Field experiments in markets

There is a new NBER working paper from Omar Al-Ubaydli and John A. List:

This is a review of the literature of field experimental studies of markets. The main results covered by the review are as follows: (1) Generally speaking, markets organize the efficient exchange of commodities; (2) There are some behavioral anomalies that impede efficient exchange; (3) Many behavioral anomalies disappear when traders are experienced.

This is the best survey article on these claims that I know.


A related open access paper here (pdf): Do Anomalies Disappear In Repeated Markets?

My understanding was that the winner's curse was never driven from oil lease auctions, but apparently here in repeated auction tests it could be reduced. Perhaps different conditions of uncertainty.

One of the known oddities of experimental economics, like experimental psychology, is that many of the "real world experiments" consist of non-repeated games oft played by undergraduates. Useful, to be sure, but often quite far from the real world as well.

That's certainly true, but the same external validity question arises with regular econometric studies as well. They analyze one-time historical events in history and try to draw conclusions for other times with other players.

One country's market for apples or fish is not so different from another, at least within certain limits. A 20-year old in a one-shot game for a few dollars is very different from several million participants in a market.

"They analyze one-time historical events in history" they don't.

That is to say, some do, but most empirical papers don't.

OK, 2 or 3 or 5 times and/or places.

Very little empirical work is historically or internationally comprehensive. Perhaps justifiably so at times, since it's often reasonably to think that things will be different in different times and places. However, when things are the same across times and places, then you've got a pretty strong finding.

They had a similar paper comparing field experiments in markets to lab markets a few years back. Will have to check this one out too. Thanks TC.

My general observation is that individual irrational biases often fail to manifest in the broader market, not because traders become rational, but because they come up with other superstitions that tend to cancel out. For example people tend to anchor to the price they bought a stock at, and rationalize not selling losses. So stop-loss orders are elevated to near mythical status. On their own stop-losses are irrational, and there's no decent justification for why they'd generate alpha. In fact, they violate the Copernican principle. Why should the future returns of a stock be so heavily dependent on the price that a single trader bought at?

A perfectly rational trader would re-evaluate with all current information, not slavishly follow a preset price limit set in the past. But most successful traders are very disciplined about stop-losses. Completely banishing irrationality is very difficult, especially in a field that requires some degree of subjective judgement. Instead we often see "good-enough" kludges, which aren't exactly rational, but basically work to cancel the original bias.

But commodities amount to what portion of the economy? 5%? 10%?

Depends how you define commodity. Some definitions include just natural resources, others include any goods and services with close substitutes.

Disappointed, paper was too dry. I expected cites of all the literature on every page, but it was more of a 'big picture' 30k foot high survey.

OT--does anybody know of laboratory or field experiments that show there's money illusion? I am only aware of one lab experiment in a college setting involving students and real money, which is subject to manipulation, that showed indeed there's money illusion.

There are cites in the paper, just not as compact and as many as I like. The paper is too dry, too conversational for me...not worth saving on my HD.

Surveys show that people prefer high raise in high inflation over lower raise with even lower inflation. That one is commonly cited as money illusion in the wild.

I was able to reproduce the results of List (2004) with random agents suggesting that the results aren't really evidence in favor of the neoclassical model with optimizing agents ...

Comments for this post are closed