In a recent article by Özgür Gürek, Bernd Irlenbusch, and Bettina Rockenbach entitled “The Competitive Advantage of Sanctioning Institutions” (published in Science in 2006), the authors study human institution-choice behavior and cooperative behavior in a setting where there is permanent competition between a sanctioning institution and a sanction-free institution.  In the sanctioning institution, subjects may choose (at a cost) to punish or reward others’ behavior (subjects play the public goods game with other subjects in their chosen institution).  In the sanction-free institution, there is no ability to punish or reward others’ behavior.  The subjects play 30 rounds.  A key feature of the study is that subjects can choose their institution at the beginning of each period (subjects also learn outcomes from both institutions at the end of each period).  Thus, this is a setting in which there is “voting with one’s feet,” where subjects can freely move from one institution to another.

Although the article reports a lot of interesting findings, two in particular seem noteworthy.  First, only about one-third (36.9%) choose the sanctioning institution in the first period, suggesting some initial apprehension towards “living” in such a world.  Second, subjects overcome such apprehension, as 92.9% of the subjects choose the sanctioning institution by the last period.  Further, 100% convergence (i.e., all subjects choosing the same institution) is never reached in any of the 30 rounds of play.  I think these results speak to the issue of whether convergence is likely to happen, a question posed by Rob Bloomfield in an earlier post (see here).  Even in this stark setting, where subjects are homogeneous (at least within the experiment design), there are no costs from moving to one institution to another, and subjects can clearly see the benefits that accrue from choosing the sanctioning institution over the sanction-free institution, we do not see complete convergence.  Admittedly, the experiment setting does not map perfectly into the world in which accounting standards convergence is being considered, but the evidence noted here seems to cast at least some doubt on whether such convergence will take place.