obligations (Brennan and Buchanan 1980: 172).
Since rational egoists always seek to minimize their costs, why can’t minimally extensive obligations be arrived at in the absence of fully competitive markets? After all, members have an incentive to find ways either to reduce costs (if their initial estimate of the production cost was too high) or to increase costs (if their estimate was too low). Even though they seek to minimize their obligations, there are at least three reasons why members are unable to do so in the absence of frictionless market conditions. In the first place, each collective consideration of the level of obligations entails time and other costs of decision–making (Buchanan and Tullock 1962; Buchanan 1975), and these costs rise geometrically with the size of the membership. Rather than incurring such costs, members will settle for greater than minimal obligations. In the second place, to the degree that members place a high value on continued access to the immanent good, they may be reluctant to risk suboptimal provision of it (this may account for the sanctity of the defense budget in the eyes of American voters). Finally, if there are no alternative sources of the same good, then initial members may levy higher obligations on all subsequent ones and consume the resulting surplus themselves. This will create a two– (or multi–) tiered tax structure within the membership and raise the average extensiveness of obligations for the group as a whole.
The problem is that the assumption of frictionless markets is extremely stringent and unrealistic, and to the degree that it does not hold, the minimal production cost cannot be inferred. Hence, it is more reasonable to regard the dependence of members rather than the cost of production as the key determinant of the extensiveness of group obligations.
Rational egoists choose to belong to a group because they are dependent (Thibaut and Kelley 1959; Emerson 1962; and Blau 1964) on other members for access to some desired joint good. If they could attain this good without incurring the obligations of membership, they would always prefer to do so. Yet their degree of dependence varies widely. Our nomads may have only one source of security ‒ their local protective association. But similar individuals living on the populated plain may have a multitude of security options; on this account they are likely to be less dependent on any one protective association. The more dependent people are, the more tax they must pay for access to the same quantity of a given good. This variable degree of dependence is indicated by the opportunity cost of leaving the group, or what may be termed (after Hirschman 1970) the members’ cost of exit. This is the difference between the value received from membership in the group and the value that is gained from the member’s best alternative, taking into account any costs incidental to the transfer.
To the degree that members face a high cost of exit, they are dependent on that group. As exit costs approach prohibitive levels, dependence on the group increases; ultimately, members may become beholden to it for their very survival. In this extreme situation (analogous to a monopoly) members will accept the most extensive obligations to gain access to a given immanent good. Correlatively, when the cost of exit decreases, members’ dependence on the group diminishes. When the average dependence of members declines significantly, the extensiveness of obligations begins to approach the minimum as specified in the neoclassical analysis.
Ultimately, dependence is affected by environmental shifts, many of which are beyond the control of group members themselves. It is increased by limits on the supply of close substitutes available outside group boundaries, a lack of information about these alternatives, moving costs, and the existence of strong personal ties among members. Let us consider each of these factors separately.
1. The supply of close substitutes. If the number of distinct sources of substitutable goods in the environment is not large, the chance of finding a better alternative generally decreases. Groups may collude and levy the same tax for providing a given joint good to their members. If there are only a small number of groups, this collusion may be easy to organize and enforce. As the number of alternative sources increases, however, the costs of organizing and enforcing collusive agreements grow disproportionately.
2. Lack of information about alternatives. On the one hand, since groups are more or less exclusive, knowledge of their internal workings may be relatively difficult to obtain (Goffman 1959: 77–105). Uncertainty about the relative advantages of alternative groups breeds inertia. On the other hand, many groups are able to restrict information about available alternatives. To the degree that members are unaware of the existence of alternative sources of joint goods outside group boundaries, they will be willing to bear the cost of more extensive rules. If people are unaware of the existence of a better alternative, they can hardly choose to take advantage of it. And information is always costly to gather.
3. Costs of moving. Transfer costs, which must be taken into account in any decision to join or leave a group, are seldom zero. Since moving costs among the nomads are minimal, they can be very responsive to marginal differences in the cost of protection. It is not very costly to pack everything on your camels and move to another part of the valley where security is a better bargain. Were the nomads to opt for a sedentary lifestyle, however, then their moving costs would rise, thereby increasing their dependence on the initial protective association. It is far more costly to move a house than a tent. Beyond this, however, groups sometimes impose entry and exit costs. Entry to the most rudimentary rotating credit association requires character references, which are developed over a lifetime and thus are costly to obtain. Many intentional communities demand that exiting members leave some part of their personal assets with the group. Finally, groups vary widely in their exclusiveness; for example, younger groups are more likely to be open than older ones. To the degree that there are barriers to entry/exit (and to the degree that moving between groups entails costs), this increases the dependence of members.
4. The strength of personal ties. Sociability is one of the most important immanent goods that groups provide. Since personal ties tend to arise with repeated interaction– and thus only in the course of time ‒ they are akin to an irredeemable investment (or sunk cost) in the group. The probability of repeated interaction increases with limitations of supply and with costly information and mobility, lowering the chance that close substitutes can be found outside group boundaries.
If these are some of the factors that create dependence, constitutional and legal arrangements have decisive implications for generating them. For most of human history the number and composition of voluntary associations has been regulated by political authorities. As Adam Smith (1961 [1789]) was at pains to emphasize, without effective constitutional guarantees of individual private property rights and the freedoms of mobility and association, dependence flourishes. It is also strengthened whenever rights are granted to groups, rather than to individuals. This occurs in western feudalism, but the Indian caste system provides perhaps the classic example (Weber 1946 [1916–17]; Bougle 1971; Leach 1962; Dumont 1970; Barth 1962; Berreman 1972).
While there has been much disagreement about the precise definition of caste, nevertheless it is usually held that caste implies three things: hereditary occupational specialization, a hierarchical ranking of groups, and great social distance between them. Interaction between the members of different castes is governed by canons of purity and impurity. If an untouchable so much as gazes at the dinner of a Brahman, it will be considered impure. Exogamy is prohibited, and imbalanced sex ratios among the members of higher castes lead to hypergamy as well as female infanticide. The caste system restricts intergroup mobility and thereby limits the availability of benefits outside group boundaries. Less extreme instances follow from the distinctive legal status of groups such as Jews and Greeks under the Ottoman millet system, Indians in North America, or blacks in South African Bantustans. In each case the state makes individuals dependent on corporate membership by allocating resources to groups qua groups (Van Dyke 1977).
Just as laws that limit the individual’s alternatives outside group boundaries promote dependence, those that provide new alternatives lessen it. Here the development of the welfare state takes pride of place. The welfare state provides citizens with a wide range of benefits, or entitlements, that were customarily supplied by mutual benefit associations, churches, trade unions, political parties, and other kinds of voluntary associations. Once these benefits are offered to all as public goods, the incentives to belong to these other kinds of organizations erode. In this way, the growth of government