Does Beauty Actually Pay? What Thirty Years of Research Found

The number you have read is wrong. Attractive people do not earn twenty per cent more.

The figure that actually comes out of the founding study is four to five per cent — and the penalty for being rated below average is larger than the bonus for being rated above it. Most of what gets written about this cites one 1994 paper and rounds it up.

But there is a genuinely interesting finding buried a layer down, and it is not the one you would expect. When researchers separated grooming from bone structure, grooming did most of the work. For women in one study it did all of it.

Here is what thirty years of economics and psychology actually established, including the parts that cut against the thesis.

What Hamermesh and Biddle actually measured

The founding paper is Daniel Hamermesh and Jeff Biddle, “Beauty and the Labor Market”, American Economic Review 84(5), 1994.

They used three surveys in which interviewers rated respondents’ appearance on a five-point scale, from strikingly handsome or beautiful down to homely. Then they ran hourly-earnings regressions with controls for demographics, education, occupation and labour market.

Two things jump out. The effect is small. And the penalty for looking plain is larger than the reward for looking good. The paper everyone cites as evidence of a beauty premium is at least as much evidence of a plainness penalty.

The authors were careful about what this was. These are conditional associations, not a randomised experiment on beauty. An interviewer’s rating picks up grooming, dress, weight, health, demeanour and class-coded presentation alongside facial structure. Below-average-looking women were also less likely to be in the labour force at all, so the wage estimates among working women may not represent all women.

The $230,000 figure is not a finding

Hamermesh’s 2011 book Beauty Pays is where the lifetime number comes from. His summary benchmark is that the most attractive third earns about five per cent more than average-looking people, and roughly ten to twelve per cent more than the least attractive third.

The widely quoted figure of around $230,000 to $250,000 in extra lifetime earnings is an illustrative extrapolation from that. It rests on assumptions. Assumed wages, assumed career length, assumed employment differences, compounding. It is not a measured lifetime outcome and should never be quoted as one.

Hamermesh himself is more careful than his coverage. He notes that beauty can be genuinely productive in customer-facing and persuasion work, that it can produce discrimination by employers and customers, that it affects confidence and communication, and that it is partly endogenous through investment in appearance — which is the thread the next section pulls.

The experiment that shows the effect is causal

Markus Möbius and Tanya Rosenblat, “Why Beauty Matters”, American Economic Review 96(1), 2006, is the study that moves this from correlation to cause.

They built a laboratory labour market. Workers completed a real-effort maze task, and employers set wages under different information conditions: résumé only, photograph, voice interaction, or both.

Attractive workers were not more productive at the task. Employers paid them more anyway, as soon as appearance or interaction became observable. The premium ran roughly twelve to seventeen per cent — absent with résumés alone, about twelve to thirteen per cent with a photo or a phone call, and around seventeen per cent face to face.

Then they decomposed it.

That middle row is the uncomfortable one. Forty per cent of the premium was employers being wrong about who was good at the job.

The finding that changes what you would do about it

Jaclyn Wong and Andrew Penner, “Gender and the returns to attractiveness”, Research in Social Stratification and Mobility 44, 2016.

Their move was simple and nobody had made it. They used a dataset in which interviewers rated two separate things: physical attractiveness, and grooming — the presentation of hair, clothes, makeup and general neatness.

Before adjusting for grooming, very attractive people earned roughly twenty per cent more than average-looking people. After adjusting for grooming, that fell to about thirteen per cent. And being very well groomed rather than average was independently associated with around twenty per cent higher earnings on its own.

Then it splits by gender, and this is the sentence the paper is remembered for.

For women, grooming accounted for the entire attractiveness premium. Once grooming was in the model, the remaining coefficient on attractiveness was no longer statistically distinguishable from zero. For men, grooming explained about half, and a substantial association with attractiveness survived.

Now the caveat, because it matters more than the headline. This is observational. Grooming was measured by an interviewer’s judgement, and it could easily be standing in for occupation, income, conscientiousness, workplace norms, employer expectations or social skill. It does not show that buying better clothes raises anyone’s pay by twenty per cent.

What it does show is that the thing being measured as “attractiveness” is not a fixed inherited trait. A large share of it is what people do in the morning.

Now the case against

Satoshi Kanazawa and Mary Still, “Is there really a beauty premium or an ugliness penalty on earnings?”, Journal of Business and Psychology 33(2), 2018, is the strongest challenge in the literature.

They argued that earlier work pooled “very unattractive” with “unattractive”, hiding a non-monotonic relationship, and that it lacked controls for health, intelligence and personality. Then they produced this.

The very unattractive group out-earned everyone except the very attractive. In their regressions, the already-weak premium disappeared entirely once they controlled for health, intelligence, Big Five personality, education and family background.

It is a serious paper and it has two real weaknesses. The very unattractive cell was small, and the paper itself reports unusually poor agreement between raters about who belonged in it. More fundamentally, controlling for personality may remove the causal pathway rather than a confounder: if being treated as attractive from childhood shapes your confidence, then holding confidence constant is not isolating the effect of appearance. It is deleting it.

The best current anchor is meta-analytic: roughly 4.3 per cent higher earnings per one standard deviation of beauty, across 1,159 estimates from 67 studies. Small, positive, real — and not proof of a direct causal mechanism.

Where it flips: the women who did better with no photo

The premium is not universal, and in some hiring contexts it reverses.

Bradley Ruffle and Ze’ev Shtudiner, “Are good-looking people more employable?”, Management Science 61(8), 2015, sent matched fake applications with and without photographs. Attractive men received roughly twice the callback rate of plain-looking men.

Attractive women received no premium at all. The highest callback rate went to women who submitted no photograph — 22 per cent above women with a plain photo, and 30 per cent above women with an attractive one.

Dan-Olof Rooth’s 2009 study in the Journal of Human Resources manipulated photos to appear obese and found callbacks fell 6 percentage points for men and 8 for women. Experimental hiring research has also documented a “beauty is beastly” pattern in male-typed roles, where attractive women are rated as worse fits.

So there is no rule that says looking better always helps. There is a set of context-dependent findings, and at least one where the safest move for a woman applying for a job is to send no picture at all.

Does makeup make you look more competent?

Yes, and it costs you something.

Nancy Etcoff and colleagues, PLoS ONE 6(10), 2011, photographed 25 women barefaced and in natural, professional and glamorous makeup. Then 149 participants viewed the faces for 250 milliseconds and 119 viewed them without a time limit.

At a quarter of a second, cosmetics raised ratings of attractiveness, competence, likability and trustworthiness, all significant at p < .0001.

Given unlimited time, attractiveness, competence and likability stayed up. The trustworthiness effect vanished entirely.

And it split by style. Under longer viewing, natural makeup increased trustworthiness, professional makeup left it unchanged, and glamorous makeup reduced it while still raising attractiveness and competence ratings.

Which is a genuinely useful finding: the look that maximises perceived competence is not the look that maximises perceived trust, and if people have time to look properly, the glamorous end costs you.

Two honest limitations. The study used only female North American faces with professionally applied, digitally adjusted cosmetics and modest sample sizes. And it was funded by P&G Beauty & Grooming, though the paper states the funder did not control the design or the analysis. A later review also cautions that the within-person effect of cosmetics on attractiveness ratings can be small — one estimate puts it at about two per cent of the variance.

So what does this actually mean

Five things, and they fit in a paragraph each.

The premium is real and it is small. Four to five per cent, not twenty. The penalty for looking plain is bigger than the reward for looking polished, which means most of the effect is a floor rather than a ceiling.

The mechanism is not mostly your face. Forty per cent of the experimental premium was employers guessing wrong from a photograph, and twenty per cent was the confidence of the person being looked at.

The part that moves is grooming. That is the finding with the most practical content in it, and for women in one major study it was the whole effect.

More is not better. The glamorous end of the range buys competence ratings and spends trust, and in some hiring contexts a photograph makes a woman’s odds worse rather than better.

And none of this licenses a shopping trip. The research measures presentation, not products. Nobody has shown that any cream, serum or fragrance changes an earnings trajectory, and anyone telling you otherwise is selling something.

The unglamorous conclusion

If you want a shopping list out of this, the research points somewhere boring: the everyday maintenance layer, done consistently. Not the expensive end. Below are four things we inspected on Amazon this month while building our deodorant and shampoo ingredient guides — chosen because they have the strongest buyer evidence in their categories, not because any study says they will do anything for your career. None of them will.

• Secret Clinical Strength antiperspirant — the best-reviewed product in its category, read across 27,245 reviews averaging 4.6 as of 24 August 2026
• Olaplex N°.4 Bond Maintenance Shampoo — the largest evidence base in haircare, read across 83,622 reviews averaging 4.6 as of 24 August 2026
• Vanicream fragrance-free shampoo — the cheapest thing on either of our ingredient pages, and the one with no fragrance or dye in it
• Each & Every fragrance-free deodorant — fragrance-free and baking-soda-free, both stated on the listing rather than implied

Those four links are affiliate links and sit outside the inline-versus-button test we run on our ranking pages. We would rather tell you that than have you wonder why an article about labour economics has product links in it at all.

Related reading: how we rank products, the deodorant ingredients worth avoiding, and the shampoo ingredients with real regulatory records behind them.