Charles Tilly, in this eloquent manifesto, presents a powerful new approach to the study of persistent social inequality. How, he asks, do long-lasting, systematic inequalities in life chances arise, and how do they come to distinguish members of different socially defined categories of persons? Exploring representative paired and unequal categories, such as male/female, black/white, and citizen/noncitizen, Tilly argues that the basic causes of these and similar inequalities greatly resemble one another. In contrast to contemporary analyses that explain inequality case by case, this account is one of process. Categorical distinctions arise, Tilly says, because they offer a solution to pressing organizational problems. Whatever the "organization" isas small as a household or as large as a governmentthe resulting relationship of inequality persists because parties on both sides of the categorical divide come to depend on that solution, despite its drawbacks. Tilly illustrates the social mechanisms that create and maintain paired and unequal categories with a rich variety of cases, mapping out fertile territories for future relational study of durable inequality.
About the Author
Charles Tilly is Joseph L. Buttenwieser Professor of Social Science at Columbia University and former Distinguished Professor and Director of the Center for Studies of Social Change at the New School for Social Research. Among his recent books are Roads from Past to Future (1997), Work Under Capitalism (with Chris Tilly, 1997), Popular Contention in Great Britain (1995), and European Revolutions (1993).
Read an Excerpt
By Charles Tilly
University of California PressCopyright © 1999 Charles Tilly
All right reserved.
Of Essences and Bonds
We could reasonably call James Gillray (1757-1815) Britain's first professional cartoonist (George 1967, 57; Hill 1976). He left us unforgettable images of public and private affairs under George III. Very few handsome people figure in Gillray's caricatures. In the savage portrayals of British life he drew, etched, and colored toward 1800, beefy, red-faced aristocrats commonly tower over other people, while paupers almost invariably appear as small, gaunt, and gnarled. If Gillray painted his compatriots with malice, however, he also observed them acutely.
Take the matter of height. Let us consider fourteen-year-old entrants to the Royal Military Academy at Sandhurst to represent the healthier portion of the aristocracy and gentry, and fourteen-year-old recruits for naval service via London's Marine Society to represent the healthier portion of the city's jobless poor At the nineteenth century's start, poor boys of fourteen averaged only 4 feet 3 inches tall, while aristocrats and gentry of the same age averaged about 5 feet 1 inch (Floud, Wachter, andGregory 1990, 197; for the history of the Marine Society as an aristocratic benefaction, see Colley 1992, 91-93). An average beginning military cadet stood some 10 inches taller than anewly recruited mariner. Because poor youths then matured later than rich ones, their heights converged an inch or two by adulthood. Nevertheless we can imagine their counterparts in the army: aristocratic officers glowering down half a foot or more at their plebeian troops. Such an image vivifies the phrases "high and mighty," "haughty," and "look down on someone."
Poor people have few good times. But the years around 1800 brought Britain's low-income families especially bad times. In the short run, massive diversion of resources and labor power to French Revolutionary and Napoleonic wars depleted domestic production as it drove up consumer prices. Over the longer run, the urbanization, industrialization, and sharpened inequality promoted by capitalist expansion were then aggravating the hardships faced by Western Europe's poorer households. As poor people ceased producing their own food faster than agricultural productivity rose, hardship extended to their daily bread.
In his Nobel Prize lecture, economist and economic historian Robert Fogel points out that at nutritional levels prevailing toward the end of the eighteenth century, from 3 to l0 percent of the English and French work forces had too little food to sustain any effective work at all, while a full fifth of the population commanded too little for more than a few hours of light work per day (Fogel 1994, 371-374). At those low nutritional levels, furthermore, English and French workers were extremely vulnerable to chronic disease, hence liable to work lives disrupted by illness and early death. Fogel speculates that malnutrition itself thereby accounted for the stunning proportion of beggarsup to 20 percent of the entire populationreported in various regions of eighteenth-century Europe.
Over population categories, regions, and countries, as Fogel and other researchers have recently established, material well-being and stature vary in strong relation to each other (Floud, Wachter, and Gregory 1990; Fogel 1993, 1994; Fogel and Costa 1997; Komlos 1987, 1990, 1994). Richard Steckel sums up:Stature adeptly measures inequality in the form of nutritional deprivation; average height in the past century is sensitive not only to the level of income but to the distribution of income and the consumption of basic necessities by the poor. Unlike conventional measures of living standards based on output, stature is a measure of consumption that incorporates or adjusts for individual nutritional needs; it is a net measure that captures not only the supply of inputs to health but demands on those inputs. (Steckel 1995, 1903)
Well-being and height link through food consumption; victuals invigorate. Although genes set variable limits to height distributions in human populations, childhood nutrition strongly affects the degree to which any individual approaches her or his genetic limit. Low birth weight, which typically results from a mother's illness and malnutrition, predicts reliably to a child's health problems, diminished life expectancy, and smaller adult size.
Within a given population, furthermore, short stature itself generally predicts to higher levels of morbidity and mortalitymost likely not because of height's inherent advantages but because, on the whole, short stature correlates with unfavorable childhood health experiences and lesser body strength. Rising height across an entire population therefore provides one of our clearest signs that the well-being of that population is increasing, and marked adult height differentials by social category within the male or female population provide a strong indicator of durable inequality.
That average heights of adults in Western countries have typically risen 6 inches or so over the past century and a half reflects a significant rise in living standards. That even in egalitarian Sweden recent studies reveal lower birth weights for the newborn of less-educated women (in this case, most likely a joint outcome of smoking and nutrition) tells us that material inequalities persist into prosperity (Dagens Nyheter 1996). That at my modest altitude I easily see over the heads of many adult males with whom I travel on New York subwaysespecially those speaking languages other than Englishsignals that in capitalist countries we still have profound inequalities of life experience to identify and explain.
Since sexual dimorphism prevails among primates and since humans commonly live in mixed-sex households whose members share food, one might suppose that female/male height differences, unlike class inequalities, derive almost entirely from genetic predisposition. Not quite. Nature and nurture are disentangled with difficulty when it comes to such matters as sex differences in body size. As James Tanner puts it:
Variation between the heights of individuals within a subpopulation is indeed largely dependent on differences in their genetic endowment; but the variation between the means of groups of individuals (at least within an ethnically homogeneous population) reflects the cumulative nutritional, hygienic, disease, and stress experience of each of the groups. In the language of analysis of variance, most of the within-group variation is due to heredity, and most of the between-group variation is due to childhood environment. (Tanner 1994, 1)
What counts, however, as a subpopulation, or group? Surely not any cohabiting population, regardless of social divisions within it. For "group," read "category" to recognize that class, gender, race, ethnicity, and similar socially organized systems of distinction clearly qualify. (I will follow current conventions by speaking of "sex" in reference to X and Y chromosome-linked biological differences, "gender" in reference to social categories.) In each of these cases, differences in "nutritional, hygienic, disease, and stress experience" contribute to differences in adult stature. Researchers in the field have so far done much more with class differences, national differences, and change over time than with male / female differences.
Still, gender likewise marks distinctive childhood experiences, even when it comes to nutrition. When children in pastoral and agricultural economies begin serious work in their household enterprises, they almost always take on gender-differentiated tasks. That means their daily routines give boys and girls unequal access to food. Most of the time girls get less, and their food is of lower quality. Where men fish or hunt while females till and gather, however, the division of labor often attaches girls and women to the more reliable and continuous sources of calories. Thus in some circumstances females may actually get better nourishment than males.
The fundamental fact, then, is gender differentiation in nutrition, with the usual but not universal condition being inferior nutrition for females. We have enough episodic documentation concerning gender discrimination with respect to health care, feeding, infanticide, and general nurture, as well as slivers of evidence suggesting gender-differential patterns of improvement or decline in nutrition under the influence of broad economic fluctuations, to support hypotheses of widespread unequal treatment of males and females, of inequality in their resulting life chances, hence of a social contribution to gender differences in weight and height as well.
Below a certain threshold of food supply, most households make regular if implicit choices concerning which of their members will have adequate nourishment. Contemporary capitalist countries seem to have risen above that threshold, although we lack reliable evidence concerning nutritional inequality among capitalism's currently increasing share of poor people. But the hungry world as a whole still features gender discrimination in nutrition.
Here Fogel's line of investigation crosses the inquiries of Amartya Sen (Sen 1981, 1982, 1983, 1992). From his analyses of poverty and famine onward, Sen has sniffed out deliberately unequal treatment in the presence of resources that could ensure more general welfare. He recurrently detects gender-differentiated claims on such resources. "There is a lot of indirect evidence," he comments, "of differential treatment of women and men, and particularly of girls vis-'-vis boys, in many parts of the world, e.g., among rural families in Asia and North Africa. The observed morbidity and mortality rates frequently reflect differential female deprivation of extraordinary proportions" (Sen 1992, 123). The most dramatic observations concern female infanticide through direct attack or (more often) systematic neglect, which analysts have frequently reported for strongly patrilineal regions of Asia (Johansson and Nygren 1991; Langford and Storey 1993; Lee, Campbell, and Tan 1992; Lee, Feng, and Campbell 1994; Muhuri and Preston 1991; Yi et al. 1993).
People of Western countries have not much practiced selective female infanticide. But Western states have often reinforced gender distinctions in nutrition and nurture, notably by confining military service to males, diverting food stocks from civilian to military use, providing superiorhealth care for troops, and ensuring that soldiers receive better rations than the general population. Florence Nightingale, after all, more or less invented professional nursing as we know it while organizing the health care of British fighting men during the Crimean War. In the absence of powerful drugs and diagnostic instruments, Nightingale's nursing stressed cleanliness, warmth, and nourishment, comforts many women back home in Britain did not then share. If military men at war have historically faced exceptional risks of violent death and disabling disease, in recent centuries they have also typically received three square meals a day when civilians, especially female civilians, were tightening their belts.
Such socially organized differences in well-being illustrate this book's main subject: the causes, uses, structures, and effects of categorical inequality. The book does not ask what causes human inequality in general. Instead it addresses these questions: How, why, and with what consequences do long-lasting, systematic inequalities in life chances distinguish members of different socially defined categories of persons? How do categorical inequalities form, change, and disappear? Since all social relations involve fleeting, fluctuating inequalities, let us concentrate on durable inequalities, those that last from one social interaction to the next, with special attention to those that persist over whole careers, lifetimes, and organizational histories.
Let us concentrate, furthermore, on distinctly bounded pairs such as female / male, aristocrat / plebeian, citizen / foreigner, and more complex classifications based on religious affiliation, ethnic origin, or race. We focus on categories rather than on continua such as [rich . . . poor], [tall . . . short], [ugly . . . beautiful], and so on. Bounded categories deserve special attention because they provide clearer evidence for the operation of durable inequality, because their boundaries do crucial organizational work, and because categorical differences actually account for much of what ordinary observers take to be results of variation in individual talent or effort.
As Max Weber noted almost a century ago, the creation of what he called "social closure" advances efforts by the powerful to exclude less powerful people from the full benefits of joint enterprises, while facilitating efforts by underdogs to organize for the seizure of benefitsdenied (Weber 1968, 1:43-46, 1:341-348; Parkin 1979, 44-116). A relationship is likely to be closed, Weber remarked,
in the following type of situation: a social relationship may provide the parties to it with opportunities for the satisfaction of spiritual or material interests. If the participants expect that the admission of others will lead to an improvement of their situation, an improvement in degree, in kind, in the security or the value of the satisfaction, their interest will be in keeping the relationship open. If, on the other hand, their expectations are of improving their position by monopolistic tactics, their interest is in a closed relationship. (Weber 1968, 1:43)
Organizations such as firms and clans use closure by drawing complete boundaries around themselves and then monitoring flows across those boundaries with care. Contrary to Weber, however, I argue that at a scale larger than a single organization completely bounded categories are rare and difficult to maintain, that most categorical inequality relies on establishment of a partial frontier and defined social relations across that frontier, with much less control in regions distant from the frontier. Yet in other regards my analysis resonates with Weber's discussion. It builds a bridge from Max Weber on social closure to Karl Marx on exploitation, and back. Crossing that bridge repeatedly, this book concerns social mechanismsrecurrent causal sequences of general scopethat actually lock categorical inequality into place. The central argument runs like this: Large, significant inequalities in advantages among human beings correspond mainly to categorical differences such as black/ white, male/female, citizen/foreigner, or Muslim/Jew rather than to individual differences in attributes, propensities, or performances. In actual operation, more complex categorical systems involving multiple religions or various races typically resolve into bounded pairs relating just two categories at a time, as when the coexistence of Muslims, Jews, and Christians resolves into the sets Muslim/Jew, Muslim/Christian, and Jew/Christian, with each pair having its own distinct set of boundary relations.
Even where they employ ostensibly biological markers, such categories always depend on extensive social organization, belief, and enforcement. Durable inequality among categories arises because people whocontrol access to value-producing resources solve pressing organizational problems by means of categorical distinctions. Inadvertently or otherwise, those people set up systems of social closure, exclusion, and control. Multiple partiesnot all of them powerful, some of them even victims of exploitationthen acquire stakes in those solutions. Variation in the form and durability of inequality therefore depends chiefly on the nature of the resources involved, the previous social locations of the categories, the character of the organizational problems, and the configurations of interested parties.
Through all these variations, we discover and rediscover paired, recognized, organized, unequal categories such as black/white, male/female, married/unmarried, and citizen/noncitizen. The dividing line between such categories usually remains incomplete in two regards: first, some people (persons of mixed race, transsexuals, certified refugees, and so on) do not fit clearly on one side of the line or the other; and, second, in many situations the distinction between the members of any particular pair does not matter. Where they apply, however, paired and unequal categories do crucial organizational work, producing marked, durable differences in access to valued resources. Durable inequality depends heavily on the institutionalization of categorical pairs.Roots of Categorical Inequality
How and why does the institutionalization of categorical pairs occur? Since the argument is unfamiliar and complicated, it may help to lay out its major elements and their causal connections even before defining crucial terms. The list will serve as a preliminary map of the wilderness this book will explore:
1. Paired and unequal categories, consisting of asymmetrical relations across a socially recognized (and usually incomplete) dividing line between interpersonal networks, recur in a wide variety of situations, with the usual effect being the unequal exclusion of each network from resources controlled by the other.
2. Two mechanisms we may label exploitation and opportunity hoarding cause durable inequality when their agents incorporate paired and unequal categories at crucial organizational boundaries.
3. Two further mechanisms we may title emulation and adaptation reinforce the effectiveness of categorical distinctions.
4. Local categorical distinctions gain strength and operate at lower cost when matched with widely available paired and unequal categories.
5. When many organizations adopt the same categorical distinctions, those distinctions become more pervasive and decisive in social life at large.
6. Experience within categorically differentiated settings gives participants systematically different and unequal preparation for performance in new organizations.
7. Much of what observers ordinarily interpret as individual differences that create inequality is actually the consequence of categorical organization.
8. For these reasons, inequalities by race, gender, ethnicity, class, age, citizenship, educational level, and other apparently contradictory principles of differentiation form through similar social processes and are to an important degree organizationally interchangeable.
Whatever else it accomplishes, the book will make dear what is at issue in such an organizational view of inequality-producing mechanisms. At a minimum, it will challenge other analysts to clarify the causal mechanisms implied by their own preferred explanations of durable inequality and then to search for evidence that those causal mechanisms are actually operating.
Although the word "organization" may call to mind firms, governments, schools, and similar formal, hierarchical structures, I mean the analysis to encompass all sorts of well-bounded dusters of social relations in which occupants of at least one position have the right to commit collective resources to activities reaching across the boundary. Organizations include corporate kin groups, households, religious sects,bands of mercenaries, and many local communities. Durable inequality arises in all of them. All of them at times incorporate categorical distinctions originating in adjacent organizations.
Humans invented categorical inequality millennia ago and have applied it to a wide range of social situations. People establish systems of categorical inequality, however inadvertently, chiefly by means of these two causal mechanisms:
Exploitation, which operates when powerful, connected people command resources from which they draw significantly increased returns by coordinating the effort of outsiders whom they exclude from the full value added by that effort.
Opportunity hoarding, which operates when members of a categorically bounded network acquire access to a resource that is valuable, renewable, subject to monopoly, supportive of network activities, and enhanced by the network's modus operandi.
The two mechanisms obviously parallel each other, but people who lack great power can pursue the second if encouraged, tolerated, or ignored by the powerful. Often the two parties gain complementary, if unequal, benefits from jointly excluding others.
Two further mechanisms cement such arrangements in place: emulation, the copying of established organizational models and/or the transplanting of existing social relations from one setting to another; and adaptation, the elaboration of daily routines such as mutual aid, political influence, courtship, and information gathering on the basis of categorically unequal structures. Exploitation and opportunity hoarding favor the installation of categorical inequality, while emulation and adaptation generalize its influence.
A certain kind of inequality therefore becomes prevalent over a large population in two complementary ways. Either the categorical pair in questionmale/female, legitimate/illegitimate, black/white, citizen/noncitizen, and so onoperates in organizations that control major resources affecting welfare, and its effects spread from there; or the categorical pair repeats in a great many similar organizations, regardless of their powerIn the first case, organizations that produce work and wield coercive powercorporations and states, plantations and mercenary forces, textile mills and drug rings, depending on the contexttake pride of place because they ordinarily control the largest concentrations of deployable resources within large populations. In some settings of ideological hegemony, religious organizations and their own categorical distinctions can also have similar effects on inequality around them.
In the second case, households, kin groups, and local communities hold crucial positions for two reasons: within a given population, they form and change according to similar principles, and they strongly influence biological and social reproduction. Gender and age distinctions, for example, do not ordinarily separate lineages from one another, but the repetition of these distinctions in many lineages lends them influence throughout the population. The basic mechanisms that generate inequality operate in a similar fashion over a wide variety of organizational settings as well as over a great range of unequal outcomes: income, wealth, power, deference, fame, privilege, and more.
People who create or sustain categorical inequality by means of the four basic mechanisms rarely set out to manufacture inequality as such. Instead they solve other organizational problems by establishing categorically unequal access to valued outcomes. More than anything else, they seek to secure rewards from sequestered resources. Both exploitation and opportunity hoarding provide a means of doing so. But, once undertaken, exploitation and opportunity hoarding pose their own organizational problems: how to maintain distinctions between insiders and outsiders; how to ensure solidarity, loyalty, control, and succession; how to monopolize knowledge that favors profitable use of sequestered resources. The installation of explicitly categorical boundaries helps to solve such organizational problems, especially if the boundaries in question incorporate forms of inequality that are already well established in the surrounding world. Emulation and adaptation lock such distinctions into place, making them habitual and sometimes even essential to exploiters and exploited alike.
To be sure, widely applicable categories accumulate their own histories and relations to other social structures: male/female distinctionshave acquired enormous, slow-moving cultural carapaces yet reappear within almost all social structures of any scale, whereas in the United States the distinction Hispanic/white remains a disputed, politically driven division of uncertain cultural content. Such categorical pairs therefore operate with characteristic differences when imported into new settings. The distinction citizen/foreigner, for instance, does a variety of organizational workseparating temporary from long-term employees, differentiating access to public benefits, managing rights to intervene in political processes, and so onbut everywhere and always its existence and effectiveness depend on the present capacity of a relatively centralized government. The power of a differentiator based on membership or nonmembership in a political party (notable cases being communist parties in state socialist regimes) similarly depends on the existence of a hegemonic party exercising extensive state power and controlling a wide variety of valued resources.
Divisions based on preference for sexual partnersgay, lesbian, straight, and so ondepend far less on governmental structure. As compared to those who differentiate based on citizenship or party membership, those who install sexual preference as a local basis of inequality have less access to governmental backing as well as a lower likelihood of governmental intervention. Sexual preference distinctions, however, do import extensive mythologies, practices, relations, and understandings that significantly affect how the distinctions work within a new setting.
Categorical inequality, in short, has some very general properties. But one of those properties, paradoxically, is to vary in practical operation with the historically accumulated understandings, practices, and social relations already attached to a given set of distinctions.
Consider some quick examples. Josef Stalin knits together an effective political machine by recruiting ethnically identified regional leaders, training them in Moscow, making them regional party bosses, and giving their ethnic identifications priority within semiautonomous political jurisdictions. When the Soviet center later relaxes its grip, political entrepreneurs within regions mobilize followings around those ethnic identifies, others mobilize against them, and ostensibly age-old ethnic conflicts flame into civil war.
Again, the founder of a small manufacturing firm, following models already established in the trade, divides the firm's work into clusters of jobs viewed as distinct in character and qualifications and then recruits workers for those jobs within well-marked categories. As turnover occurs and the firm expands, established workers pass word of available jobs among friends and relatives, collaborating with and supporting them once they join the work force. Those new workers therefore prove more reliable and effective than others hired off the street, and all concerned come to associate job with category, so much so that owner and workers come to believe in the superior fitness of that category's members for the particular line of work.
Another case in point. Householders in an urban neighborhood build up a precarious system of trust on the basis of common backgrounds and shared relations to third parties, live with persons and property at risk to that system of trust, and then react violently when newcomers whom they cannot easily integrate into the same networks threaten to occupy part of the territory. In the process, members of the two groups elaborate compelling stories about each other's perfidy and utter incompatibility.
Members of an immigrant stream, finally, peddle craft goods from their home region on big-city streets, and some of them set up businesses as suppliers, manufacturers, or retail merchants. New immigrants find work in the expanding trade, and not only an immigrant niche but an ethnically specific international connection provides exclusive opportunities for the next generation. In all these cases, organizational improvisations lead to durable categorical inequality. In all these cases, but with variable weight, exploitation and opportunity hoarding favor the installation of categorical inequality, while emulation and adaptation generalize its influence.
When it comes to the determinants of durable inequality, are these special cases or the general rule? This book gives reasons for thinking that categorical inequality in general results from varying intersections of exploitation, opportunity hoarding, emulation, and adaptation. It goes farther, claiming that much of the inequality that seems to result from individual or group differences in ability actually stems from the same causes: Authoritatively organized categorical differences in current performance (e.g., categorically differentiated cooperation or sabotage by fellow workers, subordinates, and supervisors)
Authoritatively organized categorical differences in rewards for performance (e.g., systematically lower pay for blacks than for whites doing similar work)
Authoritatively organized differences in the acquisition of capacities for performance (e.g., categorically segregated and unequal schools)
It also argues that the social mechanisms which generate inequality with respect to a wide range of advantageswealth, income, esteem, protection, power, and moreare similar. Although historical accumulations of institutions, social relations, and shared understandings produce differences in the day-to-day operation of various sorts of categories (gender, race, citizenship, and so on) as well as differences in various sorts of outcomes (e.g., landed wealth versus cash income), ultimately interactions of exploitation, opportunity hoarding, emulation, and adaptation explain them all.
Nutrition turns out to provide a useful general model for categorical inequality, since in most settings feeding differs with categorical membership, and since in many cases the cumulative effects of feeding elsewhere help to explain categorical differences in performance in the current case. In direct parallel, the information and social ties that individuals and groups can currently acquire differ categorically, but previous categorical experience also strongly affects the information and social ties these individuals and groups already have at their disposal, not to mention the means they have of acquiring new information and social ties. Unequal treatment of females and males in a wide range of social lives creates female/male differences in the qualifications and social ties prospective workers bring to workplaces; those differences interact with (and generally reinforce) gender distinctions built into the allocation and supervision of work.
Again, categorically differentiated family experience strongly affects children's school performance and teachers' evaluations of that performance, which in turn channel children into categorically differentiated,career-shaping educational streams (Hout and Dohan 1996; Taubman 1991). To the extent that teachers, employers, public officials, and other authorities differentiate their responses to performances categorically, they contribute to durable, authoritatively organized categorical differences. More generally, apparent third parties to the inequality in questionstate officials, legislatures, owners of firms, and other powerholderssignificantly influence the operation of categorical inequality and sometimes take the initiative in creating it. Authorities do, in fact, frequently solve their own organizational problemshow to sort students, whom to hire, what rights to honorin categorical ways.
Feelings of identity, on one side, and intergroup hostility, on the other, may well accompany, promote, or result from the use of categorical differences to solve organizational problems. But the relative prevalence of such attitudes plays a secondary part in inequality's extent and form. Mistaken beliefs reinforce exploitation, opportunity hoarding, emulation, and adaptation but exercise little independent influence on their initiationor so I will argue. It follows that the reduction or intensification of racist, sexist, or xenophobic attitudes will have relatively little impact on durable inequality, whereas the introduction of certain new organizational formsfor example, installing different categories or changing the relation between categories and rewardswill have great impact.
If so, the identification of such organizational forms becomes a significant challenge for social scientists. It also follows that similar organizational problems generate parallel solutions in very different settings, in articulation with very different sets of categories. Thus matches of positions with categories, and the justifications for such matches, vary much more than recurrent structural arrangementsfor example, when similar clusters of jobs acquire contrasting racial, ethnic, or gender identifications in different labor markets. Causal mechanisms resemble each other greatly, while outcomes differ dramatically, thus inviting very different rationalizations or condemnations after the fact. Social scientists dealing with such durable forms of inequality must hack through dense ideological overgrowth to reach structural roots.
Obstacles to Understanding
The essential machete work presents a serious challenge. The literature is vast, evidence mixed, current controversy therefore intense. My personal expertise falls laughably short of sufficing for the effort. Although my ideas about exploitation spring from the Marxist tradition, I have no talent or inclination for the sorts of point-by-point critique and reconstruction of Marxist models that John Roemer, Jon Elster, Samuel Bowles, Herbert Gintis, Howard Botwinick, and Melvin Leiman have undertaken. I cannot hope to provide here a comprehensive review and synthesis of current thinking concerning inequality.
Nor does it seem profitable to proceed chiefly by attacking established models of status attainment, gender inequality, or segmented labor markets, since the critique of each model would require a separate move onto its own terrain, and since I hope to identify the common ground of these ostensibly incompatible accounts of inequality rather than destroying them. My self-appointed task is instead to address the problems that emerge from the crossing of these literatures. Serious trouble begins when we try to synthesize understandings of these different kinds of inequality, when we move from description to explanation, when we search for the actual causal mechanisms that produce, sustain, or alter durable inequality. Trouble comes in four packages labeled particularism, interaction, transmission, and mentalism.
First, particularism. Observers often ground explanations for each form of inequality separately in perennial but peculiar forces. Each one seems sui generis, constituting its own mode of existence. If sexism springs from age-old patriarchy, racism from the heritage of slavery, denigration of noncitizens from xenophobic state traditions, however, it is hard to see why the mechanisms of inclusion and exclusion in all these cases have such striking resemblances. They must have more common causal properties than particularistic accounts suggest.
Our second trouble comes from the weakness of all available explanations for the interaction among various forms of categorical inequality. Despite illuminating analyses of ethnic niche formation and variable principles of citizenship taken one at a time, no one has provided acompelling explanation for what actually occurs: simultaneous differentiation of jobs and entrepreneurial niches by gender, race, ethnicity, and citizenship. What intersection of employers' and workers' preferences, for example, could possibly explain sharp segregation in all these regards at once, not to mention the interchangeability of one basis of segregation for another? How do similar sets of jobs end up all-female in one setting, all-black or all-immigrant in another?
Trouble number three concerns the transmission of categorical inequality to new members of the related categories. Do instantaneous configurations of interest or impulse that plausibly seem to account for the short-run creation of inequality continue at work from one generation to the next, or do some other mechanisms congeal categories that have unequal relations? Do genes and shared environments so powerfully reproduce individual propensities and capacities? In none of these well-documented fields do we have a convincing explanation of inherited inequality.
Finally, mentalism which relies in the last instance on shared interests, motivations, or attitudes as the bases of inegalitarian institutionsalso causes serious trouble. Resort to mental states as fundamental sources of inequality leaves mysterious the cause-effect chains by which these states actually produce the outcomes commonly attributed to themespecially considering how rarely we humans accomplish the precise ends we consciously pursue (Merton 1936, 1989). If collectively a whole population sustains a set of preferences simultaneously ordered by gender, race, ethnicity, and citizenship, whose mental processes contain these preferences, how do they order the preferences, and what translates preferences into a wide range of structural inequalities?
Let us look more closely at the last difficulty, which besets much of today's social science, not just the study of inequality. Most people seeking to explain any sort of social process choose among three ontological foundations; all three presume the existence and centrality of self-propelling essences (individuals, groups, or societies). Two of the ontologies center on the mental processes of such essences. First, methodological individualism presumes that social life results chiefly or exclusively from the actions of self-motivated, interest-seeking persons. Second,phenomenological individualism posits conscious minds as the ultimate social reality; with sufficient doubt about the possibility of reliable communication among minds, phenomenological individualism becomes solipsism.
The third ontology eschews mentalism, but at the price of other high-risk assumptions about social reality. Systems theories impute self-maintaining logics to social structures, from groups, organizations, or institutions to that big, vague structure that analysts refer to as "society." Some theorists, to be sure, combine two of these ontological foundations, as in Emile Durkheim's recurrent image of an individual face to face with a society or Alfred Marshall's representation of a calculating buyer or seller who confronts an impersonal market. But alone or in combination, methodological individualism, phenomenological individualism, and systems theories all nevertheless assume self-sustaining essences, whether individual, collective, or both.
A fourth possibility exists, however, a possibility assuming not essences but bonds: relational models of social life beginning with interpersonal transactions or ties. Since Charles Peirce and Georg Simmel, relational models have haunted social science; phenomenological individualists such as George Herbert Mead who wanted to represent the effects of social interaction on consciousness and action have recurrently heard relational voices. The unjustly neglected institutional economist John R. Commons (1934) insisted sixty years ago that economics should begin its analyses with transactions, not individuals. Economists, alas, did not heed him.
More recently, Norbert Elias's "configurations" were largely relational. Although Elias tended to ground his configurations in shared attitudes, he stressed collective connections among the social positions involved. In 1965, for example, Elias and John Scotson published a study of two nearly identical neighborhoods in Winston Parva (a locality near Leicester) whose members had organized hostile, unequal conceptions of each other. In the study's introduction, Elias remarked:
At present the tendency is to discuss the problem of social stigmatisation as if it were simply a question of people showing individually apronounced dislike of other people as individuals. A well-known way of conceptualising such an observation is to classify it as prejudice. However, that means perceiving only at the individual level something which cannot be understood without perceiving it at the same time at the group level. At present one often fails to distinguish between, and relate to each other, group stigmatisation and individual prejudice. In Winston Parva, as elsewhere, one found members of one group casting a slur on those of another, not because of their qualities as individual people, but because they were members of a group which they considered collectively as different from, and as inferior to, their own group. Thus one misses the key to the problem usually discussed under headings such as "social prejudice," if one looks for it solely in the personality structure of individual people. One can find it only if one considers the figuration formed by the two (or more) groups concerned or, in other words, the nature of their interdependence. (Elias and Scotson 1994, xx )
In Winston Parva, the study shows, the old-timers of one neighborhood had greater cohesion, hence more strategic power, than the newcomers of the other; old-timers collectively translated their organizational advantage into successful stigmatization of the neighboring population. Thus Elias reached at least halfway to a full-fledged relational account.
These days the program of "structural sociology" as variously advocated by such theorists as Mark Granovetter, Alejandro Portes, Pierre Bourdieu, Paul DiMaggio, and Harrison White most aggressively advances relational models (see, e.g., Portes 1995; Powell and DiMaggio 1991; Wellman and Berkowitz 1988). In economics, institutionalists also make allowances, generally more grudging, for relational effects (e.g., Akerlof 1984; Jacoby 1990; Lazonick 1991; North 1991; Osterman 1993; Simon 1991; Williamson 1991). In his Foundations of Social Theory (1990), the late James Coleman feinted repeatedly toward relational accounts of norms, commitments, and similar phenomena but pulled his punches as they approached the target. Although his verbal accounts mentioned many agents, monitors, and authorities who influenced individual actions, his mathematical formulations tellingly portrayed a single actor's computations rather than interactions among persons.
Structural and institutional analyses of relations clarify and emphasize the significance of culture in social life. Instead of imagining cultureas an autonomous sphere in which ideas change ideas, which then constrain behavior, structural and institutional analyses treat culture as shared understandings and their representations; actors operate within frames of understanding constructed by previous interactions, anticipating one another's responses on the bases of those frames, and modifying their strategies as a consequence of shared experiences. In such a view, culture intertwines unceasingly with social relations; culture and structure are simply two convenient abstractions from the same stream of transactions.
The four ontologies lead characteristically to different ways of accounting for categorical inequality. Following methodological individualism, analysts typically treat inequality by gender, race, ethnicity, or citizenship as a special case of inequality in general, a case in which (1) members of a category come to share attributes (e.g., educational levels) that place them in similar relations to markets and/or (2) other participants in markets build categorically defined preferences (e.g., an aversion to working with foreigners) into the utility schedules guiding their decisions Methodological individualists who seek to explain social inequality have so far faced an insurmountable obstacle. Their causal mechanisms consist of mental events: decisions. But they have not formulated a plausible theory of how such mental events produce their consequences in the always erratic behavior of human beings, much less in the complexities of social structure.
Phenomenological individualists find it easier to imagine that categories themselves have meaning and that people express their own identities by acting categorically; consumption of commodities and services, for example, becomes a way of broadcasting one's self-conception to the world at large. That sort of phenomenological individualism, however, has produced no coherent account of the interactions among conscious states of different actors or of the processes by which such states produce alterations in social structure.
Systems theorists generally derive categorical distinctions from collective relations between members of categories and some larger social structurefor example, explaining gender differences by their expression of society-wide values or their service to the reproduction of thewhole system. It has proved impossible either to identify those relations to larger structures concretely or to assemble convincing evidence for functional explanations of this kind. Despairing of functional explanations, other systems theorists have commonly derived categorical distinctions from a vague, autonomous entity called "culture" or even "the culture." Such accounts relabel the phenomenon instead of explaining it. For the explanation of durable inequality, systems theories look like a cul-de-sac.
Relational analysis, as we shall see in detail, typically treats categories as problem-solving social inventions and/or by-products of social interaction (Elster 1983, 25-88). Relational analysts characteristically conceive of culture as shared understandings that intertwine closely with social relations, serving as their tools and constraints instead of constituting an autonomous sphere. Strongly relational analysis remains a minority movement in social science as a whole; individualisms and holisms continue to reign. In the choice between essences and bonds, nevertheless, I want to hold high the banner of bonds. I claim that an account of how transactions clump into social ties, social ties concatenate into networks, and existing networks constrain solutions of organizational problems clarifies the creation, maintenance, and change of categorical inequality.
Let me state that claim with care. Since the fading of systems theories a generation ago, methodological individualism and phenomenological individualism have dominated analyses of inequality. Individualistic analyses have in that time accomplished a great deal. They have documented the outcomes (e.g., stark racial differences in income and wealth) that any adequate account of inequality must explain. They have also ruled out all commonly voiced one-cause explanations of in-equalitygenetic capacity alone, effort alone, educational achievement alone, point-of-hire discrimination alone, and more. They have thereby greatly clarified what any sound theory of inequality must explain. They have, however, relied on obscure, implausible, or insufficient causal mechanisms grounded in individual experience and action. They have centered thinking about inequality on the image of individuals with variable attributes who pass through a screening process that sorts them according to those attributes into positions that give them differentialawards. In various explanations, these attributes may include human capital, ambition, educational credentials, gender, race, or even personal connections, but they remain individual properties. The screening processes often considered range from market competition to employers' selection of workers on the basis of prejudice or favoritism, but they always involve selection among individuals as a function of those individuals' attributes.
Take an example of first-rate research on inequality. A University of California group has published a sustained theoretical, methodological, and factual critique of Richard Herrnstein and Charles Murray's well-known 1994 book The Bell Curve. The Herrnstein-Murray book argues, among other things, that in the contemporary United States innate intelligence deeply affects success or failure, that inherited class and racial disparities in intelligence account for the major part of differential accomplishment by children born into different classes and races, and that remedial measures such as affirmative action will therefore inevitably fail or will even compound the inequalities they are designed to mitigate. The California group's critique is judicious, skilled, and ultimately devastating for the Herrnstein-Murray argument. In company with abundant material from elsewhere, it deploys the very same body of evidence on which The Bell Curve is based to identify errors in the Herrnstein-Murray analysis and to reach distinctly different and better-reasoned conclusions.
The California study makes a powerful contribution to our understanding of American inequality and to the destruction of widely held misunderstandings. But consider the authors' summary of their alternative explanation:
Children may start out with different "natural" advantages useful for economic advancement (and such advantages probably include far more than just the sort of narrow intelligence psychometricians dwell on, advantages such as energy and good looks). But children certainly do start off with different social advantages, some with more parental resources and better conditions in their communities than others, and some with the advantage of being male. Children in better-off families and better-off places then receive better schooling and develop theircognitive skills further. Having good schooling and skills, added to the original advantages of gender, family, and neighborhood, combines with contemporary circumstances, such as being married and living in an economically booming area, to reduce substantially people's risk of poverty. Young adults who have lost outin family advantages, in earlier community conditions, in gender, in schooling, or in current community conditionssuffer a heightened risk of poverty. (Fischer et al. 1996, 93)
Every statement in this summary carries conviction. When we examine it more closely, however, the causal links it identifies run something like this:
Community location, household position, and parental resources affect (a ) individual cognitive skills and (b ) quality of schooling, which (interacting with gender) jointly affect (c ) individual educational outcomes and (d ) certain other unspecified adult characteristics. Educational outcome, gender, and other unspecified adult characteristics affect economic outcomes, notably the likelihood of being poor.
Such an argument derives collective outcomes (e.g., racial differences in poverty) entirely from individual effects. It also fails to specify the causal mechanisms by which community location, household position, parental resources, and gender produce educational outcomes or the other relevant adult characteristics. Nor does it say how and why educational accomplishment, gender, and other characteristics produce their sorting effects. An individualistic framework leads the California authors to slight organizational, relational, and collective processes.
The authors actually say as much:
And yet we have not accounted for most of the inequality in income. Perhaps some of the remaining 63 percent of unexplained variation can be accounted for by other, unmeasured attributes of individualsenergy, looks, charm, whateveror by other, unmeasured attributes of their social situationsgrandparents' legacies, social contacts, the industry they work in, and so onor, as Christopher Jencks has suggested, by simple luck. But much of that remaining inequality can only be understood by leaving the individual level of analysis and looking at the social structure of inequality. (Fischer et al. 1996, 99)
They then attempt to deal with that deficiency by calling attention to institutional influences on opportunity and mobility: government policies concerning taxation, investment, redistribution, and services; the organization and operation of schools; the activities of unions and employers' associations; residential segregation; categorical hiring and other categorical forms of discrimination. All these factors do, indeed, enter into the production of durable inequality. The questions are how and why.
Those questionshow and why?drive my inquiry. The crucial causal mechanisms behind categorical inequality, I argue, do not consist of individual mental events, states of consciousness, or self-sustaining actions of social systems. They operate in the domains of collective experience and social interaction. The remainder of this book explicates and defends that claim. For the most part, my analysis accepts the definition of what analysts of inequality must explain that has emerged from a generation of individualistic investigation. But it complements and clarifies the findings of individualistic analyses by looking at the social structure of inequality.
We pay a price for concentrating on well-documented outcomes. Recent students of inequality under capitalism have, unsurprisingly, focused on wages, a topic that lends itself both to measurement and to explanation in individual terms. They have neglected wealth, health, nutrition, power, deference, privilege, security, and other critical zones of inequality that in the long run matter more to well-being than wages do. They have also drawn evidence disproportionately from wage-paying firms, while giving little attention to family enterprises, contracting, the informal economy, and other settings whose categorically differentiated personnel and operations contribute significantly to aggregate differences in well-being. To document such differences, explain them, and relate them to each other remains a major task for analysts. To pursue it here, however, would enormously complicate and lengthen an already dense analysis. Pages to come take up these other forms of inequality when possible, for example, in discussions of South African racial divisions, power within American health care, and nationalism in the contemporary world. Yet on the whole I leave their complexities for later inthe hope that the book's analysis will provide a model for their treatment.
Although concern about inequalities in contemporary capitalist countriesespecially my own country, the United Statesmotivates my inquiry, my plan is not to close in immediately on today's inequalities. Instead I am pursuing an indirect strategy, stepping back from current American discussions of comparable worth, white racism, or immigrant/native differentials to place durable categorical inequality in historical, comparative, and theoretical perspective. A relational view identities common causal mechanisms beneath the bewildering variety of concrete inequalities.Elements of Inequality
Before undertaking the necessary reconstruction, however, let us think about inequality as such. Human inequality in general consists of the uneven distribution of attributes among a set of social units such as individuals, categories, groups, or regions. Social scientists properly concern themselves especially with the uneven distribution of costs and benefitsthat is, goods, broadly defined. Relevant goods include not only wealth and income but also such various benefits and costs as control of land, exposure to illness, respect from other people, liability to military service, risk of homicide, possession of tools, and availability of sexual partners. Students of social inequality have paid little attention to the uneven distribution of other attributes such as genetic traits and musical tastes except as they correlate with the uneven distribution of goods in this broad sense.
Goods vary in the extent to which they are autonomous (observable without reference to outside units, as in accumulations of food) or relative (observable only in relation to other units, as in prestige). Wealth, income, and health exemplify autonomous goods, while prestige, power, and clientele exemplify relative goods. (Some analysts prefer to call relative goods "positional," on the grounds that they attach to positions rather than to persons, but that usage draws attention away fromtheir relational character.) On the whole, inequalities with respect to autonomous goods reach greater extremes than inequalities with respect to relative goods.
Analysis of exploitation by the elite, opportunity hoarding by the nonelite, emulation, and adaptation makes it dear that autonomous and relative goods depend intimately on each other. Although people come to value them for their own sakes, relative goods generally occupy a subordinate, derivative position: they serve as a means of creating or maintaining categorical inequality with respect to autonomous goods. Possession of prestige, power, clientele, and status-marking goods then justifies the superior position of favored categories ex post facto, just as the perquisites of favored categories give autonomous goods such as well-built housing, luxurious automobiles, comfortable workspaces, fine foods, good liquor, or rich entertainment the patina of relative goods as well. The chief reversals in the priority of autonomous over relative goods occur in such public displays as potlatch, charitable donations, and ostentatious weddings, where wealthy or powerful people incur great expenditures in the short run to mark their superiority over other people. Even there, successful displaysfor example, magnates parading great clienteles in the public rituals of Renaissance Florencecharacteristically enhance the longer-run advantages of those who mount them (Paige and Paige 1981; Trexler 1981).
I certainly did not discover the interaction between autonomous and relative goods. Pierre Bourdieu has spent much of his career exploring it, with his analytic division among economic, cultural, and social capitals representing the interdependence of autonomous goods narrowly conceived, valued information, and the social ties that provide differential access to those goods and information (Bourdieu 1979; Bourdieu and Wacquant 1992, 118-119; Buchmann 1989, 31-42). When ever-relational Karl Marx traced back relative goods (not his term!) to origins in relations of production, he likewise portrayed prestige, power, clientele, and possession of status-marking goods as instruments and products of categorically based exploitation. Categorical inequality with respect to autonomous goods gains strength from and generates parallel differences in relative goods.
How can we judge the equality or inequality of two social unitspositions, persons, categories, organizations, networks, countries? Estimating the inequality of any set of social units presents three major problems: to identify and bound the units under comparison, to weigh the importance of different goods, and to decide whether the weighted differences are "large" or "small." Generally speaking, all three judgments require a theory of the larger social structures in which the units are embedded.
The difficulty compounds with summary measurement of inequality and its changes among many unitsfor example, among all households in a national population (as in many analyses of long-term change) or among all the world's states (as in many world-system analyses). In such cases, analysts usually adopt two linked strategies: first choosing a single criterion good (such as current income) that seems to correlate with a number of other inequalities, and then comparing the actual distribution of that good with a standard of absolutely equal distribution. Such widely used devices as the Gini index and the Duncan dissimilarity index illustrate the combined strategy. In this approach, inequality implicitly becomes a one-dimensional phenomenon. Individual units vary in position along the chosen dimension.
Although analysts sometimes apply the term loosely to all sorts of inequality, stratification properly designates the rare form of disparity that clusters social units by layers, or strata, which are homogeneous with respect to a wide range of goods (both autonomous and relative) and which occupy a single, well-defined rank order. A true system of stratification resembles a pyramidal skyscraper, with its summit and base, its distinct levels, its elevators and stairways for movement from level to level, and its array of multiple graded niches.
One of my own great teachers, I fear, introduced abiding mischief into sociological discussions of inequality and mobility. Pitirim Sorokin's Social Mobility, first published in 1927, popularized not only representations of inequality as stratification but also ideas of vertical and horizontal mobility. Sorokin said explicitly that "social stratification means the differentiation of a given population into hierarchically superposed classes" (Sorokin 1959, 11). Stratification implies social strata:upper, middle, and lower, or some other bounded vertical division. Thus Sorokin committed his followers to the suppositions of continuous, consistent hierarchies transecting whole populations, of discrete individual locations within those hierarchies, and of well-marked boundaries between classes.
Summarizing the causes of stratification, moreover, Sorokin gave them a distinctly individualistic cast:
First, the very fact of living together; second, innate differences of individuals, due to the differences in the complements of their chromosomes; third, differences in the environment in which individuals are placed since the moment of their conception. (Sorokin 1959, 337)
Although the first cause, the "very fact of living together," sounds groupish, it turns out to consist for Sorokin of an inevitable division between (few) leaders and (many) led. Intergroup and interpersonal processesof struggle, conquest, or otherwiseplay no part.
Sorokin's analysis of vertical and horizontal mobility compounds the difficulty of judging inequality by fostering the illusion of a continuous, homogeneous two-dimensional grid within which individuals and aggregates of individuals occupy specific cells and move along geometric paths. The seductive spatial metaphor misleads analysts to the extent that inequality consists of organized ties among groups, categories, or individuals; that different forms of inequality order the same groups, categories, or individuals differently; that changes in patterns of inequality result from intergroup processes. Since all these conditions actually obtain, sociologists would have benefited if Sorokin had never mentioned vertical and horizontal mobility.
Large organizations such as armies, to be sure, sometimes stratify internally: they create bands of homogeneous rank that reach across the whole organization, establish segregation among ranks, and perform rituals of succession from rank to rank. As a consequence, localities such as company towns and military bases, which depend on large, stratified organizations, likewise fall into ranked strata. But no general population larger than a local community ever maintains a coherent system of stratification in a strong sense of the word; even the so-called caste systemof India accommodated great variation in rank orders from village to village. In general, rank orders remain inconsistent, apparent strata contain considerable heterogeneity, and mobility blurs dividing lines. Stratification is therefore a matter of degree.
Inequality is likewise a matter of degree, but for the opposite reasonbecause it is ubiquitous. Whatever the criterion of equivalence, no two social units ever command precisely equivalent arrays of goods for more than an instant. Possession of different sorts of goods, furthermore, couples loosely enough that the same social unit moves in several directions simultaneously; inequality is always in flux. Any unified, fixed model of inequalityand, a fortiori, of stratificationthat we impose on social life caricatures a dynamic reality, etches a Gillray portrait of social interaction. As with other useful caricatures, then, the secret is to sketch a model that brings out salient features of its object, but never to confuse model with reality.
Since the late nineteenth century, individualistic models of inequality have crowded out categorical models. From Adam Smith to Karl Marx, classical economists generally analyzed categories and relations among them: chiefly land, labor, and capital for Smith, capital and labor alone for Marx. They examined returns to these factors considered collectively and situated socially rather than returns to individual effort. Discussing returns to labor, for example, Smith reasoned:
What are the common wages of labour, depends everywhere upon the contract usually made between those two parties, whose interests are by no means the same. The workmen desire to get as much, the masters to give as little as possible. The former are disposed to combine in order to raise, the latter in order to lower the wages of labour. It is not, however, difficult to foresee which of the two parties must, upon all ordinary occasions, have the advantage in the dispute, and force the other into a compliance with their terms. The masters, being fewer in number, can combine much more easily; and the law, besides, authorises, or at least does not prohibit their combinations, while it prohibits those of the workmen. (Smith 1910 , 1:58-59)
Although Smith certainly saw market conditionsin this case, especially rates of growth in demand for laboras crucial to the advantageof one party or the other, he reasoned about categories, groups, institutions, and ties. Those ties emphatically included collective, categorical, unequal power.
The neoclassical revolution, however, diverted economic attention from categories to individuals and markets. As Suzanne Shanahan and Nancy Tuma remark:
Theories about allocations among social groups dominated economic and sociological thought in ... the eighteenth and nineteenth centuries. With growing industrialization and the development of Fordist labor organization in the nineteenth and twentieth centuries, the rise of the individual as the unit of analysis swept through the social sciencesnot only psychology and economics, but even sociology, the field that proclaims to study social groups and societies. Whatever the defects in the classical theories of factor distribution, it is telling, we think, that by the middle of the twentieth century, social scientists had almost completely switched their gaze from intergroup distributions to interindividual distributions. (Shanahan and Tuma 1994, 745)
Their switched gaze focused social scientists on situations of choice among relatively well-defined alternatives within known constraints on the basis of clear preference criteria. It meant they acquired little knowledge of the processes by which such choices produced consequences, of indirect and environmentally mediated effects, of situations of choice that did not meet these stipulations, of the influence of shared meanings over action.
On the assumption that the market itself operates impartially, since the late nineteenth century economists and their imitators explaining categorical phenomena have usually tried to reduce them to individual causes and effects. It has become a habit: faced with male/female differences in wages, investigators look for average human-capital differences among the individuals involved. Noticing that school performances of children correlate with the social positions of their parents, researchers attribute those differences in performance to "family background" rather than considering that teachers and school officials may shape those performances by their own categorical responses to parental social positions. Encountering racial differences in job assignments,researchers ask whether members of distinct racial categories are distributed differently by residential location. Uncovering evidence of sharp ethnic differences in industrial concentration, analysts begin to speak of discrimination only when they have factored out individual differences in education, work experience, or productivity.
Familiarity has made these methodological precautions seem natural. Yet few students of social processes would employ logically similar procedures in trying to determine whether or why Jews and Catholics have differing views of divinity, whether some geographic boundary really separates French people from Spaniards, or why white South Africans, on average, enjoy much higher incomes and greater political power than their black fellow citizens. In such cases, we generally assume that category membership and collective ties to nonmembers, rather than individual variation in propensities and capacities, produce group differences. Yet in the world of work and labor markets, the presupposition prevails that inequality results from variation at the individual level. Ruth Milkman and Eleanor Townsley sum up the literature on gender discrimination:
Typically, a study will examine a variety of factors that might explain wage differences between men and women, such as education, experience, interruptions in work histories, and so forth. The unexplained residual is then attributed to "discrimination," which is implicitly presumed to be a willful act on the employer's part (or sometimes on that of the co-workers, customers, or unions). This approach, while valuable for demonstrating the existence of a serious inequality problem, fails to capture the depth with which gender segregation and the norms associated with it are embedded in the economic orderin fact, they are embedded so deeply that a willful act of discrimination is not really necessary to maintain gender inequality. (Milkman and Townsley 1994, 611)
Similarly, the idea of "statistical discrimination" (Bielby and Baron 1986; Mueser 1989) individualizes a collective process radically: it portrays an employer who avoids hiring members of a whole category on the basis of beliefs or informationhowever well foundedthat on average workers belonging to the category contribute less to productivity than their counterparts from outside the category.
Even as representations of individual derision-making, characteristic models of methodological individualism fall short. "The conventional story about individual behavior" reflects economist Michael Piore on standard economic reasoning, "is built around the notion that human actions are the product of purposive decision making in which the actors make a sharp distinction among means, ends, and the causal models which lead from one to the other." Recent theories based on pragmatic philosophy, hermeneutics, and linguistics, continues Piore, "suggest that these distinctions are at best vague and imprecise, if not completely absent, and that they emerge at all only in practice through the processes in which people first discuss the situation and then eventually act" (Piore 1996, 750; see also Conlisk 1996; Lewin 1996). The century-old move of economics from relational to individualistic accounts simplified analysts' work at the cost of losing verisimilitude.
When they adopted status attainment models of mobility and inequality, sociologists accentuated the shift from collective to individual effects. "In the most brilliant destructive paper in the history of sociology," Arthur Stinchcombe remarked some time ago,
Otis Dudley Duncan (1966) stopped research into the relation between the labor market in which fathers had attained their status and the labor market in which sons attained theirs. His solution to the difficulties of such analysis was to regard the father's achievement only as a feature of the biography of sons, to be related to other features of that biography (such as later status attainment by the son) by regression analysis or qualitative loglinear models for sons. This tradition has however given a very queer tone to the mobility literature, since it deliberately starts off by talking as if people promoted themselves instead of being promoted by employers, or as if failure and success in self-employment depended on fathers rather than on success in a modem market. (Stinchcombe 1978a, 1)
Duncan's ingenious solution greatly simplified the problem of representing mobility statistically. But the representation of son's present occupational rank as an outcome of father's occupational rank in combination with son's other characteristicsfor example, years of school completedradically individualized the mobility process whileobscuring such causes as changes in hiring practices and the formation of job-finding networks by migrants.
In an earlier review of the Christopher Jencks et al. Inequality (1972), Stinchcombe had made the essential distinction between two ways of representing inequality among paired persons: first as a difference in the positions of the two individuals with respect to similar variables, second as a characteristic of the relationship between them. "The second kind of analysis," he then pointed out, "requires the comparison of social systems (at the very least, social systems containing the pair), since data on variables describing pairs cannot be derived from data on isolated individuals" (Stinchcombe 1972, 603; for a recent review of status attainment analyses, see Breiger 1995). The Duncan solution stresses comparison of individuals with external standards rather than examining relations among individuals.
Human-capital theory offers a closely related individualistic account of inequality, with the additional twist of radical depersonalization. In strict human-capital models, neither the worker nor the worker's effort earns the rewards of work; instead, previous investments in the quality of workers command current returns. Again Stinchcombe's remark applies: such analyses rule out ties among workers or between bosses and workers as independent causes of inequality. They rely on an almost magical belief in the market's ability to sort out capacities for work.
Still, individualistic analyses of inequality have all the attractions of neoclassical economics: nicely simplified geometric analogies, reassuring references to individual derision-making, insistence on efficiency, avoidance of inconvenient complications such as beliefs, passions, culture, and history. They lend themselves nicely to retroactive rationalization; confronted with unequal outcomes, their user searches the past for individual differences in skill, knowledge, determination, or moral worth that must explain differences in rewards. These analyses fail, however, to the extent that essential causal business takes place not inside individual heads but within social relations among persons and sets of persons. That extent is, I claim, very large. If so, we have no choice but to undertake relational analyses of inequalitywhether ornot we finally couple them with individualistic elements of relevant decision processes.
Two disclaimers on that very point. First, I consider persons to possess as many identities as the number of social relations they maintain, one identity per relation, and to acquire their individuality through interactions among genetic capabilities and social experiences. Yet I also recognize the existence of sentient individuals whose actions depend mightily on their physiological functioning; all of us perform differently in fatigue than in freshness, in illness than in health, in old age than in childhood. Over the long run, any valid social science must therefore be compatible with known regularities in the operation of individual human organismsbrains, nervous systems, viscera, and all the rest. In the short run, however, our great deficit lies on the interactional side of the individual social ledger. Instead of reducing social behavior to individual decision-making, social scientists urgently need to study the relational constraints within which all individual action takes place.
My second disclaimer is intended chiefly for specialists in the study of inequality. Insiders may find baffling and irritating my tendency to draw heavily from the results of research based on individualistic assumptions and then to trash those assumptions. They may also feel that this book, without attaching specific offenses to any named offender, dismisses status attainment analysts, neoclassical economists, and specialists in wage determination as scoundrels or nitwits. Let me reassure my many friends and collaborators in these fields: I am building gratefully on their work, indeed trying to codify qualifications, objections, findings, anomalies, and hypotheses coming directly from their work.
Analysts of inequality occupy something like the position of seismographers. In the explanation of earthquakes, the recognition that the shifting of great tectonic plates beneath the earth's surface causes much of the heaving and cleaving in that surface has not made small-scale geology irrelevant. On the contrary, it has greatly clarified the causal mechanisms behind the flux of sand, gravel, rock, and soil while also providing partial accounts of such puzzling phenomena as continental drift. In current investigations of deep encounters between Asian plates and those of the Indian subcontinent beneath southern Tibet, for example,seismologists, geologists, and geophysicists are combining evidence from "CMP reflection, wide-angle reflection, broadband earthquake, magnetotelluric and surface geological data" in ways that not only use surface distributions to verify hypotheses concerning tectonic processes but also help to explain those surface distributions (Nelson et al. 1996, 1684). Seismologists draw on these complementary efforts to explain patterns of Tibetan earthquakes but cannot simply reduce those patterns to plate tectonics. Similarly, extension of relational analyses within the study of social inequality does not deny the existence of individuals or individual-level effects. It does, however, place individualistic processes in their organizational context. It does, finally, challenge any ontology that reduces all social processes to the sentient actions of individual persons.Alternatives to Individualism
Over the past few decades, the large body of research based on individualistic assumptions and adopting persons as units of analysis has done a superb job of specifying what analysts of inequality must explainfor example, by showing how much of male/female income differences springs not from unequal pay within the same jobs but from job segregation. My complaint with the literature chiefly concerns available explanations. Prevailing accounts of inequality strongly emphasize, first, the one-time decisions of powerholders to allocate rewards in a given manner and, second, the attributes and performances of individuals that attract differential rewards. Drawing dues from existing studies of wage determination, occupational careers, hiring, and labor market segmentation, I hope to show the great importance of cumulative, relational, often unnoticed organizational processes in the actual creation of durable inequality.
No doubt this way of putting it reminds you of the old saw about economics explaining how people's ordered choices produce collective effects and sociology explaining why people have no choices to make. But I urge no such deterministic view; on the contrary, my account ofinequality builds relentlessly on counterfactuals, on could-bes and might-have-beens. For social life in general, valid explanations proceed from specifications of possible configurations to specifications of circumstances differentiating those configurations from others that could, in principle, also form.
My central counterfactuals concern organizational problem-solving and collective acquisition of stakes in organizational arrangements. They concern different ways that connected, powerful people draw to their own advantage on the efforts of excluded outsiders and different ways that less powerful people form segregated niches affording them privileged access to more limited but genuine advantages. Such counterfactuals play down the importance of attitudes, prejudices, and mistaken beliefs in unequal arrangements while playing up the significance of convenience, transaction costs, and contingent opportunities. By the same token, they point to organizational innovationsrather than changes in preferences, attitudes, and personal qualificationsas a means of reducing durable inequality.
Although it dovetails with abundant recent work by institutional economists and economic sociologists, my approach, even where it stands on solid logical and empirical ground, will encounter three significant barriers to acceptance. First, as institutional economists themselves have taught us, established solutions generally have the advantage over innovations in the short run because the transaction costs of devising, perfecting, installing, teaching, and integrating new solutions to problems exceed the costs of maintaining old ways, more so to the extent that old ways articulate well with a wide range of adjacent beliefs and practices. As in organizational life, so in social science: the well-developed apparatus of individualistic analysis will not be displaced easily.
Second, and less obvious, the sort of relational analysis I am advocating clashes with the narrative mode in which people ordinarily think and speak about social processes. At least in Western countries, people learn early in life to tell stories in which self-motivating actors firmly located in space and time produce all significant changes in the situation through their own efforts. Actors in narratives need not be rational orefficient, but their own orientations cause their actions. Deliberate individual actions then cause individual reactions, which cause further deliberate individual actions, on to story's end (Bower and Morrow 1990; Somers 1992; Steinmetz 1993; Turner 1996).
Narratives feature essences, not bonds. They therefore favor individualistic analyses, whether rational-choice, phenomenological, or otherwise. The supposition that social processes actually behave as narrative would require motivates (frequent) complaints against social scientists for practicing inferior versions of the novelist's craft and (infrequent) praise of ethnographies because they "read like a novel." By-products of social interaction, tacit constraints, unintended consequences, indirect effects, incremental changes, and causal chains mediated by nonhuman environments play little or no part in customary narratives of social life. Relational analyses of inequality affront narrative common sense by insisting that just such subtle ramifications of social interaction produce and sustain unequal relations among whole categories of persons.
Proximity of social-scientific analysis to moral discourse erects a third barrier to acceptance of relational explanations for inequality. In a narrative mode, social science closely approaches the prevailing discourse of morality. That discourse judges both intentions and actions of self-motivated individuals according to certain criteria of adequacy: goodness, fairness, authenticity, or something else. Every social-scientific narrative invades moral ground, a fact that helps account for the passion such narratives often stir in the hearts of people outside the profession.
Even nonnarrative explanations for inequality, however, touch moral discourse. Doubly so, in fact, both because rejecting self-motivated actors as sufficient causes for social outcomes challenges the standard premises of moral discourse and because relational analyses invoke counterfactuals. Counterfactuals say that other arrangements were, are, or will be possible, hence that present circumstances do not embody the best of all possible worlds. If we assemble known causal links plausibly into previously unknown longer causal chains, we challenge any claims for the inevitability of currently operating chains. If we show that similar causal sequences have in fact operated elsewhere in other eras, we simultaneously strengthen the case for contingency and sharpen thespecification of what needs explaining here and now. Thus new worlds of equality or inequality unfold counterfactually from causal analyses of present circumstances.
In principle, the deployment of valid counterfactualsvalid in the sense that they incorporate causal sequences known to be possiblenot only advances the work of explanation but also equips social scientists to criticize and compare moral doctrines, political programs, and ideologies very effectively. Every such doctrine includes assertions, implicit or explicit, of possible social conditions; most also include assertions concerning paths from the present state of affairs to those possible conditions. Those assertions invite social-scientific criticism and comparison. If a politician argues that the United States should exclude Asian and Latin American immigrants in order to improve the job prospects of existing U.S. workers, social scientists may of course discuss the value premises of such a recommendation, but they have special expertise in examining the causal reasoning it involves. Is Asian and Latin American immigration actually diminishing the job prospects of American-born workers?
By the same token, we can anticipate heated resistance to causal analyses, however well founded, that contradict the possibility arguments built into cherished doctrines. Here the second and third obstacles combine. Widely accepted American moral, legal, and political doctrines enshrine the individual as a conscious, responsible, efficacious agent of her or his own actions. Here, too, individualism reigns. Anyone who locates efficacious social action in the contract rather than the signers, in the plot rather than the players, in the conversation rather than the speaker invites intuitive rejection on behalf of cherished creeds. This book will succeed if it makes credible and useful a relational account of social inequality. It will succeed doubly if it clarifies what other forms of inequality or equality are possible, and how.
I have undertaken my inquiry because I believe the intensity of capitalist inequality causes unnecessary suffering and becauseon the grounds I have just identifiedsocial scientists can help to discover means of alleviating inequality and its attendant suffering. I have written much of the book in egalitarian Sweden, where increasingly confident voices are currently urging reduction of government-guaranteedbenefits on the argument that the logic of free-flowing capital now conflicts with the premises of welfare states, thus sharpening the contradiction between redistribution and economic growth. Historical experience persuades me that abundance and equality are, on the contrary, compatible, and in some circumstances even complementary. Yet in this book I do not try to formulate remedies or to persuade readers of their efficacy. If the book helps produce better understanding of causes, effects, and crucial questions yet to answer, that will suffice for a preliminary inquiry.
One final warning. A lifetime of densely empirical work based on large (often obsessively systematic) bodies of evidence bound to particular times and places has given me an irresistible compulsion to illustrate general points with specific cases. Hardly a page of this book goes by without irruption of at least one example. Yet the book's handling of evidence differs from most of my previous work in two crucial ways First, it aims not at drawing time- and place-bound generalizations from careful comparisons of cases but at identifying very general processes that produce inequality, at singling out causal mechanisms that operate in an enormous variety of times, places, and social settings. Second, it introduces illustrations chiefly to clarify theoretical points rather than to establish their empirical generality. My previous investigations give me some confidence that the book's main assertions apply widely. Still, for the moment they stand not as empirically validated generalizations but as working hypotheses about the interplay of exploitation, opportunity hoarding, emulation, and adaptation. Those hypotheses invite verification, refinement and, no doubt, refutation.
Given the way I have constructed the argument, indeed, skeptics can challenge it at several different levels:
1. The book describes categorical differences with respect to goods and social relations so badly as to undermine its explanations of those differences.
2. Its explanations of those categorical differences fail in significant ways.
3. Its explanations work for some kinds of categories but not for others.
4. Its argument fails to explain significant differences in the operation of various sorts of categories (e.g., gender versus citizenship).
5. Dimensions of categorical difference that it does not explore extensively (e.g., in regard to political power or deference) do not behave as the argument implies they should.
6. Noncategorical inequalities among individuals are larger or weightier than the argument claims, and the argument provides no means of explaining them.
Objections 1 and 2, if both founded, would destroy the book's credibility. Objections 3, 4, and 5, if well defended, would point to elaborations or revisions of its arguments. Objection 6, if sustained, would leave the main arguments intact but reduce the scope I have claimed for them. Although I expect all six types of criticismand suppose that some versions of all six will identify weaknesses in the argument as I have stated itI will be disappointed if critics formulate effective combinations of objections 1 and 2, not surprised to face effective statements of objection 6, and actually encouraged if the book stimulates constructive variants of all six objections, especially constructive objections that advance relational explanations of inequality. A provisional synthesizer of so vast a phenomenon as durable inequality, after all, can hope no more than to push new inquiries in a fruitful direction.
The book's chapters follow from its program. Chapter 2 lays out the relational concepts with which the rest of the book works, while Chapter 3 reassembles them into a preliminary analysis of mechanisms generating categorical inequality. Chapters 4, 5, and 6 take up exploitation, opportunity hoarding, emulation, and adaptation in turn before putting them back together in a general account of how categorically organized advantage and disadvantage cumulate and endure. Chapter 7 examines categorical identities and their activation in politics, especially contentious politics. A final chapter treats relations between categorically defined and individual-to-individual inequality and then considers implications of the argument for deliberate intervention to change inequality and for future varieties of inequality.
Excerpted from Durable Inequality by Charles Tilly Copyright © 1999 by Charles Tilly. Excerpted by permission.
All rights reserved. No part of this excerpt may be reproduced or reprinted without permission in writing from the publisher.
Excerpts are provided by Dial-A-Book Inc. solely for the personal use of visitors to this web site.