Assist or desist? Conditional bailouts and fiscal discipline in local governments

Whenever a central government faces a sub-unit in financial distress, the unpleasant question arises of whether to assist the unit or not. On the one hand, resisting to bail out the unit may lead to default or bankruptcy, which could be very costly both economically and politically. On the other hand, bailouts may increase problems of soft budget constraints: noting that the central government steps in in times of trouble, sub-units may come to expect that bailouts will be available when needed. Restricting attention to the relation between central and local governments, bailouts have been found to be followed by deteriorating fiscal discipline in case studies of several fiscally decentralized countries (e.g. Bordignon and Turati, 2009; Pettersson-Lidbom, 2010; Fink and Stratmann, 2011). Little is however known about whether such tendencies can be dampened if assistance is made conditional upon the local governments' own fiscal efforts. We examine a case in which the Swedish central government provided conditional grants to 36 financially troubled municipalities: to receive the grant, municipalities first had to cut costs and balance their budgets. There were also 23 municipalities that applied for the grant, but whose applications were rejected. We examine the program effect on fiscal discipline over the decade after the launch of the program. Two major identification problems are that the program effect may vary between municipalities, and that selection into the program was not random. We address these two problems by employing the synthetic control method, a case study method developed in Abadie and Gardeazabal (2003) and Abadie et al. (2010), and also combine this method with fixed effects estimations. Our results reveal substantial heterogeneity in the municipal responses to the program and also highlight the importance of the choice of comparison group. For the average effect on admitted municipalities, the main bulk of the evidence point to no increases in the level of per capita costs in the longer run. Fixed effects estimations show consistently better net revenues on average. Moreover, admitted municipalities do better on average than rejected municipalities, who arguably received a signal of hard budget constraints. We conclude that conditional bailouts need not erode fiscal discipline.