Conference on Second
Generation Reform Home Page

Conference Agenda

© International Monetary Fund. All rights reserved. For requests for permission to reproduce articles, see Rights and Permissions

Social Capital and Civil Society
Francis Fukuyama
The Institute of Public Policy
George Mason University
October 1, 1999

Prepared for delivery at the IMF Conference on Second Generation Reforms


  1. What is Social Capital?

  2. What Functions Does Social Capital Play in a Free-Market Liberal Democracy?

  3. How Do We Measure Social Capital?

  4. Where Does Social Capital Come From?

  5. How Can We Increase the Stock of Social Capital?


Social capital is important to the efficient functioning of modern economies, and is the sine qua non of stable liberal democracy. It constitutes the cultural component of modern societies, which in other respects have been organized since the Enlightenment on the basis of formal institutions, the rule of law, and rationality. Building social capital has typically been seen as a task for "second generation" economic reform; but unlike economic policies or even economic institutions, social capital cannot be so easily created or shaped by public policy. This paper will define social capital, explore its economic and political functions, as well as its origins, and make some suggestions for how it can be cultivated.

I.  What is Social Capital?

While social capital has been given a number of different definitions, many of them refer to manifestations of social capital rather than to social capital itself. The definition I will use in this paper is: social capital is an instantiated informal norm that promotes cooperation between two or more individuals. The norms that constitute social capital can range from a norm of reciprocity between two friends, all the way up to complex and elaborately articulated doctrines like Christianity or Confucianism. They must be instantiated in an actual human relationship: the norm of reciprocity exists in potentia in my dealings with all people, but is actualized only in my dealings with my friends. By this definition, trust, networks, civil society, and the like which have been associated with social capital are all epiphenominal, arising as a result of social capital but not constituting social capital itself.

Not just any set of instantiated norms constitutes social capital; they must lead to cooperation in groups and therefore are related to traditional virtues like honesty, the keeping of commitments, reliable performance of duties, reciprocity, and the like. A norm like the one described by Edward Banfield as characterizing southern Italy, which enjoins individuals to trust members of their immediate nuclear family but to take advantage of everyone else, is clearly not the basis of social capital outside the family.1

James Coleman, who was responsible for bringing the term social capital into wider use in recent years, once argued that it was a public good and therefore would be underproduced by private agents interacting in markets.2 This is clearly wrong: since cooperation is necessary to virtually all individuals as a means of achieving their selfish ends, it stands to reason that they will produce it as a private good (see Section IV below). In Partha Dasgupta's phrase, social capital is a private good that is nonetheless pervaded by externalities, both positive and negative.3 An example of a positive externality is Puritanism's injunction, described by Max Weber, to treat all people morally, and not just members of the sib or family.4 The potential for cooperation thus spreads beyond the immediate group of people sharing Puritan norms. Negative externalities abound, as well. Many groups achieve internal cohesion at the expense of outsiders, who can be treated with suspicion, hostility, or outright hatred. Both the Ku Klux Klan and the Mafia achieve cooperative ends on the basis of shared norms, and therefore have social capital, but they also produce abundant negative externalities for the larger society in which they are embedded.

It is sometimes argued that social capital differs from other forms of capital because it leads to bad results like hate groups or inbred bureaucracies. This does not disqualify it as a form of capital; physical capital can take the form of assault rifles or tasteless entertainment, while human capital can be used to devise new ways of torturing people. Since societies have laws to prevent the production of many social "bads," we can presume that most legal forms of social capital are no less "goods" than the other forms of capital insofar as they help people achieve their aims.

Perhaps the reason that that social capital seems less obviously a social good than physical or human capital is because it tends to produce more in the way of negative externalities than either of the other two forms. This is because group solidarity in human communities is often purchased at the price of hostility towards out-group members. There appears to be a natural human proclivity for dividing the world into friends and enemies that is the basis of all politics.5 It is thus very important when measuring social capital to consider its true utility net of its externalities.

Another way of approaching this question is through the concept of the "radius of trust."6 All groups embodying social capital have a certain radius of trust, that is, the circle of people among whom cooperative norms are operative. If a group's social capital produces positive externalities, the radius of trust can be larger than the group itself. It is also possible for the radius of trust to be smaller than the membership of the group, as in large organiza-tions that foster cooperative norms only among the group's leadership or permanent staff. A modern society may be thought of as a series of concentric and overlapping radii of trust (see Figure I). These can range from friends and cliques up through NGOs and religious groups.

Virtually all forms of traditional culture-social groups like tribes, clans, village associations, religious sects, etc.-are based on shared norms and use these norms to achieve cooperative ends. The literature on development has not, as a general rule, found social capital in this form to be an asset; it is much more typically regarded as a liability. Economic modernization was seen as antithetical to traditional culture and social organizations, and would either wipe them away or else be itself blocked by forces of traditionalism. Why should this be so, if social capital is genuinely a form of capital?

The reason, in my view, has to do with the fact that such groups have a narrow radius of trust. In-group solidarity reduces the ability of group members to cooperate with outsiders, and often imposes negative externalities on the latter. For example, in the Chinese parts of East Asia and much of Latin America, social capital resides largely in families and a rather narrow circle of personal friends.7 It is difficult for people to trust those outside of these narrow circles. Strangers fall into a different category than kin; a lower standard of moral behavior applies when one becomes, for example, a public officials. This provides cultural reinforcement for corruption: in such societies, one feels entitled to steal on behalf of one's family.

Traditional social groups are also afflicted with an absence of what Mark Granovetter calls "weak ties,"8 that is, heterodox individuals at the periphery of the society's various social networks who are able to move between groups and thereby become bearers of new ideas and information. Traditional societies are often segmentary, that is, they are composed of a large number of identical, self-contained social units like villages or tribes. Modern societies, by contrast, consist of a large number of overlapping social groups that permit multiple memberships and identities. Traditional societies have fewer opportunities for weak ties among the segments that make it up, and therefore pass on information, innovation, and human resources less easily.

II.  What Functions Does Social Capital Play in a Free-Market Liberal Democracy?

The economic function of social capital is to reduce the transaction costs associated with formal coordination mechanisms like contracts, hierarchies, bureaucratic rules, and the like. It is of course possible to achieve coordinated action among a group of people possessing no social capital, but this would presumably entail additional transaction costs of monitoring, negotiating, litigating, and enforcing formal agreements. No contract can possibly specify every contingency that may arise between the parties; most presuppose a certain amount of goodwill that prevents the parties from taking advantage of unforeseen loopholes. Contracts that do seek to try to specify all contingencies-like the job-control labor pacts negotiated in the auto industry that were as thick as telephone books-end up being very inflexible and costly to enforce.

There was a period when social scientists assumed that modernization necessarily entailed the progressive replacement of informal coordination mechanisms with formal ones. There was presumably a period in human history in which formal law and organizations scarcely existed, and in which social capital was the only means of achieving coordinated action; Max Weber argued that, by contrast, rational bureaucracy constituted the essence of modernity.

The fact of the matter is that coordination based on informal norms remains an important part of modern economies, and arguably becomes more important as the nature of economic activity becomes more complex and technologically sophisticated.9 Many complex services are very costly to monitor and are better controlled through internalized professional standards than through formal monitoring mechanisms. A highly educated software engineer often knows much more about his or her own productivity than his or her supervisor; procurement is often more efficient when left to the judgment of an experienced procurement officer, rather than being done "by the book" as in the case of a good deal of government procurement. A number of empirical studies suggest that high-tech R&D is often dependent on the informal exchange of intellectual property rights, simply because formal exchange would entail excessive transaction costs and slow down the speed of interchange.10

Even in non-hi-tech environments, social capital often leads to greater efficiency than purely formal coordination techniques. Classical Taylorism, which organized workplaces in a highly centralized, bureaucratized manner, created many inefficiencies as decisions were delayed and information distorted while moving up and down hierarchical chains of command. In many manufacturing facilities, Taylorism has been replaced by much flatter management structures which in effect push responsibility down to the factory floor itself. Workers who are much closer to the sources of local knowledge are authorized to make decisions on their own, rather than referring them up a managerial hierarchy. This often leads to great gains in efficiency,11 but is totally dependent on the social capital of the workforce. If there is distrust between workers and managers, or widespread opportunism, then the delegation of authority required in a typical "lean" manufacturing system will lead to instant paralysis. This is in effect what happened to General Motors during the strikes of 1996 and 1998, when a single dissident local (angry, in the first instance, over the outsourcing of brake parts) was able to shut down the company's entire North American operations.12

The political function of social capital in a modern democracy was best elucidated by Alexis de Tocqueville in Democracy in America, who used the phrase the "art of association" to describe Americans' propensity for civil association. According to Tocqueville, a modern democracy tends to wipe away most forms of social class or inherited status that bind people together in aristocratic societies. Men are left equally free, but weak in their equality since they are born with no conventional attachments. The vice of modern democracy is to promote excessive individualism, that is, a preoccupation with one's private life and family, and an unwillingness to engage in public affairs. Americans combated this tendency towards excessive individualism by their propensity for voluntary association, which led them to form groups both trivial and important for all aspects of their lives. This stood in sharp contrast to his native France, which was beset by a much more thoroughgoing individualism than the United States. As Tocqueville explained in The Old Regime and the French Revolution, on the eve of the Revolution "there were not ten Frenchmen who could come together for a common cause." It was only by coming together in civil associations that weak individuals became strong; the associations they formed could either participate directly in political life (as in the case of a political party or interest group) or could serve as "schools of citizenship" where individuals learned the habits of cooperation that would eventually carry over into public life.

An abundant stock of social capital is presumably what produces a dense civil society, which in turn has been almost universally seen as a necessary condition for modern liberal democracy (in Ernest Gellner's phrase, "no civil society, no democracy").13 If a democracy is in fact liberal, it maintains a protected sphere of individual liberty where the state is constrained from interfering. If such a political system is not to degenerate into anarchy, the society that subsists in that protected sphere must be capable of organizing itself. Civil society serves to balance the power of the state and to protect individuals from the state's power.

In the absence of civil society, the state often needs to step in to organize individuals who are incapable of organizing themselves. The result of excessive individualism is therefore not freedom, but rather the tyranny of what Tocqueville saw as a large and benevolent state that hovered over society and, like a father, saw to all of its needs. Low levels of social capital lead to a number of political dysfunctions, which have been extensively documented. Following Tocqueville's analysis of France, many observers have noted how administrative centralization has led to an excessively rigid and unresponsive political system, one that can be changed only through anti-systemic upsurges such as the évenements of 1968.14 Low levels of social capital have been linked to inefficient local government in southern Italy, as well as to the region's pervasive corruption.15 In many Latin American societies, a narrow radius of trust produces a two-tier moral system, with good behavior reserved for family and personal friends, and a decidedly lower standard of behavior in the public sphere. This serves as a cultural foundation for corruption, which is often regarded as a legitimate way of looking after one's family.

It is of course also possible to have too much of a good thing. One person's civic engagement is another's rent-seeking; much of what constitutes civil society can be described as interest groups trying to divert public resources to their favored causes, whether sugar-beet farming, women's health care, or the protection of biodiversity. The public choice literature has analyzed the baleful consequences of rent-seeking for modern democracies at great length; Mancur Olson has argued that Britain's long-term economic decline was due to the long-term buildup of entrenched interest groups there.16 There is no guarantee that self-styled public interest NGOs actually represent real public interests. It is entirely possible that too active an NGO sector may represent an excessive politicization of public life, which can either distort public policy or lead to deadlock.17

Despite the possibility that a society has have too much social capital, it is doubtless worse to have too little. For in addition to being a source of spontaneously-organized groups, social capital is vital to the proper functioning of formal public institutions. It is sometimes argued that it is more useful to compare societies in institutional rather than cultural terms. Chalmers Johnson, for example, argues that differences in Japanese and American economic policy is not culturally based, but simply the result of the fact that Japan had MITI and the United States did not.18 The implication is that were the US to create an equivalent of MITI in Washington, it would have similar consequences. But there are any number of reasons for thinking that different societies have different cultural capacities for institution-building. Japan's deployment of an economic planning agency with enormous power over credit allocation did not lead to the same levels of rent-seeking and outright corruption that comparable agencies have brought about in Latin America or Africa (or indeed the US, were it to follow Japan's example). This is testimony to a number of Japanese cultural characteristics: the respect given bureaucrats, their high level of training and professionalism, the general deference to authority in Japanese society, etc., and it suggests that some institutions cannot be readily transferred to other societies lacking social capital.

III.  How Do We Measure Social Capital?

One of the greatest weaknesses of the social capital concept is the absence of consensus on how to measure it. At least two broad approaches have been taken: the first, to conduct a census of groups and group memberships in a given society, and the second, to use survey data on levels of trust and civic engagement. At the end of this section, I will suggest a third metric that may point to a measure of social capital within private firms. Robert Putnam has tried to measure social capital by counting groups in civil society, using a number n to track size of memberships in sports clubs, bowling leagues, literary societies, political clubs, and the like as they vary over time and across different geographical regions. There are, in fact, a large number of n's in any given society, n1..t. Hence the first measure for the total social capital (SC) in a society is the sum of the membership of all groups,

   (1)   SC = n1..t.

Both n and t are important measures of civil society. A small value for n may limit the kinds of ends a group can achieve; families, for example, are good at socializing children and running family restaurants, but not very good at exerting political influence or manufacturing semiconductors. The variable t itself constitutes a separate of measure of civil society; unfortunately, limitations in the data prohibit our knowing what t is for a given society, or how many missing or undercounted data elements there are between n1 and nt. A number of attempts have been made to produce censuses of groups and associations in the United States. One was done by the US Department of Commerce in 1949, which estimated that there were 201,000 nonprofit voluntary trade and business organizations, women's groups, labor unions, civic service groups, luncheon clubs, and professional groups at all levels of American society.19 Lester Salamon estimates that by 1989 there were 1.14 million nonprofits in the US, indicating an overall rate of growth much higher than that of the population as a whole.20 The near-impossibility of producing a complete census that catalogues the whole range of informal networks and cliques in a modern society is suggested by the Yankee City study, which counted some 22,000 different groups in a community of 17,000 people.21 Changing technology changes forms of association: how do we account for the proliferation of on-line discussion groups, chat rooms, and e-mail conversations that have exploded with the spread of personal computers in the 1990s?22

N and t may also be inversely correlated (that is, the larger the average size of groups, the fewer there are); on the other hand, because individuals can hold overlapping memberships in multiple groups, they need not be.

It is clear that each of these n1..t groups is characterized by a different level of internal cohesion and therefore collective action. Bowling leagues are not capable of storming beaches or lobbying Congress, so some qualitative coefficient must be added to provide some measure of cohesion. Let us call this coefficient c. Unfortunately, there is no accepted method for measuring the internal cohesiveness of groups; each one of the c coefficients would have to be determined subjectively by an outside observer who would note the types of activities the group could undertake and their difficulty, its cohesion under stressful circumstances, and other factors. Despite the subjective nature of its derivation, it is clear that c varies across groups and is a critical qualitative measure of social capital. Hence a society's total stock of social capital would be expressed as

   (2)   SC = (cn)1..t.

As noted above, social capital is more heavily pervaded by externalities than other forms of capital, so measurement of a nation's stock of social capital must take these externalities into account. The radius of trust can be thought of as a type of positive externality (which we will therefore designate as rp) because it is a benefit that accrues to the group independently of the collective action that the group formally seeks to achieve. For example, a sect that encourages its members to be honest and reliable will foster better business relationships when they deal with each other economically, in addition to the sect's religious objectives.

For many groups, the radius of trust would extend to the whole group; this is true of most families, for example. The rp coefficient in this case is 1, and the total amount of social capital in the society would therefore be expressed as

   (3)   SC = (rpcn)1..t.

Certain groups, particularly larger ones, are characterized by internal hierarchy, a division of labor, status and functional distinctions, etc. While the group may be united around some common interest or passion, the degree to which individual members are capable of collective action on the basis of mutual trust depends on their relative position within the organization. Putnam rightly distinguishes between what he calls a "membership organization" like the American Association of Retired People (AARP), which, at more than 33 million members, is second only to the Catholic Church in size. Such a group has a very large n-value, but most of its members simply contribute yearly dues, receive a newsletter, and would have little reason for cooperating with one another on any issue not related to pensions or health benefits. For such an organization, the rp coefficient may be very small, limited to (for example) those people who work full-time in its national headquarters (though even there, there are presumably many employees who are simply wage-earners and not part of the trust network).

On the other hand, it is possible for a group to have an rp coefficient larger than 1. To take the earlier example of the religious sect that encourages honesty and reliability, if these traits are demanded of its members in their dealings not just with other members of the sect but generally in their dealings with other people, then there will be a positive spillover effect into the larger society. Again, Weber argued in effect that sectarian Puritans had an rp value greater than 1.

The final factor affecting a society's supply of social capital concerns not the internal cohesiveness of groups, but rather the way in which they relate to outsiders. Strong moral bonds within a group in some cases may actually serve to decrease the degree to which members of that group are able to trust outsiders and work effectively with them. A highly disciplined, well-organized group sharing strong common values may be capable of highly coordinated collective action, and yet may nonetheless be a social liability. I earlier noted the fact that strongly familistic societies like China and central-southern Italy were characterized by an absence of a broader, generalized social trust outside the family. At best, this prevents the group from receiving beneficial influences from the outside environment; at worst, it may actively breed distrust, intolerance, or even hatred for and violence toward outsiders. Certain groups may be actively harmful to other parts of society-criminal organizations like the Mafia or the Crips and Bloods come to mind. A society made up of the Ku Klux Klan, the Nation of Islam, the Michigan Militia, and various self-regarding ethnic and racial organizations may score very high in terms of last three of the four variables given in expression (3), and each group may have an rp of 1, and yet overall it would be hard to say that such a society had a large stock of social capital.

Group affiliation can therefore produce a negative externality which we can think of as the radius of distrust, or rn. The larger the rn value, the greater the liability that group represents to the surrounding society; hence the measure for a single group's social capital, rpcn, needs to be multiplied by the reciprocal of rn. (All rn values, we assume, must be 1 or greater.) The final value for a society's total stock of social capital would then be:

   (1.4)   SC = ((1/rn)rpcn)1..t.

To some extent, we could expect that c and rn might be positively correlated with one another. That is, internal cohesiveness is often based on strongly shared norms and values within a group: both the Marines and the Mormon Church are examples. But the very strength of those internal bonds creates something of a gulf between members of the group and those on the outside. Latitudinarian organizations like most contemporary mainline Protestant denominations in the United States, by contrast, easily coexist with other groups in the society, and yet are capable of a much lower level of collective action. Ideally, one would like to maximize the c and and minimize the rn values: such would be the case, for example, in a professional organization that socializes its members into the values of its particular profession, while not at the same time breeding distrust of other professions or being closed to influences from them.

As this exercise indicates, producing anything like a believable census of a society's stock of social capital is a nearly impossible task, since it involves multiplying numbers that are either subjectively estimated or simply nonexistent. This leads us to the other source of data that has been used as a proxy for social capital, survey data on trust and civic engagement. There are a number of data sources that are useful here, such as the National Opinion Research Council's General Social Survey (for the US) and the University of Michigan's World Values Survey (for international data). Each of these surveys asks a series of question concerning trust in various political and social institutions, as well as others probing the respondents' level of participation in voluntary organizations. There are manifold problems with survey data, of course, beginning with the fact that responses will vary according to the way the question is phrased and who is asking it, to the absence of consistent data for many countries and many time periods. A general question such as "Generally speaking, would you say that most people can be trusted or that you can't be too careful in dealing with people?" (asked on both the General Social Survey and World Values Survey) won't give you very much precise information about the radius of trust among the respondents, or their relative propensities to cooperate with family, co-ethnics, co-religionists, complete strangers, and the like.

A third possible way of measuring social capital in specific organizations may be to look at changes in market valuations of a company before and after takeover offers. The market capitalization of any company represents the sum of both tangible and intangible assets; among the latter is, presumably, the social capital embodied in the firm's workers and management. There is no accepted methodology for separating out the social capital component of the intangible assets, which include other things like brand names, good will, expectations of future market conditions, and the like. Firms being taken over by other firms, however, are usually bought at a premium to their pre-takeover price. In such a situation, we can assume that part of the premium being offered is a measure of the degree to which the new owners believe that they can manage the new firm better than the old owners, with all other factors like tangible assets, expectations about market conditions, etc. being held constant. In many cases, part of the premium being offered represents the cost savings that the new owners expect to achieve through realization of economies of scale and scope; one would have to deduct this from the actual premium to get a measure of the net value of the new management alone. This management premium is not a pure measure of social capital; it may consist partly of human capital rather than social capital. But social capital must constitute a significant part of the residual, since effective management is, after all, nothing more than efficient coordination of the firm's activities.23

IV.  Where Does Social Capital Come From?

If we define social capital as instantiated, informal norms that produce cooperation, economists have a straightforward explanation of where it comes from: social capital arises spontaneously as a product of iterated Prisoners Dilemma (PD) games. A one-shot PD game does not lead to a cooperative outcome because defection constitutes a Nash equilibrium for both players; if the game is iterated, however, a simple strategy like tit-for-tat (playing cooperation for cooperation and defection for defection) leads both players to a cooperative outcome. In non-game theoretic turns, if individuals interact with each other repeatedly over time, they develop a stake in a reputation for honesty and reliability. Market interactions in a commercial society leads, as Adam Smith observed, to the development of bourgeois social virtues like honesty, industriousness, and prudence.24 A society composed entirely of Kant's "rational devils" will develop social capital over time, simply as a matter of the devils' long-term self-interest.

Social capital is clearly spontaneously generated all the time through the playing of iterated PD games. Both Robert Ellickson25 and Elinor Ostrom26 have catalogued many empirical cases of cooperative norms arising as a result of repeated community interaction. The latter's database of instances in which communities have successfully dealt with common pool resource problems is particularly interesting because this class of problems constitutes an n-sided PD game which should theoretically be much harder to solve through iteration than a two-player game.

The economists' approach to understanding how social capital is generated is ultimately very limited, however. The problem is that social capital more often than not is produced by hierarchical sources of authority, which lay down norms and expect obedience to them for totally a-rational reasons. The world's major religions like Buddhism, Hinduism, Christianity, or Islam, or large cultural systems like Confucianism, are examples. Not only do norms from such sources not come about through decentralized bargaining; they are transmitted from one generation to the next through a process of socialization that involves much more habit than reason. Path dependence-another word for tradition-means that norms that are clearly socially suboptimal can persist for very long periods of time.

It is, of course, possible to try to give economic or rational explanations for religious and cultural phenomena, and thus to try to fit them into some larger theory of social behavior based on rational choice. There was for some time a school of "functionalist" sociology and anthropology that tried to find rational utilitarian reasons for the most bizarre social rules. The Hindu ban on eating cows was ascribed, for example, to the fact that cows were resources that had to be protected for other uses like plowing and dairy-farming. Similarly, one could try to explain the Protestant Reformation in terms of the economic conditions prevailing in central Europe in the sixteenth century that led people to respond to religious reformers like Luther, Calvin, and Melanchthon. But ultimately, these accounts prove to be unsatisfying because they are too reductionist; all such historical developments usually incorporate a substantial measure of chance, genius, accident, or creativity that cannot be explained in terms of prior conditions. Max Weber stood Marx on his head by arguing that the cultural "superstructure" actually produced the economic "substructure": it was the moral values inculcated by Puritanism, and particularly the fact that virtues like honesty and reciprocity now had to be practiced beyond the family, that made the modern capitalist world possible in the first place. In Weber's account, culture was the uncaused cause, the product of "charisma."

Religion continues to be a factor in economic development. One of the most important and underrated cultural revolutions going on in the world today is the conversion of Catholics to Protestantism by (largely) American evangelicals and Mormons. This process, which has now been under empirical observation for nearly two generations, has produced social effects in the poor communities where it has occurred not unlike those ascribed to Puritanism by Weber: converts to Protestantism find their incomes, education levels, hygiene, and social networks expanding.27

Apart from religion, shared historical experience can shape informal norms and produce social capital. Both Germany and Japan experienced considerable labor unrest and conflict between workers, managers, and the state in the 1920s and 30s. The Nazis and Japan's military rulers ultimately suppressed independent labor unions and replaced them with "yellow" ones. After their defeat in World War II, the democratic successor regimes opted for a much more consensual approach to management-labor relations that produced Germany's postwar Sozialmarktwirtschaft and Japan's lifetime employment system. Whatever their current dysfunctions, these institutions played a critical role in allowing the two societies to return to growth after the war, and constituted a form of social capital.

V.  How Can We Increase the Stock of Social Capital?

The discussion of where social capital comes from should be informative to policymakers who want to increase the stock of social capital in a given country. States can both do some positive things to create social capital, and forebear from doing others that deplete a society's stock. We can make four observations.

First, states do not have many obvious levers for creating many forms of social capital. Social capital is frequently a byproduct of religion, tradition, shared historical experience, and other factors that lie outside the control of any government. Public policy can be aware of already existing forms of social capital-for example, the social networks used to develop information for microlending-but it cannot duplicate the effect of religion as a source of shared values. Policymakers also need to be aware that social capital, particularly when associated with groups that have a narrow radius of trust-can produce negative externalities and be detrimental to the larger society.

Second, the area where governments probably have the greatest direct ability to generate social capital is education. Educational institutions do not simply transmit human capital, they also pass on social capital in the form of social rules and norms. This is true not just in primary and secondary education, but in higher and professional education as well. Doctors learn not just medicine but the Hippocratic oath; one of the greatest safeguards against corruption is to give senior bureaucrats high-quality professional training and to create an esprit de corps among this elite.

Third, states indirectly foster the creation of social capital by efficiently providing necessary public goods, particularly property rights and public safety. Diego Gambetta has shown that the Sicilian Mafia can be understood as a private protector of property rights in a part of Italy where the state has historically failed to perform this function.28 Something similar to this has sprung up in Russia during the 1990s. Private property rights protection is very inferior to the state-supplied version, since there is nothing to prevent these private providers from getting into a host of other illegal activities as well. There are also economies of scale in the deployment of coercive force used to enforce property rights. People cannot associate, volunteer, vote, or take care of one another if they have to fear for their lives when walking down the street. Given a stable and safe environment for public interaction and property rights, it is more likely that trust will arise spontaneously as a result of iterated interactions of rational individuals.

Fourth, states can have a serious negative impact on social capital when they start to undertake activities that are better left to the private sector or to civil society. The ability to cooperate is based on habit and practice; if the state gets into the business of organizing everything, people will become dependent on it and lose their spontaneous ability to work with one another. France had a rich civil society at the end of the Middle Ages, but horizontal trust between individuals weakened as a result of a centralizing state that set Frenchmen at each other through a system of petty privileges and status distinctions. The same thing occurred in the former Soviet Union after the Bolshevik Revolution, where the Communist Party consciously sought to undermine all forms of horizontal association in favor of vertical ties between Party-State and individual. This has left post-Soviet society bereft of both trust and a durable civil society. There are, of course, good reasons why countries should restrict the size of their state sectors for economic reasons. On top of this, one can add a cultural motive of preserving a sphere for individual action and initiative in building civil associations.

If we look beyond the role of the state, there remain at least two additional sources of social capital. The first is religion. General social science theories about the inevitability of secularization appear to apply primarily to Western Europe; there is little evidence that religion is losing its grip elsewhere, including the United States.29 Religiously-inspired cultural change remains a live option in many parts of the world; the Islamic world and Latin America have both seen the growth of new forms of religiosity in recent decades. Obviously, not all forms of religion are positive from the standpoint of social capital; sectarianism can breed intolerance, hatred, and violence. But religion has also historically been one of the most important sources of culture, and is likely to remain so in the future.

The second source of social capital in developing countries is globalization. Globalization has been the bearer not just of capital but of ideas and culture as well. Everyone is well aware of the ways in which globalization injures indigenous cultures and threatens longstanding traditions. But it also leaves new ideas, habits, and practices in its wake, from accounting standards to management practices to NGO activities. It is not just investment bankers who can take advantage of the global communications and information revolution; activists of all sorts from environmentalists to labor organizers can now operate transnationally to a much greater extent than before.30 The issue, for most societies, is whether they are net losers or gainers from this process, that is, whether globalization breaks down traditional cultural communities without leaving anything positive in its wake, or rather, is an external shock that breaks apart dysfunctional traditionals and social groups and becomes the entering wedge for modernity.

1Edward Banfield, The Moral Basis of a Backward Society (Glencoe, IL: Free Press, 1958).
2James S. Coleman, "Social Capital in the Creation of Human Capital," American Journal of Sociology Supplement 94 (1988): S95-S120.
3Partha Dasgupta, "Economic Development and the Idea of Social Capital," unpublished paper, March 1997.
4According to Weber, "The great achievement of ethical religions, above all of the ethical and asceticist sects of Protestantism, was to shatter the fetters of the sib." The Religion of China (New York: Free Press, 1951), p. 237.
5See Francis Fukuyama, Trust: The Social Virtues and the Creation of Prosperity (New York: Free Press, 1995), chapter 9.
6To my knowledge, the first person to use this term was Lawrence Harrison in Underdevelopment is a State of Mind: The Latin American Case (New York: Madison Books, 1985), pp. 7-8.
7Fukuyama (1999).
8Mark S. Granovetter, "The Strength of Weak Ties," American Journal of Sociology 78 (1973): 1360-80.
9For a fuller treatment of this issue, see Francis Fukuyama, The Great Disruption: Human Nature and the Reconstitution of Social Order (New York: Free Press, 1999), chapter 12.
10See, for example, Annalee Saxenian, Regional Advantage: Culture and Competition in Silicon Valley and Route 128 (Cambridge, MA: Harvard University Press, 1994), which gives numerous examples of informal intellectual property exchange in Silicon Valley.
11See James P. Womack and D. Jones, The Machine that Changed the World: The Story of Lean Production (New York: Harper Perennial, 1991).
12See Fukuyama (1999), chapter 13.
13Ernest Gellner in Conditions of Liberty: Civil Society and Its Rivals (London: Hamish Hamilton, 1994).
14See, for example, Michel Crozier, The Bureaucratic Phenomenon (Chicago, IL: University of Chicago Press, 1964); Stanley Hoffmann, Decline or Renewal? France since the 1930s (New York: Viking Press, 1974); Stanley Hoffmann and Charles Kindleberger, In Search of France (Cambridge, MA: Harvard University Press, 1963).
15 Edward Banfield in The Moral Basis of a Backward Society (Glencoe, IL: Free Press, 1958); Robert D. Putnam, Making Democracy Work : Civic Traditions in Modern Italy (Princeton, NJ: Princeton University Press, 1993).
16Mancur Olson, The Rise and Decline of Nations (New Haven, CT: Yale University Press, 1982).
17Larry Diamond, "Rethinking Civil Society," Journal of Democracy 5 (1994): 5-17.
18Johnson, Chalmers, MITI and the Japanese Miracle (Stanford, Ca.: Stanford University Press, 1982).
19Calvert J. Judkins, National Associations of the United States (Washington, DC: US Dept. of Commerce, 1949). I am very grateful to Marcella Rey for this and other references concerning measures of group memberships.
20Lester M. Salamon, America's Nonprofit Sector (New York: The Foundation Center, 1992).
21W. Lloyd Warner, J. O. Low, Paul S. Lunt, and Leo Srole, Yankee City (New Haven, CT: Yale University Press, 1963).
22Apart from the difficulty in counting the number of such groups, there are a number of complex issues in assessing the quality of the relationships engendered by them. Ladd contests Putnam's dismissal of many new advocacy groups as mere "membership groups." He shows that not only have memberships in large environmental organizations like the Nature Conservancy or the World Wildlife Fund grown substantially, but the quality of the relationships formed among these groups' members goes well beyond writing a yearly dues check. He points to one study that shows how a single local chapter of a single environmental organization sponsored countless hikes, bike trips, backpacking classes, and the like, all of which presumably fostered personal relationships and had spillover effects on social capital.
23The case could probably be made that social capital is the most intangible of all intangible assets, and tends to be consistently undervalued by markets because it is so difficult to measure. Many mergers and acquisitions have involved radical downsizing of company work forces. This achieved cost savings in terms of current wages, but undermined trust and therefore social capital among the firm's remaining workers-a process popularly known as "dumbsizing." Wall Street is obviously better able to measure the immediate labor cost savings than the longer-term impacts of such actions on the firm's social capital.
24Adam Smith, The Theory of Moral Sentiments (Indianapolis, IN: Liberty Classics, 1982), Part One, I.4.7; Part Seven, IV.25
25Robert C. Ellickson, Order Without Law: How Neighbors Settle Disputes (Cambridge, MA: Harvard University Press 1991).
26Elinor Ostrom, Governing the Commons: The Evolution of Institutions for Collective Action (Cambridge: Cambridge University Press, 1990).
27David Martin, Tongues of Fire. The Explosion of Protestantism in Latin America (Oxford: Basil Blackwell, 1990); David Stoll, Is Latin America Turning Protestant? The Politics of Evangelical Growth (Berkeley, CA: University of California Press, 1990).
28Diego Gambetta, The Sicilian Mafia: the business of private protection (Cambridge: Harvard University Press, 1993).
29David Martin, A General Theory of Secularization (Oxford: Blackwell, 1978); Peter L. Berger, "Secularism in Retreat," National Interest no. 46 (1996): 3-12.
30Jessica Matthews, "Power Shift," Foreign Affairs 76 (1997): 50-66.