Yes, economics has a problem with women. In the news recently we’ve heard about the study of the Economics Job Market Rumors (EJMR) on-line forum. Student researcher Alice H. Wu found that posts about women were far more likely to contain words about their personal and physical issues (including “hot,” “lesbian,” “cute,” and “raped” ) than posts about men, which tended to focus more on academic and professional topics. As a woman who has been in the profession for over three decades, however, this is hardly news.
Dismissive treat of women, and of issues that impact women more than men, comes not only from the sorts of immature cowards who vent anonymously on EJMR, but even from men who probably don’t think of themselves as sexist. And because going along with professional fashion may be necessary for advancement, women economists also sometimes play along with the dominant view.
Consider a few other cases I’ve noticed during my thirty years in the profession:
- A literature has grown up “explaining” the greater achievements men have obtained in labor markets as the result of men and women having “fundamentally” different preferences—in spite of the empirical evidence for this being weak to nonexistent. (More about this below.)
- Articles about how to get Chief Executive Officers (mostly men) to act in the interest of others (shareholders) prescribe incentivizing them with generous bonuses and stock options. Articles about how to make sure that nurses (mostly women) have the interest of others at heart suggest keeping the pay low, reasoning that then the job will only appeal to true altruists. Yet few seem to notice that such stereotyped reasoning might be relevant to discussing the gender wage gap…
- An economist department tenures a man who had, as his strongest publication, a piece in World Development, a journal not among the top 100 in the profession. The next year, the department denies tenure to a woman purportedly on the basis of inadequate research, in spite of her having published in top-flight journals including Econometrica, JPE, AER, JEP, and REStats.
- A book by an economist claims that statistical discrimination against women is fair and allowable, because it treats women fairly as a group even if it disadvantages some individual women. Affirmative action for women, that attempts to even out the advantages historically enjoyed by men as a group, however, is considered by the author to be unfair, because it could disadvantage individual men.
- Two well-known (male) economists discuss the problem of widening income and wealth inequality at a conference, offering progressive taxation and an extension of opportunities for higher education as possible solutions. Afterwards, I suggest to another economist that public support of high quality subsidized childcare and early education should also be considered a possible remedy for inequality. Well-designed policies could reduce inequality by supporting low-income working parents (especially, given current patterns, mothers). They could also make crucial investments—arguably more important, for many, than higher education—in the next generation. His reply is, “Oh, but that panel wasn’t about life course issues.”
Additional sexists comments, discriminatory decisions, biased research, and, of course, a high preponderance of all-male panels at conferences and high-level policy events would be reported by many economists I know—both men and women. Some of us also see gender bias in the values underpinning the discipline, and especially those values that have governed the choices of the particular models and methods.
As economists, we like to think we are very sophisticated when it comes to research. But are still human beings, and so still subject to the cognitive biases that have explored by psychologists and, more recently, by behavioral economists. Thinking simplistically is one bias-creating cognitive shortcut. For example, we may think of groups as being either “the same” or “different” without paying attention to the degree of difference. Another form of bias is “confirmation bias,” or the tendency to take more notice of evidence that confirms our pre-existing beliefs.
A few years ago, I became curious about claims such as “Women are more risk-averse than men”—or “are less competitive than men” or “are more altruistic than men”—that were popping up in the economics scholarly literature. Some articles claim that these differences are “fundamental.” They also often make allusions to stories about evolution, hormones, or genetics as possible causes. Some draw the conclusion that men and women therefore need to be treated in categorically different ways in employment, investment advising, and so on. I decided to do a meta-analysis of the literature, and describe my discoveries in Gender and Risk-Taking: Economics, Evidence, and Why the Answer Matters.
I find that both simplistic thinking and confirmation bias are rife in the literature on gender and risk-aversion. Reexamining the data used in the economics and finance literature—drawn from survey questions, experiments such as lottery games, and actual investment portfolios—I find that the empirical results are actually quite mixed. Some studies find women to be the bigger, on average, risk-takers than men. Very many studies fail to find a statistically significant difference.
But even more importantly, I find that the best (that is, most precise) estimates of the substantive size of the difference between measures of men’s and women’s risk aversion to be about 0.13 of a standard deviation. To put it another way, you can think of the men’s and women’s distributions having about a 95% overlap. Or expressing it in even simpler terms, suppose you were presented with a random pair of people, about whom you know nothing. In that case, there is a 50% probability that your guess about which one will be the bigger risk-taker will be correct. My meta-analysis implies that if you were presented with a random man and women, and predict that the man will be the bigger risk-taker, your chance of being correct rises all the way to…55%.
While this is not a finding of “sameness,” neither is it by any stretch of the imagination a finding of sort of 0% overlap and 100% predictive, categorical “difference” that is often extrapolated from the research studies. I suggest that, as a corrective, we researchers include measures of the size of a difference, as well as of similarity and overlap, in our studies, and remember to talk about them when addressing our colleagues and the media. We should also entertain alternative explanations. For example, while substantively larger degrees of “differences” are often found in studies of preferences regarding competition, the degree of overlap still tends to run about 60%. And in a world where people tend to judge competitive behavior by women more harshly than the same behavior in men, acting less competitively may be largely a learned and defensive (as opposed to innate) strategy.
What about the second type of bias I mentioned? Sadly for the profession, I find ample evidence of risk-aversion researchers slanting the results in ways that confirm their own (stereotypical) beliefs. For example, one article claims to find a “robust” “victory for gender difference” in spite of the fact that the authors only find a statistically significant difference in 5 out of the 12 measures examined. Four of the 5 were significant only at a 10% level. Another study merged the results from studies in which women on average took more risks than men with studies that leaned the other direction. The published article, as a result, failed to acknowledge the existence of most of this non-confirming evidence. Across the board, findings of “difference” are highlighted, much discussed, and often featured in the titles of articles. Meanwhile—driven by our profession’s fascination with p-values, as well as its gender bias—failures to find statistically significant gender differences are relegated to brief mentions and footnotes, at best, or else to the file drawer.
It’s long past time for our profession, which rightly aspires to do rigorous and scientific work, to wake up to its gender biases. Unfortunately, at least one study has suggested that the more people believe themselves to be rational and objective, the more likely we are to be fooled by our own prejudices.