National research assessment exercises: the effects of changing the rules of the game during the game

National research evaluation exercises provide a comparative measure of research performance of the nation's institutions, and as such represent a tool for stimulating research productivity, particularly if the results are used to inform selective funding by government. While a school of thought welcomes frequent changes in evaluation criteria in order to prevent the subjects evaluated from adopting opportunistic behaviors, it is evident that the "rules of the game'' should above all be functional towards policy objectives, and therefore be known with adequate forewarning prior to the evaluation period. Otherwise, the risk is that policy-makers will find themselves faced by a dilemma: should they reward universities that responded best to the criteria in effect at the outset of the observation period or those that result as best according to rules that emerged during or after the observation period? This study verifies if and to what extent some universities are penalized instead of rewarded for good behavior, in pursuit of the objectives of the "known'' rules of the game, by comparing the research performances of Italian universities for the period of the nation's next evaluation exercise (2004-2008): first as measured according to criteria available at the outset of the period and next according to those announced at the end of the period.