Thursday, June 24, 2010

Publish school rankings, and bad schools improve

Does publishing the performance of schools work to improve them? Schools certainly resist such disclosure, as they are afraid of being exposed or misjudged, given their particular circumstances. The case of the Netherlands is interesting, as a newspaper regularly lists a ranking of secondary schools along with various scores, and has been doing so for eleven years.

Pierre Koning and Karen van der Wiel use this data and notice that schools that score badly systematically improve, once one takes into account the delays in publishing and disseminating the survey results (three years!) and in registrations in schools. This helps also in resolving issues with endogeneity. Interestingly, they find indicators that cannot be manipulated by schools (for example by preventing poor students from getting tested). Finally, one important issue here is the mean reverting bias, as schools that do badly one year may just have been hit by a bad shocks and should be doing next year anyway. Long term effects of the responses to the publication of the rankings are in the order of 10 to 30% of the standard deviation, which is substantial giving that the publisher has no way to sanction bad schools. It is all about naming and shaming.

PS: the actual text of this paper starts on page 11. What a waste!

No comments: