Debt bailouts and constitutions

Economic Inquiry, July, 2008 by Emanuel Kohlscheen

Judging from the statements of senators, there seems to be an understanding that there is little to gain in the chamber from casting a dissenting vote in measures that favor other states of the federation when the outcome is clear. (12) Nevertheless, two senators did express their negative votes during the session that led to the approval of the precedent-setting debt agreement with Sao Paulo state when it reached the voting floor of the Brazilian Federal Senate in November 1997. They represented the State of Parana and the Federal District. The theory presented predicts that these were exactly the two most likely states to oppose a bailout at the time: they are identified by the circles close to the origin in Figure 2. These are the states furthest away from the indifference line, that is, those whose support for a bailout would be most expensive to buy through compensating deals. States with such a locus are the least likely to take part in a pro-bailout coalition as they do not benefit from a bailout in either dimension: neither via debt relief nor via increased income through Constitutionally mandated transfers. The likelihood of two randomly selected senators belonging to the two most bailout adverse states is just 0.46%. These states were also among the few that saw their contribution to federal revenues increase in the period 1997-2001.

The consideration of the revenue sharing arrangement gave us the correct outcome predictions. Moreover, these predictions are fully consistent with the idea that rational politicians individually vote in the best interest of their constituencies (which includes themselves). In the next section, alternative explanations are discussed.

V. ALTERNATIVE EXPLANATIONS AND SENSITIVITY

A. Was Sao Paulo "too Big to Fail"?

Some authors have argued that the State of Sao Paulo may have been too large to fail. Models like Wildasin (1997) indeed provide a theoretical underpinning for such argument. This prediction is not unambiguous, however. The model of Sanguinetti and Tommasi (2004), for instance, implies the contrary. (13) Furthermore, it is often the case that small federation units are politically overrepresented. Seitz (1999), for instance, argues that in the German Federation, two Laender were too small to fail in 1992.

To address the concern that the State of Sao Paulo was too large to fail, consider the alternative hypothesis that it was common knowledge that a share of the debt of Sao Paulo--of which part was owed to its arguably beleaguered state bank Banespa--would have to be assumed by the federal government either way. Table 2 shows the proportion of the 27 states that would have been predicted to oppose the 1997 bailout if we considered that all senators took it for granted that x% of Silo Paulo's debt would have to be assumed by the federal government. By taking the revenue sharing mechanism into consideration, the opposition to a bailout decreases from 21-24 to 6-7 states. Hence, even if the State of Sao Paulo was widely perceived as being too large to fail, we would still not have a plausible explanation for the approval of the general bailout by the Senate.