The Political Embeddedness of Voluntary Action: The Case of Local Philanthropic COVID-19 Relief Funds

Abstract

Scholars and policymakers have long been interested in the complex relationships between political institutions and voluntary collective action. However, the reciprocal nature of their relationships complicates empirical analysis: voluntary action supports democratic institutions and political institutions enable voluntary action. This article examines the relationship between political institutions and the activation of local voluntary action in the context of COVID-19 funds managed by community philanthropic organizations. We find that political engagement, policy signaling, and political competition all support the emergence of a COVID-19 fund. The findings advance our understanding of the significant role that political institutions play in activating voluntary action.

Introduction

The relationship between voluntary collective action and political institutions is a matter of both theoretical and practical significance. While economic theory foregrounds government, viewing voluntary action such as nonprofit provision of public goods and services as a response to government failure (Weisbrod, 1977), political theory presents a more complex story. Since Tocqueville’s early description of associational life in the United States, political theorists have argued that strong democratic political institutions both emerge from and sustain rich voluntary associational life (Paxton, 2002). In other words, strong associational life and various forms of voluntary action bolster political institutions (Cuthill & Fien, 2005; Paxton, 2002) and healthy democratic institutions support voluntary action (Andrews, 2012; Skocpol et al., 2000) in a virtuous cycle.

This theorized reciprocity makes it difficult to study the relationship between voluntary action and political institutions. Disasters, as strategic research sites, provide opportunities to test these theories (Merton, 1987; Stinchcombe, 2005) as they shed light on how preexisting institutions influence the emergence of specific forms of voluntary action (Dutta, 2017). While most disasters are geographically isolated events, COVID-19 simultaneously affected communities around the world, allowing for a rich comparison across place.

In this paper, we leverage the COVID-19 context to address the research question: what is the relationship between political institutions and voluntary action? We argue that voluntary action is not merely a substitute or response to government (in)action. Rather, voluntary action is embedded in local political institutions. At the most basic level, government facilitates voluntary action in that nonprofits and private foundations are legally constituted by government policies and agencies (Clemens, 2020). However, other, less apparent, political factors shape voluntary action including local government capacity and policy, residents’ propensity to engage with government, residents’ political ideology, and local political competition. We test our hypotheses using data on the emergence of COVID-19 philanthropic funds hosted by local community philanthropic organizations (CPOs) during the first quarter of the pandemic. The emergence of COVID-19 funds activated voluntary action by allowing local individuals and organizations to pool their private contributions to address community needs. We find that while voluntary action in time of disaster may address problems in the short-term, the capacity and willingness to take voluntary action is rooted in local political institutions, which may affect the resilience of communities in the long-term.

Leveraging the COVID-19 context, our paper makes both theoretical and practical contributions. While theories of government failure and sector interdependence have largely dominated the study of voluntary action in the public administration literature, our findings emphasize that philanthropic response to support the public good are embedded in the broader political context. These findings have important implications for practice. Public policy increasingly relies upon voluntary action to respond to not only disasters, but the general needs of local communities (Dutta, 2017; Rivera & Nickels, 2014; Storr & Haeffele-Balch, 2012). However, voluntary action is closely entwined with political institutions, raising important questions about the resilience of local communities, especially in situations where political engagement is weak, disaster is politicized, and/or government response is limited.

We begin by reviewing the literature on the relationship between voluntary action and political institutions to develop our conceptual framework. We then describe the context of COVID-19, our methods, and findings. We conclude with a discussion of the implications for theory and practice.

Voluntary Action and Political Institutions

Political theorists have long been interested in how vibrant associational life and other forms of voluntary collective action support the emergence and sustainability of democratic political institutions. Putnam et al. (1992) described how civic cultures in Northern Italy, facilitated by voluntary associations, supported the growth of democracy in the region. Similarly with a focus on the United States: as Putnam (1995) described, “. . .it was the Americans’ propensity for civic association that most impressed [Tocqueville] as the key to their unprecedented ability to make democracy work” (p. 65). Associational life fosters civility and dense networks, helping ensure that citizens will cooperate for mutual benefit through local government (Foley & Edwards, 1996).

Other scholars challenge the “causal arrows,” arguing that political factors shape social structures (Krishna, 2002; Paxton, 2002). Voluntary action is embedded in political institutions and behaviors and well-designed democratic institutions and voluntary action are complements, not substitutes (Bowles & Gintis, 2002; Clemens, 2020; Foley & Edwards, 1996). At the very least, formal voluntary organizations and associations are formally sanctioned by government (Clemens, 2020) and supported by a variety of tax incentives (Reich, 2018). Skocpol (2008) argued that organized civil society developed alongside and in synergy with government, rather than apart from government.

Figure 1 illustrates five political factors that we theorize will shape the activation of local voluntary action, which we measure as the establishment of a COVID-19 fund by a CPO.

Figure 1.

Local political factors shaping the activation of voluntary action.

Government Size

Institutional perspectives from economics and political science offer divergent explanations for the relationship between the size of government and voluntary collective action. Economic theory widely used by nonprofit scholars views voluntary action as resulting from government and/or market failures (Weisbrod, 1977). In contrast, drawing upon concepts of sectoral interdependence, a growing number of nonprofit studies posit that government and voluntary action are complementary; government grants and contracts support the development and maintenance of a robust nonprofit sector, facilitating private voluntary action (Cheng, 2019; Lecy & Van Slyke, 2013; Paarlberg & Zuhlke, 2019).

Political institutional theories propose a broader perspective: that voluntary collective action is rooted in robust governance structures. This perspective focuses on how government builds capacity to engage in collective action by directly fostering the development of the human and physical resources that support voluntary action. For example, government capacity fosters social trust and opportunities for engagement, which supports voluntary action (Skocpol, 2008). Taking a historical perspective, Skocpol (1995) described how the provision of mass public education taught common civic virtues and civic participation skills, supporting the rise of middle-class reformers, and fostering participation in voluntary associations. Skocpol et al. (2000) further proposed that public infrastructure, such as mail and transportation systems, facilitated voluntary organizing in the United States. Finally, government service provision incentivizes participation by encouraging the public to feel that they have a stake in civic life, encouraging people to pay attention and engage (Mettler & Soss, 2004). Existing literature leads us to expect that larger local government would be associated with greater capacity and incentives for voluntary action.

• H1: The size of local government is positively related to the emergence of voluntary action.

Political Engagement

While concepts of sectoral interdependence and institutional approaches highlight the government resources and incentives enabling voluntary action, research on political engagement focuses on the consequences of public participation in political activities (Ekman & Amnå, 2012). Political engagement refers to the actions of ordinary people aimed towards influencing political and policy outcomes (Ekman & Amnå, 2012, p. 286). Through participation in public meetings, elections, letter writing, surveys protests, and other less obvious forms of political activities citizens form an understanding of their ethical responsibility (Cuthill & Fien, 2005) to “act in concert” (Sobieraj & White, 2004, p. 743) for the common good. Political participation creates a sense of agency and collective self-efficacy (Ikeda et al., 2008) and fosters social responsibility for the community’s well-being (Youniss et al., 1997).

While nonprofit research has often focused on nonprofits as “schools of democracy” (H. O. Jeong, 2013; Jo, 2021), there is also evidence that political participation effects voluntary engagement. Politically active counties have a higher density of nonprofits (Kim, 2015) and greater collaboration between government and nonprofits (Fusi & Feeney, 2020). In a study of civic engagement in Western Europe during the COVID-19 pandemic, Borbáth et al. (2021) found that people with stronger political interests were more likely to become civically engaged during the pandemic by helping in their neighborhood and donating money. We expect that political engagement reflects a shared sense of responsibility to the community and collective efficacy, enduring effects that would support voluntary action.

• H2: Community level political engagement is positively related to the emergence of voluntary action.

Policy Signaling

Recent work on policy feedback highlights the complex material and interpretive effects of public policy on voluntary collective action (Béland & Schlager, 2019; Mettler & Soss, 2004). Public policy that allocates instrumental resources and specifies rules that restrict or enable the participation of diverse groups in political processes can have direct effects on voluntary action by signaling the importance of an issue or the ways in which the public can engage (Goss et al., 2019; Schneider & Ingram, 1993). Government attention to a particular issue may increase the benefits of collective action (Bevan, 2013). Government provision of veteran benefits after the U.S. Civil War encouraged the formation of a variety of veterans groups working to expand these benefits (Skocpol, 1995). Similarly, anti-discrimination laws in the 1960’s, mobilized women in a variety of collective organizations to protect and expand their rights (Béland & Schlager, 2019).

Policy also shapes the interpretation of issues. Through feedback loops, policy defines problems and solutions (goals, rules, and incentives), and sends normative signals about both the kinds of problems and solutions that are in government’s purview, and the actors and activities that are appropriate in a democratic society. “These messages are interpreted and internalized by target groups and other players in the policy arena, shaping the way citizens view the problem and shaping their perceptions of whether their interests are of legitimate public concern” (Licari & Meier, 2000, p. 111). Policy signals produce shared understandings, norms, and identities. In the case of COVID-19, the emergence of state COVID-19 policies may have signaled the public health risk of the pandemic (Curley et al., 2021), highlighting its’ seriousness and legitimizing voluntary action.

• H3: Being located in a state with stronger COVID-19 policies is positively associated with the emergence of voluntary action.

Political Ideology

A small body of research examines the relationship between political ideology and voluntary action. Political ideology is a commitment to different values, such as individualism and openness to change, that reflect “beliefs about the proper order of society and how it can be achieved” (Erikson & Tedin, 2019, p. 64). Political ideologies shape preferences for government spending and service provision (Allan & Scruggs, 2004; Feldman & Zaller, 1992; McAllister & Makkai, 2021), support for private, philanthropic goods (Bielefeld et al., 2005; Brooks, 2006), and, according to Jack and Anderson (2002), conceptions of need, the prioritization of community problems, and the recognition of opportunities. Political values and preferences are important drivers of civic engagement (Bekkers, 2005).

Individual political ideology may shape charitable behavior (McAllister & Makkai, 2021) and similar contextual processes could affect the activation of voluntary action at the community level (Bielefeld et al., 2005; Lecy et al., 2019; Paarlberg et al., 2019). Paarlberg et al. (2019) found that charitable giving in U.S. counties increased with the percent voting Republican. Following the assumption that Republican communities have an inherent preference for voluntary versus government action, we expect Republican dominant counties will be more likely support the emergence of voluntary action.

• H4: Community-level Republican political ideology is positively related to the emergence of voluntary action.

Political Competition

Political competition, or the degree to which no political party dominates the electoral process, may shape the extent to which political preferences are reflected in voluntary action. Market models of democracy posit that citizens hold distinct preferences for public policies and that competition for votes drives levels of public expenditures (Downs, 1957). In the face of diverse voter preferences, private philanthropic activity develops to meet diverse demands that exceed the median voter’s preferences for government provision (Weisbrod, 1977). Although contemporary nonprofit research drawing on the median voter concept emphasizes socio-demographic indicators of diversity (Paarlberg & Zuhlke, 2019), historical notions of “the median voter” focus on competition between political parties as an indicator of diverse preferences (Downs, 1957).

Specifically, nonprofit historians have described how political competition early in the U.S. republic drove the development of a robust civil society (Hall, 1987; Neem, 2003). Voluntary control ensured that if government leadership were to shift to the “other party,” as was more likely in a competitive political environment, that “losing” interests would be protected. Neem (2003) argued that as political competition between Federalists and Republicans increased, leaders of each party feared the other’s control over elite institutions and were eager to move governance of civic institutions to the “voluntary sector.” Political competition may also expose disagreements over problem definitions and identification of potential solutions, discouraging government response and facilitating the emergence of voluntary action. Both perspectives suggest that, in the face of elevated levels of political competition, government fails to reach consensus and the nonprofit sector becomes the preferred mechanism for action. In support of these arguments, Paarlberg et al. (2019) found that charitable contributions are highest when political competition is greatest. We therefore expect that in counties with high levels of political competition, voluntary action may be more likely to emerge.

• H5: County-level political competition is positively related to the emergence of voluntary action.

The Context: Community Philanthropic Organizations

We test our hypotheses in the context of the activation of COVID-19 funds by two types of community philanthropic organizations (CPOs)—community foundations (CFs) and United Ways (UWs)—during the COVID-19 pandemic. CPOs are public charities, registered as 501(c)3 organizations, that raise funds from the public to make grants to other organizations. UWs and CFs are distinct from other public grant making organizations in that their fundraising and grant making occurs in a defined geographic place, often a single or multiple contiguous counties. UWs and CFs are in both rural and urban areas across the United States and support a broad range of efforts and interests aimed at improving the quality of life in the community (Brilliant, 1990; Hammack, 1989; Perry & Mazany, 2014). Their historical and social legitimacy in the community and place-based focus enable UWs and CFs to play key coordinating and philanthropic roles in local communities, particularly during times of disaster (Simo & Bies, 2007).

Many CPOs acted quickly during the pandemic, establishing funds to raise and distribute financial resources to respond to the immediate and long-term needs of local individuals and organizations impacted by COVID-19. These funds were sustained by contributions from a variety of sources, including businesses, individuals, and other foundations. Although institutional actors, such as the United Way Worldwide (UWW), the Council on Foundations, and regional philanthropic support organizations, provided public relations support, peer learning opportunities, and disaster response related resources, there were no central mandates for UWs or CFs to establish COVID-19 funds. The voluntary nature of these local responses offers a valuable context in which to study the political determinants of voluntary action. Unlike other disasters, which are often geographically isolated, COVID-19 affected all communities in the United States and across the globe. Although the pandemic was a global phenomenon, response was often local and, particularly in the United States, there was great variation in how states and communities responded (Curley & Federman, 2020).

Data and Methods

Data Collection

We began by identifying the population of U.S.-based CFs and UWs compiled through institutional lists, such as those of the UWW and Candid (created through the merger of Foundation Center and GuideStar). After cleaning, this included 1183 CFs (51.6%) and 1110 UWs (48.4%). Early in the pandemic, CPOs rapidly established COVID-19 funds, which took many forms. We used two criteria to define a COVID-19 fund. First, the fund was held or managed by a CPO on our list. Second, the fund was dedicated specifically to COVID-19 relief in the local community. In most cases, the description of these funds explicitly pledged that 100% of the donations would be used for pandemic response. Pandemic response efforts included emergency funding to individuals, local businesses, and organizations, as well as long and short-term support for a variety of community human, health, and education needs. Sometimes funds were launched in cooperation with other community partners. Supplemental Appendix A provides greater detail on the identification of CPOs and COVID-19 funds.

The four-member research team began tracking the emergence of COVID-19 funds managed by CPOs in March 2020 and continued identifying funds through August 2020. The team identified fund existence in several ways including: (1) sorting through the list of funds identified by philanthropic support organizations such as the Council on Foundations and the UWW, (2) searching the websites of CPOs on our list (1,615 organizations had websites), (3) monitoring social media posts using key word searches to identify COVID-19 funds, and (4) using Google alerts to monitor local press coverage announcing the launch of a fund by a UW or CF. Two members of the research team cross-checked the listings to verify that the fund was CPO-managed and that the funds were explicitly identified for COVID-19. Through this process, the team identified 1,032 CPOs managing a fund, with 165 funds jointly managed by two or more CPOs.

Dependent Variable: Fund Existence

The dependent variable is the existence of a fund (binary indicator of yes/no) as of August 2020. In total, we identified 1,032 organizations (45% of the total population of CFs and UWs) that had a COVID-19 relief fund.

Independent Variables: County Measures of Political Institutions

Our geographic unit of analysis for each case is the county in which the managing UW and/or CF was located. The county was identified based upon the organization’s mailing address as reported in the 2018 Business Master File (BMF) produced by the National Center of Charitable Statistics. U.S. counties are used as the unit of analysis for several reasons. First, while some UWs and CFs serve a portion of a county or multiple counties, more than 50% of CPOs serve a single county. Second, the devolution of power to local governments in the United States (Kelleher & Yackee, 2004) makes county government a key actor (Lobao & Kraybill, 2005) in the implementation of local, state, and federal policies. Counties may be a particularly important unit of government in non-metropolitan areas, where towns and villages have limited capacity. Third, a county-level analysis allows for better comparisons between this research and other work that uses counties to examine the relationship between political structures and voluntary action (Dutta, 2017; J. Jeong & Cui, 2020; Kim, 2015; Paarlberg et al., 2019; Paarlberg & Zuhlke, 2019; Van Puyvelde & Brown, 2016). In the following subsections, we describe how we operationalized independent variables to test each of our hypotheses.

Government Size

Previous studies have operationalized the size of government in various ways, including the total number of government employees, government revenue per capita, and government payroll (Kim, 2015; Lecy & Van Slyke, 2013; McCall et al., 2020). We measured government size as county government payroll expenditures on the assumption that it best captures the ability of government to act while excluding grants or contracts to private organizations. This measure includes both full-time and part-time employees. We collected this data from the U.S. Census Bureau, Annual Survey of Public Employment & Payroll, which occurs every 5 years. We supplemented the missing 2017 survey data with 2012 data. Dollar values from the 2012 survey were adjusted for inflation.

Political Engagement

We operationalized political engagement as the percent of the local population completing a Census return in 2010. Drawing on Ekman and Amnå’s (2012) distinction between civic and political engagement, we focus on Census return rates as a political activity. Census population counts are critical for the distribution of federal funds and Congressional redistricting, influencing policy or election outcomes and shaping resource allocations. Unlike voting, which may represent self-interested political action, Vigdor (2004) posited that Census completion is a form of individual political engagement that benefits the larger community. Unlike voting, all U.S. residents are required to complete the Census and the process is designed to require limited resources (Denny, 2022), but completion of the return reflects a willingness to engage with government (Desouza & Bhagwatwar, 2012). As the 2020 Census was being fielded as the pandemic began and COVID-19 funds were being launched, we used county-level 2010 Census return rates. In addition, Census completion rates are available for all counties, while other measures of political engagement, which might be available through the U.S. Census Bureau supplemental surveys provide rates for many urban counties but do not provide geographic identifiers for rural places. Studies that have analyzed Census return rates find these rates are positively correlated with other measures of civic norms of engagement including voting, trust, and participation in civic organizations (Martin & Newman, 2015).

Policy Signaling

We include a state-level measure of COVID-19 restrictions to capture government policy. COVID-19 policies potentially shaped perceptions of the risk and severity of the pandemic as a public problem, especially given the politicized nature of the COVID-19 response (Curley et al., 2021). We used Curley et al.’s (2021) state-level measure of COVID-19 executive orders adopted between February and May 2020. Their team kept track of the executive orders issued in each state and qualitatively coded the category of restriction (e.g., restaurant closures, business closures, school closures, bans on mass-gathering). Then they used elements of the Institutional Grammar Tool to scale the stringency of the language within each order. Higher numbers indicate more state-level restrictions were adopted during this time. We anticipate that higher scores will send stronger signals about the perceived threat of the pandemic and the legitimacy of voluntary action.

Political Ideology

We measured political ideology as the percentage of the county population that voted Republican in the 2016 presidential election. Because the 2020 election occurred after our identification of COVID-19 funds, we selected 2016 voting records. This measure captures a community’s position on the liberal-conservative spectrum as reflected by citizen voting behavior, rather than the ideological position of the office holder (Berry et al., 1998). Although there is the danger that voting behavior in national elections does not reflect attitudes toward local issues, studies have found a relationship between this measure and dimensions of local voluntary action (Lecy et al., 2019; Paarlberg et al., 2019). Further, while the 2016 U.S. presidential election was unusual because of its populist tone, research on the pandemic has found that county political voting patterns in the 2016 presidential election shaped individual and collective responses to the pandemic (Allcott et al., 2020).

Political Competition

There are several ways to measure political competition (Plotnick & Winters, 1985). For example, Paarlberg et al. (2019) used a Gini-Simpson index, calculated using the proportion of people voting Republican, Democratic or other. However, Plotnick and Winters (1985) describe competition as the balance between parties. In the U.S. two-party system, in which national third parties are weak, a high percentage voting for a single party suggests low political competition. Such places are less likely to switch back and forth between elections and the county is more likely to be dominated by a particular party. We operationalized political competition as a nonlinear measure—the square of the percent voting Republican in the 2016 presidential election. This is also more intuitive to interpret than a Gini index.

Control Variables

Organizational Controls

We included several organizational controls. The existence of other local funds in the community might influence an organization’s decision to launch a fund, thus we controlled for the existence of a COVID-19 fund established by another UW or CF in the same county. Forty percent of CPOs in our sample were located in counties in which another CPO had launched a fund. Since UWs and CFs have different business models- annual fund raising versus endowment building- which might affect their capacity to take quick action, we controlled for organizational type. Fifty-two percent were UWs and 48% were CFs. Finally, we expect that a CPO’s capacity to quickly launch a fund depends on the existence of paid staff. We used payroll data from the National Center for Charitable Statistics’ (NCCS) core files to create a binary variable measuring paid staff. If the organization reported a non-zero value for compensation or total salary within the latest three core files (2017, 2018 and 2019), then the payroll/employee status was coded as 1, otherwise as 0. Seventy-nine percent of CPOs in our sample reported payroll compensation.

County Level Controls

Drawing upon a diverse body of research on the contextual determinants of voluntary action, particularly in the face of disaster (Aldrich, 2012, 2019), we incorporated the six following county level controls. Detailed descriptions of the construction of these variables can also be found in Supplemental Appendix A. We gathered the data for our measures of concentrated advantage, stability, racial homogeneity, and population size from the 2018 American Community Survey 5-year estimates.

Disaster-Related Need: Severity of disaster may predict the emergence of voluntary action (Aldrich, 2012, 2019). Because of the unique nature of the pandemic, we included two controls for pandemic-related needs consistent with Borbáth et al.’s (2021) study of civic engagement in Western Europe. First, we included the number of COVID-19 cases in May 2020, at the county level (usafacts.org/visualizations/coronavirus-covid-19-spread-map). Second, as the pandemic was not only a health crisis, but also an economic crisis, we included changes in unemployment rates. We gathered monthly county-level unemployment rates for 2020 from the U.S. Bureau of Labor Statistics.

Socioeconomic Resources: Empirical studies of contextual determinants of voluntary action suggest a positive relationship between levels of place-based socioeconomic resources and the co-production of public services (Cheng, 2019; Gazley et al., 2020). Socioeconomic resources support voluntary action by creating a shared sense of efficacy (Sampson, 2012), greater capacity for engagement, and greater awareness of community needs and invitations to participate in voluntary activities (Verba & Nie, 1987). We used a principal component score consistent with Sampson’s (2012) measure of socioeconomic advantage.

Community Stability: Community stability increases sense of community and commitment to place (Sampson, 2012) and is positively associated with nonprofit engagement in the planning and design of public services (Cheng, 2019). In times of crisis, stability facilitates information sharing and trust, which are necessary for rapid action (Aldrich, 2012). In contrast, high rates of residential mobility may stress a community’s social organization as social ties and trust take time to form. Our measure of community stability is a principal component score adapted from Sampson (2012).

Racial Homogeneity: Racial homogeneity figures prominently in research on various forms of voluntary action (Alesina et al., 1999; Habyarimana et al., 2007) and may be particularly important in facilitating collective response to disaster (Aldrich, 2012; Dutta, 2017). We measured racial homogeneity using the sum of the shares of each major ethnic-racial group in the population of each county. The homogeneity index ranges from zero to one, with one indicating that a county’s population is homogeneous (composed of one group) and zero indicating that it is heterogeneous (all groups equally represented).

Population Size: We controlled for the total population of each county. Not only might population size affect a community’s willingness to take voluntary action, but it provides a control for the raw measures of the other county level variables, such as payroll and number of COVID-19 cases.

Social Capital: Social capital is an individual and community-level concept that captures the value that accrues from social networks among individuals and organizations (Putnam, 2000). Empirical research supports a positive relationship between social capital and generosity (Brooks, 2006; Brown & Ferris, 2007), including per-capita giving to community foundations (Graddy & Wang, 2009). Communities with high levels of social capital are better able to come together and engage in voluntary action during a disaster leading to improved outcomes (Aldrich, 2012; Monteil et al., 2020). We used a modified form of the county-level social capital index developed by Rupasingha et al. (2006). This index has been used in studies of a variety of social phenomena, including community resilience (Sherrieb et al., 2010) and political opposition (Hopkins, 2010). We modify this measure to capture the structural dimensions of social capital, which previous research suggests supports disaster response (Aldrich, 2012).

Table 1 summarizes the descriptive statistics of all non-categorical variables and reports the correlations between the non-categorical independent and control variables.

Table 1.

Descriptive Statistics of Non-categorical Independent and Control Variables.

Variables		1	2	3	4	5	6	7	8	9	10	11
Mean		14.84	0.75	7.53	0.55	11.63	4.72	2.58	−0.05	0.05	0.73	−0.03
SD		1.69	0.07	2.95	0.16	1.43	2.10	1.40	0.98	0.96	0.17	0.97
Min		0.00	0.27	1.00	0.10	6.54	0.00	0.04	−1.83	−3.29	0.18	−1.95
Max		20.39	0.95	16.50	0.88	16.13	10.73	12.50	4.92	2.50	0.99	20.99
Independent variables
1	Gov’t payroll (log)	1.00
2	2010 census return rate	0.08	1.00
3	Gov’t COVID19 restrictions	−0.11	0.04	1.00
4	Voting republican	−0.52	0.10	0.07	1.00
Control variables
5	Population (log)	0.81	0.10	−0.05	−0.63	1.00
6	COVID−19 cases May (log)	0.67	0.06	0.04	−0.60	0.86	1.00
7	Unemployment change rate	0.20	−0.04	0.14	−0.24	0.30	0.32	1.00
8	Concentrated advantage index	0.44	0.07	−0.03	−0.63	0.54	0.50	0.25	1.00
9	Residential stability index	−0.47	0.12	0.25	0.50	−0.57	−0.47	−0.16	−0.38	1.00
10	Population homogeneity	−0.59	0.19	0.16	0.57	−0.63	−0.61	−0.14	−0.35	0.60	1.00
11	Social capital index	−0.48	0.03	0.03	0.26	−0.56	−0.49	−0.18	−0.14	0.30	0.39	1.00
N = 2,052

Results

We used logit analysis (Table 2) to examine the relationship between local political factors and the emergence of a fund among the sample of UWs and CFs. We tested our hypotheses in two separate models. Model 1 examines the effects of governmental capacity, civic engagement, policy signaling, and political ideology on the activation of COVID-19 fund. McFadden’s pseudo R² of .20, while more difficult to interpret than the R² for a linear regression model, suggests a reasonably good fit for our base model. Model 2 extends the analysis by examining the effect of political competition (squared term of voting Republican) on fund activation.

Table 2.

Logistic Regression Model of Predictor of CPO Establishment of COVID 19.

	Model 1	Model 2
Independent variables
Gov’t pay roll (log)	0.06	0.05
	−0.06	−0.06
2010 census return rate	1.72*	1.34
	−0.79	−0.8
Gov’t COVID 19 restrictions	0.05*	0.04
	−0.02	−0.02
Voting Republican	−1.12 *	7.77 ***
	−0.5	−2.17
Voting Republican (squared)		−8.43 ***
		−2.01
Control variables
Existence of other funds	−0.28 *	−0.33 **
	−0.12	−0.12
Organization type-UW	0.27 *	0.27 *
	−0.11	−0.11
Organization payroll status	2.99 ***	2.95 ***
	−0.21	−0.21
Population (log)	0.18	0.18
	−0.1	−0.1
COVID-19 cases may (log)	0.03	0.03
	−0.05	−0.05
Unemployment change rate	0.08 *	0.10 *
	−0.04	−0.04
Concentrated advantage index	0.19 **	0.23 **
	−0.07	−0.07
Residential stability index	−0.19 *	−0.20 **
	−0.08	−0.08
Population homogeneity	0.79	0.81
	−0.48	−0.48
Social capital index	0.16 *	0.17 *
	−0.07	−0.07
Constant	−7.68***	−9.33***
	−1.13	−1.2
Observations	2,052	2,052
Log likelihood	−1,133.26	−1,124.48
Akaike Inf. Crit.	2,296.52	2,280.95

Note. *p < 0.05. **p < 0.01. ***p < 0.001.

In support of existing research on voluntary action and our hypotheses, we found that voluntary action (measured by the activation of a CPO COVID-19 fund) is deeply embedded in local political context. We begin with a brief discussion of the control variables. Examining the relationship between organizational and community-level characteristics and fund activation (Model 1), we found that the existence of other CPO-managed COVID-19 funds in the county where the organization is located reduces the likelihood of the organization establishing a COVID-19 fund (β = −0.28, p < .05). UWs are more likely to launch a fund than CFs (β = .27, p < .05). Having paid staff is also associated with launching a fund (β = 2.99, p < .01). As the change in unemployment rate increases, CPOs are more likely to launch a fund (β = .08, p < .05); however, the relationship between COVID-19 rates and the emergence of a fund is not statistically significant. Concentrated advantage (β = .19, p < .01) and social capital (β = .16, p < .05) are positively related to launching a fund. However, stability is negatively associated with launching a fund (β = −0.19, p < .05). Population size and homogeneity have no statistically significant relationship to launching a fund. The direction and significance of these controls are consistent after we introduce the quadratic function of proportion voting Republican in model 2.

While we do not offer hypotheses about control variables, it is important to mention that our results are consistent with previous empirical research on determinants of voluntary action. While multicollinearity may be a concern, variance inflation factors are below 2.5 for all predictor variables in our models, except for our control for population size, which is below the threshold of 10.

Turning now to the relationship between political institutions and the activation of a fund, we found support for hypotheses 2, 3, and 5. As shown in model 1, governmental size, measured by county government payroll, is not significantly associated with existence of a COVID-19 fund, contrary to hypothesis 1. Because the government payroll is strongly negatively correlated with the measure of political ideology, we also examined whether the size of government is significant when political ideology is not included in the model. It was not.

Consistent with hypothesis 2, as Census response rates increased, the likelihood of launching a fund increased (β = 1.72, p < .05). This suggests that political engagement is positively associated with the launch of a COVID-19 fund. In support of hypothesis 3, we found that policy signaling is positively associated with the establishment of a COVID-19 fund. As state-level COVID-19 policy restrictions increased, the likelihood of launching a fund increased (β = .05, p < .05). The effect of proportion voting Republican is also statistically significant but its negative relationship to launching a fund (β = −1.12, p < .05) contradicts hypothesis 4. As the proportion voting Republican increases, the likelihood of launching a fund decreases.

Model 2 (Table 2) displays the results of the effect of political competition on launching a fund. The introduction of the quadratic function supports a non-linear relationship between political ideology and the emergence of a fund. This non-linear relationship, displayed in Figure 2, supports hypothesis 5. The likelihood of launching a fund is greatest for those CPOs located in counties with the highest levels of political competition, then for counties more equally split between Republican and Democratic voters. When proportion voting Republican approaches 50%, a highly competitive political environment, CPOs were most likely to sponsor a fund. It is important to note, however, that when the quadratic term is introduced into the model, the other political variables, COVID-19 restrictions, and the Census return rate, are no longer significant. While we address this further in our discussion, this highlights the politicized nature of the COVID-19 response.

Figure 2.

Marginal effect of political competition on establishment of COVID-19 fund.

Finally, because CPO capacity is a strong predictor of the activation of a fund, we checked the robustness of our model by testing the relationship between having paid staff and the political variables. The results of this robustness check are included in Supplemental Appendix B. Two key findings emerge. First, political competition remains a strong predictor of this basic indicator of capacity for action. CPOs located in communities with more intense political competition are more likely to have paid staff. In addition, consistent with theories of government failure, even when controlling for community size, the size of local government is negatively related to having paid staff. While it is beyond the scope of this article, it is conceptually and statistically reasonable to assume that organizational capacity mediates the relationship between political factors and the activation of a COVID-19 fund.

Discussion

Our study extends ongoing discussions in public administration, political science, and nonprofit studies about the reciprocal relationship between voluntary action and political institutions by examining the effect of various political institutions on the emergence of voluntary action in response to the COVID-19 pandemic. While the pandemic may have been unusual in its scope, examining the relationships between political institutions and voluntary action in the context of COVID-19 provides an important and distinctive opportunity to untangle the effect of political institutions on voluntary action in local communities (Dutta, 2017). We found empirical support for the notion that voluntary action, the launch of a CPO fund in response to the COVID-19 pandemic, is not merely a product of government failure or reliance on government funding. Voluntary action is embedded in political institutions (Clemens, 2020). We begin by reviewing key findings and their implications for theory and practice. We conclude by discussing the limitations of our study and opportunities for future research.

Three key findings emerge from our study. First, we found no support that government size directly affected the launch of a fund during COVID-19. However, the negative relationship detailed in Supplemental Appendix B between the size of county government and the initial capacity to launch a fund (having paid staff) does support the concept of government failure. All else being equal, in places with limited government CPOs may already be playing a more active role in the community, so when a disaster strikes, they are poised to respond. Our context is a valuable test for this theory because unlike other tests of government failure and interdependence, our dependent variable (paid staff in a CPO) is unlikely to emerge because of government grants or contracts that may be associated with nonprofit service delivery systems. This provides initial support for a mediated model that is beyond the scope of this study. In the face of limited government, CPOs play leadership roles, which can be activated in times of disaster. These complex relationships warrant future empirical testing.

Second, CPOs in communities with high levels of political engagement, as evidenced by Census return rates, were more likely to launch funds. Local traditions of political engagement normalize individual and collective action for the public benefit, emphasizing the importance of preexisting roles, relationships, and capacities in determining voluntary organizational response in times of crisis (Aldrich, 2012, 2019; Dell et al., 2018; Rivera & Nickels, 2014). While our study occurred in the context of an extreme event, it is highly unlikely that CPOs that responded to COVID-19 were previously passive actors in their community. Existing norms of political engagement shape the roles that CPOs, and other nonpartisan organizations, play in mobilizing collective action in non-disaster situations.

Third, consistent with policy signaling theory (Goss et al., 2019; Schneider & Ingram, 1993), CPOs were responsive to state-level policies. Much of the existing policy signaling literature focuses on how the policy environment creates incentives for political participation by creating “interest groups” or rules that directly restrict or facilitate participation. Our findings extend this literature by emphasizing that policy signals matter outside of interest group formation. State restrictions were weakly correlated with county infection rates, suggesting that state policy restrictions had a symbolic signaling effect, legitimizing COVID-19 responses in local communities. Additional research is needed to understand how nonprofits perceive and use public policy to legitimize and mobilize voluntary action. Signaling may be especially important in the context of politically polarized issues, such as the pandemic.

However, the effects of policy signaling and pre-existing political engagement did not remain significant in the face of intense political competition. Statistically, although the linear term of political ideology is not strongly correlated with political signaling and political engagement, political competition may be strongly correlated with both variables. This leads us to the most intriguing finding for theory development: high political competition is associated with the establishment of a fund and having paid staff prior to the pandemic. There are a couple of explanations for these findings.

We hypothesized that in the face of political competition, nonprofits may serve as a means of moderating political conflict by allowing interest groups to retain control over institutions (Hall, 1987; Neem, 2003). Alternatively, one might posit that high levels of political competition may lead to more moderate political positioning in a community. In communities with high levels of political competition, individuals are more likely to encounter someone from the other political party and, through such contact, be more willing to “work across the aisle” to respond to community needs.

However, a small but growing body of literature from political science and organization studies suggests that contact may not help people bridge differences in political ideology because, in the United States, differences in political ideology shape identities and affective categorizations. Iyengar et al. (2012, 2019; Iyengar & Westwood, 2015) have demonstrated that individuals categorize others based upon their political affiliations. Those on the “other side” of the political divide are often viewed more negatively, perceptions characterized by distrust and animosity; while co-partisans are viewed more positively (Iyengar et al., 2012, 2019; Iyengar & Westwood, 2015). The effect of political ideology on categorizations and affective response to out-group members in turn diminishes collective action. Castiglia (2022) found that political competition reduced collaboration within firms and increased feelings of animosity, which lowered firm innovation. Using field experiments, McConnell et al. (2018) found that consumers prefer dealing with co-partisans.

If political competition does not lead to compromise, an alternative explanation, more aligned with historical perspectives of control and contestation (Hall, 1987; Neem, 2003), is that CPOs, and other nonprofits, may serve as collective action structures that allow in-groups to form across party identities. Associations, such as CPOs, recruit and attract like-minded individuals through social networks (McPherson et al., 2001), allowing CPOs to be internally homogenous within a politically polarized community. Similar backgrounds or value orientations may enable small groups of organization members to act collectively on behalf of the larger divided community.

Regardless, both explanations align with the proposition that nonprofits fill the important political role of finding compromise and resolving competing interests. Foley and Edwards (1996, p. 49) argue, civil society organizations “. . .mediate conflict by hearing, channeling, and mediating the multiple citizen demands. . ..” Perry and Mazany (2014, p. 14) describe a similar role for CPOs. “If you accept the premise of antagonism or identity-based contestation as a defining reality of community, then community foundations are the one institution with the standing, credibility, and capacity to convene the collaborations required to move beyond the paralysis of competing interests and forge agreement. . ..”

Our findings extend these observations by suggesting that in the face of political competition, which may be another form of government failure, CPOs, and other nonpartisan nonprofit organizations, may play mediating roles in divided communities, even in the face of highly contested issues. In contrast to government failure theory’s focus on the role of nonprofits in responding to unmet service needs, our findings focus attention on the political role of CPOs in mediating conflict, which may facilitate cooperation even in increasingly polarized environments. Although we did not start out with the goal of exploring the role of nonpartisan organizations in local governance, our findings suggest future research on the topic is warranted. Future empirical research should examine the relationship between political competition and voluntary action and the processes by which CPOs mediate such conflict. Research emerging from political science and organization studies (Castiglia, 2022; McConnell et al., 2018) may provide insights into theories and methods that can be used to understand the role of voluntary action in politically polarized environments.

Conceptually our study suggests that the debate between government failure theory and sectoral independence theory may not be an either/or choice. Our findings suggest that voluntary action is activated in the face of political polarization (a potential form of government failure) and that political institutions and voluntary action are interdependent. Taking an institutional rather than a resource-dependence perspective allows us to acknowledge that both concepts may be at play in local communities, pointing to the need for additional theory building.

Our findings have implications for policy and practice as well. In the United States, public policy favors voluntary action as a response to public needs; however, political institutions, particularly political engagement and policy signaling, are important in activating voluntary response. Voluntary action embedded in political institutions may reproduce disparities as those communities that foster vibrant political institutions may reap strong voluntary action that supports resiliency. This reproductive cycle may be even more salient when disaster response is politicized as disasters (e.g., wildfires, flooding) are increasingly associated with the contested issues of global warming and climate changes. Reliance upon voluntary action in response to a politically charged disaster may exacerbate inequities that exist in disaster response more generally.

Our study has limitations that suggest the need for future research. First, our study tracks the establishment of a fund but does not consider the timing of the fund (early vs. late actors) or the size of the fund; however, political factors may affect both timing and intensity. Methodologically, we selected the mailing address of the CPO as the geographic unit of analysis. CPOs operate within clearly defined geographic boundaries, which is most often a county or multiple contiguous counties within a state. However, it is possible that their actions are shaped by a regional political culture that aligns with a broader service region. For example, a small number of CPOs serve multiple contiguous counties in a state, and we have thus assumed that neighboring counties share similar political and social characteristics. In addition, there may be other ways to measure these political characteristics. For example, we selected Census return rates as a measure of political engagement. While Census return rates are highly correlated with other measures of cooperative behavior (Martin & Newman, 2015), scholars should examine how other measures of political participation, such as voting rates or attendance at public meetings affect voluntary action. Testing alternative measures might provide insight into the mechanisms by which the political variables support voluntary action. Third, our study examined the political embeddedness of two distinct forms of CPOs, UWs and CFs, which have historically played key roles in convening and facilitating community conversations on a variety of issues (Mazany & Perry, 2013). Additional research is needed to understand how these diverse roles are shaped by political contexts, particularly political polarization. Our model should also be tested in the context of other forms of voluntary action.

In conclusion, the idea that associational life supports the development of democratic institutions has had a global impact on research and practice. This assumption has shaped international development efforts, focusing government air and philanthropic funding on building “strong third sectors” (Krishna, 2002). In the United States and other western democracies, a variety of initiatives have focused on building civil society to strengthen democratic governance and improve service delivery (Bowles & Gintis, 2002; Ressler et al., 2021). However, our findings suggest a more nuanced relationship between government and the nonprofit sector that emphasizes the importance of political institutions in activating voluntary responses. Our findings point to the need for additional theorizing and research to unpack the role that non-partisan organizations play in mediating political conflict in local communities.

Author Biographies

Laurie E. Paarlberg is a professor of philanthropic studies at the Lilly Family School of Philanthropy at IUPUI. She serves as the Charles Stewart Mott Foundation Chair on Community Foundations.

Jin Ai is a doctoral student at the Lilly Family School of Philanthropy at IUPUI. Her research focuses on network analysis.

Megan LePere-Schloop is an assistant professor of public and nonprofit management at the Glenn College of Public Affairs at OSU. She studies organizational change and behavior.

Marlene Walk is an assistant professor of nonprofit management at the Universität Freiburg, Germany. She studies organizational change, human resource management, and organizational behavior.