Systemic Racism — Vicious Circles

by Lawrence Blume and Raji Jayaraman

The United States is undergoing a long-overdue reckoning, in the highest echelons of government, with the problem of systemic racism. The new Biden-Harris administration has declared that “The moment has come for our nation to deal with systemic racism…” The wide span of policy remedies it goes on to propose reflects the breadth of the problem. A further challenge is that systemic racism runs deep, operating not only at the level of interpersonal interactions but from the workings of our social systems. It may be sustained by formal structures such laws or rules, in which case an obvious remedy is to reform those structures. But it can equally be sustained by amorphous structures that are harder to combat.

One of the most insidious mechanisms through which systemic racism can be sustained is the vicious circle of self-fulfilling beliefs. Here is the idea. Suppose there are two groups, Blacks and Whites. Blacks hold beliefs about Whites, and Whites hold beliefs about Blacks. They interact with each other, choosing behaviours based on their beliefs. Everybody’s beliefs are validated by the behaviours they see.

We say that these beliefs are “in equilibrium” because each group’s belief enables the other. The representation each group holds about the other is correct because and only because each group holds these representations. When a large number of individuals in a market or some other social structure interact—and this is where the “systemic” part of racism enters—the beliefs turn into a self-fulfilling prophesy. Positive, “pro-social”, behaviours in this context are sustained by positive beliefs, resulting in a virtuous circle. Negative, “anti-social”, behaviours are sustained by negative beliefs, resulting in a vicious circle.

In his brilliant book, An American Dilemma, Gunnar Myrdal saw racial disparities in the United States in precisely this light. “White prejudice and discrimination”, he writes, “keep the Negro low in standards of living, health, education, manners and morals. This, in its turn, gives support to White prejudice. White prejudice and Negro standards thus mutually ‘cause’ each other.” While Myrdal implicated deliberate discrimination in this process, subsequent work by Kenneth Arrow and others has taught us that the vicious circle operating in systems with no deliberate prejudice, can still generate discriminatory outcomes.

How does this happen?

A good starting point is to imagine a hypothetical world with no systemic racism. Take the example of a labour market and imagine two groups of people: Blacks and Whites. Suppose employers are Whites, and no one is deliberately racist in the sense that they harbour personal animus against Blacks. Jobs require skills and different workers have different skills, but a Black worker and a White worker are equally likely to have any given skill set. Suppose too that some skills can only be seen by the employer once the worker is on the job; they cannot be observed at the point of hire. Since skill levels are not correlated with race, race tells an employer nothing about a worker’s unobservable skills or productivity. In this world, labour market outcomes—wages and employment—would be the same for all Blacks and Whites with the same set of observable skills.

So far in this hypothetical world, workers have been magically endowed with skills that make them productive on the job, but the truth is that these skills need to be acquired through schooling and families. Suppose now that parents can make investments that boost their children’s skills. For example, they move to a neighbourhood with a better elementary school, or enrol their child in a preschool or after-school program. These investments make children more productive later on as adults but are costly to parents in terms of time, money, and effort. Employers cannot tell which parents have made these investments, and many of the “soft” skills acquired through these investments are not readily observed on a job application or in an interview. Employers can only guess at applicants’ skills they cannot see; they have beliefs about the unobserved skills of each applicant. What these beliefs will be conditioned on depends on what the employer can see, and this includes race.

To see where vicious circles come from, suppose for the moment that an employer faced with a Black applicant and a White applicant whose applications are otherwise identical, believes that the Black candidate has fewer unobserved skills and will therefore be less productive than the White candidate. In this case, Blacks will be less likely to be hired, or will be paid less, than Whites. This is the first half of the circle. Here is the other half. If Blacks are not going to be rewarded for these additional investments at the same rate that Whites are, then it is rational for Black families to invest less on this dimension than do White families. Consequently, a Black worker with given observable skills is less likely to be as productive as a White worker with the same observable skill set. Tragically, the circle is closed: the employers’ beliefs that Blacks have lower productivity are validated. Race has become a signal for productivity, and so the market sorts on it. The result is higher unemployment and lower wages for Blacks than Whites.

We told this story as if the “cause” of labor market discrimination was employers’ beliefs. We could just as easily have begun our story by writing, “Blacks believe that the return to skill investments is lower than do Whites”, and followed the chain to the same conclusion. It really is a circle; no beginning and no end. This is what we mean when we say that beliefs are “in equilibrium”. Blacks’ and Whites’ beliefs “cause” each other.

This story may be simplistic, but that is what makes vicious circles so perfidious: it does not take much to end up with this outcome. In the equilibrium of our hypothetical world, people making totally reasonable individual choices free of any deliberate racism, perpetuate a racist society. This society is propped up by a system of beliefs that are self-reinforcing: each group’s behaviours validate the other’s beliefs, and each group’s beliefs reinforce their behaviour.

The hypothetical world does, of course, support a different story where race does not trigger beliefs at all and there is no racial disparity in outcomes. In principle, this too would have been feasible. But history tipped the scales in one inexorable direction, so that today a correlation of skin colour with unobservable workplace skills is enough to sustain discriminatory outcomes even in the absence of deliberate racists.

In the system we described, White workers are the winners and Blacks are both the losers and the victims. This much is obvious. Somewhat less obvious is who the perpetrator is. White employers hold negative beliefs, but these beliefs are consistent with what they see. Blacks reinforce these beliefs, but their behaviour is a rational response to the situation they are stuck in. A White employer who chooses to ignore her labor-market perceptions will be less profitable than her competitors, perhaps unsustainably so. A Black family who spends money on after-school programs that may not matter much to a child’s future earnings will have less money for rent or medical care. In our view, the actors making these decisions are blameless. Nonetheless the system is clearly immoral. Moreover, it acts counter to the economic interests of society as a whole.

Amplification is another important feature of vicious circles. Rewind to the equilibrium of our baseline hypothetical world, where beliefs and behaviours supported equal outcomes for Black and White workers. Now introduce into the employer pool some deliberate racists who will always choose White applicants over Black applicants. This lowers the return to Black families’ investment in skills, and raises it for White families. Blacks should now be less willing to invest than are Whites and the skill difference becomes pervasive. Employers who are not deliberately racist will adjust their beliefs accordingly, and the labor market moves to the discriminatory equilibrium. The deliberate racism of the few has infected the beliefs of the race-neutral majority. It has made no one deliberately racist who was not already, but it destroyed the validation of race-neutral beliefs, leaving only the discriminatory outcome. As Myrdal puts it, the introduction of a few racist employers “start[s] a process of interaction where the change in one factor will continuously be supported by the reaction…. The whole system will be moving in the direction of the primary change, but much further.”

How is the circle to be broken? In the labor market, changing beliefs is not an option because it is neither affordable nor effective at the scale necessary to shift the entire market. Changing behaviour sounds more promising, but inventing programs to subsidize skill acquisition cannot move the needle, for a subtle reason. Suppose we could simply endow Black workers with the additional skills that underinvestment lost. What happens next? If employers are not willing to hire Blacks, or if Blacks are hired only into jobs where the skills are not used, employers will never see these additional skills, and beliefs remain stuck. We can imagine that over time the additional skills are recognized, and beliefs begin to move as more and more employers have better-than-expected experiences with Black employees, but the needle will move very slowly.

Another possibility is to subsidize employers to give Black workers a shot. Again, the needle will move only slowly, this time because Black families will need to be convinced that the investment will pay off. The remaining possibility is a two- pronged strategy, one which lowers Blacks’ cost of skill acquisition and increases Whites’ incentive to hire Blacks. And both of these programs will have to operate on a massive scale. The two-pronged attack is hugely expensive. This should not be surprising. Consider how long it took and how much effort was expended to create our racist social structure to begin with. We should expect that fixes will be neither quick nor cheap.

For Myrdal, vicious circles are a negative interaction of behaviours and the social construction of beliefs. The behaviours and beliefs are self-reinforcing, in a positive-feedback loop. The labour market is one of many places in which it is manifested. Moreover, such loops are not confined to behaviour–belief interactions alone. We can find positive-feedback loops—vicious circles—in housing markets, college admissions, policing, and elsewhere throughout society. The blunt reality is that vicious circles are pervasive, pernicious, and hard to break out of. The adaptive capacity of individuals in such systems to adjust to their circumstances, be they good or bad, makes system outcomes resilient to change. If indeed “The moment has come for our nation to deal with systemic racism”, the Biden-Harris administration will have to address this kind of systemic racism with social policies that work at large enough scale to be truly disruptive. It’s hard to imagine that the political will needed to accomplish this will match the soaring rhetoric.


Lawrence E. Blume is a Goldwin Smith Professor of Economics and professor of information science at Cornell University. He is a visiting research professor at IHS Vienna and a member of the external faculty at the Santa Fe Institute, where he has served as co-director of the economics program and on the institute’s steering committee. He teaches and conducts research in general equilibrium theory and game theory, and also has research projects on natural resource management and network design. A Fellow of the Econometric Society, he received a BA in economics from Washington University and a PhD in economics from the University of California, Berkeley.

Raji Jayaraman is a regular 3QD columnist and her bio can be seen here on our Mondays page or by clicking her name just under the title of the post.