Debt bailouts and constitutions

Economic Inquiry, July, 2008 by Emanuel Kohlscheen

B. Partisan Considerations and Interest Groups

Another possible explanation for the observed pattern could in principle come from the partisanship of elected governors. National parties may internalize cross-state spillover effects. Therefore, party discipline may have important consequences in the discussion of wide-ranging bailouts. However, it appears unlikely that the partisan alignment was decisive in the case of Brazil. According to Rodden (2003b), the striking feature of the political process in Brazil is the lack of party loyalty. A recent event illustrates this: in September 2003, a recommendation of the leadership of the main opposition party to oppose a tax bill being proposed by the incoming government was largely ignored. In total, 24 senators voted yes and 26 no. Commenting on the outcome, the president of the party explained that "... representatives responded to the appeals of state governors," while a governor of the same party added, "It is natural that there is no unanimity. There are many regional aspects at play." (14) In the case of the 1997 bailout mentioned earlier, the dissenting votes were cast by representatives of opposing parties: one of the party of the federal government (PSDB) and one of the main opposition party at the time (PT).

Still another possibility is that shifting state liabilities to the federal sphere would bring benefits to bondholders. This would create incentives for lobbying activities upon state representatives. Such explanation, however, does not seem to apply to the case highlighted in the preceding sections. First, if a bailout were the result of lobbying activities, the cost-minimizing strategy for lobbyists would be to ensure a minimum winning coalition of just 41 of the 81 senators. This was clearly not the case. Moreover, as the bailout carries considerable cost for some states (direct debt transfers alone exceeded 10% of Brazil's GDP), incentives to expose such lobbying activities would be large. Second, the above analysis suggests that such lobbying activities of compensation payments would have been redundant.

C. Revenue Sharing Avoidance

Another objection might come from the fact that not all federal revenues are shared with states. This creates an incentive for the federal government to tilt taxation toward nonshared taxes (which typically are more distortionary). Indeed, members of the Constitutional Assembly had already thought of this possibility and included an article that establishes that 20% of revenues of any federal tax that was not previewed in the Constitution at the time of enactment would be directed to the revenue sharing fund. This figure lies quite close to the estimated value of [mu] (20.49%). The analysis presented was performed under the assumption that states expected the central government to resort to nonshared taxes in the future at the same proportion as it was doing at the time. A sensitivity analysis reveals that the predicted support for a bailout would still have assured a comfortable winning margin for bailout supporters even if g were expected to be halved. A bailout would only be rejected if g were expected to fall below 7.8%. This was certainly not a realistic scenario. Finally, the formal specification assumed that increased federal debt service expenditures have to be met by an increase in taxation. This assumption was made due to downward rigidities in federal government expenditures, as the bulk of federal tax revenues are earmarked for specific uses. Figure 3 shows that this was in fact the case: contrary to the revenues of subnational governments, federal revenues as a share of GDP did increase markedly following the bailout.