I. Introduction
The needs of vulnerable individuals in the United States fluctuate over the business cycle, with measures such as food insecurity, poverty, and homelessness rates increasing during bad times (Sard
2009; Kneebone and Holmes
2016; Lombe et al.
2018). The public at large expresses a desire for countercyclicality in US nonprofits, hoping that nonprofits will expand their activities during downturns in the face of rising need.
1 This hope motivates the rich literature on the drivers of giving to nonprofits. Yet there is little comprehensive evidence on whether nonprofits—particularly nonprofits that the public especially hopes will expand during bad times—are indeed countercyclical.
In this paper we establish a set of key facts about nonprofits in good and bad times that demonstrate that the public’s hopes are disappointed. The expenditure, revenue, and balance sheet size of US nonprofits are procyclical, declining rather than expanding during downturns at the national and local levels. We uncover procyclicality not only among all nonprofits but also among a select group of charities—such as food banks and housing assistance organizations—for which the public most intensely reports a desire for countercyclicality.
We build our analysis on micro data drawn from millions of tax returns of nonprofit organizations in the United States from 1990 to 2013—covering the near universe of nonprofits in the United States for all but the smallest organizations. While US nonprofits are exempt from taxation, Internal Revenue Service (IRS) guidelines generally require the filing of annual returns to maintain tax-exempt status. This legally mandated disclosure offers a useful window into financials across the distribution of nonprofit activity.
2 Crucially, the returns of tax-exempt organizations include information on revenue and expenditure, in addition to a wide range of data on the characteristics and type of each organization. This tax return database offers substantial advantages for the study of nonprofits, primarily through its measurement of a nonprofit’s full financial position and because of its impressive coverage of this large and growing sector of the economy. Indeed, from 2000 to 2013, nonprofits in our data grew from 5% to 8% of US businesses while their revenue grew from 10% to 13% of US GDP.
3Nonprofits vary widely in purpose and type, ranging from hospitals and universities to golf clubs and soup kitchens. The National Taxonomy of Exempt Entities (NTEE)—used by researchers and by the IRS for classification—includes over 600 detailed codes grouping nonprofits. While these classifications are informative for a range of purposes, NTEE codes themselves do not provide a direct mapping to the “type” of nonprofit that may provide a public safety net during bad times. For example, wineries and food banks are both subgroups of the “Food, Agriculture, and Nutrition” NTEE major group.
Given our desire to comprehensively examine whether nonprofits countercyclically expand during bad times—while still being able to narrow in on the aforementioned types of nonprofits—we build a novel classification scheme that organizes nonprofits according to the degree to which the public hopes they expand their programs and services during bad times. We recruited thousands of individuals to complete an online survey. In this survey, after respondents are presented with a randomly selected NTEE code and corresponding description, they indicate—on a scale from 1 (strongly disagree) to 7 (strongly agree)—whether they believe nonprofits with that detailed NTEE code should expand their programs and services during economic downturns. We then construct an average desired countercyclicality rating (DCR) for each NTEE code. Quite intuitively, the highest DCR nonprofits provide critical assistance such as food, housing, or medical care to indigent populations, while the lowest DCR nonprofits include organizations such as nonprofit golf clubs. Indeed, rather than food banks and wineries being grouped similarly because they both fall under the same NTEE major group, food banks secure the second-highest DCR whereas wineries are ranked 631 out of 655. Not only does the DCR measure allow us to distinguish between nonprofits within the same major groups; the DCR measure also allows us to link nonprofits with vastly different NTEE codes. The top ranked DCR—just above food banks (K31)—is emergency assistance (P60), defined as “organizations that provide food, clothing, household goods, cash, and other forms of short-term emergency assistance.” Our construction of the DCR measure uniquely positions our paper to provide comprehensive insight into the cyclicality of the types of nonprofits that the public hopes to see expand to provide more services during bad times.
We organize our empirical analysis around five main questions. We frame each question in a manner that sheds light on our key motivation. Do nonprofits weather adverse economic conditions as the public hopes, expanding during downturns? Or do they instead contract during bad times? We first ask whether nonprofits adjust their spending in bad times by increasing their expenditure (Q1) or by reallocating their expenditure toward core programs (Q2). We then investigate the sources and uses of nonprofit financial resources in bad times, asking whether their revenue increases (Q3), whether their assets decline (Q4), and whether their liabilities increase (Q5). Leveraging our comprehensive data on nonprofits, the resulting facts, that is, answers to these questions, describe cyclicality at nonprofits in the face of nationwide business cycles and local economic fluctuations. We emphasize at the outset that each of our facts is descriptive rather than causal in nature, with our analysis purposefully targeted toward the documentation of observed behavior.
In Q1 we ask whether nonprofit expenditure is countercyclical. To answer this question—as well as the following four questions—we employ cyclicality regressions measuring the elasticity of nonprofit outcomes to income at the national and local level. We find that the answer to Q1 is no. Instead of expanding during economic downturns as desired by the public, we instead observe procyclicality in nonprofit expenditure with an elasticity of around 0.5 to local income. Even the “high DCR” nonprofits—those with a DCR in the top decile—cut their expenditure during downturns with an elasticity of 0.4 to local income.
After documenting a reduction in nonprofit expenditure during downturns, in Q2 we ask whether nonprofits reallocate their (lower) expenditure during bad times. We are motivated by a debate in the nonprofit sector about the importance of spending on two categories: core programs and services versus overhead costs. Historically, there has been a push for nonprofits to spend little on overhead costs under the belief that high overhead costs are indicative of waste and not instrumental to achieving their mission. Under this belief, if nonprofit expenditure falls, it would be less harmful, perhaps even helpful, for such reductions to be disproportionately borne by lower administrative expenditure. We find little to no reallocation in the data. While there is some evidence that high-DCR nonprofits reallocate their expenditures toward core programs during national downturns, we do not find similar shifts for high-DCR nonprofits during local downturns, and we find no cyclical shifts for all nonprofits. Overall, the answer to Q2 is no: the share of spending on core programs and services is mostly acyclical and does not shift over the business cycle. We note, however, that our result need not be viewed as a “failure” of the nonprofit sector. Business leaders and academics have reasonably argued that, as detailed in Gregory and Howard (
2009), a focus on decreasing the share of spending on overhead costs can lead to a “nonprofit starvation cycle” in which charities lack the necessary talent or infrastructure to implement their goals.
4Motivated by evidence that nonprofit outlays or expenditure decline in bad times, we then move to an analysis of nonprofit financial resources. In Q3 we ask whether nonprofits secure higher revenue during bad times. We find that the answer to Q3 is no. Revenue for nonprofits is procyclical, declining during bad times and increasing during good times for both the nationwide and local economies. We estimate an elasticity of revenue to local income of 1.1 for all nonprofits and 0.7 for high-DCR nonprofits. We also document the procyclicality of various revenue streams, including donation-based revenue and non-donation-based revenue. While prior work similarly finds procyclicality of donation-based revenue (List
2011), our investigation of non-donation-based revenue—facilitated by tracking organization-level rather than donor-level outcomes—is informative since the average nonprofit in our data receives more than 80% of its revenue from nondonation sources including the sales of products (e.g., discounted clothes) and fees associated with services (e.g., job training or medical care).
Declining revenue during economic downturns motivates our next two questions on nonprofit finances or balance sheets: Do nonprofit assets decline during bad times (Q4)? Do nonprofit liabilities increase during bad times (Q5)? The answer to Q4 is yes: nonprofits do in fact experience declines in their assets during economic downturns with a cyclical elasticity of assets to local income of 0.5 for all nonprofits and the same for the high-DCR group. The answer to Q5 is no: nonprofit liabilities decline during bad times with a cyclical elasticity of liabilities to local income of 0.2 for all nonprofits and the same for the high-DCR group. Our findings of procyclicality for both assets and liabilities imply that the size of nonprofit balance sheets shrinks during economic downturns. These patterns are consistent with the idea that financial constraints may impact nonprofit decision making.
5Taken together, our empirical facts reveal that the public’s desire for countercyclicality is disappointed in practice, both for the nonprofit sector as a whole and even in the highest DCR nonprofits such as food banks or homeless shelters, which all fluctuate procyclically. The pronounced disconnect between hopes and empirical outcomes makes it important to conduct a set of comprehensive robustness checks, subsample analyses, and extensions of our baseline analysis. In a series of these checks we ask whether our findings differ by the exact DCR level, by size, by broad NTEE categories, by nonprofit legal structure, by census region, by local urbanization level, by revenue streams, by measure of economic fluctuations, and under alternative specifications of our cyclicality regressions. Motivated by a desire to understand differences between nonprofit and for-profit firms and the push for nonprofits to become more business-like (Hwang and Powell
2009; Bloom et al.
2015; Tsai et al.
2015; McConnell et al.
2016), we also compare nonprofit cyclicality to that of for-profit businesses. These exercises uncover some interesting heterogeneity across groups of nonprofits in the magnitude of their cyclicality. But in no case do we uncover evidence of the nonprofit countercyclicality desired by the public: procyclicality among US nonprofits is a robust phenomenon.
Our results complement the rich literature on charitable giving. See Vesterlund (
2006), List (
2011), Andreoni and Payne (
2013), and Gee and Meer (
2020) for excellent reviews of that work. Much of this literature focuses on the manner in which micro conditions influence individual giving decisions, for example, how donations are influenced by social pressure (Ariely, Bracha, and Meier
2009; DellaVigna, List, and Malmendier
2012; DellaVigna et al.
2013; Andreoni, Rao, and Trachtman
2016), by matching donations (Eckel and Grossman
2003; Karlan and List
2007; Meier
2007), by seed money or lead donors (List and Lucking-Reiley
2002; Karlan and List
2020), by household income (Randolph
1995; Auten, Sieg, and Clotfelter
2002; List
2011; Kessler, Milkman, and Zhang
2019; Meer and Priday
2020a), and by tax policy (Duquette
2016,
2019; Meer and Priday
2020b). A smaller set of studies focuses on the relationship between macro conditions and giving, such as papers relating to giving after large, tragic events (Lilley and Slonim
2016; Bergdoll et al.
2019) and work relating to redistribution and fairness views at the societal level (Almås, Cappelen, and Tungodden
2020).
6 An even smaller but important and emerging body of literature seeks to understand aggregate giving in response to nationwide economic fluctuations.
7 This existing body of research compellingly documents the procyclicality of giving in relation to nationwide economic fluctuations (List
2011; Reich and Wimer
2012; Meer, Miller, and Wulfsberg
2017) and includes evidence that such procyclicality is smoothed during recessions (List and Peysakhovich
2011). Relative to this prior work on procyclicality, we differ by focusing on the behavior of nonprofits themselves rather than individuals giving to nonprofits. Not only does this approach allow us to investigate the cyclicality of total nonprofit revenue (combining both donations and nondonation sources); our data allow us to investigate other nonprofit outcomes such as the cyclicality of nonprofit expenditure, which we view as useful to understanding whether nonprofits expand their programs and services during bad times. For all of our cyclicality analyses—on expenditure, program expenditure share, revenue and its subcomponents, assets, and liabilities—we start by investigating nationwide economic fluctuations, like this prior literature on the cyclicality of donations. But, in addition, we further examine local economic fluctuations. These local economic fluctuations are larger in magnitude than nationwide cycles, providing substantially more variation in our data and crucially linking our notion of “good versus bad times” more closely to the lived experiences of individuals in a given local area.
Our results also complement a nonprofits literature that uses nonprofit tax return data similar to ours or relies on related surveys. One stream (Froelich
1999; Carroll and Stater
2009; Duquette
2017) studies the sources of nonprofit revenue and emphasizes differences between volatile revenue sources, such as contributions, versus other sources of income. A second stream of work (Tuckman and Chang
1991; Greenlee and Trussel
2000; Trussel
2002) studies nonprofit financial vulnerability, finding wide heterogeneity across organizations in their ability to withstand financial shocks. Most closely related to our paper, a third stream of work (Salamon, Geller, and Spence
2009; Brown et al.
2014; Lin and Wang
2015) directly examines fluctuations in nonprofit outcomes during the Great Recession and often focuses on a small sample or survey of nonprofits. This work typically finds meaningful declines in revenue and financial resources during the Great Recession. Relative to this prior work, we broaden the scope beyond the nationwide Great Recession by studying a longer time period with multiple economic cycles, by analyzing local rather than only nationwide economic fluctuations, and by studying a comprehensive sample of the near universe of tax returns across nonprofits.
We also highlight that this prior work on nonprofits often focuses on a small subset of organizations precisely because those studies correctly recognize the diversity of nonprofits of different types. This diversity—and indeed the associated challenge in classifying whether nonprofits are more “like food banks,” which the public might hope will expand during bad times, or more “like wineries”—is exactly what motivates our construction of the novel DCR measure. Our constructed DCR measure is publicly available online.
8 We hope this ranking will prove to be a useful methodological resource for research on nonprofits. Growing a literature in economics on nonprofits themselves—to complement the existing, rich but distinct literature that focuses instead on giving to nonprofits—would substantially improve our understanding of this large and diverse sector of the economy.
Finally, our results complement work on for-profit firms. One stream of papers documents the relative cyclicality of sales at small versus large firms (Gertler and Gilchrist
1994; Crouzet and Mehrotra
2020). Another set of studies analyzes firms in recessions (Kehrig
2015; Moreira
2017; Bloom et al.
2018). A third body of work measures overall volatility and trends (Davis et al.
2006; Decker et al.
2014,
2020). A fourth line of research covers local responses to policy (Nakamura and Steinsson
2014). The contribution of our paper is to extend such analysis to the large, qualitatively distinct context of nonprofit organizations.
Section II describes our data.
Section III presents our results.
Section IV reports our heterogeneity and robustness checks.
Section V concludes. Online appendices provide more detail on our data, additional results, and our DCR survey.
V. Conclusion
Our survey evidence reveals a public hope that US nonprofits, especially those assisting with critical needs such as food or housing, will expand during economic downturns. Using data from millions of nonprofit tax returns, we lay out a series of facts about nonprofit behavior in the face of nationwide and local economic fluctuations. We find that—far from increasing their scope as the public hopes—nonprofits exhibit robust procyclicality with their expenditure, revenue, assets, and liabilities declining in bad times.
By providing descriptive facts on nonprofit outcomes in good times and bad times, this paper seeks to improve our understanding of the nonprofit sector and to motivate further work on it. In light of our descriptive facts, at least two avenues for further research into the nonprofit sector suggest themselves. As is often the case following the establishment of descriptive facts, these avenues are motivated by a desire to narrow in on a particular mechanism or a particular counterfactual question that requires different analyses than in this paper, for example, causal identification or structural modeling.
First, many questions remain open around the counterfactual impact of policy on the nonprofit sector. The existing level of government subsidies to nonprofits during economic downturns does not prevent their revenue from declining, because our revenue measure includes government grants. But one might reasonably speculate that—and future work could investigate through explicitly causal analysis whether—government subsidies to nonprofits during bad times might cause nonprofits to expand their services or to, at least, contract less.
Second, we note that cuts to nonprofit expenditure during bad times could in principle stem from multiple sources including but not limited to manager preferences or financial constraints. As one example, if the leaders of charities were biased on average toward organizational survival rather than maintenance of service provision, or if managers were averse to expansion, then the natural implication would be a failure of charities to expand during times of increased need. In the for-profit sector, evidence exists that such motives might be widespread (Bertrand and Mullainathan
2003; Hurst and Pugsley
2011). As another example, nonprofit managers may hold beliefs that are not accurate about their optimal strategies during bad times. See, for example, inaccurate beliefs about fund-raising strategies documented in Samek and Longfield (
2019).
30 Exploring the extent to which these alternative explanations contribute to the procyclicality of nonprofit expenditure is a natural avenue for future work.
Third, we note that the literature on for-profit businesses documents a decline in measures of net entry, churn, and dynamism in the United States in recent decades (Decker et al.
2014). Although our work in this project does not explicitly focus on the entry and exit of nonprofits, future empirical work might profitably investigate the impact of competition in the nonprofit sector (Harrison and Thornton
2014; Gayle, Harrison, and Thornton
2017).
