Right Now | Attractive Returns
The Beauty Bounty
Beauty is Natures coin, John Milton wrote in 1634. It is currency in todays labor market as well. Since 1994, numerous studies have found that workers of above- average beauty earn 5 to 15 percent more than those with below-average looks. Those differences are of a similar order of magnitude as the premiums we associate with race and gender, says associate professor of economics Markus Mobius. Most research on what creates this beauty gap has relied on observational and empirical data. But in Why Beauty Matters, published recently in The American Economic Review, Mobius and Tanya Rosenblat, professor of economics at Wesleyan University, designed a novel experiment to decompose the root causes of the beauty premium.
We attribute all kinds of positive things to people who look good, says Mobius. We think they are smarter, more sexually active, wealthier, and generally happier people, so why not think that theyre more productive and better at doing certain things? Or pure discrimination could be a cause: take the case of a boss who likes having attractive workers around and so passes over the plain and ordinary. There may also be skills associated with beauty: pretty children, favored by teachers and peers, may acquire better social skills that enable them to negotiate higher wages, explains Mobius. Social psychologists have shown that these positive expectations in turn boost confidence, he adds, citing another quality known to increase wages.
Determining which factors produce the beauty premiumand to what degreemeans going inside the black box of the negotiation process. We dont really know whats going on when a worker is being interviewed, notes Mobius. Does the employer decide the worker has certain attributes, or does the worker convince the employer of those attributes? In an experiment, because you can create an artificial labor market, you can very easily control how employer and worker interact. You can also know precisely what the worker’s confidence is when he goes into the negotiation, because you can ask him [beforehand].
To create their experimental market, Mobius and Rosenblat used a computer game for which beauty was not a factor in the labor involvedsolving as many mazes as possible in 15 minutes. They had passport-style photos taken of all participants (330 university students in Argentina), who were rated on a scale of 1 to 5 (plain to above-average beautiful) by high-school students. Each experimental session involved 10 participants, half playing workers and half employers. Workers solved a practice maze and then estimated how many they could solve during their quarter-hour employmenta measure of confidence.
To isolate various channels in the negotiation process, Mobius and Rosenblat assigned each employer a different mode of reviewing workers: one saw only rsums (baseline), while the others also utilized either a photograph (visual), a phone interview (oral), a photograph and phone interview (visual and oral), or a photograph and a face-to-face interview. Employers then estimated the expect- ed productivity of each worker, which counted as his or her wages.
Who won this game? Beautiful workers earned wages 12 to 17 percent higher than their ordinary-looking coworkers, Mobius and Rosenblat found, even though their maze-solving productivity was no greater. There was no beauty premium when employers saw rsums only, but there was a 12 to 13 percent wage increase when they used photographs, phone calls, or both. In the case of face-to-face interviews, the beauty premium was 17 percent.
Mobius and Rosenblat identified three channels through which physical attractiveness increased an employers estimate of a workers ability, and their relative importance. The visual stereotype, that more attractive people are more productive, accounted for 40 percent of the beauty premium. Attractive workers also displayed greater confidence, by giving higher estimates of their predicted productivity, a channel producing 20 percent of the beauty gap. The most surprising result, however, was that beautiful people perform better even in telephone conversations, where their beauty isn’t obvious, says Mobius. The oral stereotype, by which the conversational skills of attractive workers suggest capability sight unseen, accounted for the entire remaining 40 percent of the beauty premium. There hasn’t been any hard evidence on the last two channels before, Mobius notes.
But the policy implications of these findings are less clear. Moving to blind phone interviews, for instance, would block the face-to-face channel for visual stereotyping (40 percent), but leave open the channels for confidence and oral stereotyping (20 and 40 percent), Mobius points out. Employers could eliminate interviews entirely and hire only via print rsums (without pictures), but that, he believes, would be too extreme. Although such a policy would eradicate most of the beauty premium, the losses would outweigh the gains.
The tradeoff is: How important is it to reduce discrimination against not-so-beautiful people, [compared with] How important is it to allocate people efficiently in the labor market? Mobius says. Eradicating it completely by going to typed CVs without any human interaction wouldnt be worth it. The productivity loss of assigning applicants wrongly would be much higher than the reduction in discrimination against the less beautiful.
The right policy conclusion for Why Beauty Matters may be to do nothing, Mobius concludes. But he also imagines a follow-up experiment with an additional consciousness-raising component. We could use a training manual for a second group of employers, giving them evidence on what sort of factors tend to matter. If we show them pictures of four different employees and [report] that their productivity was the same, then perhaps the employers would be less likely to fall victim to these stereotypes. If so, perhaps the labor market would function according to another famous adage: Handsome is as handsome does.
~Harbour Fraser Hodder
Markus Mobius e-mail address: firstname.lastname@example.org
Mazes used in experiment: http://games.yahoo.com/games/kidsmz.html