Prejudice,Exclusion,and Economic Disadvantage

Key Concepts

The concept of discrimination is used in several important ways, but two need crucially to be distinguished. The first we call economic discrimination or disadvantage—having worse jobs or lower pay than others with comparable skills, qualifications, and ability (this is also economists’ technical use of the term discrimination; Becker 1971; Cain 1986; Lang and Lehmann 2012). A second aspect of discrimination, equally important in sociology, focuses on exclusion and social distance: To discriminate against a group is to exclude them from jobs. (The term can also cover barriers to participation in school or to friendship and marriage, but we deal only with jobs here.) This we call exclusionary discrimination. Many studies of discrimination and many legal cases involving it consider the underrepresentation of the foreign-born (or other minority groups)—namely, exclusionary discrimination—and implicitly assume that it leads to economic disadvantage—that is, to economic discrimination. The widely used “audit” method is one example (Heckman 1998; Pager 2007).

Thus, the classic prejudice-discrimination hypothesis contains two distinct claims: first, that prejudice leads to exclusionary discrimination and, second, that exclusionary discrimination leads to economic discrimination. That prejudice leads to exclusionary discrimination is both logical and well documented (Evans and Kelley 1991; Lang and Lehmann 2012 and the literature reviewed there). Nor is there anything irrational (as opposed to immoral) about prejudiced people discriminating. Full-time workers, for example, spend about a third of their waking life at work, so if they dislike minorities, working with them is a real subjective cost. Money is only one motive for working, so even if it costs to discriminate, it is still rational if people are willing to pay for their prejudices: In Durkheim’s terms, if people treasure mechanical solidarity based on ethnicity, they will discriminate even if it is costly.

Assumptions: Common Elements in the Theories

Secondary Labor Markets Cause Lower Pay

The central assumption here is that there are two distinct labor markets, a “primary” market with high wages, security, and benefits protected by law or power, the other a “secondary” (informal, segmented, dual, unprotected, unregulated) labor market with poorly paying, insecure jobs and few legal protections. The primary labor market is typically created by government labor market regulation (although there are alternative possibilities; Reich et al. 1973), with the secondary market arising spontaneously in small firms and self-employment niches exempt from regulation (World Bank 2006 : Chapter 9). The argument is that prejudice then creates discrimination and economic disadvantage by restricting access to the primary labor market (Bowles and Gintis 2002; Pastore and Haller 1982; World Bank 2006). Thus, it is argued, outgroups are rejected in many roles, including employees and co-workers. ¹³ Because of prejudice by employers and fellow workers in the more desirable and better paid primary labor market, they are not hired, promoted, or paid on their merits. As a result, they are pushed into the secondary labor market where rewards are meager, promotion opportunities limited, and unemployment recurrent. They therefore get worse jobs and lower pay than they would in an unprejudiced world.

Note that there are two separate hypotheses here: (1) Prejudice leads to exclusion and (2) exclusion leads to disadvantage.

Competition Cures Economic Discrimination

In sharp contrast to secondary labor market arguments, neoclassical economic theory implies that in the long run, economic discrimination cannot exist in free and competitive labor markets, according to Becker’s (1971, 1993) arguments—for a detailed review of the large literature flowing from this, see Lang and Lehmann (2012). ¹⁴ If members of an ethnic outgroup (for example) are paid less than dominant-group workers with similar productivity—or, equivalently, if dominant group workers are paid more than the going wage for equally productive outgroup workers—then the profit motive provides a strong incentive for an employer to break ranks with their discriminatory peers and profit by hiring (cheap) outgroup members instead of (expensive) dominant-group workers. The employer can pay the outgroup members less, or can hire more skilled—and hence more productive and profitable—outgroup members at the going wage. That raises the maverick firm’s profits, while employers with all dominant-group workers make less profit: They have to sell their products at the same price as the maverick firm, but are paying more for labor.

On Becker’s argument, it does not matter if the great majority of employers want to discriminate so long as a few are willing to get rich instead and cannot be prevented from doing so. Discrimination by workers or customers leads to the same result (Cain 1976:1232). So if at least some employers are concerned to maximize their profits, economic discrimination cannot long endure in Adam Smith’s competitive market: Unless discrimination is practiced by every single employer (e.g., because of legislation, ostracism, or the threat of violence), competitive pressures will eventually force the market back into equilibrium by reducing natives’ wages, or increasing immigrants’ wages, or both (Agesa and Brown 1998; Cymrot 1985; Lang and Lehmann 2012). Population ecology theory of ethnic labor markets makes the important complementary point that minority entrepreneurs can play an especially effective role in this (Evans 1989, 2005; Wilson and Portes 1980), so long as there is sufficient information flow between niche economies and the mainstream economy (Sanders and Nee 1987; Sanders, Nee, and Sernau 2002).

Existing Evidence on Prejudice and Discrimination: Enough to Raise Doubts

A vast literature has developed on the “prejudice-exclusion-disadvantage” thesis. Examples concern the United States, Britain, Northern Ireland, Canada, Australia, Germany, Israel, and many others (Andriessen, Dagevos, and Iedema 2008; Charles and Guryan 2008; Elliott and Smith 2001; Evans 1987; Evans and Kelley 1991; Kraus and Yonay 2000; Lang and Lehmann 2012; Mendelsohn and Vicziany 1998; Neidert and Farley 1985; Pager et al. 2009; Pager and Shepherd 2008; Pendakur and Pendakur 1998).

We do not provide a comprehensive review of this literature here because it would take us far afield, and in any case, it has been well done elsewhere (Cain 1976; Lang and Lehmann 2012; Pager and Shepherd 2008). A limited number of solid studies suffice to make the point that motivates our theoretical investigation. In particular, a number of good studies using well-conceived models and appropriate data have sought but not found economic discrimination in settings where secondary labor market theory would predict it. For example, in Australia, there is prejudice against immigrants from Italy, Greece, and other Mediterranean nations; many employers (as well as workers) prefer jobs to go to native-born Australians rather than equally well qualified Mediterranean immigrants; many Mediterranean immigrants believe they have missed out on jobs because of their ethnicity. And yet Mediterraneans get just as good jobs and just as high pay as native-born Australians (Evans and Kelley 1991).

The paradigmatic analytic approach in this literature, stemming from Siegel’s (1965) classic analysis, “On the Cost of Being a Negro,” estimates models predicting occupational status or earnings separately for several groups, for example, Latinos and non-Latinos; estimates the attainments to be expected in each group on the basis of their education, experience, skills, and other relevant characteristics; and then compares the results to see whether one group is getting less than their expected rewards, attributing the difference to discrimination. The context has usually been the general population of large free-market economies, such as the United States, Canada, or Australia.

The earlier literature claimed strong evidence of discrimination. This is still a highly influential view, but the analytic issues are difficult (Jones and Kelley 1984; Lang and Lehmann 2012). And more recent evidence, often using more comprehensive and appropriate models, sometimes goes against the discrimination hypothesis. For example, there is no sign of a social distance gradient in ethnic groups’ occupational and income returns to education in Australia (Evans and Kelley 1991 : Figure 2), as seems also to be true of the United States (Heckman 1998; Neidert and Farley 1985 : Tables 1 and 2), even in the past (Jensen 2002). In Northern Ireland, where strong residential segregation and governmental policies led to Catholic disadvantage as late as the 1970s, the situation now is complex and obscure (Breen 2000; Kelley and McAllister 1984). There were differences in intergenerational mobility patterns for Catholics and Protestants in the 1970s (Kelley and McAllister 1984: Figure 2) but not in the 1990s, with the loci of earlier discrimination and subsequent change primarily in the educational system and in access to public service jobs, rather than with actions by private sector employers or co-workers (Breen 2000). Differences in workforce outcomes between majority and minority group Israelis also seem primarily to reflect differences in family background, the educational system, and direct government policies rather than discriminatory actions by private employers (Kraus and Yonay 2000 : Tables 5 and 8). Even the huge disparity in occupational status between workers of Spanish and Aymara origin in rural Bolivia turns out to reflect their family backgrounds and education rather than differences in returns to education (Kelley 1988: Table 2). Little or no economic discrimination in returns to education is evident for Native Americans in Canada (de Silva 1999). Nonetheless, these issues remain controversial, complex, and often obscure (Altonji and Blank 1999; Heckman 1998; Jensen 2002). Note that any disadvantages in education are prior to the labor market and hence belong in separate analyses of discrimination and disadvantage in the education system rather than in the economy.

The studies just mentioned are important because they investigated occupational and income returns to education in a variety of countries where prejudice is clearly demonstrated (settings where traditional theories would predict a steep gradient); where their data, methods, and modeling are sound; but where the results are far from uniformly finding the predicted social distance gradient in occupational and income returns to education. This suggests that we must take seriously the possibility that such economic discrimination might be rarer and smaller than we had once supposed.

In addition to contemporary data, there are also well-documented cases in the historical record where substantially segregated subeconomies existed without much evidence of economic discrimination—real cases where “separate but equal” is an accurate description rather than the lie it was in the American South in past generations. For example, there were separate Catholic and Protestant economies in the Netherlands until recent decades—the verzuiling or “pillarization” system (Bax 1988; Sturm et al. 1998)—and perhaps in Northern Ireland still (Doherty and Poole 1997; Shirlow 2006). There were distinct ethnic economies in the Ottoman Empire (Inalcik and Quataert 1994; Lewis 1971); substantially separate Muslim and Hindu economies in pre-partition India (Papanek 1972); and a variety of separate ethnic economies within Mediterranean cities during medieval times, for example, the Mudejar communities of Spain during the high middle ages (Bartlett 1993).

Thus, there is good evidence in a number of sites, both past and present, where split labor market theory would predict economic disadvantage, but it is not found. This is a surprise, since the link between prejudiced feelings, exclusion from jobs, and economic disadvantage seems eminently logical. ¹⁵ The link between prejudice and exclusion from jobs is clear both from employment patterns showing ethnic segregation (found almost universally, e.g., Andriessen et al. 2008) and from direct questioning of potential employers and employees (Bobo and Zubrinsky 1996; Heckman 1998; McLaren 2003). For example, in Australia Evans and Kelley (1991) found that prejudiced managers and workers are more than willing to discriminate in hiring. ¹⁶ What is less clear is whether the (seemingly logical) link between exclusion and economic disadvantage is always to be found.

Why Exclusion Does Not Imply Economic Discrimination

Exclusionary discrimination does not generally cause economic discrimination because equally good opportunities are not necessarily the same opportunities (Becker 1993:387-388). ¹⁷ To take an extreme example, suppose the U.S. economy were arbitrarily split into two, all companies with names beginning with the letters A through N belonging to the AN economy and hiring only people whose names begin with A through N and all the rest belonging to the OZ economy and hiring only people with names from O to Z (Evans and Kelley 1991). There would then be complete segregation: universal exclusionary discrimination against OZ people in the AN economy and against AN people in the OZ economy. But so long as capital was equally available to entrepreneurs in both economies, ¹⁸ there would be no economic discrimination: There would be plenty of jobs for AN people in the AN economy, for OZ people in the OZ economy, and the two economies would pay equally well ¹⁹ —in effect, separate but equal economies that happen to share a single country.

Such extreme segmentation does not correspond to many modern societies, but more limited exclusionary discrimination does exist and need not produce economic discrimination. Suppose an employer does not want to associate with the foreign-born and prefers to hire only workers of native stock. That still gives the prejudiced employers a large part of the workforce to choose from, and so long as they do not lower their standards, their workers will be just as productive as anyone else’s, and they will be at no competitive disadvantage. They might not find as many employees as they would want, since they refuse to look at part of the labor force, so they might have to content themselves with running smaller businesses or substituting technology for labor. But even that is not likely, for some workers typically share these prejudices and so would prefer a segregated workplace—indeed, there seem to be at least as many prejudiced workers as prejudiced employers (Evans and Kelley 1991). But this will not stop immigrants getting commensurate returns to their skills and efforts, so long as prejudiced employers do not have a stranglehold on the economy. And in typical modern economies they do not.

And even if there were initially a dearth of jobs with non-discriminating employers, in a free and competitive market, new ones would soon be created. The opportunity to hire cheap immigrant labor instead of expensive natives gives existing non-discriminating employers a strong incentive to expand their businesses and opens up opportunities for ethnic entrepreneurs to set up an enclave economy (Evans 1989; Portes, Haller, and Guarnizo 2002). In short, an efficient capital market ensures that immigrants (or other groups subject to prejudice) have job opportunities no worse than other groups.

Documentation

Table 1 lists the variables in the model. A clearly documented Excel file implementing the model is freely available for download on the authors’ website, www.international-survey.org.

OPEN IN VIEWER

Table 1.

Variables in the Theory.

Name	Source and Value	Concept	Units
Assumptions
PrejudicePerson a	Assumptions: 0 through 1.00	Prejudice against outgroups. Expressed as the income necessary to compensate for the disutility of having outgroup members in the same firm. Zero if no outgroup members in the firm.	Number of minimum incomes
Labor_incomeWorker Labor_incomeBoss	Assumptions: Worker = 1, boss = 1.5	Minimum income: earnings from one’s own labor when self-employed or unemployed.	Index number. Sets the units for other variables.
WorthwhileWorker WorthwhileBoss	Assumptions: Worker = 1.20, boss = 3.0	Utility needed to make it worthwhile to change jobs (e.g., a self-employed farmer will shift into factory work only if offered at least 20 percent more than he currently has; potential bosses will set up a firm only if they can earn at least three minimum incomes).	Equivalent number of minimum incomes
Divison_Labor	Assumption: 2	Division of labor: increase in a worker’s productivity when they join a firm (e.g., 2 implies that productivity doubles).	Number
Skill_Boss	Assumption: 1.40 or 1.50	Amount by which a skilled and established boss increases a worker’s productivity: by 50 percent (Skill_Boss = 1.50). New and inexperienced bosses increase it only by 40 percent.	Number
Capital_cost	Assumption: 0.5	Cost of capital needed to support an extra worker. For example, 0.50 = annual payment of half a minimum income.	Number of minimum incomes
Theoretical implications
Output_per_Worker	Implied by model; Equation 2	Value of a worker’s output when employed by a firm	Number of minimum incomes
Firm_profit	Implied by model; Equation 3	Firm’s profit	Number of minimum incomes
IncomeBoss IncomeWorker a	Implied by model	Income (and economic discrimination)	Number of minimum incomes
UtilityBoss UtilityWorker a	Implied by Equations 4 and 5	Utility, well-being. Expressed as equivalent number of minimum incomes. Equal to Income –Prejudice.	Equivalent number of minimum incomes
Exclusionary discrimination a	Implied by model; Equation 6	Firm refuses to hire an outgroup worker.	Yes or no

a

Key variable.

Income and Utility

The baseline in our model is workers’ usual income when they are self-employed. In Adam Smith’s time, these would mainly be small farmers, day laborers, self-employed craftsmen, and the like; in the modern world they would be in a variety of service and small business jobs, mostly reasonably well paid, earning perhaps 70 percent to 80 percent as much as employees (Kelley et al. 2010; World Bank 2012). We assume self-employment is available to anyone who wants it (or alternatively and equivalently, that the unemployed who choose not to take up self-employment have similar income from unemployment support, other transfer payments, and informal earnings ²⁰ ). We set earnings for the self-employed to one minimum income as we call it (Labor_incomeWorker = 1); this sets the units for the rest of the model. If bosses choose to be self-employed rather than setting up a firm, we assume they would earn half again as much (Labor_incomeBoss = 1.50).

We assume that everyone, bosses and workers alike, seeks to maximize their own utility (Assumption 3). Thus, everyone is selfish, without any altruistic concern for others. Subjective well-being or utility (Utility) depends both on income (as in classical economics) and on satisfying one’s prejudices (following Becker). For convenience, utility is also expressed in income units. Thus, a self-employed worker’s utility is 1 (viz an income of 1, less nothing for prejudice since there are no co-workers). We assume that prejudice exists (Assumption 1) and that it creates a motivation for exclusion by reducing the utility of prejudiced people in firms where there are outgroup members (Assumption 2). Prejudice against outgroups is expressed in the same metric as income, specifically as the income necessary to compensate the prejudiced for the disutility of associating with outgroups. Thus, a boss or majority group worker might be unprejudiced and so require no compensation for working alongside a minority worker (Prejudice = 0); or might be slightly prejudiced and require, say, an extra 10 percent (so Prejudice = .10); or very prejudiced and require an extra 50 percent (Prejudice = .50); and so on. The logic is the same as the neoclassical notion of compensating differentials in pay (Kilbourne et al. 1994; Ragan and Tremblay 1988). Thus, for all persons in the firm, workers and bosses alike, utility is:

U t i l i t y P e r s o n = I n c o m e P e r s o n − P r e j u d i c e P e r s o n

where Prejudice = 0 unless the person is prejudiced and the firm has an outgroup worker.

For simplicity, we assume the same level applies to every prejudiced person, but the model easily generalizes to allow for a range of prejudices.

The level of prejudice in society matters a good deal, so we model a wide range of possibilities. To get some intuitive feel for the magnitudes involved, imagine that you yourself had to work alongside someone you disliked (perhaps a chain smoker). The lowest level of prejudice we model (Prejudice = .10) implies that you would require 10 percent of a minimum income to fully compensate for that. For Western nations with minimum incomes in the neighborhood of $20,000 a year, that comes to about an extra $1 per hour. If that $1 makes it worth putting up with the aggravation, then your level of prejudice against chain smokers is .10. For a more extreme example, suppose you had to work alongside someone you really despised (perhaps a left-wing fanatic endlessly declaiming about big business conspiracies and ecological risks to endangered beetles or a right-wing lunatic who praises Hitler and hates foreigners). If it would take $5 an hour to get you to willingly put up with such a co-worker, then your level of prejudice is .50. And if it would take $10 an hour then your level of prejudice is 1.00, the highest level we model.

Potential employers (“bosses”) decide who they are willing to employ as workers and how much they are willing to pay them, making these choices purely in their own self-interest (Assumption 3). Bosses then choose whether or not to set up firms. If it is sufficiently advantageous, they do so; otherwise, they remain self-employed. Following Adam Smith, Durkheim, and ample empirical evidence, we assume that the division of labor increases a worker’s productivity greatly when they shift from self-employment into a firm, perhaps doubling or tripling. Exactly how much it increases will depend on the inherent advantages of the division of labor (Division_Labor), the ability and experience of the boss who runs the firm (Skill_Boss), and the cost of capital (Capital_cost). We assume that the division of labor inherently doubles productivity (Division_Labor = 2), so that each worker in a firm produces two minimum wages worth of goods (viz Labor_income_Worker times 2), and that a typically able and experienced boss in an established firm increases that by half again (Skill_Boss = 1.50). A new and inexperienced boss, trying to set up a business for the first time, would not do quite so well, say increasing things by 40 percent (Skill_Boss = 1.40); see (World Bank 2012 : Chapter 3). For simplicity, we assume the boss borrows the required capital, perhaps from a bank, and repays interest on the loan out of the firm’s takings—say, half a minimum income per worker (Capital_cost = .5). With these assumptions, a worker increases their output (Output_per_Worker) by two and a half times by joining an established firm, from 1 minimum income to 2.5. Adam Smith’s pin makers did even better. Specifically:

O u t p u t _ p e r _ W o r k e r = L a b o r _ i n c o m e W o r k e r × D i v i s i o n _ L a b o r × S k i l l _ B o s s − C a p i t a l _ c o s t .

The firm’s profit (Firm_profit) is the output of each worker less the income it pays each, added up for all the workers, plus the boss’s contribution from his own labor (Labor_IncomeBoss). We assume the boss keeps it all:

F i r m _ p r o f i t = Σ W o r k e r s _ i n _ f i r m ( O u t p u t _ p e r _ W o r k e r − I n c o m e W o r k e r ) + L a b o r _ I n c o m e B o s s .

Bosses will require substantial compensation for taking up the stress, aggravation, and risk of running a firm—say twice the 1.5 they would get from self-employment (Worthwhile_Boss = 3). So a boss’s income will then be either what comes from setting up a firm (assuming it is enough to be worthwhile) or, failing that, whatever they would earn in self-employment. And the boss’s utility will be (by Equation 1) that income less any adjustment for prejudice. Thus, a boss has at most three options: setting up a firm including a minority group worker; setting up a firm but excluding minority group workers; or remaining self-employed. We assume he or she chooses the one that provides the greatest utility (Assumption 3):

U t i l i t y B o s s = ( F i r m _ p r o f i t − P r e j u d i c e B o s s ) i f > W o r t h w h i l e B o s s o t h e r w i s e n o f i r m e s t a b l i s h e d , a n d b o s s r e m a i n s s e l f − e m p l o y e d a n d U t i l i t y B o s s = L a b o r _ I n c o m e B o s s .

Similarly, we assume that a worker will not take up a job unless it is beneficial for them, say with a utility 20 percent greater than self-employment (Worthwhile_Worker = 1.20). This models the cost of change and the uncertainties and disadvantages of wage labor. For simplicity, we assume the unemployed get about the same as self-employed workers (see World Bank 2012 Chapter 1 and note 20), so the same logic applies for them.

Which job has the greatest utility will depend both on the pay offered and on a possible loss of utility due to prejudice (from Equation 1). Prejudice matters only if the worker is prejudiced and the firm employs an outgroup member. Workers choose among their job prospects and self-employment, selecting whatever maximizes their utility (Assumption 3):

U t i l i t y W o r k e r = ( I n c o m e f r o m t h e j o b o f f e r w i t h g r e a t e s t u t i l i t y − P r e j u d i c e W o r k e r ) i f > W o r t h w h i l e W o r k e r o t h e r w i s e w o r k e r r e m a i n s s e l f − e m p l o y e d a n d U t i l i t y W o r k e r = L a b o r _ i n c o m e W o r k e r .

For simplicity, we assume firms remain small, as most are in the real world (World Bank 2012), with one boss and two or three workers. But much the same logic would apply to larger firms. Each worker in our model could represent a thousand workers, each paid the same and with the same self-employment opportunities. Each boss could represent a thousand administrators, managers, professionals, and the like whose joint efforts are necessary to perform the employer’s functions in a large firm. If they are all ultimately paid from the firm’s profits in some fixed proportion and have alternative opportunities in self-employment, government, personal services, or the like, then their joint incentives are much the same as the boss’s incentives in our model.

We assume the firms operate in a small open economy so that the price they receive for their goods is fixed, with nothing they can do change it. There may or may not be competition between firms in the wages they offer. For our purposes, that is the key distinction among economies.

Constrained Economies Dominated by Monopoly, Oligarchy, Tradition, or Government: The Sociological Stereotype

We consider two ideal types: constrained economies and competitive ones. In constrained economies (or constrained sectors of a mixed economy), the labor market is dominated by monopoly, oligarchy, tradition, or government. Wage rates are fixed, all bosses offering exactly the same wages, with no competition between employers in the wages they offer. At the other extreme are competitive economies dominated by individual self-interest in a free market in the style of Adam Smith: Numerous would-be employers compete for workers by offering different wages, those who offer too little getting no workers and therefore no profits. These two types of economy have fundamentally different implications for discrimination.

“Constrained” economies (as we shall call the first type) could arise from monopoly when a single firm drives out, or buys out, almost all competition (as, e.g., J.P. Morgan arguably did in the U.S. steel industry in the 1900s). This could be a single firm, or several similar firms acting in collusion. What matters is that they have effective control over the labor market. ²¹ That control could also arise from oligarchy, where a dominant economic elite restricts competition (as in some Third World nations). It could arise from religious and other traditions (as, e.g., customary wage rates in medieval Europe or the “just wage” in modern Catholic doctrine). And most commonly, it could arise from government regulation. Modern governments typically have control over wages in government administration and the army and often in education, health care, public safety, and other areas. Many also have strong influence on wages in the private sector, especially in European-style welfare states.

In an ideal-typical constrained economy, employers—monopolists, or oligarchs, or tradition, or government—have control over the labor market: They set wage rates for everyone, all offering exactly the same wages. Workers have no choice other than accepting that fixed wage or forgoing employment altogether, remaining self-employed or unemployed. Crucially, no competing would-be employer is allowed to set up a new firm in competition with existing employers.

Empirically, economies in the lowest quintile or two of economic freedom in Gwartney, Lawson, and Hall’s (2012) widely used economic freedom index would qualify as constrained as we are using the term; this includes Malaysia, China, India, Nigeria, and Venezuela, among many others, and Communist nations in times past.

For simplicity, we model only the classical type of constrained economy, a monopoly with a single firm that tries to maximize its own profits and is free to do so without legal, moral, or other constraints. A monopolist could have other goals, and government monopolists often do, but this is the ideal-typical case and the simplest one. We model a single firm, a prejudiced employer, initially with two workers (Table 2, Panel A). The boss is a majority group member (e.g., English) and prejudiced against outgroups (Irish, say); both workers are majority group members, one prejudiced and one not. This is roughly the traditional sociological stereotype.

OPEN IN VIEWER

Table 2.

Income: Theoretical Predictions Separately for Economies Dominated by Monopoly, Oligarchy, Tradition, or Government (Panel A) and Competitive Economies Dominated by Individual Self-interest (Panel B). ^a

		Identity	Prejudiced against Outgroups	Income (Index: 1 = Solo Worker) and Level of Prejudice in Society
				Prejudice = 0	Prejudice = .10	Prejudice = .25	Prejudice =.50	Prejudice = .75	Prejudice = 1.00
		(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)
Panel A. Constrained economies dominated by monopoly, oligarchy, tradition, or government b
All firms: Prejudiced
1	Owner	Majority	Prejudiced	5.40	5.30	5.15	4.90	4.10	4.10
2	Worker	Majority	Prejudiced	1.20	1.30	1.45	1.70	1.20	1.20
3	Worker	Majority	Not	1.20	1.20	1.20	1.20	1.20	1.20
4	Worker	Outgroup	Not	1.20	1.20	1.20	1.20	No job	No job
Outcome
5	Exclusionary discrimination?			No	No	No	No	Yes	Yes
6	Higher pay for some workers?			No	Yes .10 more	Yes .25 more	Yes .50 more	No	No
7	Economic disadvantage for outgroup?			No	No	No	No	Yes c .20 less	Yes c .20 less
8	Worse job for outgroup (utility)?			No	No	No	No	Yes d	Yes d
Panel B. Competitive free-market economies dominated by individual self-interest
Firm 1: Prejudiced
9	Owner	Majority	Prejudiced	4.35	4.25	4.10	3.40	3.40	3.40
10	Worker	Majority	Prejudiced	1.55	1.65	1.80	1.55	1.55	1.55
11	Worker	Majority	Not	1.55	1.55	1.55	1.55	1.55	1.55
12	Worker	Outgroup	Not	1.55	1.55	1.55	No job e	No job e	No job e
Firm 2: Mixed
13	Owner	Majority	Not	4.35	4.25	4.10	3.85	3.60	3.40
14	Worker	Majority	Prejudiced	1.55	1.65	1.80	2.05	2.30	1.55
15	Worker	Majority	Not	1.55	1.55	1.55	1.55	1.55	1.55
16	Worker	Outgroup	Not	1.55	1.55	1.55	1.55	1.55	No job f
Firm 3: Unprejudiced, or enclave economy
17	Owner	Either	Not	4.35	4.35	4.35	4.35	4.35	4.35
18	Worker	Majority	Not	1.55	1.55	1.55	1.55	1.55	1.55
19	Worker	Either	Not	1.55	1.55	1.55	1.55	1.55	1.55
20	Worker	Outgroup	Not	1.55	1.55	1.55	1.55	1.55	1.55
Outcome (whole economy)
21	Exclusionary discrimination?			No	No	No	Yes	Yes	Yes
22	Higher pay for some workers?			No	Yes .10 more	Yes .25 more	Yes .50 more	Yes .75 more	No
23	Economic disadvantage for outgroup?			No	No	No	No	No	No
24	Worse job for outgroup (utility)?			No	No	No	No	No	No

Note: Level of prejudice in society varies from none to very high. Key implications noted (highlighted for exclusionary discrimination, dot-dash outline for pay inequalities, and solid outline for economic discrimination).

a

Evaluated at the baseline settings. Details in Table 1 and in the text.

b

Technically a “monopsony.” a sole buyer of labor. Informally, a “monopoly.” This could be a single firm, or several similar firms in collusion, with monopoly control over the labor market, many firms subject to strong control over hiring by tradition or government regulation.

c

Disadvantage for outgroup worker compared to their income in an unprejudiced setting. This comes about because they are not employed.

d

Solo self-employed or unemployed, thus earning less than employees.

e

Employed in Firm 2 or Firm 3.

f

Employed in Firm 3.

On our baseline assumptions, workers are two and a half times as productive in a firm as outside one, after taking into account the advantages of the division of labor, the ability and experience of the boss, and the cost of capital. So they produce 2.5 minimum incomes for the firm (Equation 2). From that, the boss must subtract their pay; the rest is profit (Firm_profit, Equation 3).

The key question is how much workers are paid. If the boss offers three-quarters of a minimum income, workers are better off self-employed and will not join the firm; if he or she offers one minimum income, they would be as well off financially working for him but gain nothing, so accepting the job would still not be worthwhile; nor would an offer of 1.10 minimum incomes be worthwhile. On our baseline assumptions, the least the boss can successfully offer is 1.20 minimum incomes, which is the minimum workers would accept (Worthwhile_Worker = 1.20). There is no need for him to offer more—that would cut into his profits—so 1.20 is what he offers (Table 2, column 3, rows 2 to 4). With that offer, it is in the workers’ interest to join the firm. The boss then makes a handsome profit of 1.30 minimum incomes from each worker. ²² Workers would like more, of course, but the boss has no reason to offer more and the workers have no better alternative. Thus, in a monopolistic labor market controlled by the bosses, workers’ wages are fixed at the minimum workers will accept.

What happens if an outgroup worker (an Irish immigrant in our example) shows up and applies for a job (Table 2, row 4)? If prejudice did not rear its ugly head, the boss would be happy to hire the Irishman: He would simply provide extra 1.30 profit. ²³ But prejudice cuts into that profit. If prejudice is minor, say it takes .10 of a minimum income to compensate, then the prejudiced boss’s utility would drop by .10 (Equation 4). It would also reduce the prejudiced majority group worker’s utility by .10, bringing him down below the minimum he is willing to accept (as now Utility_Worker = 1.10 while Worthwhile_Worker = 1.20). We assume the boss is not willing to lose his existing workers and the profits they create, ²⁴ so he has to give the prejudiced worker a small raise, .10, to keep him from quitting. Thus, a perverse effect of prejudice is to raise a prejudiced worker’s pay in proportion to his prejudice (row 2, column 4, dashed outline). However, it does not raise his utility (Table 2, row 8): That is still the income he gets minus the prejudice he feels: 1.30 – .10 = 1.20, just the same as every other worker’s utility, majority and outgroup alike.

More prejudice makes these effects larger (row 2, columns 5 and 6, dashed outline). With prejudice at .25, the Irishman still gets hired, but the boss loses .25 in utility and has to pay another .25 extra to the prejudiced majority worker. With prejudice at .50, the boss is barely willing to hire the Irishman: As before, hiring him adds 1.30 to output but also costs an extra .50 in pay for the prejudiced majority worker and another .50 in lost utility for the prejudiced boss, leaving a gain in utility of only .30. At an extreme of .75 or 1.00 (perhaps like Muslim hostility to Jews in modern Iran or Brahman aversion to untouchables in 19th-century India), the boss is unwilling to hire an outgroup worker (row 4, columns 7 and 8, shaded cells). Between his personal prejudice (a loss of .75) and the huge raise he would have to give the prejudiced majority worker (another .75), it is just not worth it—the 1.50 loss from those two outweighing the 1.30 gain from having another worker. In sum, in the absence of legal or other constraints on hiring:

Σ P e r s o n s _ i n _ f i r m ( P r e j u d i c e P e r s o n s ) > ( O u t p u t _ p e r _ W o r k e r − I n c o m e O u t G r o u p W o r k e r )

This outcome (Table 2, Panel A) is much like the conventional sociological stereotype. At middle levels of prejudice, there appears to be economic discrimination, with outgroup workers getting lower pay than some majority workers, more so when there is more prejudice (row 2, columns 4 to 6, dashed outline). At high levels of prejudice, there is exclusionary discrimination (row 4, columns 7 and 8, shaded cells).

However, this exclusionary discrimination is only at very high levels of prejudice, .75 or above. Prejudice at that level implies that it would take 75 percent of a minimum income to fully compensate for having to work alongside an outgroup member. For a prejudiced American in recent decades, that comes to something like an extra $7 or $8 an hour (around $15,000 a year). Few Americans nowadays would be that prejudiced, although more would have been in the past. And even now in the Middle East, the Balkans, and other areas where ethnic and religious tensions run high, such levels may well exist.

There also appears to be economic discrimination. But it is mostly, although not entirely, illusory. The difference in pay comes about mainly because prejudiced majority workers are, perversely, paid more than the going rate (Panel A, row 2, columns 4 to 6, dashed outline). Furthermore, most workers, majority and outgroup alike, get equal utility from their jobs—there is no inequality in overall subjective well-being derived from work. The exception to all this is outgroup workers in very prejudiced societies (row 4, columns 7 and 8, shaded cells). They are subject to exclusionary discrimination and because of it to economic discrimination, earning less because they remain self-employed (or perhaps unemployed ²⁵ ) rather than getting a job with the monopoly firm. In our model, their loss would come to something like 20 percent of a minimum income (row 7, last two columns), which is around $4,000 a year for modern American workers (viz Worthwhile_Worker – Labor_income_Worker = .20).

Competitive Economies Dominated by Individual Self-interest: Adam Smith’s Ideal Type

Rather than a monopoly or other constraints on the economy, suppose there are many firms in competition in a free market, each pursuing its own self-interest. Empirically, this would usually include economies (or competitive sectors of a mixed economy) in the highest quintile or two of economic freedom in Gwartney et al.’s (2012) widely used economic freedom index—for example, Switzerland, the United States, the United Kingdom, Taiwan, South Korea, or Ghana.

There are many types of firms, but we model three that cover the main possibilities: a mostly prejudiced firm as before, with a prejudiced owner and one prejudiced worker; a mixed firm with an unprejudiced owner but one prejudiced worker; and finally an entirely unprejudiced firm, whose members are either unprejudiced majority, or outgroup members in an enclave economy, or some mixture of those (Table 2, Panel B). Other assumptions remain the same. This is a rough approximation of Adam Smith’s free and competitive market.

In this competitive market, worker’s wages will be much higher. Suppose things begin with Firm 1 paying workers 1.20 minimum incomes, as in the monopolistic economy. Then another boss sets up a firm, say Firm 3, and offers workers 1.30 minimum incomes. That would be enough to lure the workers away from Firm 1, but still leave the new boss better off than before he set up his firm, and by a worthwhile amount. But other bosses have an incentive to make even better offers, say 1.40 minimum incomes or 1.50. The best they can offer is 1.55; beyond that, there is not enough profit to make it worthwhile for them to establish a firm. We are assuming that the new bosses are not quite as skilled and experienced as established bosses (in Equation 2, Skill_Boss = 1.40 rather than 1.50), so their best offer is a little lower than an established boss could offer.

Thus, 1.55 becomes the going rate for workers since some boss will find it worthwhile to make that offer and none will be willing to offer more. So workers do well while owners make very little from each worker, just half a minimum income. Thus, in a competitive economy, workers’ pay is fixed by the minimum profit potential bosses will accept. ²⁶

The outcome is that there is almost always exclusionary discrimination (Table 2, rows 12 and 16, last few columns, shaded cells). Except at low levels of prejudice, the prejudiced firm (No. 1) will not hire outgroup workers (row 12) since the profit from an outgroup worker is so small that it rarely makes up for the disutility they cause the boss and for the extra pay the boss has to give to prejudiced majority workers. Even in a mixed firm (No. 2) where the boss is unprejudiced and only one worker prejudiced, outgroup workers won’t be hired when prejudice is exceptionally high (row 16, last column)—the extra profit they bring is less than the extra pay needed to compensate the prejudiced majority worker.

Thus, a competitive market fosters exclusionary discrimination. This comes about (strangely) because it pays workers more than they would get in a monopoly economy and hence leaves bosses with little profit. With little profit, prejudiced bosses have little room to compensate themselves or their prejudiced workers for the disutility of hiring an outgroup worker. So good pay for workers often leads to exclusionary discrimination.

In this situation, there is no genuine economic discrimination (row 23): Wherever they are employed, outgroup workers get the going rate, 1.55 (rows 12, 16, and 20). Moreover, that is exactly what they would get in an entirely unprejudiced society (which would be like firm No. 3). And regardless of the level of prejudice in the society, all workers, majority and outgroup alike, get the same utility from their jobs (row 24).

There is, however, apparent economic discrimination. This is because outgroup workers in societies that are only mildly prejudiced get a little less than some majority workers (viz the perversely overpaid prejudiced majority workers in firms with outgroup employees; rows 10 and 14, cells outlined in dashes). Thus, outgroup workers are not disadvantaged compared to an unprejudiced society.

This sort of competitive economy is the model underlying our basic argument in Hypotheses A through C. Together with a variety of side assumptions about the market and social conditions that foster or inhibit competition between employers, it also underlies Predictions 1 through 7, below. With side assumptions about government’s role in fostering or inhibiting competition, it leads to Predictions 8 through 13, below. Throughout, the key is wage competition or the lack thereof.

Generalization

Our theory implies that any competitive economy will look very much like the particular example in Table 2, Panel B. Specifically:

In a competitive economy where workers and employers are entirely self-interested, where employers are allowed to compete in the wages they offer, and where it is possible for a new entrepreneur to establish a viable firm . . .  ²⁷

(Theoretical implication 1) Outgroup and majority workers get equal utility from their jobs; ²⁸

(Theoretical implication 2) Outgroups are paid the same as they would get in an unprejudiced society;

(Theoretical implication 3) There is apparent economic discrimination within any firm that has both prejudiced and outgroup workers, because the prejudiced workers are paid more than other workers;

(Theoretical implication 4) There will be exclusionary discrimination in any firm where the disadvantages imposed by prejudice exceed the gains from employing an outgroup worker, specifically whenever:

∑ P e r s o n s _ i n _ f i r m ( P r e j u d i c e P e r s o n s ) > ( O u t p u t _ p e r _ W o r k e r − I n c o m e O u t G r o u p W o r k e r ) ;

(Theoretical implication 5) Exclusionary discrimination will be more common than in an otherwise similar constrained or monopoly economy.

Predictions: The Social Locus of Discrimination

Modern societies typically have many non-discriminatory employers—if only because education and cosmopolitanism undermine most forms of prejudice, because outgroup members themselves are often employers, and because capital markets are generally reasonably efficient so unprejudiced employers can start and grow their businesses. Hence, economic discrimination will be rare, with workers from disparaged groups finding ample jobs with the non-discriminatory employers, including outgroup members. ²⁹ Thus, economic discrimination will, we predict, in fact be rare in precisely the modern, market economies where past research has mostly looked for it.

Anything that leads to the growth of an efficient capital market will, we have argued, undermine economic discrimination by facilitating the entry, or expansion, of non-discriminating employers. Thus, the expansion of the market into previously non-monetized societies would undermine economic discrimination (as, e.g., in the Middle Ages where discrimination against Jews was less in mercantile cities than in the largely non-monetized Polish hinterland or in Africa in recent times, where tribal discrimination is less in urban areas than in rural). So would the development of a competitive market in what was previously a bureaucratic command economy (e.g., in Eastern Europe after the fall of Communism, or in China). So would abandonment of bureaucratic controls over the economy in quasi-capitalist countries with legally sanctioned ethnic discrimination (e.g., Malaysia now or South Africa in years past). So would globalization (Black and Brainerd 2004).

In the absence of legal coercion by the government, it is difficult to prevent market competition from undermining economic discrimination. But in suitably intimate and isolated settings, social pressure has considerable force and can, in the extreme, coerce compliance. So economic discrimination is more likely when employers are a small and cohesive class, or where tradition and social pressure are strong, as for example discrimination against untouchables in village India or Amerindians in rural Bolivia (Kelley and Klein 1982), rather than where employers are a polyglot and competitive group in an individualistic society.

Economic discrimination can, we argue, only rarely appear in small firms exposed to market competition because entrepreneurs have little leeway to set wages. But large bureaucracies generally have less clear evidence on performance—for example, no one can tell whether any one clerk out of hundreds in a large insurance company matters much to the company’s profits. Furthermore, economies of scale and other advantages of size more easily absorb the economic cost of discrimination. So it is in large, bureaucratic settings that economic discrimination should appear, not in small competitive firms.

But the opposite is true of exclusionary discrimination. Prejudice matters more in small settings—social contact matters more, with sociability extending not only among co-workers but between workers and the boss; feelings matter more in intimate, less impersonal settings. But in large, bureaucratic settings, prejudice matters less, behavior is more impersonal, and recruitment more rule bound.

In all, we have argued:

Predictions: Governments and Economic Discrimination

We have argued that the competitive market corrodes economic discrimination, unless some compulsion can be used to prevent unprejudiced employers from hiring workers from the excluded group. Laws passed by governments provide an extreme instance of such barriers to market competition: They make competition illegal and impose compliance by force of law. Examples are to be found in Malaysia, South Africa under apartheid, and nativist immigration policies in many developed nations, among many others. Moreover, discriminatory actions by governments as employers have been clearly documented in a variety of developed societies, ranging from Ireland (Breen 2000) to Israel (Kraus and Yonay 2000). And the more powerful and efficient the government, the better able they are to enforce discrimination (Lenski 1966). So, for example, modern governments in developed nations will be better able to discriminate economically than were their less efficient predecessors in medieval times or their peers in poor countries today.

Because the central government’s writ runs throughout the country, it has more opportunity for economic discrimination than do local governments, which are subject to competition from other local jurisdictions. For example, if San Francisco imposes a tax surcharge on foreign-owned banks, they can all move to Los Angeles. So given equal will to discriminate (e.g., against foreigners and resident aliens or in favor of politically powerful minorities and special interest groups), the federal government would have the greatest opportunities to do so; state governments less; and local governments least. Similarly, governments can discriminate more effectively in areas where they have more control (e.g., the civil service or government purchasing) rather than where they have less control (e.g., over private businesses). For example, during much of the pillarization period in the Netherlands, Catholics were barred from public office in the national government (Bax 1988).

Conversely, the elimination of exclusionary discrimination is easier for more authoritarian systems. For example, in the U.S. South, the courts and not locally elected officials played the leading role in dismantling discrimination at a time when most Southerners supported it. Exclusion is thus an enduring threat in a democracy.

In contrast, economic discrimination could be more common in Communist systems where there is no free market to undermine it (except for the informal economy: Portes and Haller 2005); authoritarian societies able to overcome market pressures; and settings where economic considerations are not salient. Examples are discrimination in Russia against Muslims or against blacks in South Africa in times past.

Thus, we have argued:

Summary

This is a theory about discrimination arising from prejudice (in the sense of aversion, hostility, or social distance) or its converse (favoritism for one group can be thought of as prejudice against those who are not in that group). It applies to prejudice arising from differences in ethnicity, tribe, or place of birth; or from religion or moral values; or from class, accent, or appearance; or from sexual preferences, values, or lifestyle; or from age, weight, or cigarette smoking; or from many other sources besides. In both sociology and economics, prejudice was the focus of the foundational analyses, and it is still the focus of the large and growing literatures that stem from them. Our theory does not apply directly to gender discrimination (since prejudice is rare), nor to racial, ethnic, or political discrimination whenever they are based on factors other than prejudice. We are concerned with the effects of prejudice and discrimination in their classic milieu, the economy; hiring and pay are the key issues.

On theoretical grounds, we have argued that exclusionary discrimination and economic discrimination are fundamentally distinct. One does not imply the other. Only in exceptional circumstances and in “constrained” sectors of the economy dominated by monopoly, oligarchy, tradition, or government will exclusionary discrimination cause economic discrimination. In competitive economies, the fact that ethnics, or Catholics, or homosexuals, or immigrants, or smokers are excluded from some jobs will not, we argue, in general mean that they get worse jobs or lower incomes than other people. ³⁰

While this article builds on familiar theories in sociology (on prejudice) and neoclassical economics (on competition), combining them leads in some new directions. Our argument suggests that economic discrimination is not likely to be found where it is usually sought—among the general population in large free-market economies. Rather, it is to be found elsewhere in constrained economic settings where the free market plays little part: in government, in highly regulated industries and large oligopolistic corporations, in hospitals and armies, and in authoritarian nations.

Furthermore, we argue that exclusionary discrimination and economic discrimination flourish in quite different settings, produced by quite different causes. Competition undermines economic discrimination since the very fact of discrimination itself creates opportunities for a non-discriminating entrepreneur. This is true both of economic competition among firms and of competition among governments or bureaucratic jurisdictions. Thus, if one wishes to undermine economic discrimination, the natural course is to reduce monopoly by employers, if it exists; to reduce any government regulation or control that hinders the formation of new businesses or expansion of old, if such are present; to prefer small-scale, independent organizations to large, bureaucratic ones; and to encourage competition among employers, not reduce it. In short, creating a competitive world in the image of Adam Smith is a sovereign remedy for economic discrimination.

We have argued that the natural home of economic discrimination is, instead, in government and large corporations little restrained by market competition, and in Communist and other constrained economies where the market plays little role, and also in institutions where the market plays little part, such as the civil service, hospitals, and armies.

Exclusionary discrimination is quite a different matter. It is more at home in small-scale, intimate settings; in small businesses rather than in large bureaucratic ones. It is more likely in competitive, free-market settings—where wages are generally high and employers’ margins thin—than in constrained sectors of the economy where the free market plays little part. And, it is more to be expected in democratic countries than in authoritarian ones.

In competitive free-market settings—which include much of the economy in the United States and many other advanced nations—our theory implies that outgroups are paid the same as they would get in an unprejudiced society: The economy is not the problem. But if outgroups have preexisting disadvantages relevant to economic success (weaker language skills, fewer books, less education, disadvantaged family background, disadvantageous peer groups, etc.), the free-market economy will not cure those disadvantages. But neither will it make them worse. The difficulties lie elsewhere, and it is elsewhere that the remedies are most likely to be found.

Implications for Future Research

Tests of the prejudice-exclusion-discrimination hypothesis have traditionally been done on general population samples in democratic market economies—for example, in the United States, Britain, Australia, and Germany. Our argument, however, suggests that this is precisely the wrong place. Rather, research should in future focus on places where market competition is weak and the concern for profit limited. More promising are surveys of large, relatively closed labor markets in institutions not oriented toward profit—such as armies, hospitals, and the government service. Rather than the United States and Europe, more promising are surveys of more bureaucratically controlled economies (e.g., in China, India, Ukraine, Venezuela, or parts of Africa). In short, we argue that the prejudice-exclusion-discrimination paradigm should shift its analytic focus from open market economies to regulated, bureaucratic domains far from competition.

Another implication of our argument is that new and much richer survey data are required. Traditional analyses have relied mainly on data from the census and government surveys. These are typically high in quality and rich in cases but impoverished in the range of variables measured. We need to know more about workers and about jobs: Income is not enough. Following Becker, our logic implies that equal incomes do not imply equally good jobs, nor do equally good jobs necessarily have equal incomes: Prejudice must be taken into account. We also need explicit measures of other influences on job quality (security, autonomy, complexity, prestige, etc.) in addition to the “usual suspects” of prejudice, gender, education, ability, effort, experience, and the like. Theory assumes that we know what makes jobs good, but that may be presumptuous: In many ways, what really matters is what workers themselves feel. In particular, if they believe themselves fairly paid, that satisfies the main public policy concern. But if they believe they are paid too little, or if others believe they are paid too much, that is a concern regardless of what objective estimates say about their income. Fair treatment is inherently subjective; it can and should be measured. In short, research on prejudice and discrimination needs extensive new survey data in order to clad its elegant theoretical reasoning with a real-world panoply of fact and measurement. ³¹ We offer some suggestions elsewhere (see www.international-survey.org).

Implications for Law and Public Policy

Law and public policy generally assume that a demonstrated shortfall in pay for ethnics, immigrants, and other groups subject to prejudice and exclusionary discrimination imply that they are getting less than they should. But our theory suggests that pay differences between outgroup and dominant group may come about because prejudiced members of the dominant group are paid too much, not because outgroup members are paid too little. A competitive market with compensating differentials (perversely) implies higher pay to prejudiced workers. Thus, a shortfall in pay does not necessarily imply that the lower paid worker is getting less than they should. Instead, the analysis needs to figure out what the job would be paid in the absence of prejudice and compare actual pay to that standard. That is a different and much more difficult analytic task.

Much law and public policy focuses on exclusionary discrimination, on who is hired and who is not; “audit” studies are a common and familiar example (e.g., Pager et al. 2009; Tilcsik 2011; Pager 2007 gives an extensive review). Exclusion may be undesirable in and of itself. But it does not imply economic disadvantage. Indeed, our theory suggests that workers, majority and minority alike, are better off in competitive economies where exclusionary discrimination is relatively common rather than in constrained economies where exclusionary discrimination is rare. ³² Competition among owners increases workers’ wages and eliminates economic discrimination. But whenever prejudice exists, it also increases exclusionary discrimination.

Law and public policy’s usual analytic approach to documenting economic disadvantage—one stemming from Siegel’s classic 1965 analysis—is questionable. That approach relies on models predicting earnings on the basis of education, skills, ability, effort, experience, and such. It then assesses whether, these things being equal, outgroup members earn less than those from the dominant group. If so, the discrepancy is attributed to economic discrimination against outgroup members. But we have argued, following Bentham and Becker, that this is the wrong comparison: utilities (as economists would call it) or contributions to subjective well-being (as sociologists and social psychologists would put it) should be compared, not incomes alone. Equal treatment implies equally good jobs for comparable workers but not equal incomes. Income is only one desirable feature of a job and, at least in prosperous nations, not even the most important one. Autonomy at work, complexity and prestige of the tasks, security, employer quality, and other things also matter. An analysis of job quality, not of income alone, is required. Equal treatment does not imply equal pay. Nor, conversely, does equal pay imply equal treatment.