Uncertainty about the expected recovery value is a main caveat when pricing credit contingent claims in reduced form models. This article introduces an empirical model of the bond price which allows the simultaneous estimation of both, default intensity parameters and recovery rates. The model addresses some peculiarities of risky sovereign debt. While under distress sovereign bond prices do not necessarily converge to par at maturity, recent experience from coercive restructurings show that recovery is comprised mostly of an equal compensation to all bondholders. A parsimonious model with these features is applied in a case study of the Argentine debt crisis 2000–2002. Theresulting recovery value estimated from Argentine global bonds starts out above 50% and falls to 25% after default. The arrival of an USD 40 billion aid package arranged by the IMF appears to raise the recovery ratio while leaving the implied default parameters mostly unaffected. This provides evidence that bond investors did not believe in this cure.
When economy minister Roberto Lavagna announced the result of the Argentine restructuring offer this March, investors had tendered 76% of all eligible claims into the exchange. The restructuring forced creditors to accept the largest haircut seen so far on the sovereign bond market. Even though the offer was scolded by most players, including the IMF, it neatly matches what investors expected from the Argentine default in 2001. This analysis shows how the recovery expectation implied from Argentine global bond prices evolved during the crisis and finally hit the after-default price range in the mid-20s.
Once an issuer faces distress, the expected recovery value makes up a major portion of the total bond value. Traders are aware that after a deterioration of an issuer’s credit quality, bonds are traded on a “price basis” rather than a “spread basis”. This simply means that individual bond features like coupon and maturity become less important for bond pricing. The intuition behind this is not of an economic, but of a legal nature becausein default a minority of bond holders can accelerate the respective issue, making it immediately repayable at par. The core legal claim against the sovereign consists therefore of the debt’s nominal value. This fact gets also reflected by coercive debt restructurings. Besides Argentina, Ecuador and Pakistan recently offered the same bond instruments to all holders of their original, defaulted securities so that all investors received roughly the same compensation.
In terms of a yield curve analysis, this supplements the reasoning why yield curves are steeply inverted when default is close. This is especially the case if there is a bullet amortization due in the near future, making default likely at that point in time and causing the bond price not to converge to par. Conceptually, this behavior is not well described by today’s most popular bond pricing framework which defines recovery as a fraction of market value (“recovery of market value”, RMV) and presumes bond prices to converge to par (see Duffie and Singleton [1999], p. 691). Therefore, this paper reinvigorates he “recovery of face value” (RFV) concept which furthermore allows disentangling recovery and default probability parameters. This approach fits Argentine global bond data very well because Argentina’s solvency was under dispute for quite a while. Its USD 128 billion public debt, which squeezed the fiscal budget during a prolonged recession, consisted by 73% of bonds, mostly foreign denominated. When down payment is not a viable option (especially under an unsustainable currency regime) and debt swaps only succeed in rolling over due debt at increased yields, default and a subsequent “hard” restructuring are a country’s final choice. The estimates of the expected recovery ratio yielded in this paper illustrates this evolution.
The estimates gained are useful for two purposes. First, expected recovery rates are necessary ingredients when pricing other instruments, such as derivatives contingent on the recovery value. For instance for credit default swaps, the presumable irrelevance of the recovery assumption (see Duffie [1999]) does not longer hold when bonds trade far below par. Second, the relation between the parameter estimates and fundamental factors help to establish a better picture of financial crises and appropriate policies to contain those. The results show for instance that the USD 40 billion rescue package put together by the IMF could not reduce the likelihood of insolvency, but lead to higher recovery rates. This means that the partial bailout made investors believe that they could pocket more if the government finally decides to restructure.
This paper draws on the standard literature on reduced form models which assume an exogenously specified bankruptcy process (see Heath et al.[1992], Jarrow and Turnbull [1995], Lando [1998], Duffie and Singleton [1999], among others.) Within this approach, equilibrium models follow the idea of term structure models for the risk free rate and obey the no-arbitrage rule (See Dai and Singleton [2000], and Duffee [2002]). While tractable under the RMV scheme, the closed form solution for the RFV scheme applied in this paper is much more complex. To provide a sufficient fit, affine models require at least a two-factor stochastic process for the default rate, making the model even more burdensome.
Download
Default and Recovery Rates of Sovereign Bonds: A Case Study of the Argentine Crisis
