Corporate Political Donations: Influences from Directors’ Networks

Abstract

Motivated by contemporary debates concerning whether directors inappropriately deploy corporate funds for corporate political donations and the limited research into managerial influence on corporate political donations, we examine the impact of director influences from a network perspective. Using a sample of large listed Australian corporations and their political party donation activity during 2000–2007, we find that both the professional and non-professional networks of directors influence corporate political donations. We observe these influences in relation to donations at the federal and state levels, and with respect to the choice of recipient political parties.

Introduction

Contemporary media coverage indicates ongoing public and investor interest in understanding the motives for corporate political donations. Public concerns centre on whether corporate donations distort the democratic process by inducing political participants to pursue policies and actions that are favorable to the donors. Shareholder concerns focus on whether managers inappropriately deploy corporate funds in the form of political donations to serve interests other than that of the corporation; for example, the Australian Shareholders Association (2004) described corporate political donations as ‘inappropriate disbursements of shareholders’ funds’ and called for ‘greater disclosure of political donations such as the amount and the reasons the donations are made’. The business and economics literature emphasizes corporate interests (Hillman et al. 2004) but yields inconsistent explanations (Mathur and Singh 2011). While there has been little study of the impact of managers’ interests on corporate political donations, such influences can be inferred from the studies that report significant relations between corporate donations and corporate governance attributes (Mathur and Singh 2011; Ramsay et al. 2001). For example, it has been argued that managers’ interests, rather than corporate interests, are reflected in the apparent free cash flow characteristics of donor corporations (Aggarwal et al. 2008) and the negative relation between blockholders and donations (Bartkus et al. 2002). Consistent with these concerns and the absence of direct evidence, Hillman et al. (2004) call for more direct research on managerial influences on corporate political donations. We address this deficiency by examining the impact of directors’ influences from a network perspective. Specifically, we examine whether directors’ networks affect individual corporations’ political donations.

A large body of literature indicates two means by which individuals’ networks can affect organizations’ decisions. The first emphasizes information transfers among network members, whereby a network provides a channel for communication between and learning from network members, which affects decisions made in members’ own organizations (Davis 1991; Maak 2007; Useem 1984). The second focuses on the individuals’ compliance with group norms when their desires for reputation and status in their networks (Lin et al. 2008; Olson 1965) encourages them to make organizational decisions that are expected by other members in their networks (Burt 1992; Galaskiewicz 1985; Olson 1965; Useem 1984)

We examine two networks pertaining to directors; we label these as their professional and non-professional networks. We identify their professional networks as those established through director interlocks. Director interlocks are recognized as important network devices that can influence organizational practices (Davis 1991; Haunschild 1993; Palmer et al. 1993). We identify non-professional networks as those arising from directors’ memberships of non-profit boards; this is consistent with prior studies that suggest political activities and social responsibility may be characterized by an obligation and accountability to the wider society (e.g., Berkowitz and Lutterman 1968; Cole and Stewart 1996; Gough et al. 1952) while charitable contributions and political donations can reflect views on particular issues (Kahn 1997; Ramsay et al. 2001).¹ We distinguish network effects for executive and non-executive directors because executive directors are generally expected to be more influential than non-executive directors with respect to a corporation’s donating behavior (Bond 2004; Davis 1991; Haunschild 1993; Haunschild and Beckman 1998; Palmer et al. 1995).

Our study uses Australian data for two reasons. First, unlike many jurisdictions, Australia does not cap the amounts that can be directly contributed by corporations, which we suggest results in a less noisy setting in which to examine corporate political donations and network effects. Second, Australia provides a reliable and readily available data source because any entity’s aggregate political donations that exceed a modest annual threshold must be reported to the Australian Electoral Commission. Out study period commences in 2000, when data for our political exposure variables became available, and ends in 2007 prior to a change of government. This period gives the advantage of including federal election cycles while avoiding potential complications pertaining to changes in donation patterns that may arise with a change in the governing parties.

Our main analysis reveals a robust positive association between corporations’ political donations and: (1) the incidence of such donations in directors’ professional networks; and (2) the extent of directors’ participation on non-profit boards. The analysis controls for other factors expected to affect corporate political donations, including some specific corporate interests, and use a variety of methods to address potential endogeneity concerns associated with directors’ interlocks and potential selection biases associated with donation decisions. In further tests, we find evidence of network effects with respect to federal and state donations to particular political parties. Overall, our results are consistent with the proposition that directors’ networks have significant influences on corporate political donation activity.

This study contributes to the literature in two main ways. First, we show that directors’ networks influence corporate political donations. The omission of managerial influences on corporate political donations might help explain why prior studies concerned with corporate interest motives, such as regulatory exposure or the outcomes in contests for government contracts, report inconsistent results. Second, our results complement studies that suggest firms’ investments in corporate social responsibility activities reflect attempts to enhance managers’ or directors’ reputations, contrary to the interests of shareholders (e.g., Barnea and Rubin 2010; Cespa and Cestone 2007; Harjoto and Jo 2011). Our results may thus inform future debate regarding the appropriateness of corporate political donations. Thus, our study has important ethical and public policy implications.

The remainder of the paper is organized as follows. In the next section, we review the relevant literature to develop our hypothesis with respect to network effects. The “Research Methods” section describes our empirical model, sample selection and descriptive statistics, and multicollinearity diagnostics. The “Results” section reports our empirical results, robustness checks and further tests. The last section concludes the paper with discussions of limitations to our analysis and the main implications of our findings.

Hypothesis Development

Networks connect sets of people (Ibarra 1993) who, by interacting with each other, generate solidarity, goodwill and mutual influence (Maak 2007). Studies identify two ways in which a person’s membership of a network can affect his or her decisions at the organizational level. The first emphasizes information transfers among the network members (e.g., Davis 1991; Haunschild 1993; Useem 1984; Westphal 1999) and the second focuses on individual benefits derived from their status within a social network (e.g., Galaskiewicz 1985; Haley 1991).

The information transfer argument is based on networks providing channels for communication and learning so that, through network engagement, members learn from each other, which then influences their contributions to decisions made in their organizations. For example, it has been argued that director interlocks provides a mechanism for managers to identify business practices that they might elect to imitate in their own firms (e.g., Davis 1991; Haunschild 1993; Useem 1984). Consistent with these arguments, Davis (1991) finds that managers are more likely to adopt takeover defense strategies when they sit on the boards of firms that had already adopted such strategies and Haunschild (1993) finds that managers are more likely to engage in acquisitions when they sit on other boards that had experience of acquisitions.

There is substantial evidence that individuals value their status or acceptance in their networks (e.g., Bearden and Mintz 1987; Burris 1992a; Galaskiewicz 1985, p. 21; Mizruchi 1992, 1996; Useem 1984), which may also yield extra-network benefits such as promotion (Haley 1991). To acquire, retain or enhance their status or acceptance, individuals demonstrate conformity to the norms of their social group (Galaskiewicz 1985; Olson 1965; Useem 1984); for example, Galaskiewicz (1985) argues that corporate charitable contributions enable corporate managers to enhance their status among the other corporate leaders, while Useem (1984) argues that peer pressure and the threat of exclusion from business networks motivate managers to make corporate charitable donations. Consistent with the established network literature and its application to corporate charitable contributions, we argue that the incidence of corporate political donations by corporations linked through directors’ professional networks indicates the extent to which they are network norms, and that directors’ are motivated to demonstrate conformity with such network norms, as expressed in Hypothesis 1:

Hypothesis 1

There is a positive association between the level of political donations by directors’ employing corporations and the incidence of corporate political donation in directors’ professional networks.

The broader network literature indicates that managers are influenced by networks additional to their immediate professional connections. Involvement in non-profit institutions may enhance a person’s public or social status and prestige, and may convey tangible benefits such as better pay and future job opportunities (DiMaggio and Useem 1978; Haley 1991; Ostrower 1997, 2002). There is no established link between corporate political donation behavior and non-professional networks but both charitable contributions and political donations have been associated with individuals’ desires to express preferences on particular issues (Kahn 1997; Ramsay et al. 2001) and prior studies report a positive association between an individual’s participation in political activities and their levels of social responsibility (Berkowitz and Lutterman 1968; Pancer et al. 2007; Watts and Guessous 2006). Therefore, we use directors’ non-profit board memberships as a proxy for the influence arising from directors’ non-professional networks, as expressed in Hypothesis 2:

Hypothesis 2

There is a positive association between the level of political donations by the directors’ employing corporations and the extent of directors’ non-professional networks.

Research Methods

In this section, we describe the models and variables used to test the hypothesized relations and effects. Further details on the methods used in collecting our data are provided in Appendix.

Regression Models

To test our hypothesis, we regresses the level of firm-year donations against proxies for directors’ network influences, while controlling for corporate interests that may motivate political donations, corporate governance effects and other variables, as shown in Model 1. We employ several regression methods (OLS, Fama–Macbeth, Tobit and Fixed effects).

$$\begin{aligned} {\text{Donations}}_{it} = & a_{0} + \sum {d_{j} {\text{Network}}\,{\text{Influence}}\,{\text{Indicators}}_{it} } + \sum {b_{j} {\text{Corporate}}\,{\text{Interest}}\,{\text{Controls}}_{it} } \\ & + \sum {bj\,{\text{Corporate}}\,{\text{Governance}}\,{\text{Controls}}_{it} } + \sum {b_{j} {\text{Other}}\;{\text{Controls}}_{it} } + \varepsilon_{it} , \\ \end{aligned}$$

(Model 1)

where DONATIONS _it is the logarithm of total federal level political donations made by corporation i and its subsidiaries during financial year t, plus 1.

Network Influence Indicators NET_DON_EX _it is the sum of individual executive director’s and chairperson’s directorships of other donating firms, divided by one plus the number of executive directors and chairperson on the board at the end of year t multiplied by 100. NET_DON_NONEX _it is the sum of individual non-executive directors directorships of other donating listed companies, divided by one plus the number of non-executive directors on the board at the end of year t multiplied by 100. NPNET_EX _it is the sum of non-profit board memberships held by executive directors, divided by one plus the number of executive directors sitting on the board at the end of year t multiplied by 100. NPNET_NONEX _it is the number of non-profit board memberships held by non-executive directors, divided by one plus the number of non-executive directors sitting on the board at the end of year t multiplied by 100.

Corporate Interest Controls PARL_EXPOSURE _it is the the number of times that corporation i is cited in the Federal Parliamentary Hansards during year t. GAZETTE_EXPOSURE _it is the the number of times that corporation i is cited in the federal gazettes during year t. CONTRACT_AWARDED is the number of federal contracts awarded to corporation i during year t.

Corporate Governance Controls BDINDEP _it is the number of non-executive directors, divided by the number of executive directors serving on the board at the end of year t, plus 1. BLOCK _it is the number of shareholders, excluding directors, who own five percent or more of ordinary shares in the corporation at the end of year t. SHARES_EX _it is the percentage of ordinary shares held by executive directors at the end of year t. SHARES_NONEX _it is the the percentage of ordinary shares held by non-executive directors at the end of year t.

Other Controls SIZE _it is the natural logarithm of market capitalisation at the end of year t. ROA _it is the operating profit defined by Aspect FinAnalysis scaled by the average total asset in year t. MEDIA _it is the number of references to corporation i by media during year t, as recorded in ParlInfo Search. BDSIZE _it is the number of directors serving on the board at the end of year t. INDUSTRY _it is the indicator variables to identify the firm’s sector based on 2-digit GICS. YEAR _it is indicator variables indicating the financial year of the observation.

Measure of Corporate Political Donations

We measure corporate political donations (DONATIONS) as the total federal level political donations made by a corporation and its subsidiaries during a financial year. We focus on federal level donations because the substantially different political issues in states and territories may induce jurisdictional differences in corporations’ strategies and effort to influence government decisions in Australia (Moon and Sharman 2003), and the relatively low incidence of corporate political donation in any given state and territory is statistically noisy. We use total donations rather than donations to individual political parties because of the presence of party coalitions and because the legislative process in Australia requires the assent of both houses, which is rarely achievable without the support of multiple parties (Woodward et al. 1997). Throughout our study period 2000–2007, the executive government was formed by a coalition of the Liberal Party of Australia and the National Party of Australia, who jointly controlled the House of Representatives. However, this coalition held only 35 of the 76 Senate seats for most of this period (2000–2005). The other Senate seats were variously held by the Australian Labor Party (29 seats), the Australian Democrats (8–9 seats), the Australian Greens (1–2 seats) and independent senators. For the remainder of the period, 2005–2007, the coalition government parties jointly held 39 Senate seats. Consistent with prior studies, we include donations made by a corporation’s wholly-owned subsidiaries (e.g., Mitchell et al. 1997; Aggarwal et al. 2008) and use the natural logarithm of total corporate political donations (Hansen and Mitchell 2000; Hart 2001).

Measures of Directors’ Network Influences

We develop two sets of proxies for the unobservable political donation influences from directors’ networks, differentiated for executive and non-executive directors (Bond 2004; Geletkanycz and Hambrick 1997; Haunschild 1993; Palmer et al. 1995). The first set of proxies (NET_DON_EX _it and NET_DON_NONEX _it ) measure the incidence of corporate political donations in professional networks attributed to director interlocks. Consistent with the arguments concerning the establishment of network norms presented in our hypothesis development, we interpret the incidence of corporate political donations as indicative of the network norm in this regard. The second set of proxies (NPNET_EX _it and NPNET_NONEX _it ) measure the relative size of directors’ non-professional networks that we infer from their memberships of boards of non-profit institutions, such as charities, clubs, hospitals and universities.

Using the incidence of corporate political donation in directors’ professional networks based on their memberships of other corporations’ boards is consistent with other studies of director networks (e.g., Bearden and Mintz 1987; Burris 1992b, 2005; Davis 1991; Mizruchi 1992, 1996; Useem 1984). For each donor corporation in our firm-year sample, we sum the number of executive (non-executive) directors’ interlocks with other donor corporations. We then deflate the total by the number of executive (non-executive) directors (plus one, to avoid zero divisors) because, the smaller the proportion of interlocks with other donor corporation, the less the expected influence on donation decisions. We include the chairperson in the measure of executive directors’ interlocks because, compared to non-executive directors, both the chair and executive directors are generally expected to have more influence over a corporation’s donating behavior (Bond 2004; Davis 1991; Haunschild 1993; Haunschild and Beckman 1998; Palmer et al. 1995) Thus, NET_DON_EX (NET_DON_ NONEX) equals the sum of executive (non-executive) director interlocks to other donor corporations divided by the number of executive (non-executive) directors on the board at the end of year (plus one), as reported in the annual report.

Our proxy for corporate political donation influence from directors’ non-professional networks is the number of non-profit board memberships held by directors. Our measure of the size of directors’ non-professional networks is consistent with our approach for professional networks and corporate charitable contribution studies (Bartkus et al. 2002; Galaskiewicz 1985; Werbel and Carter 2002). NPNET_EX (NPNET_NONEX) is calculated as the total number of non-profit board memberships held by executive (non-executive) directors, deflated by the number of executive directors (non-executive directors) in the focal corporation (plus one).

Measure of Control Variables

We control for factors that, based on a broad literature, may affect corporate political donations and whose omission may bias results for our test variables.

Corporate Interest Related Control Variables

Political scientists have long argued that corporate political donations are an effective means of influencing government decisions (Jacobson 2001; Kagan et al. 2008, p. 15; Shipper and Jennings 1984).² This is more likely if groups with interests contrary to those of a corporation can influence government decisions that result in costs to the corporation (Austen-Smith 1995; Dahl and Lindblom 1953; Milbrath 1963).³ Therefore, following prior studies (e.g., Hansen and Mitchell 2000; Mitchell et al. 1997; Schuler et al. 2002), we control for firm’s exposure to government decision-making.

In Australia, federal government decision-making in both the legislature and executive (cabinet) is subject to strong party influences (Singleton et al. 2006) such that both parliamentary and executive decisions may be influenced by donations to political parties. Therefore, we control for a corporations’ exposure to both parliamentary decision-making and government decision-making outside parliament.

We measure a corporation’s exposure to parliamentary decision-making (PARL_EXPOSURE), in a manner consistent with prior research (e.g., Hansen and Mitchell 2000; Mitchell et al. 1997; Schuler et al. 2002), by summing the number of citations of a corporation’s name(s) in the official record of parliamentary proceedings (Hansard) for the relevant financial year.⁴ We identify citations by individually searching each corporation’s name (and known variants) in the electronic Hansard service.

We measure a corporation’s exposure to executive government decision-making in two ways: (1) summing the number of times that the corporation’s name(s) is referenced in notifications of government decisions and actions recorded in Government gazettes, excluding government tendering notices (GAZETTE_EXPOSURE)⁵; and (2) the number of contracts the corporation was awarded during the financial year (CONTRACT_AWARDED), as evidenced by notices of successful tenders reported in the Gazette. We measure only successful tenders because there is no public record of unsuccessful tenders.

Corporate Governance Control Variables

Corporate political contributions can be a manifestation of managerial preferences, which reflect their expected utility from perquisite consumption (e.g., Haley 1991; Navarro 1988; Ramsay et al. 2001). The corporate governance literature contends that board independence and external blockholders can play important roles in monitoring the behavior of managers and that managerial equity holdings can increase the alignment of the interests of managers and shareholders (Hambrick and Finkelstein 1987). This is because an independent board is less likely to be influenced by managers, and thus more likely to constrain managerial behavior that is detrimental shareholder interests (Hermalin and Weisbach 2003; Roberts et al. 2005; Walsh and Seward 1990). External blockholders have sufficient incentive to influence potentially inefficient decisions (Shleifer and Vishny 1986) and may do so through voting power (Maug 1998) or trading behavior (Edmans and Manso 2011). Directors’ equity holdings in a corporation may increase the alignment of their interests with those of external shareholders (Berle and Means 1932; Fama 1980; Fama and Jensen 1983) but can also give directors sufficient power to pursue private objectives without endangering their employment and remuneration.

We measure board independence (BDINDEP) as the ratio of the number of non-executive directors to the number of executive directors (plus one), which is consistent with prior studies (Dahya et al. 2002; Weir and Laing 2003; Westphal and Zajac 1995). We separately measure the levels of executive and non-executive directors’ shareholdings as SHARES_EX and SHARES_ NONEX, calculated as the ratios of shares held by executive or non-executive directors to total shares outstanding (Buchholtz and Ribbens 1994; Cochran et al. 1985; Jensen and Murphy 1990). We measure external blockholder monitoring (BLOCK) as the number of external shareholders who hold five per cent or more of a corporation’s ordinary shares (Demsetz and Lehn 1985; Kroll et al. 1997; Wright et al. 2002).

Other Control Variables

Our other control variables are continuous measures of firm size, media exposure and board size, and industry and year fixed effects.

Firm size and profitability may influence a firm’s capacity to make donations (Cooper et al. 2010; Hansen and Mitchell 2000; Hillman and Hitt 1999; Schuler 1996; Yoffie 1987). We control for firm size effects using SIZE, calculated as the natural logarithm of a firm’s market capitalisation (Cooper et al. 2010; Faccio et al. 2006; Kiel and Nicholson 2003).⁶ We control for firm profitability using return on assets (ROA), calculated as net operating income divided by average total assets for the financial year.

Media exposure may affect a firm’s political costs and therefore influences its political activities, including donations (Hansen and Mitchell 2000; Mitchell et al. 1997). We measure a corporation’s media exposure (MEDIA) as the total number of references to the corporation in newspapers, radio, television, and periodicals in a particular year, as recorded by the parliamentary media coverage section of the ParlInfo Search; a database managed by Australian Federal Parliament. This is more extensive than measures used prior studies, which tend to focus on major newspaper references (Hansen and Mitchell 2000; Mitchell et al. 1997). We expect MEDIA to be correlated with PARL_EXPOSURE.

Board size may affect the likelihood of private interests arising or prevailing, or may influence the effects of governance mechanisms (Golden and Zajac 2001; Ruigrok et al. 2006; Ryan and Wiggins 2004). Board size (BDSIZE) is measured as the number of non-executive and executive directors sitting on the donor’s board.

Industry membership (Grier et al. 1994; Hart 2001) or year effects (Claessens et al. 2008; Snyder 1990) may affect donation patterns. We use dummy variables indicating firms’ 2-digit (sector-level) Global Industry Classification Scheme (GICS) classification to control for industry effects, and also include year fixed effect dummy variables.

Sample Selection and Descriptive Statistics

We study federal political donations by listed corporations during a period of eight years 2000–2007, during which time a coalition of two conservative parties (the Liberal and National parties) remained in government. This period has the advantage of including federal election cycles but avoids potential complications arising from a change of government. We first describe the donations by listed corporations during this period. Total annual corporate political donations made at the federal level by the top 100 and top 500 listed corporations for the period 2000–2007 are graphed in Fig. 1. The top 500 is defined by market capitalization at the end of the financial year. The top 100 comprises all corporations ranked as such by either total market capitalisation or total revenue. The percentage of firms in each tier of 100 (by market capitalisation or revenue) within the top 500 (by market capitalisation) that reported political donations in each year is shown in Fig. 2. Donations by our top 100 account for over 90 % of the donations made by the top 500 listed corporations. Therefore, we restrict our sample to the top 100 Australian corporations (ranked by either total market capitalisation or total revenue at the end of any financial year during 2000–2007). Any corporation in the top 100 (by either market capitalisation or revenue) in any sample year is included in our sample for the entire eight-year period. We do not require that all 8 years data are available for our sample firms, leading to unbalanced panels. This sample selection yields 1,339 firm-year observations, which was reduced to 1,049 cases by missing data (with 160–170 observations each year).⁷

Fig. 1

Total annual corporate political donations made at the federal level by the top 100 and top 500 listed corporations for the period of 2000–2007

Fig. 2

Percentage of firms in each tier of 100 with the top 500 corporations that reported political donations for the period of 2000–2007

We next explore the demographics of our sample. The industry distribution of our sample is shown in Fig. 3. The financial sector was the biggest contributor, consistent with evidence for earlier periods reported by Ramsay et al. (2001). The Australian financial services sector is one of the most heavily regulated sectors at the federal level and they represent a larger proportion of the firms in our top 100. Corporate donations to the coalition (Liberal and National) parties, the Australian Labor Party, and the total for all other (minor) parties by year are graphed in Fig. 4 and tabulated in Table 1 Panel A. Total donations were higher in federal election years and the incumbent coalition parties received the largest contributions.⁸ Our sample contains 878 firm-years (80 % of our sample) with zero donations, which might distort our regression results; therefore, we also report the results of Tobit regressions that allowing for this implied censoring.

Fig. 3

Industry distribution of the top 100 corporations that reported political donations for the period of 2000–2007

Fig. 4

Corporate political donations made to the coalition parties and the Labor Party by the top 100 corporations that reported political donations for the period of 2000–2007

Table 1 Corporate political donations by year and the number of firm-year observations

The annual distribution of reported donations is described in Table 1 Panel B. The mandatory disclosure threshold for political donations was $1,500 for 2000–2005, increasing to $10,000 after 2005 (with annual indexation). The reported annual donations between these two thresholds are summarized in Panel B; this shows 34 firm-year donations between $1,500 and $10,000 across the sample period. Although the disclosure of donations less than $10,000 is voluntary after 2005, the number of reported donations between $1,500 and $10,000 did not decline; nonetheless, in our further testing, we re-estimate our main models for all years excluding donations less than $10,000.

Descriptive statistics for the continuous variables are reported in Table 2 for the full sample and two subsamples: (1) the 65 unique corporations (contributing 447 firm-years) that made at least one donation during the sample period; and (2) the 123 unique corporations (contributing 647 firm-years) that made no donations during the sample period. The means for most of our continuous variables are significantly different between the two subsamples, except for GAZETTE_EXPOSURE, BLOCK, NONEX_SHARES and ROA. Compared to non-donating firms, donating firms tend to be larger with more active boards, and greater parliamentary and media exposure, and receive more government contracts. While smaller mean percentage of shares held by executive directors for donor corporations is consistent with size, the higher mean percentage of non-executive director shareholdings in donor corporations might warrant further investigation but is beyond our current scope.⁹

Table 2 Descriptive statistics for each continuous variable (excluding the interaction terms) for donors and non-donors

Multicollinearity Diagnostics

Table 3 reports the correlation matrix for our explanatory variables. As expected, MEDIA and PARL_EXPOSURE are highly correlated (0.547). Otherwise, the correlation matrix does not indicate any particular collinearity concerns. Likewise, the Variance Inflation Factor diagnostics do not indicate any multicollinearity issues, as no factor exceeds 2.5 and the benchmark is 10.0 (Kutner et al. 2004).

Table 3 Correlation matrix between each continuous explanatory variable, excluding the interaction terms, for the sample period 2000–2007

Results

We test our hypotheses using a range of regression methods, which are collectively suitable for addressing the main sources of potential bias in our data. We use OLS regressions with standard errors adjusted for within-firm clustering, Fama–Macbeth regressions, Tobit regressions specifying left-censoring of the dependent variable at zero and firm fixed effects regressions to address the possibility that corporations that never made donations during the sample period are inherently different from those that made donations during the period. Because our Fama–Macbeth and Tobit regressions results are substantively similar to our OLS results, our discussion of results focuses on the OLS models.

Tests of the Impact of Directors Networks (H1 and H2)

The results, as reported in Table 4, are consistent with our hypotheses. All measures of network influences (NET_DON_EX, NET_DON_NONEX; NPNET_EX, and NPNET_NONEX) are significant and positive in all regression specifications, except NPNET_EX in the FE regression for the sample of 65 firms (447 firm-year observations) that made at least one donation during the sample period (Column 4). The non-significant results for NPNET_EX in the FE regression might arise because the average number of non-profit board memberships held by executive directors varies little within firms.

Table 4 Regressions of DONATIONS against corporate and private interest proxies (model 1)

The positive coefficients for NET_DON_EX and NET_DON_NONEX indicate that, as the incidence of corporate political donations in directors’ professional networks increases, the level of political donations made by the directors’ corporation increases. The positive coefficients of NPNET_NONEX and NPNET_EX indicate that increased director involvement on non-profit boards also increases the level of political donations by the directors’ corporation. Overall, these results suggest a positive influence of managers’ and directors’ social network on the corporate political donations in their employing corporations.

With regard to the corporate interests control variables, PARL_EXPOSURE is positive and significant for all regression methods, suggesting that, on average, corporations with greater exposure in parliamentary proceedings make larger political donations. The results for the two indicators of a corporation’s exposure to government decision-making outside parliament, GAZETTE_EXPOSURE and CONTRACT_AWARDED, are less persuasive. GAZETTE_EXPOSURE is not significant in any regression and CONTRACT_AWARDED is is positive and weakly significant only in the Fama–Macbeth regression.¹⁰

Our corporate governance controls (board independence, BDINDEP or blockholder monitoring, BLOCK) are not significant in any of the regressions. SHARES_EX and SHARES_NONEX are significant and positive under all regression methods. The coefficients for SHARES_EX and SHARES_NONEX must be interpreted cautiously because larger directors’ shareholdings may indicate stronger interest alignment between directors and shareholders, thus reducing the likelihood of donations for private interests, or they might increases directors’ power to pursue their private interests.

Robustness Tests

Prior studies suggest a potential endogeneity problem regarding the director interlock variables, NET_DON_EX and NET_DON_NONEX (Hermalin and Weisbach 2003; Stuart and Yim 2010). Two potential sources of endogeneity are: (1) reverse causation boards that wish to engage in corporate political donations recruit directors with corporate political donating experience to their boards (e.g., Baker and Gompers 2003) and (2) common directorships may reflect an unobserved similarity or ownership connection between the interlocked firms, and it may be this commonality that causes each firm to make corporate political donations. To address these endogeneity concerns, we specify NET_DON_EX and NET_DON_NONEX as endogenous variables when re-estimating Model 1 using a two-stage least squares approach.¹¹ The first stage uses four instrumental variables (LIST_EX, LIST_NONEX, UNLIST_EX, and UNLIST_NONEX) to predict the likelihood of executive directors’ and non-executive directors’ interlocks to donating firms, where: LIST_EX (LIST_NONEX) is the number of executive directors’ (non-executive directors’) directorships of listed companies and UNLIST_EX (UNLIST_NONEX) is the number of executive directors’ (non-executive directors’) directorships of unlisted entities. Standard diagnostic tests confirm that our instruments are both exogenous to the errors in the structural equation, and collectively relevant to the prediction of the endogenous variables.¹² The results for the two-stage selection models are reported in Table 5. The first stage results indicate executive directors’ listed company directorships is significant in predicting executive directors’ interlocks to donating firms (Column 1) and both listed and unlisted corporate directorships are significant in predicting non-executive directors’ interlocks to donating firms (Column 2). However, directors’ interlocks to donating firms (NET_DON_EX and NET_DON_NONEX) remain significant in the two-stage models, consistent with the OLS regressions.

Table 5 Two-stage selection models to account for the potential endogeneity associated with NET_DON_EX and NET_DON_NONEX (Model 1)

The decision to make corporate political donations may not be random. Corporations that made donations during the sample period can be inherently different from those that did not make donations during the period. Therefore, we use the Heckman two-step approach to re-estimate Model 1. In the first step, we use a probit model to predict the probability that a company donated in a particular year. We include the inverse Mill’s ratio (or Heckman’s λ) obtained from the first step as an additional regressor in the second step estimation of Model 1. Because the factors that affect likelihood of donations are also likely to affect the level of donations, we use the same variables in both steps (probit and OLS) (Hart 2001).¹³ The results, as reported in Table 6, are similar to the OLS results; all indicators for managerial influences are significant in predicting the likelihood and levels of donations.

Table 6 Heckman two-stage estimation of Model 1

We also examine the potential impact of the change in the disclosure threshold from $1,500 to $10,000 after 2005 by comparing the OLS results for Model 1 for the two sample periods, as reported in Table 7 Panel A. Although significance levels are weaker, the results for NPNET_EX and NPNET_NONEX are consistent with our main results. It appears the change in disclosure in 2005 introduces noise to the regression results but, overall, does not substantially affect the associations between corporate political donation levels and either corporate interest or private interest indicators.

Table 7 Subsample results for test variables in Model 1

Our main tests focus on total donations to all parliamentary political parties but it is possible that recipient parties differ according to managerial influences. Therefore, we re-estimate Model 1, alternatively re-defining the dependent variable as: (1) DONATIONS_OPP = the natural logarithm of donations to the Australian Labor Party +1, which was the main opposition party for the duration of our sample period; and (2) DONATIONS_GOV = the natural logarithm of donations made to the governing Coalition parties +1. The results, as reported in Table 7 Panel B, are broadly similar to the main model. NET_DON_EX and NPNET_NONEX are significantly positive for donations to both the government and the opposition. NET_DON_NONEX is not significant in either of the party-specific regressions. A plausible reason for the difference in relation to NET_DON_EX and for NET_DON_NONEX is that non-executive directors are more likely to have multiple corporate directorships in other corporations than executive directors; the increase in professional networks may increase incentives for corporate political donations preferred by non-executive directors, particularly with regard to the recipient of donations, relative to executive directors. Alternatively, because executive directors are more directly involved in the daily operations of firms, interlocked executive directors of the focal corporation are more likely than non-executives to influence the occasional donating behavior of that firm, as suggested by literature on interlocking directorates (e.g., Bond 2004; Davis 1991; Haunschild 1993; Haunschild and Beckman 1998; Palmer et al. 1995).

We also examine whether managers and directors have influences over corporate political donations made to the state-level branches of the major parliamentary parties. We re-estimate Model 1 using alternative dependent variables based on the donations made to the political parties in four states: New South Wales, Victoria, South Australia and Western Australia.¹⁴ Donations and exposure variables (PARL_EXPOSURE, GAZETTE_EXPOSURE, CONTRACT_AWARDED) are measured at the state level using the same methods as for the main analysis. The results, as reported in Table 8, exhibit varying significance levels but are generally consistent with the federal level results in our main analysis.

Table 8 Regressions of donations made to state branches

Summary and Conclusion

We examine managerial influences of corporate political donations from a social network perspective. We find that both executive and non-executive directors’ networks influence corporate political donations, with influences arising from both professional and non-professional networks. The affects appear to be strongest for executive directors’ professional networks. This is robust to endogeneity treatments, sub-period analysis and donations to both the government an opposition parties. The further tests for donations at the state level do not exhibit a clear pattern of influence with respect to executive versus non-executive networks; further research is necessary to reveal whether differences between states reflect differences in state-level political interests or non-systematic effects.

There may be some uncertainty as to the best interpretation of our measure on non-professional networks, and why involvement in non-profit organizations may influence political preferences. Nonetheless, our results in this regard appear robust and are stronger effects for executive directors’ are consistent with the results obtained for professional networks. This reinforces the observations in prior studies of charitable donations that executive directors are more influential than non-executive directors in this regard.

Our study also extends the political science, economics, and management literature linking corporate political donations and government decisions but which do not directly include managerial private interests (e.g., Aggarwal et al. 2008; Ramsay et al. 2001), and do not establish a conclusive link between donations and favorable government decisions (Rehbein and Schuler 1999). Given the strong evidence of directors’ network influences presented here, we suggest that omitting these influences from donation studies impedes reliable detection the relations between donations and corporate motives.

Our results also give substance to shareholder concerns regarding corporate political donations. The evidence of strong influence from directors’ networks in this regard suggests external shareholders cannot rely on internal monitoring. While further research is needed in this regard before advocating particular future policy developments, it lends weight to the calls by the Australian Shareholders’ Association (2004) for improved accountability with respect to corporate political donations to facilitate shareholder monitoring, or possibly requiring approval of corporate political donations.

Appendix: Details of data collection and variable measurement

Data on corporate political donations were obtained from the Australian Electoral Commission’s (AEC) webpage (http://www.aec.gov.au). Because the AEC does not identify whether a donor is a listed corporation, and the names reported by the AEC are those provided by the donors, using various abbreviations or truncated forms of current and former names. Thus, to identify donors that are listed corporations, and consolidate donations made by wholly-owned subsidiaries and their parent company, we used the following data collection process: (1) we obtained all the names used by a corporation, including their abbreviated forms, during our study period from Aspect FinAnalysis and assign each corporation a unique identity using the company group name code from the Centre for Research in Finance (CRIF) SPPR database; (2) we identified the names of the fully owned subsidiaries for each of our sample firms in a particular year from Connect4’s companies’ annual reports database and Thomson SDC Platinum takeover database; (3) we compared these names with those on the AEC annual donor list to identify donations made by our sample firms and the subsidiaries of the sample firms in a particular year¹⁵; and (4) we consolidated the donation data between subsidiary and parent company.

Because a corporation’s name referenced in Parliamentary Hansards can vary, we search all identifiable variants of names including its current and previous Australian Stock Exchange listing codes as the search terms in the ParlInfo Search (http://parlinfo.aph.gov.au/), a search engine provided by the Parliament of Australia that covers Australian Parliamentary information resources, including Hansards. For each search of a corporation’s name, we manually checked all resulting citations to verify they actually applied to the corporation in our sample and was not merely an entity with a similar name.

To collect data on the number of times a corporation is referenced in the Government gazettes (excluding the Government Tendering Gazette, AusTender), we first obtained a list of all current and previous series of Government Gazette publications at the Commonwealth level from the Australian Government Publication Guide (National Library of Australia 2012). Each series comes with a link to its relevant website, which we searched using the same protocol as for the Hansard search to obtain and verify the citations, and summed the number of citations of a corporation across all series as the measure of exposure.

Government contract information is normally published in AusTender, a centralized government procurement information system that administers the advertisement and award of government contracts for information on all the applicants. However, because the data on government contract(s) prior to 2007 are not available online, these data were obtained with a file provided by the federal Department of Finance and Deregulation, which manages the government contract databases. The data are of the same type as those are available online. We compared each corporation’s name(s) with the government contract information to identify the number of government contracts awarded to a corporation in a particular year.

To collect data on NET_DON_EX; NET_DON_NONEX; NPNET_EX; and NPNET_NONEX, we first collected the information on each director’s corporate and non-corporate directorships from company annual reports available on Connect4, Aspect DatAnalysis, the Australian Stock Exchange’s website (http://www.asx.com.au/) or individual, company websites. To ensure that each director is correctly identified, we reconciled each director’s surname, middle name and given name and other information, such as their employment history and age as reported by Connect4, with similar data in DatAnalysis and the company’s annual reports and websites. Each director was then assigned a unique identifier that was used in the measurement of interlocks with other companies and the number of non-profit boards.

The information on a director’s corporate and non-profit directorships disclosed in the annual report is not standardized across firms, so different corporations may disclose such information differently for the same director.¹⁶ To address this issue, we used the following procedure: (1) we first collected the corporate and non-profit directorship information for each director from each corporation’s annual report; (2) for each director, we consolidated all the corporate and non-profit directorship information across the sample firms, and grouped this information under each director identifier on a yearly basis, to identify all the profit and non-profit board membership a director has during the year; and (3) we reconstructed director’s corporate and non-profit directorship information for each firm-year observation in our study. This process of consolidating information aims to correctly identify the directorship information for directors of each corporation.

The corporate directorship information indicates to which corporation(s) a director of the focal corporation has interlocks. We followed the steps described taken to collect donation data to check whether the interlocked corporation made a political donation in the relevant year. We then calculated the total number of interlocked corporations that had made political donations by executive directors (non-executive directors), for each focal corporation, for each year and deflated the total by one plus the number of executive directors (non-executive directors).¹⁷

Data on corporate governance mechanism measures were collected from company’s annual reports, available on Connect4, Aspect DatAnalysis, the Australian Stock Exchange’s website (http://www.asx.com.au/) or individual company websites. This information includes the director’s positions (executive director, non-executive director or chairperson), executive and non-executive directors’ shareholdings, and the number of blockholders. Data on firm size, the market capitalisation and the 2-digit GICS code were obtained from the CRIF SPPR database. Board size and ROA were obtained from company annual reports. Data on the number of times a corporation was referenced in the media were collected by searching a corporation’s name(s) in the media coverage section of the ParlInfo Search, which covers press releases, newspaper clippings, and radio and television broadcasts.