Abstract
What drives organizational nonconformity to global corporate governance norms? Despite the prevalence of such norms and attendant conformity pressures, many firms do not adhere to them. We build on a political view of corporate governance to explore how different national institutional contexts and organizational conditions combine to produce over- and underconformity to global board independence norms. Using configurational analyses and data from banks in OECD countries, we identify multiple equifinal combinations of conditions associated with over- and underconformity. We also find that over- and underconformity have different drivers. We conjecture that while overconformity is associated with a shareholder–management coalition in liberal market economies, underconformity results from multiple complex combinations of national and organizational conditions that often include dominant blockholders, strong labor rights, and small organizational size. We leverage these findings to abduct theoretical insights on nonconformity to global corporate governance norms. Doing so sheds light on the role of power in conditioning the adoption of global practices and contributes to research on international corporate governance by informing discourse surrounding the globalization of markets.
Keywords
What drives organizational nonconformity to global corporate governance norms? 1 Despite evidence-based prescriptions and attendant pressure to adopt “best practices,” many firms do not adhere to global corporate governance norms, instead choosing to diverge by either underconforming or overconforming (Aguilera, Judge, and Terjesen, 2018). While there is research on the antecedents and outcomes of various governance practices, little is known about when and why firms do not conform to these practices. Understanding the drivers of such nonconformity is important not only because corporate governance has significant implications for organizational outcomes (Krause, Semadeni, and Cannella Jr., 2014) but also because deviations from norms relate directly to theories of firm heterogeneity and organizational change (Deephouse, 1999).
In this study, we explore how national institutional contexts and organizational conditions form configurations that lead to over- or underconformity to global board independence norms. We focus on board independence because it is the most prominent corporate governance dimension in academic research and practical prescriptions (Neville et al., 2019; OECD, 2019). With increasing prevalence in practice and weight in academic literature (Dalton et al., 1999; García-Meca and Sánchez-Ballesta, 2009; Post and Byron, 2015), majority independent boards are now seen as “synonymous with good governance” in corporate governance research and in regulatory circles (Neville et al., 2019: 2539). However, while a global norm of majority board independence has emerged as a result (Zattoni et al., 2017), corporate boards’ independence levels vary widely around the globe.
To facilitate theoretical understanding of nonconformity to global norms, we draw on a political perspective inspired by prior corporate governance research anchored in sociological institutionalism (Palmer, Jennings, and Zhou, 1993; Dobbin, 1994; Fligstein and Freeland, 1995) and by comparative corporate governance research based on historical institutionalism (Jackson, 2001; Aguilera and Jackson, 2003). Albeit differing in important aspects, both perspectives hold that corporate governance structures are the result of political processes, wherein “coalitions with different interests and capacities for influence vie for dominance” (Palmer, Jennings, and Zhou, 1993: 103; see also Roe, 1994). Viewed through this lens, board composition reflects power differentials among actors with influence over director appointments, and these power differentials are shaped by the wider institutional environment (North, 1990; Fligstein and Freeland, 1995; Hall and Soskice, 2001). Prior research suggests that in the United States, managers and shareholders are the central actors shaping board composition (Fligstein and Brantley, 1992; Roe, 1994; Aguilera, 2005). In international contexts, labor often emerges as a third core actor (Aguilera and Jackson, 2003; Adams, Licht, and Sagiv, 2011). Given that these actors may have diverse and often diverging interests with respect to board independence, we argue that the institutionally derived power of each of these three actors represents a key layer of drivers for nonconformity to global norms (Connelly et al., 2010).
In addition, we suggest that organizational factors might act as boundary conditions to these political processes. Firm-level characteristics can themselves be sources of power and thus shape “conflict among rival groups for scarce resources” (Hall and Taylor, 1996: 937; see also Roy, 1997). That is, organizational factors may increase or decrease the extent to which managers, shareholders, and labor can draw on power derived from the national institutional context to influence board composition. Hence, national institutional and organizational sources of power are systematically interdependent and may form distinct, context-specific combinations that shape corporate governance practices (Greckhamer, 2016). Prior studies have examined how interactions between institutions and organizational factors might shape nonconformity (e.g., Westphal and Zajac, 2001; Westphal and Graebner, 2010) but have paid limited attention to the nature of these interactions across distinct national institutional contexts. This has resulted in limited scholarly understanding of how firms pursue nonconformance and what kinds of firms do not conform to global norms under various institutional conditions.
With this lacuna in mind, we suggest that a theoretical prism of political power that integrates national institutional variety and the distinct ways in which different firms respond to it can advance theory of international corporate governance and, more broadly, of organizational heterogeneity. Accordingly, to accommodate the causal complexity and multilevel nature of our proposed comparative institutional framework, we apply a configurational analytical approach to data on banks from OECD countries. Our findings show that multiple equifinal configurations of conditions are associated with over- and underconformity to global board independence norms, indicating that different types of firms exhibit nonconformity in distinct institutional contexts and that configurations leading to underconformity are distinct from those resulting in overconformity. We leverage these findings to abduct theoretical insights on the political power configurations that drive nonconformity to global corporate governance norms.
Our study makes two primary contributions. First, our comparative cross-national approach to nonconformity allows us to identify factors leading to under- and overconformity in a global context, and our politics-driven theoretical framework offers a novel perspective on corporate governance theory. Delineating why some firms deviate from global norms in different national contexts enriches theoretical understanding of variance in governance practices across countries and firms. Second, our configurational approach enables theorizing on equifinal yet asymmetric combinations of nonconformity drivers, thus revealing the complex configurations shaping nonconformity to global norms. Doing so further sheds light on the political power of key actors as an underlying mechanism enabling or impeding the adoption of global industry norms, thus informing theory in the sociology of globalization literature surrounding the extent to which firms abide by global practices (Guler, Guillén, and Macpherson, 2002; Fiss and Zajac, 2004; Ahmadjian and Robbins, 2005; Davis and Marquis, 2005).
As in prior studies employing an abductive approach along with configurational theorizing and analysis (e.g., Misangyi and Acharya, 2014; Campbell, Sirmon, and Schijven, 2016; Dwivedi, Joshi, and Misangyi, 2018), this paper is organized differently from a deductive study. In keeping with Furnari et al. (2020), we first synthesize empirical and theoretical literature on comparative corporate governance and nonconformity to scope the institutional and organizational factors that shape actors’ power to influence board independence. We then subject these factors to empirical analysis to uncover the configurations that explain nonconformity to global board independence norms, enabling us to elaborate theory of the political power dynamics underpinning this nonconformity. We close with an exposition of the implications of our study for future research.
Nonconformity Drivers: A Guiding Framework
Many firms do not follow global norms with respect to board independence levels. Some firms are governed by a majority of insider directors, thereby underconforming to the widely accepted norm of majority independent directors on the board. Other firms choose not to conform in the opposite direction, thus overconforming. Although the prescribed norm of “a majority of independent directors” does not intuitively indicate the possibility of overconformity, in practice, the descriptive norm converges on a “50+1” independence ratio, which leaves ample room for overconformity. Our study is centered on nonconformity in the context of comparative practices rather than on conformity to the loose notion of “majority independence.” For example, some listed U.S. firms have a board in which the CEO is the only insider, going well beyond the recommended and descriptive norm (Zorn et al., 2017). This practice is not necessarily desirable, as it might enhance CEO power by enabling this individual to filter and control firm-specific information (Lewellyn and Fainshmidt, 2017: 1606), possibly resulting in CEO capture of the board. While such lone-insider boards do not necessarily imply nefarious CEO action and may well reflect overzealous observance of the board independence norm, regardless of the underlying motivations, these cases represent overconformity to global norms.
Nonconformity to norms has been extensively researched in sociology, especially focusing on why organizations sometimes defy norms or common practice as well as why they adhere to global versus local institutional pressures (Sassen, 2007). In corporate governance research, Aguilera, Judge, and Terjesen (2018) noted that there are typically norms or common practices for corporate governance in a given industry. In the case of the board of directors, governance norms bound the typical composition and responsibilities of the board, which subsequently influence the allocation of resources, profits, and risk among the firm’s stakeholders. Because norms define which governance practices and, in particular, board structures are legitimate, many firms tend to operate with boards that approximate global descriptive norms in an industry (Aguilera, Judge, and Terjesen, 2018: 96). Yet institutional theory indicates that organizations can find room not to conform to norms (Oliver, 1991).
The corporate governance literature has provided extensive documentation of variance in board structure and composition across firms and countries (Neville et al., 2019; Wang et al., 2019). However, although much work in corporate governance examines the adoption and diffusion of new or imported governance practices (Aguilera and Cuervo-Cazurra, 2004; Haxhi and Aguilera, 2017), much less is known about who exhibits nonconformity under what circumstances to established, global corporate governance norms. This reflects two limitations in current theory. First, very little work has explored nonconformity in the corporate governance context (Greenwood et al., 2011; Miller, Le Breton-Miller, and Lester, 2013). As a result, we have a robust picture of the drivers of board composition but a very limited understanding of why and when such composition fails to conform to industry norms.
Second, as Thornton, Ribeiro-Soriano, and Urbano (2011) suggested, nonconformity may be exhibited by different types of firms under different institutional contexts, implying that institutional and organizational factors work in conjunction to shape nonconformity (Greenwood et al., 2011). For instance, Miller, Le Breton-Miller, and Lester (2013) found that in contexts in which transparency and “good governance” are expected, family-owned and family-managed firms tend to gain legitimacy in the eyes of wary stakeholders by conforming more closely to institutional norms. Such firms might be less keen to conform in contexts in which local institutions empower blockholders whose governance goals may not align with global norms. Hence, within different institutional contexts, different types of actors may engage in nonconformity, highlighting the need to allow for equifinality in explaining nonconformity. This notion is conceptually consistent with findings that interrelated characteristics of both the organization and the environment may shape conformity (e.g., Miller and Chen, 1996), but prior research has tended to theorize unifinal, isolated relationships between nonconformity and its drivers.
These limitations are exacerbated by the fact that corporate governance nonconformity can take two distinct forms. As Aguilera, Judge, and Terjesen (2018) argued, prior research overlooks the distinction between firms that fall short of a given governance norm (underconform) and firms that exceed it (overconform). Existing theory does not afford insights into how the organizational and institutional drivers of these two distinct forms of nonconformity differ.
National Institutional Drivers of Nonconformity
In exploring the drivers of nonconformity, we follow the approach adopted by recent configurational studies (e.g., Campbell, Sirmon, and Schijven, 2016; Greckhamer, 2016) to identify the most salient conditions, synthesizing both theoretical and empirical literature. We focus on factors that influence board composition as well as nonconformity, adopting a perspective that sees corporate governance structures as the result of political processes shaped by both national and organizational factors. The role of political processes in influencing national institutions is well documented in the literature, both for institutions in general (Hall and Soskice, 2001; Thelen, 2004; Iversen and Soskice, 2009) and corporate governance institutions in particular (Jackson, 2001; Aguilera and Jackson, 2003). Politically created institutions tend to benefit those who set them up and become a source of path dependency that reinforces and cements their creators’ power over time (North, 1990; Hall and Taylor, 1996).
In current corporate governance models and the wider institutional contexts in which they are embedded, shareholders and management have been the key forces, especially in the United States (Roe, 1994; Fligstein and Freeland, 1995; Hall and Soskice, 2001). Internationally, labor has often been involved as a third force in the creation of corporate governance regimes and the wider institutional context (Hall and Soskice, 2001; Jackson, 2001; Aguilera and Jackson, 2003). Since these actors do not always share the same interests with respect to board independence, as we discuss below, national differences in their relative power configurations may lead to deviations from global norms. Hence, the nature of nonconformity to global board independence norms is likely to be related to the national institutional structure and the attendant power differentials among capital (shareholders), labor (employees), and management. We discuss each of these in turn.
Shareholder rights
Shareholders usually represent a powerful force in terms of determining board composition. Shareholder influence reasserted itself with the rise of the shareholder conception of control in the United States in the 1980s (Davis and Thompson, 1994; Fligstein, 2001) and its subsequent international diffusion (Aguilera and Jackson, 2003; Fiss and Zajac, 2004; Ahmadjian and Robbins, 2005), with attendant institutional reforms strengthening shareholders’ statutory rights (Aguilera and Cuervo-Cazurra, 2004; Aoki, Jackson, and Miyajima, 2007; Yoshikawa and Rasheed, 2009). In general, we would expect shareholders to pressure firms to adopt corporate governance practices perceived to be aligned with their monitoring interests, usually implying higher levels of independence (Klapper and Love, 2004). It is possible that shareholder power may substitute some corporate governance structures—for example, if shareholder interests are well-protected through the legal system, formal adherence to global corporate governance norms may become less important. However, extant comparative corporate governance research suggests that shareholders institutionally empowered by strong rights tend to favor higher levels of board independence (Tuschke and Sanders, 2003; Aguilera and Cuervo-Cazurra, 2004; Kim, Kitsabunnarat-Chatjuthamard, and Nofsinger, 2007). Still, interdependencies with other institutional and organizational conditions may determine whether shareholders push for conformity or overconformity, as we discuss below.
Labor rights
The power of labor (employees) may likewise codetermine corporate governance arrangements. In Anglo-Saxon countries, the destruction (UK) or absence (U.S.) of traditional forms of labor organization and the subsequent political repression of unions from industrialization (Thelen, 2004) to the recent past (Jung, 2017) have effectively suppressed labor as a factor in corporate governance (Jackson, 2001). By contrast, non-liberal market economies such as Germany and Japan have evolved patterns of corporate citizenship that brought labor into the boardroom (Roe, 1994), whether to enable the creation of long-term organizational capabilities or to coopt labor so as to reduce popular support for a communist revolution (Jackson, 2001). Legal frameworks for labor rights have coevolved with and reflect these historical processes (Locke and Thelen, 1995; Aguilera and Jackson, 2003; Thelen, 2004).
Drawing on these rights, labor has the power to influence board composition in ways that may shift board independence toward over- or underconformity. On one hand, to the extent that independent directors are at arm’s length from the firm and bring broader conceptions of shareholder or stakeholder value to the board, labor may see these directors as more aligned with its interests than insiders who fulfill a double role of directors and managers (Luoma and Goodstein, 1999; Aguilera, 2005). This may prompt labor to pursue higher levels of board independence and, potentially, overconformity. On the other hand, labor may be concerned that independent directors, especially those with executive positions elsewhere and in jurisdictions with a stronger shareholder orientation, may prioritize shareholders’ interests over those of employees. In these cases, labor may seek to secure board appointments of sympathetic insiders (Aguilera and Jackson, 2003; Aguilera and Jackson, 2010) and thus advocate for underconformity. Overall, prior literature leads us to expect that labor power will more likely be associated with lower board independence, though labor’s stance toward the board might depend on conflict and mutual interests with other powerful actors.
Managerial discretion
In many institutional contexts, managers—especially CEOs—are powerful actors with influence over board structure. The foundation of this power is managerial discretion (Fligstein, 2001; Crossland and Hambrick, 2007) resulting from the historical rise of large, managerial hierarchies (Chandler, 1984) and the attendant separation of ownership and control in the modern firm (Fligstein, 2001). Defined as the range of plausible options open to management and shaped, in part, by the external institutional environment (Finkelstein and Hambrick, 1990; Finkelstein, Hambrick, and Cannella, 2009), “managerial discretion is the primary mechanism through which national institutions influence CEO effects on organizational outcomes” (Crossland and Hambrick, 2011: 798). While this suggests the power to influence board composition, prior research offers no clear guidance on the outcomes management is likely to pursue. CEOs with high discretion may support a conforming board composition to alleviate legitimacy or agency concerns (Wangrow, Schepker, and Barker, 2015). Conversely, powerful CEOs might promote a CEO-lone-insider board or an insider-dominated board to facilitate more favorable board decisions and enhance CEO power over other actors such as shareholders and labor (Lewellyn and Fainshmidt, 2017). Overall, while we expect higher managerial discretion to be consequential for nonconformity to global board independence norms, prior research does not offer clear guidance on the direction of that nonconformity, again suggesting that interdependencies with other powerful actors might be at play.
Organizational Boundary Conditions
Organization-level characteristics have the potential to alter the power that shareholders, labor, and management can bring to bear on shaping corporate governance structures. The management literature has extensively documented how powerful constituencies inside and outside the organization shape its structure and behavior (Finkelstein and Hambrick, 1990; Bigley and Wiersema, 2002; Hambrick, 2007; Aguilera and Crespi-Cladera, 2016). Importantly, where organizational characteristics strengthen or weaken the relative power of shareholders, labor, and management as provided for by national institutions, the empowered actors are likely to demand structural and behavioral modifications within the organization that reflect their interests. Depending on the direction of these interests, the outcome may be closer to or further from conformity to global norms. Prior research in international corporate governance suggests three organizational conditions that may alter the institutionally driven relative power configuration: firm ownership, organizational size, and board leadership structure.
Firm ownership
The corporate governance literature has identified ownership as the backbone of corporate governance, with studies focusing on the separation of ownership and control, principal–agent problems, and principal–principal conflicts (Jensen and Meckling, 1976; Young et al., 2008). In the corporate governance conformity context, both the extent of ownership concentration and the type of ownership are salient. Cross-national governance research suggests that ownership concentration fundamentally changes how an organization is governed and the extent to which it conforms to institutional pressures (La Porta et al., 2000). Blockholders with interests that diverge from those of other actors may use their power to create governance structures that reflect their interests, possibly at the expense of other shareholders and stakeholders more generally. Consistent with this notion, Aguilera et al. (2015) argued that in firms with ownership concentration, blockholders tend to dictate governance practices, while ownership dispersion allows for more interested parties to influence how the firm will be governed. Similarly, Anderson and Reeb (2004) found that family blockholders prefer lower levels of board independence, while the opposite is true for non-family shareholders. Compared to firms with dispersed ownership, firms with concentrated ownership may be relatively insulated from institutional pressures and may therefore more easily reflect blockholder preferences when these differ from norms. Similarly, ownership concentration may reinforce shareholder power, potentially dampening the extent to which labor and management are able to exercise meaningful power to define board composition.
Ownership type is also theoretically relevant. In particular, foreign investors tend to favor the diffusion of corporate governance practices across national borders (Aguilera et al., 2017), which may induce firms to conform to global norms. For example, Desender et al. (2016: 350) argued that with high levels of foreign ownership, “independent directors will have greater incentives to protect shareholders’ interests by monitoring,” and “we expect such behavior to be absent when foreign ownership is low, as . . . shareholders employ a different set of governance mechanisms to protect their interests.” Further, foreign owners relate differently to national institutions, potentially leading them to promote corporate governance practices that may differ from those of domestically owned firms (Yoshikawa, Rasheed, and Del Brio, 2010). That is, foreign owners with goals and concerns that vary from those of domestic owners may leverage power rooted in shareholder rights differently than their domestic counterparts. In sum, we expect nonconformity to global board independence norms to be associated with the extent to which blockholders and foreign owners are present, with the presence of blockholders leading to underconformity and the presence of foreign owners to conformity. That is, blockholders and foreign owners introduce causal complexity to the relative institutional power of shareholders, labor, and management in shaping board composition.
Organizational size
Political processes in large firms are inherently different from those in small firms. Small firms tend to attract less stakeholder pressure, have fewer connections to existing institutional arrangements, and hold relatively lower status (Porac et al., 1995; Greenwood et al., 2011). Larger firms are often made exemplars of institutional pressures from stakeholders (Wry, Lounsbury, and Greenwood, 2011), while smaller firms are less likely to receive social nudging (Zuckerman, 1999; Westphal and Zajac, 2001). In terms of board independence, these notions suggest that even if political processes among key governance actors favor nonconformity, larger firms might find it difficult not to conform because their size subjects them to stronger institutional pressures. Smaller firms have less reason to conform to such pressures, thus offering a greater scope for nonconformity (Leblebici et al., 1991; Kraatz and Zajac, 1996; Nee and Ingram, 1998). However, large organizations may deviate from prevailing norms if they are beyond regulatory agents’ control (Wry, Lounsbury, and Greenwood, 2011) or if institutions empower actors interested in deliberately leveraging the firm’s sizable resource base and visibility as a platform to challenge prevailing practices, either for instrumental or moral reasons (Greenwood and Suddaby, 2006). Overall, the literature largely supports the notion that small organizations are more likely to nonconform, but it is possible for some large firms to nonconform as well, depending on the powerful actors steering them.
Board leadership structure
Corporate boards may implement a combined leadership structure whereby the CEO also chairs the board of directors (Wang et al., 2019). This structure, known as CEO duality, can exert significant influence on political processes that shape board independence. Although board independence can reduce the likelihood of introducing CEO duality to begin with, once in place, duality dampens board monitoring of the CEO and enhances CEO power to shape future board appointments (Wang et al., 2019). As a result, CEO duality extends the power of management (Krause and Semadeni, 2014), possibly at the expense of shareholders and employees. However, as with managerial discretion, it is not clear how CEO duality affects board independence. CEO duality might bring about greater scrutiny of a firm and thus restrict nonconformity to norms, but it can also provide the latitude for CEOs to shape the board in ways favorable to management. The outcome likely depends on the susceptibility of the firm and its most powerful actors to external scrutiny of its governance practices.
In sum, viewed through our political perspective, prior research has suggested a range of factors that may shape nonconformity to global corporate governance norms (see Figure 1). While each of these factors may be salient in its own right, board composition will be determined by the overall power constellation these factors create—hence, it is the combinations among the institutional and organizational factors that lead firms to over- or underconform. This pattern is consistent with our earlier argument that a compelling explanation of nonconformity requires that we accommodate the asymmetric nature of nonconformity and the systematic interdependencies among national institutional and organizational factors shaping distinct power configurations within firms. Given the insufficient theoretical basis to hypothesize which combinations of these factors will result in over- or underconformity, we do not posit any specific hypotheses a priori. Instead, we use this repertoire of factors to explore empirically the following research question: Which configurations of national and organizational factors lead firms to over- or underconform to global board independence norms?
Theoretical Framework of Nonconformity to Global Board Independence Norms
Data and Method
Data
Sample
We explore our research question using a sample of organizations from the banking industry (SIC codes 602 and 603). Banks represent a suitable context for several reasons. First, the global banking sector offers a particularly visible corporate governance context characterized by transnational connectivity and strong global common governance and compliance practices, thus allowing us to explore nonconformity conservatively. Corporate governance in the banking industry has received much public scrutiny, which suggests that conformity pressures are likely to be higher here than in other industries. This makes deviations from norms likely to be the result of deliberate action (for overconformity) or inaction (for underconformity). At the same time, banks are incorporated under the jurisdiction of a single country and thus embedded in national institutions as well, placing them at the nexus of global industry practices and their local institutional contexts. Second, focusing on one industry allows us to hold constant some factors influencing industry dynamics, such as the type and level of regulations, competition, and government intervention. Third, unlike in most other industries, there is a sufficiently large number of banks among listed corporations internationally to make a meaningful analysis possible.
We obtained data for the period from 2014 to 2017. Our measure of labor power is not available before 2014, and the latest company results available at the point of data collection were for 2017. We obtained data for all listed banks in major advanced industrialized countries, as signified by membership of their home countries in the Organization for Economic Cooperation and Development (OECD). We chose this group of countries for three reasons. First, the focus on OECD countries allowed us to maximize data reliability and comparability. Second, by focusing on advanced industrialized countries, we were conservative in setting the institutional boundaries of our sample, allowing us to alleviate concerns about alternative explanations such as lack of rule of law or non-democratic political regimes. Third, the OECD context presents a hard test for drivers of nonconformity because there are clear expectations about corporate governance within the OECD, paired with generally good implementation of regulations and laws. We included only countries for which we had data for a minimum of three banks per year on average. We set this threshold because our analysis involved some comparisons within countries, for which our method—explained below—required at least three data points.
Our initial sample meeting the above criteria contained 1,787 observations across 18 countries and 4 years (i.e., an average of 447 banks for each year). However, 74 percent of the sample consisted of U.S. banks, as there is a very high number of small, listed banks in that country. This implied a risk that our results might be driven predominantly by differences within the United States rather than by a combination of differences within and across countries as intended. We consequently trimmed the number of U.S. observations to the same level as that of the next largest country subsample, Japan, by removing the smallest U.S. banks. Our final sample thus comprised 521 observations across 18 countries. Table 1 shows the number of observations by country.
Observations by Country
Measures
Table 2 provides an overview of all measures with descriptive statistics as well as the sources from which we obtained our data. The basis of our outcome condition, the proportion of independent directors on the board, is standard in the corporate governance literature. We obtained this measure from the BoardEx database. As indicated in Table 2, board independence levels run the gamut from 0 to 1, with a median of 0.58 (i.e., majority independent). We conceptualize board independence as ranging from high levels of underconformity, to conformity in the middle of the distribution, to high levels of overconformity. Low levels of independence thus represent underconformity; high levels, overconformity; and medium levels, neither underconformity nor overconformity (i.e., conformity). The empirical thresholds for membership in these three categories were determined by the calibration process, which we detail in the Method section below.
Measures, Data Sources, and Descriptive Statistics before Calibration*
N = 521. Summary statistics for revenues are unavailable because revenues are denominated in local currency units and thus not comparable across the entire sample.
In terms of causal conditions, at the country level, we drew on the World Bank’s shareholder protection measure to capture the strength of shareholder rights. For labor rights, we used the International Trade Union Congress (ITUC) Global Rights Index to capture employee strength, which we inverted so that higher values indicate more rights and thus greater strength. A relatively recent measure, the Global Rights Index has been adopted in a number of comparative studies (Zattoni et al., 2017; Witt et al., 2018) because it expresses rights that workers enjoy in practice. This gives it higher construct validity than traditional measures such as unionization rates, which say little about how much influence labor actually has. For instance, unionization rates in Germany are fairly low, at about 20 percent, but labor enjoys strong rights based on legal features such as the right to determine half of the membership of supervisory boards.
We further included a national-level measure of managerial discretion. Crossland and Hambrick (2011) presented a national-level managerial discretion index for 15 countries. Since only 12 of these countries overlapped with our sample comprising 18 OECD countries, we drew on their work to recalculate and extend the managerial discretion index.
While the original index was survey-based, we calculated our extension on the basis of the seven correlates that the original study identified and validated: individualism; uncertainty tolerance (i.e., the inverse of uncertainty avoidance); power distance and cultural looseness to capture informal institutions; and ownership dispersion, legal origin, and employer flexibility to account for formal institutions. In line with the original, we used Hofstede (2001) and Gelfand et al. (2011) for informal institutions, as well as La Porta and colleagues (1999) for legal origin. For the remaining formal institutional measures, the original sources lacked up-to-date information for our countries of interest. We consequently drew on the OECD for ownership information (De La Cruz, Medina, and Tang, 2019) as well as data for employer flexibility, operationalized as in the original as the inverse of employment protection (OECD, 2020).
To maximize comparative validity of our calculations, we included all 27 OECD countries for which we could obtain the requisite data. Since the cultural looseness index is not available for four of these 27 countries, we adopted the imputed values for three of these from Crossland and Hambrick (2011): Canada received the same score as the United States; Sweden, the same score as Norway; and Switzerland, the mean of France, Germany, and Italy. For Denmark, which has a reputation as culturally more relaxed than its Scandinavian neighbors (e.g., Worrall, 2015), we assigned the average for Germany and the Netherlands.
We applied factor analysis as implemented in Stata 16.1. The analysis yielded one factor with an eigenvalue larger than the required threshold of 1. The resultant predicted values are highly correlated with Crossland and Hambrick’s original measure (r = 0.90) and significantly extends their index from 12 to 27 OECD nations, thus providing an empirical contribution that opens comparative research opportunities drawing on this construct.
Table 3 shows the original and newly calculated values, standardized between 0 and 1 to aid comparison. Countries included in our analysis are marked with shaded background. Countries showing larger deviations are Austria, Italy, and Japan. We believe that for these countries, the new index values are more plausible than the original, as they bring their values into closer alignment with the societies in the sample that are institutionally most similar: Germany for both Austria and Japan, France and Spain for Italy (see Witt et al., 2018).
Managerial Discretion Scores*
Values are standardized between 0 and 1 to aid comparison. Countries included in our analysis are highlighted.
At the organizational level, we included revenues to capture firm size, which is standard in the literature and was obtained from the Eikon database. For firm ownership, we used the proportion of closely held shares to indicate ownership concentration and the proportion of foreign holdings to express the presence of foreign shareholders. We obtained these measures from the Eikon database. We further included a binary measure of CEO duality as supplied by the BoardEx database, with a value of 1 indicating that the CEO is also the chairperson of the board.
Method
We applied fuzzy-set qualitative comparative analysis (fsQCA) as implemented in the software package fsQCA 3.1b (Ragin, Drass, and Davey, 2019). This method has been increasingly adopted in the management literature (e.g., Fiss, 2007; Bell, Filatotchev, and Aguilera, 2014; Judge, Fainshmidt, and Brown III, 2014; Judge et al., 2015; Witt and Jackson, 2016). FsQCA relies on set theory and Boolean algebra to determine which combinations of causal conditions are sufficient for an outcome (Ragin, 2000). Outcomes and causal conditions—the fsQCA equivalents of dependent and independent variables—assume values between 0 and 1, where 0 stands for “fully not a member of this set” (short: “fully out” or “fully absent”) and 1 denotes “fully a member of this set” (short: “fully in” or “fully present”). FsQCA utilizes “fuzzy sets,” which permit for gradations in set membership. Generally, 0.33 means “more out than in,” 0.50 “neither out nor in,” and 0.67 “more in than out.”
We chose fsQCA over conventional regression analysis techniques because fsQCA offers a range of analytical advantages relevant to this study. First, we expect the causal factors included in our framework to combine in complex patterns to produce either overconformity or underconformity. FsQCA is well-suited to accommodate such complexity because it allows for causal asymmetry between outcomes. Given the abductive nature of this research, we do not have a clear prior as to which configurations we expect to emerge. In standard regression analysis, we would thus need to model all possible interaction terms to cover all possibilities. Such a model would probably fail to compute, and the higher-order interaction coefficients would be challenging to interpret (Fiss, 2011; Campbell, Sirmon, and Schijven, 2016; Fainshmidt et al., 2020).
Second, it is possible that multiple solutions exist for the same outcome. This phenomenon, known as “equifinality,” is difficult to model in regression analysis. For instance, one configuration for a given outcome may require the presence of a given causal condition, while a second configuration for the same outcome may require the absence of the same causal condition. In regression analysis, these effects may cancel each other out, leading to the (incorrect) conclusion that said causal condition does not matter (Fainshmidt et al., 2020). FsQCA, in contrast, is designed to permit and identify equifinal solutions.
Third, since no underlying distributions are assumed and analysis is not correlational, the method is not vulnerable to outliers (Fiss, 2011; Vis, 2012) or endogeneity resulting from omitted variable bias (Ragin, 2008; Schneider and Wagemann, 2012). Outliers in fsQCA would either fail to meet the minimum numerical threshold for inclusion in fsQCA results (see below) or, if present in sufficient numbers, define an additional configuration with limited explanatory power (low “coverage,” i.e., low proportion explained of membership in the outcome). Omitting relevant variables would reduce the explanatory power of the analysis, as expressed in total coverage. However, it would not produce bias—in the sense of an included variable picking up part of the effect of a missing, correlated variable—because fsQCA is not a correlational method and does not estimate a coefficient for individual explanatory factors.
Calibration
As mentioned, causal conditions and outcomes in fsQCA assume set membership values from 0 (“fully absent”) to 1 (“fully present”). To achieve this, fsQCA requires a process called “calibration” (Ragin, 2008). Tables 4 and 5 provide a summary overview of our calibrations.
Calibration
Calibration Values by Country*
Values for organization-level measures are country averages. Organizational size is not shown because it was calibrated within countries and is not comparable across countries.
For ownership concentration and foreign shareholdings, we calibrated the data manually (Ragin, 2008; Fainshmidt et al., 2020) on the grounds that different shareholding levels establish qualitatively different levels of control rights in firms. Values assigned for these causal conditions were 1 for shareholding levels from 0.50, which imply full control of a firm; 0.67 for holdings from 0.25 up to but excluding 0.50, a level that is usually associated with extended control rights; 0.33 for levels from 0.10 up to but excluding 0.25, which are usually associated with at least a board seat; and 0 for all holdings below that level.
For our remaining measures, we applied the “direct” calibration method (Ragin, 2008; Greckhamer, 2016; Greckhamer et al., 2018). This approach requires specifying three anchors for each condition: a threshold below which observations are “fully out,” a crossover point of “maximum ambiguity” (neither in nor out), and a threshold above which observations are “fully in.” Calibration then involves using a logistic function to assign values based on these anchors (Ragin, 2008). In the absence of established knowledge of what specific values may represent these three anchors for our measures, we followed Greckhamer (2016) and used the 10th, 50th, and 90th percentiles.
We calibrated across the entire sample for board independence (as we are interested in deviations from global norms). As a result of the calibration process, board independence levels below 47 percent are regarded as underconforming and levels above 67 percent as overconforming. Levels in between these thresholds are effectively “neither in nor out” and thus neither underconforming nor overconforming. 2 In contrast, we calibrated revenues within countries, as we expected the impact of size to be context dependent. For example, the largest bank of a given country will always be a major actor in that country’s financial system even if it is small by international standards. Had we calibrated revenues across the entire sample, information about this relative importance would have been lost.
Model specifications
FsQCA was designed for cross-sectional analysis. When applied to panel data such as ours, fsQCA may calculate inaccurate consistency and coverage scores (Garcia-Castro and Ariño, 2016). One remedy would have been to focus on any one of the four years in our sample, but doing so would have meant discarding valuable data. Instead, we conducted five separate analyses: one for each of the four years in our sample and one with data pooled across all four years (see Aversa, Furnari, and Haefliger, 2015; Fainshmidt, Smith, and Guldiken, 2017). For our final results, we retained as viable only configurations present in the pooled analysis and in at least three of the four annual analyses. Our rationale is that the absence of a configuration from the pooled analysis or more than one annual analysis reduces the likelihood of that configuration being more broadly valid rather than an analytical anomaly. The results of the yearly runs are reproduced in the Online Appendix (http://journals.sagepub.com/doi/suppl/10.1177/00018392211022726).
Before conducting our main configurational analysis reported below, we explored whether any one condition was necessary, that is, whether it had to be present or absent for a given outcome to occur. To this end, we performed necessity analyses of all conditions and their negations, looking for high levels of coverage paired with a consistency threshold of 0.9 or higher (Ragin, 2008; Schneider and Wagemann, 2012; Greckhamer et al., 2018), with consistency expressing how reliably a given solution is associated with the outcome. We found one condition meeting the consistency threshold: the absence of foreign ownership had a consistency level of 0.96 for overconformity. However, with a coverage of 0.55, this condition is unlikely to be necessary for overconformity (Schneider and Wagemann, 2012).
FsQCA permits researchers to integrate their priors about the relationship between a given causal condition and the outcome in the form of “simplifying assumptions.” Given our exploratory empirical approach and the absence of strong priors for the conditions in our framework, we did not specify such simplifying assumptions for our analyses.
FsQCA calculates a so-called truth table of all 2 k possible combinations of the causal conditions, where k is the number of causal conditions. Normally, only some of these combinations are observed in reality, and the final analysis usually includes only those accounting for a meaningful number of cases (Ragin, 2008). While the literature offers no clear definition of “meaningful,” we followed recommended best practice by setting our thresholds to exclude rare configurations while retaining at least 80 percent of the original sample (Greckhamer et al., 2018). The resultant thresholds are two cases for our year-by-year analyses and six cases for our pooled analysis.
To be retained, solutions also had to meet a double threshold for two consistency measures. The generally accepted minimum threshold for a solution to be accepted as reliable is a consistency of 0.75 to 0.80 (Ragin, 2008; Fiss, 2011; Schneider and Wagemann, 2012). We adopted the more stringent threshold of 0.80. To reduce the risk of simultaneous subsets—that is, the emergence of solutions that explain both the presence and absence of a given outcome—we further followed best practice by requiring that each solution have a proportional reduction of inconsistency (PRI) score of 0.70 at minimum to be retained (Greckhamer, 2016; Greckhamer et al., 2018), thus ensuring that the configurations associated with a given outcome were not simultaneously associated with the negation of the outcome as well.
Results
Our analysis yielded three viable configurations for overconformity and seven viable configurations for underconformity (see Tables 6 and 7). As discussed, we consider a configuration “viable” if it resulted from both the analysis pooling of all four years of data and at least three of the four separate annual analyses. While our tables report all solutions for the sake of transparency, our discussion will focus on configurations that have been identified in at least four of the five total analyses, as indicated by the numbers in the last rows in Tables 6 and 7. 3 We further follow standard practice by indicating “core conditions”—i.e., conditions that are included in the “parsimonious solution” of fsQCA, suggesting a relatively strong connection with the outcome—with larger symbols. Smaller symbols indicate “peripheral conditions” with weaker connections to the outcome. 4
Configurations for Overconformity*
● = condition present; ⊗ = condition absent; blank = condition either present or absent. Large symbols = core conditions. Coverage and consistency statistics are based on four-year pooled analysis. Configuration shown in gray is present in fewer than four annual analyses and thus not discussed in the text.
Configurations for Underconformity*
● = condition present; ⊗ = condition absent; blank = condition either present or absent. Large symbols = core conditions. Coverage and consistency statistics are based on four-year pooled analysis. Configuration shown in gray is present in fewer than four annual analyses and thus not discussed in the text.
We use the standard symbols for reporting fsQCA results (Fiss, 2011): a solid dot (•) indicates that the configuration requires the presence of the given condition; a crossed-out circle (⊗), that the configuration requires its absence; and a blank space, that the configuration requires neither its presence nor its absence (“don’t care”).
FsQCA expresses the quality of a configuration and of the overall solution containing all configurations for an outcome in terms of two metrics: coverage and consistency. Coverage refers to the proportion of membership in the outcome explained by individual configurations or, at the level of the solution, all configurations jointly. Consistency expresses how reliably the combination of causal conditions in a given configuration, or the combination of configurations in a solution, is associated with the outcome. Our solutions perform well for both dimensions. The solution for overconformity has a total coverage of 0.55 and a consistency of 0.94—that is, the configurations described by the solution account for 55 percent of instances of overconformity, and 94 percent of the individual cases in the configurations contained in the solution exhibit overconformity. The solution for underconformity has a total coverage of 0.54 and a consistency of 0.90.
FsQCA results commonly identify multiple equifinal configurations of causal conditions sufficient for a given outcome and often exhibit asymmetry between configurations sufficient for the presence of an outcome and those sufficient for the absence of the same outcome (Ragin, 2008; Fiss, 2011; Schneider and Wagemann, 2012). In practical terms, this means that a given causal condition may be present in one configuration but absent in another configuration producing the same outcome. Similarly, the presence (or absence) of a given causal condition may be part of configurations sufficient for both the presence and absence of a given outcome. These configurational outputs are not inconsistencies in the analysis. Rather, they stem from the fact that fsQCA permits conjunctural causation whereby all causal conditions of a configuration are jointly sufficient to produce a given outcome, rather than identifying the average effect of each condition.
Configurational Paths to Nonconformity
Turning to the characteristics of individual configurations, we find that all overconformity configurations shown in Table 6 share three conditions: the absence of ownership concentration and of foreign shareholding at the organizational level and the presence of managerial discretion at the national level. 5 To generate overconformity, these three common conditions can combine with the absence of duality and labor rights (O1), the absence of duality and the presence of shareholder rights (O2a), or the presence of shareholder rights and the absence of labor rights (O2b).
Configurations for underconformity shown in Table 7, in contrast, do not share any single across-the-board condition. Configuration U1a requires the absence of virtually all conditions except concentration (“don’t care”) and the presence of labor rights. Configuration U1b is very similar but requires the absence of concentration instead of duality. Configuration U1c combines the absence of size, foreign shareholdings, duality, and managerial discretion with the presence of concentration and shareholder rights. Configurations U2a and U2b share the presence of large size and the absence of duality, managerial discretion, and labor rights. U2a additionally requires the presence of concentration and the absence of shareholder rights. Conversely, U2b requires the absence of concentration and foreign shareholdings as well as the presence of shareholder rights. Configurations U3 and U4 are relatively rare, as indicated by their low coverages. They share the absence of size and the presence of concentration, managerial discretion, and labor rights. U3 additionally requires the absence of foreign shareholdings and shareholder rights and the presence of duality, while U4 requires the presence of foreign shareholdings and shareholder rights but the absence of duality.
We confirmed the robustness of these results in five tests, as reported in Online Appendix B. First, we re-ran the pooled analysis with all small U.S. banks included, which increased our sample size to 1,787 observations across 18 countries and four years (i.e., an average of 447 banks for each year), with 74 percent of the sample representing U.S. data points. The results were effectively identical. The only substantive change involved Configuration O2b, in which concentration changed from absent to “don’t care.”
Second, we modified the calibration anchors for all conditions calibrated using the direct method to the 20th, 50th, and 80th percentiles and re-ran the pooled analysis for our main sample. The results were again very similar. We obtained the same results for all configurations for overconformity, plus one additional configuration not previously identified. For underconformity, six of the eight original configurations were replicated. Configuration U1b was missing. This robustness test also yielded an additional configuration for underconformity not previously identified.
Third, we changed the calibration method for the two ownership conditions from manual to the direct method with anchors at the 10th, 50th, and 90th percentiles and re-ran the pooled analysis for our main sample. These results were identical to our main analysis.
We further tested the sensitivity of our results by increasing the numerical and consistency thresholds for inclusion. For overconformity, increasing the threshold from at least six to at least seven cases left about 78 percent of the original sample. Configuration O3, which we do not treat as viable because it appeared in only three of the five analyses, disappeared, as expected given its low unique coverage. Setting a higher consistency threshold of 0.90 did not affect the results for overconformity. For underconformity, configurations U1b and U4, with the lowest unique coverages, disappeared with the increased numerical threshold. Increasing the consistency threshold left configurations U1a through U4 unchanged. In addition, the original version of U1c was joined by a logically consistent variant that inverts the presence of shareholder rights and “don’t care” for labor rights.
In sum, extensive robustness tests yielded results that are logically consistent with our main results.
Discussion
Our results suggest much greater heterogeneity among drivers of underconformity to global norms than among drivers of overconformity. While our results for overconformity and underconformity have approximately the same explanatory power as indicated by comparable coverage scores, this coverage is distributed across three viable configurations for overconformity but seven viable configurations for underconformity. This suggests that the drivers of underconformity are likely to be more multifaceted than those for overconformity.
Our results further indicate the presence of multiple equifinal configurations for over- and underconformity. This finding represents an important theoretical insight as it suggests that different combinations of organizational and institutional factors can yield different pathways toward the same nonconformity outcomes. In addition, our results for overconformity and underconformity are asymmetrical, that is, the configurations driving overconformity are not the inverse of those driving underconformity. Such asymmetry is common in fsQCA-based research (Ragin, 2008; Fainshmidt et al., 2020). In our context, it suggests the presence of distinct mechanisms for underconformity versus overconformity, a possibility that recent work has surmised (Aguilera, Judge, and Terjesen, 2018) but that has remained underexplored in the literature.
We now turn our attention to generating insights into how and why organizational and institutional factors combine to bring about nonconformity. Following Furnari et al. (2020), this iterative theorizing process entails specifying the jointly causative conditions as well as articulating the holistic, configurational themes in which conditions work together, with the latter including a process of attaching meaningful labels to the different configurations.
Configurational Patterns of Nonconformity
Table 8 summarizes seven emerging patterns evident in the viable configurations detailed in Tables 6 and 7: one pattern for overconformity and six patterns for underconformity. As mentioned, “viable” refers to configurations that were present in at least four of the five separate analyses we conducted.
Over- and Underconformity Patterns
Overconformity patterns
Configurations O1, O2a, and O2b are logically consistent with one another and can be combined into a single, more stringent pattern. 6 We summarize the result of this combination as Pattern 1 in Table 8. This pattern is institutionally associated with Anglo-Saxon liberal market economies (Hall and Soskice, 2001). Conceptually, it is consistent with prescriptions derived from agency theory, which include high levels of board independence, as well as the absence of both CEO duality and influential blockholders. We label this pattern Shareholder–Management Coalition. Exemplars include Citibank, Wells Fargo, National Australian Bank, Barclays, and Royal Bank of Canada.
Viewed through our political lens and consistent with prior corporate governance research, this pattern suggests that overconformity is likely to occur when shareholders and management are institutionally empowered at the national level and neither labor nor blockholders or foreign shareholders stand in their way. Dispersed shareholders are likely to support high levels of board independence as a tool that is perceived to enable higher levels of monitoring, which might also explain the absence of CEO duality (Setia-Atmaja, 2009). In principle, management might be expected to resist the monitoring resulting from higher board independence. However, management may support high independence levels as a signal of good corporate governance, enabling it to leverage independent directors as providers of valuable resources (Pfeffer, 1972) and to reduce board monitoring effectiveness through information asymmetries (Zorn et al., 2017). The latter is a relatively new insight that probably had not yet become salient in shareholders’ minds in the years covered by our sample. Dispersed shareholders and management may thus have an interest in securing the same outcome, overconformity, though possibly with different underlying motivations. Overall, in national contexts with high levels of shareholder rights and managerial discretion but weak labor rights—typical of Anglo-Saxon liberal market economies—firms with dispersed ownership, few foreign owners, and no CEO duality exhibit overconformity to global board independence norms.
Underconformity configurations
Table 8 shows six patterns for underconformity. Pattern 2 suggests Labor Control. Banks in this category have dispersed domestic ownership. They are small, have no CEO duality, and operate in institutional contexts with strong labor rights but weak shareholder rights and low managerial discretion. Smaller German, Italian, and Japanese banks, such as Comdirect Bank, Credito Emiliano, and Aozora Bank, are typical of this pattern. With labor representing the most powerful actor here, it seems plausible that this predominance may have allowed labor to capture the board through inside directors to gain control of the organization and its resources (see Aguilera and Jackson, 2003, 2010).
Patterns 3.1 and 3.2 represent variations of Blockholder Capture of the board. In Pattern 3.1, banks are small with concentrated domestic ownership and no CEO duality. They exist in institutional environments that empower shareholders but disenfranchise management. Exemplars are relatively small French, Israeli, Spanish, and Turkish banks, such as Crédit Agricole de Ile-de-France, Union Bank (Israel), VakıfBank (Turkey), and Liberbank (Spain). In this pattern, blockholders seem to be able to control the board and keep independence levels low, thus hindering outsider monitoring, even though the institutional context empowers minority shareholders (see Aguilera et al., 2015). Blockholders may be able to do so because banks in this pattern are small, which probably reduces the chances of determined shareholder pressure to increase independence levels.
Pattern 3.2 represents blockholder capture of the board in large banks. These banks have no CEO duality that might empower management, and the national institutional context disenfranchises shareholders, management, and labor alike. Exemplars include large Greek and Polish banks, such as National Bank of Greece and Alpha Bank. Even though these organizations are large and thus more readily subject to institutional conformity pressures, in the absence of any countervailing power, predominance may enable blockholders to gain control over the board through inside directors (see Aguilera et al., 2015).
Pattern 4, by contrast, seems to be indicative of Shareholder Indifference. It comprises large banks with dispersed domestic ownership and no CEO duality, operating in environments that empower shareholders and disenfranchise management and labor. Large Israeli banks such as Bank Hapoalim and Bank Leumi are exemplars. This pattern is counterintuitive in that conventional wisdom based on agency theory suggests that dispersed shareholders should have an interest in high levels of independence so as to improve monitoring. In this pattern, no other actor is empowered, which suggests that shareholders should, in principle, have the power to attain higher levels of board independence. Since the banks included in this pattern are large, it is unlikely that lack of awareness deters shareholder pressure for more independence. Instead, with the presence of high levels of shareholder power and the absence of any countervailing power especially on the part of management or blockholders, one possible interpretation is that dispersed shareholders may consider their interests in these banks secured and may thus not see a need to increase monitoring.
Intriguingly, in patterns 2, 3.2, and 4, underconformity is associated with the presence of a single type of actor with power—labor, blockholders, or dispersed shareholders—while all other actors are relatively weak. The crux of the former two patterns seems to be that labor or blockholders, in effect, control boards and keep them staffed with insider directors, as prior conceptual research has implied (Aguilera and Jackson, 2003, 2010; Aguilera et al., 2015). Dispersed shareholders seem to acquiesce to insider boards, possibly resting secure in the knowledge of institutionally derived shareholder power relative to all the other actors. Taken together, this suggests that in the presence of a single actor with power and the simultaneous absence of other powerful actors, firms will tend to exhibit underconformity to global board independence norms. Pattern 3.1 shows that size may serve as a contingency in this relationship: Even when dispersed shareholders are empowered, they seem to lose out to the interests of blockholders in terms of insider control when those banks are small. Small organizational size and, presumably, attendant reduction in conformity pressures resulting from less attention from the investor community may thus enable powerful actors, such as strong blockholders, to defend insider boards against the interests of dispersed shareholders.
Pattern 5 represents a rarer configuration of Domestic Insider Control. It encompasses small banks with domestic concentrated ownership and CEO duality, operating in environments with strong management and labor and weak shareholders. Typical examples are small Swiss banks such as Glarner Kantonalbank and Bank Linth. Blockholders and labor are likely to favor insider boards, so underconformity here might represent the confluence of blockholder and labor interests. A comparison with Configuration O3 suggests that blockholders are likely to be crucial: O3 effectively represents the same configuration of conditions, albeit with two changes: the presence of size and the absence of concentration. It thus seems likely that similar to the mechanisms discussed earlier, small firms’ ability to eschew conformity pressures may enable the formation of an insider board through a de facto coalition, be it formal or informal, of blockholders and labor.
Pattern 6 is rare and unusual, as evident in its low coverage and narrow capture of cases/contexts, so we do not think it generalizes beyond the specific cases in which it appears. Thus we do not label or discuss it here, though we provide details about this pattern in Online Appendix C.
Advancing Comparative Corporate Governance and Nonconformity Research
Our study offers novel insights on comparative corporate governance and the workings of nonconformity in the sociology of globalization. It suggests that understanding nonconformity in general and international corporate governance in particular benefits from a multilevel theoretical lens that accommodates the ways in which firms are interdependent with their institutional context and are guided by political forces. Nonconformity is not the same all over; it can take different forms (overconformity versus underconformity), and there are distinct patterns across countries in which firms exhibit nonconformity. While much of the corporate governance literature focuses on shareholders and management, our findings provide a strong reminder that in many parts of the world, labor has considerable influence on corporate governance practices. Research on nonconformity, whether in corporate governance or otherwise, that does not account for institutional variety and the resulting power dynamics within firms is thus likely to be incomplete.
The empirical patterns we uncover suggest that overconformity is more likely shaped by powerful shareholders and management, and underconformity is more likely shaped by dominant blockholders and labor. This overarching pattern also implies that firm owners can support either overconformity or underconformity depending on whether their power emanates from institutions or from equity holdings and depending on the power of other key actors with whom owners might form coalitions. In contexts in which dispersed shareholders and management are powerful, overconformity might serve to assuage concerns about board impartiality and empower managers through reduced oversight and enhanced access to resources. By contrast, in contexts in which labor holds bargaining power and blockholders are present, there might be mutual interest in appointing trusted insiders to reinforce owner control and prevent conflict with labor. These emerging patterns highlight the complex, multilevel, and asymmetric nature of differences in corporate governance practices among firms within and across national contexts, opening promising opportunities to advance corporate governance theory.
The complexity of these findings further highlights the gestalt-like nature of nonconformity. We show that firms have latitude to under- or overconform and that the specific manifestation of that latitude depends on the configuration of relative power among key actors shaped by firm-specific conditions and national institutions. Our results reveal that the same condition can lead to conformity, underconformity, or overconformity depending on its combination with other conditions, which helps remedy some equivocal associations in prior research, such as the one related to firm size (Greenwood et al., 2011). Our study shows that both larger and smaller firms can exhibit nonconformity, and whether they overconform or underconform is contingent upon interrelated conditions surrounding political power dynamics within the firm. While causal complexity is more challenging to study and theorize, it offers a more holistic account of why some firms engage in nonconformity. Advancing nonconformity research warrants a more serious accommodation of the complex nature of organizational phenomena.
We also find that the relationship between power and corporate governance outcomes may be more complex than the literature has considered. Our findings suggest a need to examine not only whether actors have power but also whether they are willing or able to leverage it to attain their interests. This willingness is usually assumed, but the Shareholder Indifference pattern we identified suggests that, at times, this may not be the case. As proposed in our discussion, the presence of functional substitutes serving the same interests in different ways—such as strong legal protection paired with the weakness of other actors making strong board monitoring less necessary—may represent a key contingency. This further highlights the importance of national context in how key actors leverage their power within firms.
In addition, our findings underline the configurational nature of power, a conceptual dimension we believe is difficult to capture in correlational research that examines independent variables whose effects can be established in isolation. All of the configurations we have identified simultaneously entail the presence or absence of power in other actors, which is consistent with the notion that power is an inherently relative concept defined as a relation among actors (Dahl, 1957). In this way, the power of one or even a few actors can lead to divergent outcomes, such as overconformity versus underconformity, depending on the overall constellation of powerful actors. Studying power dynamics in corporate governance and in management more generally thus seems to be a natural field for configurational research.
While the theoretical focus of our study is on the various configurations of different conditions, our findings also shed light on the relationships between individual explanatory factors and nonconformity. Some of these relationships are relatively consistent across configurations. For instance, ownership concentration tends to lead to underconformity, while foreign shareholdings and CEO duality tend to associate with conformity to the global norm. Consistent with expectations in the literature, smaller firms seem to find it easier not to conform to global norms, yet our findings also highlight the conditions under which larger organizations may do so. We uncover a consistent relationship between managerial discretion and over- or underconformity, in that all configurations for overconformity require the presence of managerial discretion, while all configurations for underconformity (with two rare exceptions) include its absence. In contrast, the relationships between shareholders and labor rights and nonconformity are complex, governed by sets of interrelationships. This suggests that the labor–shareholder balance of power and how it is reflected in board composition is sensitive to organizational conditions and thus to complex causation.
Limitations and Future Research Directions
Our study has several limitations, some of which also open up avenues for future research. First, we do not know when each firm decided to engage in nonconformity, so inferences of causality should be made with caution. While simultaneity or reverse causality does not cause bias in fsQCA, it somewhat complicates the interpretation of our results. Similarly, governance practices like board independence tend to be sticky over time; hence, examining how levels of board independence and power change over time to inform nonconformity would be worthwhile.
Second, our study focused on OECD countries, all of which are advanced industrialized countries with relatively well-functioning institutional and legal contexts when benchmarked against the rest of the world. It would be instructive to extend this study to emerging markets and study nonconformity under weaker formal institutions more prevalent in many other countries. The importance of extending this study to a larger variety of institutional contexts is further highlighted by the fact that local and regional differences matter in conformity and social comparisons (Marquis and Battilana, 2009). Our study examines nonconformity to global norms, but it is not clear whether conforming or even nonconforming firms in our study abide by local norms. Explicitly studying the interplay between global and local norms would be valuable, particularly in terms of how this interplay shapes corporate governance nonconformity.
Third, we are interested in one key aspect of corporate governance: board independence. But nonconformity in other dimensions, such as gender representation, might be driven by partially different conditions and configurations. It would be meaningful to extend this study to other core corporate governance practices. Similarly, while our framework of conditions is anchored in prior research and theory, there may be other potentially relevant factors, such as top management team composition. While fsQCA is robust against omitted variable bias, exploring other factors that may be relevant to corporate governance nonconformity could yield important additional insights.
Fourth, while we define and measure board independence in line with prior research, other dimensions exist that co-determine to what extent a board is truly independent, in the meta-conceptual sense of executing its tasks unfettered by social or material strictures. For instance, we are not able to capture the role of social ties in the independence measure. There is potential for deception here, as formally “independent” directors may have hidden social ties with management. Similarly, the presence or absence of CEO duality may affect independence as a meta construct. Although we examine the formal state of board independence rather than independence as a meta construct or board effectiveness, we acknowledge this potential limitation as an opportunity for future research.
Fifth, our results may not generalize to other industries beyond banking. We deliberately selected an industry that is highly regulated and global. As a result, our test is conservative because these banks have comparatively less discretion to engage in nonconformity. Nonconformity in banking is thus likely to require relatively strong drivers, which in principle would support the notion that those same drivers could apply in other industries as well. Whether this is indeed the case is a promising avenue for further empirical study.
Finally, our interpretation of the configurational patterns is based on substantive knowledge of the literature as well as case data, but as with any abductive effort, some interpretations remain exploratory, particularly when the same condition can be interpreted in different ways. For example, we suggest that different sources of power are sometimes substitutes and at other times complements. These apparent inconsistencies reflect a broader need for research that disambiguates multiple plausible explanations of key corporate governance phenomena. For instance, our interpretations point to important future research opportunities regarding the underlying intentions of powerful actors, which corporate governance research has tended to assume rather than examine directly.
Supplemental Material
sj-pdf-1-asq-10.1177_00018392211022726 – Supplemental material for Our Board, Our Rules: Nonconformity to Global Corporate Governance Norms
Supplemental material, sj-pdf-1-asq-10.1177_00018392211022726 for Our Board, Our Rules: Nonconformity to Global Corporate Governance Norms by Michael A. Witt, Stav Fainshmidt and Ruth V. Aguilera in Administrative Science Quarterly
Footnotes
Acknowledgements
We thank Christopher Marquis for his constructive editorial guidance and the three anonymous reviewers for their insightful feedback. We are grateful for helpful input received at AIB, ICGS, and SMS conferences and at seminars at the Program on US-Japan Relations at Harvard University and the Center for Emerging Markets at D’Amore-McKim School of Business, Northeastern University. We further thank the International Corporate Governance Society for honoring a previous version of this paper with its best paper award. Michael Witt also thanks the Program on US-Japan Relations and its then-director, Susan J. Pharr, for hosting him at Harvard University during the formative period of this project.
Supplemental Material
Notes
Authors' Biographies
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
