Resistance to Reform: Status Quo Bias in the Presence of Individual-Specific Uncertainty: Comment

Why do governments so often fail to adopt reforms that economists consider efficiencyenhancing? This is the question addressed in an influential paper by Raquel Fernandez and Dani Rodrik (1991). They argue that one of the reasons is that individual winners and losers of reform can often not be identified beforehand. This individual-specific uncertainty leads to a bias against reform. Before turning to their model, Fernandez and Rodrik illustrate their argument with an example. The example assumes that voters are risk neutral and that reforms need the support of a majority of the electorate. Fernandez and Rodrik first argue that individual-specific (IS) uncertainty does not lead to a bias against reform as long as the electorate only gets to vote once. Their argument is easiest to explain with the help of Figure 1. The ISU set captures all reforms passed with IS uncertainty. The NO-ISU set captures all reforms passed without IS uncertainty. It can be seen that some reforms (those marked by A) will be enacted with IS uncertainty although they would have been rejected without uncertainty. To see how this may happen suppose that a majority of voters will turn out to lose L from reform and a minority will turn out to gain G from reform. If voters know whether they win or lose for certain when they have to decide, reform will be rejected. If voters only know about the distribution of winners and losers in the population, reform will be passed as long as gains G are large relative to losses L. Figure 1 also shows reforms (those marked by C) that will be enacted without IS uncertainty although they would have been rejected with uncertainty. A necessary condition for this to happen is that a majority will turn out to gain G from reform and a minority to lose L. If voters know who wins and loses when they make their decision, reform will be accepted. If voters only know about the distribution of winners and losers, reform will be rejected as long as G is small relative to L. As reforms may be enacted with IS uncertainty but not without IS uncertainty and vice versa, Fernandez and Rodrik argue that IS uncertainty does not lead to a bias in favor of or against reform when the electorate only gets to vote once. To show how IS uncertainty gives rise to a bias against reform, Fernandez and Rodrik introduce a dynamic element into the example by allowing voters to reconsider reform in a second period. They argue that this does not change the NO-ISU set. This is because in the case without IS uncertainty, voters have all the relevant information when they first decide on reform. As there is no news in the second period, voters have no reason to change their decisions. Second period voting will however eliminate all reforms in the ISU set that are not contained in the NO-ISU set (reforms marked by A) according to Fernandez and Rodrik. Their argument is that IS uncertainty will have resolved when voters get to reconsider reform in the second period. All reforms where a majority of workers turn out to lose will therefore be revoked in the second period. Fernandez and Rodrik argue that these are exactly those reforms that would have been rejected in the first period in the case without IS uncertainty. This implies that reforms passed in the first period and sustained in the second period with IS uncertainty (reforms marked by B) are a proper subset of the reforms passed and sustained without uncertainty (the NO-ISU set). This is the sense in which IS * Department of Economics and Business, Universitat Pompeu Fabra, Ramon Trias Fargas 25-27, 08005 Barcelona, Spain. I thank Reza Baqir, Antonio Cabrales, Brad DeLong, Giovanni Peri, David Romer, Ilya Segal, and two referees for their comments. I also acknowledge partial financial support from CREI and Spanish Ministry of Science and Technology grants SEC2001-0792 and SEC200201601.