Bryan Caplan  

The Temptation of Discrimination

Gordon on Growth: A Critique... Marginal Revolution University...
Why doesn't everyone just follow the rules?  The obvious answer, in many cases, is that breaking the rules has concrete advantages... if you don't get caught. 

Why do people jaywalk?  Because it's quicker than schlepping to the crosswalk.  Why do people steal?  Because it's easier than working for a living.  Why do people adulterer?  Because forbidden sex is exciting.  As long as rules like these exist, many human beings will try to furtively evade them.

On the surface, rules against labor market discrimination have the same structure as rules against jaywalking, stealing, and adultery.  We've created a whole system of public and private lawsuits to ferret out discrimination.  We encourage people to come forward and expose acts of discrimination.  Our implicit assumption, apparently, is that human beings yearn to discriminate against others.  In the absence of harsh punishment, we'll succumb to temptation.

My question: What is supposed to make discrimination so tempting?  For adultery, we've got a crisp evolutionary story: Cheating on your spouse without getting caught has a massive genetic payoff.  I just don't see how discrimination is remotely comparable.  How many people really enjoy inflicting unmerited suffering?  I can easily believe that people enjoy discriminating when everyone else is doing it; people are sheep, after all.  But once discrimination is publicly unacceptable, our evolved desire for conformity ought to push in the opposite direction.

My favorite explanation: what popular culture impugns as "hate" is, by and large, merely misunderstood statistical discrimination.  Firms are tempted to discriminate because stereotypes save time and money.  If you don't buy this story, though, I'd like to hear your alternative.  If discrimination isn't lucrative, why are employers continuously tempted to break the law?

Comments and Sharing

COMMENTS (27 to date)
John Thacker writes:

Most people, though not you, do enjoy group identification. Discriminating tends to be justified and viewed as helping other people "like them." It's about helping the in-group rather than some out-group.

Of course people do this with alma maters, or favorite sports or hobbies, or close family relation, or other things beside very distant supposed genetic relation, and we don't stigmatize it nearly as much in those cases.

David Friedman writes:

John says, about other forms of in-group discrimination, "and we don't stigmatize it nearly as much in those cases."

In quite a lot of cases we not only don't stigmatize it, we praise it. Think about all the people who take it for granted that buying locally is a virtuous act--because it helps people near you instead of helping people somewhere else.

RPLong writes:

In a way, I think some people (especially the ones who don't like to think) are unable to differentiate between a friend's success and an enemy's failure. They think if one person wins, someone else must have lost. So they conclude that it is better to determine who the losers are. From there, the rest follows.

We as human beings don't like to admit that a great many of us literally don't like to think, but I can't tell you how many people I know who have said as much to me, using those very words.

Hume writes:

I'm sure there's an evolutionary explanation along the lines of (1) evolved in small groups (tribal), (2) evolutionarily optimal, at the group level, for members with strong feelings of group identification and member priority. Modern citizens in a pluralistic world are, unfortunately, still subject to tribal moral intuitions and preference for the Similar over the Other.

John Thacker writes:

Right, as David says.

Bryan, didn't you write a whole book about the prevalence of anti-foreign bias? It's a related phenomenon.

For me, the libertarian attraction has never been that people aren't motivated to discriminate (though I grant the role of statistical discrimination that is based on things reasonably accurate in the aggregate but painfully unfair to the individual), but that the free market makes it expensive to discriminate and discourages it. It also allows a diversity of approaches, so that no one winning in-group coalition is able to impose its rules on all other groups.

Some people, however, find it unacceptable because it still allows some people to discriminate when they find a particularly high value in doing so, and when that form of discrimination would be distasteful to the ruling political coalition of the time, it would be prohibited and driven underground.

david writes:

If you're looking for genetic payoffs, think about contagious disease transmission, which rewards literal isolationism.

Foobarista writes:

Some situations:

1. In a "discriminationist" society, you don't have an incentive to not discriminate in general. If you have a society of blue and gray people, the blue people are 90% of the population, and they will not generally buy stuff from a gray salesperson, you'd only hire a gray salesperson as a "niche hire" to sell to gray people.

2. In a post-discriminationist society with lots of laws making it hard to fire people and general social disapproval of formal discrimination, you have less business incentives to discriminate. But you perversely have dangerous exposure in hiring.

Let's say that we now have laws that award big bux if you fire a gray person if it can be proved by clever lawyers that you fired them over their color. Let's also say that with immigration, we have lots of red, orange, and green people running businesses, but who are also subject to laws covering firing gray people. They are now incented to only hire can't-miss gray rock-star hires, as hiring marginal gray people could expose them to big-bux settlements that they don't have to pay if they hire marginal workers from "non-protected" groups.

So, the net is that you have gray-person unemployment that is not helped by the anti-discrimination laws.

david writes:

"The free market punishes discrimination" is an odd way to look at it.

No, in the Beckerian analysis, the free market charges a fee for discrimination. It 'punishes' discrimination to the exact extent it 'punishes' a like for, say, chocolate.

Maybe some people might be willing to pay that fee, for reasons of taste, or for efficiency reasons. It is the choice between these two that Caplan is asking about.

Milton Recht writes:

Harvard's Robert Putnam takes the view, based on his research, that diversity leads to mistrust. Putnam found mistrust in both the majority and the minority populations in diversity situations.

Trust between an employer and an employee is necessary in all work environments. Employers cannot wait until there is an established long job history to trust an employee.

Greater trust between worker and employer has economic and non-economic benefits to both the firm and the employee. Lower monitoring costs, fewer agency problems, more delegation and responsibility with less supervision, lower employee turnover due to perceived fairness, safety, etc. by employees.

Even within firms that hire a diverse workforce, many job functions and specializations within the firm will group by some identity characteristic, such as gender, race, geographical origin, age, etc.

Since mistrust has a cost to the employer, that cost is balanced against the cost of diversity non-compliance.

The open question is not the lack of trust between different identity groups, but whether the lack of trust develops from a history of living in non-diverse neighborhoods, towns, etc. or whether is is a more fundamental human trait.

Tracy W writes:

How many people really enjoy inflicting unmerited suffering?

Judging by history, an awful lot. The Stanford prison experiment springs to mind, for a start.

Chops writes:

Two hypotheses:

(a) While statistical discrimination is not hateful, it can have self-fulfilling results when broadly applied. A society aiming for equal individual opportunity may rationally force employers to do costly individual-based hiring, forbidding the use of shortcuts in judging talent. Of course, enforcing this is tricky: one may hire systematically fewer Hispanics if Hispanics have systematically lower-quality education, without once discriminating.

(b) Suppose government implements various 'affirmative action' programs to give blacks unfairly favorable treatment in education. Employers who are interested in the role of educational attainment as a signal of underlying ability will rationally discriminate against blacks to 'wash out' the effect of affirmative action. In this case, the market is making the government's first action ineffective, so government may respond by outlawing statistical discrimination, simply in order to make its own discriminatory practices more effective.

egd writes:

I think I am a good worker. I think I am a good worker because of my background. I want to hire good workers. Therefore I want to hire people who are like me.

Where "like me" translates to "Harvard Graduate," "female," "middle-class," "hispanic," or "Democrat," depending on individual bias.

I also suspect there's a lot of misunderstood statistical discrimination. Middle class college-educated white men may have a reputation as disproportionately successful. Therefore firms will tend to hire middle class college-educated white men on the theory that they are more likely to be successful.

Maybe the first is a subset of statistical discrimination.

roystgnr writes:

What is the optimal solution in an iterated "Prisoner's Dilemma" problem? It's something like "Tit-for-tat" (with minor adjustments depending on the exact version of the problem). In other words: "behave honorably towards them until/unless they behave dishonorably towards you.

What is the optimal solution in a non-iterated "Prisoner's Dilemma" problem against an opponent of unknown character? It's "always defect". In other words: you have no incentive not to screw them, and don't feel bad because they'll be doing the same to you.

How do you tell whether your problem is going to be iterated or not? As evolutionary stories go, "they and their kin look so different from me and my kin that obviously nobody's been sticking around long enough to interbreed" is probably a half-decent clue.

ChrisG writes:

I would not look to game theory or evolutionary psychology to explain discrimination. Although you might come up with some tribal unity "just-so story" if you're feeling speculative. I suspect it is simply a characteristic of how the neocortex operates. I recommend watching some lectures by Jeff Hawkins from Numenta - where they are busy modeling the neocortex in software.

andy writes:

Anti-discrimination law = thou shall not commit thought-crime.

Anyway, I don't think people are disciminating so much [except statistical discrimination]. The more people were discriminating, the more could you profit by hiring the discriminated people...unless the government comes with equal-pay-for-equal-work laws, of course, that would slow down market forces a little...

Yes, there are people who want others to suffer - I just wonder, how is that supposed to work. "I won't hire you, I want you to suffer"? It would seem to me to be quite a stupid idea having somebody like this as a hiring person in a company... It's more likely "I don't like this person, I think he will be less productive". Of course people tend to think some particular people are on average less productive...therefore if you want to hire more productive people, you would tend to hire people who.. well, tend to be more productive given the 'lack of information' one faces.

Anti-descrimination laws = you are not allowed to judge applicants by their expected productivity.

Floccina writes:

The workers have reason to favor discrimination and this could lead management to want to discriminate to placate them.

guthrie writes:


"...contagious disease transmission... rewards literal isolationism."

...unless you're the microbe! :)

Am I the only one old enough to remember when it was a compliment to be called a 'discriminating person'? It meant that you were someone who made wise choices.

And I think this;

The more people were discriminating, the more could you profit by hiring the discriminated people....

is exactly to the point. My brother and I built a successful small business virtually from scratch doing exactly this. We looked for people who had talents useful to us who were underemployed.

The first such was a bartender at a restaurant we frequented. After observing him in action a few times, my brother said to me, 'You know, we could use that guy's talents.'

Which we did. We ended up tripling his income. Btw, he was a black, high school graduate and most of our customers were white blue collar workers.

Raphael writes:

Other commenters expounded on the mechanisms that trigger racial/ethnic/religious discrimination, so let me instead write about the next step in reasoning: Since some forms of diversity, ceteris paribus, impose costs on the society, by decreasing trust, increasing brittleness, overall reducing economic efficiency, the interesting question is how to deal with diversity where it is for historical reasons, and unfortunately so, present. I agree with Bryan that government-directed anti-discrimination laws affecting private businesses are a cure worse than the disease. But what are the superior alternatives? Voluntary geographic separation followed by trade across borders? Extensive voluntary reciprocal behavior monitoring and shunning of hateful (i.e. "motived by hate", not "hated) individuals? An explicit quota system in hiring by government?

Ken B writes:


I think I am a good worker. I think I am a good worker because of my background. I want to hire good workers. Therefore I want to hire people who are like me.

An interesting question is, what if you are right? Right I mean for all the different groups you mention and more. Imagine you are a Xlian in a large city with a very heterogenous mix. New York in 1850 say. You might trust Xlians more then Qlians, and Xlian workers might trust you more then Plian bosses. You might also be better at communicating even if you all speak English, not Xlish, due to shared presumptions. Assortative mixing becomes rational.

Ken B writes:

Patrick R Sullivan

Am I the only one old enough to remember when it was a compliment to be called a 'discriminating person'?

Nope, but you're the only one honest enough to admit it. :) I bet you were once a gay young blade in rude health too. Interesting how language changes over time! Stigma bleeds to secondary meanings and words disappear.

Although I do still hear about a 'discriminating palate' with wine. But elitism is less of a stigma with *ahem* oenology.

Silas Barta writes:

Wait, you can understand the evolutionary history rewarding adultery, but not behind showing preference for the in-group?

MingoV writes:

My wife and I have experienced age discrimination in hiring. Employers do not want to hire people older than 50 because they believe: 1. that their health insurance costs will increase, 2. that older employees will have more absences, 3. that older employees will expect more pay, 4. that older employees are too fixed in their ways, 5. that older employees who experienced different workplace cultures will be disruptive, 6. that older employees will readily recognize how bad the managers are, etc. Those factors outweigh the very low risk of successful prosecution for agism, so unjust treatment of older applicants is rampant.

*I, too, agree with Patrick R Sullivan. Discrimination should be replaced by the phrase unjust treatment.

JVA writes:

@Silas Barta

How does evolution explain existance of 1 billion strong in-group? Also, isn't there evolutionary pressure to win over superior members of out-group?

Ken B writes:

@JVA: Very simply. 1B sized groups are a new phenomenon, the mechanism evolved in smaller groups. You are right to note that, with groups that big won't there be an incentive to discriminate more finely? There will be but evolution works on the brain rather more slowly. Huge groups are a very, very recent thing.

I would say rather that the demonstrated preference we see on the huge scale, is evidence for some inate mechanism. Just as Silas suggests.

(Of course you should expect and do see cultural evolution having a similar effect. Culture changes faster than genes.)

regularjoeski writes:

I am unconvinced that absent governmental pressure that "discrimination" lasts. If group a is systematically discriminated against absent actual market data at some point someone will notice and take advantage of the price discrimination, cf Moneyball. What we call discrimination is probably using statistical data to increase your odds of hiring a successful candidate for the job. If discrimination worked without a need for government than there would have been no need for Jim Crow laws. BTW- Firms hire older workers at lower costs because they can, ie those firms who do not discriminate get a better worker at a lower cost. The health insurance costs for older workers just decrease the payroll component of compensation. All things being equal an older employee must be more efficient to justify an increased total cost of employment at equal salaries.

Zippy writes:

I think that andy has it basically right: most businesses don't engage in discrimination to any meaningful extent. Indeed, if we're talking about race, most bend over backwards for marginally-competent black employees.

Indeed, they probably don't discriminate as much as they should. Affirmative action is so deep and so pervasive at law schools, for example, that simply not hiring black lawyers would be the rational strategy. This is doubly true in a world of anti-discrimination laws, because one's chances of being sued for firing a marginally-competent black lawyer are virtually 100%.

Comments for this entry have been closed
Return to top