The Impact of Revenue Diversification on Nonprofit Financial Health: A Meta-analysis

Abstract

This study reviews the influence of revenue stream diversification on financial health. It is a meta-analysis of previous studies that have studied the relationship. This literature variously demonstrates that nonprofit financial health is improved, not influenced or harmed by diversifying reliance on different revenue streams. Our analysis of 40 original studies reporting 296 statistical effects demonstrates a small, positive, yet statistically significant association between revenue diversification and nonprofit financial health. In addition, we show that granularity of measurement of revenue diversification influences effect size, that this effect has shifted over time, and that studies on U.S. nonprofits demonstrate weaker (or more negative) effects. However, few other prominent suspects, including diversity of financial health measure or methodology choices, explain variations in effects across the literature on revenue diversification. Overall, the study supports the contention that both analysts and practitioners should make strategic considerations that have generally escaped scholarship on revenue diversification or shift attention to revenue optimization considerations that have been raised by portfolio theory.

Keywords

revenue diversification nonprofit financial health meta-analysis modern portfolio theory

The logic is appealing and easily communicated and digested by scholars and practitioners alike: Nonprofits should diversify revenue sources to hedge against uncertainty. If a nonprofit relies exclusively on grants, it might wish it had pursued an additional earned income strategy when its funder priorities change. More revenue sources provide flexibility. Reliance on a limited number of sources leaves them vulnerable. A widely-cited formulation by Tuckman and Chang (1991) defines a financially healthy nonprofit organization as one with adequate equity balances, sufficient administrative allocations, positive operating margins, and diversified revenue sources, all of which provide slack flexibility in the face of financial downturns. That is, service continuity can be achieved by effective financial management and sound financial capacity, which ensure a nonprofit organization has a greater flexibility to withstand financial shocks. A financially healthy nonprofit organization is more likely to weather financial difficulties and, most importantly, fulfill its mission (Carroll & Stater, 2009; Gronbjerg, 1992; Trussel, 2002; Trussel & Greenlee, 2004; Tuckman & Chang, 1991).

Due to the ubiquity of this argument, revenue diversification is a central feature of many financial health and sustainability models. In that seminal paper, Tuckman and Chang (1991) argued that nonprofit organizations that cultivate equal amounts of revenues from disparate sources are less vulnerable than those that derive all revenues from a single type of source. A decline in one revenue source can be offset by increases in other revenue sources. For example, a decrease in government contracts to a social service organization might be offset by an increase in donations from a private foundation. In short, revenue diversification is regarded as a cushion strategy for nonprofit organizations to battle against financial instability and uncertainty (Kingma, 1993; Tuckman & Chang, 1991).

However, empirical studies on the relationship between revenue diversification and financial health have produced mixed results. Some indeed find that nonprofit organizations increase financial health by diversifying revenue portfolios (Carroll & Stater, 2009; Greenlee & Trussel, 2000; Hager, 2001; Trussel, 2002; Wicker & Breuer, 2014; Wicker, Longley, & Breuer, 2015). More recent studies tend to contest the relationship (Chikoto & Neely, 2014; Chikoto-Schultz & Neely, 2016; de Andrés-Alonso, Garcia-Rodriguez, & Romero-Merino, 2015; de los Mozos, Duarte, & Ruiz, 2016; Lin & Wang, 2016; Prentice, 2016a; Wicker & Breuer, 2013). Nuanced treatments regard revenue diversification as a double-edged sword (Froelich, 1999): Higher reward comes with higher risk (Frumkin & Keating, 2011). The competing arguments based on discrete empirical findings have not provided a definitive conclusion regarding whether nonprofit managers can enhance financial health of their organizations by cultivating multiple revenue sources. A synthesis that evaluates the relationship across the empirical literature is needed.

The study reported in this article aims to examine the effect of revenue diversification on nonprofit financial health through meta-analysis. It contributes to the literature by summarizing the results of foregoing empirical studies, comparing reported magnitude and direction of the statistical “effects” between revenue diversification and financial health. Our key question, however, is why the influence of revenue diversification—positive, null, or negative—varies across studies. So, we dedicate substantial focus to an evaluation of how effects might be influenced by various “moderators,” or differences between studies or models. We begin with a general survey of the literature that advances competing explanations of the relationship between revenue diversification and financial health. We then sketch the process of conducting the meta-analysis, including methodological strategy, study selection criteria, and selection of moderator variables that we believe might influence the relationship. After presentation of findings, we conclude with a summary and discussion of results.

Balancing Revenue Streams: Pros and Cons

Portfolio theory seeks to provide a lens through which to consider the optimal mix of revenue streams (Mayer, Wang, Egginton, & Flint, 2014). For the broad purposes of this article, modern portfolio theory is valuable to the extent that it considers the complex interaction of revenue streams (Kingma, 1993; Qu, 2016) and management decisions (Jegers, 1997), rather than considering single sources in isolation. Other perspectives have embraced this complexity as well, resulting in a variety of expectations regarding revenue diversification, discussed variously in the empirical literature. The sections below outline arguments both for and against a revenue diversification strategy.

Theoretical Advantages of Revenue Diversification

Flexibility

As described in our introduction, a common argument in the nonprofit scholarly literature is that revenue diversification provides options to organizations that suffer swings in dedicated revenue streams. Carroll and Stater (2009) note that these swings can be idiosyncratic (such as an unhappy major donor) or result from environmental shifts (such as a global financial crisis). Diversifying revenue portfolios is said to reduce financial risks and increase organizational stability: A decline in one revenue source might be offset by increases in other revenue sources (Bingham & Walters, 2013; Chang & Tuckman, 1994). Conceptually, a nonprofit organization with equal revenues from several sources might be said to be healthier (or at least safer) than those with revenue from a single source due to the flexibility that affords it greater potential to weather shocks.

Autonomy

The obverse of flexibility is vulnerability: A nonprofit organization might experience financial difficulties if it becomes overly dependent on any single revenue source (Carroll & Stater, 2009). Diversifying revenue streams is said to reduce resource dependence and enhance nonprofit autonomy (Froelich, 1999; Mitchell, 2014), which in turn enables nonprofits to make better use of their resources and carry out missions (Chang & Tuckman, 1994; Pfeffer & Salancik, 1978). For example, a nonprofit organization that depends heavily on government funding might experience delays in the receipt of funding or lose a contract entirely. In the donor case, major contributors might exert control over organizations by restricting donations to very specific purposes. Nonprofit organizations with these resource constraints may have no choice but to follow the instructions from donors. This loss of autonomy threatens to diminish financial capacity and sustainability and may have a negative effect on efficiency in providing program services (Froelich, 1999).

Income and growth potential

Revenue streams available to nonprofits represent opportunities for growth. Enterprising nonprofit organizations are sometimes drawn to pursue new types of income streams for their potential to enhance overall organizational revenue. For example, when efforts to raise contributions are maximized, boards might encourage staff to wade into earned income strategies to fund new programs and staff. If successful, these forays might foster growth or greater penetration of mission.

Community connection and embeddedness

Network connectedness provides greater penetration of nonprofits into their communities, expanding their mission reach and increasing survival chances (Hager, Galaskiewicz, & Larson, 2004). The effort that nonprofits take to diversify their revenue streams provides exposure to new audiences. For example, an organization well known to local grantmakers might gain unique connections, greater embeddedness, and increased reputation in their community through the development of a new social enterprise. Diversification of revenue sources might serve to stabilize revenues in a financial downturn, but increased points of community connection might provide additional protections that allow an organization to weather the storm.

Theoretical Disadvantages of Revenue Diversification

Increased complexity and risk

Portfolio theory warns that any additional source of revenue gained by an organization comes with risk of losing other revenue streams or organizational character. Grasse, Whaley, and Ihrke (2016) suggest that nonprofit organizations can maximize revenue growth by mixing revenue streams, but only at a level of risk they are comfortable with. The difficulty lies in the fact that different revenue streams come with different risks, and nonprofit managers lack the information required to evaluate the risk associated with each revenue stream. This complexity compromises the ability to reliably achieve an optimal balance that might maximize any advantage of pursuing greater numbers of streams.

Increased administrative costs

Pursuit of new income streams might require new management systems and expertise, both of which incur costs of money and time (Frumkin & Keating, 2011; Kingma, 1993; Wicker & Breuer, 2013). This investment threatens to substitute one form of uncertainty for another. Nonprofit organizations that rely on sole-source revenue spend less on administrative and fundraising costs (ceteris paribus), which indirectly produces administrative efficiency (Foster & Fine, 2007) and might help keep faith with donors under sway of the overhead myth. For example, nonprofit managers with professional fundraising skills developed from the process of relationship building with a few certain funders are able to concentrate on this core competency (de los Mozos et al., 2016; Frumkin & Keating, 2011), thereby maximizing efficiency. Similarly, obtaining government funding requires a high level of time and human resource commitment, which invariably increases the administrative costs of nonprofit organizations that might already have reliable or stable revenue sources (Gronbjerg, 1992).

Crowd out of private donations

Receipt of support from singular sources, primarily government, can “crowd out” funding from other sources in at least two different ways. One is that donors might question whether nonprofits need other sources of revenue, especially, when grants or contracts are seen as long term or permanent. When government or grantmaking foundations provide seed money to nonprofit organizations, the imprimatur of the gift might stimulate private donations. However, major long-term support threatens to chase away donors instead. In his investigation of American symphony orchestras, Brooks (2000) found that whether public subsidies substitute private donations depends on their magnitude relative to one another. Public subsidies might crowd in private donations at low levels of government support; however, the effect is opposite at high levels of government support.

Major donors can also crowd out the attention that nonprofits themselves pay to other revenue streams. Andreoni and Payne (2011) explored why public subsidies crowd out private donations and found that the crowding out effect is due primarily to reduced efforts of fundraising after receiving public subsidies. Instead of improving financial health, diversifying revenue streams might have a negative impact on nonprofit financial health in the way that nonprofit organizations end up putting less effort into fundraising and heavily relying on government contracts or other major singular sources of income.

Mission drift

Kearns, Bell, Deem and McShane (2014) warn that nonprofit organizations are in danger of losing reputation and financial support in the long run when they undertake tasks seen as inconsistent with their missions. Despite any value that might come from increased community embeddedness, heterogeneous sources of revenue might shift nonprofits’ attention to funders rather than their clients or members (de los Mozos et al., 2016; Froelich, 1999). Different revenue sources carry with them different expectations (Wilsker & Young, 2010), and organizations sometimes compromise their image or mission direction in exchange for contributions or contracts (Gronbjerg, 1993). This problem is amplified when nonprofit organizations compete for scarce resources in pursuit of financial stability (Cooley & Ron, 2002). Mission drift occurs when nonprofit organizations manipulate their program offerings to meet donor expectations rather than beneficiary needs (Froelich, 1999; Mitchell, 2014; Oster, 1995).

In short, revenue diversification might have a positive effect on nonprofit financial health, especially, in the short run. However, to the extent that diversification complicates organizational processes and diverts attention from missions, the strategy might have a negative effect on financial health in the long run. We might not be surprised that empirical studies demonstrate mixed results on the association. The questions that motivate our study are what generalizations we can draw from the accumulated competing research on revenue diversification and whether differences in studies explain the differences in results. However, before turning to these questions, we first outline our conception of financial health, the outcome under study.

Financial Health

The outcome of interest in this meta-analysis is the financial health of nonprofit organizations, a multifaced conception that scholars have operationalized in a variety of ways. Accumulation of research on financial health for meta-analysis is problematic (and perhaps even controversial) because of this variety, which raises questions of the commensurability of studies that purport to study similar phenomena (Prentice, 2016b). For example, a primary source of model effects for our meta-analysis is Chang & Tuckman’s (1994) paper, which reports zero-order correlations between revenue diversification and fundraising spending (Table 3; for 26 separate National Taxonomy of Exempt Entities (NTEE) subsectors, plus overall), gross assets (Table 4; same 27 correlations), and operating margin (Table 5; same 27 correlations). Some observers might characterize all three outcome measures as proxies for financial health, but our conception of financial health leads us to consider revenue diversification correlations with gross assets (Chang and Tuckman (1994): “The . . . test involves non-profit gross assets as . . . a measure of financial strength” (p. 285) and operating margin (a traditional measure of financial flexibility) as two exemplifications of financial capacity measures (among others). Accordingly, these 52 subsector correlations and two overall sector correlations constitute 54 of the 296 “effects” included in our meta-analysis. In contrast, we do not take Chang and Tuckman’s fundraising spending as a prima facie measure of financial health and do not include it in the present study because Chang and Tuckman do not defend it that way.

Such a determination requires rules (and some judgment calls) for which measures in the revenue diversification literature constitute financial health and which do not. Bowman (2011a) divides financial health into two broad camps: financial capacity and financial sustainability. We include both these conceptions in our measure of overall measure of financial health.¹ Financial capacity is described as resources that organizations obtain and maintain to seize opportunities and weather unexpected crises. These measures are static representations of the resources available to organizations at the time that the value of revenue diversification is considered. Measures of financial capacity include assets (e.g., de Andrés-Alonso et al., 2015; Trussel, 2002), operating margin (e.g., Chang & Tuckman, 1994), revenues (e.g., Mayer et al., 2014), program expenses (e.g., de Andrés-Alonso et al., 2015), and even survival (e.g., Hager, 2001).

Financial sustainability is regarded as dynamic change in financial capacity. That is, whereas financial capacity indicates the level of a financial indicator, financial sustainability represents its fluctuation over time (Wicker et al., 2015). The fluctuation of the financial indicators used as proxies of financial sustainability include the difference between actual revenue and expected revenue volatility (e.g., Carroll & Stater, 2009), percentage growth in total revenue (e.g., Chikoto & Neely, 2014), variance of revenues (e.g., Mayer et al., 2014; Wicker et al., 2015), and change in months of liquidity (e.g., Lam & McDougle, 2016). These related conceptions of financial capacity and sustainability guided our search for literature and relevant outcome models of revenue diversification as it relates to financial health.

Method

We employ a standard meta-analysis approach (Ringquist, 2013) to summarize and compare results from empirical studies regarding the impact of revenue diversification on nonprofit financial health. Meta-analysis is “the statistical analysis of a large collection of analysis results from individual studies for the purpose of integrating the findings” (Glass, 1976, p. 3). This methodology is based on the “effect sizes” of the statistical relationships reported in the literature. A primary task of meta-analysis is to harvest these effects, usually correlation coefficients in the fields of nonprofit and public management. First, we identified relevant studies and models in these studies whose outcome measures conform to our definition of financial health. Second, we gathered the effects (each statistical measure of the association between revenue diversification and financial health) reported in these studies. Third, we compared the effect sizes across the studies to estimate an overall effect. Fourth, we coded study characteristics, so we could investigate the moderating influence of study characteristics on the overall effect, as outlined by Card (2012). Sample selection, measure specification, and moderator choices are the defining characteristics of any meta-analysis.

Sample Selection

Meta-analysis is regarded as a form of literature review, which confronts the question of which studies to include in the review. Lipsey and Wilson (2001), Ringquist (2013), and Card (2012) provide guidance on the literature searches and reality checks we employed in sample selection. First, we searched EBSCO, Social Sciences Citation Index (SSCI), and ProQuest databases for relevant studies (last search performed January 17, 2017). We used the following search strings: (nonprofit) AND (financial health OR financial vulnerability OR financial stability OR financial sustainability) AND (revenue diversification OR revenue concentration OR funding source). This search yielded 45 candidate studies. Second, although EBSCO and ProQuest include unpublished studies (e.g., dissertations), we searched programs of relevant academic conferences (e.g., ARNOVA and International Society for Third-Sector Research [ISTR]) and archives of working papers (e.g., Social Science Research Network [SSRN]) for relevant unpublished studies (last search performed on January 20, 2017). The searches yielded 30 candidate studies. For conference presentations not archived online, we contacted three authors for the papers or presentations; two responded to our request, yielding two more candidate studies. Third, we performed backward and forward searches to identify literature that previous steps might have missed. Backward searching involves reviewing bibliographies of candidate studies for references to other potential candidate studies. Forward searching utilizes Google Scholar (in our case) to search for later studies that cite the candidate study. These searches yielded 71 additional candidate studies. Fourth, we conducted reality checking: first, a skim of online tables of contents of Nonprofit and Voluntary Sector Quarterly (NVAQ), Nonprofit Management & Leadership, Public Administration Review, and Voluntas, and second, a Google scholar search of (revenue diversification OR revenue concentration). One additional candidate study was uncovered through the reality checking, and one more was suggested by an NVSQ reviewer. In total, our searches and checks uncovered 150 candidate studies, effective March 18, 2017.

Most candidate studies did not qualify for our meta-analysis. The decision of which original studies should be included is based on the criteria suggested by Lipsey and Wilson (2001), Ringquist (2013), and Card (2012). We included both published and unpublished studies that sampled in and across all nations. We culled studies that did not document a statistical relationship between revenue type diversification (or concentration) and financial health. We removed candidate studies not written in English. We removed studies or excluded models where we did not have sufficient information to record or compute effect sizes. The 28 published and 12 unpublished studies included in our meta-analysis are listed in Table 1.

Table 1.

Studies (40) and Effects (296).

Study	Number of effects	Average effect size	Weight (%) from random effects analysis
Allen, Henley, and Doss (2014)	1	.00	0.11
Baba, Ishida, and Okuyama (2014)	2	−.04	0.89
Carroll and Stater (2009)	1	.01	0.48
Chang and Tuckman (1994)	54	.06	21.66
Chikoto and Neely (2014)	9	−.02	4.26
Chikoto, Ling, and Neely (2016)	32	−.02	14.82
Chikoto-Schultz and Neely (2016)	1	.01	0.47
Cordery, Sim, and Baskerville (2013)	2	.14	0.24
Daniel and Kim (2018)	6	.03	0.96
de Andrés-Alonso, Garcia-Rodriguez, and Romero-Merino (2015)	4	−.10	0.78
de los Mozos, Duarte, and Ruiz (2016)	18	−.10	3.95
Despard, Nafziger-Mayegun, Adjabeng, and Ansong (2017)	3	.09	0.42
Faulk (2010)	6	−.05	2.41
Frumkin and Keating (2011)	84	−.01	25.70
Gordon, Fischer, Greenlee, and Keating (2013)	5	−.02	2.26
Greenlee and Trussel (2000)	1	.05	0.43
Hager (2001)	8	.07	2.69
Harvey (2012)	1	.01	0.33
Keating, Fischer, Gordon, and Greenlee (2005)	4	.04	1.91
Lam and McDougle (2016)	7	.06	1.34
Lin and Wang (2016)	3	−.07	0.87
Matthiesen (2009)	1	.17	0.09
Mayer, Wang, Egginton, and Flint (2014)	2	.02	0.95
Merk (2014)	3	−.06	0.79
Myser (2016)	1	−.03	0.46
Prentice (2016a)	2	.01	0.93
Searing (2015)	4	.18	1.32
Silva and Burger (2015)	2	−.02	0.29
Spyker and Deol (2014)	1	−.01	0.47
Tevel, Katz, and Brock (2015)	1	.40	0.05
Trussel (2002)	1	.06	0.47
Trussel and Greenlee (2004)	2	.01	0.88
von Schnurbein and Fritz (2017)	1	−.15	0.17
Wang and Liu (2013)	1	.22	0.07
Webb (2015)	6	.00	2.04
Wicker and Breuer (2013)	1	−.06	0.47
Wicker and Breuer (2014)	4	.08	1.47
Wicker, Longley, and Breuer (2015)	6	.18	1.21
Yan, Denison, and Butler (2009)	2	.22	0.76
Zhai (2012)	3	.08	0.12
Total (overall)	296	.008	100

Measure Specification

As prescribed in Ringquist’s (2013) guide to meta-analysis, we use Pearson’s correlation coefficient (r) as the index of effect size for combining and comparing measures (revenue diversification and financial health) across studies. For those studies that did not report r, we followed strategies for calculating the effect sizes suggested by Card (2012) and Ringquist (2013, pp. 105-109):

We used t statistics to calculate r effect sizes for original studies that reported partial correlation coefficients and standard errors associated with regression parameter estimates in regression models (e.g., Chikoto & Neely, 2014; Trussel & Greenlee, 2004).

We used corresponding t statistics to compute r when original studies reported a level of statistically significant association between financial health and revenue concentration, but not standard errors (e.g., Gordon, Fischer, Greenlee, & Keating, 2013). These are low-bound effect estimates, which mean the actual r could be larger.

We record r as zero when original studies report no statistically significant relationship between financial health and revenue diversification but do not report standard errors (e.g., Daniel & Kim, 2018; Prentice, 2016a). They are low-bound effect estimates as well.

We use Wald statistics to calculate r for original studies that test hypotheses using the Wald test and report Wald test results (e.g., Cordery, Sim, & Baskerville, 2013).

We regard standardized regression coefficient estimates as r if only standardized β is reported in an original study (e.g., Wicker & Breuer, 2013).

This procedure yielded 296 effect magnitudes (and directions) from the 40 studies. Following Card (2012) and Ringquist (2013), we transformed each r into Fisher’s z, ranging from z = 0.490 (in Chang & Tuckman, 1994) to z = −0.485 (in Frumkin & Keating, 2011). Of the 296 z scores, 142 indicate positive association, 31 indicate no association, and 123 indicate negative association.

Moderator Choices

What explains this variation? Studies differ in terms of data quality, measurement and analysis choices, subsector focus, and definition of financial health, potentially compromising their comparability. Consideration of such differences across studies provide the opportunity to moderate our global conclusion regarding the relationship between revenue diversification and financial health across these 40 studies. We investigate a series of study or model characteristics that might systematically predict variation in the effect sizes within and across studies. That is, we explore whether the effect sizes varied as a function of a variety of prospective moderators. Such moderators are not drawn from theory, per se: Card (2012) asserts that moderator choices “need to be heavily informed by your knowledge of the content area in which you are performing a meta-analytic review” (p. 64) and provides a list (Table 4.1, p. 66) of aspects to consider, including sample, measurement, design, and source characteristics, as well as study quality. We consider the following 14 moderators in our study.

Nature of financial health measure

Our foregoing discussion of financial health foreshadows our calculation of this conceptually vital moderator variable. Following Bowman (2011a, 2011b) and as detailed above, we differentiate models that conceive of financial health as a static measure of current organizational capacity (e.g., assets, operating margin, revenues, program expenses) versus those that conceive of financial health as a dynamic measure of organizational sustainability (e.g., revenue volatility, growth in revenue over time, change in months of liquidity). This distinction provides some basis for evaluating Prentice’s (2016b) conclusion that interchanging dissimilar measures of financial performance carries with it a hazard of drawing faulty conclusions. Does the influence of revenue diversification vary according to how analysts measure financial health? The financial health measurement moderator is coded 1 for studies that use a capacity measure of financial health and 0 for studies that use a sustainability measure of financial health.

Number of revenue streams in diversification measure

Most studies in our review use a Herfindahl–Hirschman Index (HHI) to estimate revenue diversification. Qu (2016) notes this as a point of divergence from modern portfolio theory as this crude measure of diversification (the HHI) does not account for how revenue sources move in relation to each other and ignores the influence of expenses required to generate various revenue streams. Nonetheless, use of the HHI is ubiquitous. Some studies include more streams than others, which may influence the sensitivity of measurement of financial health (Chikoto, Ling, & Neely, 2016; Kim, 2014). That is, variation in effect sizes might vary with the different number of revenue streams used in the calculation. Many studies use three revenue streams to calculate the index: donative support, earned income, and investment revenue. However, Hager (2001) used five and Wicker et al. (2015) used 21 revenue streams to constructing their HHI indices of revenue diversification. For this moderator variable, effects generated from an HHI calculated from three revenue streams are coded as 1, whereas effects generated from an HHI with more than three revenue streams (a “finer grain”) are coded as 0.²

Bivariate versus multivariate effects

When studying the value of revenue diversification, analysts may choose to control for any number of forces theorized to influence financial health. Above, we note the large number of effects contributed by Chang and Tuckman’s (1994) tables of simple bivariate correlations, where no controls are employed. Critics contend that the magnitude and direction of these associations may be artifacts of other organizational characteristics and forces, whose marginal influences can be measured and controlled by multivariate models. For example, some measure of an organization’s size is frequently controlled in studies of organizational processes. Small nonprofit organizations might have less financial, human, and information management resources needed to achieve social mission (de los Mozos et al., 2016; Tuckman & Chang, 1991), and smaller nonprofit organizations are less likely to survive than other nonprofit organizations (Searing, 2015). Organization age is another common control as the life stage of an organization might contribute to effect size heterogeneity (Chikoto-Schultz & Neely, 2016; Hager, Galaskiewicz, Bielefeld, & Pins, 1996). The liability of newness thesis holds that young nonprofit organizations lack resources needed to develop in their early stage and are more likely to fail (Singh, Tucker, & House, 1986). Other controls, often other financial measures, are common. Our moderator compares effects derived from bivariate associations (coded 0) versus those that derive their effects from multivariate models (coded 1).

Adjustment for unobserved heterogeneity

Even the most complete multivariate models are unlikely to account for important influences on financial health, resulting in endogeneity (where a measure of revenue diversification is correlated with a model’s error term). Those important influences, some of which are described in the review above, might be excluded because they are not measured (or measurable) in a given study. A popular method of accounting for these unobserved influences is the use of a fixed effects regression model. Carroll and Stater (2009) use the fixed model in their study to incorporate the assumption that error terms are correlated with the model’s independent variables. Daniel and Kim (2018) characterize the value of the fixed effect model in capturing “time-invariant heterogeneity within organizations” (p. 961). To evaluate whether use of fixed effect models moderates the relationship between revenue diversification and financial health, we code our dummy variable 1 for values generated from fixed effect models and 0 otherwise.

Longitudinal data

Variation in effect sizes across the original studies might derive from the data structure of the original studies as well (Lu, 2016; Ringquist, 2013). Data in our subject studies are either cross-sectional or longitudinal. Cross-sectional data are measured at one point in time, whereas longitudinal data are collected more than once for the same organizations over time. Longitudinal data provide the means to track revenue diversification over time and may be superior for measuring inherently dynamic processes, such as financial health. Most importantly, parameter estimates from cross-sectional and longitudinal data use different variance components (Ringquist, 2013), with longitudinal data deemed superior to cross-sectional data. For our moderator, effects derived from longitudinal data are coded 1, and effects derived from cross-sectional data are coded 0.

Publication bias

Moderator tests for publication bias are common in meta-analyses (Card, 2012; Ringquist, 2013). A recent example in the field of nonprofit and philanthropic studies is Lu’s (2016) analysis of the influence of government grants on private donations, which considers the difference in effects between studies published in academic journals and unpublished papers or theses. Studies may not be published because they report null effects or because they are not aligned with reviewer expectations, resulting in inflated effect sizes in the published literature. A way to evaluate bias is to include a moderator that tests whether the effect sizes in the published studies are different from those in the unpublished ones. In his study, Lu (2016) reported no effect of publication bias on the model using clustered robust standard errors, but a small significant effect on the model using generalized estimating equations, one he characterizes as “not . . . seriously contaminated by publication bias” (p. 392). We implement the same test in our meta-analysis. Our moderator is the same as that of Lu, coded as 1 for published studies and 0 for unpublished ones.

Publication era

We observe in our introductory remarks that early studies tended to support a positive relationship between revenue diversification and financial health, whereas more recent studies tend to challenge this finding. Frumkin and Keating’s (2011) negative findings seem to be a marking point, with later studies more likely (or more willing) to report a null or negative relationship. To test for differences between these two eras, we code a publication era moderator as 1 for studies published in or after 2011, and 0 for studies published before 2011.

United States versus the world

We include a moderator to test for differences between studies based on nonprofits in the United States versus nonprofits studied in other parts of the world. Most of the effects in our study (265) reflect results on U.S. nonprofits. Our moderator (1 for United States effects, 0 otherwise) tests whether effect sizes differ according to this geographic difference.

Subsectors (6)

Nonprofit organizations operate in a variety of fields or “subsectors,” and studies of revenue diversification differ in their breadth and subsector focus. Wilsker and Young (2010) offer a wrinkle that is not often appreciated in the revenue diversification literature: Revenues streams are frequently tied to particular programs. For example, by their nature, hospitals are more reliant on commercial income streams. Professional associations rely on dues and contributions almost exclusively from their membership. We assert that these natures fundamentally influence the ability to diversify into new revenue streams. Diversity of natures is only crudely captured by subsector, but Wilsker and Young’s application of benefits theory at least supports an expectation that the potential for revenue diversification might differ across types of nonprofit. Chang and Tuckman (1994) report a range of correlations between revenue diversification and nonprofit financial health (operating margin) across the breadth of the sector, from .45 (for mutual benefit nonprofits) to –.19 (for housing nonprofits).³ Other studies focus on a more limited range of types, or even single subsectors: Wicker and Breuer (2013, 2014) focus on sports clubs, which we categorize, per the National Center for Charitable Statistics, under the major category of human service. Faulk (2010) studies theaters. We created subsector moderator variables when (a) at least four studies report effects for a particular subsector and (b) we had at least 10 total effects to represent that subsector. This resulted in six subsector moderators: arts, culture, and humanities; education; health; human services; international and foreign affairs; and religion-related. Each subsector moderator is coded 1 when the effect represents results for a particular subsector, and 0 otherwise.

Analysis and Results

Following Ringquist (2013), we conduct a Q test and calculate an I² statistic to identify the correct framework for computing the average effect size. The Q test indicates whether variation among effect sizes can be explained solely by sampling error. We rejected the null hypothesis that the study effects are homogeneous (Q = 10,875.6; p < .001). However, Cochran’s Q statistic has two shortcomings. First, it has low power to detect effect size heterogeneity when the number of studies is small, and excessive power when the number of studies is large. Second, it does not report the degree of effect size heterogeneity (Huedo-Medina, Sánchez-Meca, Marín-Martínez, & Botella, 2006). Therefore, we calculated the I² statistic to obtain the magnitude of the random-effects variance. An I² value of 97.29% indicates that 97.29% of total variability is explained by heterogeneity among the studies and their reported effects (Higgins & Thompson, 2002). With heterogeneity established, we selected a random-effects model to represent the overall (global) effect. The mean effect size was .008 (z = 2.028, p < .05), with a 95% confidence interval of [.0005, .0154]. This result indicates a statistically significant positive overall relationship between revenue diversification and nonprofit financial health. However, the magnitude of the relationship is relatively small, a statistically significant association that may not be substantial enough to guide practical decision making. How this result differs across study and effect characteristics is the subject of our moderator analyses.

Random-Effects of Moderators From Metaregression

We consider the 14 moderators described above but include only 13 in each model due to concerns regarding multicollinearity stemming from high correlations between effects generated from the revenue diversification measure calculated from three revenue sources and those effects drawn from panel studies, correlated at .65. So, we do not include the three-revenue-streams moderator in the same models with the longitudinal data moderator. No other moderator intercorrelations approached .60.

Modern metaregression calculates cluster-robust variance estimation (CRVE) and generalized estimating equations (GEE) models. Both models designed to address potential methodological issues: effect-size heteroscedasticity and nonindependence of observations (Liang & Zeger, 1986; White, 1980). In addition, GEE models place less emphasis on individual studies that contribute many effects to the meta-analysis (Liang & Zeger, 1986). We report both in Table 2, but believe the GEE models are a better fit for our situation. One feature of our data is notably large contributions (number of effects) from two studies: Chang and Tuckman (1994; 54 of 296 effects, 18.2%) and Frumkin and Keating (2011; 84 of 296 effects, 28.4%). GEE models place less emphasis on effects from studies that dominate the data, which help to temper conclusions that might be driven by these two studies. The dependent variable in the moderator analysis is effect size, transformed to Fisher’s z by methods described above. Results of the moderator analyses are presented in four models in Table 2.

Table 2.

Clustered Robust Variance Estimations and Generalized Estimating Equations, Random-Effects Metaregression (n = 40 Studies, 296 Effects).

Moderator	CRVE Model 1	CRVE Model 2	GEE Model 1	GEE Model 2
Capacity financial health measure	−.02 (.02)	−.02 (.02)	−.03 (.02)	−.02 (.02)
Three revenue streams		−.03^† (.02)		−.03* (.02)
Multivariate	−.01 (.02)	−.02 (.01)	−.01 (.02)	−.02 (.01)
Fixed effect model	−.02 (.04)	−.02 (03)	−.02 (.04)	−.01 (03)
Longitudinal data	.01 (.02)		−.00 (.02)
Published study	.00 (.04)	.01 (.04)	.00 (.04)	.01 (.04)
Released in or after 2011	−.07*** (.02)	−.05* (.02)	−.08*** (.02)	−.05* (.02)
U.S. data	−.08* (.04)	−.07^† (.04)	−.08* (.03)	−.07^† (.04)
Arts, culture, humanities	.00 (.02)	.01 (.02)	.00 (.02)	.01 (.02)
Education	−.00 (.03)	.00 (.03)	−.00 (.03)	.00 (.03)
Health	.04 (.03)	−.03 (.03)	−.04 (.03)	−.03 (.03)
Human services	.01 (.02)	.01 (.02)	.01 (.02)	.01 (.02)
International and foreign affairs	−.07 (.04)	−.06 (.04)	−.07^† (.04)	−.06 (.04)
Religion-related	−.01 (.06)	−.01 (.06)	−.02 (.06)	−.01 (.06)
Constant	.16** (.05)	.14** (.05)	.16*** (.05)	.14** (.05)
F statistics	16.12***	17.64***
Wald χ²			490.94***	1,311.44***
R squared	.1063	.1155

Note. Robust standard errors in parentheses. CRVE = clustered robust variance estimations; GEE = generalized estimating equations.

The coefficients in boldface type indicate statistical significance.

†

p ⩽ .1. *p ⩽ .05. **p ⩽ .01. ***p ⩽ .001.

Influence of financial health measure on effect sizes

The literature reflects a broad array of measures that we have characterized generally as financial health, including fundraising efficiency, revenue growth, months of liquidity, and fund balance. Following Bowman (2011a), we code this array as either a capacity or a sustainability measure and test whether this dichotomy helps explain the influence of revenue diversification on financial health. Notably, the difference is not statistically significant in any of our models. So, despite the diversity of ways that subject studies operationalize financial health, we find no moderation of revenue diversification effect size due to our broad dichotomy of financial health measures.

Influence of diversification measure grain (more than three revenue streams) on effect sizes

Chikoto et al. (2016) study diversification with fewer and more income streams and conclude that “a loss of information occurs through aggregation and this in turn affects estimation results in important ways” (p. 1439). Our results underscore this finding. For both CRVE Model 2 and GEE Model 2, effects derived from a revenue diversification constructed from three revenues sources are smaller, on average, than effects reported from studies that construct their diversification measure from more revenue sources.

Influence of bivariate versus multivariate models on effect sizes

As with choices in construction of measures, choice of statistical technique might also influence a study’s ability to accurately measure the magnitude of effects. Some studies report simple correlation coefficients with no statistical controls, whereas others utilize multivariate methods that incorporate the influence of other organizational characteristics. Our moderator analysis indicates that effects calculated from bivariate statistics are not significantly different from effects calculated from multivariate models, where controls (such as the size and age of organization) are employed.

Influence of effects from fixed effect models (adjustment for unobserved heterogeneity)

Efforts to adjust for unobserved heterogeneity should produce better estimators, which in turn might be more sensitive in calculating the magnitude and direction of the relationship between revenue diversification and financial health. The results of our moderator analysis, which differentiate effects from fixed effect models from nonfixed effect models, cannot confirm the value of this effort. Effects from both approaches return similar estimates.

Influence of longitudinal data on effect sizes

“Data structure” is a descriptor adopted by Lu (2016) in his meta-analysis of the relationship between government grants and private donations, where use of longitudinal data is shown to produce larger effect sizes. This relationship is not indicated in the relationship we study, however. Null coefficients for longitudinal data in CRVE Model 1 and GEE Model 1 indicate that effect sizes in analyses of the relationship between revenue diversification and nonprofit financial health do not differ according to whether researchers rely on longitudinal or cross-sectional data.

Influence of publication bias on effect sizes

Similarly, our moderator for published study is not statistically significant in any of the four models in Table 2. The results of these GEE models support the conclusion that the effects from published studies are not different from unpublished working papers, conference papers, or theses.⁴

Influence of publication era on effect sizes

As described above, we set 2011 as a marking point due to the publication of Frumkin and Keating’s (2011) contrary findings that year. Our observation that later studies are more likely to challenge the positive relationship between revenue diversification and financial health is corroborated by our systematic comparison of the effect sizes across these studies. The coefficient for those released in or after 2011 is significant in all four models in Table 2. The negative coefficient is expected, indicating that effects are smaller in later studies.

Influence of U.S. data on effect sizes

Our moderator for U.S. data tests for differences between effect sizes in studies conducted with U.S. data versus studies conducted with nonprofits in other countries. The coefficients across the models in Table 2 indicate that studies that focus on nonprofits in the United States return smaller (or more negative) effects of revenue diversification than studies reporting on other countries. This finding raises the question of whether effect sizes in U.S. studies are partly an artifact of revenue categories in Form 990, from which data on most U.S. studies are drawn.

Influence of subsector on effect sizes

Research articles on revenue diversification frequently concentrate on a specific subsector, or report different effects by subsector. The six subsector variables in Table 2 test the extent to which the relationship between diversification and financial health differs by subsector type of nonprofit organization. Of these six tests, only one result is marginally suggestive of a subsector effect: international and foreign affairs nonprofits. Among the 296 effects in the meta-analysis, only 14 focus on this subsector, but those effects are spread across five different studies (Chang & Tuckman, 1994; de Andrés-Alonso et al., 2015; de los Mozos et al., 2016; Despard, Nafziger-Mayegun, Adjabeng, & Ansong, 2017; Silva & Burger, 2015). The negative coefficient in GEE Model 1 hints that the effect size between diversification and financial health may be smaller, on average, for nonprofits operating in this subsector. In contrast, effect sizes for nonprofits operating in the arts, culture, and humanities; education; health; human services; and religion-related subsectors do not differ enough to register in our statistical tests.

Summary and Conclusions

The question of whether nonprofit organizations should pursue new and more income sources has the attention of both scholars and practitioners. The strategic decision to diversify these streams is potentially influenced by a variety of forces that escape consideration in most portfolio models, such as the value that organizations place on autonomy or community connectedness or the concerns they harbor regarding mission drift or increased costs in time and resources. These forces are largely unobserved, or at least unmeasured. Scholars measure what they can, and often this is just the simple association between revenue diversification and some measure of financial health. The diversity of findings regarding this relationship motivates this meta-analysis, although the moderator analysis ultimately sheds only a dim light on why the field has produced disparate findings on the revenue diversification question over the past quarter-century. Our study of 40 reports of 296 effects supports the popular principle that revenue diversification holds value for nonprofit organizations. The overall effect is small, with negative and null effects largely counter-balancing the positive assessment. So, although we conclude that the overall literature supports the practice of cultivating and balancing diverse revenue streams, this may not make a difference in many cases and could be harmful under some conditions. Under what conditions, revenue diversification contributes to financial health is an appropriate direction for future study.

Some of these conditions are explored in our moderator analyses, although the preponderance of null effects suggests that explanation of differences across studies lies in forces that are not easily measured. The first conclusions from our moderator analysis are primarily aimed at future scholarship on this topic rather than managers seeking advice on revenue portfolio management. Our analysis casts doubt on the proposition that measurement and methodology choices are important in detecting the direction and magnitude of the relationship between revenue diversification and financial health. An exception (validated in GEE Model 2) regards revenue diversification measures that consider more than three revenue sources: These report stronger (or less negative) effects. Most differences lie in characteristics that are beyond the control of the researcher, with more recent studies, and those conducted on non-U.S. nonprofits, exhibiting weaker (or more negative) returns from revenue diversification. These observations return us to the theoretical pros and cons we outline at the beginning of this paper. Future scholarship might benefit from greater attention to the influence of forces that the literature has identified as directly relevant to the relationship, such as autonomy, flexibility, risk-tolerance, and the value that an organization places on community embeddedness, among others discussed above.

The main advice for practitioners is to consider the strategic value of revenue diversification but to not place too much faith in the embryonic scholarly literature on this topic. The moderator in the current study that provides a modicum of insight into practice regards subsector of operation. We find that the small but positive influence of revenue diversification holds across models that test for effects in major subsectors, including arts, education, health, human services, and religion. Nonprofits operating in international and foreign affairs, however, return marginally smaller effects than nonprofits operating in other subsectors. Future research on this question might employ a more fine grained measure of organizational purpose, to better test Wilsker and Young’s (2010) contention that revenue options follow social purposes.

Two aspects of our study give us pause, one a reflection on our choices and the other a reflection on the state of the field. The first concerns the requirement of meta-analysis to define studies as inside or outside an area of inquiry according to diversity in how different studies define both their independent and dependent variables. Other analysts might make different decisions on whether a given study’s measure qualifies as revenue diversification or whether its measure is commensurate with other studies included in the meta-analysis. However, the pressing measurement question regards the dependent variable, financial health, which is defined differently in most extant studies, or not characterized as financial health at all. Although we are comfortable with our overall choice of representing these disparate conceptions as financial health, we believe that other analysts might make different or more nuanced tests for how revenue diversification might influence different interpretations or dimensions of this concept. We divide the literature broadly into capacity and sustainability measures, but other differentiations (e.g., flexibility, volatility, growth, efficiency) may be more useful or fruitful. In any case, our effort to gather the financial vulnerability literature under a colorful umbrella of financial health reflects Prentice’s (2016b) conclusion that accounting constructs are multidimensional, even if the disconnected efforts by various scholars to measure the constructs are otherwise incommensurable. Instructively, however, our finding that revenue diversification effect sizes are not influenced by whether a study defines its outcomes according to static capacity or dynamic sustainability measures provides hope for this incommensurability. Measurement of effect sizes is relatively similar regardless of disparate measurement of financial performance.

Finally, our foray through this literature underscores a disconnection between modern portfolio theory and prevailing analysis of the relationship between revenue diversification and nonprofit financial health, a point that motivates Qu’s (2016) critiques of the HHI as a reliable measure of revenue stream optimization. As we assert early in this article, portfolio theory can provide a valuable lens through which to view the complexity of the risks and rewards associated with multiple income streams, with conceptual advances featured in the work of Grasse et al. (2016), Mayer et al. (2014), and Kingma (1993). This perhaps explains why scholarship on nonprofit revenue concentration frequently refers to and purports to stand on portfolio theory as a scholarly pedestal. However, we question the extent to which portfolio theory has truly guided or informed research on this topic and the extent to which prevailing research on nonprofit revenue diversification tests or seeks to elaborate portfolio theory. We urge future scholars to take one of two paths in future scholarship in this area. One option is to bring the conceptual contributions of portfolio theory more squarely into scholarly inquiries of the value of revenue stream diversification, possibly by re-considering the way that it measures such diversification. A second, perhaps more fruitful option is to blaze a new conceptual path that recognizes the variety of forces that influence a nonprofit organization’s decision to pursue multiple disparate income streams. As suggested by Qu (2016), this path might include a shift in scholarly inquiry from the value of revenue diversification to the ways in which nonprofits can optimize revenue streams to achieve their missions.

Footnotes

Authors’ Note

The authors would like to thank the anonymous reviewers for their advice and comments, and Dr. Jiahuan Lu for his advice on technical issues examined in this article. The authors, however, bears full responsibility for the article.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

Notes

Author Biographies

ChiaKo Hung is a PhD and MPA student at Arizona State University. His areas of interest are nonprofit finance, fundraising, nonprofit competition, social entrepreneurship, and meta-analysis. He is currently working on his dissertation. The dissertation considers how meta-analyses advance knowledge development in the fields of nonprofit and public management.

Mark A. Hager is an associate professor of Nonprofit Leadership and Management at Arizona State University. He is editor-in-chief at Nonprofit Management & Leadership. His published research includes studies of the struggles and failures of nonprofit organizations, fundraising strategies, volunteer management practices, and financial reporting.

References

*Allen

A. A.

Henley

Doss

D. A.

(2014). Vulnerability and resource dependency of nonprofit organizations in Alabama’s black belt: Call for a better measure of perceived revenue diversification. In Arduin

(Ed.), Proceedings of the association for small business and entrepreneurship (pp. 195-230). Memphis, TN: 40th Annual Conference.

Andreoni

Payne

A. A.

(2011). Is crowding out due entirely to fundraising? Evidence from a panel of charities. Journal of Public Economics, 95, 334-343.

*Baba

Ishida

Okuyama

(2014). Revenue strategies and financial viability for emerging nonprofit sector in Japan: Commercialization or diversification? Kansai University Review of Business and Commerce, 15, 1-14.

Bingham

Walters

(2013). Financial sustainability within UK charities: Community sport trusts and corporate social responsibility partnerships. Voluntas, 24, 606-629.

Bowman

(2011a). Finance fundamentals for nonprofits: Building capacity and sustainability. Hoboken, NJ: Wiley.

Bowman

(2011b). Financial capacity and sustainability of ordinary nonprofits. Nonprofit Management & Leadership, 22(1), 37-51.

Brooks

A. C.

(2000). Public subsidies and charitable giving: Crowding out, crowding in, or both? Journal of Policy Analysis and Management, 19, 451-464.

Card

N. A.

(2012). Applied meta-analysis for social science research. New York, NY: Guilford Press.

*Carroll

D. A.

Stater

K. J.

(2009). Revenue diversification in nonprofit organizations: Does it lead to financial stability? Journal of Public Administration Research and Theory, 19, 947-966.

10.

*Chang

C. F.

Tuckman

H. P.

(1994). Revenue diversification among non-profits. Voluntas, 5, 273-290.

11.

*Chikoto

G. L.

Ling

Neely

D. G.

(2016). The adoption and use of the Hirschman–Herfindahl index in nonprofit research: Does revenue diversification measurement matter? Voluntas, 27, 1425-1447.

12.

*Chikoto

G. L.

Neely

D. G.

(2014). Building nonprofit financial capacity: The impact of revenue concentration and overhead costs. Nonprofit and Voluntary Sector Quarterly, 43, 570-588.

13.

*Chikoto-Schultz

G. L.

Neely

D. G.

(2016). Exploring the nexus of nonprofit financial stability and financial growth. Voluntas, 27, 2561-2575.

14.

Cooley

Ron

(2002). The NGO scramble: Organizational insecurity and the political economy of transnational action. International Security, 27(1), 5-39.

15.

*Cordery

C. J.

Sim

Baskerville

R. F.

(2013). Three models, one goal: Assessing financial vulnerability in New Zealand amateur sports clubs. Sport Management Review, 16, 186-199.

16.

*Daniel

J. L.

Kim

(2018). The scale of mission-embeddedness as a nonprofit revenue classification tool: Different earned revenue types, different performance effects. Administration & Society, 50, 947-972.

17.

*de Andrés-Alonso

Garcia-Rodriguez

Romero-Merino

M. E

. (2015). The dangers of assessing the financial vulnerability of nonprofits using traditional measures. Nonprofit Management & Leadership, 25, 371-382.

18.

*de los Mozos

I. S. L.

Duarte

A. R.

Ruiz

Ó. R

. (2016). Resource dependence in non-profit organizations: Is it harder to fundraise if you diversify your revenue structure? Voluntas, 27, 2641-2665.

19.

*Despard

M. R.

Nafziger-Mayegun

R. N.

Adjabeng

B. K.

Ansong

(2017). Does revenue diversification predict financial vulnerability among non-governmental organizations in sub-Saharan Africa? Voluntas, 28, 2124-2144.

20.

*Faulk

(2010). Edifice complex: Building ownership and financial strength of nonprofit theaters (Working paper). RGK Center, University of Texas. Retrieved from https://repositories.lib.utexas.edu/handle/2152/61854

21.

Foster

Fine

(2007). How nonprofits get really big. Stanford Social Innovation Review, 5, 46-55.

22.

Froelich

K. A.

(1999). Diversification of revenue strategies: Evolving resource dependence in nonprofit organizations. Nonprofit and Voluntary Sector Quarterly, 28, 246-268.

23.

*Frumkin

Keating

E. K.

(2011). Diversification reconsidered: The risks and rewards of revenue concentration. Journal of Social Entrepreneurship, 2, 151-164.

24.

Glass

G. V.

(1976). Primary, secondary, and meta-analysis of research. Educational Researcher, 5, 3-8.

25.

*Gordon

T. P.

Fischer

Greenlee

Keating

E. K.

(2013). Warning signs: Nonprofit insolvency indicators. International Research Journal of Applied Finance, 4, 343-378.

26.

Grasse

N. J.

Whaley

K. M.

Ihrke

D. M.

(2016). Modern portfolio theory and nonprofit arts organizations: Identifying the efficient frontier. Nonprofit and Voluntary Sector Quarterly, 45, 825-843.

27.

*Greenlee

J. S.

Trussel

J. M.

(2000). Predicting the financial vulnerability of charitable organizations. Nonprofit Management &Leadership, 11, 199-210.

28.

Gronbjerg

K. A.

(1992). Nonprofit human service organizations: Funding strategies and patterns of adaptation. In Hasenfeld

(Ed.), Human services as complex organizations (pp. 73-97). Newbury Park, CA: SAGE.

29.

Gronbjerg

K. A.

(1993). Understanding nonprofit funding: Managing revenues in social services and community development organizations. San Francisco, CA: Jossey-Bass.

30.

*Hager

M. A.

(2001). Financial vulnerability among arts organizations: A test of the Tuckman-Chang measures. Nonprofit and Voluntary Sector Quarterly, 30, 376-392.

31.

Hager

M. A.

Galaskiewicz

Bielefeld

Pins

(1996). Tales from the grave: Organizations’ accounts of their own demise. American Behavioral Scientist, 39, 975-994.

32.

Hager

M. A.

Galaskiewicz

Larson

J. A.

(2004). Structural embeddedness and the liability of newness among nonprofit organizations. Public Management Review, 6, 159-188.

33.

*Harvey

(2012, November). Mission and Mammon: Revenue diversification in the church. Paper presented at the meeting of the Association for Research on Nonprofit Organizations and Voluntary Action, Indianapolis, IN.

34.

Higgins

Thompson

S. G.

(2002). Quantifying heterogeneity in a meta-analysis. Statistics in Medicine, 21, 1539-1558.

35.

Huedo-Medina

T. B.

Sánchez-Meca

Marín-Martínez

Botella

(2006). Assessing heterogeneity in meta-analysis: Q statistic or I² index? Psychological Methods, 11, 193-206.

36.

Jegers

(1997). Portfolio theory and nonprofit financial stability: A comment and extension. Nonprofit and Voluntary Sector Quarterly, 26, 65-72.

37.

Kearns

K. P.

Bell

Deem

McShane

(2014). How nonprofit leaders evaluate funding sources: An exploratory study of nonprofit leaders. Nonprofit and Voluntary Sector Quarterly, 43, 121-143.

38.

*Keating

E. K.

Fischer

Gordon

T. P.

Greenlee

J. S.

(2005). Assessing financial vulnerability in the nonprofit sector (Faculty Research Working Paper Series). Cambridge, MA: John F. Kennedy School of Government, Harvard University.

39.

Keating

Elizabeth K.

Fischer

Mary

Gordon

Teresa P.

Greenlee

Janet S.

, Assessing Financial Vulnerability in the Nonprofit Sector (January 2005). KSG Working Paper No. RWP05-002; Hauser Center for Nonprofit Organizations Paper No. 27. Available at SSRN: https://ssrn.com/abstract=647662 or https://dx-doi-org.web.bisu.edu.cn/10.2139/ssrn.647662

40.

Kim

(2014). Does revenue diversification really matter? The power of commercial and donative distinction in the nonprofit arts. Workshop paper, University of Pennsylvania Social Impact Fellowship, 1-34.

41.

Kingma

B. R.

(1993). Portfolio theory and nonprofit financial stability. Nonprofit and Voluntary Sector Quarterly, 22, 105-119.

42.

*Lam

McDougle

(2016). Community variation in the financial health of nonprofit human service organizations: An examination of organizational and contextual effects. Nonprofit and Voluntary Sector Quarterly, 45, 500-525.

43.

Liang

K.-Y.

Zeger

S. L.

(1986). Longitudinal data analysis using generalized linear models. Biometrika, 73, 13-22.

44.

*Lin

Wang

(2016). What helped nonprofits weather the great recession? Nonprofit Management & Leadership, 26, 257-276.

45.

Lipsey

M. W.

Wilson

D. B.

(2001). Applied Social Research Methods Series: Vol. 49. Practical meta-analysis. Thousand Oaks, CA: SAGE.

46.

(2016). The philanthropic consequence of government grants to nonprofit organizations. Nonprofit Management & Leadership, 26, 381-400.

47.

*Matthiesen

J. A.

(2009). Searching for organizational effectiveness by examining financial vulnerability and nonprofit failure (Master’s dissertation). University of Oregon, Eugene.

48.

*Mayer

W. J.

Wang

H. C.

Egginton

J. F.

Flint

H. S.

(2014). The impact of revenue diversification on expected revenue and volatility for nonprofit organizations. Nonprofit and Voluntary Sector Quarterly, 43, 374-392.

49.

*Merk

A. E.

(2014). Weighing in on the hallmarks of nonprofit marketization: Organizational factors impacting financial performance indicators of scalability and sustainability (Doctoral dissertation). University of Texas, Arlington.

50.

Mitchell

G. E.

(2014). Strategic responses to resource dependence among transnational NGOs registered in the United States. Voluntas, 25, 67-91.

51.

*Myser

(2016). Financial health of nonprofit organizations (Doctoral dissertation). University of Kansas, Lawrence.

52.

Oster

S. M.

(1995). Strategic management for nonprofit organizations: Theory and cases. Oxford, UK: Oxford University Press.

53.

Pfeffer

Salancik

G. R.

(1978). The external control of organizations: A resource dependence approach. NY: Harper and Row Publishers.

54.

*Prentice

C. R.

(2016a). Understanding nonprofit financial health: Exploring the effects of organizational and environmental variables. Nonprofit and Voluntary Sector Quarterly, 45, 888-909.

55.

Prentice

C. R.

(2016b). Why so many measures of nonprofit financial performance? Analyzing and improving the use of financial measures in nonprofit research. Nonprofit and Voluntary Sector Quarterly, 45, 715-740.

56.

(2016). Two essays on nonprofit finance (Doctoral dissertation). Indiana University, Indianapolis.

57.

Ringquist

(2013). Meta-analysis for public management and policy. Hoboken, NJ: John Wiley.

58.

*Searing

E. A.

(2015). Beyond liabilities: Survival skills for the young, small, and not-for-profit (Doctoral dissertation). George State University, Atlanta.

59.

*Silva

Burger

(2015). Financial vulnerability: An empirical study of Ugandan NGOs (CIRIEC Working Paper 2015/15). Liège, Belgium: International Centre of Research and Information on the Public, Social and Cooperative Economy.

60.

Singh

J. V.

Tucker

D. J.

House

R. J.

(1986). Organizational legitimacy and the liability of newness. Administrative Science Quarterly, 31, 171-193.

61.

*Spyker

C. L.

Deol

H. S.

(2014, January). Doing too much? Financial vulnerability of Canadian registered charities. Paper presented at the meeting of the Canadian Academic Accounting Association, Edmonton, Alberta, Canada.

62.

*Tevel

Katz

Brock

D. M.

(2015). Nonprofit financial vulnerability: Testing competing models, recommended improvements, and implications. Voluntas, 26, 2500-2516.

63.

*Trussel

J. M.

(2002). Revisiting the prediction of financial vulnerability. Nonprofit Management & Leadership, 13(1), 17-31.

64.

*Trussel

J. M.

Greenlee

(2004). A financial rating system for nonprofit organizations. Research in Governmental and Nonprofit Accounting, 11, 105-128.

65.

Tuckman

H. P.

Chang

C. F.

(1991). A methodology for measuring the financial vulnerability of charitable nonprofit organizations. Nonprofit and Voluntary Sector Quarterly, 20, 445-460.

66.

*von Schnurbein

Fritz

T. M

. (2017). Benefits and drivers of nonprofit revenue concentration. Nonprofit and Voluntary Sector Quarterly, 46, 922-943.

67.

*Wang

H.-C. J.

Liu

Y.-C. A.

(2013). The cause of private universities’ financial vulnerability and performance evidence from Taiwan. Actual Problems of Economics, 11, 308-315.

68.

*Webb

(2015). A path to sustainability: How revenue diversification helps colleges and universities survive tough economic conditions. Journal of International & Interdisciplinary Business Research, 2, 69-97.

69.

White

(1980). A heteroskedasticity-consistent covariance matrix estimator and a direct test for heteroskedasticity. Econometrica, 48, 817-838.

70.

*Wicker

Breuer

(2013). Understanding the importance of organizational resources to explain organizational problems: Evidence from nonprofit sport clubs in Germany. Voluntas, 24, 461-484.

71.

*Wicker

Breuer

(2014). Examining the financial condition of sport governing bodies: The effects of revenue diversification and organizational success factors. Voluntas, 25, 929-948.

72.

*Wicker

Longley

Breuer

(2015). Revenue volatility in German nonprofit sports clubs. Nonprofit and Voluntary Sector Quarterly, 44(1), 5-24.

73.

Wilsker

A. L.

Young

D. R.

(2010). How does program composition affect the revenues of nonprofit organizations? Investigating a benefits theory of nonprofit finance. Public Finance Review, 38, 193-216.

74.

*Yan

Denison

D. V.

Butler

J. S.

(2009). Revenue structure and nonprofit borrowing. Public Finance Review, 37, 47-67.

75.

*Zhai

(2012). Assessing the vulnerability of not-for-profit organisations in Australia (Doctoral thesis). University of Western Australia, Perth.