1 Introduction
The nonprofit sector is a vital part of the U.S. economy. In 2016, nonprofit organizations (NPOs) contributed an estimated $1.047 trillion to the economy, comprising 5.6 percent of the country’s gross domestic product, and, in 2018, total private giving from individuals, foundations, and businesses totaled $427.71 billion (Urban Institute).1 It is well documented that funders, boards of directors, and regulators use financial information when making giving, pay, and enforcement decisions (e.g., Parsons 2003) and that these stakeholders have a particular interest in how NPOs spend their money. The only financial report all NPOs are required to file and make public (and is easily accessed through GuideStar, a nonprofit data consolidator) is Form 990, Return of Organization Exempt from Income Tax. On this Form, the Internal Revenue Service (IRS) requires NPOs to report their expenses in a matrix format on the Statement of Functional Expenses. These expenses are classified into over 20 natural expense classifications.2 Example natural expense classifications include salaries and wages, accounting fees, and travel expenses. NECs are then allocated among the program, fundraising, and administrative functional expense categories.
The most common financial metric used to evaluate NPOs is the program ratio – program expense to total expense – which measures how efficiently NPOs maximize programmatic spending and minimize administrative and fundraising spending.3 Given the emphasis that donors, for example, place on the program ratio (hereafter also referred to as the ratio) (e.g., Parsons 2003), it is well documented that some nonprofits intentionally misreport expenses among the functional expense categories. (See Ling and Roberts 2023 for a literature review.) As such, auditors have issued qualified opinions when they cannot verify the rationale behind the allocation of expenses.4
Advertisement
Although evidence suggests some NPOs manipulate reported ratios, a considerable gap exists in the academy’s understanding of how to identify NPOs that have potentially managed their ratios and how they accomplish this.5 The objectives of this study are two. The first is to identify a red flag stakeholders can use to detect nonprofits that have potentially managed cost allocations to influence how ratios are reported. The second is to identify cost allocation techniques and the expenses likely used to manipulate cost allocations. The red flag we identify is the reporting of an identical program ratio—that is, the NPO reports the same ratio in multiple years while reporting a large change in total spending. We focus on large changes in total spending—i.e., at least 5 percent—because large changes presumably reflect a change in operations,meaning identical ratios do not reflect stable operations.6 Note that we do not suggest that, when changes in total spending are relatively small and when nonprofits report identical ratios, that this does not reflect ratio management. But we acknowledge that because small changes in spending may reflect stable operations, it is empirically difficult to determine whether identical ratios are the outcome of stable operations or ratio management.
We study 4,063 relatively large nonprofit organizations (17,451 NPO-years) between 2003—2015. We study the NPOs from the Internal Revenue Services’ Statistics of Income (SOI) database that report at least a five-percent change in total spending. The average NPO in our sample changes total spending by $11.9 million or 15 percent (the median is $2.7 million or 9 percent) within a year.
To determine whether identical ratios reflect ratio management, we hypothesize and find that NPOs are likely to report identical ratios when incentives to manipulate ratios are strong. Specifically, after controlling for organizational sophistication, complexity of operations, and prior period ratios, we find NPOs are more likely to report identical ratios when resource providers – donors or government grantors – use program ratios in their giving decisions; executive pay is affected by changes in the program ratio; and when the NPO is domiciled in a state where nonprofit regulation is relatively strong. Our results are robust to several ways of computing ratio changes, alternatives for a large change in total spending, and the number of times an NPO reports an identical ratio. We also find associations between incentives and the likelihood NPOs report identical ratios can depend on industry affiliation, whether the NPO wants to avoid reporting a decrease in their ratio (versus an increase), and whether managers can easily justify reasons for ratios to change.
We find NPOs use one of three cost allocation techniques to manage ratios (i.e., the study’s second objective). NPOs either: 1) keep program allocation rates the same as the prior period for all natural expense classifications 2) manipulate program allocation rates for one such classification, or 3) manipulate program allocation rates for multiple classifications. We find the most popular technique in our sample is to manipulate program allocation rates for multiple natural expense classifications, and most nonprofits use the same allocation technique over time. Those that manipulate multiple classifications and those that use multiple allocation techniques are more sophisticated (in terms of size). Research concludes that sophisticated nonprofits are less likely to manage ratios (Krishnan et al. 2006; Keating et al. 2008). Yet our finding suggests that, once a nonprofit decides to manage ratios, its sophistication dictates how this is achieved.
Advertisement
We find that the natural expense classifications likely manipulated depend on the allocation technique used.7 However, regardless of the allocation technique, managers use natural expense classifications where the degree of flexibility over how much to allocate to programs is relatively high. Nonprofits that manipulate only one classification mostly use the other expense line to manage reported ratios. This provides them with a good opportunity to manipulate reporting because other expenses include various expenses, making evaluating the appropriateness of allocations difficult. Nonprofits that change multiple natural expense classifications mostly manipulate lines associated with salaries and wages (such as employee benefits), office expenses (such as supplies, printing, telephone and postage, and shipping), and travel. Research documenting managers use a high degree of discretion when deciding how to allocate these expenses supports our findings (Wing and Hager 2004b; Jones and Roberts 2006; Ling and Neely 2013). Accounting and legal expenses are two natural expense classifications least likely manipulated, probably because the instructions to Form 990 require nonprofits to allocate most of these costs to the administrative category.
We acknowledge the complexity inherent in assessing changes in financial ratios within the context of NPOs. Specifically, if NPOs adhere to a policy of maintaining the same allocation percentages for all expenses without annual adjustments, this rigidity may lead to unchanged ratios over time. Additionally, nonprofits can either omit entire natural expense classifications (Burks 2015) or underreport such expenses. In essence, the measurement of ratio changes in NPOs is a multifaceted endeavor, and thus we acknowledge that there are alternative reasons contributing to NPOs reporting identical ratios and alternative methods they may employ to report identical ratios, extending beyond the scope of this study.
Nevertheless, our study is important for several reasons. First, with one exception, despite the emphasis placed on ratios, there are no red flags to help stakeholders detect when ratios are potentially managed.8 This differs from the many red flags available to help identify publicly traded firms that have managed earnings (Dichev et al. 2013).9 We contribute to the nonprofit sector by identifying the reporting of identical ratios (over multiple years and when changes in spending are relatively large) as a red flag of ratio management.
Second, the academy has yet to determine how the average NPO manipulates cost allocations to manage how ratios are reported. We close this gap by identifying cost allocation techniques and the expense lines NPOs most likely use.
Finally, beginning with fiscal years after December 15, 2017, GAAP requires all NPOs to use Form 990 style functional expense reporting in their financial statements (ASC 958–720-45). The FASB stated that this requirement makes information about expenses more comparable and useful to “help donors, creditors, and others in assessing an NFP’s [nonprofit’s] service efforts, including the costs of its services and how it uses resources.” Because the GAAP standard focuses on cost allocations, it has refueled a long-standing debate on the reliance of using ratios to measure nonprofit effectiveness and efficiency (Klotz 2019). As a result, the accounting standard itself may heighten incentives to manipulate ratios. Hence, our findings benefit all auditors who must attest to the appropriateness of how natural expense classifications are allocated among the functional expense categories.
The remainder of this paper is organized as follows. The next section describes why we believe an identical program ratio signals ratio management. Section three develops the hypotheses. Section four discusses the research design, and Section five describes the data. Section six discusses the results and additional analyses. Section seven examines techniques and expenses nonprofits use to manage ratios, and the last section concludes.
2 Identical ratios, a signal of ratio management
We observe nonprofits reporting program expenses such that program ratios do not change in 23 percent of cases.10 This percentage is significant, given ratios are expected to change when organizations grow, contract, pursue new activities, or face changes in input prices. The expectation that program ratios change resembles the expectation analysts (of for-profit firms) have when evaluating ratios, such as gross margins. As it relates to nonprofits, we anticipate, for three broad reasons, that, when total spending changes, ratios should change as well.
First, economists advise nonprofit managers to think at the margin—in other words, to ignore ratios to make socially optimal choices when allocating resources (Steinberg 1986; Young and Steinberg 1995). For example, if optimal, NPOs should use additional resources to give highly effective executives bonuses so they do not seek other jobs. If executive pay is overhead, the program ratio will decrease.
Second, if changes in spending reflect changes in activity levels and if any component of program or overhead costs is fixed, assuming activity levels are within the relevant (i.e., a normal) range of activity, basic managerial accounting concepts suggest the proportion of spending on programs and overhead will change as well. For example, if program spending for a meals-on-wheels charity is mostly fixed—because the geographical area and the number of clients the charity serves do not vary much—the NPO may use additional resources to increase executive pay. If executive pay is overhead, then the program ratio will decrease. If overhead costs are mostly fixed, additional resources are used on programs, increasing the program ratio (Kitching et al. 2012). In the case of spending cuts, contracting constraints, or other adjustment costs inhibit cutting budgets to programs and overhead at exactly the same rate in the short term (Jones et al. 2013); thus ratios will change.
Lastly, ratios should change when changes in total spending are due to changes in input prices. For example, if food costs for the meals-on-wheels charity increase during the period, the program ratio will increase. The top half of Appendix Table 11 illustrates these general points.
We posit that, when an NPO reports a large change in total spending and simultaneously reports an identical program ratio, it has potentially managed its cost allocations to achieve this outcome. We first show that identical ratios are a credible ratio management signal and then demonstrate techniques and expenses nonprofits use to report identical ratios.
3 Hypotheses development
We consider incentives to avoid reporting changes in the program ratio to develop our hypotheses. We acknowledge that, for some incentives, we hypothesize that managers’ desire to avoid reporting program ratio decreases is likely greater than their desire to avoid reporting program ratio increases. (We test this later.) However, we presume that, with each incentive, nonprofits have motives to avoid reporting ratio increases and decreases. Hence we focus on ratio changes with the incentives we examine.
3.1 Resource incentive
Resource dependency theory (Pfeffer and Salancik 1978) posits that funding sources are associated with different levels of engagement. Managers, recognizing this, respond to each key resource provider’s demand to obtain funding. We hypothesize two key providers—donors and government grantors—influence decisions to report identical ratios. We predict that, when resource providers use ratios in their giving decisions, NPOs depending on these resources, are more likely to report ratios in ways to influence these decisions positively.
A wide body of research suggests donors use program ratios in their giving decisions. Specifically, research finds donations are positively associated with program ratios (e.g., Weisbrod and Dominguez 1986; Harvey and McCrohan 1988; Posnett and Sandler 1989; Callen 1994; Tinkelman 1999; Greenlee and Brown 1999; Okten and Weisbrod 2000; Tinkelman and Mankaney 2007; Jacobs and Marudas 2009; and Kitching 2009). As such, there is evidence consistent with the notion that NPOs manage how costs are allocated or spending decisions to report high program ratios or to keep program ratios from declining (Trussel 2003; Krishnan et al. 2006; Tinkelman 2009; Kitching et al. 2012; Ling and Neely 2013; Parsons et al. 2017). Jones and Roberts (2006) argue (but do not test) that NPOs may wish to avoid reporting program ratio increases (in addition to decreases) to avoid raising donors’ expectations concerning future program spending. Also, there is mixed evidence that donors respond to changes in ratios (Bowman 2006).11 Accordingly, our first hypothesis is:
H1a: Nonprofits with donors more sensitive to changes in the program ratio are more likely to report identical program ratios.
The U.S. federal government is concerned about how much nonprofits spend on overhead. For this reason, it has detailed rules about how its grantees spend federal grants. In particular, it restricts the percentage of funds grantees can use for administrative purposes and does not permit the use of the funds for fundraising (OMB Circular A-122, Cost Principles for Non-profit Organizations). Research supports the idea that nonprofits receiving government funding try to avoid reporting decreases in the program ratio (Kitching et al. 2012).
The percentage of funds grantees can use for administration—the indirect cost rate—is negotiated between the NPO and the federal granting agency (OMB Circular A-122); thus indirect cost rates are inconsistent across all grant recipients. We speculate that, because grant recipients negotiate indirect cost rates, managers must consider how current performance will impact future negotiations of indirect cost rates. As such, they have an incentive to avoid reporting increases in the program ratio. For example, suppose that the contracted indirect cost rate in the current grant cycle is 15 percent and the NPO reports an increase in the program ratio. This might suggest to the federal granting agency that the recipient can meet its administrative needs using less than 15 percent of the grant.12 Given that nonprofits are continuously challenged to find ways to keep administrative spending low (Wing and Hager 2004a; Gregory and Howard 2009), they would be particularly reluctant to report ratios that would make federal granting agencies aware of their reservation price and ratchet up the starting point for future negotiations (Thompson 2000), thus lowering future contracted indirect cost rates. Taken together, we expect NPOs with government grants to have incentives to minimize program ratio changes. Thus, we hypothesize:
H1b: Nonprofits that receive grant money from the federal government are more likely to report identical program ratios.
3.2 Pay incentive
Research finds changes in executive pay are positively associated with changes in the program ratio (Baber et al. 2002; Aggarwal et al. 2012). Moreover, Krishnan et al. (2006) find that the more sensitive executive pay is to changes in the program ratio, the more likely nonprofits are to report zero or underreport fundraising expenses so that reported program ratios are high. These findings suggest NPOs are more likely to avoid reporting decreases in program ratios when executive pay is sensitive to changes in ratios.
There are at least two reasons executives, whose pay depends on reported program ratios, may avoid reporting increases.13 The first is when an increase in the program ratio in the current period is small (and may not have even reached the threshold to impact pay positively) and management anticipates a large increase in the program ratio in the next period. Thus, to maximize pay, managers wait until the next period to report the program ratio increase. This resembles Healy’s (1985) argument and finding that for-profit managers whose pay depends on earnings have incentives to reduce current earnings if target earnings will not be met, thus increasing the probability of meeting future earnings’ targets. Another reason is if managers expect compensation contracts to change in the near future and they would benefit from “saving” the program ratio increase for later. This idea resembles the situation of for-profit executives who miss earnings targets via downward earnings management when large subsequent stock-option grants are expected (McAnally et al. 2008).14 Given these considerations, we hypothesize:
H2: Nonprofits with executive pay more sensitive to changes in the program ratio are more likely to report identical program ratios.
3.3 Regulatory incentive
Jones and Roberts (2006) suggest nonprofits might avoid reporting changes in program ratios to avoid regulatory intervention. They suggest (and get confirmation from a state regulator) that regulators are more likely to question a nonprofit’s operating and financial reporting decisions when it reports a change, including an increase, in its program ratio. Thus, given the costs associated with regulatory intervention, NPOs aim to avoid reporting ratio changes.
We acknowledge that research finds regulation diminishes the occurrence of ratio management; specifically, it reduces the likelihood nonprofits report zero or underreport fundraising expenses (Yetman and Yetman 2012). We suspect that, although regulators prevent NPOs from reporting zero (or underreporting) fundraising expenses, they can do so because underreporting fundraising expense is known to be a red flag of ratio management (Tinkelman 1999; Froelich et al. 2000; Wing and Hager 2004b; Krishnan et al. 2006; Yetman and Yetman 2013). However, regulators may not be aware of identical ratios yet; thus managers arguably have a greater opportunity to manage perceptions of performance by reporting identical ratios.15 In fact, research (in the for-profit domain) finds that regulation does not fully deter earnings management but impacts how firms manage earnings (Evans et al. 2014). Because states are the locus of legal authority over nonprofits and vary in their degree of regulation (Desai and Yetman 2015), we hypothesize:
H3: Nonprofits domiciled in highly regulated states are more likely to report identical program ratios.
4 Research design
We use the following logistical regression model, cluster-corrected by the nonprofit, to determine whether NPOs with the incentives described increase the probability that they will report an identical ratio. All variables are constructed using data on Form 990.where Identical Ratio is a dichotomous variable equal to 1 if the absolute value of the period t change in the ratio of program expense to total expense is less than 0.5% (no change), and 0 otherwise (change).16 Note that most sample NPOs are coded one in some periods and zero in others. By allowing NPOs to switch between groups, we essentially control for unobserved organizational characteristics that might relate to NPOs reporting identical ratios.17
$$\begin{array}{c}P{(Identical\; Ratio)}_{it}={\beta }_{0}+{\beta }_{1}Donor\; Sensitivityit+{\beta }_{2}{Grants}_{it}+{\beta }_{3}Pay\; {Sensitivity}_{it}\\ +{\beta }_{4}{Regulation}_{i}+{Controls}_{it}+{\varepsilon }_{it},\end{array}$$
(1)
Donor Sensitivity is the organization-specific coefficient of the change in the program ratio from running the regression %∆Donations = α0 + α1 ∆Program Ratio + ε, using the eight most recent years of change data before year t.18,19 The higher the coefficient, the more sensitive donors are to changes in ratios. We anticipate β1 > 0. Grants is equal to 1 if government grants are reported in period t and 0 otherwise. We anticipate β2 > 0. Pay Sensitivity is the organization-specific coefficient of the change in the program ratio from running the regression %∆Executive Compensation = α0 + α1 ∆Program Ratio + ε, using the eight most recent years of change data before year t. The higher the organization-specific coefficient, the more sensitive executive pay is to changes in ratios. We anticipate β3 > 0. Regulation is the sum of the number of state-level governance and disclosure laws where the nonprofit is domiciled (obtained from Desai and Yetman 2015). Values range from 0 to 17. We anticipate β4 > 0.
Our controls are those that explain why, absent ratio management, program ratios may not change. Reported ratios may not increase simply because ratios are already high. That is, if the ratio is 90 percent, despite how much more the NPO spends on programs (compared to overhead) it is difficult to raise the percentage, resulting in an identical ratio. Incentives to report identical ratios may also depend on the ratio reported in the prior period (Kitching et al. 2012). Thus we include the ratio of program expense to total expense at period t-1, Lag Program Ratio. Nonprofits with a variety of operations are more likely to report internal control problems (Petrovits et al. 2011). Thus, to the extent organizations with weaker internal controls are more likely to misreport accounting information, we include Complexity, which represents the number of ways the NPO generates revenue (public support and program revenue).20 Relatively small changes in total spending may represent stable operations. If so, organizations that experience smaller changes in total spending are more likely to report identical ratios. We include the absolute Δ Total Spending to control for this possibility.21
Identical ratios may result from misreporting due to a lack of organizational or managerial sophistication (Krishnan et al. 2006; Keating et al. 2008; Petrovits et al. 2011; Parsons et al. 2017) and not ratio management. Specifically, small organizations might report identical ratios simply because they lack resources to track program, administration, and fundraising costs (Wing and Hager 2004b) or because their stakeholders do not emphasize the proper allocation of costs (Wing and Hager 2004b). Additionally, reputation may impact management’s decision to manage ratios (Tinkelman 1999 and Kitching 2009). We include Age and Size to control for organizational (and managerial) sophistication and reputation. We do not predict the direction of the associations for the control variables.22,23 Following convention, we control for industry (using the National Taxonomy of Exempt Entities, NTEE) and year. Variable definitions and data sources are described in Table 1.
Table 1
Variable Definitions
Variable | Definition | Data Source |
|---|---|---|
Identical Ratio | equal to one if the absolute value of the change in the ratio of program expense to total expense from period t-1 to period t is less than 0. 5% | Form 990 |
Donor Sensitivity | the coefficient of the firm-specific change in program ratio variable from running the regression %∆Donationsit = β0 + β1 ∆Program Ratioit + ε | Form 990 |
Grants | equal to one if the organization reports federal government grants at the end of the year and zero otherwise | Form 990 |
Pay Sensitivity | the coefficient of the firm-specific change in program ratio variable from running the regression %∆Total Payit = β0 + β1 ∆Program Ratioit + ε | Form 990 |
Regulation | the sum of the number of state-level governance and reporting laws where the nonprofit is domiciled; values can range from 0 to 17 | Desai and Yetman [2015] |
Lag Program Ratio | prior period ratio of program expense to total expense | Form 990 |
Complexity | the number of the revenue sources from 0 to 2. Revenue sources include public support and program revenue | Form 990 |
Age | the number of years since the organization filed for 501c(3) exemption | Form 990 and IRS Exempt Organizations Business Master File Extracts |
Size | total assets in $100 million | Form 990 |
∆ Total Spending | the absolute change in total expense | Form 990 |
∆ Total Revenue | the absolute change in total revenue | Form 990 |
5 Data
We use nonprofits included in the IRS’s Statistics of Income (SOI) dataset that report program expenses between 1994 and 2015 (the most recent year available at the time we collect data). As previously described, at least nine years of data before year t are required to construct the Donor Sensitivity and Pay Sensitivity variables; thus the sample period begins in 2003. The SOI dataset includes Form 990 data for the largest NPOs (plus a random sample of smaller nonprofits) in the United States (Feng et al. 2014). The advantage of studying mostly large nonprofits is we study relatively sophisticated organizations with stakeholders who care about and with resources to track cost allocation.24
We eliminate observations with zero fundraising expense to ensure relationships between the dependent variable and the incentive variables are driven by incentives to report identical ratios and not incentives to report zero fundraising expense (Krishnan et al. 2006). We also eliminate observations with errors on Form 990 and without data needed to calculate our variables.25
Table 2
Sample Selection
nonprofit- years | |
|---|---|
501c(3) nonprofits in SOI datasets between 1994–2015 with at least 10 years of data | 45,150 |
Less observations: | |
fundraising expense equals zero | 11,030 |
without data needed to calculate regression variables | 2,255 |
errors on Form 990 and influential observations | 449 |
the absolute change in total spending is less than 5% | 13,965 |
Final sample: | 17,451 |
Without changes in program ratios between years, no ∆ group | 4,022 |
With changes in program ratios between years, ∆ group | 13,429 |
Unique nonprofits | |
Total: | 4,063 |
Without changes in program ratios between years, no ∆ group | 2,180 |
With changes in program ratios between years, ∆ group | 1,883 |
Finally, because relatively small changes in spending might reflect stable operations and warrant identical ratios (and thus not reflect ratio management), we eliminate observations where the absolute change in total spending is 5 percent or lower. (We consider the reasonableness of the 5 percent cutoff in a sensitivity test.) Although we restrict the sample to include NPOs with at least a 5 percent (absolute) change in total spending, the average NPO in our sample reports a 14.8 percent (median = 9.3 percent) change in total spending and spends an additional $11.9 million (median = $2.74 million) between periods. Changes of this magnitude make it unlikely for a team of managers to control spending in ways such that the NPO could spend the exact same proportions on programs and overhead to report an identical ratio. It is also difficult to make real spending decisions to keep ratios the same given that spending decisions occur over the course of a year and not all at the beginning or end of a period. These are important points, given the focus of our study is whether NPOs manage cost allocations to report identical ratios.
Also, we require nonprofits to report at least a 5 percent change in total spending and not a 5 percent change in total revenue. This distinction is important because nonprofits save additional revenue to smooth spending and keep operations stable (Jones et al. 2013). Thus, when they report large changes in total spending (especially increases), this likely reflects managers’ decisions to either change the level of involvement in existing activities or to engage in new activities, that is, to change operations.
The final sample includes 17,451 NPO-years (4,063 unique NPOs), of which 4,022 observations (2,180 unique NPOs) have identical ratios, labeled the no change group, and 13,429 observations (1,883 unique NPOs) have ratios that change, labeled the change group.26 Table 2 reports the sample selection process.
Table 3 reports the sample distribution. Panel A reports industry distributions. Charitable organizations are the largest category (52.1%), followed by educational (34.1%) and medical (13.8%) institutions. Educational (27.3%) and medical (26.0%) institutions have a greater propensity to report identical ratios than charities (19.5%). We explore later whether incentives to report identical ratios are influenced by industry. Panel B reports the year distribution. Each year is fairly represented in the sample.27 Panel C reports the number of times nonprofits are in the sample and the number of times they report an identical ratio. Of the 2,180 nonprofits that report identical ratios, 454 do so in three or more periods, and, of the 454, 37 report the same ratio every year throughout the sample period.28,29
Table 3
Sample Distribution
Panel A: Industry distribution | ||||||||||||||||||
Industry | Frequency | % | No change nonprofit-years | % No change nonprofit-years | ||||||||||||||
Charitable Organizations | 9,095 | 52.1% | 1,770 | 19.5% | ||||||||||||||
Arts, Culture, and Humanities | 1,776 | 10.2% | 242 | 13.6% | ||||||||||||||
Environment and Animals | 810 | 4.6% | 145 | 17.9% | ||||||||||||||
Health and Medical Research | 418 | 2.4% | 70 | 16.7% | ||||||||||||||
Human Services | 2,646 | 15.2% | 658 | 24.9% | ||||||||||||||
International, Foreign Affairs | 380 | 2.2% | 87 | 22.9% | ||||||||||||||
Public and Societal Benefit | 3,065 | 17.6% | 568 | 18.5% | ||||||||||||||
Educational (includes universities) | 5,956 | 34.1% | 1,628 | 27.3% | ||||||||||||||
Medical (hospitals) | 2,400 | 13.8% | 624 | 26.0% | ||||||||||||||
Total | 17,451 | 100.0% | 4,022 | 23.0% | ||||||||||||||
Panel B: Year distribution | ||||||||||||||||||
Year | Frequency | % | No change nonprofit-years | % No change nonprofit-years | ||||||||||||||
2003 | 369 | 2.1% | 99 | 26.8% | ||||||||||||||
2004 | 619 | 3.5% | 163 | 26.3% | ||||||||||||||
2005 | 957 | 5.5% | 232 | 24.2% | ||||||||||||||
2006 | 1,215 | 7.0% | 316 | 26.0% | ||||||||||||||
2007 | 1,472 | 8.4% | 369 | 25.1% | ||||||||||||||
2008 | 1,467 | 8.4% | 293 | 20.0% | ||||||||||||||
2009 | 1,423 | 8.2% | 288 | 20.2% | ||||||||||||||
2010 | 1,472 | 8.4% | 337 | 22.9% | ||||||||||||||
2011 | 1,492 | 8.5% | 319 | 21.4% | ||||||||||||||
2012 | 1,547 | 8.9% | 344 | 22.2% | ||||||||||||||
2013 | 1,642 | 9.4% | 367 | 22.4% | ||||||||||||||
2014 | 1,834 | 10.5% | 433 | 23.6% | ||||||||||||||
2015 | 1,942 | 11.1% | 462 | 23.8% | ||||||||||||||
Total | 17,451 | 100% | 4,022 | 23.0% | ||||||||||||||
Panel C: Nonprofit distribution by number of times nonprofits are in the sample and number of times nonprofits report an identical ratio | ||||||||||||||||||
Number of years nonprofit is in sample | Number of years nonprofit reports identical ratio | |||||||||||||||||
0 | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 | Total | % | |||||
1 | 554 | 161 | - | - | - | - | - | - | - | - | - | - | 715 | 17.6% | ||||
2 | 364 | 181 | 56 | - | - | - | - | - | - | - | - | - | 601 | 14.8% | ||||
3 | 283 | 160 | 73 | 19 | - | - | - | - | - | - | - | - | 535 | 13.2% | ||||
4 | 195 | 171 | 85 | 22 | 9 | - | - | - | - | - | - | - | 482 | 11.9% | ||||
5 | 162 | 139 | 98 | 37 | 13 | 3 | - | - | - | - | - | - | 452 | 11.1% | ||||
6 | 114 | 130 | 80 | 54 | 28 | 10 | 3 | - | - | - | - | - | 419 | 10.3% | ||||
7 | 87 | 78 | 64 | 39 | 25 | 5 | 3 | 1 | - | - | - | - | 302 | 7.4% | ||||
8 | 53 | 65 | 39 | 22 | 11 | 14 | 7 | 2 | - | - | - | - | 213 | 5.2% | ||||
9 | 29 | 38 | 32 | 24 | 15 | 10 | 3 | 3 | - | - | - | - | 154 | 3.8% | ||||
10 | 28 | 28 | 14 | 12 | 14 | 3 | 5 | 1 | 1 | 1 | 1 | - | 108 | 2.7% | ||||
11 | 9 | 8 | 11 | 5 | 4 | 4 | 3 | 1 | 1 | - | - | 1 | 47 | 1.2% | ||||
12 | 5 | 8 | 5 | 7 | 1 | - | - | 2 | 1 | - | - | - | 29 | 0.7% | ||||
13 | - | 1 | 1 | 1 | - | 2 | - | - | - | 1 | - | - | 6 | 0.1% | ||||
Total | 1,883 | 1,168 | 558 | 242 | 120 | 51 | 24 | 10 | 3 | 2 | 1 | 1 | 4,063 | 100.0% | ||||
% | 46.3% | 28.7% | 13.7% | 6.0% | 3.0% | 1.3% | 0.6% | 0.2% | 0.1% | 0.0% | 0.0% | 0.0% | 100.0% | |||||
Table 4 presents descriptive information about the variables included in the regression model. Mean Donor Sensitivity is 6.1 (median = 0.2), indicating a one percentage-point change in the program ratio results in a 6.1 (0.2) percent change in donations. Over half of the sample have government grants. Mean Pay Sensitivity is 1.2, indicating a one percentage-point change in the program ratio results in a 1.2 percent change in executive pay. Although the median is zero, the standard deviation is 2.75, indicating ample cross-sectional variation. Refer to Appendix Table 12 for descriptive information about the variables used to compute the Donor and Pay Sensitivity measures. The mean (median) value for Regulation is 11.5 (13.0), but values range from 3 (low regulatory environment) to 16 (high regulatory environment). Finally, our sample’s average NPO spends about $108 million per year. (The median spends $26 million.) The no-change and change groups differ on each attribute identified (p < 0.01 and p < 0.10 for Regulation). Thus the inclusion of the control variables is warranted. Except for the correlation between Size and Δ Total Spending (the correlation coefficient is 0.77), correlation coefficients among the other variables do not exceed 0.30, and variance inflation factors (for all variables) do not exceed 1.3. Thus multicollinearity is not of concern.
Table 4
Sample Descriptive Statistics
Full Sample | No-Change Group | Change Group | |||||||
|---|---|---|---|---|---|---|---|---|---|
(N = 17,451) | (N = 4,022) | (N = 13,429) | |||||||
Mean | Median | Std. Dev | Mean | Median | Std. Dev | Mean | Median | Std. Dev | |
Donor Sensitivity | 6.138 | 0.244 | 15.898 | 8.697 | 0.126 | 22.170 | 5.371 | 0.270 | 13.369 |
Grants | 0.569 | 0.569 | 0.495 | 0.629 | 1.000 | 0.483 | 0.551 | 1.000 | 0.497 |
Pay Sensitivity | 1.167 | 0.000 | 2.749 | 1.585 | 0.000 | 3.677 | 1.042 | 0.000 | 2.389 |
Regulation | 11.499 | 13.000 | 3.000 | 11.579 | 13.000 | 2.980 | 11.475 | 13.000 | 3.006 |
Lag Program Ratio | 0.816 | 0.833 | 0.100 | 0.855 | 0.865 | 0.080 | 0.805 | 0.822 | 0.102 |
Complexity | 1.839 | 2.000 | 0.368 | 1.900 | 2.000 | 0.301 | 1.821 | 2.000 | 0.384 |
Age in years | 48.301 | 48.000 | 20.090 | 50.349 | 52.000 | 20.067 | 47.688 | 47.000 | 20.057 |
Size (Total assets, in $100 M) | 3.061 | 0.857 | 12.565 | 4.654 | 1.118 | 19.640 | 2.584 | 0.793 | 9.417 |
∆ in Program Ratio | 0.028 | 0.014 | 0.044 | 0.002 | 0.002 | 0.001 | 0.036 | 0.020 | 0.048 |
absolute %∆Total Spending | 0.148 | 0.093 | 0.237 | 0.095 | 0.079 | 0.062 | 0.163 | 0.099 | 0.266 |
absolute ∆Total Spending | $11,886,373 | $2,738,301 | $48,860,803 | $15,502,132 | $3,827,787 | $43,270,029 | $10,803,449 | $2,487,926 | $50,365,743 |
Total Spending | $108,275,894 | $25,842,697 | $357,864,159 | $174,882,649 | $44,810,709 | $479,090,502 | $88,327,100 | $21,473,374 | $309,785,594 |
absolute %∆Total Revenue | 0.540 | 0.128 | 30.010 | 0.207 | 0.102 | 1.125 | 0.639 | 0.141 | 34.204 |
absolute ∆Total Revenue | $17,270,280 | $3,620,662 | $77,686,009 | $22,837,583 | $4,841,057 | $79,536,847 | $15,602,867 | $3,316,373 | $77,047,799 |
Total Revenue | $119,477,408 | $28,938,664 | $392,067,642 | $194,528,477 | $49,531,313 | $536,567,958 | $96,999,531 | $23,808,353 | $333,696,845 |
6 Results
6.1 Main results
Table 5 reports results of the logistic regression based on Eq. (1). Recall that a positive coefficient indicates NPOs are more likely to report an identical ratio. The coefficients on the resource incentive variables – Donor Sensitivity and Grants – are positive (p < 0.0001 for Donor Sensitivity and p < 0.0215 for Grants), supporting H1a and H1b, our hypotheses that predict NPOs are more likely to report identical ratios when they rely on resource providers who use ratios in their giving decisions. The coefficient on Pay Sensitivity is positive and significant (p < 0.0001), supporting H2, our hypothesis that the more sensitive executive pay is to changes in the program ratio, the more likely the NPO is to report an identical ratio. The coefficient on Regulation is positive and significant (p < 0.0411), supporting H3, our hypothesis that nonprofits domiciled in states with relatively strong regulation are more likely to report identical ratios.
Table 5
Logistic Regression of the Likelihood of Reporting Identical Program Ratios
Dependent variable: Identical Ratio | Hypothesis | Predicted Sign | Coefficient (Std. Error) | |
|---|---|---|---|---|
Constant | -7.911 | *** | ||
(0.266) | ||||
Donor Sensitivity | H1a | + | 0.005 | *** |
(0.001) | ||||
Grants | H1b | + | 0.082 | ** |
(0.040) | ||||
Pay Sensitivity | H2 | + | 0.026 | *** |
(0.006) | ||||
Regulation | H3 | + | 0.011 | ** |
(0.006) | ||||
Lag Program Ratio | 6.313 | *** | ||
(0.261) | ||||
Complexity | 0.476 | *** | ||
(0.063) | ||||
Age | 0.002 | ** | ||
(0.001) | ||||
Size | 0.004 | ** | ||
(0.001) | ||||
∆ Total Spending | -0.126 | ** | ||
(0.050) | ||||
Industry and Year effects | YES | |||
Pseudo R2 | 6.94% | |||
Number of obs. in no ∆ group | 4,022 | |||
N | 17,451 |
The coefficients on the control variables Lag Program Ratio, Complexity, Age, and Size are positive and significant, indicating that NPOs with high program ratios, complex operations, and sophistication (as indicated by age and size) are more likely to report identical ratios.30 The coefficient on Δ Total Spending is negative and significant, indicating NPOs are less likely to report identical ratios the larger the change in total spending.
The model’s pseudo R-squared is 6.9 percent (and is sometimes higher, depending on robustness and additional testing described later). Following Yetman and Yetman (2012), we evaluate the joint economic significance of our results by comparing predicted outcomes across different values of the model variables. We increase the values by one-half of a standard deviation for continuous variables and change the value from zero to one for indicator variables. Our analysis (untabulated) indicates a 14 percent increase in the probability of reporting an identical ratio. This means NPOs with characteristics included in the model are 61 percent (14.0%\(\div\) 23.0%, the percentage of sample NPOs that report identical ratios) more likely to report identical ratios.
6.2 Sensitivity analyses
We conduct several tests to examine the robustness of our main results. First, we test whether our requirement that observations have at least a 5 percent change in total spending is large enough to reflect a change in operations. Recall that we make this restriction to better ensure NPOs have experienced a change in its operations and thus a change in ratios is warranted. Although the mean (median) NPO in our sample has a 14.8 (9.3) percent change in spending between periods, which translates to an additional $11.9 ($2.7) million of spending, the 5 percent cutoff may be too small to reflect a change in operations. Results (untabulated) are generally the same when we increase the 5 percent restriction in 1 percentage point increments up to 10 percent and re-estimate the model.
Next we test whether the results are robust to our assumption that, when an nonprofit reports a large change in total spending and simultaneously reports an identical program ratio, it has intentionally managed cost allocations to achieve this outcome. However, ratio management may be reflected by how frequently NPOs report identical ratios, and when an NPO reports an identical ratio once, this has occurred by chance and not because of ratio management. To address this concern, we partition the sample and separate no-change observations into two groups and re-estimate the model. In particular, one partition includes NPOs that report identical ratios multiple times, and the other includes NPOs that report identical ratios one time during the sample period. Results (untabulated) support our hypotheses when NPOs report identical ratios multiple times during the sample period, but not when identical ratios are reported only once. These results suggest that by classifying NPOs that report identical ratios once into the no change group, we have introduced noise and the primary results are biased downward. Further analyses (untabulated) suggest the likelihood of ratios being managed strengthens the more times the NPO reports identical ratios.
Next recall we include Lag Program Ratio in the model to control for the possibility that nonprofits with very high program ratios may not report program ratio increases. Thus, regardless of the magnitude of the change in spending, these nonprofits will report an identical ratio. To the extent that nonprofits with very high program ratios are the same nonprofits with high donor sensitivity, for example, these nonprofits may be more likely to report identical ratios simply because they cannot increase the ratio. To rule out the possibility that nonprofits with very high program ratios drive our main results, we exclude observations with program ratios above 90 percent (and then 95 percent) and re-run the model. Results (untabulated) are qualitatively the same as the main results, indicating nonprofits with very high ratios do not drive our results.
6.3 Additional analyses
In this section, we further explore specific conditions under which nonprofits report identical ratios. First, we examine whether incentives to report identical ratios vary by industry. Following prior research (e.g., Krishnan et al. 2006), we partition the sample into three primary nonprofit industries, i.e., charitable, educational, and medical (hospitals) institutions.
Table 6 presents the results and shows that incentives to report identical ratios are influenced by industry. In particular, charities versus medical and educational institutions are more likely to report identical ratios when donors are sensitive to program ratio changes. These differences are likely because charities depend more on donors as a revenue source: the percentage of donor revenue to total revenue is 45.4 percent, 20.6 percent, and 15.4 percent for charities, educational institutions, and medical institutions, respectively. Medical and educational institutions are more likely to report identical ratios when they have government grants. Educational institutions, in particular, depend more on government grants than charities. The mean educational institution in the sample receives $12.2 million annually in government grants (versus $5.7 million for the mean charity), and 64.2 percent of sample educational institutions receive government grants (versus 52.8 percent of charities).
Table 6
Logistic Regression of the Likelihood of Reporting Identical Program Ratios: by Industry
Charitable | Educational | Medical | ||||||
|---|---|---|---|---|---|---|---|---|
Dependent variable: Identical Ratio | Hypothesis | Predicted Sign | Coefficient (Std. Error) | Coefficient (Std. Error) | Coefficient (Std. Error) | |||
Constant | -17.867 | -7.323 | *** | -10.622 | *** | |||
(201.300) | (0.476) | (0.751) | ||||||
Donor Sensitivity | H1a | + | 0.009 | *** | 0.002 | 0.003 | ||
(0.002) | (0.002) | (0.002) | ||||||
Grants | H1b | + | 0.010 | 0.108 | * | 0.208 | ** | |
(0.060) | (0.070) | (0.105) | ||||||
Pay Sensitivity | H2 | + | 0.031 | *** | 0.027 | *** | 0.001 | |
(0.010) | (0.010) | (0.015) | ||||||
Regulation | H3 | + | 0.014 | * | 0.008 | 0.016 | ||
(0.010) | (0.010) | (0.016) | ||||||
Lag Program Ratio | 6.326 | ** | 5.397 | *** | 8.582 | *** | ||
(0.400) | (0.416) | (0.757) | ||||||
Complexity | 0.240 | *** | 0.751 | *** | 1.131 | *** | ||
(0.079) | (0.157) | (0.189) | ||||||
Age | 0.001 | 0.003 | 0.000 | |||||
(0.001) | (0.002) | (0.002) | ||||||
Size | 0.035 | ** | 0.004 | ** | 0.020 | ** | ||
(0.008) | (0.002) | (0.006) | ||||||
∆ Total Spending | -0.418 | ** | -0.180 | * | -0.227 | ** | ||
(0.160) | (0.103) | (0.096) | ||||||
Industry and Year effects | YES | YES | YES | |||||
Pseudo R2 | 7.32% | 5.07% | 11.60% | |||||
Number of obs. in no ∆ group | 1,770 | 1,628 | 624 | |||||
N | 7,335 | 4,328 | 1,766 |
Interestingly, we find that charities and educational institutions, but not medical institutions, are more likely to report identical ratios when executive pay is sensitive to changes in ratios. This result is somewhat surprising, given that both hospitals and colleges and universities use metrics in addition to program ratios to evaluate organizational performance (e.g., Eldenburg and Krishnan 2003; Parsons and Reitenga 2014). Finally, we find that only charities are more likely to report identical ratios when domiciled in states that heavily regulate nonprofits. Our state regulation measure excludes schools and hospitals, as these are frequently exempted from many state-level nonprofit laws and are subject to other specific laws and regulations (Desai and Yetman 2015). Our result does not necessarily indicate regulation does not influence reporting for medical and education institutions; it merely suggests that our measure cannot be used to determine its effects.
Next, given research documenting a positive association between donations (and to some extent executive pay) and program ratios, we examine whether the association between reporting an identical ratio and incentive variables strengthens when nonprofits want to avoid reporting ratio decreases (versus increases). To accomplish this, we partition observations into two groups and re-estimate the model. The first (second) group in Table 7 includes observations where program ratio changes are negative (positive). Hence this test examines the choice between reporting a decrease (increase) in the program ratio and reporting an identical ratio.
Table 7
Logistic Regression of the Likelihood of Reporting Identical Program Ratios: by Direction of Program Ratio Change
Incentive to avoid reporting ratio decreases | Incentive to avoid reporting ratio increases | |||||
|---|---|---|---|---|---|---|
Dependent variable: Identical Ratio | Hypothesis | Predicted Sign | Coefficient (Std. Error) | Coefficient (Std. Error) | ||
Constant | -5.385 | *** | -8.928 | *** | ||
(0.287) | (0.302) | |||||
Donor Sensitivity | H1a | + | 0.006 | *** | 0.005 | *** |
(0.001) | (0.001) | |||||
Grants | H1b | + | 0.158 | *** | 0.006 | |
(0.045) | (0.046) | |||||
Pay Sensitivity | H2 | + | 0.034 | *** | 0.020 | ** |
(0.007) | (0.007) | |||||
Regulation | H3 | + | 0.012 | * | 0.008 | |
(0.007) | (0.007) | |||||
Lag Program Ratio | 4.101 | *** | 8.400 | *** | ||
(0.279) | (0.300) | |||||
Complexity | 0.495 | *** | 0.458 | *** | ||
(0.068) | (0.070) | |||||
Age | 0.003 | ** | 0.002 | ** | ||
(0.001) | (0.001) | |||||
Size | 0.005 | ** | 0.011 | ** | ||
(0.002) | (0.003) | |||||
∆ Total Spending | -0.068 | -0.260 | ||||
(0.048) | (0.073) | |||||
Industry and Year effects | YES | YES | ||||
Pseudo R2 | 4.94% | 10.72% | ||||
Number of obs. in no ∆ group | 4,022 | 4,022 | ||||
N | 10,555 | 10,918 | ||||
Results indicate that it matters whether nonprofits want to avoid reporting a ratio decrease or increase. Specifically, the coefficients on Grants and Regulation are positive and significant only when nonprofits want to avoid reporting a decrease in the program ratio. This suggests managers are more inclined to report identical ratios than to report program ratio decreases, given decreases might cause them to miss benchmarks established in grant contracts or set by regulators. The coefficients on Donor Sensitivity and Pay Sensitivity are positive and significant for both increases and decreases. These results support the arguments made by Jones and Roberts (2006) that, in addition to the desire to avoid reporting program ratio decreases, nonprofits may want to avoid reporting ratio increases so as to not raise donors’ or boards’ expectations concerning future program spending.
Next we examine whether nonprofits are more or less likely to report identical ratios—that is, manage cost allocations—when managers can easily explain changes in ratios and when stakeholders will readily accept managers’ explanations. One such condition is during an economic downturn. During downturns, donations decrease, and, depending on philanthropic objectives, program service delivery may change, causing a decrease (or increase) in the program ratio. This provides an easy explanation for changing ratios that stakeholders can accept. This explanation is analogous to for-profit firm managers attributing a decline in earnings to the economy during recessions.31 To examine whether identical ratios depend on economic conditions, we include an indicator variable Recession, which is equal to 1 if the year is between 2007 and 2009 (years of Great Recession, as indicated by the National Bureau of Economic Research) and 0 otherwise. Results (untabulated) indicate the coefficient on Recession is negative and significant (p < 0.05), suggesting NPOs are less likely to report identical ratios during economic downturns.
In our final additional test, we examine directly whether NPOs report identical ratios when ratios are at benchmarks set by ratings agencies. There are three major charity ratings agencies – Charity Navigator, the Better Business Bureau’s (BBB) Wise Giving Alliance, and Charity Watch. Charity Navigator’s ratings are based on the distance between an NPO’s score to perfect scores in two areas – financial health and accountability and transparency—using an approach similar to the two-dimension Euclidean distance.32 The minimum program ratio requirement, which is one part of the financial health score, is 50%. Nonprofits earn points toward the financial health score for program ratios above 50% and up to 85%. Charity Navigator does not provide a bright-line ratio. But, based on how its point system works, NPOs may view 85% as Charity Navigator’s ratio benchmark. Note that Charity Navigator uses other ratio ranges for particular subsets of nonprofits—i.e., 50 percent–92 percent for food banks, food pantries, food distribution, and humanitarian relief supplies; 50 percent–83 percent for museums; and 50 percent–82 percent for public broadcasting and media. Charity Watch requires charities spend between 75 percent–79 percent on programs for an A- rating, 80 percent–89 percent for an A rating, and over 90 percent for an A + rating. And the BBB’s standard requires program ratios equal at least 65 percent.
We plot the distribution based on program ratios for the full sample and find a cluster statistically significant at 82%. Although this percentage does not correspond to any rating agency benchmark, we cannot conclude NPOs do not manage ratios to meet rating agencies' benchmarks.
7 Techniques and expenses used to manage ratios
On Form 990, the IRS requires nonprofits report their expenses in a matrix format. Expenses are classified into over 20 natural expense classifications. Examples include salaries and wages, accounting fees, and travel expenses. These expense classifications are then allocated among the program, fundraising, and administrative functional expense categories.
7.1 Cost allocation techniques
We describe each technique using the examples (with selected expenses) described in the bottom half of Appendix Table 11. We focus on deviations from the base case where a $10 increase in total spending is due to wage increases (for example) and employees maintain the time they spend on program activities. (The analysis is the same when we consider the case where the $10 increase is due to a change in activity levels.) Because salaries and wage expense is $70 in period t-1 and $80 in period t and the nonprofit properly allocates 86 percent of wages to programs in both periods (and the other 14 percent is allocated to overhead), the overall program ratio will change from 73 percent to 74.2 percent. We describe three techniques NPOs may use to achieve a target ratio, which in this case, is a ratio identical to the one reported in the prior period. Note that NPOs can use one of the techniques for a specific period but can switch to another in another period.
Technique #1 (T1), Keep program allocation rates the same as the prior period for all natural expense classifications—Nonprofits that use this technique keep allocation rates the same as the previous year but change every natural expense classification by the same percentage. In our example, all expenses in period t increase 10 percent, and the percentage allocated from each expense line to programs is just like the prior year.33
Technique #2 (T2), Change the program allocation rate for one natural expense classification— Nonprofits that use this technique correctly report the natural expense classification, but, if they use the appropriate rates to allocate costs to programs, the overall program ratio will change. To counteract this potential effect, the nonprofit changes the allocation rate of one of the classifications. In our example, salaries and wage expense equals $80 and 86 percent of it is allocated to programs, but the allocation rate changes for other expenses.
Technique #3 (T3), Change allocation rates for multiple natural expense classifications—This resembles T2, but nonprofits that use this technique change the allocation rates for multiple natural expense classifications (e.g., salaries and wages and other expenses) to achieve an overall identical program ratio. We distinguish T3 from T2 because, in our view, the former requires more effort on the part of the preparer. It requires the preparer to alter multiple allocation rates rather than just one.34
To study techniques, we start with the 2,180 NPOs that report identical ratios (see Table 2) and use SOI data from 1994 through 2015.35 Beginning with the 2008 tax year, the IRS redesigned Form 990, which resulted in changes to the statement of functional expenses—specifically, the newer statement aggregates some expense lines from the old form and adds several new expense lines. We perform analyses separately using the new and the old Form 990 s. Further, because changes in natural expense classification allocations may result from changes in the form and not changes in allocations, we do not use data from the 2008 Form 990 (the first year of the new Form 990). As a result, we lose 74 NPOs because they only report an identical ratio in tax year 2008. The final sample for the technique analyses is 2,106 unique NPOs.
Table 8 presents nonprofit level descriptives by whether the NPO uses T1, T2, or T3 solely during the sample period or a combination of the techniques to manipulate cost allocations. Approximately 4 percent (89 of 2,106) of sample NPOs use a combination of techniques during the sample period. (The sum of NPOs that use one technique and multiple techniques equal 2,106). These NPOs are much larger (and have higher program ratios) than those that use a single technique. The mean total spending for NPOs that use multiple techniques is $247 million compared to $126 million for those that use a single technique. This suggests that the most sophisticated NPOs, in terms of size, rely on various methods to manage ratios.
Table 8
Nonprofit-Level Descriptive Statistics by Ratio Management Technique
One technique nonprofits (N = 2,017) | Multiple technique nonprofits (N = 89) | Diff | T1 nonprofits (N = 43) | T2 nonprofits (N = 38) | T3 nonprofits (N = 1,936) | Diff. T1 vs. T2 | Diff. T 1 vs. T3 | Diff. T2 vs. T3 | |
|---|---|---|---|---|---|---|---|---|---|
Donor Sensitivity | 7.331 | 9.102 | 13.269 | 6.522 | 7.215 | ||||
Grants | 0.608 | 0.602 | 0.270 | 0.343 | 0.621 | *** | *** | ||
Pay Sensitivity | 1.399 | 1.828 | 0.958 | 1.091 | 1.415 | ** | |||
Regulation | 11.559 | 11.000 | 10.535 | 11.763 | 11.578 | * | ** | ||
Lag Program Ratio | 0.836 | 0.874 | *** | 0.835 | 0.837 | 0.836 | |||
Complexity | 1.896 | 1.876 | 1.583 | 1.857 | 1.904 | ** | *** | ||
Age | 49.649 | 50.419 | 46.410 | 43.323 | 49.845 | ** | |||
Size | 3.299 | 8.419 | * | 3.090 | 3.437 | 3.301 | |||
ΔTotal Spending | 0.119 | 0.115 | 0.135 | 0.125 | 0.118 | ||||
Total Spending | $125,833,206 | $247,223,283 | ** | $83,921,506 | $137,150,208 | $126,541,965 |
The data indicate that most NPOs prefer to use one technique to manage ratios and prefer to change the allocation rates for multiple expense items (T3). Descriptive statistics suggest incentives may influence technique choice. The most striking differences are between T1 (or T2) and T3. Those that use T3 have greater incidences of government grants, have executives whose pay is more sensitive to changes in the program ratio, are domiciled in more regulated states, and have more complex operations. Note that our analysis suggests most nonprofits use T3. But, because our sample is biased toward large organizations experiencing large spending changes, our analysis likely understates the use of T1 and T2.
Both sets of results suggest organizational sophistication influences methods nonprofits use to manage ratios. Research has concluded that more sophisticated nonprofits are less likely to manage ratios (e.g., Krishnan et al. 2006; Keating et al. 2008). Our finding suggests that, once nonprofits decide to manage financial information, their sophistication dictates how they do so.
7.2 Natural expense classifications
Nonprofits that keep program allocation rates for each natural expense classification the same as the previous year use T1. Those that change allocation rates use T2 or T3. Allocation rates change if they differ by at least 1percent from the prior period. In this section, we explore which, if any, natural expense classifications NPOs favor (i.e., change) when using T2 or T3.
To do this, for each NPO that reports an identical ratio, we classify NPO-years into either a manipulation or non-manipulation period, based on whether the NPO reports an identical ratio in that particular period. We do this separately for the new and old Form 990. We then compare how frequently the NPO changed natural expense classification allocation rates in its manipulation period with how frequently it changed allocation rates in its nonmanipulation period. We use the frequency of changes in natural expense classification allocation rates in nonmanipulation periods to benchmark the expected allocation activity, absent ratio management. We focus on the case when, on average, changes in the percentage allocated to programs, for a given natural expense classification, happen more frequently in manipulation periods (when the nonprofit reports an overall identical ratio) than in non-manipulation periods (when the same nonprofit reports a change in its overall ratio). For example, if a nonprofit changes the allocation rate for salaries and wages 20 percent of the time in its non-manipulation periods but changes it 50 percent of the time in its manipulation periods, we presume the nonprofit uses the natural expense classification “salaries and wages” to manage program ratios, which in this case, is to report an identical program ratio. We evaluate the occurrence of these cases for every natural expense classification.
To be clear, we do not suggest changes in natural expense classification allocation rates are good or bad or how frequent they should be. We suggest changes in allocation rates in non-manipulation periods indicate a given NPO’s normal allocation pattern. Our expectation is that, absent manipulation, the frequency of changes in allocation rates should be equal in manipulation and non-manipulation periods. When nonprofits report identical ratios, we anticipate that program allocation rates change more frequently for natural expense classifications where managers have more discretion regarding how much to allocate to programs.
Tables 9 and 10 list all natural expense classifications that can be allocated among the three functional expense categories. Column 2 reports the number of NPOs that change program allocation rates more frequently in manipulation periods than in non-manipulation periods for each classification, and column 3 ranks classifications based on the numbers in columns 2. We perform this analysis separately using the new Form 990 (panel A) and the old Form 990 (Panel B). Note that the number of NPOs on panel A (panel B) is smaller than the number reported in Table 8 because an NPO must be included in both the manipulation and non-manipulation period between 2009–2015 (1994–2007). Thus nonprofits are excluded from the analysis if they report identical ratios for the entire time in the sample.
Table 9
Nonprofit-Level Analysis of Changes in Allocation Rates by Expense Line Item for Nonprofits that Change Allocation Rates for One Expense Item
Panel A, New Form 990, N = 24 | ||||
Form 990 line number | Natural Expense Classification | Number of nonprofits* | Ranking | |
line 5 | Compensation: current officers etc | 2 | 7 | |
line 6 | Compensation: disqualified persons | 1 | 14 | |
line 7 | Other salaries and wages | 2 | 7 | |
line 8 | Pension plan | 2 | 7 | |
line 9 | Other employee benefits | 2 | 7 | |
line 10 | Payroll taxes | 2 | 7 | |
line 11a | Fees for services: management | 0 | 22 | |
line 11b | Fees for services: legal | 2 | 7 | |
line 11c | Fees for services: accounting | 2 | 7 | |
line 11d | Fees for services: lobbying | 0 | 22 | |
line 11f | Fees for services: investment management fees | 0 | 22 | |
line 11 g | Fees for services: other | 0 | 22 | |
line 12 | Advertising and promotion | 2 | 7 | |
line 13 | Office expenses | 1 | 14 | |
line 14 | Information Technology | 1 | 14 | |
line 15 | Royalties | 0 | 22 | |
line 16 | Occupancy | 3 | 2 | |
line 17 | Travel | 1 | 14 | |
line 18 | Payments of travel or entertainment for public officials | 0 | 22 | |
line 19 | Conferences | 1 | 14 | |
line 20 | Interest | 0 | 22 | |
line 21 | Payments to affiliates | 0 | 22 | |
line 22 | Depreciation | 1 | 14 | |
line 23 | Insurance | 1 | 14 | |
line 24 | Other expenses | 5 | 1 | |
Panel B, Old Form 990, N = 19 | ||||
Form 990 line number | Natural Expense Classification | Number of nonprofits* | Ranking | |
line 25a | Compensation: current officers etc | 3 | 2 | |
line 25b | Compensation: former officers etc | 0 | 14 | |
line 25c | Compensation: disqualified persons | 0 | 14 | |
line 26 | Salaries and wages | 0 | 14 | |
line 27 | Pension plan | 0 | 14 | |
line 28 | Other employee benefits | 1 | 4 | |
line 29 | Payroll taxes | 0 | 14 | |
line 30 | Professional fundraising fees | 0 | 14 | |
line 31 | Accounting fees | 0 | 14 | |
line 32 | Legal fees | 0 | 14 | |
line 33 | Supplies | 0 | 14 | |
line 34 | Telephone | 0 | 14 | |
line 35 | Postage and shipping | 0 | 14 | |
line 36 | Occupancy | 1 | 4 | |
line 37 | Equipment rental and maintenance | 0 | 14 | |
line 38 | Printing and publications | 0 | 14 | |
line 39 | Travel | 0 | 14 | |
line 40 | Conferences | 0 | 14 | |
line 41 | Interest | 1 | 4 | |
line 42 | Depreciation, depletion, etc | 0 | 14 | |
line 43 | Other expenses | 7 | 1 | |
Table 10
Nonprofit-Level Analysis of Changes in Allocation Rates by Expense Line Item for Nonprofits that Change Allocation Rates for Multiple Expense Items
Panel A, New Form 990, N = 1,834 | ||||
Form 990 line number | Natural Expense Classification | Number of nonprofits* | Ranking | |
line 5 | Compensation: current officers etc | 589 | 8 | |
line 6 | Compensation: disqualified persons | 104 | 22 | |
line 7 | Other salaries and wages | 599 | 6 | |
line 8 | Pension plan | 609 | 5 | |
line 9 | Other employee benefits | 648 | 1 | |
line 10 | Payroll taxes | 613 | 4 | |
line 11a | Fees for services: management | 112 | 21 | |
line 11b | Fees for services: legal | 370 | 17 | |
line 11c | Fees for services: accounting | 309 | 18 | |
line 11d | Fees for services: lobbying | 119 | 20 | |
line 11f | Fees for services: investment management fees | 144 | 19 | |
line 11 g | Fees for services: other | 505 | 12 | |
line 12 | Advertising and promotion | 478 | 13 | |
line 13 | Office expenses | 632 | 2 | |
line 14 | Information Technology | 433 | 14 | |
line 15 | Royalties | 56 | 23 | |
line 16 | Occupancy | 591 | 7 | |
line 17 | Travel | 630 | 3 | |
line 18 | Payments of travel or entertainment for public officials | 10 | 25 | |
line 19 | Conferences | 378 | 16 | |
line 20 | Interest | 399 | 15 | |
line 21 | Payments to affiliates | 54 | 24 | |
line 22 | Depreciation | 561 | 10 | |
line 23 | Insurance | 559 | 11 | |
line 24 | Other expenses | 566 | 9 | |
Panel B, Old Form 990, N = 1,591 | ||||
Form 990 line number | Natural Expense Classification | Number of nonprofits* | Ranking | |
line 25a | Compensation: current officers etc | 575 | 9 | |
line 25b | Compensation: former officers etc | 88 | 19 | |
line 25c | Compensation: disqualified persons | 19 | 21 | |
line 26 | Salaries and wages | 548 | 11 | |
line 27 | Pension plan | 524 | 13 | |
line 28 | Other employee benefits | 585 | 6 | |
line 29 | Payroll taxes | 541 | 12 | |
line 30 | Professional fundraising fees | 79 | 20 | |
line 31 | Accounting fees | 262 | 18 | |
line 32 | Legal fees | 326 | 17 | |
line 33 | Supplies | 640 | 2 | |
line 34 | Telephone | 606 | 5 | |
line 35 | Postage and shipping | 631 | 3 | |
line 36 | Occupancy | 577 | 8 | |
line 37 | Equipment rental and maintenance | 571 | 10 | |
line 38 | Printing and publications | 642 | 1 | |
line 39 | Travel | 617 | 4 | |
line 40 | Conferences | 425 | 15 | |
line 41 | Interest | 369 | 16 | |
line 42 | Depreciation, depletion, etc | 506 | 14 | |
line 43 | Other expenses | 578 | 7 | |
Table 9 report results for T2—nonprofits that alter allocation rates of only one natural expense classification. Due to the small number of NPOs that use T2 exclusively—24 for the new 990 and 19 for the old 990—we focus on the natural expense classification most frequently used to manage cost allocations.
When NPOs change allocation rates for only one classification, they favor using the other expenses line to alter program allocation rates. Of the 24 (19) NPOs that use this technique on the new (old) Form 990, 21 (37) percent change program allocation rates more frequently in the manipulation periods than in the non-manipulation periods. No other natural expense classification approaches this number. “Other expenses” include a variety of expenses, making evaluating the appropriateness of allocations difficult. Thus this particular natural expense classification allows managers to use this expense line item to manage ratios.36
Table 10 reports results for T3—nonprofits that alter allocation rates of multiple natural expense classifications. When NPOs change the allocation rates for multiple classifications, they favor classifications where the degree of discretion associated with how much to allocate to programs is relatively high. On the new Form 990, program allocation rates change most frequently, in manipulation periods, for lines associated with salaries and wages (employee benefits, payroll taxes, and pensions), office expenses, and travel. On the old Form 990, program allocation rates change most frequently, in the manipulation periods, for printing and publications, supplies, postage and shipping, travel, and telephone expenses, most of which are combined as office expense in the new form. And then salary-related (other employee benefits, etc.) are close behind. There is a high degree of discretion, for each of these expenses, regarding how much organizations should allocate to programs. For example, because employees rarely track their time by functional expense category at each payroll period and many make retrospective judgments at year-end about how they spent their time, the accuracy of such judgments is open to question (Wing and Hager 2004b). In other words, it is difficult to evaluate the appropriateness of the percentage of salaries and wages (and natural expense classifications associated with salaries and wages) allocated to programs, which provides an opportunity for nonprofits to use this expense line to manage ratios.
Further, research finds that, due to discretion afforded by ASC 958–720-45, nonprofits manage printing and publication costs connected to fundraising appeals (Jones and Roberts 2006). In fact, one could argue that managers have a lot of discretion when determining how much of an expense to allocate to programs for the top 10 ranked natural expense classifications. These expenses differ from the lowest-ranked natural expense classifications. As examples, Form 990 instructs NPOs to allocate most accounting and legal expenses to the administrative category and professional fundraising fees to the fundraising category. Our results suggest that NPOs are likely to manage program allocation rates for natural expense classifications where managers have more discretion regarding how much to allocate to programs.
Note that natural expense classifications most frequently manipulated on the old Form 990—i.e., printing and publications, supplies, postage and shipping, and telephone expenses—are now combined and included in office expenses on the new Form 990. And, because the new Form 990 combines these lines, it shifts the rankings of the other lines on the old form upward. In other words, if the new and old Form 990 lines were the same, salary expense, for example, would be ranked either seven or eight (and not 11) on the old Form 990. Thus, the natural expense classifications nonprofits likely use to manage cost allocations have been fairly consistent.
8 Conclusion
Various nonprofit stakeholders rely on financial reports to make decisions. To prevent suboptimal decision-making and the potential misallocation of resources, stakeholders require financial information to accurately communicate the organization's underlying economics. Accurate financial information, however, is not supplied when nonprofits intentionally report identical program ratios. Our findings reveal a red flag stakeholders can use to assess whether ratios have been potentially managed. This red flag is when an nonprofit reports the exact same ratio in multiple years while simultaneously reporting a large change in total spending. By focusing on NPOs that report identical ratios, we can isolate the expense lines NPOs most likely use to manipulate cost allocations.
Our sample includes larger and older NPOs that experience relatively large changes in total spending. The advantage of studying these NPOs is that we study relatively sophisticated organizations with stakeholders who care about cost allocation and with resources to track that allocation. The advantage of studying NPOs that experience relatively large changes in total spending is that we focus on instances where it is presumed an NPO has experienced a change in its operations (i.e., operations are evolving); thus a change in the reported program ratio is warranted. These advantages, however, also limit our study. Small nonprofits also report identical ratios and may do so intentionally or erroneously. We do not study these cases.
Further, we know nonprofits report identical ratios when changes in total spending are relatively small, and, regardless of the size of a change in total spending, it is difficult for a team of managers to control spending in ways such that the NPO could spend the exact same proportions on programs and overhead to report an identical ratio. Thus, NPOs with small changes in total spending and identical ratios may have managed cost allocations as well. We do not study these cases. Also, by studying mostly large NPOs that report relatively large changes in spending, we likely understate the use of the simpler (and potentially more prevalent) cost allocation techniques nonprofits use to manipulate cost allocations—that is, keeping allocation rates the same for all expense lines and changing the allocation rate for one expense line only.
This paper identifies one red flag that donors, auditors, researchers, and others can use to detect whether a nonprofit has potentially managed its reported financial information. There are other red flags, such as program ratios continuing to increase when circumstances suggest otherwise and nonprofits using the exact same ratio to allocate costs on every expense line. We call for future research to investigate these and others. Finally, this study and others in the literature identify characteristics of nonprofits likely to manage financial information. We also call for future research to determine the particular circumstances in which nonprofits engage in this behavior.
Acknowledgements
We thank Bill Baber, Patricia Derrick, Chris Jones, Adam Koch, Linda Parsons, Susan Perry-Williams, Christine Petrovits, and Kevin Rich for their insightful and helpful comments. We also thank participants at the 2016 GNP Conference, the 2016 Telfer Annual Conference on Accounting and Finance by the University of Ottawa, and the 2016 AAA Annual Conference.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.
Publisher's note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
