point. Furthermore, the access standard perhaps allows hospitals too much flexibility to offer whatever access-enhancing service they want without having to consider any social needs. The access standard might, for example, allow a hospital to merit tax exemption by opening a new cosmetic surgery wing that permitted a semi-rural community to access procedures never before available. That may not be an activity deserving of a valuable tax exemption, not just because cosmetic surgery is a luxury, but also because no one in that particular community wants it. Markets supply services in response to consumer demand. If a market has not previously offered a health service, that may be because no demand for it exists. Health procedures for which no demand exists may not be worthy of tax subsidies.

An entirely different solution to defining community might be drawn from Professor Berg’s proposal that tax-exempt hospitals should have to provide population health benefits to the communities in which they operate.[195] This approach offers the upside of asking hospitals even in resource-rich communities to consider activities that provide broad population-wide benefits. Furthermore, even communities where individuals have ample health insurance can still use public health interventions around issues like nutrition and sunscreen. The data presented above, however, point to one downside of this approach: it departs from what many hospitals are already doing with regard to community benefit. Dollar-wise, most community benefits presently take the form of free or discounted care. Asking those hospitals that are not currently spending much on community benefit to devote funds to population health might be an improvement over the status quo. However, for those hospitals that are already devoting substantial resources to free care, asking for additional work on population health might be unduly burdensome. In addition, the population health approach would not replace the community benefit standard entirely, so it would not remedy many of the problems that the standard currently presents.

A completely different solution would be to offer particular tax benefits for certain health-enhancing activities. The idea here would be to tie the amount of the tax subsidy to the amount of benefit that a hospital provided. For example, the IRS could offer a refundable tax credit equal to the amount of free care provided to patients below the poverty line. This is similar to law professor Nina Crimm’s proposal to eliminate the tax exemption for hospitals and instead grant some form of tax-favored treatment to for-profit and not-for-profit health care organizations that engage in worthy activities.[196] One could envision different variants of this proposal that would subsidize different specified activities—perhaps just free care, perhaps every community benefit or community building activity, perhaps some separate list of health-related benefits.

On the upside, this approach would directly address the vertical equity problem as well as the longstanding critique that the tax exemption for hospitals has little relationship to the level of benefit that hospitals actually provide. Furthermore, it would allow Congress not only to identify and respond to the most pressing health needs that the country currently faces, but to design a tax program that directly meets those needs. However, on the downside, as mentioned above, Congress does not appear to have the political will to overhaul the community benefit framework entirely, which is what this plan would entail. A less radical version of this proposal would be to maintain the current framework, but to add a credit on top of it. However, tax-exempt hospitals, which do not currently pay income tax on much of their net income, would not be able to use credits against taxable income, and Congress has not yet experimented with offering substantial refundable credits in excess of tax liability to tax-exempt organizations.

Moreover, a credit program would represent a large new federal expense on top of the existing valuable tax exemption for hospitals. Still further, Congress might experience problems in deciding exactly what to subsidize via a credit. The federal government already has Medicaid in place to provide health care for the poor. Would the proposed tax-credit serve mostly as a tax benefit for hospitals’ unreimbursed Medicaid? If so, why not just increase Medicaid reimbursements? Or would the credit take a broad approach and subsidize each of the community building activities as well? If yes, that raises the question of why the federal tax code would be paying hospitals, which are supposed to specialize in health care, to develop workforces and clean up the environment. These questions might have good answers, but Congress would have to agree on them before enacting a credit plan.

As these comments make clear, each potential approach to solving the existing difficulties with the community benefit standard has advantages and disadvantages of its own. Among these alternatives, no approach emerges as the clear winner, although each one perhaps has elements that lawmakers might want to consider. Regardless of how they choose to proceed, however, lawmakers should recognize and contend with the central fact highlighted in the foregoing analysis of the Schedule H data: namely, the enormous extent to which tax-exempt hospitals’ legal obligations and activities depend on how hospitals define their communities. Even more, the data suggest that tax-exempt hospitals, on the whole, are already working to respond to the needs of, and to deliver benefits to, their specific communities. What the data show, in other words, is that the community benefit standard, for all of its flaws, has given hospitals the flexibility to assess and address the needs of their immediate communities. That is exactly what most tax-exempt hospitals are doing.

VII. Conclusion

According to current federal tax law, hospitals merit tax-exempt status insofar as they meet the standard of “community benefit,” a standard that has long been controversial. The Article has reported new data on more than 2,100 tax-exempt hospitals. These data on how hospitals actually meet the community standard provide a first-time opportunity to analyze in some depth how hospitals are currently earning their tax exemptions, and to consider how well the legal requirements for tax-exempt hospitals are working. This analysis is particularly important because Congress and the IRS have recently implemented the Affordable Care Act’s new rules for tax-exempt hospitals. Lawmakers are still evaluating the extent to which these rules address existing problems with the community benefit standard, but they have been conducting this evaluation in an information vacuum.

This Article examined the new Schedule H data to determine what tax-exempt hospitals really are doing to benefit their communities and what problems may be emerging which tax law might address. The data analysis suggested that some hospitals, primarily large ones in urban areas with populations living close to the poverty line, are providing substantial amounts of all of the community benefits that tax law currently envisions. At the same time, a different group of hospitals, especially those in areas with high private insurance rates, are engaging in what the IRS has called community building activities. In addition, some hospitals have adopted particularly high thresholds for free and discounted care, thresholds that lead to higher amounts of community benefit. Further, while hospitals across all of these communities are adopting financial policies designed to comply with the new legal requirements, smaller hospitals in less diverse communities are still engaging in debt collection practices that the new legal requirements aim to discourage.

These patterns suggest that, in determining what community work they should undertake, the great majority of tax-exempt hospitals are (at least partially) responding to the needs of their immediate communities. This finding raises a range of policy questions, however, about how hospitals should conceptualize their communities. Is it appropriate to allow tax-exempt hospitals to fulfill their community benefit obligations by benefiting communities that already have substantial resources? This Article considers that question, and probes several possible ways that lawmakers might address it going forward.

Acknowledgments: Thanks to Chas Camic, Fred Goldberg, Andy Grewal, Heinz Klug, Alex Reid, Alex Tahk, Bill Whitford, the participants in the University of Wisconsin Law School faculty workshop, and the members of the Fall 2008 Tax Policy & Health Care in America seminar at New York University School of Law for useful suggestions and discussions. In addition, thanks to the Graduate School at the University of Wisconsin and a University of Wisconsin Summer Research Fellowship for funding; and to Jordan Behmke, Patrick Kearney and Angela N. Muñoz for excellent research assistance. All errors are my own.

[1] Sara Rosenbaum & Josh Margulies, Tax-Exempt Hospitals and the Patient Protection and Affordable Care Act: Implications for Public Health Policy and Practice, 126 Pub. Health Rep. 283, 283 (2011).

[2] See generally Susannah Camic Tahk, Everything is Tax, 50 Harv. J. on Legis. 67 (2013).

[3] See I.R.C. §§ 501(c), 527–28 (2012).

[4] Boris I. Bittker & Lawrence Lokken, Federal Taxation of Income, Estates & Gifts ¶¶ 100.1.1–100.1.2 (2014), available at 1997 WL 440008.

[5] I.R.C. § 170; (West Supp. 2013); see also Staff of Joint Comm. on Tax’n, 109th Cong., Historical Development and Present Law of the Federal Tax Exemption for Charities and Other Tax-Exempt Organizations (Joint Comm. Print 2005), available at

[6]I.R.C. § 501(c)(3) (2012); Treas. Reg. § 1.501(c)(3)-1(d)(2) (as amended in 2014).

[7] Bittker & Lokken, supra note 5, ¶ 100.2.

[8] Treas. Reg. § 1.501(c)(3)–1(b), (c) (as amended in 2014).

[9] I.R.C. § 501(c)(3) (2012); see also id. § 4958; Treas. Reg. § 1.501(c)(3)-1(f)(2)(ii) (as amended in 2014).

[10] I.R.C. §§ 501(c)(3), 4958. (2012).

[11] I.R.C. § 501(c)(3) (2012). See also Lloyd Hitoshi Mayer & Joseph R. Ganahl, Taxing Social Enterprise, 66 Stan. L. Rev. 387, 409 (2014).

[12] Treas. Reg. § 1.501(c)(3)-1(d)(1)(ii) (as amended in 2014).

[13] Id.

[14] See, e.g., Asmark Inst., Inc. v. Comm’r, 101 T.C.M. (CCH) 1067 (2011).

[15] Nina J. Crimm, Evolutionary Forces: Changes in For-Profit and Not-for-Profit Health Care Delivery Structures; A Regeneration of Tax Exemption Standards, 37 B.C. L. Rev. 1, 10–11 (1995) (citing Charles Rosenberg, The Care of Strangers: The Rise of America’s Hospital System 5, 18, 116–65 (1987)); Marshall W. Raffel & Norma K. Raffel, The U.S. Health System: Origins and Functions 241–46 (3d ed. 1980); Stanley Joel Reiser, Medicine and the Reign of Technology 152 (1978); Paul Starr, The Social Transformation of American Medicine 145–79, (1982); William H. Williams, America’s First Hospital: The Pennsylvania Hospital, 1751-1841, at 2 (1976); Robert S. Bromberg, The Charitable Hospital, 20 Cath. U. L. Rev. 237, 239 (1970); Henry Hansmann, The Evolving Law of Nonprofit Organizations: Do Current Trends Make Good Policy?, 39 Case W. Res. L. Rev. 807, 810–11 (1988-89); Rosemary Stevens, “A Poor Sort of Memory”: Voluntary Hospitals and Government Before the Depression, 60 Milbank Memorial Fund Q.: Health & Society 551, 552–55 (1982).

[16] See, e.g., Crimm, supra note 15, at n.140 (citing I.T. 2421, 7-2 C.B. 150 (1928)).

[17] Gabriel O. Aitsebaomo, The Nonprofit Hospital: A Call For New National Guidance Requiring Minimum Annual Charity Care To Qualify For Federal Tax Exemption, 26 Campbell L. Rev. 75, 86 (2004) (citing J. Rogers Hollingsworth & Ellen Jane Hollingsworth, Controversy About American Hospitals: Funding, Ownership, and Performance 66–67 (1987)).

[18] Rev. Rul. 56-185, 1956-1 C.B. 202.

[19] Id.

[20] Id.

[21] John D. Colombo, The Role of Access in Charitable Tax Exemption, 82 Wash. U. L.Q. 343, 348 (2004).

[22] Id. at 348-349.

[23] Id. at 348 (citing Daniel M. Fox & Daniel C. Schaffer, Tax Administration as Health Policy: Hospitals, the Internal Revenue Service, and the Courts, 16 J. Health Pol., Pol’y & L. 251, 261–62 (1991)).

[24] Fox & Schaffer, supra note 23 at 261–62.

[25] Rev. Rul. 69-545, 1969-2 C.B. 117.

[26] Id. For use of the term “community benefit standard” and information about how it has taken hold, see, eefor example, James J. Fishman & Stephen Schwarz, Nonprofit Organizations Cases & Materials 384–85 (2d ed. 2000); Thomas K. Hyatt & Bruce R. Hopkins, The Law of Tax-Exempt Healthcare Organizations 15, 529–32 (2d ed. 2001); Colombo, supra note 21, at 347; and Crimm, supra note 15, at 44–45.

[27] Rev. Rul. 69-545, 1969-2 C.B. 117.

[28] Id.

[29] Rev. Rul. 83-157, 1983-2 C.B. 94.

[30] I.R.S. Nat’l Office Field Serv. Mem., 200110030, (Feb. 5, 2001) [hereinafter FSA 200110030]. For discussion of this FSA, see Douglas M. Mancino, The Impact of Federal Tax Exemption Standards on Health Care Policy and Delivery, 15 Health Matrix 5 (2005).

[31] FSA 200110030, supra note 30.

[32] Lawrence M. Brauer et al., Internal Revenue Serv., Exempt Organizations Continuing Professional Education (CPE) Technical Instruction Program For Fiscal Year 2002, Topic D: Update on Health Care 173 (2002), available at

[33] Mary A. Crossley, Non-Profit Hospitals, Tax Exemption and Access for the Uninsured, 2 Pitt. J. Envtl. & Pub. Health L. 32–36 (2008).

[34] Corey S. Davis et al., Leveraging the Patient Protection and Affordable Care Act’s Nonprofit Hospital Requirements to Expand Access and Improve Health in Low-Income Communities, 45 Clearinghouse Rev. 403, 406 (2012).

[35] Cecilia M. Jardon McGregor, Comment, The Community Benefit Standard for Non-Profit Hospitals: Which Community and for Whose Benefit?, 23 J. Contemp. Health L. & Pol’y 302, 318 (2007).

[36] The Tax-Exempt Hospital Sector: Hearing Before the H. Comm. on Ways & Means, 109th Cong. 9 (2005) (statement of Mark Everson, Comm’r of Internal Revenue Serv.) [hereinafter Tax-Exempt Hospital Sector].

[37] Colombo, supra note 21, at 369.

[38] Jessica Berg, Putting the Community Back into the ‘Community Benefit’ Standard, 44 Ga. L. Rev. 375, 390 (2010).

[39] See generally Internal Revenue Serv., IRS Exempt Organizations (TE/GE) Hospital Compliance Project Final Report (2009) [hereinafter IRS Hospital Report], available at

[40] Id. at 3.

[41] Id. at 4.

[42] See U.S. Gov’t Accountability Office, GAO-08-880, Nonprofit Hospitals: Variation in Standards and Guidance Limits Comparison of How Hospitals Meet Community Benefit Requirements (2008).

[43] Berg, supra note 39, at 388.

[44] See generally Leah Snyder Batchis, Comment, Can Lawsuits Help the Uninsured Access Affordable Hospital Care?: Potential Theories for Uninsured Patient Plaintiffs, 78 Temp. L. Rev. 493, n.104 (2005) (citing Complaint at 6, Hutt v. Albert Einstein Med. Ctr., No. 04-3440, 2004 WL 1732445 (E.D. Pa. July 21, 2004)).

[45] Id. at 506–07.

[46] Lucette Lagnado, Anatomy of a Hospital Bill: Uninsured Patients Often Face Big Markups on Small Items; ‘Rules are Completely Crazy’, Wall St. J. (Sept. 21, 2004, 12:01 AM), [hereinafter Lagnado, Anatomy]; Lucette Lagnado, Hospitals Try Extreme Measures to Collect Their Overdue Debts: Patients Who Skip Hearings On Bills Are Arrested; It’s a ‘Body Attachment’, Wall St. J. (Oct. 30, 2003, 12:01 AM), [hereinafter Lagnado, Extreme Measures].

[47] Lagnado, Anatomy, supra note 46.

[48] Id.

[49] Id.

[50] Id.; Lagnado, Extreme Measures, supra note 46.

[51] See generally Taking the Pulse of Charitable Care and Community Benefits at Nonprofit Hospitals: Hearing Before the Senate Comm. on Fin., 109th Cong. (2006) [hereinafter Taking the Pulse]; Tax-Exempt Hospital Sector, supra note 36.

[52] Taking the Pulse, supra note 51, at 16-18 (statement of Ray Hartz, Executive Director, Legal Aid Society of Eastern Virginia, Inc., Norfolk, VA).

[53] H.R. 790, 102d Cong. (1991).

[54] H.R. 1374, 102d Cong. (1991).

[55] H.R. 1374, 102d Cong. (1991).

[56] H.R. 6420, 109th Cong. (2006).

[57] See Senate Committee on Finance—Minority, Tax Exempt Hospitals: Discussion Draft, Tax Notes Today, July 18, 2007, at 140.

[58] I.R.C. § 501(r)(3)(A)(i) (2012).

[59] I.R.C. § 501(r)(3)(A)(ii) (2012).

[60] I.R.C. § 501(r)(3)(B)(i) (2012).

[61] I.R.C. § 501(r)(3)(B)(ii) (2012).

[62] I.R.C. § 4959 (2012).

[63] Community Health Needs Assessments for Charitable Hospitals, 78 Fed. Reg. 20523 (proposed Apr. 5, 2013).

[64] David van den Berg, IRS Hopes to Publish Final Charitable Hospital Regs by Year-End, Tax Notes Today, Apr. 15, 2014.

[65] Community Health Needs Assessments for Charitable Hospitals, 78 Fed. Reg. at 20529.

[66] Id. at 20528.

[67] Id. at 20529.

[68] Id.

[69] Id.

[70] Id.

[71] Id.

[72] Id.

[73] Id.

[74] Id.

[75] Id. at 20531-23222.

[76] Id. at 20532.

[77] I.R.C. § 501(r)(4) (2012).

[78] I.R.C. § 501(r)(5)(A) (2012).

[79] I.R.C. § 501(r)(5)(B) (2012).

[80] I.R.C. § 501(r)(6) (2012).

[81] Additional Requirements for Charitable Hospitals, 77 Fed. Reg. 38148 (proposed June 26, 2012).

[82] Van den Berg, supra note 64.

[83] Additional Requirements for Charitable Hospitals, 77 Fed. Reg. at 38151.

[84] Id. at 38152–53.

[85] Id. at 38151–55.

[86] Id. at 38155–56.

[87] Id. at 38156.

[88] Id.

[89] Id.

[90] Id.

[91] Id.

[92] Id. at 38156-59.

[93] Internal Revenue Serv., Sched. H, Form 990 (2013), available at–2013.pdf.

[94] Schedule H: Prior Year Products, Internal Revenue Serv. (last visited Aug. 24, 2014).

[95] Internal Revenue Serv., Instructions for Form 990 and Form 990-EZ (2007), available at–2007.pdf.

[96] Gary J. Young et al., Provision of Community Benefits by Tax-Exempt U.S. Hospitals, 368 New Eng. J. Med. 1519 (2013).

[97] Id. at 1519.

[98] Id. at 1526.

[99] Id. at 1519.

[100] Id.

[101] Id.

[102] Id. at 1523.

[103] Id.

[104] Id.

[105] Id.

[106] Id.

[107] Erik Bakken & David Kindig, Does Nonprofit Hospital Community Benefit Vary by State?, J. Pub. Health Mgmt. & Prac. (forthcoming 2014).

[108] See generally id.

[109] Id. at 3.

[110] Id.

[111] Simone Rauscher Singh, Community Benefit in Exchange for Non-Profit Hospital Tax Exemption: Current Trends and Future Outlook, 39 J. Health Care Fin. 32, 35 (2013). See generally Daniel B. Rubin et al., Evaluating Hospitals’ Provision of Community Benefit: An Argument for an Outcome-Based Approach to Nonprofit Hospital Tax Exemption, 103 Am. J. Pub. Health 612 (2013) (exploring Schedule H critically, but not independently analyzing its data).

[112] Bradford H. Gray & Mark Schlesinger, Charitable Expectations of Nonprofit Hospitals: Lessons From Maryland, 28 Health Aff. w809, w810 (2009).

[113] Id. at w814.

[114] Id. at w815.

[115] Simone Rauscher Singh, Not-for-Profit Hospitals’ Provision of Community Benefit Beyond Charity Care: Is There a Trade-off Between Free Medical Care and Other Health Services Provided to the Community?, 39 J. Health Care Fin. 42 (2013).

[116] Gloria J. Bazzoli et al., Community Benefit Activities of Private, Nonprofit Hospitals, 35 J. Health Pol., Pol’y & L. 999 (2010).

[117] Gloria J. Bazzoli et al., The Influence of Health Policy and Market Factors on the Hospital Safety Net, 41 Health Services Research 1159 (2006).

[118] Amy J. Davidoff et al., The Effect of Changing State Health Policy on Hospital Uncompensated Care, 37 Inquiry 253 (2000).

[119] Cong. Budget Office, Pub. No. 2707, Nonprofit Hospitals and the Provision of Community Benefits, (2006), available at

[120] Kenneth E. Thorpe et al., The Impact of HMOs on Hospital-Based Uncompensated Care, 26 J. Health Pol., Pol’y & Law 543 (2001).

[121] Helen Schneider, Paying Their Way? Do Nonprofit Hospitals Justify Their Favorable Tax Treatment?, 44 Inquiry 187 (2007).

[122] Michael A. Morrisey et al., Do Nonprofit Hospitals Pay Their Way?, 15 Health Aff. 132 (1996).

[123] Sean Nicholson et al., Measuring Community Benefits Provided by For-Profit and Nonprofit Hospitals, 19 Health Aff. 168 (2000).

[124] Edward C. Norton & Douglas O. Staiger, How Hospital Ownership Affects Access to Care for the Uninsured, 25 RAND J. of Econ. 171 (1994).

[125] Janet P. Sutton et al., The Role of Private Hospitals in the California Healthcare Safety Net: A Comparison of Charitable Contributions Reported by the For-Profit and Non-Profit Hospitals in 1998, 2 Pub. Pol’y Brief H Series 6 (2002).

[126] See generally Jill R. Horwitz & Austin Nichols, What Do Nonprofits Maximize? Nonprofit Hospital Service Provision and Market Ownership Mix, (Nat’l Bureau of Econ. Research, Working Paper No. 13246, 2007), available at; Jill R. Horwitz, Does Corporate Ownership Matter? Service Provision in the Hospital Industry, (Nat’l Bureau of Econ. Research, Working Paper No. 11376, 2005), available at; Jill R. Horwitz, Nonprofit Ownership, Private Property, and Public Accountability, 25 Health Aff.W308 (2006); Jill R. Horwitz, Why We Need the Independent Sector: The Behavior, Law, and Ethics of Not-for-Profit Hospitals, 50 UCLA L. Rev. 1345 (2003).

[127] Regina E. Herzlinger & William S. Krasker, Who Profits from Nonprofits?, Harv. Bus. Rev., Jan.-Feb. 1987, at 93.

[128] Barbara Arrington & Cynthia Carter Haddock, Who Really Profits from Not-For-Profits?, 25 Health Services Res. 291 (1990).

[129] Alva O. Ferdinand et al., Community benefits provided by religious, other nonprofit, and for-profit hospitals: A longitudinal analysis 2000-2009, 39 Health Care Mgmt. Rev. 145 (2014).

[130] Id. at 151–52.

[131] Gregory O. Ginn & Charles B. Moseley, Community Health Orientation, Community-based Quality Improvement, and Health Promotion Services in Hospitals, 49 J. of Healthcare Mgmt. 293 (2004).

[132] E. Jose Proenca et al., Community Orientation in Hospitals: An Institutional and Resource Dependence Perspective, 35 Health Serv. Res. 1011 (2000).

[133] Frances A. Kennedy et al., Do non-profit hospitals provide more charity care when faced with a mandatory minimum standard?:Evidence from Texas, 29 J. Acct. & Pub. Pol’y 242 (2010).

[134] Gregory O. Ginn & Charles B. Moseley, The Impact of State Community Benefit Laws on the Community Health Orientation and Health Promotion Services of Hospitals, 31 J. Health Pol., Pol’y & L. 321 (2006).

[135] Id. at 321.

[136] Id.

[137] Gregory O. Ginn et al., Community benefit laws, hospital ownership, community orientation activities, and health promotion services, 34 Health Care Mgmt. Rev. 109 (2009). See also Charles B. Moseley et al., The Long-term Coercive Effect of State Community Benefit Laws on Hospital Community Health Orientation, 7 Nev. J. Pub. Health 14 (2010).

[138] Peter G. Szilagyi et al., Evaluating Community Engagement in an Academic Medical Center, 89 Acad. Med. 585 (2014).

[139] Lloyd Michener et al., Aligning the Goals of Community-Engaged Research: Why and How Academic Health Centers Can Successfully Engage With Communities to Improve Health, 87 Acad. Med. 285 (2012); J. Lloyd Michener et al., Improving the Health of the Community: Duke’s Experience with Community Engagement, 83 Acad. Med. 408 (2008).

[140] Jessica G. Burke et al., What Can Be Learned from the Types of Community Benefit Programs that Hospitals Already Have in Place?, J. Health Care for Poor & Underserved, Feb. 2014, at 165.

[141] Jeffrey A. Alexander et al., Community Accountability Among Hospitals Affiliated with Health Care Systems, 78 Milbank Q. 157 (2000).

[142] From this point forward, when referring to tax-exempt hospitals, I will use the shorthand “hospitals,” because all hospitals in the dataset are tax-exempt.

[143] American Community Survey: 2012 Data Release, U.S. Census Bureau, (last visited Aug. 24, 2014).

[144] Univ. of Md.-Balt. Cnty., Hilltop Inst., Community Benefit State Law Profiles Comparison, onal size, teaching facilities,Hilltop Inst.,, (last visited Aug. 24, 2014).

[145] Univ. of Md.-Balt. Cnty., Hilltop Inst., About the Community Benefit State Law Profiles, Hilltop Inst., (last visited Aug. 24, 2014).

[146] Id.

[147] Id.

[148] Internal Revenue Serv., Instructions for Form 990 Return of Organization Exempt From Income Tax, at 9-10 (2013) available at

[149] Internal Revenue Serv., Instructions for Schedule H (Form 990) at 12 (2013), available at

[150] Id. at 14.

[151] Id. at 15.

[152] Id. at 17.

[153] Id.

[154] Id. at 18.

[155] Id.

[156] Id. at 20.

[157] Id. at 4.

[158] Id.

[159] Id.

[160] Id.

[161] Id.

[162] Id.

[163] Id. at 20–21.

[164] Id.

[165] Id.

[166] For a description of this model and how to interpret its results, see David S. Moore et al., Introduction to the Practice of Statistics 611–42 (8th Ed. 2014) and Edward R. Tufte, Data Analysis for Politics and Policy 65–163 (1974).

[167] Tufte, supra note 166, at 78.

[168] The correlation coefficient was 0.13.

[169] In each of these models, I applied a square-root transformation to the dependent variable to reduce the skew in the dependent variable. This has the benefit of reducing the impact of what appeared to be a handful of substantial outliers.

[170] The reason the effect size is 0.02 rather than 0.132 is because, as discussed, to reduce the effect of outliers, I used a square-root transformation of the dependent variable. That means that the reported coefficients do not by themselves represent the marginal effects of the independent variables. Instead, that I have to take the derivative of the OLS function to determine the marginal effect of the dependent variable in each case, which is how I arrived at these effect sizes.

[171] The effect was significant using a one-tailed test.

[172] Young et al., supra note 96, at 1526.

[173] Gray & Schlesinger, supra note 112 at w814.

[174] Taking the Pulse, supra note 51 (statement of Scott A. Duke, CEO, Glendive Medical Center, Glendive, MT).

[175] IRS Hospital Report, supra note 39, at 39–41.

[176] See infra Part V.

[177] See generally Melissa B. Jacoby & Elizabeth Warren, Beyond Hospital Misbehavior: An Alternative Account of Medical-Related Financial Distress, 100 Nw. U. L. Rev. 535 (2006);

Melissa B. Jacoby et al., Rethinking the Debates Over Health Care Financing: Evidence from the Bankruptcy Courts, 76 N.Y.U. L. Rev. 375 (2001).

[178] Additional Requirements for Charitable Hospitals, 77 Fed. Reg. 38148, 38159 (proposed June 26, 2012) (to be codified at 26 C.F.R. pt. 1).

[179] Id.

[180] For a discussion of this issue, see Kristin E. Hickman, Coloring Outside the Lines: Examining Treasury’s (Lack of) Compliance with Administrative Procedure Act Rulemaking Requirements, 82 Notre Dame L. Rev. 1727, 1734 (2007).

[181] Bradford H. Gray & Mark Schlesinger, The Accountability Of Nonprofit Hospitals: Lessons from Maryland’s Community Benefit Reporting Requirements, 46 Inquiry 122, 122 (2009).

[182] Community Health Needs Assessments for Charitable Hospitals, 78 Fed. Reg. 20523, 20529 (proposed Apr. 5, 2013) (to be codified at 26 C.F.R. pts. 1, 3).

[183] Id.

[184] See, e.g., Holy Cross Hosp. of Silver Spring, Md., Community Health Needs Assessment 4 (2012), available at; Meriter Health Serv., 2012 Community Health Needs Assessment 5 (2012), available at

[185] See, e.g., Advocate Trinity Hosp., 2011–2013 Community Health Needs Assessment 4 (2011), available at; Ctr Prof’l Research Consultants, Inc., 2012 PRC Community Health Needs Assessment Report 7 (2012), available at

[186] See, e.g., Bronx-Lebanon Hosp. Ctr., 2013 Community Needs Assessment 1–13 (2013), available at; Ngozi Moses, Brooklyn Perinatal Network et al., The Need for Caring in North and Central Brooklyn: A Community Needs Assessment 1 (2013), available at….PDF.

[187] See, e.g., Bassett Med. Ctr., Community Health Needs Assessment 1 (2013), available at!/2013-BMC-CHNA.pdf?item_id=182938650&version_id=182938651; Kaiser Permanente, 2013 Community Health Needs Assessment: Kaiser Foundation Hospital—Redwood City 12 (2013), available at

[188] See, e.g., Ann & Robert H. Lurie Children’s Hosp. of Chi., Community Health Needs Assessment 2013, at 10 (2013), available at; Motion Picture & Television Fund, Community Health Care Needs Assessment 11 (2013), available at—part/2013-MPTF-Community-Health-Needs-Assessment.pdf.

[189] Joseph J. Cordes, Vertical Equity, in The Encyclopedia of Taxation and Tax Policy (Joseph J. Cordes et al. eds., 2d ed.), available at

[190] Id.

[191] Susannah Camic Tahk, The Tax War on Poverty, 57 Ariz. L. Rev. (forthcoming 2014) (draft at 44–46), available at

[192] Community Health Needs Assessments for Charitable Hospitals, 78 Fed. Reg. at 20529.

[193] Id.

[194] Colombo, supra note 21, at 345.

[195] See generally Berg, supra note 38.

[196] See generally Crimm, supra note 15.