Saddam promoted incompetents in his army deliberately, believing they would be less likely to sponsor a coup.  There is a similar process that can operate within firms, the Peter Principle:  If firms automatically promote the best performer at level k of the hierarchy to the level k+1, people will be promoted till they find their level of incompetence.  Saddam’s promotion policy can be justified on rational choice grounds and similarly we might ask how firms can counteract the logic underlying the Peter Principle.

The New York Times magazine has a section on interesting ideas of the year.  One of them concerns the Peter Principle.  A group of Italian physicists did a computer simulation with various promotion policies.  Random promotion outperformed a “promote the best” policy.  It increases the chance that someone who is actually good at the job makes it to the next level.  This seems pretty straightforward and eminently amenable to a simple analytical model.  But peer review is even better than random promotion: ask the co-workers who might be good at the higher level job.  If they have big incentives to lie, at worst you can ignore them and get random promotion as the optimal policy.  Or better, share some of the rents from promoting the right person with the reviewers and get some useful information out of them.

These are old ideas from contract theory but we are clearly not doing a good job at getting our insights to the New York Times.  On that note, let me congratulate Dan Ariely and his co-authors who have at least three of the best ideas of 2009.  The experiment involving the drunks playing the ultimatum game was the most fun – won’t give the point away so you can enjoy it yourself!  But it makes me think Jeff and I should do some experiments in our wine club.  I wonder if we can get the NSF to support it so I can finally taste a Petrus.