Risk, rationality, and religious portfolios.

Iannaccone, Laurence R.

I. INTRODUCTION

This paper builds on two assumptions - the first, self-evident; the second, more controversial. Together they yield a provocative theory of religious behavior and organization.

The first assumption is that religion involves a great deal of risk. Those who engage in religious acts cannot know that their efforts will have the desired effects. Religious acts thus have the character of risky investments, exacting resource costs in return for uncertain rewards. The second assumption is that people choose their religious behavior rationally, weighing the costs and benefits of religious behavior and seeking to limit the risk associated with religious choices.

Churches and other "religious firms" employ a variety of institutional arrangements that mitigate religious risk. Most arrangements follow one of two major risk management strategies, which in turn give rise to two different styles of religion. One centers on collective production, the other on private production and diversified consumption. Collective religion is congregationally oriented and tends to employ exclusivity and costly prohibitions to limit free riding and maintain high levels of commitment. Private religion is less susceptible to free-rider problems, but also less capable of generating commitment. Private religion tends to center on client/practitioner relationships or fee-for-service transactions, and consumers tend not to form exclusive attachments to a single religious firm. Outright payments for goods and services overshadow costly demands for an austere, deviant, or sacrificial life-style.

A large body of descriptive, historic, and anecdotal evidence confirms these predictions and illustrates their range of application. The concluding sections of the paper apply the proposed theory to a series of contrasts: Asian versus Western religion, mainstream "churches" versus exclusive "sects," New Age versus standard American religion, and Greco-Roman paganism versus Judeo-Christian monotheism.

II. RELIGIOUS COMMODITIES AND THE PROBLEM OF RISK

Economic models of religious behavior typically begin by assuming that individuals allocate their time and money so as to maximize their utility from religious commodities, R, and all other (secular) commodities, S. The production of religious commodities depends on personal inputs of time and money, but may also depend on past experience (religious human capital), styles of behavior (such as diet, dress, and sexual practices), and the time and money inputs of others (such as fellow church members). See Azzi and Ehrenberg [1975], Sullivan [1985], and Iannaccone [1992] for formal analyses and empirical tests.

With the exception of Durkin and Greeley [1991] and Montgomery [1992], previous models take belief as given, thereby sidestepping the issue of risk. But religious commodities do involve risk; sermons and scriptures do emphasize the need for faith; and people do wrestle with, and often yield to, doubt.

Insights from secular markets and the economics of uncertainty explain a broad range of behaviors and institutions that arise in response to religious risk. In particular, they highlight two different risk-reducing strategies, one of which centers on exclusivity and collective production, the other centers on diversification and private production.

Managing Risk through Collective Action: Religious Congregations

The would-be believer faces a situation not unlike the would-be used car buyer. Used cars are classic "experience" goods [Nelson 1970], since potential buyers are hard-pressed to determine car quality in advance of purchase. Sellers thus have an incentive to overstate the value of their merchandise and to disguise its true nature. Used car buyers, in turn, are aware of the sellers' incentive to misrepresent their products and so will attempt to increase their stock of information and/or reduce their risk in many ways. They may demand guarantees, seek information from third parties, investigate the seller's reputation, and so forth. For their part, sellers also have an incentive to provide, or at least appear to provide, proof that their claims are true.

The uncertainty surrounding most religious goods far surpasses that which surrounds used cars. Indeed, religion qualifies as the ultimate "credence good" [Darby and Karni 1973] since no amount of experience suffices fully to evaluate most religious claims. Hence we can predict the emergence of institutions and arrangements designed to increase information (or at least the appearance of information) and reduce fraud.(1)

Examples are not hard to find. Testimonials are commonplace in religion and, predictably, are more common in those variants that place greater emphasis on material blessings. Testimonies are more likely to be believed when they come from a trusted source, such as a personal acquaintance or a respected figure. They are especially credible when testifiers have relatively little to gain (or better yet much to lose) from having their claims heard and believed. This helps to explain why the character of religious activity is so often collective and structure of religious organizations is so often congregational. Fellow members are more trustworthy than strangers. They also have less incentive to overstate the benefits of the religion than do members of the clergy, whose livelihood depends on a steady stream of "sales." The clergy, in turn, are more persuasive when they do not benefit materially from their followers' faith (or have large sunk costs invested in their reputation). This may help to explain why clergy receive low salaries relative to their level of training. Citizens fully expect the managers of successful secular corporations to earn large salaries, bonuses, and stock options; and they rarely boycott a firm simply because its chief officers have grown rich from product sales. Yet the same citizens have little but scorn for religious leaders who parlay successful ministries into personal fortunes.

Several distinctive features of religious institutions are thus explained in terms of their ability to reduce the risk of consumer fraud. These include: a minimal professional staff whose financial compensation is low and independent of customer contributions/payments; heavy reliance on part-time and volunteer workers (and thus reliance on payments of time and service rather than money); a congregational structure, which limits the need for full-time professionals and provides a source of credible product endorsements; and collective activities, which provide continuous assurance through the enthusiasm, devotion, conviction, and testimony of fellow members.

Free-rider problems. There is, however, a downside to congregational structure and collective activities: free-rider problems. These problems lead to an organizational response that I have labeled "sacrifice and stigma."(2)

To understand the temptation to free ride, consider the situation confronting the typical member of a Protestant church. The pleasure and assurance that he (or she) derives from Sunday services depends not just on his own inputs but also the inputs of others - how many others attend, how warmly they greet him, how well they sing, how enthusiastically they read and pray, how deep their expressed commitment, and so forth. This leads immediately to two related problems. The first is that members with low levels of participation (and perhaps also low levels of faith) are tempted to free ride off those with higher levels, since given the choice they are better off in groups whose average level of participation is greater than their own. The second is that even in a homogeneous group, opportunistic behavior leads to an inefficient equilibrium with suboptimal participation, since individuals maximize personal welfare by ignoring the external benefits of their participation.

Costly demands offer a solution to free-rider problems. These costs are not the standard costs associated with the production or purchase of secular commodities. Rather, they are apparently gratuitous costs - sacrifice and stigma - foreign to most secular contexts: burnt offerings, which destroy valued resources; distinctive dress and grooming that invite ridicule or scorn; dietary and sexual prohibitions that limit opportunities for pleasure; restrictions on the use of modern medicine or technology. Sacrifice and stigma discourage free-riding by prohibiting or penalizing activities (such as interaction with nonmembers) that otherwise compete for members' resources. In heterogeneous populations, sacrifice and stigma tends to screen out the less committed members. And even in homogeneous populations, it induces a substitution effect that can raise average levels of group participation and group utility.

Deviant norms thus mitigate the externality problems faced by religious groups. Distinctive diet, dress, grooming, and social customs constrain and often stigmatize members, making participation in alternative activities more costly. Potential members are forced to choose: participate fully or not at all. The seductive middle-ground of free-riding and low participation is eliminated. Paradoxically, those who remain find their welfare increased. It follows that perfectly rational people can be drawn to decidedly unconventional groups. It also follows that any collectively oriented religion will benefit from some level of sacrifice and stigma, although the optimal level declines as the average member's income and education rise.

We have thus traced the consequences of religious risk down a lengthy path: from collective activities and congregational structures, to free-rider problems, to sacrifice and stigma.

Managing Risk through Diversification: Religious Portfolios

There is, however, an alternative response to risk - portfolio diversification. To apply this insight to religious behavior, note that most people have access to a variety of different religions or different classes of religious acts within a single religious tradition.(3) It follows that rational religious investors have reason to avoid specializing in a single religion or class of religious acts, preferring instead to diversify their investments over a number of them.(4) In fact, since risk is most effectively reduced by investing in assets that vary independently or even negatively, rational investors may hedge their bets in one religion with other bets in very different religions. For example, one might go to confession on Sunday, consult a medium on Monday, and engage in transcendental meditation on Wednesday. Hence:

PROPOSITION 1. In order to reduce the risk associated with religious commodities, consumers will seek to hold diversified portfolios of competing religious assets.

How do we square this prediction with reality? After all, the typical church-goer is firmly wedded to a single religious tradition, a single denomination, and a single congregation. This tendency to specialize can not be dismissed merely by noting that most religions do not allow diversification. If the demand for diversification is as great as the preceding argument suggests, then why does competition not force churches to accommodate it? How do churches demand, and get, the sort of brand loyalty that secular producers only dream of?

The answers lie in the contrasting production technologies employed by different types of religion. These technologies may be either "collective" or "private," and only the latter lend themselves to diversification.

Most Western religions rely on collective, congregational production. In describing the risk-reducing advantages of the collective production approach, I have emphasized that these religions must limit free riding. Many do so through costly demands that effectively isolate members from competing groups. The demands can be sustained, despite competition from other religions, because the costs of exclusivity are for many people more than balanced by the gains from collective action. But exclusivity and diversification do not mix; a church can not isolate its members while at the same time permit them to participate in competing groups. Hence, congregationally oriented religions can (and generally do) limit their members' involvement in competing religions. These limitations are most extensive in highly sectarian groups such as the Mormons, Jehovah's Witnesses, Adventists, Krishnas, Moonies, and fundamentalist Christian denominations.

But not every religion is congregationally oriented. Some deal in private commodities that can be transferred directly from an individual producer to an individual consumer without recourse to a mediating group. Most "New Age" products fall into this category - books, tapes, and seminar lectures; crystals and pyramids; and techniques for meditation. Production and distribution of these products are not hampered by free-rider problems, and thus the costly strategies described in the previous section are neither necessary nor profitable. Indeed, a private commodity producer who demands exclusivity or stigma will suffer a competitive disadvantage relative to other private producers. Firms dealing in private commodities must therefore sell their products "with no strings attached." It follows that their customers will patronize other firms so as to diversify their portfolio of religious assets. In short,

PROPOSITION 2a. Wherever religious firms exist to provide private commodities, competitive forces and risk aversion will lead consumers to patronize multiple firms, thereby diversifying their religious portfolios.

PROPOSITION 2b. Wherever religious firms exist to facilitated the production of collective goods, the firm and its patrons will demand exclusivity to mitigate the free-rider problems.

Proposition 2b does not assert that a collective setting alters people's desire for a diversified portfolio, only their opportunity to indulge that desire. In the absence of institutional constraints, people will yield to the temptation to free ride and to diversify, and this in turn will undermine the institution. Hence, only those firms that maintain sufficient exclusivity (by means of strictness, monitoring, and the like) can remain viable. The members' impulse to diversify will be accommodated only in a limited sense and only within the individual firm. It follows that these firms will be under pressure to take a department store approach to religion - comprehensive belief systems, cradle to grave services, extensive and varied social networks, and so forth. In contrast, religious firms that produce private commodities will be under no pressure to provide a full range of products. The forces of competition will in fact drive them to specialize, much as secular firms tend to specialize. Over time, private commodity firms will come to resemble highly specialized boutiques.

If propositions 1 and 2 are correct, then the degree to which a religious firm provides its customer-members collective versus private goods should correlate with its tendency to make exclusive versus nonexclusive demands. The comprehensiveness of firm's product line should also correlate. Moreover, the content of the religions and their financial strategies should also follow different patterns. Private religion will usually deal in specific goods and services, items that can be produced by one set of individuals and sold to another. Collective religion will deal in collective action and collective services - commodities that blur the line between producer and consumer and that do not lend themselves to buying and selling. The distinction between private and collective goods thus helps explain why some religions organize around client/practitioner or buyer/seller relationships whereas others organize as collectives in which all members are, to varying degrees, coworkers and co-consumers. Table I summarizes these and other predictions of the theory.

III. EXAMPLES AND APPLICATIONS

Although the predictions in Table I are straightforward, statistical tests must await new data. Existing surveys largely [TABULAR DATA FOR TABLE I OMITTED] ignore the issues of collective versus private religion, sacrifice and stigma, religious risk, and fee-for-service religion. They especially ignore diversification, assuming instead that religious affiliation is exclusive and unitary. Respondents are asked to state "their religion," and all subsequent questions about religious practice are assumed to refer back to that religion. The resulting data radically underestimate levels of involvement in non-conventional religion and provide no insight into diversification.(5)

Fortunately, there exists a large body of qualitative evidence relevant to the theory and its range of application.

Example 1: Western Religions versus Asian Religions

At its most general level, the theory of collective versus private production suggests a fundamental distinction between Western and Asian religions.

Western religions - Christianity, Judaism, and Islam - emphasize collective action. Congregations form the basic unit of religious production, and congregational events (such as weekly worship services) underpin the standard cycle of religious activity. Responsibility for religious production is, to varying degrees, shared by everyone. As the theory predicts, exclusivity and membership are fundamental to each of these religions. There is an "in" group and an "out" group, and the distinction between the two is of great, even eternal, consequence. Defining and guarding the membership boundary is a central concern, and transitions across the membership boundary - conversion and apostasy - are major events. Members are expected to remain loyal to their particular "brand" of religion, and loyalty is, in fact, the norm [Roof and McKinney 1987, 162-181]. Those who attempt to "diversify" their religious practices through participation in different groups are condemned as backsliders or even excommunicated as heretics. Indeed, in these religions, renouncing all other faiths is as important as affirming the group's faith. The archetypal example appears in the ten commandments: "I am the Lord your God ... You shall have no other gods before me ... for I the Lord your God am a jealous God" [Deuteronomy 5:6-9]. In like manner, Muslims are expected to reaffirm at every private or congregational prayer, "There is no God but Allah, and Muhammad was his last messenger on Earth."

Western religions often employ sacrifice and stigma, restricting the diet, dress, or social and sexual behavior of their members. Religious goods and services are not usually sold for money; rather they are freely offered to all members (although contributions are expected in return). The scope of services is very broad, comprehending virtually every aspect of this life and the next.

The most popular forms of Asian religions are quite different. Collective, congregational production takes a back seat to private practice. Household rituals, individualistic exchanges with the gods (as when one makes a money offering in order to effect a specific outcome), the purchase of charms and other religious artifacts, and client/practitioner exchange relationships are pervasive. Brand loyalty is weak, and diversification is common. It is not unusual to patronize more than one priest, more than one temple, or even more than one religion. According to Maraca [1975, 7-8] dual religious membership is "observable everywhere in rural and traditional Japan":

A household keeps a family Buddhist altar where Buddha and the family's ancestral spirits are enshrined, a family Shinto altar for charms distributed by local shrine and the Ie Grand Shrine; in addition, charms issued by prayer temples are posted on the walls and pillars of the house.

In this context, "membership" is at best a matter of secondary concern and at worst a meaningless construct. According to McFarland [1967, 20], most Japanese religious institutions "do not require or even provide for the kind of profession of faith or commitment to membership that is the sine qua non of meaningful religious statistics in the West. Thus, even if one turns to the people themselves$and asks them to identify themselves religiously, he will find that many of them cannot do so with any certainty." Indeed, the official statistics cited by Morioka [1975, 4] show that the combined number of Japanese Shintoists, Buddhists, Christians, and sect members sums to nearly twice the national population!

Along with the phenomena of private production, nonexclusivity, and diversified plural membership, there is less emphasis on the sort of behavioral restrictions that I have labeled "sacrifice and stigma." The blessings of religion are often purchased outright. Obtaining them in return for self-mortification is far less common.(6)

Example 2: The New Age Movement

Within the United States and Europe, the "New Age" movement and its many precursors epitomize what I have called private religion - specialized sellers, no congregational structure, client/practitioner relationships, fee for service transactions, no brand loyalty, and acknowledged diversification. At New Age fairs, for example, dozens of exhibitors will set up stalls, side by side, and hawk their wares to consumers who literally shop around for products - astrology, numerology, psychic readings, hand-analysis, channeling, runes, tarot, past-life regression, crystals, metaphysical tapes and books, aura photographs, massage, chiropractic treatment, and natural foods. See Jorgensen [1992, 90-100] for details.

Although the fairs are but one facet of New Age religion, similar features characterize the entire movement. Wilson [1987, 41-42] observes that "therapeutic movements are vulnerable because their hold on the clientele is itself so narrow. The basic client-practitioner relationship of all such movements makes difficult the creation of a solid core of support." Johnson [1987, 255] states that "such organizations have difficulty building a stable pool of devoted members because the bulk of them are essentially fee-paying clients ... There is a wide-open market in the area of commercialized therapeutic services, and brand loyalty is weak." And according to McGuire [1987, 125] the typical "seeker" goes about "permanently searching for particles of truth from a wide variety of sources ... drawling] upon several different groups, either simultaneously or serially. ... [T]here is thus no way to handle 'heresy' and therefore no clear boundaries between members and nonmembers."

Example 3: The Spectrum of Protestant Denominations

Protestant Christianity is on the whole collectively oriented. But the degree of commitment to congregational organization and collective action varies systematically across different Protestant denominations. This gives rise to a church-sect spectrum, with the so-called "liberal" or "mainline" denominations at one end and the "sectarian" denominations at the other. The Episcopal, Congregational, Presbyterian, and Methodist denominations are among the most liberal groups; the Southern Baptist, Pentecostal, Fundamentalist, Adventist, and Jehovah's Witness denominations are among the most sectarian.

Intra-Protestant comparisons thus provide a small-scale test of the theory of collective versus private production. Table II summarizes empirical results from previous studies of Protestant groups [Iannaccone 1992; 1994]. The observed differences are large, significant, and easily replicated. Compared to members of other Protestant denominations, sect members are poorer and less educated, contribute more money and attend more services, hold stronger and more exclusive beliefs, belong to smaller congregations, are more likely to choose their closest friends from within their congregation, and are less involved in activities and organizations outside of their congregation. The spectrum of Protestant denominations thus provides statistical evidence that the attributes of religious organizations and the behavior of their members cluster as predicted.

Other studies extend these findings to the realm of religious portfolios, proving that members of sect-like denominations are much less likely to "diversify" than are their more mainline counterparts. For example, Stark and Bainbridge's [1985, 381-386] survey of about 1,500 university students found that "Born Again" Christians were far less likely that other Christians to approve of astrology, psychic practices, or Eastern meditation. Donahue [1991] finds that Southern Baptist church members are much less likely to hold New Age beliefs than are members of more liberal Protestant denominations. And Tamney et al. [1991] find that experimentation with New Age religious activities is more common among members of church-like denominations than sect-like denominations.

TABLE II

Mean Values for Members of Different Protestant Denominations(*)

 LIBERAL MODERATE SECT-LIKE SECTS

HOUSEHOLD INCOME: 10,140 9,435 8,399 6,944
(dollars per year)

RESPONDENT EDUCATION: 14.5 13.9 12.3 12.5
(years)

SUNDAY ATTENDANCE: 33.8 38.4 44.2 49.1
(services/year)

CHURCH CONTRIBUTIONS: 244 288 369 497
(dollars/year)

CHURCH FRIENDS: 1.32 1.51 1.80 3.15
(number out of 5 closest)

NON-CHURCH MEETINGS: 48.6 41.1 30.6 29.0
(number/year)

NON-CHURCH MEMBERSHIPS: 3.59 3.07 2.13 1.72
(total number)

EXCLUSIVE BELIEFS: 2.06 2.70 3.79 4.15
(3-item, belief scale)

NUMBER OF RESPONDENTS: 963 882 194 248
Notes:

Source: Iannaccone's [1992] analysis of data from the Northern
California Church Member Study, 1963.

Denominational groups: LIBERAL = members of Congregational,
Methodist, Episcopalian, and Disciples of Christ churches. MODERATE
= Presbyterians, American Lutherans, and American Baptists.
SECT-LIKE = Missouri Lutherans and Southern Baptists. SECTS =
members of Assemblies of God, Church of God, Church of Christ,
Nazarene, Seventh Day Adventist, Gospel Lighthouse, Foursquare
Gospel churches.

* For all variables, the mean difference between members of sects
and the liberal denominations is significant at the 1% level.

Example 4: Paganism versus Early Christianity and Judaism

A final example, drawn from classical history, suggests that the theory of private versus collective religion may be as applicable to ancient times as it is to modern societies.

The Roman Empire was filled with competing religions: civic worship of the old Greco-Roman gods, such as Zeus, Apollo, Artemis, Dionysus, and the like; Caesar worship; mystery religions of Isis, Mithras, Adonis, and others; magic and astrology; and numerous schools of philosophy. All these existed, side by side, in an atmosphere of mutual toleration. Adherence to one did not preclude involvement in another. Citizens could, and many did, participate in multiple cults - sacrificing to several gods, worshipping the emperor, and undertaking a variety of different rites and initiations. Pagan worship exemplified the portfolio approach to religion.

Christianity and Judaism were utterly different, and it was this difference that occasioned their persecution in an empire that otherwise tolerated virtually any and all religions. Unlike the pagan cults, Christianity and Judaism demanded not only acceptance of new rites and new beliefs, but also renunciation of old ways. As Robbins [1988, 65; see also Nock 1964] notes, they "were characterized by exclusivity whereby the convert was cut off from his past lifestyle and identity and from other religious groups. One was thus converted to the intolerant faiths of Judaism and Christianity while one merely adhered to the cults of Isis, Orpheus, or Mithra." Pagans might use Mithraism or consult the Delphic Oracle, but Christians and Jews belonged to their faiths.(7) Consistent with this paper's theory of exclusive religion, Christianity and Judaism placed far more rigorous demands on the moral behavior and life-style of their members and offered a far broader range of benefits in return. Moreover, Christianity and Judaism were collective, congregationally oriented religions. According to Dodds [1965, 136-138], a "Christian congregation was from the first a community in a much fuller sense than any corresponding group of Isiac or Mithraist devotees. Its members were bound together not only by common rites but by a common way of life and ... by their common danger.... The Church provided the essentials of social security: it cared for widows and orphans, the old, the unemployed, and the disabled; it provided a burial fund for the poor and a nursing service in time of plague... Christians were in more than formal sense 'members one of another'." The contrasts between paganism and Christianity thus fit the pattern of Table I in every detail.

IV. CONCLUSIONS

This paper has interpreted many aspects of religious institutions as attempts to limit religious risk. In particular, it has identified two major risk management strategies that give rise to two different styles of religion. One centers on collective production, the other on private production and diversified consumption. Collective religion is congregationally oriented and thus susceptible to free-rider problems. Exclusivity and costly prohibitions are employed by such religions to limit free riding and maintain high levels of commitment. Western religions, particularly their more sectarian forms, exemplify this pattern. Private religion is less susceptible to free-rider problems, but is also less capable of generating commitment. It may attract many "customers" but few real "members." New Age religions in America and Shinto and Buddhism in Japan exemplify this style of religion. The practice of these religions tends to center on client/practitioner relationships or fee-for-service transactions, and consumers tend not to form exclusive attachments to a single religion or religious firm. Payments for goods and services tend to overshadow costly demands for an austere, deviant, or sacrificial life-style.

The analysis is by no means complete. The argument builds on earlier, formal models of religious participation, but is itself intuitive. The concept of religious risk needs to be specified more precisely and its implications derived formally. As things stand, I can not explain why market forces do not always drive religions toward one style of production, private or collective. (One should, however, note that collective religions such as evangelical Protestantism and Islamic fundamentalism have tended over time to gain market share at the expense of less demanding alternatives.)

The theory of religious portfolios and of collective versus private religion is rich in testable predictions. Existing surveys do not address these implications, but the implications do accord with much of what is known about religious movements within the United States, across the globe, and throughout history. That in itself is a very promising start.

1. I am indebted to Hull and Bold [1989] for drawing my attention to this fact. Although they do not discuss congregational structure and collective activity, their basic insights regarding risk and fraud provide much of the motivation for the following two paragraphs.

2. See Iannaccone [1992] for a formal model of religious groups in which sacrifice and stigma works to reduce free-riding and raise utility.

3. The type and range of available options will, of course, differ from one time and place to another. Contemporary America and ancient Rome are examples of societies offering an exceptionally wide range of choices. Societies dominated by a single, state-supported religion (e.g., medieval Europe and contemporary Iran) do much to limit religious choice. Still, we tend to underestimate the extent to which even the medieval Catholic monopoly struggled with competition from paganism, witchcraft and schism [Stark and Bainbridge 1985, 111-122].

4. One may recoil at a comparison of religious acts and stock market acquisitions, but it should be noted that investment metaphors are common to most religions. Pascal's "wager" is an investment-oriented appeal to faith, as is Matthew 6:19-20.

5. For example, projections from a recent telephone poll of over 113,000 American households [Kosmin 1991] imply that only 20,000 Americans identify their religion as "New Age." Nevertheless, subscriptions to the New Age Journal number more than 140,000. Evidently, the overwhelming majority of New Age patrons describe themselves as members of some other religious tradition or no religion at all.

6. The distinction that I have attempted to draw between Western and Asian religion is, of course, highly generalized. Within the religious traditions of virtually every Asian country, one does find some collective, congregationally oriented groups. Hindu communes and Buddhist monasteries come readily to mind. Some of Japan's new religions, such as Shokagakkai, also fall into this category as do America's best-known imported "cults" - Hare Krishna and the Unification Church. As the theory predicts, all these groups emphasize membership, demand exclusivity, maintain high levels of participation, enforce strict codes of behavior and provide a comprehensive range of services.

7. Some pagans attempted to add Christianity and Judaism to their religious portfolios. For example, Dodds [1965, 107] reports that "the Emperor Alexander Severus, kept in his private chapel statues of Abraham, Orpheus, Christ and Apollonius of Tyana" and that the Gnostic Carpocrates preached a "comprehensive cult" that "worshipped images of Homer, Pythagoras, Plato, Aristotle, Christ and St Paul." Christians' steadfast refusal to practice any element of civic religion was thus grounded in the realistic fear that "even the most formal concession to pagan cult would lead in the end to Christianity being swallowed up and digested in the all-embracing maw of Graeco-Roman paganism" [Dodds 1965, 113].

REFERENCES

Azzi, Corry, and Ronald G. Ehrenberg. "Household Allocation of Time and Church Attendance." Journal of Political Economy, February 1975, 27-56.

Darby, Michael R., and Edi Karni. "Free Competition and the Optimal Amount of Fraud." Journal of Law, and Economics, April 1973, 67-88.

Dodds, E. R. Pagan and Christian in an Age of Anxiety: Some Aspects of Religious Experience from Marcus Aurelius to Constantine. London: Cambridge University Press, 1965.

Donahue, Michael J. "Prevalence of New Age Beliefs in Six Protestant Denominations." Working paper, Search Institute, 1991.

Durkin, John T., Jr., and Andrew M. Greeley. "A Model of Religious Choice under Uncertainty: On Responding Rationally to the Non-Rational." Rationality, and Society, April 1991, 178-196.

Finke, Roger, and Rodney Stark. The Churching of America 1776-1990: Winners, and Losers in Our Religious Economy. New Brunswick, N.J.: Rutgers University Press, 1992.

Hull, Brooks B., and Frederick Bold. "Towards an Economic Theory of the Church." International Journal of Social Economics, 16(7), 1989, 5-15.

Iannaccone, Laurence R. "Sacrifice and Stigma: Reducing Free Riding in Cults, Communes, and Other Collectives." Journal of Political Economy, April 1992, 271-91.

-----. "Why Strict Churches Are Strong." American Journal of Sociology, March 1994, 1180-211.

Johnson, Benton. "A Sociologist of Religion Looks at the Future of New Religious Movements," in The Future of New Religious Movements, edited by David G. Bromley, and Phillip E. Hammond. Macon, Ga.: Mercer University Press, 1987, 251-60.

Jorgensen, Danny L. The Esoteric Scene, Cultic Milieu, and Occult Tarot. New York: Garland, 1992.

Kosmin, Barry A. "Research Report: The National Survey of Religious Identification 1989-90." The Graduate School and University Center of the City University of New York, 1991.

McFarland, H. Neill. The Rush Hour of the Gods: A Study of New Religious Movements in Japan. New York: Macmillan, 1967.

McGuire, Meredith B. Religion: The Social Context, 2nd ed. Belmont, Calif.: Wadsworth, 1987.

Montgomery, James. "Pascal's Wager and the Limits of Rational Choice: A Comment on Durkin and Greeley." Rationality and Society, January 1992, 117-22.

Morioka, Kiyomi. Religion in Changing Japanese Society. Japan: University of Tokyo Press, 1975.

Nelson, Phillip J. "Information, and Consumer Behavior." Journal of Political Economy, March/April 1970, 311-29.

Nock, Arthur Darby. Early Gentile Christianity and Its Hellenistic Background. New York: Harper and Row, 1964.

Robbins, Thomas. Cults, Converts and Charisma: The Sociology of New Religious Movements. London: Sage, 1988.

Roof, Wade Clark, and William McKinney. American Mainline Religion: Its Changing Shape and Future. New Brunswick: Rutgers University Press, 1987.

Stark, Rodney, and William Sims Bainbridge. The Future of Religion. Berkeley: University of California Press, 1985.

Sullivan, Dennis H. "Simultaneous Determination of Church Contributions and Church Attendance." Economic Inquiry, April 1985, 309-20.

Tamney, Joseph B., Barbara A, Bunch, Paul R. Stieber, and Deborah L. Zigler-Geis. "The New Age in Middletown: An Exploratory Study." Working paper, Ball State University, 1991.

Wilson, Bryan R. "Factors in the Failure of the New Religious Movements," in The Future of New Religious Movements, edited by David G. Bromley, and Phillip E. Hammond. Macon, Ga.: Mercer University Press, 1987, 30-45.

LAURENCE R. IANNACCONE, Associate Professor, Department of Economics, Santa Clara University, Santa Clara, CA. Early drafts of this paper were presented at the meetings of the Society for the Scientific Study of Religion, November 1991, the American Economic Association, January 1992 and the Western Economic Association, June 1992. I am indebted to many colleagues for encouragement and suggestions, particularly Douglas Allen, Corry Azzi, Gary Becker, Thomas Borcherding, Timur Kuran, John Lott, John Murray, S. Ramachandran, Rodney Stark, Gordon Winston and two anonymous reviewers. This research was supported in part by a grant (#90-0004) from the Lilley Endowment's Louisville Institute for the Study of Protestantism and American Culture.