Experimentally Disentangling Donors’ Perceptions of Government-Supported Nonprofits: Cost-Efficiency,Program Impact,and Shared Services

Abstract

This study advances understanding of the relationship between government support and private donations, by further investigating the mechanisms underlying that relationship and by examining a nonmonetary form of government support, namely, shared services. We use a survey experimental design to highlight U.S. donors’ perceptions of government-supported nonprofits. The results suggest that donors are less willing to give to government-funded nonprofits. This is not only because donors see government funding as a substitute for their donations but also because donors perceive government-funded nonprofits as cost-inefficient. The results also suggest that donors’ relative reluctance to donate to government-funded nonprofits is not because donors perceive government-funded nonprofits as less impactful and that donors’ decisions about giving do not vary according to the forms of government support nonprofits receive. Overall, our findings stress the importance of governments simplifying administrative procedures for nonprofits to apply for and manage government funding.

Keywords

government support charitable giving government funding shared services crowding out

Nonprofits rely on multiple sources of revenue to fulfill their missions (Froelich, 1999; Hung & Hager, 2019; Shon et al., 2019). However, these revenue sources are correlated with, rather than independent from, each other (De Wit & Bekkers, 2017; Hung, 2020; Tinkelman & Neely, 2018). An increase in one revenue source might lead to increases or decreases in other revenue sources; the increases represent crowding-in effects, while the decreases represent crowding-out effects. As many nonprofits possess limited resources, crowding out is of considerable concern to them. On one hand, crowding out leads to a decrease in revenue. Without sufficient revenue at their disposal, nonprofits are not able to fulfill their missions. On the other hand, crowding out leads to a concentrated revenue structure, which makes it difficult for nonprofits to hedge against financial crises (Hung & Hager, 2019). Previous discussion of crowding out has focused particularly on the effects of government funding on private donations (De Wit & Bekkers, 2017; Lu, 2016) due primarily to the fact that government funding often constitutes the lion’s share of a nonprofit’s revenue structure (Lu, 2015), and private donations represent the public’s willingness to stand behind nonprofits and their causes (Froelich, 1999).

As the potential crowding-out effect is associated with the loss of financial resources and the public’s support, it is important for nonprofits to understand how potential donors react to government-funded nonprofits. The classic explanation for the crowding-out effect links donors’ giving decisions to the total supply of public goods (Roberts, 1984). It posits that donors respond to an increase in government funding by decreasing their donations because donors treat government funding as a substitute for their donations. However, this explanation not only ignores factors derived from the act of giving by donors (Andreoni, 1989; Hughes et al., 2014) but also overlooks donors’ perceptions of government-funded nonprofits’ costs and impact. Government-funded nonprofits may give potential donors an impression of high costs and low impact (Frumkin & Keating, 2011; Gazley, 2010), which in turn influences potential donors’ giving decisions.

Existing discussion of the crowding-out effect of government support on private donations has also largely focused on direct government funding (De Wit & Bekkers, 2017; Lu, 2016). There are, however, various nonmonetary forms of partnership between nonprofits and governments. One form of nonmonetary partnership that has been much discussed in the recent literature is shared services (Elston & Dixon, 2020; Elston & MacCarthaigh, 2016). The term “shared services” refers to centralizing administrative functions of governments and nonprofits that were previously performed separately. Unlike government funding, which emphasizes revenue increases to nonprofits, shared services focus on overhead savings. As donors are sensitive to nonprofit overhead costs (Gneezy et al., 2014; Hung et al., 2022), averse to risks when faced with the same gain (Kahneman & Tversky, 1979), and easily influenced by nonprofit image (Michel & Rieunier, 2012), potential donors’ decisions to give to nonprofits that run a shared service center with governments may be different from their decisions to give to nonprofits that receive government funding.

This study uses a survey experimental design to examine theoretical links between local government support and private donations to an environmental nonprofit, by highlighting U.S. donors’ perceptions of government-supported nonprofits.¹ Based on the literature (Andreoni, 1989; Frumkin & Keating, 2011; Gazley, 2010; Hughes et al., 2014), the crowding-out effect might occur when potential donors directly treat government funding as a substitute for their donations and when potential donors indirectly perceive government-funded nonprofits as cost-inefficient and/or less impactful. We therefore test these direct and indirect effects. Moreover, as potential donors are sensitive to nonprofit overhead costs, averse to risks, and easily influenced by nonprofit image (Gneezy et al., 2014; Hung et al., 2022; Kahneman & Tversky, 1979; Michel & Rieunier, 2012), we further test whether potential donors are more willing to give to nonprofits that run a shared service center with governments, compared with nonprofits that receive direct government funding.

This study advances our understanding of the crowding-out effect by further examining the indirect effects of perceived costs and impact. Despite arguments that working with governments increases nonprofit costs and restrict nonprofits’ ability to advocate for their own issues (Frumkin & Keating, 2011; Gazley, 2010; Jones, 2007), to the best of our knowledge, no prior research has examined those indirect effects. Moreover, this study further tests potential donors’ perceptions of shared services run by governments and nonprofits. A good deal of recent scholarly attention has shifted to this form of partnership (Elston & Dixon, 2020; Elston & MacCarthaigh, 2016), but we know little about whether and how it influences potential donors’ giving decisions. Findings from this study offer important implications for nonprofits and governments. We detail these implications in the “Discussion” section.

Government Support and Charitable Giving

Nonprofits typically seek a variety of sources of revenue to finance their service programs. This is because not many nonprofits are able to find and rely solely upon a single stable and predictable revenue stream (Foster & Fine, 2007) and also because diversifying revenue streams is believed by scholars and practitioners alike to help hedge against uncertainty (Carroll & Stater, 2009; Hung & Hager, 2019). However, most nonprofits do not have a diverse structure from the beginning. They start and grow a revenue stream and then seek other sources of revenue to stabilize their budgets. Some nonprofits start by soliciting private donations and then seek earned income, while others start by securing government funding and then access private donations. In most cases, nonprofit revenue portfolios reflect a combination of private donations, government funding, earned income, and investment returns, with support from individuals (both donations and earned income) comprising a major part (McKeever & Pettijohn, 2014). Compared with other sources of revenue, private donations are considered important to many nonprofits as they represent the public’s willingness to stand behind nonprofits and their causes (Froelich, 1999).

The importance of having support from the public, combined with the necessity of obtaining additional sources of revenue to weather crises and continue service programs, leads to the concern of whether other sources of revenue crowd out private donations. The discussion of this concern in the literature has leaned heavily on government funding (De Wit & Bekkers, 2017; Lu, 2016). A classic explanation of the crowding-out effect of government funding posits that donors are purely altruistic, and their giving decisions depend on the total supply of public goods (Roberts, 1984). In this view, donors perceive government funding as a substitute for their donations to nonprofits to finance the supply of public goods, so their giving decreases as government funding to nonprofits increase. As a result, private donations might be completely substituted by government funding. Andreoni (1989), however, argues that donors’ consideration goes beyond total provision of public goods to the positive reward of helping others (i.e., warm glow). In this view, government funding only partially crowds out private donations. A large number of empirical studies have examined the crowding-out effect in the past decades.

Previous empirical studies (e.g., Grasse et al., 2022), along with two recent meta-analyses (De Wit & Bekkers, 2017; Lu, 2016), offer valuable insights into the crowding-out effect. This study encapsulates those insights in the following five points. The first three points focus on the factors that cause the variation in previous empirical results, with research gaps identified by De Wit and Bekkers (2017), Grasse et al. (2022), and Lu (2016), and the last two points emphasize research gaps identified by this study.

First, De Wit and Bekkers (2017) find that previous empirical studies are more likely to report the crowding-out effect when using experimental designs. They attribute this to several factors, such as that experimental designs can address potential endogeneity issues, which in turn reduces estimation bias. They also argue that experimental designs allow researchers to give participants full information on the level of government funding. Researchers are able to examine whether, and the extent to which, government funding makes a difference in participants’ giving decisions. In De Wit and Bekkers’s (2017) dataset, 96% of experimental estimates show the crowding-out effect of government funding on charitable giving. This percentage is higher than that of non-experimental estimates.

Second, many previous empirical studies use all levels of government support to examine the crowding-out effect (De Wit & Bekkers, 2017). This may be partly due to disaggregated data not often being available. For example, the widely used U.S. Form 990 files released by the National Center for Charitable Statistics only offer aggregated data on government funding. The crowding-out effect of federal government support, however, may be different from that of local government support (Grasse et al., 2022). Having disaggregated data will allow researchers to examine the effect of federal and local government support separately. Another method would be to use experimental designs to deliver the two types of government support information to participants to understand their effects respectively. As public services increasingly decentralize to local levels (Klijn, 2008), we need more research to focus on local government support, to better understand whether, and the extent to which, local government support crowds out private donations. The results would not only help nonprofits manage their revenue portfolios but also help local policymakers plan their partnership strategies (De Wit & Bekkers, 2017).

Third, the relationship between government funding and private donations varies according to nonprofit subsectors (Grasse et al., 2022); the effect is negative in some subsectors while positive in others (Lu, 2016). However, according to De Wit and Bekkers (2017), a large number of previous studies have examined the relationship in the arts and culture, social services, education, international aid, and heath subsectors, but very few studies have investigated the relationship in the religion and environmental subsectors. Lu (2016) finds that studies examining the relationship in the human services subsector report more negative effects, while studies testing the relationship in the arts and culture subsector report more positive effects. De Wit and Bekkers’s (2017) and Lu’s (2016) findings on subsector differences suggest two avenues for future research. First, the discussion of the relationship needs to take subsector differences into consideration. Second, more research on subsectors such as religion and environment would offer a more holistic view of the relationship between government funding and private donations.

Fourth, while previous studies have shown that, under certain conditions, government funding crowds out private donations (De Wit & Bekkers, 2017; Lu, 2016), they have rarely demonstrated why donors reduce their contributions in response to government funding. A common argument in support of the crowding-out effect, from the donors’ perspective, is based on the theory of pure altruism: Donors reduce their contributions to nonprofits when they believe that the demand for public goods has been met or fully supplied by government (Andreoni, 1989; Roberts, 1984). Donors treat government funding as a substitute for their donations. However, donors’ decisions about giving to government-funded nonprofits are not merely affected by total supply of public goods. Their giving decisions may also be affected by their perception of government-funded nonprofits’ costs and impact. Their concerns about diminishing cost savings and program impact (Frumkin & Keating, 2011; Gazley, 2010) may affect their perceptions of government-supported nonprofits, which in turn influence their willingness to give. The existing literature has largely focused on the classic explanations for the crowding-out effect and ignored other underlying mechanisms.

Finally, despite a variety of forms of government support to nonprofits (Gazley, 2008), previous studies have largely examined the effect of direct government funding on private donations (De Wit & Bekkers, 2017; Lu, 2016). Yet government support to nonprofits can be in the form of direct funding, contracts, tax credits, shared services, and so forth. Donors may react to different government support programs differently (De Wit & Bekkers, 2017). For example, compared with nonprofits receiving government funding, potential donors may be more likely to give to nonprofits running a shared service center with governments due partly to their impression that the goal of shared services is to achieve cost savings (Elston & MacCarthaigh, 2016; Schwarz, 2014). In this case, the crowding-out effect of government funding might be stronger than that of shared services. As existing literature mainly focuses on direct government funding, little is known about the crowding-out effect of other forms of government support.

Theory and Hypotheses

The Effect of Government Funding on Private Donations

To fill the gaps in the literature, this study draws on theoretical insights from the literature on the crowding-out effect, government–nonprofit partnership, fundraising management, and behavioral economics to examine donors’ decisions about giving to government-supported nonprofits. We argue that government funding affects potential donors’ willingness to give to nonprofits, through one direct and two indirect mechanisms. First, government funding may directly crowd out potential donors’ giving to nonprofits. Potential donors may respond to an increase in government funding support by decreasing their donations because they consider government funding to be a substitute for their donations.

Second, government funding may indirectly crowd out private donations partly due to donors perceiving government-supported nonprofits as cost-inefficient. Funding from governments involves many reports and paperwork (Gazley, 2010). In some cases, it also requires new expertise and management systems (Frumkin & Keating, 2011), both of which may increase nonprofits’ costs to deliver services. Many potential donors, however, expect nonprofits to deliver services at the lowest costs (Gneezy et al., 2014; Hung et al., 2022). Nonprofits therefore strive to keep their overhead costs as low as possible (Lecy & Searing, 2015; Tian et al., 2020) and seek to conserve resources by achieving the highest level of outputs at the lowest possible costs (Coupet & Berrett, 2019). When potential donors are aware that a nonprofit receives government funding, they may perceive the government-funded nonprofit as cost-inefficient, which in turn makes donors less likely to donate to the nonprofit.

Third, government support is meant to increase nonprofit efficiency and effectiveness by attaching various strings. Moreover, there is literature arguing that government support might serve as a signal of needs and trustworthiness to some potential donors, which in turn increases donors’ support to nonprofit organizations (Lu, 2016). Despite the argument, direct government funding may indirectly crowd out private donations due partly to donors perceiving government-supported nonprofits as less impactful. Such perceptions arise because donors believe that working with governments makes it harder for nonprofits “to maintain the independence of their mission” (Gazley, 2010, p. 60), that government funding to nonprofits often comes with strings attached (Shon et al., 2019), that governments’ priorities are often inconsistent with nonprofits’ priorities (Jones, 2007), and that government–nonprofit partnership may “restrict the ability of nonprofits to advocate their own issues” (Gazley, 2010, p. 60). Potential donors expect nonprofits to make an impact in communities (Karlan & Wood, 2017). When potential donors are aware that a nonprofit receives government funding, they may perceive that the government-funded nonprofit makes little difference in the community due to their concern about mission drift, which in turn makes them less likely to donate to the nonprofit.

Based on the analyses, we propose the following hypotheses to test the total, direct, and indirect effects of local government funding on private donations.

Hypothesis 1 (H1) (total effect): Potential donors who are informed about local government funding to a nonprofit are less willing to give to the nonprofit.

Hypothesis 1.1 (H1.1) (direct effect): Potential donors who are informed about local government funding to a nonprofit are less willing to give to the nonprofit, after controlling for perceived costs and impact.

Hypothesis 1.2a (H1.2a) (indirect effect): Potential donors who are informed about local government funding to a nonprofit perceive a higher level of operating costs for the nonprofit.

Hypothesis 1.2b (H1.2b) (indirect effect): Potential donors who perceive a higher level of operating costs for a nonprofit are less willing to give to the nonprofit.

Hypothesis 1.3a (H1.3a) (indirect effect): Potential donors who are informed about local government funding to a nonprofit perceive a lower level of impact created by the nonprofit.

Hypothesis 1.3b (H1.3b) (indirect effect): Potential donors who perceive a lower level of impact of a nonprofit are less willing to give to the nonprofit.

The Effect of Shared Services on Private Donations

Despite the prevalence in the literature of examining the effect of direct government funding on private donations, a good deal of recent scholarly attention has shifted to government support in the form of shared services (Elston & Dixon, 2020; Elston & MacCarthaigh, 2016; Walsh et al., 2008). The term “shared services” was widely used by the private sector during the 1980s and by the public sector after 2000 (Elston & MacCarthaigh, 2016). The interest in shared services among governments in the United States grew at the federal level and started with information management systems. Elston and MacCarthaigh (2016) state that shared services “remove an organization’s administrative and/or professional support functions to a specialist provider, who then offers the same services to multiple clients” (p. 349). Shared services are especially common in areas such as procurement, human resources, information technology, and legal and financial services (Elston & MacCarthaigh, 2016). The goal of sharing services is to reduce overhead costs, increase efficiency, and/or promote service quality (Elston & MacCarthaigh, 2016). Shared services appear not only among governments (Elston & Dixon, 2020) but also in cases where governments partner with nonprofits (Gazley, 2010; Walsh et al., 2008).

To nonprofit organizations, working with governments to run a shared service center is quite different from receiving funding from governments, as the former is about overhead savings whereas the latter focuses on revenue increases. The difference may influence people’s decisions about giving to nonprofits, as many donors are averse to nonprofit overhead expenses (Gneezy et al., 2014; Hung et al., 2022; Lecy & Searing, 2015; Tian et al., 2020). According to prospect theory, when faced with the same gain, people are risk-averse and will choose a sure prospect over a riskier one (Kahneman & Tversky, 1979). In the case of government support where potential donors are informed that both direct government funding and shared services deliver the same gain to a nonprofit, potential donors may prefer shared services over direct government funding as running a service center with a government has a higher certainty (i.e., lower risk) of reducing nonprofit overhead costs, compared with receiving funding from a government, which has lower certainty (i.e., higher risk) of overhead savings. Moreover, given the goal of shared service centers mentioned above, partnerships with governments through shared service centers enhance nonprofits’ image of demonstrating a commitment to innovation, trying new things, using efficient business technologies, obtaining scale economies, and working collaboratively with potential donors (Dixon & Elston, 2020). This “image-enhancing” effect is rarely perceived by potential donors in cases of direct government funding.² As donors are sensitive to nonprofit overhead costs, averse to risk, and easily influenced by nonprofit image (Gneezy et al., 2014; Kahneman & Tversky, 1979; Michel & Rieunier, 2012), we posit that potential donors may be more likely to donate, and donate more, to nonprofits supported by local government in the form of shared services, compared with nonprofits that receive local government funding. We thus hypothesize:

Hypothesis 2 (H2): Potential donors are more willing to give to a nonprofit running a shared service center with local governments, compared with a nonprofit receiving local government funding.

Figure 1 shows theoretical links tested in this study. The links include total, direct, indirect, and moderating relationships.

Figure 1

Theoretical Links for the Effect of Government Support on Private Donations.

Method

Research Design and Experiment Participants

To test the hypotheses, this study utilized a between-subject design experiment to examine the effect of local government support to a hypothetical environmental nonprofit on individual charitable donations. The use of the hypothetical organization can avoid contamination by participants’ prior knowledge of certain environmental nonprofits (Coleman, 2018). Moreover, many local governments fund environmental nonprofits to help fulfill the nonprofits’ mission. For example, the State Department of Environmental Management in Indiana provides several grant programs to which local governments and nonprofits may apply. Some local governments collaborate with nonprofits to meet their environmental responsibilities and goals after being funded by the state government. We recruited 1,195 participants in the United States for a 10-min survey experiment via Amazon Mechanical Turk (MTurk) with a study compensation of US$2 (US$1) for each participant.³ The debriefing information regarding the real purpose of the experiment was provided after participants completed the survey. After the data cleaning process, we excluded invalid responses.⁴ The final sample includes 893 participants (M age = 38, SD = 11.3, range = 18–77) with a gender distribution of 42.2% female, and an ethnic distribution of 72.2% Caucasian, 10.9% African American, 7.7% Asian, 4.8% Hispanic/Latino, and 4.3% Others.⁵

Power Analysis

We used G*Power 3.1 software to conduct a power analysis to determine the sample size needed at α = .05 and a power of 0.80. A recent meta-analysis found a negative correlation between governmental support and charitable giving with a median effect size of 0.18 (De Wit & Bekkers, 2017). The sample needed to detect an effect size of f = 0.18 across three conditions at α = .05 and a power of 0.80 is 303. Our final sample of 893 participants is sufficiently powered to detect the effect size.

Condition Manipulation

The manipulation of different forms of government support to nonprofits occurred in the first section of the questionnaire, in which we embedded different forms of support from government (1 = no government support information, 2 = government funding, 3 = shared service) in the solicitation message from the hypothetical nonprofit. The message in the control (no government support information) condition was as follows:

According to the World Health Organization, air pollution leads to about 4.2 million deaths per year. Around 91% of the world’s population live in places where air quality levels threaten health.

Our organization, a local nonprofit, focuses on local environmental programs aimed at monitoring, advocating, and educating local communities about air pollution issues and organizing volunteer projects. To achieve these missions, we have to fundraise $200,000 this year.

To continue fighting for our health and environment, we ask you to consider supporting our programs. Any donation from you will help us make the changes needed to give us all fresher air and a healthier environment.

The manipulation of government funding condition has added the information, “Our organization received a government grant of $100,000 this year and works closely with the local government” to the basic solicitation message above right after the sentence in bold.

The manipulation of shared service condition has added the information, “Our organization recently runs a shared services center with the local government. Having the shared services center will help our organization save $100,000 this year” to the basic solicitation message above before the last paragraph.

Questionnaires

After participants consented to participate in this study, they were randomly assigned to one of the three conditions to take the questionnaire. The questionnaire included three sections. The first section contains the manipulation message and several main questions regarding manipulation checks, three different charitable-giving decision-making scenarios, and evaluations of the hypothetical local nonprofit based upon reading the solicitation message. The second section surveyed personal perceptions toward the government and nonprofit sectors as well as environmental issues in general. The third section surveyed respondents’ demographic information and personal experiences (e.g., previous nonprofit working, volunteering, and donation experiences). After respondents completed the survey, the true purpose of the research was disclosed, and they were asked to maintain the confidentiality of the study information and to confirm their permission to have their data included in this research.

Measures

Dependent Variables

This study measured individual charitable giving via three different scenarios. Scenario 1: “What percentage of money paid to you by this study would you like to donate to the nonprofit (From 0% to 100%)?”⁶ Scenario 2: “Would you like to make a charitable donation to the nonprofit using your own money? Yes:___ No:___”; and Scenario 3: “Imagine that you have $100 to spend; how much would you like to give to the nonprofit (From $0 to $100)?” These three scenarios target different financial options and vary the amount of money involved. Scenario 1 targets participants’ study compensation. Scenario 2 targets individuals’ own resources. Scenario 3 targets an imaginary resource of US$100 to allocate. We also added an open-ended question after these three charitable giving scenarios: “Please briefly explain why you made this decision.” A series of correlational analyses indicated that these three outcome variables were significantly correlated. Specifically, participants as nondonors in Scenario 2 were highly likely to donate 0% in Scenario 1, χ²(36) = 697.4, p < .001, and US$0 in Scenario 3, χ²(40) = 678.2, p < .001. Also, participants donating a higher percentage in Scenario 1 were likely to donate a higher amount in Scenario 3 (r = .78, p < .001).

Independent Variables

This study included three experimental conditions: 1 = no government support information, 2 = government funding, 3 = shared service. Participants were randomly assigned to one of these three experimental conditions, in which they were exposed to different solicitation information about the two forms of government support (funding and shared service) or were not given any information about government support to the local nonprofit (the control condition).

Potential Mediating Variables

Individual perceptions of government-supported nonprofits were measured in the first and second sections of the survey through different statement items. Participants rated the extent to which they agreed with the statements using a 7-point Likert-type scale (1 = strongly disagree to 7 = strongly agree). This study tested two potential mediating variables. First, we measured individual perceptions of operating costs of the government-funded nonprofit through the statement: “The nonprofit is able to minimize the costs in achieving the mission.” We reverse the values of this variable to reflect our measure of perceived increase in operating costs when conducting analyses. Second, we measured individual perceptions of program impact of government-funded nonprofits through the statement that the nonprofit “makes very little difference in dealing with major problems.”

Potential Covariates

To minimize bias in our research, we controlled for potential covariates that were statistically significant across conditions based upon randomization check results. For example, individuals’ perceptions of air pollution and their personal attitudes toward the environmental nonprofits were measured through “Air pollution is the most pressing issue in my community” and “I personally prefer environmental nonprofits over other types of nonprofits.”

Results

Descriptive Statistics

In Scenario 1 (donating a percentage from the study compensation), 889 participants reported a non-missing value, among which 41 (4.61%) decided to donate all of their study compensation, 411 (46.23%) decided to donate part of their study compensation, and 437 (49.16%) decided not to donate. In total, 452 participants decided to donate in Scenario 1 (M = 45.73, SD = 30.89, range from 1% to 100%). There were no statistically significant gender differences in the donation, χ²(31) = 37.32, p = .201, males = 51.66%, females = 49.73%. There were also no statistically significant differences by ethnicity in the donation, χ²(31) = 37.66, p = .191, Caucasian = 54.47%, Non-Caucasian = 49.46%.

In Scenario 2 (donating own money), 893 participants reported a non-missing value, among which 358 (40.09%) decided to donate and 553 (59.91%) decided not to donate. There were no statistically significant gender differences in the donation, χ²(1) = 0.025, p = .875, males = 40.31%, females = 39.79%. There were also no statistically significant differences by ethnicity in the donation, χ²(1) = 0.155, p = .694, Caucasian = 41.13%, Non-Caucasian = 39.69%.

In Scenario 3 (donating an imaginary amount of US$100), 891 participants reported a non-missing value, among which 24 (2.69%) decided to donate US$100, 546 (61.28%) decided to donate part but not all of the US$100, and 321 (36%) decided not to donate. In total, 570 participants decided to donate in Scenario 3 (M = 34.68, SD = 27.34, ranging from 1 to 100). There were no statistically significant gender differences in the donation, χ²(34) = 30.97, p = .617. There were also no statistically significant differences by ethnicity in the donation, χ²(34) = 37.86, p = .297, Caucasian = 67.74%.

Manipulation Check

To ensure that our manipulation (embedding different forms of government support into a nonprofit portfolio) was effective, we utilized the following question for manipulation check:

How would you describe the form of the support received by the nonprofit? (1 = no government support information provided; 2 = receiving a grant of $100,000 from the local government; and 3 = saving $100,000 by running a shared services center with the local government).

We checked whether information regarding different forms of government support was well-received by the participants. The results indicated that 82.51% of participants passed the first manipulation check question. There were 209 manipulation failures for this question, which led to a sample of 986 after the manipulation check. We also excluded responses that were completed in less than 3 min or more than 30 min, or had duplicated IDs, for a final sample of 893 participants.

Randomization Check

We next examined whether our randomization process was effective, by checking whether there were statistically significant differences in the following variables across conditions.

Demographic Characteristics

We investigated the differences in participants’ demographic characteristics across three conditions on 15 items.⁷ Three variables, including bachelor (whether respondents’ educational level was bachelor or above, 1 = yes, 0 = no), married (whether respondents were in a marriage or not), and number of children were significant across the conditions. Therefore, we added these three variables as controls in the regression and mediation analyses later when we investigated the effects of government support on individual decisions about charitable donations to the local nonprofit.

Potential Confounding Variables

To ensure the internal validity of our experimental design, we also checked the randomization results on a couple of potential confounding variables. The results indicated no significant differences in participants’ ranking of the severity of air pollution, air pollution severity, F(2, 892) = 1.84, p = .16, and participants’ agreement with the mission of the environmental nonprofit organization, environmental nonprofit, F(2, 892) = .38, p = .68, across three conditions.

The randomization results suggested that after the randomization process, most demographic characteristics and potential confounding variables were balanced across conditions and our randomization was effective.

Total and Direct Effects of Conditions on Decisions About Charitable Giving

The descriptive statistics on the decisions about charitable giving, above, suggest that approximately half of the participants decided not to donate in the first two scenarios. The distribution of charitable giving in the first two scenarios suggests that the variables are left censored. Therefore, we used a Tobit model to run the regressions on two continuous variables (Scenarios 1 and 3): decision-making in donation percentage (donating a percentage from the study compensation) and decision-making in donating an imaginary US$100 (donating an imaginary amount of US$100) with a lower bound of 0 (Tobin, 1958). We used a logit model for the dummy variable (Scenario 2), decision-making in donating their own money yes/no (donating own money).

We first regressed the total effect and direct effect models on these three dependent variables: donating a percentage from the study compensation (Models 1 and 2 in Table 1), donating own money (Models 3 and 4), and donating an imaginary amount of US$100 (Models 5 and 6). In these models, we set the control group (no government support information condition) as the reference group for which coefficients were automatically omitted. The results indicated that government funding had a significantly negative impact on individual charitable donations in all of the three giving scenarios with the potential covariates controlled in Models 1, 3, and 5 (total effect: H1 was supported). Compared with respondents assigned to the control group, respondents assigned to the government funding group donated at least 4.81% less paid money or US$4.63 less of the imaginary money to the nonprofit (Table 1). In addition, after we added the two potential mediators, perceived costs and impact, we found significant negative coefficients of our treatment effects on charitable giving in Models 4 and 6 (direct effect: H1.1 was mostly supported). In sum, the findings suggested that crowding-out effects of government funding on individual charitable giving—that is, government funding—leads to fewer private donations.

Table 1.

The Average Marginal Effects That Compare the Differences in Private Donations Between Government Funding and the Control Group.

Variables	(1)	(2)	(3)	(4)	(5)	(6)
Variables	Donating a percentage from the study (Tobit) compensation	Donating a percentage from the study (Tobit) compensation	Donating own money (Logistic)	Donating own money (Logistic)	Donating an imaginary amount of US$100 (Tobit)	Donating an imaginary amount of US$100 (Tobit)
Testing	Total effect	Direct effect	Total effect	Direct effect	Total effect	Direct effect
Government funding (vs. control group as reference)	−4.81*	−3.69	−0.11**	−0.09**	−4.63**	−5.98*
Government funding (vs. control group as reference)	(2.79)	(2.65)	(0.04)	(0.04)	(2.28)	(3.49)
Shared services (vs. control group as reference)	−6.48**	−4.17	−.15***	−.12***	−7.07***	−8.18**
Shared services (vs. control group as reference)	(2.80)	(2.67)	(0.04)	(0.04)	(2.29)	(3.51)
Perceived costs (R)		−0.28		0.003		0.12
Perceived costs (R)		(1.03)		(0.02)		(1.36)
Perceived impact		4.11***		0.06***		6.30***
Perceived impact		(0.53)		(0.01)		(0.72)
Potential covariates^aObservations	Controlled	Controlled	Controlled	Controlled	Controlled	Controlled
Potential covariates^aObservations	889	889	893	893	891	891

Potential covariates include individual’s willingness to support nonprofit mission, individual’s perception about nonprofit resource allocation efficiency, nonprofit professional norm accordance, nonprofit trust, nonprofit acts in the best interests of the community, government performance satisfaction, individual satisfaction with partnership between government and nonprofit, pay deduction, and individual’s demographic characteristics. Marginal effect coefficients are reported. Robust standard errors in parentheses.

p < .1. **p < .05. ***p < .01 (two-tailed).

Indirect Effects of Perceived Costs and Impact

To assess the crowding-out mechanisms underlying the effect of government funding on private donations, we examined two potential mediators by conducting mediating analyses suggested by Baron and Kenny (1986). We also employed Preacher and Hayes’s non-parametric resampling procedures to generate bootstrap confidence intervals (Preacher & Hayes, 2004, 2008). We conducted these analyses on the two potential mediating variables—perceived costs and perceived impact, respectively—in all of the three giving scenarios (donating a percentage from the study compensation, donating own money, and donating an imaginary amount of US$100).

We first examined the mediating role of individuals’ perceptions of costs (perceived costs). According to Table 2, the results suggested that the indirect coefficients of perceived costs were all statistically significant. According to the results, potential donors who are informed about government funding perceive a higher level of operating costs of nonprofits (H1.2a was supported), and potential donors who perceive a higher level of operating costs are less willing to give to nonprofits (H1.2b was supported).

Table 2.

The Results of the Mediation Analyses.

Mediation relationships	Path coefficients				Bootstrapping results for indirect effects
	Path coefficients				Indirect effects	Model summary		BCa 95% CI
	a t-val	b t-val	c t-val	c′ t-val	Indirect effects	Z score	p value	LL	UL
1. Mediation role of perceived costs
Government funding → perceived costs (R) → donate percentage	.37***	−2.21***	1.20	2.01	−.817**	−2.42	.016	−1.479	−.154
Government funding → perceived costs (R) → donate own	.376***	−.05***	.021	.039	−.019**	−3.15	.002	−.032	−.008
Government funding → perceived costs (R) → donate own dollar	.38***	−1.92***	2.23	2.96	−.73***	−2.53	.011	−1.302	−.166
2. Mediation role of perceived impact
Government funding → perceived impact → donate percentage	.168	2.06***	1.20	.849	.346	1.38	.166	−.144	.837
Government funding → perceived impact → donate own	.164	.006	.02	.021	.001	0.57	.571	−.002	.004
Government funding → perceived impact → donate own dollar	.162	1.47***	2.23	1.99	.239	1.25	.210	−.134	.612

Note. BCa bias corrected and accelerated; 5,000 bootstrap samples. a refers to the relationship between the independent variable and mediating variable, b refers to the relationship between the mediating variable and dependent variable, and c′ refers to the direct effect between the independent variable and dependent variable. c refers to the total effect between the independent variable and dependent variable. And indirect effect is a × b. So, c = c′ + ab. CI = confidence interval; LL = lower limit; UL = upper limit. See Preacher and Hayes (2004, 2008).

p < .1. **p < .05. ***p < .01.

We then repeated the same procedures by investigating the mediating roles of individuals’ perceptions of impact. However, there was no empirical evidence to indicate the mediating role of perceived impact (H1.3a was not supported and H1.3b was not supported).

To check the robustness of the mediating analyses, we ran the bootstrap with 5,000 replications for each of the six sets of mediating relationships. The bootstrap results were reported in Table 2 with BCa 95% confidence interval. The bootstrap results were consistent with the previous mediation analyses. Taken together, our mediation analyses suggested a mediating role of perceived costs in the relationship between government funding and private donations.

Government Funding Versus Shared Services

To examine whether there was a significant difference between the effects of government funding and shared services on private donations, we repeated the previous regression models again with the reference group changed to the government funding condition (see Table 3). Although the results in Table 3 indicated negative relationships, there were no significant differences, β = −1.67, SE = 1.95, p = .393 (Scenario 1); β = −.05, SE = .03, p = .123 (Scenario 2); β = −2.44, SE = 1.60, p = .128 (Scenario 3), between the total effects of government funding and shared services on private donations in all three giving scenarios (H2 was not supported).

Table 3

The Average Marginal Effects That Compare the Differences in Private Donations Between Shared Services and Government Funding.

Variables	(1)	(2)	(3)	(4)	(5)	(6)
Variables	Donating a percentage from the study (Tobit) compensation	Donating a percentage from the study (Tobit) compensation	Donating own money (Logistic)	Donating own money (Logistic)	Donating an imaginary amount of US$100 (Tobit)	Donating an imaginary amount of US$100 (Tobit)
Testing	Total effect	Direct effect	Total effect	Direct effect	Total effect	Direct effect
Shared services (vs. government funding as reference)	−1.67	−0.48	−.05	−.03	−2.44	−1.36
Shared services (vs. government funding as reference)	(1.95)	(1.89)	(0.03)	(0.03)	(1.60)	(1.54)
Control (vs. government funding as reference)	4.81*	3.70	0.11**	0.09**	4.63**	3.70*
Control (vs. government funding as reference)	(2.78)	(2.65)	(0.04)	(0.04)	(2.28)	(2.16)
Perceived costs (R)		−0.28		0.003		0.07
Perceived costs (R)		(1.03)		(0.02)		(0.85)
Perceived impact		4.11***		0.06***		3.90***
Perceived impact		(0.54)		(0.01)		(0.44)
Potential covariates^a	Controlled	Controlled	Controlled	Controlled	Controlled	Controlled
Observations	889	889	893	893	891	891

p < .1. **p < .05. ***p < .01 (two-tailed).

Discussion, Limitations, and Conclusion

Discussion

While the classic explanation of the crowding-out effect of government funding on private donations links donors’ giving decisions to the total supply of public goods and argues that donors treat government funding as a substitute for their donations (De Wit & Bekkers, 2017; Lu, 2016), this study goes a step further by examining whether U.S. donors’ perceptions of costs and impact indirectly influence their giving decisions. In addition, this study discusses an emerging nonmonetary form of government support to nonprofits, namely, shared services (Elston & Dixon, 2020; Elston & MacCarthaigh, 2016), examining whether donors’ decisions about giving to a nonprofit that receives government funding are different from their decisions about giving to a nonprofit that runs a shared service center with a government. We use an experimental design to test these theoretical links in the context of local government support to a hypothetical environmental nonprofit. We find that potential donors who are informed about local government funding are less willing to give to the nonprofit. Main underlying mechanisms of this negative relationship include the direct effect of donors viewing local government funding as a substitute for their donations and the indirect effect of perceived costs, but not the indirect effect of perceived impact. We also find that donors’ decisions about giving to nonprofits do not vary according to the two forms of local government support. We discuss these findings and their implications below.

First, local government funding crowds out private donations in all of the three giving scenarios. This finding supports the crowding-out hypothesis. Moreover, in the two scenarios where this study asks experiment participants whether they would give their study compensation or a hypothetical US$100, in terms of percentage, we observe that the crowding-out effect is more negative in the former scenario where experimental participants are compensated US$1 or US$2, compared with the latter scenario where experimental participants imagine that they have US$100 to spend. This suggests that income sources and amounts matter (de Li et al., 2019; De Wit & Bekkers, 2020). Even though people consider government funding to be a substitute for their donations to nonprofits, they can be more generous to nonprofits when imagining that they have received a relatively large amount of money. This implies that government-funded nonprofits could mitigate the crowding-out effect by focusing more of their fundraising efforts on people who win a windfall such as a lottery, capital gains from investment, or bequest (Li et al., 2019) or people who have more financial resources.

Second, the result of the indirect effect analysis shows that potential donors perceive nonprofits that receive local government funding as cost-inefficient, which in turn makes them less likely to give to those nonprofits. This finding is consistent across the three giving scenarios. Theoretically, the result offers an alternative explanation for the crowding-out effect. Potential donors’ perceptions of costs play a role in explaining the relationship between government funding and private donations. Moreover, the result suggests that donors are conservative in their view of nonprofit cost savings. Although working with governments does not guarantee cost savings by nonprofits, a study of Georgia local governments and nonprofits shows that more than two thirds of governments and roughly two thirds of nonprofits report that public–private partnerships reduce their costs (Gazley & Brudney, 2007). Governments and nonprofits seem to be more optimistic than potential donors about cost savings through partnerships. For government-funded nonprofits, it is important when fundraising to highlight potential cost savings of public–private partnerships, to increase donations.

Third, the result of the indirect effect analysis also shows that perceived impact does not mediate the relationship between local government funding and private donations. Although some nonprofits’ executive directors are concerned that working with local governments may lead to mission drift and restrict their ability to advocate for their own issues (Gazley, 2010; Jones, 2007), this concern does not appear to be shared by potential donors. There is no statistically significant evidence to support the relationship between nonprofits receiving government funding and donors’ perception of nonprofits’ impact. However, we unexpectedly find that potential donors who perceive a lower level of impact are more willing to give to nonprofits in two of the three giving scenarios where participants are asked to donate their study compensation or hypothetical money. This speaks to the notion that awareness of nonprofits’ need for support drives charitable giving (Bekkers & Wiepking, 2011). While some people only give to nonprofits that already make a big impact, others give to nonprofits that make little impact because they view those nonprofits as the ones that need more help and believe their contributions will have a positive effect on those nonprofits (Karlan & Wood, 2017).

Finally, the results from the comparison between donors’ giving decisions in cases of local government funding and in cases of shared services are not consistent with our expectations. We find that there is no statistically significant difference between government funding and shared services in terms of effect on charitable giving in any of the three giving scenarios. This may be because, practically speaking, shared services often do not deliver cost savings and sometimes even increase costs to parties involved. “There are plenty of examples . . . where a good idea that is not well implemented leads to increased costs and/or poorer service standards” (Australian Public Service Commission, 2013). Also, a survey of governments in the United States reveals that nearly half of shared service cases fail to achieve cost savings and a quarter end up with cost increases (Schwarz, 2014). Moreover, Elston and MacCarthaigh (2016) argue that the benefits of shared services are sometimes difficult to realize in practice. Elston (2021) further discusses the challenges of shared services in practice, which include, but are not limited to, “low uptake among agencies, defection by early adopters, failure to share beyond the organization’s immediate network, and the duplication of shared activates in-house” (p. 3). All of these may delegitimize nonprofits’ partnerships with local governments (Dixon & Elston, 2020), affecting people’s perceptions and decisions about giving to nonprofits that run shared service centers with governments.

Limitations

Unlike many experiments that recruit participants from college student pools, this study recruits participants from MTurk. Although our sample is much more representative than college samples, our findings still cannot generalize to all U.S. populations (Stritch et al., 2017). Future studies can use more representative samples to replicate our findings. The second limitation is about the problem of “non-naiveté.” Many MTurkers complete hundreds of studies monthly. They have been exposed to common experimental manipulations and tend to provide more acceptable rather than truthful answers—a phenomenon commonly called the social desirability issue. This in turn may affect our data quality. Another limitation is that we measured rather than manipulated the two proposed mediating variables. Therefore, we can only make causal inferences regarding the effects of governmental support on private donations; yet, it is difficult to identify causal mechanisms of mediating variables without establishing a “causal chain” (Imai et al., 2011; Spencer et al., 2005). We encourage future studies to further explore this topic. Moreover, the results found in this experiment are not necessarily reproduced in real-world phenomena; this raises the concern of external validity, especially in the case where donors are not aware of government funding to nonprofits or how governments support nonprofits, and where it is hard for donors to quantify the cost reduction of running a shared service center. Finally, the manipulations of government funding and shared service can be improved. Although ANOVA and post-pairwise test results indicated that participants assigned to the shared services condition are more “strongly agreed” that the nonprofit is able to minimize the costs in achieving the mission than participants assigned to the government funding condition, further studies can use clearer language in manipulations to distinguish the characteristics of these two forms of government support. For example, future studies could briefly explain what social service centers are and what they involve. Also, although this study uses the word “save” in the vignette where shared service centers are described, some experiment participants might still not know what shared service centers actually mean and involve. Therefore, instead of merely naming different forms of government support in the vignettes, future studies could describe the forms of management and/or organizational practices to help experiment participants better understand the information researchers provide.⁸

Conclusion

This study advances our understanding of why potential donors do not give, or give less, to local government-funded environmental nonprofits. It goes beyond the classic explanation to an alternative explanation that emphasizes potential donors’ perception of costs. Moreover, this study explores potential donors’ perceptions of shared services, compared with their perceptions of local government funding. Overall, our findings emphasize the importance, for local government-supported nonprofits, of taking potential donors’ perceptions of costs into consideration when fundraising. This does not necessarily imply that nonprofits should keep their costs as low as possible. It does, however, suggest that nonprofits should be transparent to potential donors about whether or how local government support changes their cost structures. Our findings also emphasize the importance for nonprofits of evaluating the costs and benefits of various forms of partnerships with local governments to avoid revenue decrease. For example, running a shared service center with local governments may be a better option if nonprofits can really save some money from it, given that both receiving direct local government funding and running a shared service center with local governments crowd out private donations. Finally, our findings stress the importance of local governments simplifying administrative procedures for nonprofits to apply for and manage government funding. In the long run, such an effort may help alter some potential donors’ perceptions of local government-funded nonprofits’ costs, and therefore increase their willingness to donate.

We suggest three directions for future research. First, there are many forms of government support to nonprofits. While this study further investigates shared services, there are some other forms of government support, such as government matching contributions, contracts, purchase of services, tax credits, voucher, outsourcing, and so on that have not been fully examined. Unlike shared services, which require complex bilateral administrative arrangements between governments and nonprofits, most other forms of government support shift administrative burdens to nonprofits as governments cast themselves as resource providers. Moreover, unlike the goal of shared services, which is to minimize overhead costs and increase efficiency, other forms of government support have quite different goals. All of these factors might influence donors’ perceptions of nonprofits. Future studies can explore how these other forms of government support affect private donations. Second, using experimental designs can advance our understanding of the differences between donors’ reactions to funding from different levels of governments, given that disaggregated data on government funding are rarely available. More experiments that focus on the state and local levels are needed, to give us a more holistic view on this subject. Third, we also suggest that future research uses non-experimental approach on the research question of shared services and donor giving decisions to reach higher ecological validity. Finally, this study focuses on environmental nonprofits. However, our results might not apply to other types of nonprofits (e.g., human services nonprofits). Future research can make further strides toward other subsectors.

Footnotes

Correction (May 2023):

Since the original online publication, this article has been updated for funding details. Please see for more details.

Data Availability Statement

The data and codes that produce the findings reported in this article are available at osf.io/4wpce/files/

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) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This study was supported by grants from the College of Social Sciences, University of Hawai’i at Mānoa (Primary); The National Natural Science Foundation of China, youth project (Grant No. 72004131) and (Grant No. 72004220); The Ministry of Education of Humanities and Social Science project (Grant No. 20YJC630136); Shanghai Pujiang Project (Grant No. 2020PJC070).

ORCID iDs

ChiaKo Hung

Yuan Tian

Notes

Author Biographies

ChiaKo Hung, PhD, is an assistant professor at the University of Hawai‘i at Mānoa. His areas of interest are nonprofit financial management, nonprofit–government collaboration, social entrepreneurship, and meta-analysis. His research has been featured in Nonprofit and Voluntary Sector Quarterly, VOLUNTAS, Public Performance & Management Review, Nonprofit Management and Leadership, and the Journal of Public Administration Research and Theory.

Yuan Tian, PhD, is a research associate professor at the School of International and Public Affairs, Shanghai Jiao Tong University. Her research has been specialized in philanthropic studies and cross-sector collaboration. And she has also been specialized in experimental method.

Youlang Zhang, PhD, is an associate professor at the School of Public Administration and Policy, Renmin University of China. His research interests include policy process, citizen–state interaction, and bureaucratic politics.

References

Andreoni

(1989). Giving with impure altruism: Applications to charity and Ricardian equivalence. Journal of Political Economy, 97(6), 1447–1458.

Australian Public Service Commission. (2013). State of the service report.

Baron

R. M.

Kenny

D. A.

(1986). The moderator–mediator variable distinction in social psychological research: Conceptual, strategic, and statistical considerations. Journal of Personality and Social Psychology, 51(6), 1173–1182.

Bekkers

Wiepking

(2011). A literature review of empirical studies of philanthropy: Eight mechanisms that drive charitable giving. Nonprofit and Voluntary Sector Quarterly, 40(5), 924–973.

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(4), 947–966.

Coleman

(2018). Designing experiments for the social sciences: How to plan, create, and execute research using experiments. SAGE.

Coupet

Berrett

J. L.

(2019). Toward a valid approach to nonprofit efficiency measurement. Nonprofit Management and Leadership, 29(3), 299–320.

De Wit

Bekkers

(2017). Government support and charitable donations: A meta-analysis of the crowding-out hypothesis. Journal of Public Administration Research and Theory, 27(2), 301–319.

De Wit

Bekkers

(2020). Can charitable donations compensate for a reduction in government funding? The role of information. Public Administration Review, 80(2), 294–304.

10.

Dixon

Elston

(2020). Efficiency and legitimacy in collaborative public management: Mapping inter-local agreements in England using social network analysis. Public Administration, 98(3), 746–767.

11.

Elston

(2021). Implementing shared services in the UK central government: A five-year progress report using longitudinal bureaumetrics. OECD Journal on Budgeting, 21(2), 1–26.

12.

Elston

Dixon

(2020). The effect of shared service centers on administrative intensity in English local government: A longitudinal evaluation. Journal of Public Administration Research and Theory, 30(1), 113–129.

13.

Elston

MacCarthaigh

(2016). Sharing services, saving money? Five risks to cost-saving when organizations share services. Public Money & Management, 36(5), 349–356.

14.

Foster

Fine

(2007). How nonprofits get really big. Stanford Social Innovation Review, 5(2), 46–55.

15.

Froelich

K. A.

(1999). Diversification of revenue strategies: Evolving resource dependence in nonprofit organizations. Nonprofit and Voluntary Sector Quarterly, 28(3), 246–268.

16.

Frumkin

Keating

E. K.

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

17.

Gazley

(2008). Beyond the contract: The scope and nature of informal government–nonprofit partnerships. Public Administration Review, 68(1), 141–154.

18.

Gazley

(2010). Why not partner with local government? Nonprofit managerial perceptions of collaborative disadvantage. Nonprofit and Voluntary Sector Quarterly, 39(1), 51–76.

19.

Gazley

Brudney

J. L.

(2007). The purpose (and perils) of government-nonprofit partnership. Nonprofit and Voluntary Sector Quarterly, 36(3), 389–415.

20.

Gneezy

Keenan

E. A.

Gneezy

(2014). Avoiding overhead aversion in charity. Science, 346(6209), 632–635.

21.

Grasse

N. J.

Searing

E. A.

Neely

D. G.

(2022). Finding your crowd: The role of government level and charity type in revenue crowd-out. Journal of Public Administration Research and Theory, 32(1), 200–216.

22.

Hughes

Luksetich

Rooney

(2014). Crowding-out and fundraising efforts: The impact of government grants on symphony orchestras. Nonprofit Management and Leadership, 24(4), 445–464.

23.

Hung

(2020). Commercialization and nonprofit donations: A meta-analytic assessment and extension. Nonprofit Management and Leadership, 31(2), 287–309.

24.

Hung

Hager

M. A.

(2019). The impact of revenue diversification on nonprofit financial health: A meta-analysis. Nonprofit and Voluntary Sector Quarterly, 48(1), 5–27.

25.

Hung

Hager

M. A.

Tian

(2022). Do donors penalize nonprofits with higher non-program costs? A meta-analysis of donor overhead aversion. Nonprofit and Voluntary Sector Quarterly. Advance online publication. https://doi.org/10.1177/08997640221138260

26.

Imai

Keele

Tingley

Yamamoto

(2011). Unpacking the black box of causality: Learning about causal mechanisms from experimental and observational studies. American Political Science Review, 105(4), 765–789.

27.

Jones

M. B.

(2007). The multiple sources of mission drift. Nonprofit and Voluntary Sector Quarterly, 36(2), 299–307.

28.

Kahneman

Tversky

(1979). Prospect theory: An analysis of decision under risk. Econometrica, 47(2), 263–291.

29.

Karlan

Wood

D. H.

(2017). The effect of effectiveness: Donor response to aid effectiveness in a direct mail fundraising experiment. Journal of Behavioral and Experimental Economics, 66, 1–8.

30.

Klijn

E. H.

(2008). Governance and governance networks in Europe: An assessment of ten years of research on the theme. Public Management Review, 10(4), 505–525.

31.

Lecy

J. D.

Searing

E. A.

(2015). Anatomy of the nonprofit starvation cycle: An analysis of falling overhead ratios in the nonprofit sector. Nonprofit and Voluntary Sector Quarterly, 44(3), 539–563.

32.

Liang

Liu

(2019). Does windfall money encourage charitable giving? An experimental study. VOLUNTAS: International Journal of Voluntary and Nonprofit Organizations, 30(4), 841–848.

33.

(2015). Which nonprofit gets more government funding? Nonprofits’ organizational attributes and their receipts of government funding. Nonprofit Management and Leadership, 25(3), 297–312.

34.

(2016). The philanthropic consequence of government grants to nonprofit organizations: A meta-analysis. Nonprofit Management and Leadership, 26(4), 381–400.

35.

McKeever

B. S.

Pettijohn

S. L.

(2014). The nonprofit sector in brief 2014 public charities, giving, and volunteering. Urban Institute. https://www.urban.org/sites/default/files/publication/33711/413277-The-Nonprofit-Sector-in-Brief–.PDF

36.

Michel

Rieunier

(2012). Nonprofit brand image and typicality influences on charitable giving. Journal of Business Research, 65(5), 701–707.

37.

Preacher

K. J.

Hayes

A. F.

(2004). SPSS and SAS procedures for estimating indirect effects in simple mediation models. Behavior Research Methods, Instruments, & Computers, 36(4), 717–731.

38.

Preacher

K. J.

Hayes

A. F.

(2008). Asymptotic and resampling strategies for assessing and comparing indirect effects in multiple mediator models. Behavior Research Methods, 40(3), 879–891.

39.

Roberts

R. D.

(1984). A positive model of private charity and public transfers. Journal of Political Economy, 92(1), 136–148.

40.

Schwarz

(2014). Public shared service centers: A theoretical and empirical analysis of US public sector organizations (Vol. 16). Springer.

41.

Shon

Hamidullah

M. F.

McDougle

L. M.

(2019). Revenue structure and spending behavior in nonprofit organizations. The American Review of Public Administration, 49(6), 662–674.

42.

Spencer

S. J.

Zanna

M. P.

Fong

G. T.

(2005). Establishing a causal chain: Why experiments are often more effective than mediational analyses in examining psychological processes. Journal of Personality and Social Psychology, 89(6), 845–851.

43.

Stritch

J. M.

Pedersen

M. J.

Taggart

(2017). The opportunities and limitations of using Mechanical Turk (Mturk) in public administration and management scholarship. International Public Management Journal, 20(3), 489–511.

44.

Tian

Hung

Frumkin

(2020). Breaking the nonprofit starvation cycle? An experimental test. Journal of Behavioral Public Administration, 3(1), 1–19.

45.

Tinkelman

Neely

D. G.

(2018). Revenue interactions: Crowding out, crowding in, or neither? In Seaman

B. A.

Young

D. R.

(Eds.), Handbook of research on nonprofit economics and management (pp. 35–61). Edward Elgar.

46.

Tobin

(1958). Estimation of relationships for limited dependent variables. Econometrica: Journal of the Econometric Society, 26, 24–36.

47.

Walsh

McGregor-Lowndes

Newton

C. J.

(2008). Shared services: Lessons from the public and private sectors for the nonprofit sector. Australian Journal of Public Administration, 67(2), 200–212.