Malfeasance in the Charitable Sector: Determinants of “Soft” Corruption at Nonprofit Organizations

Abstract

The causes and effects of organizational corruption have been widely examined in the literature, including malfeasance that is specific to nonprofits organizations. This article draws a distinction between outright (illegal) and “soft” corruption—the latter referring to the continued and deliberate misuse of donated funds to benefit officers of the nonprofit, with little (in some instances less than 5%) going to the nonprofit’s supposed cause. Soft corruption enables organizations to function essentially as counterfeit charities without risking legal peril. An empirical model encompassing approximately 450 randomly drawn nonprofits tests the determinants of this form of misbehavior. Ultimately it is shown that either the existence of an independent voting board or the conduct of an independent audit are the most important means of preventing soft corruption; far more important than state- or federally-mandated reporting requirements.

Keywords:

• corruption
• nonprofits
• organizational management

The economics and pathologies of corrupt organizations have been extensively scrutinized beginning with seminal work by Banfield (1975), Rose-Ackerman (1975), and Sherman (1980). Initially, attention was focused on the general topic of organizational corruption, but more recent work has addressed malfeasance at specific types of institution (e.g., educational establishments). In addition to descriptive work, some scholars have addressed the means by which—through either increased oversight or transformed managerial arrangements—corruption can be addressed.

More recent work on corruption, its causes and consequences, can be found in Jain (2001) and Aidt (2003). Organizational malfeasance has been examined by Vardi and Wiener (1996), Luo (2004), Baucus and Beck-Dudley (2005), and Pinto, Leana, and Pil (2008), while corrupt practices in the corporate sector were detailed by Baucus (1994), and Anand, Ashforth, and Joshi (2004). In addition, governmental corruption was analyzed by Banfield (1975) and Rose-Ackerman (2005). Baucus and Near (1991) and Steinberg (2010) focus on corporate governance and its relationship to ethical lapses.

Nonprofits are deemed to fulfill a special role in the economy. Given the goals of the charitable sector, it might be assumed that self-interested individuals would choose careers elsewhere, but malfeasance at nonprofits is surprisingly common. Works by Fremont-Smith (2004), Fremont-Smith and Kosaras (2003), Gibelman and Gelman (2001), Archambeault, Webber, and Greenlee (2014), Rhoades and Packel (2009), Holfreter (2008), and Greenlee, Fischer, Gordon, and Keating (2007) have all examined corruption and fraud at charities and nongovernmental organizations. Tangentially, Heyneman (2004) and MacWilliams (2002) looked at corrupt practices at educational institutions, particularly the “selling” of college degrees. Light (2008) provides an analysis of how the typical American’s views on charitable organizations have been impacted over time, noting a significant drop in the public’s confidence in the nonprofit sector. Similarly, Harris, Petrovits, and Yetman (2015) demonstrate that donations rise when charities perform well on seven objective performance criteria. O’Neill (2009) provides a somewhat contradictory analysis of views on charitable organizations (using Light’s pre-2008 data).

This article will examine the underlying causes of corruption at nonprofit institutions, arguing that differences in stakeholder relationships, notably in comparison to organizations such as private businesses, make such behavior possible. As a rule, malfeasance in the corporate sector is checked either through legal means (threat of prosecution), or by instituting control mechanisms that prevent managers from acting against interested parties. In order for such controls to work, however, stakeholders must be aware of transgressions (a feedback loop)—a situation that does not usually prevail in the nonprofit sector. Small, intermittent donors to charitable organizations are not only unlikely to demand an accounting of how their donations were spent, but the charity would, in all probability, be unable to detail the impact of any specific donation. This results in an agency problem, where contributors surrender control of their donations to nonprofit managers, potentially leading to either a redirection to a cause that is more in line with the managers’ wishes, or misuse of the contribution (see Jegers, 2009) for an overview of the literature on agency problems at nonprofits).

Corruption or Simply Misbehavior

Some actions on the part of charitable organizations can be seen as simple misbehavior, such as making inappropriate use of their donations (at least in the eyes of benefactors). While this form of malfeasance violates the trust of the contributor, it does not involve personal gain. In the flurry of donations after the September 11 terrorist attacks, the American Red Cross was flooded with contributions that, in the view of the organization, outstripped any reasonable need. According to critics of the Red Cross, the “excess” donations were used for other purposes deemed more pressing. More generally, those that give to a cause through a bequest surrender control over how the assets are put to use, and little recourse exists even if the funds are redirected to a cause that may have little to do with the interests of the benefactor, unless the donor’s decedents somehow intervene. The overall effect of misdirected use of funds may be benign, or it may actively work against the true wishes of donors, and may discourage future giving by those considering leaving money to a particular charity.

Behavior at nonprofits that is rightly labeled corruption includes the misappropriation of funds for personal use or outright theft (see below for examples of prosecutable conduct at charitable institutions). Due to the tenuous nature of the stakeholder (i.e., donor) relationships at most nonprofit organizations, however, a paradox arises: Nonprofit managers may fleece the organization through excess fundraising expenses and salaries rather than risk the commission of an actual crime (e.g., America’s Worst Charities [http://www.tampabay.com/topics/specials/worst-charities], annual report by the Tampa Bay Times). In this context, the commission of actual criminal activity seems both unnecessary and potentially self-harming. This circumstance will be referred to as “soft corruption”: practices that are unethical but not prosecutable—the characteristics of rogue, not unlawful, charities, those that collect donations but give little or nothing to any cause. It has been argued that high fund-raising expenses can represent a charity that is innovative, with a strong vision and significant accomplishments. The criteria that will be used to define “soft corruption” are a failure on seven financial measures that overall indicate the performance of an organization.

This approach is not without controversy. While the National Charities Information Bureau and the American Institute of Philanthropy both set minimum threshold levels for program expenses—a minimum of 50% at the NCIB and 60% at AIP—these standards have been questioned by others who evaluate charitable performance. Weber (1994) argues that fundraising efficiency is best utilized when comparing the performance of like organizations. Conversely, Hager, Pollak, and Rooney (2001) note that controlling for group-specific characteristics explain only a minor part of the differences in fund-raising efficiency. McLean and Coffman (2004) in “Why Ratios Are Not the Last Word” assert that program expenses should only be used in the context of other financial variables, and provide a side-by-side comparison of two hypothetical charities—one that would be deemed more effective based on financials, but less impactful based on outcomes. Charity watchdogs such as Charity Navigator and CharityWatch also caution against uncritical use of financial ratios. Batts (2013) argues that the calculation of financial ratios at nonprofits is a form of “voodoo” and should not be the only basis for evaluating charities. Batt instead argues for outcome evaluation: is the charity doing what it promised to its donors? Similarly, Larkin (2013) argues that outcome measures are more appropriate than financial measures.

It is not asserted here that exceptionally poor fundraising efficiency represents criminal malfeasance (as in prosecutable activity). If it did, the concerns outlined above would be mitigated. Charitable organizations reporting on their 990 s that 5 to 10% of donations go to their expressed cause would be prosecuted, a situation that would rapidly end soft corruption (or force nonprofit executives to commit fraud in reporting their financials). Rather, the argument is that donors, arguably the most important stakeholders of charitable organizations, are being poorly served when nearly all contributions are spent on either fundraising or management expenses.

REGULATORY CONSTRAINTS ON MALFEASANCE AT NONPROFITS

The regulatory environment within which charitable organizations operate includes multiple levels of oversight, but (in practice) ineffective disclosure requirements. Charitable nonprofits must maintain their 501(c)(3) standing with the Internal Revenue Service in order to solicit tax-deductible contributions. When charities are determined to be wholly corrupt organizations, legal action is taken by states’ attorneys general under fraud statutes. At the national level, The Federal Trade Commission (FTC) can file civil actions against sham charities when it concludes they are little more than money-churning entities for their managers (In May 2015, for example, the FTC accused four cancer charities of collecting nearly $185 million in donations and channeling little, if any, of the money to appropriate causes). Some individual states impose disclosure requirements on nonprofits, mandating that organizations that reach a threshold size (which varies by state) release audited financials. The Urban Institute’s Center on Nonprofits and Philanthropy provides a complete breakdown of the differing requirements by state (Lott, Boris, Goldman, Johns, Gaddy, & Farrell, 2016). Requiring release of Form 990 would duplicate current federal law.

The problem with these forms of oversight is that, until malfeasance reaches some threshold, a nonprofit can continue to operate. The four cancer charities closed by the FTC (noted above) had operated and received donations for years. Disclosure requirements assume that contributors will take advantage of the now public information; difficult to believe given the number of charities that have recently come under scrutiny for misuse of donated funds. Authors such as Britton (2007) argue that disclosure requirements could be made to work if government granting agencies were able to withhold grants from nonprofits that fail to provide “enhanced” financial data. Some have suggested that a Sarbanes-Oxley type of regulation (Mead, 2008) might be appropriate for the nonprofit sector. Reiser (2004) argues that even that would be insufficient, given that Sarbanes-Oxley is focused almost exclusively on financials. Mulligan (2007), however, argues that such a mandate is inappropriate to charitable organizations and will impose high costs without providing the sought-after improvement in accountability. Aprill (2007) questions the legal foundations of the law, and argues that applying it to the nonprofit sector would entail a significant (and unwarranted) expansion of federal authority over an area that is now regulated by the states. Neely (2011) examines the impact of Sarbanes-Oxley–style enforcement at the state level, detailing California’s Nonprofit Integrity Act of 2004. Ebrahim (2011) notes that various forms of nonprofit accountability exist, and none of them are appropriate for all organizations.

Nongovernmental Oversight

The most obvious form of oversight at charities is the presence of an independent voting board (IVB) that provides general guidance on the scope and aims of the organization and monitors the entity’s financial integrity (see Miller, 2002, 2007 for a critical evaluation). In the context of nonprofit management, an IVB refers to individuals who did not receive compensation from the nonprofit as an officer or employee and did not conduct any transaction with the organization that must be reported on IRS Schedule L, Transactions with Interested Persons. Hodge and Piccolo (2011) provide details on effective board management and the financial stability of a nonprofit. Yetman and Yetman (2012) examine how the level of effectiveness of a nonprofit’s board affects the accuracy of its expense reporting. Boozang (2007) argues that nonprofit boards may be excessively concerned with compliance and lose focus on the effectiveness of a charity’s activities. Conducting an independent audit provides another level of accountability, particularly if the results of the audit are included on the nonprofits’ 990 form. For many small nonprofits, the cost of an independent audit may be prohibitive. Prakash and Gugerty (2010) argue that nonprofits that voluntarily agree to additional regulatory controls can utilize their participation in such programs as a means to differentiate themselves from organizations that eschew added oversight. Finally, Williams and Taylor (2013) argue for a holistic form of accountability that takes into account the numerous stakeholders that have an interest in effective leadership at nonprofits.

In addition to this form of “internal” control, third party rating services also provide information to potential donors on the integrity of individual charities. Acting as a complement to governmental regulation, Charity Navigator and CharityWatch provide evaluations of charitable organizations that go beyond what is officially mandated. Nonprofits are ranked according to both transparency and financials. Charity Navigator downgrades an organization’s accountability and transparency score if it does not meet disclosure and administrative requirements (e.g., utilizing an independent board). Likewise, a poor financial score will be given to those charities that perform poorly on seven financial criteria, including having excessive fundraising and management expenses. Currently, Charity Navigator ranks approximately 8,200 nonprofits.

Although these rating bodies provide another level of oversight, it is still up to donors to make the effort to investigate before giving. As argued earlier, the lack of a feedback loop that informs donors about how (and how well) their donations are utilized is missing. Thus, the charitable sector remains a “donor beware” environment.

COMPARISONS OF HARD AND SOFT CORRUPTION

The term “hard corruption” will be used to refer to organizations that have committed outright fraud that ultimately led to a criminal prosecution. It is widely recognized that not-for-profits might “adjust” their Form 990 s so as to enhance their attractiveness to potential donors (note that Froelich, Knoepfle, and Pollak (2000) found no systematic manipulation of figures on Form 990 s for those nonprofits that conduct an independent audit). Yet, hard corruption involves the deliberate misuse, for personal gain, of donations intended for a philanthropic cause. The illustrations below are meant to differentiate this form of criminal behavior from cases of “soft” corruption.

Not all instances of hard corruption at nonprofit organizations are widely publicized. CharityWatch, however, provides a list of the most egregious examples of illegal activity on the part of nonprofits (CharityWatch Hall of Shame [https://www.charitywatch.org/charitywatch-articles/charitywatch-hall-of-shame/63]). The scandal involving William Aramony, who led United Way from 1970 to 1992, is perhaps the most widely known example of malfeasance. Aramony was charged with misappropriation of donated funds and filing false tax returns. He was also found to have siphoned off funds to acquire apartments in New York and Miami, and to have utilized United Way funds to pay for multiple trips to Las Vegas. Found guilty, Aramony was sentenced to five years in federal prison. The scandal forced United Way to institute new accounting rules and improve its oversight procedures. In 2009, the Federal Bureau of Investigation (acting with the IRS) raided Angel Food Ministries of Monroe, Georgia. Later (2011) a 49-count federal indictment was filed against the founders of the charity (Linda and Wesley Wingo, as well as their son Andrew). Angel Food Ministries, using a unique model, purchased groceries at wholesale and resold the food to families in need at about one-half the normal retail price. Founded in 1994, by 2008 the organization was providing assistance to hundreds of thousands of individuals. The indictment charged the Wingos both with siphoning off donated funds and with paying (criminally) excessive salaries. Ultimately, both Wesley and Andrew Wingo received seven-year prison terms (Linda Wingo received a suspended five-year sentence).

The charitable sector is capable of innovation when it comes to crafting new corrupt practices. The Foundation for New Era Philanthropy, founded by John Bennett, Jr., promised nonprofits a simple means of doubling their contributions. The charity deposited a large sum of money with New Era—after a period of time the money was matched by an anonymous donor with similar interests and returned. In reality, New Era was simply a Ponzi scheme, with new deposits being used to match old deposits (for a description of the structure of Ponzi Schemes, see Jory & Perry, 2011). When Bennett could no longer cover the inevitable shortfalls, he turned to borrowing money, eventually accumulating $50 million in unpayable loans. After being exposed in the press in 1995, Bennett was charged with 82 counts of bank/wire fraud and money laundering and sentenced to 12 years in prison in 1998. At the time of his indictment, he finally confessed that no anonymous donors ever existed.

The United States Navy Veterans Association (USNVA) was a wholly corrupt organization started and managed by John Donald Cody (aka “Bobby Thompson”). The USNVA collected nearly $100 million over a ten-year period ostensibly to assist veterans. Although the organization claimed over 60,000 members, it actually was run entirely by Cody out of a duplex in Florida, and no evidence exists that the organization ever assisted any veterans. In 2010, questions were raised about the “charity” and Cody fled the state to avoid prosecution. He was captured in Oregon by federal marshals in 2013 and eventually convicted of 23 counts of theft, money laundering, and record tampering. At the conclusion of his trial, Cody was sentenced to 28 years in prison.

Other major instances of alleged corruption occurred at Covenant House (1989), Feed the Children (2009), Central Asia Institute (2010), Hale House (2002), and Help Hospitalized Veterans (2009). Unlike the examples of malfeasance detailed above, these latter scandals led to civil, rather than criminal, prosecutions.

Examples of soft corruption in the nonprofit sector abound. As noted, the Tampa Bay Times publishes an annual summary of organizations gone bad entitled America’s Worst Charities. Of the 48 organizations analyzed by the Times, 44 (nearly 92%) give less than 5% of donations to a cause in the form of direct cash grants. Solicitation costs are frequently in the 80–90% range, indicating that these organizations are simply “mills” for generating contributions, with little or no interest in pursuing their supposed philanthropic interest. Causes “supported” by these charities are predictable—focusing on those that will elicit a strong response from potential donors: Firefighters and police (14), treating or preventing cancer (10), children’s causes (8), and support for veterans (5). The prevalence of these organizations in the philanthropic sector reflects the ease by which sham nonprofits can be formed and successfully operated within the existing legal framework. It then falls upon donors to seek out further information on charities before giving. It should be noted that New York’s Attorney General issues a generalized report (“Pennies for Charities”) that describes charitable solicitation practices that result in a high proportion of donations to fundraising, rather than the identified cause of nonprofits.

As noted above, Charity Navigator (CN) also provides assessments of nonprofits. Evaluations of each charity include a financial, transparency and overall ranking. Poor performance in one criteria (e.g., financials) can be somewhat offset by better performance in the other (transparency), and charities receive an overall ranking (0–4 stars) that reflects CN’s assessment. Nonprofits receiving “0 star” rankings generally represent entities with both poor transparency and fundraising or compensation costs that consume most of what is donated to the organization. The effectiveness of charity oversight by organizations such as CharityWatch and Charity Navigator has been examined by authors such as Cnaan, Jones, Dickin, and Salomon (2011), who question the usefulness of ratings services when they are used by few donors. Additionally, Rao (1998) argues that political leanings may tamper with the accuracy of consumer ratings of organizations in general.

MODEL

The model developed here most closely mirrors the work of Jones (1991) and Trussel (2003). Trussel used a set of financial variables (e.g., revenue growth) to predict which charities were exaggerating their program expense ratios. In particular, the author utilizes the error terms from a predictive (regression) model to uncover those nonprofits that have reported financials that differ significantly from a predicted value. Notably, the author was not concluding that these nonprofits were overstating their expenditures on programs, but rather their financial ratios looked like charities that do. Conversely, the work presented here will label as “soft corrupt” those organizations that achieve a zero-star financial rating from Charity Navigator—a score that represents poor performance on seven financial indicators. Trussel’s approach eliminates the need to accept self-reported financials. The only limitation to his methodology is there is no means to differentiate between those nonprofits that are actually misrepresenting their financials from those that simply appear to be.

A test of the determinants of soft corruption will be presented here. An empirical model using cross-sectional data on 450 charitable organizations (drawn from the over 8,000 nonprofits evaluated by Charity Navigator using a random number generator) will examine the relative importance of the presumed underpinnings of malfeasance. All charities that receive a zero-star ranking for financials will be included in the sample. The financial ranking is comprised of seven specific criteria that measure both whether donations are put to good use (e.g., Fundraising Expense Ratio) and the viability of the organization (Working Capital Ratio). Additional factors include administrative expenses (percentage), program expenses (percentage), fundraising efficiency, program expense growth, and liabilities to asset ratio.

Although there might be some disagreement about how one would define “soft” corruption, those charities receiving a zero-star financial ranking donate little or nothing to their prescribed causes. Neither the overall ranking nor the accountability and transparency ratings are appropriate, as this would create a problem with bi-causality, as some of the factors that will be used as explanatory variables (e.g., whether the organization conducts an independent audit) partially establish both a charity’s transparency score, and thus its overall ranking.

Although Charity Navigator provides, by far, the most comprehensive database of charitable organizations, some limitations should be noted. It does not rank hospitals, schools, and many smaller charities, a restriction that limits the universality of the findings presented below. In addition, the empirical model developed in this section utilizes categories (e.g., human rights) established by CN that may or may not fully reflect the work done by a particular nonprofit. This may affect the outcomes reported by either reducing the likelihood that category-specific binaries test as significant or exaggerating the importance of an organization’s focus.

A significant part of CN’s financial rating is based upon high fundraising ratios (low program expenses). It could be argued that, particularly for small, less well-known charities, high fundraising outlays are a necessary part of establishing the reputation of a new organization. It will not be argued here that “excessive” fundraising expenses are the only indicator of a poorly run organization, but simply part of what is utilized to evaluate performance. Perhaps the most significant drawback to using CN’s designation is the “black box” nature of the rankings. The seven criteria that are utilized to establish a financial ranking are available, and a diligent individual could extract a ranking from those measures, but the precise rubric used to establish a rating is still somewhat of an unknown (see https://www.charitynavigator.org/index.cfm?bay=content.view&cpid=48 for a description of the procedures used by Charity Navigator).

An additional limitation of the Charity Navigator dataset is that much of the data is self-reported, suggesting that charities that are poor performers might exaggerate financials (e.g., program expenses/total expenses) to improve their ranking within CN’s rubric. This is an inherent imperfection in the procedure used to label zero-star charities. This would suggest, however, that these organizations are even worse than reported, still placing them in the category of soft corruption. In addition, CN utilized seven criteria to establish a financial ranking, not just the figure spent on programs. The concern here, in the end, might be with one-star charities that have misreported their financials, and should have been dropped into the zero-star category, a fruitful topic for future research.

Previous work on nonprofit corruption has focused on hard corruption; those organizations where civil or criminal wrongdoing was committed and announced in the press. Unfortunately, this is an unknown when one examines a database of charitable organizations since, given the number of instances of corruption revealed each year, it is certain that additional cases of fraud are present but simply have not yet been exposed. As argued earlier, nonprofits that receive a zero-star financial ranking (soft malfeasance) from Charity Navigator can be, in most instances, regarded as corrupt in nearly every manner, except being in legal peril. In total, there are 53 charities in this category (see Table 1 for a breakdown by category). As is apparent, the prevalence of zero-star charities varies significantly by cause. Categories such as Animals, Arts, Culture and Humanities, Education, Environment, and International contain relatively few nonprofits with Charity Navigator’s lowest financial ranking. Conversely, it is far more common for charitable organizations that fall under the classifications of Community Development, Health, Human Services, Human and Civil Rights, and Religion to receive a zero-star rating. In fact, for Human/Civil Rights, over 3% of all nonprofits in this category earned the lowest financial ranking. For charities in the Health subsector, 1.5% of all organizations are zero-ranked. This variation across categories will be incorporated into the estimation.

TABLE 1 Zero-Star (financial) Charities, Number and as a Proportion of Total by Category

Categories with 0-Star charities (financial score)*
Category	Total in category	0-Star/proportion of total
Animals	507	0/0.00
Arts, culture, humanities	1111	2/0.0018
Community development	811	3/0.0037
Education	541	2/0.0037
Environment	417	1/0.0024
Health	921	14/0.0152
Human services	2329	17/0.0073
Human and civil rights	227	7/0.0309
International	588	2/0.0034
Religion	469	3/0.0064
Research and public policy	196	2/0.0102
Total		8117

Source: Charity Navigator, https://www.charitynavigator.org/.

*As of December 2015.

Variables

Corruption in the nonprofit sector can be partly attributed to agency problems. Smaller donors expect neither a return nor donor-specific accountability, enabling charitable organizations to misuse funds without immediate consequences. Hence, one of the primary determinants of potential malfeasance is the degree of oversight under which the organization operates, and whether the details of that oversight are released to the public. Those organizations that maintain an independent voting board and release the names of those board members (indboard_i) are less likely to violate donors’ intentions (consistent with cited work by Boozang, 2007; Hodge & Piccolo, 2011; Miller, 2002). Nonprofits that accept government grants (govt_i) assume an additional level of oversight, and one that carries potential legal peril for misconduct. Likewise, charitable organizations that conduct and release the results of an independent audit (audit_i) are, in most instances, voluntarily imposing an added external control (follows work by Britton, 2007; Froelich, et al., 2000 on disclosure). State-specific disclosure requirements (state_i) are presumed to deter corruption, with the probability of malfeasance higher in those states with lax reporting obligations, addressing the work by Neely (2011) and Lott et al. (2016). (For a comprehensive look at state regulations affecting charities, see State Regulation and Enforcement in the Charitable Sector [https://www.urban.org/sites/default/files/publication/84161/2000925-State-Regulation-and-Enforcement-in-the-Charitable-Sector.pdf], published by the Center on Nonprofits and Philanthropy). A binary variable will be used to distinguish between nonprofits operating in states that have reporting requirements and those operating in states that have no such mandate. No attempt was made to incorporate the differing ceilings that are used across states, as this would lead to difficulty in interpreting this variable (a potential area for future empirical work). The potential empirical relationships outlined here also reflect the more general work of Ebrahim (2011) and Steinberg (2010).

One of the potential “truisms” of misbehavior in the nonprofit sector is the idea that it is confined to smaller, less visible, organizations. The annual expenditures (expend_i), labeled “total functional expenses” in the vernacular of Charity Navigator, will be used as a measure of size. One might also assume that this form of malfeasance is more likely in upstart, rather than established, charities. Hence, the age (age_i) of the institution will be incorporated into the sample. Finally, it is possible that malfeasance is more common in certain subsectors (different giving categories) of the nonprofit universe. Binary variables (categ_{i, j}) will be used to determine if significant variation is present in the sample across categories of giving. As utilization of binary variables is confined to n − 1 cases (to avoid overspecification), preliminary testing was employed to determine within which subsectors malfeasance was more likely. Binary variables were included for “Animals,” “Health,” “Human Services,” and “Religion.”

Prior work on nonprofit corruption (e.g., Greenlee et al., 2007) has focused on the magnitude of corruption and its determinants, attributing the scale of malfeasance to the compensation, tenure, and gender of perpetrators. It is unclear whether these influences would impact the decision to engage in soft corruption. In addition, as this form of misbehavior is frequently characterized by excess salaries, bi-causality is an issue. Hence, these additional factors will be omitted from the estimation. As noted above, Trussel (2003) used accounting variables to predict which nonprofits were likely to be misrepresenting their finances. In this estimation, however, the zero-ranked charities achieved that distinction as a result of financial variables, Utilization in an estimation would make the explanatory variables a function of the dependent variable (bicausality), For an additional evaluation of which numbers matter in evaluating the strength/effectiveness of a charity, see Ritchie and Kolodinsky (2003).

In summary, the probability that a particular nonprofit receives a zero financial ranking (rank_i) from Charity Navigator will be established as a function of oversight (indboard_i, govt_i, audit_i), location (state_i), size (expend_i), category of giving (categ_i,j). A binary variable will be used to represent the “state,” with 1 representing charities that operate in states that require the release of audited financials. Ideally, the magnitude of operating expenses that leads to a required audit would be used but, for approximately half the sample, no audit is required. Overall:

(1) $$\begin{array}{rl}& {\mathrm{r}\mathrm{a}\mathrm{n}\mathrm{k}}_{\mathrm{i}=}{\mathrm{F}(\mathrm{a}\mathrm{g}\mathrm{e}}_{\mathrm{i}}{,\mathrm{i}\mathrm{n}\mathrm{d}\mathrm{b}\mathrm{o}\mathrm{a}\mathrm{r}\mathrm{d}}_{\mathrm{i}}{,\mathrm{a}\mathrm{u}\mathrm{d}\mathrm{i}\mathrm{t}\mathrm{i},\mathrm{g}\mathrm{o}\mathrm{v}\mathrm{t}\mathrm{i},\mathrm{s}\mathrm{t}\mathrm{a}\mathrm{t}\mathrm{e}}_{\mathrm{i}}{,\mathrm{e}\mathrm{x}\mathrm{p}\mathrm{e}\mathrm{n}\mathrm{d}}_{\mathrm{i}}{,\mathrm{c}\mathrm{a}\mathrm{t}\mathrm{e}\mathrm{g}}_{\mathrm{i}}{,}_{\mathrm{j},\mathrm{j}=1\mathrm{t}\mathrm{o}5})\\ & (-)(-)(-)(-)(-)(+or-)\end{array}$$(1)

• where: rank_i= 1 for 0-star (financial) charities, 0 otherwise

• age_i = Age of the nonprofit

• indboard_i = 1 for charities with named independent boards, 0 otherwise

• audit_i = 1 for organizations that release an annual audit, 0 otherwise

• govt_i = 1 for nonprofits with partial government funding, 0 otherwise

• state_i = Reporting requirement, 0 for none, 1 for legal requirement

• expend_i = Annual level of giving

• categ_{i, j} = Category of giving; j = 1, 5

As Equation (1) utilizes a binary dependent variable, it will be estimated using both logit and probit models, although it is anticipated that there will be little difference in the results (logit models provide elasticities, which might be of interest to some researchers). Missing data, particularly on the age of the nonprofit, eliminated 40 observations, resulting in a total of 410 observations in the final estimations.

Results

The results are provided in Tables 2 and 3. Both the probit and logit models produced approximately the same degree of explanatory power (pseudo-R² of approximately 38%). The most important explanatory variables in each instance (significant at the 1% level in both cases) were the presence of a transparent independent voting board and the publication of the outcomes of an independent audit. The expected negative signs indicate a measurable drop in the probability that a nonprofit will engage in soft corruption when oversight of this form is in place. These outcomes affirm the importance of external oversight. Conversely, other variables incorporated into the model to represent outside supervision were rejected in both estimations. The presence of government grants, for instance, was rejected in both the probit and logit models (z-scores of −0.53 and −0.37, respectively). This is not entirely surprising, as grants represent a different (and apparently less effective) form of oversight. It would not be difficult for a nonprofit to properly manage and report on the use of external funding, given the consequences of not doing so, and at the same time mismanage donations where no such accountability exists. The other form of oversight, state reporting requirements widely regarded as an important check on misbehavior by nonprofits, does not test as significant (rejected by the model, with a z-score of 0.27 in both models). The two variables that were incorporated to reflect the “stature” of the organizations, age, and magnitude of expenditures, were also rejected in both instances. This is a somewhat surprising result, as seasoned charitable organizations would be expected to have developed a reputation and would not want to alienate long-time donors.

TABLE 2 Results of Estimation, Probit Regression

n = 410
Pseudo R^2 = 0.3805	LR χ^2 = 120.13**
Variable	Coefficient	Standard Error
Intercept	0.9340	0.3528**
Age	−0.0037	0.0041
Indboard	−1.525	0.2754**
Gov’t	−0.1406	0.2628
Audit	−1.061	0.2363**
State	0.0609	0.2283
Expend	−3.28E^−6	6.31E^−6
categ1 (health)	0.5556	0.2682*
categ2 (human serv.)	1.1427	0.3889**
categ3 (animals.)	−0.8259	0.5381
categ4 (religion)	−0.004	0.5486

Notes: *Significant at the 5% level; **Significant at the 1% level.

TABLE 3 Results of Estimation, Logit Regression

n = 410
Pseudo-R^2 = 0.3774	LR χ^2 = 119.15**
Variable	Coefficient	Standard Error
Intercept	1.6672	0.6775*
Age	−0.0065	0.0078
Indboard	−2.711	0.5030**
Gov’t	−0.2021	0.5407
Audit	−1.964	0.4371**
State	0.1202	0.4433
Expend	−5.841E^−6	1.24E^−5
categ1 (health)	1.1605	0.4982*
categ2 (human serv.)	2.1182	0.7123**
categ3 (animals)	−1.418	0.9292
categ4 (religion)	0.1856	1.052

Notes: *Significant at the 5% level; **Significant at the 1% level.

Two of the four binary variables included in the model were significant: Those representing Health and Human Services causes. In both the probit and logit estimations, charities that pursued heath causes were significantly more likely (at the 5% level) to receive a zero ranking. Similarly, in both estimations there was a greater likelihood that human services organizations would be zero-ranked (significant at the 1% level). Conversely, there was no difference between nonprofits in the Research and Religion categories and charities in general. In the latter case, this is partly explained by a tendency of religious organizations not to utilize an independent voting board (already a highly significant variable) washing out any measurable impact.

CONCLUSION

Trust in organizations, whether corporate, governmental or nonprofits, is fragile. Scandals in the corporate world can drive consumers away from a particular firm, but seldom do buyers boycott an entire sector of the economy, such an action being neither practical nor particularly rational. Governmental corruption results in promises of reform but, once again, there are few alternatives for those that would consider avoiding a particular agency (the unending scandals at the Veterans Administration being an appropriate example). Once trust is broken in the charitable sector, however, the damage can be permanent and systemic. A contributor to a nonprofit who discovers that a donation was misused or appropriated is not only unlikely to donate to that organization again, but may foreswear charitable contributions altogether (see Light, 2008, p. 2). As the philanthropic sector plays a positive and significant role in a wide range of social programs in the U.S. economy, the damage from scandals in the nonprofit sector may be considerable.

The empirical work presented here indicates that two forms of oversight—maintaining an independent voting board and contracting for an independent audit—are the best means of avoiding organizational malfeasance at nonprofits. Governmental oversight, whether through reporting requirements or through the oversight inherent in the federal grant process, does not test as significant. This indicates that most concerns could be addressed by requiring both an independent board and an annual audit for all nonprofits with 501(c)(3) status. This is not to say this requirement would be a panacea for the problem of soft corruption—Aramony managed to manipulate the United Way Board during the 1992 scandal at the organization—but simply that this form of malfeasance would be less common with this form of oversight in place.

This article has also highlighted the unique circumstances that surround nonprofits, particularly when compared to the corporate sector. Most donors receive nothing—either informational or monetary—from organizations to which they donate, resulting in an agency problem. This enables managers of charities to misdirect, misuse, or pocket (through excess salaries) donations with few consequences. Independent audits and the presence of an independent board would reduce the number of instances of corruption arising from this agency problem.