The Antecedents of Nonprofit Board Interlock: A Longitudinal Examination on Network Structure,Homophily,and Organizational Attributes

Abstract

A board interlock creates interorganizational networks where organizations are interconnected via overlapping board of directors. Board interlock is important for nonprofits because of its potential to impact organizational performance through the flow of information, resources, and status. While much is known about the consequences of board interlock, little is known about the mechanisms underlying its antecedents. This study explores three types of predictors of board interlock: organizational, dyadic, and structural characteristics. Inferential network analysis of a 17-year-period panel of nonprofits demonstrates that network relationships are shaped by the existing network structures, such as the tendency for preferential attachment (e.g., a social preference to connect with those who are already well connected) and transitivity (e.g., a social preference to connect with friends of friends). Findings inform nonprofit leaders about how to bridge to a board interlock network by recruiting well-connected board members serving on multiple boards.

Keywords

board interlock nonprofit board governance inferential network analysis board of directors board recruitment

Introduction

The composition of a board of directors has become increasingly important as they play a crucial role in supporting organizational performance for their oversight over management, provision of knowledge and advice, and linking organizations to important stakeholders (Abzug & Galaskiewicz, 2001; Brown & Guo, 2010; Miller-Millesen, 2003). This study contributes to board governance literature through a study of board interlock, which occurs when an executive and/or board member at one organization joins the board of another organization (Mizruchi, 1996).

A board interlock creates a network relationship in which organizations are connected through these overlapping board of directors. In these interorganizational networks, overlapping board members constitute ties that link organizations. Initial studies reveal that a small group of individuals extensively sit on boards of economic, political, and nonprofit institutions and serve as carriers of rules and practices in the community (Moore et al., 2002; Vidovich & Currie, 2012). Although corporate board interlock has been studied extensively, little is known about board interlock in nonprofit literature. This is a glaring gap given the critical differences in the characteristics between corporate and nonprofit boards, including engagement in the scope of board activities, board size, and primary motivation to join boards (Epstein & McFarlan, 2011; Miller-Millesen, 2003). Thus, these sectoral differences in board characteristics could have implications on board composition, which may influence the process and the outcomes of board interlock.

A growing number of scholars have begun to explore interorganizational networks via board members in nonprofit organizations. In a recent review of literature on antecedents and consequences of board interlock in corporate and nonprofit sectors, Yoon (2021a) notes that while corporate board interlock studies focus on the role of interlocking board members in channeling information and resources, nonprofit board interlock studies pay attention to investigating the role of interlocking board members in signaling the status of the organizations. For instance, scholars report that interlocking board members influence corporate governance outcomes such as diffusion of management practices by channeling information to top management (Bouwman, 2011; Connelly et al., 2011) and financial performance by improving monitoring effectiveness of the boards (Mizruchi & Stearns, 1994). In contrast, scholars examining nonprofit board interlock attend to outcomes such as the acquisition of foundation grant funding through the signaling function of interlocking board members (Esparza & Jeon, 2013; Faulk et al., 2016, 2017; Paarlberg et al., 2020). In a recent study, Yoon (2021b) and Bloch and colleagues (2020) showed empirical evidence that nonprofits with interlocking board members report more extensive use of governance practices targeted to enhance governance operations. Related studies also emphasize how nonprofit networks contribute to organizational performance, growth, and survival through an external representation function of network ties (Galaskiewicz et al., 2006; Hager et al., 2004; Pfeffer & Salancik, 1978).

Although prior studies provide useful knowledge related to the consequences of board interlock, little is known about the formation process. To date, only one study has examined the formation process of board interlock in nonprofit organizations. Willems and colleagues (2015) provide cross-sectional evidence on predictors of board interlock by showing that homophily between the two nonprofits in terms of similarity in operational activities, size of assets, and funding structures tends to create interlock relationships in the context of nonprofit sectors in Belgium.

This study examines the predictors of board interlock in U.S. nonprofit organizations in a longitudinal context. This type of study extends the literature on the antecedents of nonprofit network relationships connected by board members (Guo & Acar, 2005; Willems et al., 2015; Yoon, 2021a). Focusing on board recruitment perspective, I test three groups of characteristics that relate to the formation of interlock relationships: organizational, dyadic, and structural characteristics. In particular, I bring attention to structural characteristics that reflect certain social mechanisms in the network formation process. For instance, some networks may be influenced by a structural tendency toward preferential attachment (e.g., a social preference to connect with those who are already well connected), while others are influenced by transitivity (e.g., a social preference to connect with friends of friends). Although much is known about organizational and dyadic characteristics, nonprofit scholars lack knowledge in understanding the structural characteristics related to board recruitment practices within a given network. This is due, in part, to the methodological challenges that impede scholarly efforts to model network relationships influenced by board composition. Collecting and analyzing longitudinal network data on boards of directors require intensive efforts to generate measures of nonprofit network relationships based on the board appointment status.

I take the first step to bring attention to the structural predictors of board interlock through board composition in a longitudinal network context. The central premise of this study is that nonprofit board governance should be understood in relation to interorganizational relationships where organizations are embedded in a broader structure that influences their actions and outcomes (Granovetter, 1985). Here, network embeddedness describes the extent to which a nonprofit organization is interconnected to other nonprofits through overlapping board members. The significance of network embeddedness has been widely recognized in a variety of network relationships such as alliance and collaboration (Ahuja et al., 2009; Gulati & Gargiulo, 1999; Powell et al., 2005). As nonprofit boards of directors play a critical role in organizational behavior and performance, it is important for both scholars and practitioners to understand how board interlock network relationships emerge and how they might be better managed to produce improved governance outcomes.

In this study, I examine the dynamics of board interlock using a unique panel of 501(c)(3) public charities in three Upstate New York cities from 1998 to 2014. An inferential network analysis known as a temporal exponential random graph model is employed to control and assess interdependencies and intertemporal dependencies inherent in the longitudinal network. The rest of this article is organized as follows. First, I provide a background on the concept of board interlock. Second, I review the relevant literature on the predictors of board interlock and present hypotheses. Third, I discuss data, measures, and empirical strategy. Finally, I conclude with a discussion of the results and their implications.

Board Interlock

A network consists of components called nodes and the relationship between the nodes called as ties (Wasserman & Faust, 1994). Figure 1 provides the visualization of two sets of nodes in a board interlock. As seen from the board member–nonprofit relationship (panel A), the first node (yellow circle) is a set of board members, and the second node (green square) is a set of nonprofits. Following existing studies (Benton, 2016; Kim et al., 2016), I transform the board member–organization relationship into an organization–organization relationship (panel B), where nonprofit organizations are the nodes and overlapping board members are the ties that connect the nodes. The inferred nonprofit–nonprofit relationship is obtained by connecting the two nonprofits if they share any overlapping board members. The focus of the study is to predict the conditions under which a board interlock occurs, that is, a tie created when an organization invites a board member from another organization.¹

Figure 1.

Inferred organization–organization relationship. (A) Board member–organization relationship. (B) Organization–organization relationship.

Predictors of Board Interlock: From Organization to Dyad to Structure

Organizational Predictors of Board Interlock

Resource dependency perspective proposes that resource characteristics induce organizations to establish network relationships to gain greater control over the resource environment (Pfeffer & Salancik, 1978). Based on this perspective, board interlock is one way to alleviate critical resource dependencies through the development of network relationships. Interlocking board members play an important role in managing resource dependencies and obtaining critical resources (Galaskiewicz, 1985; Pfeffer & Salancik, 1978). I focus on resource vulnerability and resource dependency that may influence nonprofits to create interlock relationships.

First, organizations faced with increased environmental uncertainty may likely create an interlock tie as a financial coping strategy to reduce resource vulnerability. Supporting this argument, scholars suggest that the availability of financial resources is an important consideration for studying nonprofit collaboration relationships (Gazley & Guo, 2020; Guo & Acar, 2005; MacIndoe & Sullivan, 2014). Also, evidence shows that resource vulnerability is positively associated with corporate board interlock to reduce resource constraints (Dooley, 1969; Mizruchi & Stearns, 1994). Although there is no empirical evidence on the link between resource constraints and board interlock in the nonprofit literature, based on the findings from corporate board studies, I expect to observe a positive association between resource vulnerability and the presence of any board interlock in nonprofit organizations.

Hypothesis 1 (resource vulnerability): A nonprofit with greater financial vulnerability is more likely to form board interlock.

Second, the emergence of board interlock may be influenced by resource dependence on certain sources of revenue (Mizruchi & Stearns, 1994). Studies suggest that nonprofits that heavily rely on donative sources of funding—called donative nonprofits (Chang & Tuckman, 1994; Hansmann, 1986; Weisbrod, 1988)—are different from nonprofits that rely on commercial sources of funding because of their unique funding strategy to procure external resources (Galaskiewicz et al., 2006). Board interlock may be more critical for donative nonprofits because a greater dependence on external sources of capital may lead them to seek access to potential resources by recruiting members who sit on many other boards. Furthermore, donative nonprofits could benefit from the “social status” board interlock renders (Podolny, 1994), as donor groups look to a board of directors to make charitable contribution decisions (Galaskiewicz & Wasserman, 1989). In particular, a greater level of ambiguity surrounding the evaluation of services provided by donative nonprofits may make social status rendered by the board interlock ties become even more pronounced (Podolny, 1994). Thus, I expect to observe a positive association between the level of resource reliance on donative revenue sources and the presence of board interlock.

Hypothesis 2 (resource dependency): A nonprofit with greater financial dependence on contributions is more likely to form board interlock.

Dyadic Predictors of Board Interlock

Past research suggests that dyad-specific characteristics such as similarities in resource endowments or geographic location are important predictors of network relationships (Atouba & Shumate, 2015; Willems et al., 2015). This is because similarities in organizational attributes may reduce transaction costs associated with the maintenance of network relationships (McPherson & Smith-Lovin, 1987). Moreover, organizations tend to look for relationally homophilous partners especially when the extent of environmental uncertainty arises in the decision-making process (Galaskiewicz & Shatin, 1981; Podolny, 1994). I focus on industry homophily, which occurs when interlock relationships are formed between the two nonprofits that provide services in the same industry. Industry homophily may emerge because of a board of directors’ social preference to recruit members who already have board experience at other similar nonprofits, based on a belief that they have a better understanding of the nature of services and the operations of the organizations. For instance, having board experiences in similar organizations may provide individuals with analogous frames of reference for better decision-making (Cornforth & Edwards, 1999). Evidence suggests that corporate organizations often appoint new board members with experience on boards of similarly governed organizations (Bouwman, 2011). Therefore, I expect as below.

Hypothesis 3 (industry homophily): Nonprofit board interlock is more likely to occur between nonprofits from the same industry.

Structural Predictors of Board Interlock

The emergence of a board interlock may also be shaped by structural patterns of interactions (Galaskiewicz & Wasserman, 1989) where network relationships among organizations are influenced by the presence of other ties in the network (Snijders et al., 2006). I pay attention to two structural mechanisms that describe unique social processes through which existing network ties influence the potential for future ties.

First, preferential attachment refers to an inclination to connect with well-connected actors (Barabási & Albert, 1999). Preferential attachment may occur in the nonprofit context because of an existing board of directors’ proclivity to recruit sought-after candidates serving on many other boards. Evidence suggests that well-connected actors are likely to get more referrals because of the higher social visibility and reputation transmitted through a central position in the network (Podolny, 1994). Furthermore, sitting on many boards may reflect candidates’ knowledge, experience, and social connections (Chu & Davis, 2016). Evidence shows that board members affiliated with many other boards have an advantage in receiving additional board appointments (Benton, 2016; Koskinen & Edling, 2012). Thus, I expect that nonprofit board interlock emerges through preferential attachment.

Hypothesis 4 (preferential attachment): A group of nonprofits with a higher degree of affiliation are more likely to attract additional board interlock.

Second, transitivity refers to three-node network configurations in a “closed triad” where two nodes are likely to form a tie if each of them is tied to a separate common third node (Wasserman & Faust, 1994). Evidence suggests that new interorganizational network relationships arise when two actors are introduced by a third actor with whom both of them already have a relationship with (Park & Rethemeyer, 2012). Prior corporate board interlock studies also suggest that transitivity is an important driver of board interlock formation (Benton, 2016; Koskinen & Edling, 2012). Transitivity may emerge through the recruitment process in which board members are recruited based on a referral system where candidates are selected from social contacts or recommendations of the existing boards (Lorsch & Young, 1990). This is, in part, because the cost associated with a candidate search may induce organizations to choose members based on their social networks. An indirect tie establishes a frame of reference which may enable the existing boards to vouch for prospective candidates (Isett & Provan, 2005). Furthermore, staffs and board members in nonprofit organizations often belong to the same social, religious, or cultural clubs and other local networks (Galaskiewicz, 1985), which may precipitate transitivity in the board recruitment process. Therefore, I expect as follows.

Hypothesis 5 (transitivity): Nonprofit board interlock is driven by transitivity where a group of three nonprofits is connected.

Data and Sample

I use two data sets, including board-level data which are hand-collected from the Internal Revenue Service (IRS) tax filings, and organization-level data compiled by the National Center for Charitable Statistics (NCCS) Core Files. The study sample includes 501(c)(3) public charities reporting IRS 990 tax filings in three Upstate New York cities from 1998 to 2014. As the analytic method used in this study requires a panel dataset for better estimation, I purposefully target sample nonprofits that operate every year during the study period to capture dynamic patterns of nonprofit networks through board members. Among the 424 organizations operating during a 17-year period, 30 organizations were dropped from the study sample because of implausible financial values. This process led me to a balanced panel of 394 nonprofits with complete data for analysis. A considerable longitudinal dimension of the study allows me to identify important network patterns of sample nonprofits in the regional context of the Upstate New York area.

Network visualization of how a board interlock network evolves over time in three cities is presented in Figure 2A to 2C. In the whole network maps presented, nodes are nonprofit organizations and ties are overlapping board members connecting those organizations via board affiliations. The dynamic changes observed in the network structure demonstrate that the connectedness of nonprofit organizations becomes more or less cohesive over time, depending on the city.

Figure 2.

(A) Board interlock network map, Albany, 1998–2014. (B) Board interlock network map, Syracuse, 1998–2014. (C) Board interlock network map, Utica, 1998–2014.

Measurement

Dependent Variable

The dependent variable is the existence of a board interlock tie between two nonprofits connected through overlapping board of directors within a given year from 1998 to 2014. The unit of analysis is the dyad. For each dyad-year observation by the city, I look at the dyads among the 394 sampled nonprofits during the study period. In this study, the term “board” is used inclusively to refer not only board members but also other people listed in the IRS 990 tax filings including board directors, officers, and/or trustees. Using name information listed in tax filings (Part V in old IRS 990 filings; Part VII in new filings), I identify the name of any individual in one nonprofit that appears on the list of at least another nonprofit in the study sample. In doing so, the approximate string matching approach is used to match individuals’ names between organizations by year (see Ihm & Shumate, 2019). A binary indicator is coded as “1” if the two nonprofits share any overlapping board of directors and as “0” if the two nonprofits do not share one. This yielded a total of 558,180 board interlocks during the 1998–2014 period in the three cities.

Explanatory Variables

I follow the modeling approach used in prior research to properly capture a variety of features such as organizational, dyadic, and structural characteristics that are most widely used in the existing literature (Robins et al., 2007; Snijders et al., 2006). Organizational characteristics capture the tendency for organizations with certain attributes to be more active in seeking board interlock, while dyadic characteristics capture the tendency for pairs of organizations with shared characteristics to have board interlock. Finally, structural characteristics capture the tendency for existing board interlock interconnections related to future network connections.

Organizational Characteristics

I include financial attributes that may influence the nonprofits’ propensity to initiate or maintain network relationships. To measure resource vulnerability, I follow the research tradition established by Tuckman and Chang (1991) and use two variables, including the level of liability over the total asset and the operating margin (Chang & Tuckman, 1994; Hager, 2001; Hodge & Piccolo, 2005). The first financial vulnerability variable is operationalized as the end-of-year total liability divided by the total asset. The second financial vulnerability variable is operationalized as the operating profit margin divided by the total revenue.

To measure resource dependency, I employ a binary variable of donative nonprofit that captures whether an organization heavily relies on revenue raised from contributions, gifts, and grants. This variable is coded as “1” if the share of total contributions (Part I, Line 1e from old IRS 990 filings; Part VIII, Line 1h from new filings) divided by the total revenue is over 60%, and coded as “0” otherwise (Chang & Tuckman, 1994). Although not used as key variables, I also use three additional variables to control for the extent of reliance on other types of funding sources. First, I include the program service revenue source variable operationalized as the ratio of program service revenue (Part I, Line 2 from old IRS 990 filings; Part VIII, Line 2g from new filings) divided by the total revenue. Second, I incorporate the investment income source variable operationalized as the ratio of the total investment income (sum of Part I, Lines 4, 5, and 7 from old IRS 990 filings; Part VIII, Line 3a from new filings) divided by the total revenue. Third, I add the rental income source variable operationalized as the ratio of the net rental income (Part I, Line 6c from old IRS 990 filings; Part VIII Line 6d from new filings) divided by the total revenue.

I also control for several organization-specific variables, including organization age, board size, and industry. Organization age is operationalized by subtracting the current year from the starting year (rule date) of the organization. Board size is operationalized as the total number of members serving on the board of directors of a given organization in a given year. Finally, to account for industry heterogeneity across the sample, I control for industry based on their National Taxonomy of Exempt Entities (NTEE) group classification, including Arts (AR), Education (ED), Health (HE), Human Services (HU), and Other (OT).

Dyadic Characteristics

I employ homophily attributes for industry, age, and board size to examine whether the similarity of the two nonprofits relates to network tie formation. Industry homophily is a dyadic covariate that captures whether the two nonprofits in a given dyad belong to the same industry. Although they are not discussed in my hypotheses, I also add two other dyadic covariates including organization age and board size. Organization age difference is a dyadic covariate that consists of the absolute difference in organization age operationalized by the subtraction of the current year from the starting year (rule date) of the organization. Board size difference is a dyadic covariate that consists of the absolute difference in the board of directors operationalized by the number of total board members.

Structural Characteristics

Following research tradition, I use estimated network statistics computed by the frequency of specific local network configurations such as preferential attachment and transitivity (Kim et al., 2016; Koskinen & Edling, 2012; Valeeva et al., 2020). First, the preferential attachment is measured by degree of popularity, which measures the tendency for popular organizations with more existing network connections to gain additional relationships over time. It accounts for a variation in degree distributions where highly connected nodes create more local clustering in the network (Robins et al., 2007). I apply a geometrically weighted degree distribution parameter to capture the degree centrality of nodes.

Second, the transitivity is measured by a generalized transitive closure. This variable accounts for variation in network clustering where a tie between two nonprofits—which are already tied to a common third nonprofit—is more likely to be observed in an overall network. I employ a geometrically weighted edge-wise shared partner distribution parameter to capture the distribution of shared partners of the connected nodes. Here, I note that when both the generalized transitive closure and the homophily parameters are included in the model, these two parameters can reveal whether there is a tendency of network ties to cluster beyond homophily. For example, if the generalized transitive closure parameter is not statistically significant, then it suggests that there is no tendency of ties to cluster beyond homophily (Snijders et al., 2006).

I also include a geometrically weighted dyad-wise shared partner distribution parameter in the model to control for multiple connectivity to make refined inferences about transitivity. Finally, I incorporate the edges parameter in the model to capture the overall density of the network. It controls for the overall tendency of nonprofits to form network ties (Lusher et al., 2013).

Temporal Dependency Characteristics

I test for temporal characteristics by using both intertemporal dependency and linear time trends. First, I include the intertemporal dependency covariate to control for the impact of a previous relationship on the current relationship. Importantly, the intertemporal dependency variable captures whether ties (and non-ties) at one point in time carry over to ties (and non-ties) at the next point in time. Second, I also add linear time trends to address temporal heterogeneity regarding network tie formation.

Descriptive Statistics

Table 1 presents the descriptive statistics for the sample and shows that there are remarkable similarities in the characteristics of the nonprofits across the three cities. The sampled nonprofits have about 17 to 18 total board of directors, and they represent relatively established organizations ranging from 36 to 40 years of operation with varied financial size. About half of the study sample consists of human service nonprofits, followed by health nonprofits, which make up about one-quarter of the sample.

Table 1.

Sample Nonprofit Characteristics.

Characteristics	Albany	Syracuse	Utica
Organization age (years)	36	40	38
Board of directors	17	17	18
Interlocking board of directors	3	5	4
Industry (%)
Arts	6	8	5
Education	10	13	7
Health	26	23	29
Human Services	42	44	44
Other	16	12	15
Liability (US$)	9,980,000	14,372,000	9,078,000
Asset (US$)	17,430,000	35,307,000	20,819,000
Expenses (US$)	18,325,000	22,812,000	17,432,000
Revenue (US$)	3,313,515,000	4,711,004,000	1,036,396,000
Contributions (US$)	1,107,973,000	468,300,000	82,080,000
Program service revenue (US$)	11,509,000	19,844,000	14,775,000
Investment income (US$)	119,000	327,000	185,000
Rental income (US$)	9,700	15,300	128,400
Number of nonprofit organizations	166	177	51

Temporal Exponential Random Graph Model

To examine the predictors of a nonprofit board interlock in longitudinal data, I use a type of inferential network analysis called a temporal exponential random graph model (TERGM). A network tie formation is an interdependent process influenced by the characteristics of the actors and their existing ties (Ahuja et al., 2009). Thus, standard statistical inference approaches, such as regression, are inappropriate for examining network data because they do not allow researchers to estimate structural characteristics due to interdependency of network parameters (Wasserman & Faust, 1994). An increasing number of scholars use the exponential random graph model (ERGM) approach to test a variety of interorganizational network relationships among the nonprofits, such as collaboration networks (Atouba & Shumate, 2015; O’Brien et al., 2019). TERGM is an extension of ERGM designed to accommodate intertemporal dependence in longitudinally observed networks.² I use the btergm package in R, and I run a separate analysis for each city during the study period.³

Results

Table 2 provides estimates from the analysis in each of the three cities. The first column is the baseline estimation model that includes organizational and dyadic attributes. The second column is the main model that includes structural attributes along with organizational and dyadic attributes. Compared with the results of the baseline model, the results of the main model indicate that hypothesized structural attributes have a positive and significant influence on board interlock in three cities. Below, I interpret the findings of the main model.

Table 2.

Results From Temporal Exponential Random Graph Model, 1998–2014.

Characteristics	Baseline model			Main model
Characteristics	Albany	Syracuse	Utica	Albany	Syracuse	Utica
Organizational characteristics
Resource vulnerability
Liabilities-to-assets-ratio	−0.03	−0.32*	−0.17	0.17*	−0.22	0.19
Operating margin	−0.40*	−0.46*	−0.99*	0.11	−0.28*	0.65
Resource dependency
Donative nonprofit	−0.29*	−0.02	−0.21*	−0.10	0.16*	0.16
Dyadic characteristics
Industry homophily	−0.00	0.01	0.12	−0.04	−0.04	0.13
Structural characteristics
Preferential attachment	—	—	—	4.46*	2.54*	3.06*
Transitivity	—	—	—	0.81*	1.18*	0.69*
Temporal characteristics
Intertemporal dependencies	0.53*	0.55*	0.51*	0.29*	0.42*	0.48*
Time trend	−0.04*	−0.05*	−0.01	−0.03*	−0.02*	0.03*
Control variables
Organizational control
Age	0.00*	0.00*	0.01*	0.00*	0.01*	0.01*
Board of directors	−0.01*	−0.00	−0.00	−0.00	0.00*	0.01
Education	−0.74*	−0.70*	−1.01*	0.12	0.04	0.55
Health	−0.90*	−0.68*	−0.84*	−0.00	0.07	0.40
Human services	−0.87*	−0.82*	−1.06*	0.16*	−0.09	0.39
Other	−1.04*	−0.85*	−0.63*	0.00	−0.14	0.80*
Program service revenue sources	−0.06	−0.34*	0.17	0.05	−0.27*	−0.05
Investment income sources	−0.70	−1.20*	1.31	0.79	−0.91*	−2.06
Rental income sources	1.26	−0.23	0.15	−2.74	−0.35	0.03
Dyadic control
Age difference	−0.00	−0.01*	0.00	0.00*	−0.00	0.00
Board size difference	−0.01*	−0.01*	0.00	−0.00*	−0.00*	0.00
Structural control
Multiple connectivity	—	—	—	0.05*	−0.16*	−0.11*
Edges	—	—	—	−4.13*	−2.29*	−4.56*
Number of observations	235,182	301,616	21,382	235,182	301,616	21,382

Notes. Maximum pseudo likelihood estimates are reported. All models successfully converged. Asterisks indicate that the coefficient is statistically significant at or beyond 0.05 level at 95% confidence interval of the sample of 1,000 bootstrap iterations for each model.

First, resource vulnerability covariates seem to be less consequential for board interlock. For the liabilities-to-assets-ratio variable, I find positive and significant resource vulnerability effects in Albany nonprofits only, indicated by an estimate of 0.17. This means that an additional unit in the share of liability to total asset increases a nonprofit’s likelihood to create an interlock by nearly 1.2 times (exp[0.17] = 1.19) while controlling for other associations in the model. In contrast, for the operating margin variable, I find negative and significant resource vulnerability effects in Syracuse nonprofits only. This term has an estimate of −0.28, meaning that an additional unit in the share of operating margin decreases a nonprofit’s likelihood to create an interlock by nearly 24% (exp[−0.28] = 0.76) while controlling for other associations in the model.

Based on the findings for resource vulnerability in Albany and Syracuse, I find inconsistent results for Hypothesis 1. This is consistent with the findings of previous studies that there is no clear relationship between financial characteristics and the presence of board interlock in corporate board governance (Mizruchi, 1996). For instance, while some studies report that organizations are more likely to have board interlock as their level of financial debt increases (Allen, 1974; Dooley, 1969), other studies find a null relationship between capital intensity and board interlock (Allen, 1974; Pennings, 1980).

In terms of resource dependency covariate, I find heterogeneous results for Hypothesis 2. A positive coefficient on donative nonprofits is only significant in Syracuse. These results align with the findings from a prior study (Galaskiewicz et al., 2006) which suggest that nonprofits that rely on contributions as a major revenue source tend to create network ties to gain access to external funding.

With respect to organization-level control covariates, organization age seems to have strong effects in explaining network relationships in three cities. Other resource dependency covariates used as control variables show heterogeneous results, depending on the city. For instance, program service revenue sources and investment income sources variables have negative and statistically significant effects in Syracuse nonprofits. Again, these results also suggest that financial resource dependency on a specific type of funding source may have a different effect in explaining network relationships depending on geographic context.

Second, in terms of homophily, I find no evidence for industry homophily effects across the three cities. Thus, the findings do not support Hypothesis 3. Third, the results show that structural processes, such as preferential attachment and transitivity, are important drivers of nonprofit networks via overlapping board members. Consistent with Hypothesis 4, I find that preferential attachment has notable associations on board interlock in the three cities. This is demonstrated by a positive and significant parameter for the degree of popularity (b = 4.46, 2.54, and 3.06, p < .05, for Albany, Syracuse, and Utica, respectively). The study results reveal a clear disposition toward a high degree of preferential attachment, which means that there is a tendency for highly “popular” nonprofits to connect with a large number of other nonprofits through overlapping board members. This is consistent with previous findings that dense clusters within board interlock networks are driven by highly connected organizations with multiple board memberships (Robins & Alexander, 2004).

Consistent with Hypothesis 5, I find that transitivity is an important driver of board interlock. This is demonstrated by a positive and statistically significant association of generalized transitive closure in three cities (b = 0.81, 1.18, and 0.69, p < .05, for Albany, Syracuse, and Utica, respectively), suggesting that predisposition exists between two nonprofits to have common third-party nonprofits in common. A positive transitivity, coupled with a negative multiple connectivity, adds stronger evidence to the transitivity effect. This suggests that network relationships are driven by clusters of closed triads (i.e., a set of three nodes to be closed).

Fourth, it appears that there is positive and statistically significant intertemporal dependency in the three cities. This indicates that there is a great amount of stability in network relationships over time. Overall, the major findings of the study suggest that the emergence of board interlock is influenced not just by preferential attachment (i.e., the number of network ties a nonprofit has) but also by transitivity (i.e., how those ties are connected to one another through indirect ties). Findings of structural attributes are consistent with prior research showing that corporate board interlocks are shaped by existing network structures where organizations tend to be interlocked with other highly connected organizations (Benton, 2016; Kim et al., 2016; Koskinen & Edling, 2012; Valeeva et al., 2020).

Graphical evaluations of goodness-of-fit diagnostics are conducted to ensure the overall validity of the main model (presented in Figure A1(A)–(C)).⁴ Following research tradition, statistics I used to assess the model fit include undirected degree centrality, edge-wise shared partners, undirected triad, walktrap modularity, and geodesic distance (for detailed discussion, see Leifeld et al., 2018). The visual diagnostics presented in Figure A1 show that the distribution of plotted networks randomly generated from the fitted models approximately fits well with the observed network.

Conclusion

In this article, I reveal two important, previously unverified findings related to nonprofit network connections through overlapping board members. The results reveal that board interlock networks evolve over time by responding to the structure of the network in which they are embedded. Interestingly, when including structural attributes into the estimation model, effects of resource vulnerability covariate became less conclusive. On this account, one may consider that previous studies may have overemphasized the influence of resource availability in explaining nonprofit networks without taking into account the structural characteristics of interconnections. Here, significant findings on preferential attachment seem to make sense, given that recruiting competent board members are crucial to nonprofit performance (Brown, 2007). On this account, one may conclude that board interlock occurs due to nonprofit leaders’ preference for recruiting and retaining capable board members whose human and social capital serve as critical assets in developing nonprofit effectiveness (Brown et al., 2012; Fredette & Bradshaw, 2012).

There are limitations to this study. Due to data limitations, the analysis relies on some critical assumptions about board recruitment practices related to existing boards’ preference for seeking popular candidates who tend to sit on multiple boards. However, variations may exist in board selection practices across nonprofits. Future research may incorporate data related to the recruitment structure of board of directors to further explore additional factors related to nonprofit network relationships.

Furthermore, the empirical model used does not take into account the characteristics of individual board of directors who are important players in board governance. Future research may explore fine-grained, individual-level predictors such as sociodemographic characteristics and professional experiences of individual board members. Here, scholars may build from extant studies investigating the background attributes of board members (Abzug & Galaskiewicz, 2001; Suárez, 2010), such as educational attainment and having memberships in a prestigious social club in the community (Domhoff, 2009). Given that nonprofit board of directors represents community constituents and interests (Abzug & Galaskiewicz, 2001; Guo & Musso, 2007), it will be particularly useful to explore whether individuals with certain characteristics tend to hold multiple board memberships. As Guo et al. (2014, p. 57) note in their excellent study, “future research should further examine the practices and effects of constituent participation in governance, both through and beyond the board.” This line of study would extend our understanding of the representational capacities of nonprofit organizations through board governance composition.

Finally, the study sample is limited to a panel of nonprofits that reported the IRS 990 filings in three cities in Upstate New York. Thus, the findings may not be generalizable to board interlock network connections in other contexts. Future works may look into the network pattern of organizations in different geographic locations, sizes, and time frames.

Despite these limitations, this study contributes to the literature by offering initial insights into understanding how nonprofit networks change over time through board composition. The results raise important questions regarding the structural characteristics of board interlock. For example, analyses show that preferential attachment is a strong driver of network formation. Over time, it may create structural embeddedness (Granovetter, 1985) where network relationships are more likely to be formed between organizations occupying central positions in the network structure (Galaskiewicz & Zaheer, 1999; Powell et al., 2005).

A superior network position fosters organizational performance and survival, but organizations located at the periphery of a network do not enjoy the reputational benefits associated with network positions (Gulati & Gargiulo, 1999). Recent studies report a positive association between interlock ties and foundation grant funding (Esparza & Jeon, 2013; Faulk et al., 2016, 2017; Paarlberg et al., 2020), suggesting that the status conferred through network positions contributes to grant performance. The findings from this study suggest that one way to move from the periphery to the core position in the board interlock network is through board composition. Nonprofit leaders may overcome structural barriers by recruiting board members who are actively serving on many boards in the community. By recruiting these board members with multiple board affiliations, less-central nonprofits may build network relationships with central nonprofits. Thus, the findings from the study are particularly useful for nonprofit leaders seeking to increase organizational interconnections for advanced access to resources and information.

Footnotes

Appendix

Author’s Note

This article is developed from my doctoral dissertation. My sincere gratitude goes to Dr. Joseph Galaskiewicz for insightful suggestions, which contributed to the revision of the manuscript. I am also grateful to my dissertation committee and the three anonymous reviewers for valuable comments on numerous iterations of the article. I am alone is responsible for this work and any errors or omissions.

Declaration of Conflicting Interests

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

Funding

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

ORCID iD

Nara Yoon

Notes

Author Biography

Nara Yoon received her doctorate from the Maxwell School of Citizenship and Public Affairs at Syracuse University and is currently an Assistant Professor of the School of Strategic Leadership Studies at James Madison University. Her research focuses on board governance, collaborative governance, and volunteerism in public and nonprofit management.

References

Abzug

Galaskiewicz

(2001). Nonprofit boards: Crucibles of expertise or symbols of local identities? Nonprofit and Voluntary Sector Quarterly, 30(1), 51–73.

Ahuja

Polidoro

Mitchell

(2009). Structural homophily or social asymmetry? The formation of alliances by poorly embedded firms. Strategic Management Journal, 30(9), 941–958.

Allen

M. P.

(1974). The structure of interorganizational Elite cooptation: Interlocking corporate directorates. American Sociological Review, 39(3), 393–406.

Atouba

Y. C.

Shumate

(2015). International nonprofit collaboration: Examining the role of homophily. Nonprofit and Voluntary Sector Quarterly, 44(3), 587–608.

Barabási

A. L.

Albert

(1999). Emergence of scaling in random networks. Science, 286(5439), 509–512.

Benton

R. A.

(2016). Corporate governance and nested authority: Cohesive network structure, actor-driven mechanisms, and the balance of power in American corporations. American Journal of Sociology, 122(3), 661–713.

Bloch

R. I.

Harris

E. E.

Peterson

A. N.

(2020). Interlocking boards in nonprofit organizations. Accounting Horizons, 34(2), 1–17.

Bouwman

C. H. S.

(2011). Corporate governance propagation through overlapping directors. The Review of Financial Studies, 24(7), 2358–2394.

Brown

W. A.

(2007). Board development practices and competent board members: Implications for performance. Nonprofit Management and Leadership, 17(3), 301–317.

10.

Brown

W. A.

Guo

(2010). Exploring the key roles for nonprofit boards. Nonprofit and Voluntary Sector Quarterly, 39(3), 536–546.

11.

Brown

W. A.

Hillman

A. J.

Okun

M. A.

(2012). Factors that influence monitoring and resource provision among nonprofit board members. Nonprofit and Voluntary Sector Quarterly, 41(1), 145–156.

12.

Chang

C. F.

Tuckman

H. P.

(1994). Revenue diversification among non-profits. VOLUNTAS: International Journal of Voluntary and Nonprofit Organizations, 5(3), 273–290.

13.

Chu

J. S. G.

Davis

G. F.

(2016). Who killed the inner circle? The decline of the American corporate interlock network. American Journal of Sociology, 122(3), 714–754.

14.

Connelly

B. L.

Johnson

J. L.

Tihanyi

Ellstrand

A. E.

(2011). More than adopters: Competing influences in the interlocking directorate. Organization Science, 22(3), 688–703.

15.

Cornforth

Edwards

(1999). Board roles in the strategic management of non-profit organisations: Theory and practice. Corporate Governance, 7(4), 346–362.

16.

Desmarais

B. A.

Cranmer

S. J.

(2012). Micro-level interpretation of exponential random graph models with application to estuary networks. Policy Studies Journal, 40(3), 402–434.

17.

Domhoff

G. W.

(2009). The power elite and their challengers: The role of nonprofits in American social conflict. American Behavioral Scientist, 52(7), 955–973.

18.

Dooley

P. C.

(1969). The interlocking directorate. The American Economic Review, 59(3), 314–323.

19.

Epstein

M. J.

McFarlan

F. W.

(2011). Nonprofit vs. for-profit boards: Critical differences. Strategic Finance, 92(9), 28–35.

20.

Esparza

Jeon

S. H.

(2013). Interlocking boards of trustees and grant acquisition among homeless service organizations. Public Performance & Management Review, 36(4), 637–664.

21.

Faulk

McGinnis Johnson

Lecy

J. D.

(2017). Competitive advantage in nonprofit grant markets: Implications of network embeddedness and status. International Public Management Journal, 20(2), 261–293.

22.

Faulk

Willems

McGinnis Johnson

Stewart

A. J.

(2016). Network connections and competitively awarded funding: The impacts of board network structures and status interlocks on nonprofit organizations’ foundation grant acquisition. Public Management Review, 18(10), 1425–1455.

23.

Fredette

Bradshaw

(2012). Social capital and nonprofit governance effectiveness. Nonprofit Management and Leadership, 22(4), 391–409.

24.

Galaskiewicz

(1985). Professional networks and the institutionalization of a single mind set. American Sociological Review, 50(5), 639–658.

25.

Galaskiewicz

Bielefeld

Dowell

(2006). Networks and organizational growth: A study of community based nonprofits. Administrative Science Quarterly, 51(3), 337–380.

26.

Galaskiewicz

Shatin

(1981). Leadership and networking among neighborhood human service organizations. Administrative Science Quarterly, 26(3), 434–448.

27.

Galaskiewicz

Wasserman

(1989). Mimetic processes within an interorganizational field: An empirical test. Administrative Science Quarterly, 34(3), 454–479.

28.

Galaskiewicz

Zaheer

(1999). Networks of competitive advantage. Research in the Sociology of Organizations, 16(1), 237–261.

29.

Gazley

Guo

(2020). What do we know about nonprofit collaboration? A systematic review of the literature. Nonprofit Management and Leadership, 31(2), 211–232.

30.

Granovetter

(1985). Economic action and social structure: The problem of embeddedness. American Journal of Sociology, 91(3), 481–510.

31.

Gulati

Gargiulo

(1999). Where do interorganizational networks come from? American Journal of Sociology, 104(5), 1439–1493.

32.

Guo

Acar

. (2005). Understanding collaboration among nonprofit organizations: Combining resource dependency, institutional, and network perspectives. Nonprofit and Voluntary Sector Quarterly, 34(3), 340–361.

33.

Guo

Metelsky

B. A.

Bradshaw

(2014). Out of the shadows: Nonprofit governance research from democratic and critical perspectives. In Cornforth

Brown

W. A.

(Eds.), Nonprofit governance: Innovative perspectives and approaches (pp. 47–67). Routledge.

34.

Guo

Musso

J. A.

(2007). Representation in nonprofit and voluntary organizations: A conceptual framework. Nonprofit and Voluntary Sector Quarterly, 36(2), 308–326.

35.

Hager

M. A.

(2001). Financial vulnerability among arts organizations: A test of the Tuckman-Chang measures. Nonprofit and Voluntary Sector Quarterly, 30(2), 376–392.

36.

Hager

M. A.

Galaskiewicz

Larson

J. A.

(2004). Structural embeddedness and the liability of newness among nonprofit organizations. Public Management Review, 6(2), 159–188.

37.

Hansmann

H. B.

(1986). The role of nonprofit enterprise. In Rose-Ackerman

(Ed.), The economics of nonprofit institutions: Studies in structure and policy (pp. 57–84). Oxford University Press.

38.

Hodge

M. M.

Piccolo

R. F.

(2005). Funding source, board involvement techniques, and financial vulnerability in nonprofit organizations: A test of resource dependence. Nonprofit Management and Leadership, 16(2), 171–190.

39.

Ihm

Shumate

(2019). How does a board of directors influence within-and cross-sector nonprofit collaboration? Nonprofit Management and Leadership, 29(4), 473–490.

40.

Isett

K. R.

Provan

K. G.

(2005). The evolution of dyadic interorganizational relationships in a network of publicly funded nonprofit agencies. Journal of Public Administration Research and Theory, 15(1), 149–165.

41.

Kim

J. Y.

Howard

Cox Pahnke

Boeker

(2016). Understanding network formation in strategy research: Exponential random graph models. Strategic Management Journal, 37(1), 22–44.

42.

Kono

Palmer

Friedland

Zafonte

(1998). Lost in space: The geography of corporate interlocking directorates. American Journal of Sociology, 103(4), 863–911.

43.

Koskinen

Edling

(2012). Modelling the evolution of a bipartite network—Peer referral in interlocking directorates. Social Networks, 34(3), 309–322.

44.

Leifeld

Cranmer

S. J.

Desmarais

B. A.

(2018). Temporal exponential random graph models with btergm: Estimation and bootstrap confidence intervals. Journal of Statistical Software, 83(6), 1–36.

45.

Lorsch

Young

(1990). Pawns or potentates: The reality of America’s corporate boards. Academy of Management Perspectives, 4(4), 85–87.

46.

Lusher

Koskinen

Robins

(Eds.). (2013). Exponential random graph models for social networks: Theory, methods, and applications. Cambridge University Press.

47.

MacIndoe

Sullivan

(2014). Nonprofit response to financial uncertainty: How does financial vulnerability shape nonprofit collaboration. Journal of Management and Sustainability, 4, Article 1.

48.

McPherson

J. M.

Smith-Lovin

(1987). Homophily in voluntary organizations: Status distance and the composition of face-to-face groups. American Sociological Review, 52(3), 370–379.

49.

Miller-Millesen

J. L.

(2003). Understanding the behavior of nonprofit boards of directors: A theory-based approach. Nonprofit and Voluntary Sector Quarterly, 32(4), 521–547.

50.

Mizruchi

M. S.

(1996). What do interlocks do? An analysis, critique, and assessment of research on interlocking directorates. Annual Review of Sociology, 22(1), 271–298.

51.

Mizruchi

M. S.

Stearns

L. B.

(1994). A longitudinal study of borrowing by large American corporations. Administrative Science Quarterly, 39(1), 118–140.

52.

Moore

Sobieraj

Whitt

J. A.

Mayorova

Beaulieu

(2002). Elite interlocks in three U.S. sectors: Nonprofit, corporate, and government. Social Science Quarterly, 83(3), 726–744.

53.

O’Brien

N. F.

Pilny

Atouba

Y. C.

Shumate

Fulk

J. L.

Monge

P. R.

(2019). How does NGO partnering change over time? A longitudinal examination of factors that influence NGO partner selection. Nonprofit and Voluntary Sector Quarterly, 48(6), 1229–1249.

54.

Paarlberg

L. E.

Hannibal

McGinnis Johnson

(2020). Examining the mediating influence of interlocking board networks on grant making in public foundations. Nonprofit and Voluntary Sector Quarterly, 49(4), 734–756.

55.

Park

H. H.

Rethemeyer

R. K.

(2012). The politics of connections: Assessing the determinants of social structure in policy networks. Journal of Public Administration Research and Theory, 24(2), 349–379.

56.

Pennings

J. M.

(1980). Interlocking directorates: Origins and consequences of connections among organizations’ board of directors. Jossey-Bass.

57.

Pfeffer

Salancik

G. R.

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

58.

Podolny

J. M.

(1994). Market uncertainty and the social character of economic exchange. Administrative Science Quarterly, 39(3), 458–483.

59.

Powell

W. W.

White

D. R.

Koput

K. W.

Owen-Smith

(2005). Network dynamics and field evolution: The growth of interorganizational collaboration in the life sciences. American Journal of Sociology, 110(4), 1132–1205.

60.

Provan

K. G.

Huang

Milward

H. B.

(2009). The evolution of structural embeddedness and organizational social outcomes in a centrally governed health and human services network. Journal of Public Administration Research and Theory, 19(4), 873–893.

61.

Robins

Alexander

(2004). Small worlds among interlocking directors: Network structure and distance in bipartite graphs. Computational & Mathematical Organization Theory, 10(1), 69–94.

62.

Robins

Pattison

Kalish

Lusher

(2007). An introduction to exponential random graph (p*) models for social networks. Social Networks, 29(2), 173–191.

63.

Snijders

T. A.

Pattison

P. E.

Robins

G. L.

Handcock

M. S.

(2006). New specifications for exponential random graph models. Sociological Methodology, 36(1), 99–153.

64.

Suárez

D. F.

(2010). Street credentials and management backgrounds: Careers of nonprofit executives in an evolving sector. Nonprofit and Voluntary Sector Quarterly, 39(4), 696–716.

65.

Tuckman

H. P.

Chang

C. F.

(1991). A methodology for measuring the financial vulnerability of charitable nonprofit organizations. Nonprofit and Voluntary Sector Quarterly, 20(4), 445–460.

66.

Valeeva

Heemskerk

E. M.

Takes

F. W.

(2020). The duality of firms and directors in board interlock networks: A relational event modeling approach. Social Networks, 62, 68–79.

67.

Vidovich

Currie

(2012). Governance networks: Interlocking directorships of corporate and nonprofit boards. Nonprofit Management and Leadership, 22(4), 507–523.

68.

Wasserman

Faust

(1994). Social network analysis: Methods and applications (vol. 8). Cambridge University Press.

69.

Weisbrod

B. A.

(1988). The nonprofit economy. Harvard University Press.

70.

Willems

Van Puyvelde

Jegers

Vantilborgh

Bidee

Pepermans

(2015). Exploring board interlocking behaviour between nonprofit organizations. Annals of Public and Cooperative Economics, 86(1), 73–88.

71.

Yoon

(2021a). Understanding theoretical orientation and consequences of board interlock: Integration and future directions. Nonprofit Management and Leadership, 31(4), 717–736.

72.

Yoon

(2021b). Nonprofit board governance policy adoption: Toward an integrated board interlock network and institutional perspectives. Nonprofit and Voluntary Sector Quarterly. Advance online publication. https://doi.org/10.1177/08997640211057452