Science has struggled to increase the diversity of the research community, trying to ensure that everyone has an equal opportunity to contribute to humanity’s advances. But science’s struggles are nothing compared to those of the financial industry, where only about 1% of fund managers are women or minorities. While there have been some efforts made to increase diversity, finance stubbornly remains the domain of white males, even though firms run by women and minorities have, on average, produced equivalent returns.

To find out why this disparity exists, a group of Stanford researchers collaborated with a diverse financial firm to perform a relatively simple experiment. They created fake financial firms, swapped in headshots of black and white “managers,” and asked actual asset managers to rate the firm’s performance. The results showed that when performance was good, having black managers led to lower ratings than when the same performance was supposedly delivered by a white-led firm. While there were some differences when performance wasn’t as high, the likely reasons for those differences aren’t reassuring.

Assets and allocators

For everything from hedge funds to retirement investments, it’s rare to have direct ownership of stocks. Instead, investments tend to go into funds that focus on specific aspects of the market, like energy or small capitalization firms. But these funds often don’t invest in the stocks directly, either. Instead, financial specialists called “asset allocators” identify firms that have funds with the right mix of performance and targets, and these allocators invest in a number of them.

The target of this new study was the asset allocators, since they play an outsized role in determining where money goes within the financial system. The research team created a series of four venture capital firms and developed a one-page summary of their past performance, management team’s education and experience, and other details. The four were crafted so that two would be considered high-performance, while two were considered lower; tests without headshots showed that each pair had ratings that were statistically indistinguishable from each other.

(The researchers made all of the fake fund managers male, as they didn’t want to disentangle two types of prejudice at the same time.)

Fake managers and fake funds

They then substituted in headshots of fake managers of these fake funds, using either black or white men. Asset allocators were asked to rate various aspects of the firm’s performance and management, as well as giving an overall rating and indicating whether they’d be interested in evaluating it for a possible investment. Each individual participant was asked to evaluate a single firm and told that their results would be used to evaluate the performance of an AI system being developed to choose funds. As the researchers put it, this would improve the evaluations they received, since “asset allocators had an incentive to prove they could outperform the computer.”

The results were a mix of the expected and the strange. As you’d expect, there was a roughly linear relationship between ratings of a fund manager’s competence and how much money their fund was expected to bring in. But the slope for white managers was much steeper than that for black people. As a result, less-competent black managers were expected to draw in far more money than their white counterparts, while the best black managers were expected to bring in less.

This was reflected in the overall performance ratings, where the firms that were intentionally made to look stronger were rated as higher quality when run by whites. But the two weaker firms saw this relationship reverse, with black-run firms rated as higher quality.

Incompetence or bias

There was some good news in terms of the results that showed no difference among the ratings: “We did not observe race-related differences in investment skills, attributions of social fit, expectations of how much the fund would raise, or the likelihood of taking a meeting with the team, beginning due diligence, and investing in the team.” And it might be possible to view the bias seen here as simply reflecting some bizarre inability to properly rate firms run by black managers.

But some additional data suggests that’s unlikely. The researchers found that after reviewing a strong firm run by a black male, the participants gave lower self-ratings of their own social status. “High-performing Black-male-led teams may in fact have induced some feelings of threat to the professional status of the allocators, who were predominantly White,” the researchers suggest. This wasn’t true when the black manager was leading a lower-performing fund. In any case, the asset allocators were unlikely to invest in these lower-performing funds regardless of who was leading them, lowering the threat further.

The results send a bit of a disheartening message to black people, who face ever-stronger bias as they become more successful. But there is a strong incentive for asset allocators to find ways to overcome this bias. As the researchers note, the allocators have an ethical and, in many contexts, legal requirement to find their clients the best returns possible. If they’re missing some of those returns because they underestimate the competence of the best minority fund managers, then it represents a major failure on their part.

PNAS, 2019. DOI: 10.1073/pnas.1822052116  (About DOIs).

https://arstechnica.com/?p=1551007