Ethnic diversity and price bubbles

The socializing finance blog points to a PNAS paper showing that ethnic diversity drastically reduces the incidence of price bubbles in experimental markets. This is a conclusion that I am inclined to believe on theoretical grounds and the paper itself presents the theoretical arguments very persuasively. However, the experimental evidence leaves me unimpressed.

The biggest problem is that in both the locales (Southeast Asia and North America) in which they carried out the experiments:

In the homogeneous markets, all participants were drawn from the dominant ethnicity in the locale; in the diverse markets, at least one of the participants was an ethnic minority.

This means that the experimental design conflates the presence of ethnic diversity with that of ethnic minorities. This is all the more important because for the experiments, they recruited skilled participants, trained in business or finance. There could therefore be a significant self selection bias here in that ethnic minority members who chose to train in business or finance might have been those with exceptional talent or aptitude.

This fear is further aggravated by the result in Figure 2 showing that the Southeast Asian markets performed far better than the North American markets. In fact, the homogeneous Southeast Asian markets did better than the diverse North American markets! The diverse Southeast Asian market demonstrated near perfect pricing accuracy. This suggests that the ethnic fixed effects (particularly the gap between the dominant North American ethnic group and the minority Southeast Asian ethnic group) are very large. A proper experimental design would have had homogeneous markets made out of minority ethnic members as well so that the ethnic fixed effects could be estimated and removed.

Another reason that I am not persuaded by the experimental evidence that the experimental design prevented participants from seeing each other or communicating directly while trading. As the authors state “So, direct social influence was curtailed, but herding was possible.” With a major channel of diversity induced improvement blocked off by the design itself, one’s prior of the size of the diversity effect is lower than it would otherwise be.

Posted at 5:41 pm IST on Mon, 24 Nov 2014         permanent link

Comments