Markets have been shown to be effective to a degree in combating certain types of discrimination. Anti-discrimination laws have drawbacks we shouldn’t ignore.
Miron: The topic for today is discrimination. As I’m sure you know, the U.S. and most countries have numerous policies that attempt to ban or limit discrimination in employment, in housing, admissions, sports, so on. These are based on discrimination with regard to race, gender, disability, sexual orientation, and more. Many of these policies have been quite controversial over the decades. To many people, the value of having anti-discrimination policies is self-evident. It’s taken as an unambiguous given not only that we want to eliminate discrimination, reduce it as much as possible, but that therefore, we should adopt strong policies that try to do just that. But to libertarians, the case is not so clear. While we certainly find discrimination abhorrent in most settings, we think that we need to consider the possibility of private efforts that will tend to reduce discrimination. We need to think about the magnitudes of discrimination that will occur even in the absence of policy. We need to think about whether the particular policies that we’ve used are actually effective in reducing discrimination. And as always, we need to think about the cost and the unintended consequences that might be generated by attempts to reduce discrimination. So this lecture is going to examine those issues.
The outline goes as follows. I’m going to mainly focus on employer discrimination, just to fix ideas, just to keep it narrow enough that we can cover the issues in some detail. And I’ll break that into two parts to talk about the theory, the economics of how much discrimination might occur in a market economy in the absence of government policies that try to change that, and the evidence on how much discrimination actually occurs in market economies absent government regulation. Then a look at the kinds of policies that economies might use to try to reduce discrimination, look at evidence on those policies and on private efforts to reduce discrimination, and talk about the costs of anti-discrimination policies and think about how that all balances out and where that leaves us as to what policies might be most beneficial and most respectful of liberty.
Employer discrimination has a long-standing history of being studied and modeled by economists. And there is a strong but very simple story about how much employer discrimination we should expect in a market economy, and the simple story says very little or zero. Why is that? Because an employer that discriminated, that deliberately only wanted to hire, say, whites instead of blacks would be imposing a constraint on himself or herself, would be limiting the pool of applicants, narrowing the number of people who could actually work for that employer and therefore face higher wage cost than someone who’s willing to hire everyone. In particular, as long as there are some employers who only care about the productivity of their employees and are willing to hire the, say, African American potential employees, then they will have a cost advantage. They will expand and they will drive out those employers who try to only hire white employees because those employers will face higher cost by not taking advantage of the fact that there’s lower-cost labor available in the presence of this sort of discrimination. So firms that discriminate lose money. Firms that don’t discriminate make more money. So those businesses will thrive, survive, and expand at the expense of those that discriminate. And that should limit employer discrimination quite severely. This is not a story that says that there are people who are benevolent and they go and try to hire people in a minority group simply to try to do good. There may well be people like that. But even if there are just purely self-interested businesses, they will tend to reduce the incidence and the impact of discrimination simply by trying to make more profits.
Now this might not necessarily sound compelling ex ante. But in fact, there are lots of great examples of exactly this kind of thing happening. The history of Major League Baseball is one of the most famous illustrations. In the 1940s, the Los Angeles Dodgers hired Jackie Robinson, the first African American to play in the major leagues, in the white major leagues. Why? Partially because the owner was benevolent. He thought that the discrimination by the major leagues was abhorrent. But partially because he knew there were great baseball players still playing in the negro leagues, and he could give his team a competitive advantage by hiring someone. He did that, other teams observed it, realized that they were going to be toast competitively if they didn’t follow suit. And within a decade, there were backs playing on almost all the teams in the major leagues. Nowadays, there’s a huge number, a huge diversity of backgrounds playing in all levels of sports because it was in the interest of those sports to stop discriminating. It had nothing to do with any government requirement or any government imposition of affirmative action or anything like that.
The history of the entertainment industry is similar. Clearly, blacks, Hispanics etc. are well represented within entertainment these days, not because of affirmative action or government policies mandating that they be employed to star in movies or the like but because it’s always been good business to have used people from all races in the entertainment industry.
I’ve heard anecdotes from several friends who happen to be in the law business that they stopped discriminating against Jews in their hiring practice when they realized that the major law schools were graduating lots of talented Jewish students, and they were going to lose out to the Jewish firms if they didn’t start trying to compete for those Jewish law students along with the existing Jewish firms who had been tending to hire the Jewish students from the law schools. Ivy League schools generally, other similar schools typically, had discriminated against Jews and women. Harvard had a quota for a maximum number of Jews it would admit into its class for several decades going back into previous century. Other schools never admitted women, the entire Ivy League and many other schools. And they changed for the most part long ago, well before there was government-imposed affirmative action of any kind. They changed because they were losing good students to other schools that admitted Jews, that admitted women without any prejudice.
So there are plenty of examples that illustrate this basic idea. Markets tend to put some limits on employer discrimination. Maybe they don’t eliminate it entirely, but they certainly tend to discipline it to a very significant degree.
That doesn’t mean that markets are going to eliminate everything, every kind of discrimination. In particular, markets cannot as easily solve customer discrimination. Imagine that all the customers who’ve ever tipped at a restaurant want to be served only by white waiters. Then it’s in the interest of the owner of that restaurant to only hire white waiters. Otherwise, his customers will go away. So I don’t want to claim too much. Markets don’t necessarily eliminate every kind of discrimination. Even in the customer discrimination case, there’s still some pressure on employers not to discriminate. If they only hire white waiters, they do face a smaller pool. If they can nudge their customers into accepting a broader range of waiters, then they can lower their labor cost. So the magnitude of discrimination, even under that scenario, even under customer discrimination, might be modest. But that kind of discrimination may indeed persist to some degree even under market forces.
Now let’s think about the evidence on how much discrimination there has been, how it’s changed over time, and think about that in relation to these theories of how much discrimination should occur. So how do we look at the issue of discrimination empirically? How do we measure it? The main way is by comparing the wages, or sometimes the employment rates, of different groups, of black versus white, of men versus women, and so on. You can certainly first form the raw comparison. Just look at average wages of African Americans, average wages of white Americans. More sophisticatedly, you control for a whole set of other things that might on average be different between those groups. What you find is that when you take account of these other factors, these controls, of who has more experience, of who has more education etc. because historically, blacks went to worse schools, blacks had fewer years of education, were younger on average than whites, some of the observed gap is explained away by these factors that plausibly explain wage differences even without a discrimination, just because markets reward experience, markets reward good training and background and all that.
Still, there’s a gap left in all these studies that try to measure discrimination. Even after controlling for everything you can measure, it does still appear that whites are more than blacks, men are more than women, and so on. So the question is does that show that there’s still discrimination going on in labor markets. Well, maybe, or maybe not.
The remaining issue is what’s known as statistical discrimination. Statistical discrimination means there are characteristics that you think are correlated with an observed characteristic but that you can’t measure or that at least are not in data sets that economists use to measure these things. So if you only control for the things which are measurable, you might be missing that some groups may have the same amount of experience or you’ve controlled for experience, but the quality of the experience was different or the quality of the education was different. Therefore, when you do these attempts to measure wage gaps, controlling for other factors, you do it very imperfectly. And that something like indication of your race or indication of your gender helps capture something like these other characteristics, the quality of the schooling you went to and so on. Some of the remaining gaps that we see probably are statistical discrimination, not pure, naked animus-based discrimination. But there’s no trivial way to tell exactly how much is which.
Bottom line, there probably is some discrimination in labor market still. There certainly has been in the past. But it might be a lot less than people think. At a minimum, it’s way too strong to attribute all observed differences to discrimination. There are a bunch of other factors that might be explaining some of those differences in outcome.
If discrimination does occur – and we’re going to take as given that it does occur to some degree – what can policy do? There are roughly two kinds of things. It can outlaw discrimination in hiring and promotion and salaries and the like, and/or it could adopt policies that affirmatively push employers to have a certain fraction of the employees be African American or women or so on. That is rather than just punishing employers who have specifically discriminated against a particular person, you can say whether or not there’s evidence of that. You have to have a certain fraction of your labor force that is African American or women. We have to see evidence that you are trying to hire lots of African Americans or women. You have to be affirmatively taking steps, not just defending yourself against charges from specific people.
Those might initially sound like two different policies. I want to argue that they’re closely related and we should think about them together. The first point is that bans on discrimination by themselves are probably not going to do much, good, bad, or indifferent. Why? Because in any given case, an employer who’s being charged with having discriminated against someone for whatever reason can say, “Oh, I had these applicants. I didn’t choose that applicant because I thought that their education wasn’t quite right because they had a bad interview, because they just didn’t seem very energetic,” because whatever. To prove that some specific person was not hired because of that person’s race or gender, it’s going to take close to a smoking gun, an email from a manager to a colleague saying, “Yeah, this person is better qualified, but he’s black. So I’m not going to hire him.” That would be evidence that would get you a win in court. But most of the time, you’re not going to have such evidence. So if you only have laws that ban discrimination, you might reduce discrimination in a very small amount. You might create some good outcomes. But it just can’t have that much effect. Employers could circumvent it. So it’s totally understandable and natural that people said, “We have to do more than that if we want to push for less discrimination. We have to make sure that employers are actually hiring a significant number of African Americans, women, and so on.” That’s necessary if you want this anti-discrimination set of policies to have any bite.
So now we want to think about what happens if we have… What are the cost benefits not only of having these anti-discrimination laws such as in the Civil Rights Act but also the affirmative action rules and policies that have been built up around them? So let’s first ask have they worked. Has discrimination, measured discrimination, declined? The short answer is probably. So if you look at wage gaps over long periods, you see major declines in wage gaps, obviously not zero anymore but much smaller than they used to be. If you look more casually, simply whether there are African American and female senators and congressmen, obviously things have changed. Supreme Court justices, presidents of the United States, CEOs of corporations, any measure you think of, you will see things that are very different than almost predominantly male and white set of faces that you would’ve seen 50 years ago.
So at some level, clearly, there have been major reductions in discrimination, or at least major improvements in the success of these particular groups. But we don’t know, and it’s much harder to know, what caused that because the time period over which this has occurred, multiple things happened. One thing that happened was the Civil Rights Act of 1964. The Civil Rights Act outlawed discrimination in employment, housing, etc. etc. And it led to the attempt to enforce that in the form of the Equal Employment Opportunity Commission and the affirmative action rules that it uses. But in addition, roughly similar time period, if not earlier, there were federal lawsuits and the Jim Crow laws in southern states that tried to at least reduce the legal mandates for discrimination that some states had in place to try to break that pattern, to try to make it easier for private actions to take place. There were private protests and private actions to attempt to reduce discrimination, marches in the south, protesting, boycotts of lunch counters that were segregated, and things like that. They were helping to change norms. They were putting pressure on employers and other businesses to change their practices. And they were evolving social norms for lots of reasons. So attributing exactly how much of the decline and discrimination has been due to the Civil Rights Act per se or the Civil Rights Act and its subsequent implementation versus this broad range of other policies and private actions is tricky. And I think the jury is very much still out. Most importantly, the improvements clearly start well before the Civil Rights Act. So you can measure wage differentials as far back as the 1940s, the 1950s, the 1960s, before the Civil Rights Act. And you see substantial declines controlling for appropriate things across these broad groups of minorities versus whites, men versus women that well predate the Civil Rights Act. So at a minimum, it can’t get all the credit. It may be that these private forces and other factors played very big roles. So let’s assume some of the improvement was due to private actions, and some of it was due to policy steps. And it’s hard to attribute exactly how much was which.
Therefore, that says maybe policy played some beneficial role. But let’s ask about at what cost. So as noted, the bans weren’t going to do much. So affirmative action became inevitable. The problem is affirmative action has potentially significant cost, and that has to be weighed against whatever benefits it may have achieved in addition to those of the private actions. So what are the costs of affirmative action? One thing it potentially does is confirm negative stereotypes. If people who have any tendency to be racist in any way, shape, or form see that there’s someone in a job who doesn’t seem totally qualified, who is African American, it’s easy for them to think, “Oh, that person got the job because they were black, not because they were qualified.” And that tends to reinforce their view that blacks are less qualified. And of course, in any group, there are people who are more or less qualified. Some blacks in some jobs are not going to be especially qualified, no matter how non-discriminatory a world we might’ve lived in. So that is inevitably going to happen and may confirm negative views, which make changes in people’s attitudes even more difficult. Same time, if people who think they deserve certain jobs, either people of different race or a different gender have those jobs, it breeds resentment. It lets people think to themselves, “Oh, that person got the job because of affirmative action. I should’ve gotten that job.” And that creates anger and resentment, which again is counterproductive for generating long-term improvements in attitudes towards race and gender equality. Affirmative action might have the effect of discouraging women or African Americans from becoming as qualified as possible in order to make sure that they qualify for jobs. And it didn’t appear that they got those jobs because of affirmative action. Jackie Robinson had some quotes which were in this direction, saying he knew he needed to be better than anybody. He needed to be so good, so dedicated, and so committed that no one could ever think that he got the job because of his race, and that everybody would still want him to be there and respect him because he was so amazing at what he did. That’s a danger again of affirmative action, i.e. undermining that type of attitude to changing the attitude toward I don’t need to get a really good degree because I’m going to get a job because I’m a particular race or gender – even if I don’t have the best degree, even if I don’t have the best grades, and so on.
Affirmative action has the potential to reduce economic efficiency if it is significantly causing mismatch between people’s abilities and the jobs they get if there is systematic promotion of people to jobs they’re not fully qualified for. I don’t think the evidence is particularly strong for that effect. People have looked for that, and there’s not much evidence for that. I think a more serious issue is that affirmative action is one thing that changed dramatically the norms about what’s public and what’s private. In particular, the law around affirmative action ended up saying that restaurants, businesses, hotels were public accommodations, that because they interacted a lot with the public, they had to obey these rules over affirmative action. And that spread to a more general thinking that they can be regulated and told what to do more generally. For example, the restaurants can be told they have to be non-smoking. It was never a sensible policy. People who wanted non-smoking restaurants could not go to restaurants or go to non-smoking restaurants. Lots of restaurants created non-smoking sections without any government pressure whatsoever. But once you start thinking the government has the right to tell employers how they can hire – they tell restaurants who they can hire, what customers they can let in – then it’s more natural to tell them they have to obey rules about smoking, about other things. And it’s just another way the government is intruding in a way that’s not necessary and doesn’t do any good.
So bottom line, it’s hard to know for sure what the cost-benefit ratio is of government policies to reduce discrimination. The libertarian worry is that we believe in freedom. And we believe in freedom for everyone, whether a person is acting as an individual, choosing where to live, what occupation to have, and so on, or whether that individual is running a business. If that individual wants to run a business a specific way, libertarians want that individual to have the freedom to run that business however that person wants, to hire and fire people because the owner doesn’t like the color of your socks or doesn’t change his mind about whatever and doesn’t have to be interfered with by the government. But what is clear is that some people’s views and some people’s exercise of that kind of freedom does adversely affect other people who we would like to see get the jobs for which they’re qualified, irrespective of their race or their gender or so on. So it is a messy, somewhat hard question, and libertarians are somewhat divided on that. My own personal preference is we still made a mistake in adopting the Civil Rights Act. We would be better off on net had we allowed private forces perhaps more slowly to address discrimination. But in fairness, there are certainly libertarians who take the opposite side, who at a minimum think that the Civil Rights Act was necessary for a while, even if maybe we don’t need it anymore.
One last quick thing to discuss is to ask how libertarians feel about private reverse discrimination. Should private employers, schools, whatever be able to discriminate in whatever way they want? If they want to have a lot of African American students at an elite school, an Ivy League school, should they be able to do that? If they want to discriminate in favor of women in hiring in an Ivy League school, should they be able to do that? And the libertarian view is absolutely, private actions are private actions, and those private actions may be doing some good by helping groups that have historically been penalized to have an extra chance, to have an extra opportunity to succeed and to prove their merit. So long as it’s private, there should be no government interference whatsoever. And lots of this occurs and is one reason that I tend to think we didn’t need the government pressure for affirmative action for non-discrimination because private efforts clearly do want to do al those things and have done them quite successfully in many ways.
To sum up for today, the amount of discrimination is certainly not zero, but it’s probably a lot less than is frequently portrayed. The government attempts to deter discrimination, can easily evolve into complicated, potentially intrusive affirmative action program. So there’s reason to be very concerned about that, even though we find the discrimination to be abhorrent and wish it didn’t exist. Even bans, even the more narrowly tailored policy, legitimizes a counterproductive view of what is private property and what is not by saying that businesses can’t do what they want in their own private surroundings. Some libertarians are still indeed sympathetic to anti-discrimination policies, but many are also quite skeptical. The overall balance is not entirely clear as to how libertarians come out. My own view is that policy should permit private affirmative action, but it should take no other steps to discourage discrimination. Thank you.