There are different methods to calculate default probability of a portfolio. One can name Structural method of Merton, Reduced (non-structural) form, Scoring method and Hybrid model. Although, finding default probability is really important and can help lenders to protect themselves, calculating default correlation between exposures is as critical. Calculation of default correlation enables financial institutions to take into account the effect of diversification and also, gives them a better estimate of overall defaults. The goal of this research is to find the correlation between default risks of publicly traded Canadian companies in an overall loan portfolio. For this purpose, the CreditMetrics method that was described by J.P. Morgan is utilized.
The dominant role of credit risk in total risk of banks has made financial institutions, researchers and regulators to pay special attention to credit risk. Credit risk is a risk that a debtor may not be able or willing to repay his debt. Usually great percent of banks total risk is explained by credit risk.
As stability of financial and economical system of each country is very dependent to stability of banks in that country, regulators pay special attention to banks. They make banks to calculate their risk and reserve a capital for reverse events. They set limits and conditions but allow banks to choose their own models to calculate their risks. So banks look for an accurate model for calculating their risks to immune them from crises, be accepted by regulators and also allow them to put less capital in reserve. The model does not only precisely estimates the default probability of exposures but also default correlation between exposures. Since banks do not want to simply sum up all the risks without encountering the diversification affects, that allows them to decrease their risks.
The most popular approaches in the financial literature for estimating default probability and default correlation are Structural and Reduced form models (intensity model).
Reduced form models provide statistical representation of the economic system. These models assume that a firm default time is unpredictable and driven by a default intensity, which is a function of latent state variable. In this approach, default time is the first jump of an exogenously given jump process [9]. According to reduced-form model, multiple defaults are independent, conditional on the sample paths of the default intensities. Therefore, finding the default correlation is equal to finding the correlation between default intensities. Although some researchers such as Fan Yu [22] have worked on this model to show that default correlation can be sensitive to default intensities, some authors such as Hull and White (2001) and Schonbucher and Schubert (2001) argue that the ability of this approach to estimate default correlation is limited.
Download
Estimation of default correlation in a loan portfolio
