Corporate Bond Loss Given Default: A Structural Model
This article provides a structural model in a continuous-time framework to investigate the Loss Given Default (LGD) for corporate bonds. By intro- ducing debtors to the default decision making process, both the debtors and the rm management can liquidate the rm, which makes it a stochastic game. Under the assumptions that there exists a covenant that allows them to liquidate the rm, and that they can also perfectly monitor the status of the rm, the optimizing bondholders need to solve a optimal stopping problem with a non-linear terminal payment. The best liquidation policy is intuitive: either to force liquidation the rst time rm value falls to the lowest level that guarantees a full recovery of the bond par value, or never exercise the right. The resulting LGD distribution is therefore bimodal, which agrees with empirical observations. In addition, the model predicts that coupon rates and rm growth rates affect LGDs negatively, while the rm volatility affects LGDs positively. An extension of the basic model where the recovery rates are unobserv- able is studied. It shows that creditor composition of a rm does have an important effect on LGDs, and the effect is consistant with the empirical result from Carey and Gordy (2004).
Zhipeng Zhang
Graudate School of Business, Stanford CA 94305, U.S.A.
国际会议
成都
英文
1-28
2007-07-09（万方平台首次上网日期，不代表论文的发表时间）