A new study finds that male analysts who have features that are perceived as “dominant” have a competitive advantage, while female analysts who also have “dominant” faces are at a competitive disadvantage, reports an article in The Wall Street Journal. The study, co-authored by Siew Hong Teoh from UCLA, Lin Peng of Baruch College, Yakun Wang from the Chinese University of Hong Kong, and Jiawen Yan, a student at Cornell University, found that male faces that are perceived as trustworthy receive more information and have more accurate forecasts than those with faces not considered to be dominant. For female faces, the reverse was true.

Investors—particularly institutional investors—had stronger reactions to earnings forecasts adjustments when they were issued by analysts who looked more trusting. “Our research suggests that face impressions influence analysts’ access to information and the perceived credibility of their reports,” Professor Teoh wrote in the paper. But while many human beings form fast impressions based on facial features, those first impressions are usually inaccurate, the article contends.

The study took almost 800 stock analysts working in the U.S. between January 1990 and December 2017 who also had LinkedIn profile photos as of 2018 and examined their facial features using facial-recognition software. After applying algorithms to measure facial points for trustworthiness, dominance, and attractiveness, they compared those impressions with the analysts’ quarterly forecasts. The research found that analysts whose looks ranked as the most trustworthy were 25 cents more accurate in their earnings-per-share predictions on a stock valued at $100 than analysts who looked less trustworthy. Meanwhile, analysts with dominant features were 52 cents more accurate than analysts with non-dominant faces, according to the article.

The correlation between trusting faces and accurate forecasts was much stronger before the SEC enacted the Regulation Fair Disclosure in October 2000, which sought to limit exclusive access to insiders by analysts. But while executives appeared much more willing to give “trustworthy-looking” analysts more information prior to the rule, that bias still exists, Professor Teoh told The Journal. Now, those analysts are likely getting their information from sources other than company executives, such as industry experts and their peers, but they are still getting more and better information than those with less-trustworthy features. And the correlation between having a “dominant” face and accurate forecasts was substantial both before and after the regulation was created, indicating that male analysts with dominant features will always have access to more information, the article contends.

Meanwhile, the study found a marked gender divide. Female analysts—no matter what they looked like—had more accurate forecasts than their male peers on average, but women whose facial features were perceived as “more-dominant-looking” had forecasts that were significantly less accurate than male analysts. That would suggest cultural stereotypes coming into play, with dominant-looking women being viewed in a more negative light while dominant-looking men are seen positively.