CH3 Trust in the Economy
The concept of trust, mentioned briefly in the previous chapter on norms, is important enough to deserve its own separate treatment. Trust and trustworthy behavior are critical assets for any economy, principally because they lead people to cooperate and otherwise act more benignly toward one another than the pure logic of self-interest would predict. This is one reason why I consider purely self-interest-based explanations for trust perverse. Trust matters because, as economists have often noted, the resulting cooperation saves substantial costs of precaution and monitoring that would be expended in its absence. Kenneth Arrow observed that trust “is an important lubricant of a social system. It is extremely efficient; it saves a lot of trouble to have a fair degree of reliance on other people’s word” (1974: 23), and Arthur Okun suggested that “enormous resource costs could be saved in a perfectly honest and open world that would permit do-it-yourself cash registers and communal lawn mowers” (1980: 86).
But historically, economists paid trust little attention, perhaps because, as Albert Hirschman noted in his remarkable book The Passions and the Interests (1977), from the seventeenth century on, philosophers argued that economic action was a species of calm, rational, and benevolent behavior and thus economic interests were pursued only by gentlemanly means (Hirschman 1977; and see Fourcade and Healy 2007). This assumption came to be widely accepted by classical and neoclassical economists (though not by those of socialist and other heterodox persuasions—see Hirschman 1982), and the Hobbesian question of how society contained the perils of force and fraud—which highlights the problem of trust—thus faded from the analysis of economic life.
Two related mid-twentieth-century developments stimulated a resurgence of economists’ interest in trust. One was the advent of an economics of information, which noted the difficulties that asymmetric information causes. This was first of special interest in insurance markets, which face the dual problems of “moral hazard” (insurance reduces the motivation to avoid dangers insured against, but insurers cannot know, without large search costs, which claims result from such negligence) and “adverse selection” (those at higher risk are more likely to buy insurance but do not fully disclose this risk to insurers).
The interest in asymmetric information and uncertainty accompanied and was compatible with increasing attention to the limitations of human rationality. One manifestation of this attention was closer analysis of the micro-level details of imperfectly competitive markets, peopled by small numbers of traders with sunk costs and “specific human capital” investments. In his 1975 book Markets and Hierarchies, Oliver Williamson noted that any complex contingent contract that specifies that obligations of each party depend on what has occurred faces difficulties when the parties differ, as they often do, in knowledge of relevant occurrences (31-37). This led Williamson to search for organizational devices that mitigate the tendency for actors to pursue their interests with “guile,” and in general, he and other “new institutional” economists have stressed organizational and institutional solutions and downplayed the significance of “trust” as being confined mainly to families and close personal relationships in “noncalculative” situations of minor economic significance (see esp. Williamson 1993). Williamson here implicitly makes the Hobbesian assumption that one can normally expect others to deceive and betray unless restrained by organizations and institutions and thus interprets “trust” to mean belief that this will not happen even in the absence of such restraint, which he thinks justified only in close relations unlikely in most economic transactions.
Yet many social scientists have focused heavily on the role of trust in social and economic life, largely because there are so many real-life situations in which individuals do cooperate more readily and to a greater extent than pure instrumental rationality predicts. Even those whom real-world evidence does not persuade have been impressed that experimental results on “decision dilemmas” consistently lead to outcomes that are perplexing if we try to avoid the concept of trust.1 As Elinor Ostrom pointed out, the some- times-technical discussions of these outcomes, when broadly considered, really raise the Hobbesian question: “How do communities of individuals sustain agreements that counteract individual temptations to select shortterm, hedonistic actions when all parties would be better off if each party selected actions leading to higher group and individual returns? In other words, how do groups of individuals gain trust?” (2003:19). The issue in the large experimental literature on social dilemmas is that for a Pareto optimal outcome,2 players must cooperate by selecting “strategies other than those prescribed by a subgame-perfect equilibrium solution” (23). Most relevant experimental studies have found levels of cooperation well above the predicted level of zero, and while these levels sometimes decline when experiments are repeated, face-to-face communication substantially increases these levels again, even without changes in incentives. This Ostrom links to the building of trust (34). I have reviewed some of this experimental literature in Chapter 2 with reference to the norm of reciprocity.
What then should we mean by “trust”? There are many explicit and implicit disagreements in the voluminous literature on trust, but most students of the subject agree broadly that trust is the belief that another person with whom you might interact will not cause you harm even though he or she is in a position to do so. Such a belief on the part of a “trustor” may lead to “trusting behavior”—predicated on the assumption that the “trustee” (a term that I will use in what follows to refer to the person who is trusted) will act in a “trustworthy” way. So a trustor puts herself at some risk because of her trusting belief and action, and the existence of such risk is a central element in nearly all definitions (cf. Gambetta 1988: 219), of which the following three are typical: (1) in the literature on trust in organizations, a widely cited definition is that trust is a “psychological state comprising the intention to accept vulnerability based upon positive expectations of the intentions or behaviors of another” (Rousseau et al. 1998: 395); (2) Foddy and Yamagishi (2009: 17) propose that trust is “an expectation of beneficent reciprocity from others in uncertain or risky situations”; and (3) Walker and Ostrom (2003: 382) similarly define trust as the “willingness to take some risk in relation to other individuals on the expectation that the others will reciprocate.” Despite the convergence in definitions, there is little agreement on measurement (see, e.g., McEvily and Tortoriello 2011).3
One reason for the lack of consensus on measures is that the broad definition is compatible with many different reasons one may trust another, but most scholars focus narrowly on some single such reason, which leads to a single corresponding measure. It is therefore useful in setting out a systematic account of trust in the economy to talk about the main such reasons, the implications of each, and how they relate to one another. A major theme of this chapter is my resistance to the argument of many scholars that only trust caused by their favorite reason should be called “trust” at all.
1. Trust based on knowledge or calculation of interests of the other (rational choice accounts). Perhaps the simplest possible argument about trust and trustworthy behavior is that a potential trustor assesses whether the interests of the trustee would lead her to be trustworthy, considers the benefits and risks to herself of the other’s possible actions, and then acts in a trusting way only if it is to her benefit to do so. Thus, James Coleman presents an expected utility maximization model in which a rational actor engages in trusting behavior towards another if the “ratio of the chance of gain [from trusting behavior] to the chance of loss is greater than the ratio of the amount of the potential loss to the amount of the potential gain (1990: 99). Note that this is equivalent to assuming that the expected gain to the trustor if the trustee acts in a trustworthy way is greater than the expected loss if the trustee betrays her.
There is an obvious danger of circularity in such an assumption if we observe trust only after the fact. To avoid this requires us to assume that individuals are able to make calculations of this kind—which involves questions of cognitive capacity and information acquisition—and also that the gain or loss from possible behaviors of the other can be clearly quantified. But even when all these conditions are met, as they may often not be, the trustor’s assessment of costs and benefits and the likelihood of betrayal need not assume that the behavior of the other is based on rational choice. If, for example, one knows that one’s friend will not betray because of her group membership, her normative commitments, her emotional attachments, or other nonrational causes, then only one of the two actors involved in this trust decision is in fact a rational actor, and this is troubling for any claim that this is a “rational choice” theory of trust.
While it might appear that trusting another because of their attachment to you would transcend rational choice, Russell Hardin’s account of trust attempts to re-position such an argument squarely into a rational choice framework (e.g., Hardin 2001, 2002). In order to do so, Hardin argues that the concept of trust should mainly be confined to others you know, since you can trust them if and only if you know that they have an interest in continuing your relationship. This he refers to as the idea of “encapsulated interest,” where the interests of another include (or “encapsulate”) your own. He justifies this argument by noting that if in fact trust always required more than the
rational expectations grounded in the likely interests of the trusted . . . then we are at a very early stage in the development of any theory to account for trust or even to characterize it in many contexts. If an account from interests is largely correct for a large and important fraction of our trusting relationships, however, we already have the elements of a theory of trust that merely wants careful articulation and application. . . . The sense that trust inherently requires more than reliance on the self-interest of the trusted may depend on particular kinds of interaction that, while interesting and even important, are not always of greatest import in social theory or social life—although some of them are, as is the trust a child can have in a parent. (Hardin 2002: 6-7; cf. Williamson 1993)
This account relies on a claim of parsimony but also displays an element of wishful thinking. Closer examination suggests complexities that muddy the alleged parsimony. In particular, if trust is based on the assumed interest of another in continuing our relationship, then a savvy actor would need to know more about the nature of that interest. This varies in ways that Hardin places along a dimension that he calls “richness” but does not define:
At a minimum, you may want our relationship to continue because it is economically beneficial to you. . . . In richer cases, you may want our relationship to continue and not be damaged by your failure to fulfill my trust because you value the relationship for many reasons, including nonmaterial reasons. For example, you may enjoy doing various things with me, or you might value my friendship or my love, and your desire to keep my friendship or love will motivate you to be careful of my trust. (2002: 4)
I would describe this dimension as ranging from instrumental to consumma- tory, as I discussed above in Chapter 1, referring to whether a goal (here, maintaining a relationship) is desired as a means to another end or only for its own sake. Which is the case makes a big difference in talking about trust. This is because when deciding whether to trust another, one would like to know whether she will be unconditionally trustworthy (as in the purely con- summatory case) or whether she may perhaps look for subtle and undetectable ways to betray the trust placed in her (as in the purely instrumental case). If the latter, trust must be very guarded indeed, as the human mind and complex economic institutions present a multitude of opportunities to the devious.
The problem is that only in the purely consummatory case do another’s interests truly encapsulate yours. Here, no trustee will betray because if she harms the trustor’s interests, even undetectably, then she harms her own as well—as follows from how Hardin defines “encapsulation.” But in the many cases that Hardin discusses, where the trustee wants a continuing relationship because of benefits derived from it—such as money, prestige, position, reputation, resources, or contacts—there is no true encapsulation, and in fact a rational other should want to extract the maximum benefit from the relationship regardless of harm to the trustor, provided only that she not be discovered doing so and that this extraction not reduce the trustor’s ability to benefit her.
In deciding whether to trust another, one must then assess whether the other seeks continuation of the relationship for its own sake—as in love or close friendship—or for something to be gained outside the relationship. In the latter case, where instrumentality blocks a true encapsulation of interests, one must be appropriately wary. The rational choice account is muddied because in real life, these ideal types are hard to distinguish and there is typically a mixture of motives. The decision on how much to trust must depend on understanding the relationship sufficiently to know the balance of motives and how this affects the other’s behavior.
When true encapsulation of interests results from love or deep friendship, there is a certain irony in construing this to ratify some rational choice model, based, as it ultimately is, on the least understood and most subtle emotions and passions of humankind. Whether we really want to understand this situation by referring to “interests” depends in part on what that means. With a sufficiently broad concept of “interests” we might feel more comfortable. Swedberg defines “interests,” for example, as anything that “drive[s] the actions of individuals at some fundamental level” (2003: 293-295). In this conception, “interests” seem equivalent to “motivation,” and if “rational choice” then means acting in accord with “interests” so defined, then all motivated behavior is rational choice by definition. Hardin, on the other hand, construes “interests” more narrowly but notes that interests are typically not the “whole story of a person’s motivations” because one may have “an interest in having more resources, such as money, only because they enable me to consume or experience various things. . . . The whole story is one of well being through the use of resources. Interests are merely a proxy for this whole story” (2002: 23). But this synecdoche misleads, since that piece of the story beyond “interests” requires a different theoretical argument.
2. Trust based on personal relationships. Hardin’s “encapsulated interest” argument is a special case of trust based on one’s personal relationship to another, in that it tries to assimilate such trust to a matter of interests and rational choice. A different argument about trust and personal relations is made by Zucker (1986), who proposes that industrial society has gradually shifted the grounds for trust from more personal to more institutionalized ones. For Zucker, in the early period (which she identifies for the United States as occurring before the late nineteenth century), she refers to personal sources of trust as being “process-based,” by which she means that trust is “tied to past or expected exchange such as in reputation or gift-exchange” (1986: 60). So this kind of trust depends on having exchanged previously with the potential trustee or at least knowing the reputation of that person or firm for making satisfactory exchanges. Here one might think of the way that physical markets, such as bazaars, sometimes produce stable relationships among particular buyers and sellers, of the kind that anthropologist Clifford Geertz referred to as “clientelization” (1978) because exchange partners have built up trust in one another in situations where quality of goods is very hard to assess before purchase. As anthropologists and social exchange theorists have often noted, potential exchange partners typically work their way up from minor to more major exchanges in order to test the other’s reliability in reciprocating properly (see, e.g., Blau 1964: 94 ff.).
Although this exchange-based trust and that based on “encapsulated interest” are compatible with a rational choice argument, that argument is neither a necessary nor a sufficient condition for supposing that trust depends on personal relations. Whether based on instrumental or consummatory motives, that trustworthy behavior may be a regular part of a relationship reflects one of the typically direct effects of relational embeddedness (see Ch. 1) and explains the widespread preference of many economic actors to deal repeatedly with the same others. Our information about such partners is cheap, richly detailed, and typically accurate. But assessing the balance of instrumental and consummatory motives in others is not always easy, and the trust engendered by personal relations presents, by its very existence, enhanced opportunity for malfeasance, which we must note to avoid a simplistic functionalism. In personal relations an old song reminds us that “you always hurt the one you love”4—which is possible because someone who loves you is far more vulnerable than a stranger. In the Prisoners’ Dilemma, knowledge that one’s co-conspirator is certain to deny the crime (because, e.g., she loves you) presents all the more rational motive to gain by confessing, and personal relations that abrogate this dilemma may be less intense and symmetrical than imagined by the party to be deceived. This elementary fact of social life is the bread and butter of “confidence” rackets that simulate close personal relationships, sometimes for long periods. The greater the trust, the more to be gained from malfeasance. That this occurs infrequently is a tribute to the force of personal relations and their capacity to transcend simple rational choice; that betrayals do occur shows the limits of this capacity.
Correspondingly, in her random sample of files from the Securities and Exchange Commission over the period 1948-1972, Shapiro “found the degree of intimacy of prior victim-offender relationships surprising. There are indeed more cases in the sample in which at least some of the victims and offenders were acquainted . . . than those in which they were strangers. . . . This . . . conflicts with stereotypes of white-collar crime in which a chasm of interpersonal distance, disembodied transactions, cover-up techniques, middlemen, records, papers, documents and computerization are thought to permanently separate victim and offender” (1984: 35). So individuals have reason to continuously scan relationships to determine the balance of motives behind them. One reason this is difficult, aside from instances of clever deception, is that even relationships that begin for obviously instrumental purposes may develop an overlay of social content of the kind I call consummatory— where part of the reason for maintaining the relationship becomes the value of the interaction itself.5
When personal relationships do lead to trust and trustworthy behavior, one may ask what kind of argument might best explain this outcome. In instrumental cases, it seems reasonable to suppose that the trustee is indeed protecting his interests by acting in a trustworthy way, though, as I noted above, this case calls for wariness on the part of the trustor, since, by hypothesis, there is incentive for the trustee to deceive. As I also noted, the consum- matory case, where the relationship is valued for its own sake, is an uncomfortable fit for the usual rational choice paradigm. But then what does drive it? One argument might be that in this case trustworthy behavior is driven by emotions that lead to love or other varieties of attachment to another. Behavior driven by affect is one of Max Weber’s four fundamental types of social action (Weber [1921] 1968: 24-25; cf. Elster 1999).
A different way to think about how trust and personal connections are related is to consider the idea that people act in certain ways because of their conceptions of who they are, what kind of person they want to be, and what kind of obligations they have to other individuals and groups; such arguments usually fall under the heading of “identity.” A core element of philosophical and sociological arguments about identity and the constitution of the self is that these emerge out of the interactions we have with others. As argued by classic figures such as Charles Cooley and George Herbert Mead in the early twentieth century, we have little way of judging what we are like and what our characteristics are except insofar as we learn what others think of us and how they view us (cf. Blumer 1969). A natural extension of this is to say that the specific relationships we have with others, and their contents, are building blocks of our identity or conception of self. Individuals in close relationships with others arrive at clear expectations of their behavior toward one another, that are on each person’s side, to the extent the relationship is serious and long-lasting, part of their sense of self. So, for example, I may deal fairly with you not only because it is in my interest or because I have assimilated your interest to my own but because we have been close for so long that we expect this of one another, and I would be mortified and distressed to have cheated you even if you did not find out—it would be inconsistent with how I think about myself. This would explain the severe sense of becoming partially unmoored and losing part of one’s self when one loses a loved one as the result of death, the ending of a relationship, and perhaps worst of all by unexpected betrayal.
This way of thinking is different from saying that I will be trustworthy because I encapsulate your interests in my own, although that may be true as well. But to act in a way that is consistent with your personal identity is action caused by something about yourself rather than something about the relationship between your interests and those of another. And it is also different from acting according to a moral code, though that may also be involved. It is more about acting in a way that reflects the person, or kind of person, you have decided you are or want to be.
3. Trust based on membership in groups and networks. The discussion of how trust and trustworthiness are impacted by personal relationships is at a rather micro level and depends, as I have suggested, on a concept of relational embeddedness. But pairwise relationships are nested in more complex structures of social relations, which correspond to what I have called (Ch. 1) “structural embeddedness.” The simplest argument associated with social structures beyond dyads is that trust is more likely among those who consider themselves members of the same group, however “groupness” may be defined.
Cook, Levi, and Hardin comment that in the research on trust that they coordinated, funded by the Russell Sage Foundation, a major emphasis is on “situations in which ethnic, racial or other markers facilitate certain kinds of trust relationships while inhibiting others and when they do not” (2009: 2). Foddy and Yamagishi suggest that shared group membership is particularly critical in understanding how the previously unacquainted may trust one another. They suggest two possible reasons to trust a fellow group member: (1) stereotype-based trust, where you think that your own group members are more generous, trustworthy, and fair; and (2) the “group heuristic hypothesis,” where we expect altruistic behavior from ingroup members toward one another (2009: 19). Their experiments suggest that the second mechanism is the critical one. Other experimental work confirms that strangers of the same race and nationality are more trustworthy toward one another (e.g., Glaeser et al. 2000: 814). But this empirical finding does not provide an argument for why this should be so. The dictator experiments of Habyarimana et al. (Cook et al. 2009) suggest that the ingroup effect comes from ingroup norms of reciprocity. Walker and Ostrom (2009: 105) note abundant evidence that “individuals sanction those who engage in selfish activities at the expense of other group members,” and that, moreover, norms of “fairness and reciprocity appear to shape the expectations of . . . group members beyond purely strategic responses” (2009: 107).
Here I note that although I write separate sections on particular causes for trusting and trustworthy behavior, most real situations where someone has to decide whether to trust another entail more than one of these causes, so the separation is artificial. In this particular case, part of the impact of common group membership on trust derives from commonly held norms about what group members owe to one another. The most seriously misleading arguments about trust arise from the attempt of scholars to limit explanations of trust to their own single, favorite factor, which typically leads to simplistic and irreproducible conclusions.
Zucker (1986) refers to “characteristic based trust,” which depends on characteristics such as family and ethnicity (60), which is ascriptive and cannot be invested in or purchased. She proposes that in the United States, this became more critical with economic development because the labor force became more culturally heterogeneous, and you had to interact with strangers, but you could at least assume that people with similar characteristics to your own would provide satisfactory outcomes, as in ethnic enclaves. Zucker’s take on why this works is not based on norms but on cultural familiarity with co-group members: “Many background understandings will be held in common, smoothing or eliminating the negotiation over terms of exchange and making it more likely that the outcome of the exchange will be satisfactory to both parties” (61). Broadly speaking, this is an argument about commonly held “cultures.”
Another argument about how common group membership leads to trust that is distinct from rational choice, norms, or culture is provided by “social identity” theory. Tyler (2001) notes that social exchange theory is based on the assumption that “people want resources from others and engage in organized life in order to exchange resources” and that they are “motivated by the desire to maximize their gain of resources and minimize their losses. . . . To do so they need to have an estimate of what others will do in response to their own behavior” (287). Though this is obviously sometimes the case, Tyler argues that it is “not a complete model of the psychology of trust” because people may also have feelings of obligation to a group that are “distinct from calculations about anticipated personal gain or loss owing to the actions of others” (288). He cites experimental evidence that “identification with the group to which one belongs decreases one’s propensity to engage in noncooperative behavior that removes resources from a common pool,” and this is so even without expectation of future reciprocity or current rewards or punishments or reputational consequences. Instead, people “feel an obligation to the group that develops out of identification with the group and group values. That identification shapes their behavior, leading to cooperation that is distinct from that based on expectations about the behavior of others” (288). This Tyler calls “social trust,” and he proposes that in groups where people have social connections, their trust judgments “become more strongly linked to identity concerns, and less strongly linked to resource exchange” (289).
These varying accounts show mainly that common group membership engenders trust and trustworthy behavior. Institutional economists have made similarly positive arguments. So, Ben-Porath, for example, in discussing the importance of trust in the exchange of valuable commodities, noted that “continuity of relationships can generate behavior on the part of shrewd, selfseeking, or even unscrupulous individuals that could otherwise be interpreted as foolish or purely altruistic. Valuable diamonds change hands on the diamond exchange, and the deals are sealed by a handshake” (1980: 6). His emphasis is mainly on personal relations between traders, but it seems clear as well that such transactions are possible also because they are not atomized from other transactions but embedded in a close-knit community of diamond merchants who monitor one another’s behavior closely and generate clearly defined standards of behavior easily policed by the quick spread of information about instances of malfeasance. The temptations posed by this level of trust are considerable, however, and the emergence of separate cohesive groups may bound the reach of trust, identity, and moral action.6
Frauds as well as legitimate business enterprises attempt to tap into existing membership networks in the hope of wide diffusion, more difficult if attempted through impersonal channels. In Shapiro’s study of SEC fraud investigations, she found, as I reported earlier, that victims and perpetrators typically knew one another. But the fraud usually was not just a dyadic matter but was structurally embedded: “Offenses touch victim populations containing groups of associates or portions of various social networks. The sample contains cases with victim pools composed of members of particular church congregations or ethnic associations, officers at several military bases, members of political or social clubs or recreational associations, members of a professional athletic team, a textbook editor and a network of social science professors, members of investment clubs, and networks of political conservatives” (1984: 36). Some such networks are brought into the fraud by “bird dogs”—enthusiastic investors aware of the fraud who convince others to invest; the use of celebrities or community leaders, usually innocent of the fraudulent nature of the scheme, is common as an incentive for others to participate (1984: 36-37). And, indeed, the vast Ponzi scheme of Bernard Madoff, uncovered in 2008, depended almost entirely on recruitment of investors through networks of trust, especially among wealthy members of the Jewish community.
4. Institutional sources of trust. A common theme in the literature on trust is that there are cases where one trusts another because of institutional arrangements that make deception or betrayal less likely. One of the main reasons people are at risk in such situations is that the other with whom they might deal is a stranger or at least not well known to them. To be sure, there are theorists who want to reserve the term “trust” for only those who know one another well, e.g., Hardin, who considers trust to depend upon “encapsulated interest.” In a later section, where I discuss the most reasonable scope for the concept of “trust,” I will reject this as well as other arguments for a narrow application of the term.
Arguments about the importance of institutional sources of trust sometimes make the evolutionary assumption that trust is originally personal and small-scale but that the increasing growth, complexity, and differentiation of societies make it impossible for all trust in the economy to be so derived, so that to the extent a society is economically successful, it will develop institutional supports that make it possible to take risks in relation to those about whom one knows less than they would in a much smaller-scale social situation. Cook, Levi, and Hardin suggest, for example, that as far back as Madison and Hume, it was argued that institutions like government were important in enabling cooperation and trust. If the state is reliable and neutral, it facilitates trustworthiness by “allowing individuals to begin relationships with relatively small risks as they learn about each other, and by providing insurance against failed trust” (2009: 4).
Zucker suggests that as societies progress economically, trust based on personal relations and exchange history or on group membership becomes insufficient, and institutions such as escrow accounts and credit ratings take up the slack (1986: 64-65; see also Carruthers 2013 on the history of credit ratings in the United States. But note that the history of the 2008 financial meltdown shows the severe limitations of credit ratings as a source of trust—cf. Lewis 2010). Zucker argues that between 1840 and 1920, institutionally based trust came to predominate in the United States (1986: 99) but does not actually suggest how one could measure the different kinds of trust in order to confirm this assertion. It is a common argument that a well-developed legal system that renders verdicts with some degree of impartiality in disputes is congenial to risk- taking in situations where exchanging parties might otherwise fail to reach agreement on terms. Even in situations where parties reach agreement without the use of formal institutions, the existence of these may provide a backdrop that overcomes distrust that might otherwise make this infeasible. An example is the oft-cited work of Mnookin and Kornhauser (1979) on how private negotiations on the terms of divorce amount to “bargaining in the shadow of the law,” since without legal guarantees of promises made, the “inability to make an enforceable promise may inhibit dispute settlement” (957)
So many institutional sources of trust are familiar in everyday life that it is not necessary to inventory many of them to make this point. But acknowledging their significance does not require accepting evolutionary arguments about how such sources “displace” mechanisms that assured trust in earlier periods or in less advanced societies. I consider these claims in more detail below.
5. Trust based on norms. It is fairly straightforward that one might think another trustworthy because she adheres to norms that prescribe such behavior. The scope for such trust depends on the nature of the norm. If it is a norm of reciprocity, trust another who owes you a favor. If the norm prescribes reciprocity based on group membership, trust others in your own group. If it is a norm that one should act in a trustworthy manner in general, then one might be justified in extending trust beyond merely reciprocative situations. Given that arguments about the importance of norms sound sociological, it is curious that most authors presenting this argument for trust are economists. These arguments divide into two broad streams. One is cul- turalist (as described in Ch. 1) and conceives of “norms” as pertaining not to individuals but rather to collectivities that formulate, enforce, and embody them. Like most culturalist views, this idea meshes uncomfortably with the usual methodological individualism of economics. Economists taking this view typically cite data provided by a single question asked in the World Values Survey (henceforth WVS: see http://www.worldvaluessurvey.org/): “Generally speaking, would you say that most people can be trusted or that you need to be very careful in dealing with people?” and respondents are asked to choose between two alternatives: “Most people can be trusted” or “You can never be too careful when dealing with others.” Countries vary dramatically from one another in the levels of trust, with the highest scores found in Scandinavia and the lowest in Latin America. (See, e.g., http://www .jdsurvey.net/jds/jdsurveyMaps.jsp?Idioma=I&SeccionTexto = 0404 &NOID=104).
Although standard economics is incurious about where preferences come from—treating utility functions, for example, as given rather than a dependent variable to be investigated—economists who study trust sometimes suggest that its presence is an element of the norms and “culture” of a nation, region, or ethnic, religious, or other social group, and this foray into culturalism is said to explain differences. They (e.g., LaPorta et al. 1997; Guiso, Sapienza, and Zingales 2006) cite approvingly works of non-economists who elaborate this view such as Fukuyama (1995) and Putnam (1993) (who casts trust as resulting from “social capital,” which leads some economists, such as Glaeser et al. 2000, to consider trust as a measure of social capital). But this view also leaves open the question of how the belief that others are trustworthy arises and how it may relate to other norms that may yield trustworthy behavior. (We know more about trusting behavior than about when and why people are trustworthy in part because surveys of values ask about trust but seldom if ever ask respondents whether and when they think it appropriate to cheat or deceive others, for the obvious reason that few would admit to ever thinking this appropriate.)
Economists address the question of how people come to be trustworthy by a stream of related argument that casts the passing down to children of trustworthy and cooperative or untrustworthy behavior as a decision that families or groups make, thus addressing the issue of how such norms arise. All such arguments encounter the difficulties noted in Chapter 2 in arguing that internalized norms are chosen rationally to gain advantage. So, for example, Aghion et al. (2010) define trust as “beliefs resulting from decisions about civicness made in families” (1015). Families, in this account, have two choices. They can teach their children how to behave in a “civic” way— “learning tolerance, mutual respect and independence”—or to “behave uncivicly outside the family” (1023). Uncivic children who grow up to be entrepreneurs can be expected to pollute, offer inferior risky goods, and cheat others. Societies that attain an equilibrium where everyone is civic naturally become “high-trust” societies and otherwise low-trust ones (1027-1028). Guiso et al. (2011) argue that we should focus on “investment in civic capital,” which is the “amount of resources that parents spend to teach more cooperative values to their children,” where “civic capital” is the “values and beliefs that help a group overcome the free rider problem in the pursuit of socially valuable activities” (423, emphasis in original). There is thus an “intergeneration- ally transmitted prior” that “affects each individual decision regarding whether to trust other members of the society and participate in an anonymous exchange” (424). If this trust is not well founded, then the individual could suffer a major loss. Thus, to “protect children from costly mistakes, parents transmit conservative priors to them” which can lead to an “equilibrium of mistrust” (425). People do, however, “adapt their norms and beliefs in response to the social pressure of the community they live in” (426) but differentially in relation to how strongly the norms are held: “If civic values are completely embedded in preferences, they should not be modified by socialization. If, however, civic values are supported, at least in part, by the desire to conform to others, then socialization can lead to changes” (426).
Before continuing to discuss the causes and dynamics of trust, it is useful to consider some issues about the scope conditions for use of the concept of “trust” that are closely related to the discussion of causes. Scholars of trust often argue for limiting the term to some specific circumstances. Here I argue against such limits and for construing the idea of trust very broadly, creating distinctions instead around differences in the circumstances under which trust is present and/or relevant and the different causes of trust and trustworthy behavior attaching to those different circumstances. This shifts the discussion away from what I think pointless disputes about whether a particular situation “really” involves trust to the more interesting, complex, and critical problem of how to understand under what circumstances economic actors do in fact trust one another—i.e., act in ways that make them vulnerable to others, assuming that these others will not take the opportunity to harm them or their interests.
One common argument is that because trusting someone requires the trustor to take a risk about the trustworthiness of the potential trustee, we should not apply the term “trust” to situations in which there is no such risk. This argument appears in various forms. Appealing to parsimony, Russell Hardin argues that if it is in another’s interest to avoid harming ours, our belief that they will not do so should not be called “trust” because to define trust as “nothing more than incentive compatibility or rational expectations of the behavior of the trusted” would make the term “otiose . . . because it would add nothing to the somewhat simpler assumption of compatible interests” (2002: 5). Foddy and Yamagishi similarly argue that trust is “not required when others’ interests are totally allied with our own,” and they refer to such situations as “the domain of assurance”; we only need to trust others when our interests are not allied, as this is when they can gain at our expense. Trust, they add, is most important in “uncertain, not certain relationships” (2009: 17). In their excellent study of taxi drivers, Gambetta and Hamill propose that “it is not enough to predict . . . that people will behave in a trustworthy manner if doing so is in their self-interest. This removes the problem of trust altogether” (2005: 4). Even broader is Farrell’s assertion that when “actors have good reason to be certain that others will cooperate, these expectations are better described as confidence than as trust” (2009: 25). The reason for this is that when I “know that another will behave honestly in a predetermined and well-anticipated situation,” there is not any real risk of unanticipated behavior (26).
The key assertion in all these statements that I want to contest is that there are circumstances where the behavior of the other whose trustworthiness we assess is completely predictable, without uncertainty. This is equivalent to saying that there are situations where actors have no agency whatsoever, though so rephrasing the assertion makes it seem more problematic. In the first two citations above, this certainty derives from knowing actors’ interests. But this assumes an implicit null hypothesis of the sort that I discussed in Chapter 1, that we should by default expect others whose behavior concerns us to act according to their interests. Insofar as this is not a tautology, i.e., if agents can act counter-preferentially—against their own interests—as Sen 1977 suggests, we may have uncertainty and therefore the possibility of trust even in the case where interests are aligned. Sen analyzes the example of someone acting so as to prevent or stop torture. Let us pursue this case: it may be in my colleague’s best interest to be complicit in my torture of prisoners under my care, so by the argument from interests, my expectation of his doing so should not be called “trust,” as his behavior is automatic and entirely predictable.
But this is so only if interests perfectly predict behavior. Sen’s argument suggests, however, that the other’s commitment to moral principles may intervene and cause my undoubting expectation of his silence to be misplaced. If we take counter-preferential behavior seriously, this is not simply a matter of how the balance of incentives plays out and cannot be reduced to a simple incentive-based account. My expectation of his complicit silence may indeed be a matter of trust, if there is some chance that he will pursue overriding normative principles and report my torture, even at possibly high cost to himself. To address fully the likelihood that this will take place requires a treatment of the role of norms of the sort that I undertook in Chapter 2. The point is that whether expectations of another’s behavior should be described as “trust” may depend not only on the other’s interests but also on any other factors that may cause her to support or harm my interests. So the assumption that behavior is driven by interests alone is an implicit null hypothesis that may often be plausible but at critical times can dramatically mislead, as I noted in my discussion of “sacred values” in Chapter 2.
Frequent accounts of “whistleblowers,” many of whom experience serious losses from their exposure of wrongdoing, exemplify the power of norms or identity to override self-interest. For this to matter it need not be the common or typical case, only a possibility, which can lead to serious consequences, as it did in such famous cases as the Enron frauds. I know of no systematic study of the balance of consequences for those who whistleblow, but I think it plausible from the fact that government agencies offer considerable rewards to those who flag wrongdoing that there is some presumption that without such rewards, the likely consequences for those who do so are negative on balance.
More generally, the fact that there are numerous causes to trust or distrust another besides their interests casts doubt on any argument that there are situations when another’s behavior is easily and completely predictable. This is not to say that judgments of trust are random or not plausibly based on information available to us about the likelihood of another being trustworthy, nor does it follow that new information will not reasonably give us more or less confidence in that likelihood. Indeed, one of the most important research agendas on trust should be to understand better how people make such judgments and to what extent they are accurate. Here I only mean to suggest wariness of oversocialized views of human action that depict others as acting in ways entirely determined by factors that we understand without doubt, as this discourages the detailed and subtle investigations of actual trusting behavior that are needed.
Another aspect of when it is appropriate to speak of “trust” concerns the level of social structure where the term well describes how we might assess the likely future behavior of others. While some theorists propose that we can only properly speak of trust in cases where people know each other well, others suggest quite the opposite—that the concept is helpful almost entirely in relation to how we deal with strangers. I argue against both propositions and against restricting the idea of “trust” in such ways.
As noted above, Russell Hardin has proposed that we can trust others mainly when their interests encapsulate ours, so they benefit from continuing our relationship. On this argument, trust is fundamentally a small-scale interpersonal phenomenon and cannot be of much importance in the more macro-level structures of a large industrial society. This position is elaborated by Cook, Hardin, and Levi (2005) (CHL), who assert that as societies become more complex, “the actual role of trusting relations has declined relatively” so that trust is “no longer the central pillar of social order, and it may not even be very important in most of our cooperative exchanges which we manage quite effectively even in the absence of interpersonal trust” (1). They argue that that for complex societies to work well requires institutions, such as third-party enforcement of obligations, that make exchange and other kinds of cooperation possible even when interpersonal trust is absent (2). The argument construes trust in Hardin’s sense of “encapsulated interest,” and by this definition it is “impossible . . . to trust strangers and even many of our acquaintances, and . . . virtually impossible . . . to trust institutions, governments or other large collectivities” (4-5). This being the case, “trust plays a relatively small role on the grand scale in producing and maintaining social order. We usually rely on and cooperate with each other, not because we have come to trust each other, but because of the incentives in place that make cooperation safe and productive for us” (14-15). The reach of trust cannot extend very far because it is “only beneficial for us to trust those who are trustworthy in our interactions with us, and these people constitute nowhere near all of the society” (68).
This argument is similar to that of Zucker (1986) who, however, refers to such cases as instances of trust based on institutional sources. I would take this position as well, since I believe that the fundamental dependent variable here, whether at a small scale of interpersonal relations or a larger scale where people consider the impact of institutions on the actions of others, is still whether people behave as if those in a position to cause them harm will in fact do so, and the broad question should be about what independent variables cause one assessment or another.
I believe that views of where trust is relevant turn on how scholars think about the impact on trust of interests as compared to that of norms. CHL, following Hardin’s concept of trust as based on encapsulated interpersonal interests, want to separate trust from the confidence you might have in others’ behavior based on the power you think norms have over them. Thus, they assert that in small-scale communities, “trust is generally not at issue [i.e., a relevant concept]” because in small, dense networks, “reliability can be enforced by norms that are backed by the sanctions the community would apply” (2005: 92). Thus in small towns, helping behavior is not, as one might think, caused by interpersonal reciprocity but rather by “helping or communal norms” unlike seemingly similar behavior in urban areas which really is “a matter of reciprocity” given their assumed absence of communal norms (92). This argument borrows a page from Durkheim’s (1893) concept of “mechanical solidarity” in assuming that people in small towns lack individuality or strong dyadic relations that are not completely subsumed by the “community ethos.” I suggest that this idea implicitly reflects an “oversocialized” conception of human action, as it strips away agency from actors in “communal” settings. CHL extend Durkheim’s evolutionary account by suggesting that over the course of “social evolutionary time” we may think of trust as “rising to displace control by social norms and then . . . fading to be displaced by regulation by modern social institutions” (195). Thus the “massive institutionalization of most of life makes modern society possible when mere [sic] trust could not have done so” (197).
This argument, and its separation of the influence of norms from the concept of trust, depends on conceiving of norms as pertaining not to individuals but rather to collectivities that formulate, enforce, and embody them. Such a concept has a family resemblance to the sociological exceptionalism promoted by Durkheim and others, that mental concepts are not the properties of individuals, and that society is an entity sui generis rather than a mere collection of separate persons. If norms have this effect on an entire group, then it is plausible to suppose that one can expect another to act in a trustworthy way not because of any characteristics of the other and not because of the relationship you have to her but because of your common membership in a group whose norms guarantee that behavior and eliminate all risk from the situation. Here, the criterion for not applying the concept of trust is a mixture of the imagined certainty of the situation and the fact that whether to act in a trusting manner has been stripped of any connection to assessing anything about the other as an individual (rather than as a group member) or about your relationship to him or her, which is a central part of what “trust” usually means.
If, however, you think of norms as possessed not by groups but by individuals, as a null hypothesis of methodological individualism implies, then the relationship of norms to trust can be quite different from and even opposite to the CHL proposal. This description fits well the way that economists and their sympathizers have talked about trust in the past twenty years, from which they conclude that norms are in fact a major source of trust, and that, moreover, trust is relevant mainly in one’s relations to strangers rather than between those who know one another well.
So, for example, LaPorta, Lopez-de-Silanes, Shleifer, and Vishny (henceforth LLSV) (1997) propose that trust is “more essential for ensuring cooperation between strangers, or people who encounter each other infrequently, than for supporting cooperation among people who interact frequently and repeatedly.” In order to assert this, these economists must have a very different idea about how cooperation comes about within small communal groups than the CHL idea that shared norms create unfailingly reliable behavior, and indeed, they argue that in such small and close-knit social clusters, as in families or partnerships, automatic and invariable cooperation, without deception or malfeasance, is supported by reputations and the likelihood of transgressions being punished even if levels of trust are low (333)—in other words, cooperation results from interests. That both CHL and LLSV agree on the unproblematic nature of cooperation in such small settings, though for quite different reasons, corresponds to what I have flagged in an earlier work (Granovetter 1985) as the convergence of oversocialized and undersocialized accounts, in this case to agreement that individuals in small communal settings lack agency, which makes trust irrelevant. But the views diverge because the economists conclude that trust is most needed in large organizations since you interact a lot with people you don’t know well, so that the power of reputations and the likelihood of punishment for deviations are reduced.
As I described in the previous section, it is in these circumstances that economists often appeal to the power of norms in driving trustworthy behavior, drawing on empirical data such as those provided by the World Values Survey. They often code such norms and values as being part of the “culture” of geographic units, typically nations. In so arguing, they cite scholars such as Fukuyama (1995), who believes that national culture determines the distribution of trust and, in particular, that societies vary in the extent to which people are able to trust others beyond their family circle, and that societies can be broadly dichotomized into those characterized by “low trust,” where people trust principally family members, and ‘high trust,” where it is more common to trust those outside the family. The reason this matters, he argues, is that in “low trust” societies, family bonds tower over other social loyalties, with the consequence that collections of economic actors based on mutual trust must be small. Family businesses dominate, and such societies are unable to develop large, professionally managed corporations. This implies difficulty adopting efficient, modern management practices and inability to “move into certain sectors of the global economy that require larger scale” (110). Large firms, if they exist in such societies, will not be private but can only be state-owned and managed, so there will be very large state-owned firms and small family firms without much in between. On the other hand, high-trust countries, whose cultures allow and encourage trust outside the family, make it much easier to form large firms. While legal forms like the joint-stock company allow unrelated people without trust to collaborate, nevertheless, “how easily they do so depends on their cooperativeness when dealing with nonkin” (150). Countries with well-developed patterns of association with nonkin have an emphasis on community and communitarian institutions, often referred to as “social capital,” and this eases the transition from family business to professional management.
LLSV (1997) cite this argument with approval because it is compatible with their assertion that trust in strangers is critical for large-scale organizations and economic activities to thrive, and they note that measures of the level of trust in families in survey data are negatively correlated with the significance of large firms in the economy (336).
If one thinks that differences in trust across nations or other geographic units depend largely on cultural differences, some way is needed to link such an argument to methodological individualist theory. In the previous section I reported that economists such as Aghion et al. (2010) and Guiso et al. (2011) propose that trustworthy behavior results from family decisions to teach their children to be “civic.” The link of this argument to culture lies in the assumption that cultures condition and affect the decisions of families to transmit the tendency to be trustworthy to their children, and this transmission, when aggregated up to a macro level, has a major impact on economic action. For example, the proportion of individuals who become “civic” (read: trustworthy) in the model of Aghion et al. (2010) is a major determinant of the extent of regulation of the economy, and high-trust societies “exhibit low levels of government regulation and low-trust societies exhibit high levels” because “distrust drives the demand for regulation. In low-trust societies, individuals correctly do not trust business because business is dishonest”; even government corruption is less bad than this dishonesty (1028). Note that this is an account “in which beliefs and regulations jointly influence one another” but where there is virtually no study of the behavior or sequence of events that intervenes between individual beliefs and larger-scale economic patterns. Another way to say this is that there is little interest in the mechanisms that lead from beliefs to institutions (for more on mechanisms in social theory, see Hedstrom 2005 and the essays in Hedstrom and Swedberg 1998).
Similar arguments are presented by Guiso, Sapienza, and Zingales in a series of papers on culture, trust, and economic outcomes. So they note that trust beliefs affect the probability of someone becoming an entrepreneur (where the measure is self-employment—2006: 36) and that, using the WVS measure of trust in Holland, “trusting individuals are significantly more likely to buy stocks and risky assets . . .” (2008: 2558). They conclude this implies that companies will “find it more difficult to float their stock in countries characterized by low levels of trust” (2559). In their 2011 account they extend the argument about trust to incorporate ideas about “civic capital,” which they characterize as the “missing ingredient in explaining the persistence of economic development” so that “communities/countries that, for an historic accident, are rich in civic capital enjoy a comparative advantage for extended periods” (420).
In this account, civic capital is the result of investment. It is the “amount of resources that parents spend to teach more cooperative values to their children” (423), and lest they fall victim to the usual criticism of “social capital” that one might learn to cooperate in criminal, racist or other socially undesirable activity, they define this away by stating that the definition of civic capital “purposefully excludes . . . those values that favor cooperation in socially deviant activities, such as gangs” (423). (The authors seem confident of universal agreement on what is “socially deviant,” uncomfortably reminiscent of Talcott Parsons’s mid-twentieth-century focus on social consensus, which sociologists have long since abandoned.) So the argument proceeds as one in which parents “decide how much trust to transmit to their children” and this “intergenerationally transmitted prior affects each individual decision regarding whether to trust other members of the society and participate in an anonymous exchange” (424). From a theoretical point of view, they note, we could talk about trust in family or neighbors or more generalized trust, but they argue that the latter is the right measure because for “institutions and markets to work properly, people need to trust strangers” (442). As in the work of Aghion et al., behavior and events intervening between beliefs and larger-scale outcomes are glossed over or attributed to “historic accident.”
To summarize my discussion of what is the proper scope for the concept of trust, I think it counterproductive to confine it to small-scale situations where individuals know one another well or to argue that it should only apply to large-scale situations where people interact mainly with acquaintances or strangers. To me it is more fruitful to theorize at both small- and large-scale levels under what circumstances people assume that others in a position to hurt their interests will not do so. But while saying this opens up the issue of trust to more general arguments, it does not yet clarify what, if any, relation there is between trust at a small-scale level and that in large, complex organizations that define the macro shape of an economy. If we think of trust as being at issue at both levels, then this relation becomes especially important to theorize, and I suggest a high level of caution about arguments that link individual decisions to large-scale outcomes without a detailed or plausible account of how this aggregation takes place.
The arguments on trust that I have reviewed and critiqued focused either on very small-scale instances of trust or on accounts of trust at a larger social and institutional scale that either derived such trust from historical and political developments or assumed aggregation from individual beliefs without providing theoretically detailed or coherent arguments and behavioral mechanisms to explain such developments. A fuller treatment of trust would explore this aggregation more thoroughly, and part of this discussion would concern how political, historical, macroeconomic, and other institutional contexts are critical in explaining trust at higher levels. Here I set myself the more limited goal of setting out some social network ideas that may provide an important piece of the puzzle as to when trust does or does not aggregate up from micro to macro levels.
First note, as I suggested in Chapter 1, that relational embeddedness bears heavily on trust. Consider whether I cheat a business associate with whom I have friendly personal relations. Whether I do depends in part on the nature of my relation with him. It also depends on the configuration of incentives and on those moral principles I apply to the situation, and both of these are in turn affected by this relation. But incentives and moral principles are also determined by structural embeddedness—the structure of ties within which my relation with my friend is located.7 My mortification at cheating a friend of long standing may be substantial even when undiscovered. It may increase when the friend becomes aware of it. But it may become even more unbearable when our mutual friends uncover the deceit and tell one another. Whether they do will depend on the structure of the network of relations— roughly speaking, on the extent to which the mutual friends of the dyad in question are connected to one another. When these connections are many— the situation of “high network density”—the news will spread quickly; when they are isolated from one another, much less so, as I argued in Chapter 1. So we can expect greater pressure against such cheating in the denser network; such pressures are an important part of incentives and relate directly to economic and social costs of developing a bad reputation.
But the pressure against cheating arises not only because of direct sanctions that group members would apply to me or because of reputation, both matters of interests and rational choice, but also because cohesive groups are more efficient than those with sparse relational networks at generating normative, symbolic, or cultural structures that affect our behavior. Thus, in such a group, it may never even occur to me to cheat my friend since I have absorbed a set of standards from the group that literally makes it unthinkable, at least in the group setting. So at relatively small-scale and communal levels, both interests and norms bear on trust. It is a commonplace from studies of intergroup relations, however, that the most scrupulously adhered-to norms within a well-defined group may be considered irrelevant when dealing with those outside its pale. Closely related yet importantly distinct from arguments either about information spread and sanctions or about norms is “social identity” theory, as discussed by Tyler (2001), which I noted above in my discussion of how group membership impacts interpersonal trust. The situational aspect of normative influences on behavior results from the structural embeddedness of social action and its impact on social norms as mediated by group identity. As I have noted, the power of these identities can also be harnessed on behalf of frauds that exploit the trust within identity groups or lead to conflict when such groups become fragmented.
The discussion thus far assumes that trust depends on preexisting relational and structural embeddedness and group identities but does not inquire how these arise. To assume that the situation of embeddedness is fixed and unalterable implies that the configurations of possible trust depends entirely on structure and cannot be impacted by conscious action of agents. This fatalistic view is sometimes drawn upon to show why regional differences in “social capital” are intractable, deriving as they do from many centuries of civic disengagement, “amoral familism” (Banfield 1958), or other afflictions closely related to trust or its absence. But it is important to keep in mind that social networks are themselves embedded in an economic and political institutional context that may have important impact on who comes into contact with whom and with what result.
Sabel, for example, suggests that the boundaries between trust and mistrust are blurred in practice, and that the absence of trust does not preclude discussions of conditions under which it might exist or be created. Both he (Sabel 1993) and Locke (2001) suggest that industrial policy carried out by government at various levels, with the assistance of private groups, may have the consequence of forcing actors to work together who previously had only thought of themselves as having opposed interests that made trust impossible. For Sabel’s Pennsylvania case, he notes that, in effect, the different groups he studied redefined their situation as the result of having to interact with one another. Locke’s cases come from areas often written off as culturally incapable of forming trusting economic relationships—southern Italy and northeastern Brazil; here, private associations with broad membership were the locus for generation of trust, but public policy was crucial because without its support and its encouragement for inclusiveness in association membership, the key actors would not have come together in the first place in such associations, and the suboptimal outcomes of a typical social dilemma—in this case, individual producers of cheese or melons producing adulterated or inferior products so as to free-ride on the reputation of the larger region—would have dominated, sinking local economic development (Locke 2001).
Relational ties and structural networks are embedded not only in contemporary institutions but also in a particular moment in time and space. How trust varies with these has attracted a great deal of attention, and in the empirical chapters of the sequel volume, I will develop some specific arguments about this. Here I would like to review and comment on the major positions that have been staked out.
Though much contemporary writing on trust treats the subject as if unrelated to cultural, institutional, or historical variations, there are accounts that explore this connection. Allan Silver, for example, argues that the eighteenth-century Scottish moralists, most famously Adam Smith and David Hume, considered personal relations to have been considerably changed by the increasing dominance of markets. But unlike later critics of both left and right, from socialists to Burkean conservatives, who bemoaned the deleterious effects of markets on intimate personal relations, they proposed that a vigorous market actually carved out a new and important place in society for friendships unencumbered by the calculations of social exchange (Silver 1990). Indeed, they “celebrate the liberation of friendship from instrumental concerns made possible by the advent of commercial society” (1480), arguing that such friendships are instead based on “sympathy,” an emotional tie innocent of interest calculations.
Before the market, they proposed, personal relations were necessary in order to ward off enemies or to acquire needed resources. This necessity in war, economy, or politics introduced an element of calculation into personal relations and made them “susceptible to damaging betrayal” (1487). The dominance of the market and associated legal institutions of contract as ways to provide goods and services and resolve disputes had the effect of “purifying” personal relations by “clearly distinguishing friendship from interest and founding friendship on sympathy and affection” (1487). So this was new, in the sense that only with “impersonal markets in products and services does a parallel system of personal relations emerge whose ethic excludes exchange and utility” (1494) and plays an important role in creating a “moralized civil society” In the modern, ideal conception of friendship that flows from this argument, personal trust “achieves a moral elevation, lacking in contractual or other engagements enforced by third parties” (Silver 1989: 276). And such trust is explicitly non-calculative because commitments “based on an understanding of others’ interests fall outside the moral ideal of modern friendship” (277). So this conception of trust between friends is in fact the polar opposite of trust as “encapsulated interest”
In later chapters I will assess how far this conception carries us. The sharp distinction between market relationships and the non-market relations of friendship that Silver attributes to the Scots is hard to maintain, as the empirical evidence will show, so we will need to reconsider this entire issue. The Scottish argument provides, however, an excellent reference point. But the idea that the nature of trust relations changes as institutions and culture do goes beyond the pre-market-commercial society distinction. One way this idea has been pursued has been to argue that different societies and cultures vary systematically in how much and in what way they facilitate trust among their members. Consider the argument of Fukuyama (1995), described earlier, that the existence and significance of trust at a large scale result from how it plays out at a small scale. Fukuyama argues for the crucial importance of a society’s particular culture, since he believes it determines whether people are able to trust those outside their family circle. In “low trust” societies where they cannot, collections of economic actors based on mutual trust must be small, family businesses dominate, and it is difficult to develop large, professionally managed corporations, which, however, form easily in “high trust” countries.
The most obvious criticism of this argument concerns Fukuyama’s particular classification of societies into the “low trust” or “familistic” category (China, France, and Italy being his main cases) and the “high trust” category (Japan, Germany, and the United States). Aside from whether countries in each grouping properly fit the description or belong together in such an argument, there are countries whose characteristics belie the main hypothesis—such as South Korea with its tight Confucian family system in an economy dominated by large, professionally managed and highly successful, yet typically family-based, business groups such as Samsung, LG, and Hyundai.
But I consider the more serious issue here to be the omission of inquiry into how the nature of trust at a small scale might translate into the capacity to structure larger-scale economic organizations one way or another and in particular whether it is true that societies that place a strong emphasis on family become thereby unable to construct large, private, professionally managed firms. The issue is whether and how the details of trust in small face-to- face groups provide the foundation for understanding the significance and extent of trust at more macro social levels. This issue arises as well in the work of economists who attribute trust at the macro level to culturally influenced family decisions whether to inculcate “civic” behavior into their children. I suggest that such conceptions overly privilege the micro level of analysis and that we need more detail to explain how trust at a small-scale level may aggregate up to a larger-scale level of analysis. That is, we need to understand the relationship between trust relations among individuals and in small communities and those in larger-scale networks of interaction. This question has attracted little attention.
In a related argument, commenting on community mobilization against the threat of urban “renewal” in mid-twentieth-century Boston, I proposed that local social network structure could make a big difference in whether leaders emerged whom people trusted at a larger scale. “Trust” in this context meant being willing to commit one’s time and resources to an organization run by people whose efforts one assumed would not be self-regarding but rather who would look out for community welfare. This seems consistent with the idea of trust as acting on the premise that another will not harm your interests in a situation where he could. In particular, I argued that communities with predominantly strong ties would tend to generate networks that were fragmented into closed cliques, and that the resulting problem for community organization is that
whether a person trusts a given leader depends heavily on whether there exist intermediary personal contacts who can, from their own knowledge, assure him that the leader is trustworthy, and who can, if necessary, intercede with the leader or his lieutenants on his behalf. Trust in leaders is integrally related to the capacity to predict and affect their behavior. Leaders, for their part, have little motivation to be responsive or even trustworthy toward those with whom they have no direct or indirect connection. Thus, network fragmentation, by reducing drastically the number of paths from any leader to his potential followers, would inhibit trust in such leaders. (Granovetter 1973: 1374)
This discussion of whether we can trust leaders whom we do not know personally could not be conducted at all if we defined trust as “encapsulated interest” in which one can only trust another whom one knows personally very well. But my discussion of trust in organizational leaders suggests that it is meaningful to talk about whether one trusts even individuals one does not know personally, since such individuals are entirely capable of harming your interests, whether they are aware of you or not. So my argument here is that you may trust that potential leader if there is a link or short chain of personal links to that person that conveys enough information to afford you some confidence that she will act in a trustworthy manner—e.g., will really have the interests of the community at heart and will not simply use the organization as a springboard for higher political office or as a source of funds for her country club membership or luxurious vacation. Because you have to decide whether to commit your own energies and resources to such an organization, you need to know this and can make reasonable decisions about it even though you do not know whether the prospective leader has encapsulated your own individual, personal interests—which may be impossible if there is no personal relationship to her.
The critical point here is that a little trust goes a long way: if people can trust those who are vouchsafed indirectly, then the size of structures in which trust matters expands far beyond what would be possible if only direct ties could be effective. This is why Fukuyama’s observation that some societies’ cultures are more family-oriented than others is not decisive for the structure of industrial organization. Indeed, one of the great surprises in the recent economic literature on the ownership and control of firms around the world is that the role of families has not declined nearly as significantly as mid-twentieth-century modernization theory led us to expect. It turns out that in much of the world, even most large firms are controlled by families (see LaPorta et al. 1999), and more than one-third even of the Standard and Poor’s 500 leading American industrials are “family firms” and are, by some accounts, better performers than the nonfamily firms on this list (Anderson and Reeb 2003).8
One way that families can succeed in dominating large economic networks is when they understand the need to locate trust relations strategically in a network of economic relations that may be large and complex. We see this especially clearly in the organization of large business groups (see the chapter on this subject in my sequel volume for more details). A particularly interesting case, given Fukuyama’s depiction of Chinese culture as incapable of supporting large, professionally managed firms, are Chinese business groups or “conglomerates” (as they are often called). The expansion of small family firms into large conglomerates seems common to Taiwan, Hong Kong (before 1997), various East Asian countries where ethnic Chinese business is important, and even the heartland of mainland China itself (see Keister 2000).
A representative account is offered by Kiong (1991) for Singapore, which is ethnically about three-quarters Chinese. While early Chinese entrepreneurs were in the traditional small-firm sectors of service, retail, and import/export, they gradually expanded into manufacturing, banking, and extractive industries such as rubber. The typical evolution was that an original family firm expanded not by getting larger but by setting up branches as independent companies or by buying already-established businesses. Authority, however, remained highly centralized across the component companies. Reputation and personal trustworthiness are crucial, contracts unimportant (182). Complex strategies are used to ensure the family’s control over larger numbers of legally separate firms. Nominee and trustee companies are set up to hold the family’s interests, and the structure of cross-stockholding can be very complicated. Although the number of outsiders employed exceeds that of family members, “family member and kin are put in charge of subsidiary companies” (188). Generally, family members sit on all the boards. Professional management is achieved in part by educating family members in such skills, often abroad, and also by hiring nonfamily professionals who, however, do not exercise broad control comparable to that of family members. These business groups can be very large and diversified, but control is maintained through pyramids—family firms that control other firms that control still other firms, etc.—and dense interlocking directorates.
Thus, family members who have strong trust relations with the central family group are strategically sprinkled through the many holdings in such a way as to knit the entire structure together. Employees who are not in direct touch with the core family members may nevertheless trust the motives of that group through their direct ties to the local family representatives and work harder and more effectively than if they had no commitment to the central group, and conversely, local family members can assure the central family group of the loyalty of top subsidiary employees who are nonkin. Sometimes friends of the families are called upon as investors to help accumulate the capital necessary for expansion, but the resulting networks of cooperation do not dilute the control of the families, as it is generally understood that the outside investors will be more or less “silent partners” (cf. Hamilton 2000 whose data are drawn mainly from Taiwan, Hong Kong, and Thailand). (For a detailed argument about the different types and levels of trust activated in expanding concentric circles of Chinese management, see Luo 2011).
Chung (2000) provides a detailed analysis of Taiwanese business groups. He shows that, contrary to arguments about Chinese culture inhibiting substantial-sized economic structures, the size of Taiwanese groups grew linearly during the period of study, from the 1970s to the 1990s, and by 1996 the top 113 groups contributed 45 percent of the GNP, nearly double the proportion in the 1970s (14). Closely analyzing the structure of ownership, stockholding, and leadership using social network data analysis techniques, Chung found that cohesion in these groups resulted from the same set of core leaders, typically family members such as the sons, brothers, and nephews of a founder, occupying duplicate leadership positions in various group firms. Decisions are based on the “social relationships existing among members of the inner circle. The composition of the core leaders and the way that they relate to each other are the keys to understand the management practices within Taiwan’s business groups” (82). While the proportion of key employees with professional training increased, this did not imply dilution of the importance of families. On the contrary, in 1994, 42 percent of sons had graduate degrees, a higher proportion than that of long-term employees. “In other words, sons, who are expected to succeed the founders’ enterprises, are the most ‘professionalized’ among all patrilineal core leaders” (92)
Placing trusted family members strategically across the firms in family- dominated business groups provides a way to leverage dyadic trust in such a way as to create large and viable economic structures. Do ties with relations of trust integrate even larger structures and perhaps entire national economies? This could not be plausible if very large numbers of ties that connect cohesive clique-like structures were required to create overall connectivity. But as Watts and Strogatz showed in a highly influential 1998 paper in Nature, a surprisingly small number of such connecting ties, even when inserted into a network at random, may dramatically decrease path length in the network of economic units; arguably, when such ties are placed strategically rather than randomly, the effect may be even greater. These arguments about “small worlds” are taken up more systematically in Chapter 4 on power.
I have proposed that ties featuring trust may be scattered across a large social structure in ways that make them more important than if we thought of trust as mattering only at a small-scale and localized level. The weak point of this argument is that the ties I have described, important as they might be, are more than ties of trust. In fact, the empirical accounts on which I draw are mostly not oriented to discussions of trust and emphasize other aspects such as power differentials, norms and values, the search for strategic leverage, or just information exchange. A drawback in my organization of this book into separate chapters on trust, norms, and power is that most real economic phenomena encompass more than one of these features in important ways that must be combined for fuller understanding. The ties that integrate large economic structures are good examples of this, and for that reason I return to them in subsequent chapters. In the sequel volume, organized around particular economic settings and cases, I will feel freer to pull together all the relevant theoretical arguments at one time.
Nevertheless, I argue that this discussion still properly belongs in a chapter on trust because trust is a critical feature of the ties in question, which cannot be well understood without taking it into account. It is not an accident that so many of the ties that integrate large economic structures around the world are those of kinship and that people go to considerable lengths to preserve the family business form against the typical judgment of economists and the business press, drawing on neoclassical economic arguments and mid-twentieth-century modernization theory, that families are a drag on economic development and efficiency (for a dissenting voice, see historian Harold James [2006]). Certainly part of what drives this persistence is the larger element of trust found in families than among unrelated individuals. This is not to romanticize family ties, which are often fraught with difficulty. The literature on the Chinese family, for example, often features discussions of normative obligation and power relations. Hamilton (2000), among other observers, stresses the great importance of patriarchal authority as a force in holding together large structures of Chinese economic organization and emphasizes the importance of power relations within the family. Despite this, it is hard to imagine that trust is not a necessary part of this story, and in Chapter 4, I will talk more about how trust and power are related.9
