THEORY, EMPIRICAL EVIDENCE
AND IMPLICATORS
Link between Default an. The d Recovery Rates
The article gives a detailed analysis and measurement of the relationships that exists between the rates of recovery as well as the aggregate default. It also seeks to show the empirical relationship that exists between them. The article gives a review of how models of credit risk would implicitly or explicitly work with the given recovery rate variable. The article also focuses on the rate of recovery of corporate bonds that were defaulted in the period between 1982 and 2002. Additionally, the article attempts to give an explanation of the rates of recovery by giving linear models, logarithmic model and regression models based on logistics. The article bases its argument on the rates of aggregate recovery. These are depicted as a function of demand and supply for all the given securities. The article uses the multivariate and univariate models of time series to explain a major part of the bond variance in the rates of recovery, which are aggregated through the various levels of collateral and seniority. The article also covers the resultant effects of the association between recoveries on credit and default models. It also considers the effects of procyclicality (Altman, Brady, Pesty and Sironi, 2003).
The relationship between the rate of recovery and the rate of default in the modeling of credit risks can be classified into three groups. The classes are; the first generation structural form model, The Second generation structural model and the reduced form model. The first generation structural form relies on the original set up that was designed by Merton. This was based on the principles of option pricing. The third classification is the reduced form model that does not place any predetermined set of conditions on the firm value.
There are several metrics used in explaining the aggregate recovery rates of the corporate bonds. First, there is the Bond Recovery Rate, calculated on annual basis for the years between 1982 and 2001. This is calculated with consideration of the prevailing market values of the defaulted debts of the corporate bonds that were traded publicly. A calculation of the Logarithm of BBR was done. In describing the annual variation of recovery, the factors that were considered were; the Basic explanatory variable, which were considered as the default rates, the demand and supply of the distressed securities, univariate models and the multivariate models.
After the description of the annual recovery variation, several robustness checks were conducted. These were carried out in order to accommodate any changes in the results obtained from the modifications that may have been necessary due to the approach used. To begin with, default probabilities were considered. These default probabilities were used to place the models that had been used in perspective since they were based on the default rate that was realized in the high yielding yet speculative BDR. Quarterly data was used in order to increase the frequency of the data that was being used. Additionally, risk free rates were used so as to determine their roles in explaining the rate of recovery. Furthermore, returns gained from defaulted bonds were also considered. Finally, GDP dummies were used. From the results obtained from the multivariate models, it was evident that the variables of GDP did not have any statistical influence.
The implications that Credit VaR models as well as ratios of capital and their procyclicality were also discussed. The VaR models, the link on RR/ PD and procyclicality effects are shown to be strongly linked to the rate of recovery (Altman, Brady, Pesty and Sironi, 2003).
In conclusion, the results obtained from the consideration, were found to have a profound implication on the credit risk models that relied on portfolios and the markets. These were found to be important in the determination of the recovery rates.
Works Cited
Altman E.I, Brady B, Pesty A & Sironi A (2003). The Link between Default and Recovery Rates: Theory, Empirical Evidence and Implications. Retrieved on: 10/27/2012. Retrieved from;
