Tuesday, January 9, 2018

How the classic endowment effect paper was published

One of the most powerful findings of behavioral economics is “loss aversion,” the psychological tendency to feel losses more acutely than gains. As Adam Smith (1759) put it, “Pain … is, in almost all cases, a more pungent sensation than the opposite and correspondent pleasure” (1981, III, ii, 176–77). Although Daniel Kahneman and Amos Tversky (1979) and I (1980) had earlier written about this phenomenon, its empirical validity was still very much in question when Kahneman, Jack Knetsch, and I submitted an experimental paper on the subject to the [Journal of Political Economy], later published in 1990.

In the experiment we randomly assigned half the subjects to receive some object (often a coffee mug), with the other half getting nothing. We then conducted a market for the mugs in which both buyers and sellers stated their reservation prices. Since transaction costs were negligible and the objects were randomly assigned, the Coase theorem predicts that roughly half the mugs will change hands so that subjects who value mugs the most end up owning them. Our hypothesis was that fewer than half the mugs would trade because owners would regard a trade as a loss. This hypothesis was strongly supported. In a typical experiment, the expected number of trades was 11 but the empirical average was only 3.4. As predicted by loss aversion, median reservation prices for selling the mug were roughly twice the median prices for buying the mug.

The editor handling this paper was George Stigler. He sent us back a rejection letter based on a highly critical referee report from someone Stigler described as a “heavyweight in the field.” The referee said that income effects could explain our results since those who received the mugs had received a windfall relative to those who did not. After taking a few days to calm myself down (a good self-control strategy) I wrote back on behalf of my coauthors (who were both away traveling) explaining why the referee’s comments could not be taken seriously, either theoretically or empirically. First, the marginal propensity to spend windfalls on university insignia coffee mugs must be minuscule. Second, one experiment explicitly tested and rejected this explanation. Stigler wrote back in his usual witty style saying that JPE stands for Journal of Perspicacity and Equity, and he offered to send both my letter and the original referee report to another referee to adjudicate. That referee said that if forced to choose between our view and that of the original referee, he would side with us, which is how the paper came to be accepted.*

*The self-control paper [Thaler and Shefrin (1981)] also involved quite a bit of back and forth with the editor Sam Peltzman, who somewhat reluctantly agreed to accept it rather than continue to exchange letters. Both papers were published as the last paper in the issue, which I took as a signal that they were considered the paper the editors were most ashamed to publish. It is gratifying that both papers were ranked highly on the list of most-cited papers compiled by the editors for this issue. Perhaps people read the JPE from back to front.
--Richard Thaler, Journal of Political Economy, on the slings and arrows of publishing seminal papers