Delay and Haircuts in Sovereign Debt - University of St Andrews
Delay and Haircuts in Sovereign Debt - University of St Andrews
Delay and Haircuts in Sovereign Debt - University of St Andrews
Create successful ePaper yourself
Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.
Source: Table 14 <strong>and</strong> 15 <strong>in</strong> <strong>St</strong>urzenegger <strong>and</strong> Zettlemeyer (2005); Table A.3 <strong>in</strong><br />
Roub<strong>in</strong>i <strong>and</strong> Setser (2004)<br />
Why is it so di¢ cult for sovereign debtor <strong>and</strong> creditors to restructure<br />
sovereign debts <strong>in</strong> an orderly <strong>and</strong> timely manner?<br />
Will faster <strong>and</strong> more<br />
orderly debt restructur<strong>in</strong>g not only help <strong>in</strong> restor<strong>in</strong>g economic momentum<br />
but also reduc<strong>in</strong>g the probability <strong>of</strong> which a crisis occurs <strong>in</strong> the …rst place?<br />
From a theoretical perspective, a s<strong>in</strong>gle creditor, complete <strong>in</strong>formation<br />
barga<strong>in</strong><strong>in</strong>g model <strong>of</strong> sovereign debt restructur<strong>in</strong>g (for examples, Bulow <strong>and</strong><br />
Rogo¤ (1989) <strong>and</strong> Bhattacharya <strong>and</strong> Detragiache (1994)) cannot account for<br />
delay s<strong>in</strong>ce, <strong>in</strong> this class <strong>of</strong> models, an immediate agreement occurs along<br />
the equilibrium path <strong>of</strong> play. When the size <strong>of</strong> barga<strong>in</strong><strong>in</strong>g surplus is determ<strong>in</strong>istic,<br />
the default<strong>in</strong>g country <strong>and</strong> its creditors know exactly the future<br />
barga<strong>in</strong><strong>in</strong>g surplus; thus, both barga<strong>in</strong><strong>in</strong>g parties can reach an agreement<br />
immediately after the default.<br />
In this paper, our start<strong>in</strong>g po<strong>in</strong>t is the assumption that both the recovery<br />
process <strong>and</strong> the will<strong>in</strong>gness to undertake massive …scal cuts (by runn<strong>in</strong>g a<br />
primary budgetary surplus, for example) are uncerta<strong>in</strong>. In a s<strong>in</strong>gle creditor<br />
model, we show that multi-period costly delay will exist <strong>in</strong> a Perfect Bayesian<br />
equilibrium driven by the prospect <strong>of</strong> uncerta<strong>in</strong> economic recovery <strong>and</strong> the<br />
signall<strong>in</strong>g <strong>of</strong> susta<strong>in</strong>ability concerns.<br />
In our model, the length <strong>of</strong> delay is positively correlated with the size <strong>of</strong><br />
creditor losses or haircuts. Although one-period delay occurs to permit for<br />
economic recovery, multi-period delay is essential for the debtor to signal<br />
susta<strong>in</strong>ability concerns.<br />
Relative to the case with one-period delay, with<br />
two-period delay, the creditor always receives lower payo¤s, which leads<br />
to larger haircuts although the debtor could either ga<strong>in</strong> or lose. When the<br />
debtor ga<strong>in</strong>s from two-period delay, conditional on default, the two scenarios<br />
cannot be <strong>in</strong>terim 1 Pareto ranked. However, when the debtor loses from twoperiod<br />
delay, one-period delay <strong>in</strong>terim Pareto dom<strong>in</strong>ates two-period delay<br />
conditional on default. We use data from Benjam<strong>in</strong> <strong>and</strong> Wright (2009)<br />
to provide evidence on the positive correlation between delay length <strong>and</strong><br />
haircuts.<br />
1 At the <strong>in</strong>terim stage, payo¤s are computed conditional on default but before all uncerta<strong>in</strong>ty<br />
related to the size <strong>of</strong> the barga<strong>in</strong><strong>in</strong>g surplus <strong>and</strong> the susta<strong>in</strong>ability concern is<br />
resolved.<br />
3