Fundamental legal and policy debates usually revolve not only around the goals that the law should pursue, and the appropriate means to achieve them, but also around the underlying facts. Reality is complex, and people tend to look for evidence, to perceive it and interpret it in ways that confirm their prior attitudes. Rigorous empirical research is therefore critically important. In recent years, a growing number of legal scholars have conducted empirical legal studies, often in collaboration with researchers from other social-science disciplines. The thought-provoking article reviewed here—on discrimination against women sellers by bidders in eBay auctions—belongs to this emerging genre.
Market-based allocations of goods and entitlements are presumably not only more efficient than centralized allocations, but also freer and more egalitarian. In the market, buyers and sellers freely transact on an equal footing, as opposed to being rewarded on the basis of their class, race, or gender. Of course, these ideals do not always materialize: there are market failures; transactions do not always reflect the free will of the parties; and markets are sometimes discriminatory. With regard to inequality, some have argued that what appears to be prejudice-based discrimination is actually “rational”—because race and gender are proxies for contractual performance; because profit-maximizing firms cater to the preferences of their prejudiced customers; or because cooperation works best between employees of similar social background. In theory, irrational discrimination cannot survive in a competitive market, simply because it is a faulty profit-maximizing strategy. Arguably, therefore, the continued prevalence of market discrimination indicates either that it is rational, or that our expectations of market competition are exaggerated (or both). Of course, rationality and efficiency do not imply desirability or even permissibility, but most would agree that these are relevant issues when it comes to understanding market discrimination and considering what to do about it.
To shed light on this question, one would ideally look for an environment where it is impossible to come up with any “rational” explanation for market discrimination. In the context of gender discrimination, this would mean a setting where the same products are sold and bought by men and women; contracting processes are identical; no extended, close relationships are expected between the parties; there are no gender differences in terms of perceived reputation; and so forth. Ideally, one would look at a large number of transactions involving a wide range of products. Designing such an environment experimentally would be a daunting task, and finding it in the real world practically impossible—or so one might think before reading Tamar Kricheli-Katz and Tali Regev’s fascinating study.
Having secured the cooperation of the eBay research lab, Kricheli-Katz and Regev analyzed 1.1 million transactions made through eBay—an extremely popular platform worldwide for e-commerce between private individuals (and commercial entities). The data used involved the sales of the 420 most popular products from each of the main categories of the eBay catalogue, between 2009 and 2012. Of the four sale platforms available on eBay, the study focused specifically on sales through auctions—because once a seller lists an item there, the final price is not affected by the seller’s behavior, but only by the buyers’ bidding, so that possible gender differences in bargaining behavior are irrelevant.
For each transaction, the researchers had complete data about the sales object; its presentation by the seller; the auction’s starting price; the hidden reserve price (if set by the seller); the final price; the number of bids made; the seller’s feedback-based reputation; the seller’s eBay experience; the seller’s and buyer’s gender; and more. Apart from the sellers’ feedback-based reputation—which was higher for women sellers—men and women sellers were pretty similar in most respects (with such a huge sample, almost any difference is highly statistically significant, so it is more important to look at the magnitude of the difference than its statistical significance). More importantly, when comparing the number of bids and final prices paid to male sellers with those paid to women sellers, the researchers controlled for all other variables so as to isolate the effect of gender.
Along with various findings concerning certain gender differences in item descriptions, risk-aversion, and buyers’ willingness to pay, the key findings were that women sellers received fewer bids than men, and were paid about 80% of the price paid to men when selling an identical new product, and 97% when selling an identical used one (most private sellers sell used items). The price gap varied across types of items, but there was no easily discernible regularity in this respect.
To complement the main observational study, the authors conducted two experiments. The first established that people accurately identified sellers’ gender based on their eBay actual profiles in 56% of the cases (even though the seller’s gender is not stated), misidentified sellers’ gender in only 8.5% of the cases, and could not tell what the seller’s gender was in the remaining 35%. In the other experiment, participants were asked how much they would be willing to pay for an $100 Amazon gift card, when it was sold by either “Alison” or “Brad.” Similarly to the actual eBay results regarding the sale of gift cards, when the card was sold by “Brad,” the average sum was $87.42, and when sold by “Alison,” only $83.34.
In the absence of any sensible reason to pay more for products merely because they are sold by men rather than women, the price differences found in the study appear to be neither rational nor efficient (unless one tautologically defines efficiency as maximizing the satisfaction of revealed preferences). The findings do not tell us whether such gender-based, price discrimination is conscious or not, although my hunch is that it is at least partly unconscious.
While these troubling findings are clear, their normative and policy implications are not. Arguably, bidders in eBay auctions pay less for items sold by women because other buyers offer lower prices for them. Is there anything morally wrong with that? And even if paying more for items sold by men than by women violates the moral prohibition on harming women by discriminating against them, taking legal steps to prevent price discrimination by private buyers might violate the prohibition on limiting people’s freedom. Of course, we might wish to eliminate gender-based discrimination even if there is nothing immoral in the behavior of any individual buyer. To complicate matters still further, unlike some types of discrimination, the study shows that women sellers are discriminated against by women buyers as well as men. Finally, even if one were to conclude that the government, or other entities, or eBay itself, should take measures to curb this discrimination, it is not at all clear what could be done in this regard. While the authors do not answer these questions, we should thank them for compelling us to seriously consider them.