Although the relation between the nature of P–A relations and the development of and investment in governance mechanisms such as board monitoring or incentive alignment has been extensively discussed in the agency literature (for a review, see Gomez-Mejia and Wiseman, 1997), the contextual antecedents to P–A relations have received inconsistent attention. Yet it would seem important to understand how these contextual antecedents influence the specific manifestations of agency problems as a way of enhancing our explanations for why governance structures appear to vary across social contexts (Aguilera and Jackson, 2003, 2010; Fligstein and Freeland, 1995). One model that shows promise in leading towards a more socialized agency theory is provided by Doney et al. (1998, p. 601), who propose linking cultural dimensions (e.g. power distance, individualism/collectivism, and risk orientation) to the development of interpersonal trust. Hence, improving our understanding of how P–A relations are embedded in a particular social context should enhance the explanatory power of agency theory outside of the traditional contexts in which it has been applied.
As a first step towards creating a social theory of agency, we begin by accepting the core insight of sociologists that economic behaviour is shaped by social mechanisms not just at the margin, but also at the core. That is, social mechanisms influence the conventions defining the various roles we inhabit (including that of agent and principal), as well as how we interact with one another. These mechanisms entered the lexicon, according to Dobbin (2004), under the terms institution, cognition, network, and power (see Figure 1). Within each of these mechanisms we can identify a variety of aspects of society that appear to play a role in how agents and principals see themselves, their interests, the nature of the problems that are likely to arise from their interactions, as well as the mechanisms devised to control these problems. Unlike sociologists, however, who have approached the interplay of society and economics from an inductive perspective, we adopt a deductive approach more common to economists. We favour the deductive approach because it provides the best opportunity for building a generalizable theory of governance that is not simply a reflection of characteristics idiosyncratic to one social–political context. Thus, we argue for the development of a social theory of agency as a first step towards understanding how P–A relations may vary across social contexts, which in turn can vary both across and within national boundaries.
The basic idea summarized in Figure 1 is that as long as delegation is involved, agency problems (information asymmetry, conflict of interests, and opportunistic agent behaviour) are universal, even though the explicit manifestation of these problems and ways to deal with these problems may vary depending on institutional context. Thus, for instance, information asymmetry in a highly centralized economic system (such as the old Soviet Union) may mean that plant managers would produce low quality goods in order to meet the established quota for their respective plants. Conflict of interests in a society ruled by a religious minority may translate into agents making choices for their units that differ from established dogma. And opportunistic agent behaviour may not involve personal financial gains at all, but rather the desire to acquire more power or to pursue a particularistic agenda (such as investing in social projects because of moral convictions that this is the right thing to do). Given that most of the agency literature has emanated from the United States, these agency problems have generally been portrayed in terms of CEO entrenchment, excessive CEO pay, risk minimization by top management and the like when executives enjoy substantial discretion (Gomez-Mejia et al., 2011a). But clearly issues of information asymmetry, conflict of interests, and opportunistic agent behaviours are universal concerns in any delegation situation, even though their specific manifestations may take a multitude of forms. The same can be said for the control systems to deal with these agency problems. Thus, for instance, incentive alignment in some contexts may not involve financial rewards at all but rather the provision of such non-pecuniary rewards to agents as greater political visibility, the ability to place relatives and friends in key positions, and affect-related utilities (or ‘socioemotional wealth’; Berrone et al., 2010). We now turn our attention to the role of institutional environments in shaping the nature of agency problems and control systems to deal with them.
Institutions (Institutional Environment)
For sociologists, institutions reflect the conventions of a society, often defined by law (Dobbin, 2004). Thus, government represents an institution that prescribes certain practices while circumscribing others. Recognizing this, we suggest a theory of government that could be used to explain differences in P–A relations by focusing on government's role in promoting and controlling economic exchanges. Clearly economic and political institutions are inextricably intertwined (e.g. North, 1982), although from the perspective of theory development it may be important to disentangle them in order to derive more precise and generalizable predictions about how political and economic institutions may individually and separately influence the nature of agency relations. Although several studies on agency relations have considered the role of political institutions, few have recognized that political institutions play distinguishable and to some extent opposing roles in an economy. On the one hand, political institutions promote economic exchanges by supporting an infrastructure of intermediation that increases the transparency of economic transactions and reduces the risk to transacting partners. Political institutions support intermediation and thus transparency by (a) providing the third party enforcement of contracts through an impartial judicial system, (b) collecting and disseminating financial information about transacting parties, and (c) setting and enforcing standards on independent institutional players providing and sanctifying such information including auditors, stock analysts, brokers, and so forth. The resulting transparency reduces information asymmetry between transacting partners, thus lowering transaction costs of economic exchanges and making them more likely to occur. For example, Marshall and Weetman (2002) find that the regulation of financial reporting reduces information asymmetry and thus the monitoring burden born by principals, because in this fishbowl-like environment agents are less tempted to act opportunistically lest they damage their reputation and thus their career value. Building on this argument, we speculate that increased transparency should foster more stewardship behaviours among agents. In turn principals, sensing lower risk of agent opportunism, may provide the agent with greater freedom of action and larger rewards for effort. Further, agents may respond to the relaxation of governance with greater commitment to fulfilling their responsibilities (Davis et al., 1997). That is, transparency can engender reciprocity between principals and agents, thus reducing the risk of moral hazard (cf. Irlenbusch and Sliwka, 2005).
On the other hand, political institutions simultaneously regulate economic exchanges by circumscribing and even preventing certain types of economic transactions from occurring. That is, political institutions also limit exchanges deemed socially undesirable (e.g. insider trading, loan sharking, tontines). This form of political intervention constrains managerial autonomy; the agent is faced with a powerful stakeholder having a political/social agenda that may hamper the agent's ability to satisfy other stakeholders, including the principal (Finkelstein and Boyd, 1998). These constraints could come not only by government regulation but also by government involvement in the corporate governance of individual firms through ownership and board ties (Okhmatovskiy, 2010). In looking at Italian utilities, Rienzner and Testa (2003) found that rapidly changing government regulation and policy severely burdened managerial decision making. Similarly, Finkelstein and Hambrick (1990) and Hambrick and Finkelstein (1987) suggest that in highly regulated industries (which are more subject to political meddling), managerial discretion is lower than in less regulated industries. For example, banking regulation in the USA long limited savings and loan financial institutions to home mortgages and pass-book deposits, and prevented retail banks from engaging in investment banking operations. Following the removal of these regulations, agency costs rose considerably in the Savings and Loan industry during the 1980s (Kane, 1989; White, 1991; Wiseman and Catanach, 1997) and in the retail and investment banking industry during the early 2000s (Demyanyk and Hemert, 2008). During a highly regulated environment, lower managerial discretion translates into less use of variable compensation, with executives receiving fixed salaries that are not linked to profitability. These executives can be easily dismissed, for actual or alleged incompetence, in response to pressures from regulatory commissions. In a study of the US thrift industry, Wiseman and Catanach (1997) found that principals relied on government regulatory authorities to monitor agents, and that deregulation left a void in agent monitoring that principals often failed to fill.
Beginning with a theory of government's role in private economic exchanges, one can then develop generalizable models of agency that examine how the different forms of political intervention vary and ultimately influence economic exchanges such as those between principals and agents (for an exception, see Bushman et al., 2004). It is our view that taking this deductive approach allows for the development of finer grained and more generalizable models of agency than can be derived from comparative inductive approaches. That is, simply comparing the average governance structure under one government system (often simply characterized by its philosophical or historical foundation) with the average structure under a different government system obscures the subtle differences within each setting. Beginning with a theory of institutions should also open new directions for research by providing new questions about the sources of moral hazard and its likely resolution.
Cognition (Cognitive Framework)
Cognition refers to the psychological process by which societal members make sense of the world. Thus it captures the beliefs that society uses to understand, interpret, and even generate experiences (e.g. false memories). So, the belief that leadership matters (what Meindl et al., 1985, dubbed the romance of leadership and others have called a heroic view of leadership) implies an assertion that individual executives can provoke large differences in how organizations perform. That is, some cultures strongly believe that top executives can and do have a major impact. This is especially true of the United States (for examples, see Jacobson and House, 2001; Shamir et al., 1993). This heroic view of leaders may correspond to higher absolute pay. In this line, Khurana (2002) and Tosi et al. (2004) find that perceived charisma is often used by boards in the United States to gauge and reward CEO success.
Similarly, in cultures where pecuniary rewards are important, CEOs make social comparisons of their achievements by devoting close attention to how their pay stacks up against that of other CEOs (Crystal, 1991). There is substantial evidence in the United States and England supporting this view (e.g. Eriksson, 1999; Ezzamel and Watson, 2002; Henderson and Fredrickson, 2001). In other cultural contexts, metrics other than money (such as official titles, wide media exposure, honorary appointments to prestigious universities, participation in key national committees on industrial policy, being portrayed as the embodiment of a set of moral or ethical values, and the like) may be used to determine one's value. In many Latin and Asian societies, high executive compensation packages may be seen not as symbols of personal success but as signs of avarice and hidden corruption (see Ibanez, 2005). In France, high compensation differentials between top and lower organizational levels of the order often seen in the United States would be regarded as shameful (Cala, 2005). These social values may explain why CEO compensation in most other countries (even after controlling for firm size, performance, industry, and per capita income) is only a fraction of what is paid to CEOs in the United States and England (Gomez-Mejia and Wiseman, 1997).
While there are valid reasons for comparing beliefs and values at the national level, such as those regarding the importance of leadership, pecuniary rewards, and social capital, at the national level (e.g. Spence et al., 2003; Sporer, 2004; Thompson and Yurkutat, 1999), there are also reasons for looking more deeply and considering within-country differences in these dimensions (cf. Brockner, 2003). For example, Parsons (1951) found that people living on the west coast of the USA preferred affective oriented people, while those living on the east coast preferred people who are more neutral in their disposition towards others. Mueller and Clarke (1998) provide another example in suggesting that difficulties in adopting US-style incentive systems in Central and Eastern European countries are based not only on differences in national cultures (characterized as more collectivistic in Central and Eastern Europe vs. individualistic in the USA) but also on differences in the social context within the countries. Although managers in former socialist countries have been instilled with a spirit of cooperation and benevolence towards the enterprises for which they work, younger citizens, particularly university students, have developed a sense of entitlement that leads them to favour their own interests, even at the expense of others in an organization. Brodbeck et al. (2000) describe similar cultural differences in leadership across Europe. Liberman (2001) also found that individuals' role expectations in a multinational enterprise differ not only across European country-units (suggesting influences of local cultures and other societal forces) but also between managers and employees (independent of country-office).
Building on a theory of culture that recognizes within country differences can only improve the predictive validity of comparative governance models by reducing the measurement noise created when social beliefs are measured using country-wide averages without regard to the degree of variance in beliefs within each country. To do so, however, begins with a theory of culture that provides a means for examining the cultural orientation of individual agents and principals. Although cultural differences between nations may be significant when measured in the aggregate, within nation variance tends to be large enough to undermine generalizations that link specific P–A relations to national culture (e.g. Mueller and Clarke, 1998).
Other directions for the elaboration of agency theory are found in research examining the unionization of labour, property rights of capital, managerial ideology, and consumption versus production cultures, to name but a few (Aguilera and Jackson, 2003; Dobbin, 2004; Rubach and Sebora, 1998; Smart, 2003). All of these provide interesting avenues for examining the potential for moral hazard risk. Rigorously examining these influences can not only provide insights into how P–A relations may vary, but also may help answer questions about how to design contracts that minimize the costs of governance while encouraging greater cooperation between principals and agents to maximize their joint utility. That is, recognizing how moral hazard risk varies across contexts may provide an additional explanation for why governance practices appear to vary across these contexts. In pursuing these lines of research, we may find that governance design is as much a response to the nature of the moral hazard problem as it is a consequence of cultural and legal constraints on P–A contract design. Indeed an analysis of moral hazard as reflective of institutional arrangements outside the P–A relation opens a new path of investigation into the efficacy of governance design.
Networks are institutions from which we learn and shape our identity (Durkheim, 1978). Networks occur at multiple levels of analysis that may or may not span firm and industry boundaries. At the organizational level, firm performance can benefit from the network ties of organizational members through their access to information and resources. For example, interlocking board memberships provide access to knowledge from multiple sources that can aid in addressing organizational contingencies (Galaskiewicz and Wasserman, 1989), or specific experiences and beliefs about strategic or structural changes (Betthenhausen and Murninghan, 1985; Haunschild, 1993; Palmer et al., 1993; Westphal et al., 2001).
A network's density, measured as the number of ties between actors in a network (Coleman, 1988), may influence the nature and potential for moral hazard. Network density removes roadblocks to the flow of information, thus making information widely and easily available to all members. Also, it encourages greater cooperation among network members. Since interactions among members in dense networks are likely to be observed by all other members, dense networks encourage reciprocity (Coleman, 1990) and increase the costs of defection and opportunism. Okamura and Vonortas (2006) note differences in the networking behaviour across industrial sectors, such that knowledge networks predominate in pharmaceutical sectors and alliance networks predominate in computers, electronics, and instruments. In this context, Gopalakrishnan et al. (2008) also found that the extent of financial capital that biotech firms acquire upon alliance formation was positively related to the level of familiarity across the allying partners. Thus, the degree of social network density would seem to play a role in reducing information asymmetry (Fjeldstad and Sasson, 2010) as well as increasing social pressures on agents and principals to limit opportunistic behaviours (Nielsen and Nielsen, 2009). As a result, principals do not need to invest as much in incentive alignment systems and internal monitoring mechanisms, but instead can rely on less costly and more efficient mutual monitoring by network members.
Exploring the social networks surrounding principals and agents using theories of network density, structural holes, and the creation of social capital (Burt, 2004) may open new avenues for gauging the likelihood of moral hazard and the necessity of establishing governance mechanisms to control agency costs.
Power represents our ability to shape our world. Focusing on the power of principals, we can examine how differences in governance structures are related to ownership concentration, or the nature of the principal (cf. Dixit, 1997). Ownership concentration and the power of particular stakeholders grow in tandem, since those with high ownership interests may impose their goals on the firm and agents are likely to be rewarded and/or dismissed according to the achievement of those goals. Even if we were to assume that agents are uniformly risk averse, as many agency models do (for an exception, see Wiseman and Gomez-Mejia, 1998), executives would be disciplined by principals if they pursued low risk/low return strategies (Hill and Phan, 1991).
Furthermore, as ownership concentration increases, principals are less likely to share residuals with agents, since such payments would occur largely at the expense of particular shareholders. In accord with these arguments, studies by Tosi and colleagues (Gomez-Mejia et al., 1987; Tosi and Gomez-Mejia, 1989, 1994; Tosi et al., 1999, 2004; Werner and Tosi, 1995) as well as others (David et al., 1998; Hambrick and Finkelstein, 1995; Khan et al., 2005; McEachern, 1975) find that high ownership concentration tends to be associated with lower agent compensation, a closer linkage between agent pay and firm performance, and more vigilant monitoring of the agent's decisions. In other words, as ownership concentration increases, incentive alignment and monitoring serve as complementary (rather than alternative) mechanisms of agent control (Tosi and Gomez-Mejia, 1989).
Regarding the nature of principals, Dixit (1997) pointed out the problem of heterogeneous principals, arguing that in many countries managers are supposed to be responsible not merely to shareholders, but to a more varied collection of stakeholders (such as workers, creditors, the local community, and so forth), provoking a politicization of corporate governance. In Germany, for example, labour unions and bankers have been granted formal authority over management that rivals the control by shareholders (Bruce et al., 2005). In many Arab countries, religious figures often sit on bank boards to ensure compliance with Muslim law, which proscribes certain lending practices (Abdulrahman et al., 2002; Rice, 2003). Although this situation may have its roots in legal requirements for governance, there is reason to believe that the presence and involvement of multiple principals with divergent interests should not be restricted to countries that require this involvement. Instead, we should look individually at firm practices and how different ownership structures influence governance designs. Accordingly, Young et al. (2008) suggest that although each emerging economy has a corporate governance system and creates informal mechanisms to deal with the specific ‘weak governance’ environment, all of them are characterized by conflicts between the controlling and minority shareholders in the firm. Known as the ‘principal–principal model’, recognizing the potential conflict among principals requires remedies different from those that assume unified interests among principals where shareholder wealth maximization is the most important, if not the only relevant, objective. Examples of firms that have involved non-traditional stakeholders in their internal governance structures (e.g. the Milwaukee Journal, Chrysler, Eastern Airlines, and Pan American each had union members sitting on their boards at one time) can be found in the USA as well, though this practice runs contrary to typical governance practice in the USA. Even within what are often considered a uniform investment group such as block-holders, the US government's Securities and Exchange Commission distinguishes between active (SEC code 13D) and passive (13G) investors.
This brief review of the literature on the role of principals suggests two things about governance mechanisms when the primacy of a single stakeholder is in question. When agents must satisfy multiple and diverse, often conflicting demands, performance criteria will be more ambiguous. In addition, the provision of managerial incentives and the design of a control structure require consideration of the utilities of all stakeholders, not only those who have ownership rights over the firm (cf. Tirole, 2001). Hence we would expect less use of contingent or variable pay, since it is difficult to find a common performance denominator that meets the need of all principals. Second, given that the reporting relation is not clear in the presence of several different principals, we suspect that the monitoring process is likely to be politicized.
This would encourage a more internal monitoring process using supervisory bodies that reflect the composition of external stakeholders. Under such circumstances, a key task for the agent will be to engineer compromises among principals, and the agent's evaluation is likely to hinge on convincing various parties that his/her choices best optimize the interests of all. Thus it would be advisable to develop a theory of principals, possibly built on a stakeholder view (Donaldson and Preston, 1995) in order to capture the nature of principal involvement at the firm level if we are to improve the explanatory power of models of P–A relations (e.g. Tribo et al., 2007).
In addition, family ownership itself can be further defined to reflect the dispersion of ownership among family members, and this dispersion has been argued to influence the nature of governance and risk of moral hazard (e.g. Miller and Le Breton-Miller, 2006; Schulze et al., 2003). Given that family ownership and influence over the firm's affairs varies widely (La Porta et al., 1999), it seems reasonable that the nature of the agency contracts and forms of moral hazard should vary accordingly. Several studies have pursued this line of thought and have provided several new directions for understanding the role of family ownership on moral hazard and governance design. For example, Gomez-Mejia and colleagues (Cruz et al., 2010; Gomez-Mejia et al., 2001, 2003, 2007) found that concentrated family ownership is associated with higher dismissal of non-family executives when performance drops, greater tie of non-family executives' pay to firm performance, less trust in non-family executives, stronger emphasis on the non-family executive's fit with family values as an evaluation criterion, and lower pay for family than non-family executives. Other studies find that, in such firms, families fill the void of board control (Lubatkin et al., 2005; Schulze et al., 2001), and performance criteria applied to the agent include not only a fiduciary role, but also the ability to satisfy family needs (e.g. hire family members over better qualified applicants, work effectively with family members, act altruistically towards family members, and the like) (Casson, 1999; Lubatkin et al., 2005; Schulze et al., 2001). As Gomez-Mejia et al. (2007) suggest, owner-managers willingly sacrificed firm performance in order to protect what they called ‘socioemotional’ wealth or what others have referred to as ‘familiness’ (Habbershon et al., 2003). Likewise, Gomez-Mejia and colleagues (Berrone et al., 2010; Gomez-Mejia et al., 2010, 2011b; Jones et al., 2008) find strong evidence that family principals are often driven by non-economic objectives, and this is reflected in such strategic choices as lower diversification, less pollution, culturally-related international expansion, and lower research and development expenditures. Until very recently the agency literature has neglected the special role of family principals in the pursuit of non-economic objectives, even though the vast majority of firms around the world are controlled by families (La Porta et al., 1999).
What this brief review suggests is that family ownership represents a unique context in which to view P–A relations. Thus developing a generalizable theory of principals in which family ownership is given prominence (e.g. Lim et al., 2010) could provide a first step towards understanding P–A relations across various institutional contexts in which ownership is held by founding family members.