Study: To Maximize Productivity, Affirmative Action Should Continue Indefinitely

A common view is that affirmative action is a temporary policy, designed to end once glaring inequities have been resolved. If the main intent of the practice is to correct the imbalance caused by past injustices, affirmative action will no longer be needed once minority groups are well-represented in the workforce.

The assumption is that we can “make it fair and then stop,” says Aniko Öry, an associate professor of marketing at Yale SOM. “The problem is solved and we’re done.”

But a study by Öry and Michèle Müller-Itten, an assistant professor of economics at the University of Notre Dame, suggests that this thinking is misguided when it comes to groups that are minorities in the general population. If the goal is to increase an industry’s productivity by drawing and retaining the most talented workers, then affirmative action aimed at those groups needs to continue indefinitely, they found.

Öry and Müller-Itten’s analysis did not set diversity in and of itself as the goal. Instead, they focused on how to most efficiently harness talent from majority and minority groups.

“We’re just maximizing the total productivity,” Müller-Itten says.

A critical factor in their analysis is mentoring, which tends to be weaker when the junior and senior workers are from different demographic groups. As a result, it can be more difficult for employees from underrepresented groups to find effective mentors and advance in the profession—they just have fewer good mentors available to them. According to Öry’s and Müller-Itten’s analysis, in this situation, the discontinuation of affirmative action policies leads to an inefficient underrepresentation of minority groups, unless members of the majority group can be trained to mentor them effectively.

“You need to persistently apply affirmative action,” Öry says. “It can never end.”

The idea for the study arose when Öry and Müller-Itten were PhD program classmates. They began thinking about the barriers holding back women and underrepresented minorities in professions such as economics.

One factor that clearly mattered was mentorship. Previous research had shown that these relationships arise less frequently when the junior and senior worker are of a different gender or race. And students from underrepresented minority groups tend to perform better academically when their teacher also is from such a group. But minority students often have trouble finding effective mentorship or role models because of their group’s relative sparseness in the workforce.

Öry and Müller-Itten created a mathematical model to capture these patterns. The model includes workers from majority and minority groups. Each person is assigned a certain talent level and has to choose whether to pursue a particular profession. The availability of mentorship makes them more likely to do so, but only if the representation of mentors from the same demographic group in the pool of senior workers is sufficiently large. People who do enter the profession add to that industry’s productivity.

The researchers then determined what makeup the labor force would naturally converge toward if no policies were in place to boost minority participation. They found that the industry would settle into one of a few steady states. Some are completely homogeneous—for example, 100% majority workers. A more realistic mixed workforce is also possible. But if the majority group makes up more than half the workforce at the starting point, the extra boost that its members receive from effective mentoring eventually causes the majority to be overrepresented.

Next, the team considered how the industry could achieve the most productive workforce—that is, make the most efficient use of the talents of potential workers in both the majority and minority groups.

Short-term affirmative action was not enough, they found. If these practices are eventually removed, the industry drifts back to underrepresenting the minority group. In this scenario, contributions from talented minority workers are lost, reducing the profession’s overall productivity.

“If you don’t have any policy, you will always converge to this underrepresentation of the minority,” Öry says.

In fact, the model suggested that achieving maximum productivity would often require overrepresenting the minority group in the workforce. For example, if the general population was composed of a 70% majority and a 30% minority, then the profession would need to be tilted toward the minority—say, 65% majority workers and 35% minority workers.

At first glance, such a system might seem unfair. But the researchers explain that it simply takes into account minority students’ ability to recruit talented junior members of their group in the future.

For example, if a majority and minority student have the same qualifications on paper, they might seem equally valuable. And both can effectively mentor others from their group once they rise through the workforce. But the majority student’s mentoring ability “is not in short supply,” Müller-Itten says. The minority student’s mentoring ability, in contrast, is a sparse resource. “One has a rare skill, and the other one does not,” she says.

The findings don’t hold true if the two groups have roughly equal numbers in the general population—for instance, men and women. In those cases, interventions can be discontinued once the workforce is brought into balance, Öry says.

But for racial and ethnic minorities, policymakers will need to enact long-standing affirmative action, the team says. Scholarships for all minority students would be one effective way to lower the cost of entering a profession, according to the model. Hiring quotas for minorities could work, but only if the increased demand for minority employees is reflected in higher salaries for them as well—which may not happen in reality.

Industries also could train people from the majority group to become better mentors to minority students, but the effectiveness of such interventions needs to be tested. Encouraging social mixing of junior and senior workers from different demographic groups might help.

Business schools should recognize minority students’ future mentoring capabilities, Öry says. Boosting their numbers “can generate a lot of surplus and benefits both to society and also companies,” she says. “It’s a waste to not use that talent that is out there.”