Introduction

From 2003 to 2013, the number of nonprofit organizations, which fall under 501(c) (3) tax-exempt organization category in the USA, increased 2.8% from 1.38 to 1.41 million (McKeever 2015). These organizations produced goods and services valued at $906 billion, equivalent to 5.4% of GDP in 2013 (McKeever 2015). The number of nonprofit organizations has increased overall, but the growth pattern varies by community characteristics. This led many researchers and practicing managers to examine factors that contribute to the growth of the nonprofit sector. In addition, given the continuously increasing size of nonprofit sector, it will be meaningful to explore the role of nonprofits to meet the needs of communities and find a cooperative relationship between governments and nonprofits. Therefore, this study aims to examine whether government failure, interdependence, or social capital theory better explain the variation in the size of the nonprofit sector among counties.

Three theories are presented to account for the variation in the size of the nonprofit sector. First, Douglas (1987) presents government failure theory and argues that governments are less likely to supply public goods that are driven by heterogeneous preferences. Nonprofit organizations play a role to meet the needs of heterogeneous groups (Salamon et al. 2000). Therefore, the more heterogeneous a community, the more room there is for nonprofit organizations. The second explanation is interdependence theory, which states that governments rely on nonprofits to provide public services, and at the same time, nonprofits need support from governments to fulfill its mission (Gazley 2010). Thus, government support through grants and contracts increases the size of the nonprofit sector. Third, Putnam (2001) describes social capital theory stating that participation in associational activities enhances social capital. Saxton and Benson (2005) find that social capital promotes growth of the nonprofit sector size.

The size of the nonprofit sector has been measured in different ways. Researchers have conducted an empirical analysis to study what determines the size of nonprofit sectors using the number of nonprofit organizations at the county level (Grønbjerg and Paarlberg 2001; Kim 2015; Marcuello 1998; Matsunaga and Yamauchi 2004), at the metropolitan statistical area (MSA) level (Corbin 1999; Lecy and Van Slyke 2013), and at the state level (Abzug and Turnheim 1998; Luksetich 2008). On the other hand, some researchers examine the factors contributing to the number of employees in the nonprofit sector as a proxy for the size of the nonprofit sector (Ben-Ner and Van Hoomissen 1992; Salamon et al. 2000). Ben-Ner and Van Hoomissen (1992) and Salamon et al. (2000) use cross-sectional data in New York and cross-national data to investigate the determinants of the size of nonprofit employment.

This study contributes to the debate on which measures to use in calculating the size of the nonprofit sector within a county by providing evidence that a total number of employees employed by nonprofit organizations are a better measure of the size of the nonprofit sector than the total number of nonprofit organizations. Previous studies have operationalized this variable differently. For example, Grønbjerg and Paarlberg (2001) use the number of nonprofit organizations per 10,000 residents as its measure. Marcuello (1998) defines the size as the number of nonprofits in each of the 40 counties of Catalonia, Spain. Lecy and Van Slyke (2013) use the number of human service nonprofits at the MSA level as a measure of the size. Salamon and Sokolowski (2005), however, argue that employment size is more reliable than the number of nonprofit organizations if the purpose is to capture nonprofit activities. They argue that the number of nonprofit organizations is imprecise because administrative registration systems are rarely updated to reflect an organizational birth and death. Employment data, on the other hand, are collected regularly for administrative purposes and for monitoring economic trends. Furthermore, Salamon and Sokolowski (2005) state that nonprofit organizations are labor-intensive rather than capital-intensive; most of the nonprofit organizations are concentrated in the fields as social service, nursing home care, education, art, and health care. Therefore, the number of employment may be a better indicator for measuring nonprofit activities. Accordingly, this study uses total number of employees in nonprofit organizations within each county to measure the size of the nonprofit sector.

This study is among the first to use longitudinal data to examine variations in the size of the nonprofit sector across communities. Many of the prior studies use cross-sectional data (Abzug and Turnheim 1998; Ben-Ner and Van Hoomissen 1992; Corbin 1999; Grønbjerg and Paarlberg 2001; Kim 2015; Luksetich 2008; Marcuello 1998). Using longitudinal data is more effective in controlling for unobservable county-level characteristics that are especially time-invariant (Wooldridge 2012). Moreover, previous longitudinal studies use data at the MSA level (Lecy and Van Slyke 2013) and at the state level (Matsunaga and Yamauchi 2004). This study, however, focuses on the counties in Indiana. The advantage of regional analysis is that regional samples tend to be more homogeneous and comparable than those of states because they are affected by factors that are common to each county (Bae 2015; Barro et al. 1991).

The paper proceeds as follows. The next section provides a review of the theoretical and empirical literature and presents hypotheses that are used for empirical testing of the theories. The empirical analysis section presents the data, methods, and results. The study concludes by discussing the implications of the results.

Theoretical Background and Literature Review

Government Failure Theory

Government failure theory posits that the provisions of public goods and quasi-public goods are affected by the preference of the majority (Douglas 1983; Grønbjerg and Paarlberg 2001; Weisbrod 1991; Young 2001). Individuals vote to maximize their utility, and politicians determine the direction of their policies to gain more votes (Bassett et al. 1999). Therefore, governments are less likely to supply public goods that are driven by heterogeneous preferences (Douglas 1987).

Nonprofit organizations are established to meet the needs of heterogeneous groups (Salamon et al. 2000), and oftentimes, activities of nonprofit organizations are more prevalent in communities with a more heterogeneous population (Matsunaga and Yamauchi 2004). Under government failure theory: (1) governments cater to majorities, leaving minorities dissatisfied where preferences are heterogeneous; (2) the more heterogeneous the community is, the more heterogeneous community preferences are; so (3) more heterogeneous communities have more preferences unfulfilled by governments, leaving more room for nonprofit organizations to “take up the slack” and meet the demands of those dissatisfied by political outcomes.

The previous studies that test for the government failure theory are not unanimous as to the validity of the theory. (Ben-Ner and Van Hoomissen 1992; Bielefeld 2000; Corbin 1999; Grønbjerg and Paarlberg 2001; Kim 2015; Matsunaga and Yamauchi 2004; Wolch and Geiger 1983). For example, some studies find that social heterogeneity is negatively associated with the size of a nonprofit sector (Costa and Kahn 2003; Paarlberg and Gen 2009; Putnam 2007). The rationale behind these results is that social homogeneity stimulates individual willingness to engage in collective action because people share similar values. Social diversity, on the other hand, induces people to engage in the free-ride behavior (Paarlberg and Gen 2009; Putnam 2007). Costa and Kahn (2003) further note that homogeneous communities are more likely to facilitate agreement on priorities and to provide incentives to participate in civic activities.

Contrarily, some empirical studies find that community diversity increases the size of the nonprofit sector. According to these studies, nonprofit organizations play a key role in meeting the needs of social heterogeneity in response to government failure (Matsunaga and Yamauchi 2004; Salamon et al. 2000). James (1987, p. 412) notes that “NPOs flourish where excess demand exists, in the face of limited government production, and/or where intensely felt preferences over product variety (religious, linguistic, quality) make people willing to pay a higher price for a differentiated product.”

There are some researchers that find mixed results. Ben-Ner and Van Hoomissen (1992), for instance, find that racial diversity increases nonprofit educational facilities, but decreases social services. Corbin (1999), on the other hand, finds the statistical association between racial diversity and social services only.

We posit that population heterogeneity is negatively related with the supply of the nonprofit sector and positively related with the demand of the nonprofit sector. This study hypothesizes that the positive effect of the demand side is larger than the negative effect of the supply side.

Based on the arguments presented by prior research, we test for the following hypothesis as to the government failure theory.

Hypothesis 1

Racial diversity is positively associated with nonprofit sector size.

In addition to meeting the needs of diverse population groups, nonprofits have responded to the various demands of distressed segments of the population in household income level. This perspective reflects the traditional role of nonprofits as charities. For example, Wolch and Geiger (1983) argue that people who live in affluent community support nonprofits to fulfill the needs of their distressed neighbors. And they find that a larger number of nonprofit organizations are observed in communities with a higher level of poverty. Corbin (1999) also finds a positive relationship between poverty level and nonprofit sector size. Zakour and Gillespie (1998) and Marsh (1995), however, find that nonprofit organizations are more prevalent in rich neighborhoods than in poor communities.

In this study, we test for the hypothesis that resources such as income or wealth will lead to a larger nonprofit sector because middle- and upper-income patrons are more likely to provide services or fund their community.

Hypothesis 2

Poverty level is negatively associated with nonprofit sector size.

Interdependence Theory

The government failure theory suggests that if governments fail to satisfy its citizens’ preferences, then it will promote activities of the nonprofit sector. The interdependence theory, on the other hand, hypothesizes a more cooperative relationship between governments and nonprofits (Salamon et al. 2000). The theory suggests that governments rely on nonprofits to provide public services, and at the same time, nonprofits need support from governments to fulfill its mission (Gazley 2010). Accordingly, governments provide grants and make contracts with nonprofits to deliver public services (Grønbjerg and Paarlberg 2001). Thus, the size of the nonprofit sector is more likely to increase as a result of local, state, or federal funding.

Many studies, conducted at the international, state, MSA, and county levels, test for the validity of the interdependence theory and find that the size of the nonprofit sector increases as governments spend more and provide aids to nonprofit organizations (Salamon et al. 2000; Luksetich 2008; Lecy and Van Slyke 2013; Grønbjerg and Paarlberg 2001; Kim 2015; Matsunaga and Yamauchi 2004).

Using the data from Western, Central and Eastern Europe, Latin America, and other developed countries, Salamon et al. (2000) find a positive relationship between government support and size of nonprofits. In their study, government support is defined as the share of all nonprofit revenues resulting from government grants. Luksetich (2008), defining government support in terms of the total dollar value of government grants, finds the same results.

There are some studies that examined the association between government funding and the size of the nonprofit sector using regional data (Grønbjerg and Paarlberg 2001; Kim 2015; Lecy and Van Slyke 2013; Matsunaga and Yamauchi 2004). Lecy and Van Slyke (2013) find that the amount of government grants is positively correlated with the number of nonprofits at the MSA level for years between 1998 and 2003. Matsunaga and Yamauchi (2004) also find that public subsidies are positively related to the size of nonprofits. Grønbjerg and Paarlberg (2001) further find the positive impact of federal and local funding on the size of the nonprofit sector.

Based on the aforementioned findings, this study tests for the following hypotheses as to interdependence theory.

Hypothesis 3

Federal funding is positively associated with nonprofit sector size.

Hypothesis 4

Local funding is positively associated with nonprofit sector size.

Social Capital Theory

The definition of social capital is not agreed upon among researchers. Bourdieu and Wacquant (1992, p. 119), for example, conceptualize social capital as a network-dependent resource pool and define it as “the sum of the resources, actual or virtual, that accrue to an individual or a group by virtue of possessing a durable network of more or less institutionalized relationships of mutual acquaintance and recognition.” Fukuyama (1995), on the other hand, examines social capital in the context of economic development and defines it as a “set of informal values or norms shared among members of a group that permits them to cooperate with one another” (Fukuyama 1999, p. 16). Finally, Putnam (1995) combines these two views and defines social capital as “features of social organization such [as] networks, norms, and social trust that facilitate coordination and cooperation for mutual benefit” (Putnam 1995, p. 67). Individuals manifest social capital through participation in associational activities according to the findings from Putnam (2001). Rupasingha et al. (2006) also use the number of associations such as civic groups, sports clubs, labor unions, and political organizations at the county level to measure social capital.

Most of the previous studies find that social capital is significantly related with social, economic, and political phenomena (Salamon et al. 2000). For example, social capital is found to be positively correlated with school performance (Putnam 2001), the size of governments (Putnam et al. 1994; Schafft and Brown 2000), and economic development (Rupasingha et al. 2006).

Some studies engage in examining the effects of social capital on the size of the nonprofit sector (Kim 2015; Saxton and Benson 2005). Kim (2015) finds a positive but insignificant relationship between social capital—measured by per capita crime rate, such that less crime means more social capital—and the number of nonprofit organizations. Saxton and Benson (2005) find that certain forms of social capital promote growth in the number of nonprofits. They measure social capital as diversity in friendships, political engagement, giving and volunteering, associational involvement, informal social networks, and social trust. While their study shows a positive impact of diversity in friendships and political engagement on the growth of the nonprofit sectors that are statistically significant, social trust is found to be negatively related to the nonprofit sector size (Saxton and Benson 2005).

Based on the findings mentioned above, this study hypothesizes that the number of associations is positively related with the size of nonprofits.

Hypothesis 5

The number of associations is positively associated with nonprofit sector size.

Empirical Analysis

Data

This study uses county-level data for the state of Indiana. We choose counties within Indiana as the unit of analysis because the state of Indiana is ranked 28th in the number of 501(c) 3 nonprofits per 10,000 persons, which is approximately an average size among all U.S. states. Furthermore, data regarding the number of nonprofit employees at the county level are available in the Indiana Nonprofit Employment reports.

Annual data on nonprofit employment come from the Indiana Nonprofit Employment reports (Grønbjerg and Park 2001; Grønbjerg and Eschmann 2005; Grønbjerg et al. 2007, 2009). Reports were published biannually. The report for year 2007 is not available. County library expenditure, a proxy for local funding, is obtained from the Indiana Business Research Center. Other county demographic variables are retrieved from the U.S. Census Bureau.

The Indiana Nonprofit Employment reports focus on nonprofit organizations that are exempt from federal income tax under section 501(c) in Indiana (Grønbjerg et al. 2009). Table 1 shows the number of nonprofit employees per 10,000 residents by the county as of 2009. The average number of nonprofit employees per 10,000 residents is 241.37, with St. Joseph County being the largest in the number of nonprofit employees. Figure 1 displays growth in the number of nonprofit employees in Indiana. The graph shows that the growth is quite remarkable from 2001 to 2009, and the number of employees increased from 220,617 to 243,444, a 10.3% increase.

Table 1 Number of nonprofit employees per 10,000 residents in Indiana by County in 2009
Full size table
Fig. 1
figure 1

Nonprofit employee growth in Indiana

Full size image

Measures

Dependent Variable

This study defines the size of the nonprofit sector as a total number of nonprofit employees in each county in Indiana.Footnote 1 A total number of nonprofit employees include both part-time and full-time employees but exclude nonprofits employing less than four workers in 20 weeks during a year (Grønbjerg et al. 2009, p 34). The employment data source is the Quarterly Covered Employment and Wage (QCEW) program, administered by State Employment Security Agencies and the Bureau of Labor Statistics. Thus, the number of the nonprofit employment in the report would underestimate the actual size of nonprofit employment. Note, however, that such nonprofits accounts for only 1.4% of all nonprofit employees (Grønbjerg et al. 2009). Therefore, we argue that the numbers reported in the reports are quite accurate.

Independent Variables

In order to examine the effect of many of the community factors on the size of the nonprofit sector, this study analyzes several indicators based on the government failure, interdependence, and social capital theory.

The government failure theory predicts that the number of nonprofits grow in response to unmet needs induced by preference heterogeneity and need heterogeneity. This study uses two variables—racial diversity and poverty level—as proxies for preference heterogeneity and need heterogeneity. Racial diversity is a continuous variable taking a value equal to the Blau heterogeneity index. Blau (1977) adopted the heterogeneity index to estimate diversity among groups, calculated as 1 − Σp 2 i where p is the proportion of the population consisting of members of group i. Poverty levelFootnote 2 is a continuous variable taking the value of the proportion of children in a community who are living in poverty, following Grønbjerg and Paarlberg (2001)’s study.

Interdependence theory predicts that nonprofits seek government funding in return for providing services that government would otherwise have to provide and thus increase in size as more funding becomes available. This study uses two variables, federalFootnote 3 and local funding, as proxies for funding availability. Federal funding is a continuous variable that denotes the amount of federal grants and contracts awarded in each county during the fiscal year. Local funding is a continuous variable that measures the value of public library funding in each county during the fiscal year. The previous study adopted the library funding as a proxy variable to measure the willingness of county residents as to taxing themselves to support collective actions (Grønbjerg and Paarlberg 2001, p. 695).

In order to measure social capital, this research uses the number of associations in a county. AssociationsFootnote 4 include (1) bowling centers, (2) golf clubs, (3) fitness centers, and (4) sports organizations. This study counts the number of associations based on the information provided in the North American Industry Classification System code of Indiana.

Control Variables

In addition to the factors listed above, many other macroeconomic or demographic factors not tied to any particular theory of nonprofit organizations may nonetheless affect the nonprofit sector size. Research to date has controlled for factors such as population (Kim 2015; Lecy and Van Slyke 2013; Saxton and Benson 2005), median household income (Kim 2015; Lecy and Van Slyke 2013), unemployment rate (Kim 2015; Saxton and Benson 2005), and elderly population—population aged 65 or above (Kim 2015). Moreover, to control for community characteristics, our model also includes population, median household income, the share of elderly population, and the number of hospitals and universities. In the regression model, we include population and population squared to capture a nonlinear relationship between population and the number of nonprofits, if any. A rural dummy variable, which is equal to one if the largest city’s population is less than 10,000, is further added to our models to control for county types. Finally, we add the number of hospitals and universities in each county in our models because the number of nonprofit organizations is more likely to be prevalent in a community where hospitals and universities are present.

Model Estimation

Given the discussion mentioned above, we estimate the regression model to examine the relationship between community factors and the size of the nonprofit sector using the county-level data in Indiana that span from 2001 to 2009. Based on the Hausman test results, we estimate random effects models (Wooldridge 2012). For the estimation, we use the generalized least squares model. The model specification is as follows:

$$\begin{array}{*{20}l} {Ln\left( {TOTAL \,\,NUMBER\,\, OF\,\, NONPROFIT \,\,EMPLOYEES_{it} } \right) = \beta_{0} + \beta_{1} \,\,RACIAL \,\,DIVERSITY_{it} } \hfill \\ { +\, \beta_{2} \,\,POVERTY \,\,LEVEL_{it} + \beta_{3} Ln\left( {FEDERAL\,\, FUNDING _{it} } \right) + \beta_{4} Ln\left( {LOCAL\,\, FUNDING_{it} } \right)} \hfill \\ { +\,\beta_{5} Ln(NUMBER\,\, OF \,\,ASSOCIATIONS_{it} ) + \beta_{6} \left( {POPULATION_{it} } \right)} \hfill \\ { + \,\beta_{7} \left( {POPULATION\,SQUARED_{it} } \right) + \beta_{8} Ln\left( {MEDIAN\,\, INCOME_{it} } \right)} \hfill \\ { +\, \beta_{9} PROPORTION\,\, OF\,\, OLD \,\,AGE_{it} + \beta_{10} NUMBER \,\,OF \,\,HOSPITALS \,\,AND\,\, UNIVERSITIES_{it} } \hfill \\ { +\, \beta_{11} \,\,RURAL_{it} + \beta_{12} UNEMPLOYMENT_{it} + YEAR\,\, FIXED \,\,EFFECTS + \varepsilon_{it} }. \hfill \\ \end{array}$$

Results

Table 2 reports descriptive statistics for all the variables used in this study.

Table 2 Descriptive statistics
Full size table

Table 3 displays the results. This study analyzes the effects of community factors on the size of the nonprofit sector using five models. Model 1 is a random effects model estimating the effects of racial diversity, population, population squared, median household income, proportion of old age, number of hospitals and universities, rural, and unemployment rate on the size of nonprofit sector. Model 2 also adds proportion of children below poverty level to test government failure theory. Model 3 adds federal funding and Model 4 analyzes the effect of local funding to examine interdependence theory. Model 5 tests for social capital theory by adding the number of associations.

Table 3 GLS models of the number of nonprofit employees
Full size table

Hypothesis 1 predicts that racial diversity is positively associated with nonprofit sector size. The parameter estimates on racial diversity are positive and significant across five models (Model 1: \(\beta\) = 2.471, p < 0.01, Model 2: \(\beta\) = 2.639, p < 0.01, Model 3: \(\beta\) = 2.104, p < 0.01, Model 4: \(\beta\) = 1.200, p < 0.1, Model 5: \(\beta\) = 0.943, p < 0.1). This indicates that the models support Hypothesis 1. These results are consistent with those of James (1987), Ben-Ner and Van Hoomissen (1992), Kim (2015), and Corbin (1999), but are different from that of Grønbjerg and Paarberg (2001).

Hypothesis 2 examines that poverty level is negatively associated with nonprofit sector size. The results reveal that poverty level has a negative and significant relationship with nonprofit sector size across four models (Model 2: \(\beta\) = −0.937, p < 0.1, Model 3: \(\beta\) = −1.144, p < 0.05, Model 5: \(\beta\) = −1.166, p < 0.05). Consequently, these findings support Hypothesis 2. This is consistent with the results of Zakour and Gillespie (1998), Marsh (1995), and Bielefeld (2000), which show poverty level is negatively associated with the number of nonprofit social service providers.

Hypothesis 3 predicts that the amount of federal funding is positively associated with nonprofit sector size. The parameter estimate on Federal funding is positive and significant across Model 3, Model 4, and Model 5, supporting Hypothesis 3 (Model 3: \(\beta\) = 0.070, p < 0.01, Model 4: \(\beta\) = 0.087, p < 0.01, Model 5: \(\beta\) = 0.096, p < 0.01). These results are consistent with previous findings with respect to federal funding at the state level (Luksetich 2008), MSA level (Lecy and Van Slyke 2013), and the county level (Grønbjerg and Paarlberg 2001; Kim 2015; Lecy and Van Slyke 2013; Matsunaga and Yamauchi 2004). Hypothesis 4 examines that the amount of local funding is positively associated with nonprofit sector size. Local funding is significant across the two models (Model 4: \(\beta\) = 0.131, p < 0.01, Model 5: \(\beta\) = 0.153, p < 0.01). Consequently, the findings support both Hypothesis 3 and Hypothesis 4.

Hypothesis 5 expects that the number of associations is positively associated with nonprofit sector size. The parameter estimates on the number of associations are insignificant for the GLS models. The findings do not support the social capital theory of nonprofit growth. This is inconsistent with findings by Saxton and Benson (2005) that diversity of friendships correlates positively with nonprofit growth as well as findings by Kim (2015) that crime per capita—high crime is a proxy for low social capital—has a negative relationship with the size of nonprofits.

Several control variables turned out to be statistically significant in our models. The population variable and its squared term are statistically significant, suggesting the nonlinear relationships between the population and the number of nonprofit employees. Both median household income and the rural variables are negatively related with the number of employees. These results support the argument that the number of nonprofit employees low in counties with low median household income and less population density. We also find that the number of hospitals and universities is positively correlated with the number of nonprofit employees.

Discussion and Implications

The purpose of this study is to examine the association between community factors and the number of nonprofit employees and to determine whether the findings support the government failure, interdependence, and social capital theory.

Using the panel data (92 counties in Indiana from 2001 to 2009), we find that racial diversity, federal funding, and local funding are positively associated with the size of nonprofit sectors. We also find that the share of children below poverty level is negatively associated with the size of nonprofit sectors. Our findings support the interdependence and government failure theory.

According to the government failure theory, governments provide public goods to satisfy the preference of the majority. Thus, nonprofit organizations provide additional resources and services for minority groups in communities that governments are unable to fulfill. Our findings confirm such arguments. We find that racial diversity is positively related with the size of nonprofit sectors in all models, supporting government failure theory. The size of nonprofit sectors is larger in counties with a higher racial diversity than in counties with a lower racial diversity. Previous studies that examine the government failure theory also found a negative association between the share of children below poverty level and the size of nonprofit sector (Marsh 1995; Zakour and Gillespie 1998). These studies show that the size of nonprofit sector is larger in communities with a higher proportion of the affluent residents, based on the argument that middle- and upper-income patrons play a role for human resource in each community. Our results are in the same line with previous studies that support the government failure theory.

The interdependence theory states that governments rely on nonprofit organizations to provide public services and nonprofit organizations need government support to fulfill their services (Salamon et al. 2000). Thus, governments provide grants and funds to nonprofit organizations to deliver public services (Gazley 2010). Therefore, according to this theory, increase in government funding leads to the expansion of the size of nonprofit sectors. (Grønbjerg and Paarlberg 2001; Kim 2015; Lecy and Van Slyke 2013; Luksetich 2008; Matsunaga and Yamauchi 2004; Salamon et al. 2000). Lecy and Van Slyke (2013) and Matsunaga and Yamauchi (2004) find that government funding positively correlated with the size of nonprofit sectors. Grønbjerg and Paarlberg (2001) also examine the effects of federal funding and local funding on the size of the nonprofit sector, but they do not find statistically significant results. Our findings show that federal funding and local funding is positively associated with the size of the nonprofit sector, supporting the interdependence theory. According to Austin (2000), many nonprofit organizations face fundraising challenges because of bad economic conditions and board member interests. Hence, the interdependence theory suggests that federal and local governments support will help nonprofit organizations to fulfill their jobs in a manner that is consistent with their goals.

Based on the social capital theory, we test the hypothesis that the number of associations is not associated with the size of the nonprofit sector. Previous studies supporting social capital theory argue that social capital allows individuals to form new nonprofit organizations (Saxton and Benson 2005). But our findings do now allow us to reject this hypothesis, implying that social capital does not affect the size of the nonprofit sector.

With an increase in the needs of public services, governments cannot fully provide these services to satisfy diverse needs due to budget limitation. In addition, although nonprofits increasingly address many pressing social needs like homeless or poverty, they may lack the resources to provide social services. Furthermore, governments receive support from the nonprofit sector which provides public services for diverse population groups. Nonprofits in return receive stabilized funding resources from the government which help to enhance the scope of community services.

Future research can build on this study in several ways. First, the results may not be sufficient to generalize to other context because Indiana may be different from other states, especially coastal and southern states. The racial–ethnic and/or religious demographics of these states may be different from the states of the Midwest. Although focusing on one regional sample has several advantages such as fewer confounding variables, more studies using other states are necessary to increase generalizability.

Second, the financial crisis of 2008 influenced employment both in the for-profit sector and nonprofit sector. The data used in this study include the number of employees in nonprofit sectors for 2008. Thus, this may have caused a bias in measuring the relationship between community factors and the size of nonprofit sectors. Although this study adds the unemployment rate variable to control for economic conditions, the effect of economic crisis might not have been fully controlled for. Future studies need to seek ways to control for such confounding factors through quasi-experimental research designs.

Third, the racial diversity variable may not accurately measure the racial characteristics of Indiana because the majority of residents in Indiana are Whites. Recent data show that the population of Indiana is becoming increasingly diverse. According to the estimates from U.S. Census Bureau (2010), there has been a drastic growth in the Hispanic population—43% of Indiana’s population growth over the ten-year period. The Asian population also increased sharply in recent years. Future studies, therefore, may reexamine the effect of racial diversity on the nonprofit sector size using data spanning a longer period.

Fourth, future research may focus on analyzing the effects of community factors by the type of nonprofit organizations such as arts, education, hospitals, and human services. This study has limitations in fulfilling this goal due to lack of information in the Indiana Nonprofit Employment reports.

Lastly, this study used federal funding, the amount of federal grants and contracts in each county during the fiscal year, and local funding, public library expenditure in each county during the fiscal year as proxies for funding availability based on the previous studies. Future studies, however, need to adopt the new federal and local funding variables to measure the effects of government funding on the size of nonprofit sector, such as the amount of direct federal and local funding to nonprofit organizations in each county. Furthermore, this study used the number of associations only in order to measure social capital because of the data limitation. It would be desirable to use indicators to measure social capital, such as social trust, informal social networks, and associational involvement.