Research On Default Correlation Based On Shared Frailty

Since the21st century, default correlation phenomenon, frequently occurring in each country, has seriously affected the stability of the financial system and market economy, which is causing extensive concern in terms of the practice and the theory. At present, lots of scholars not only discuss the nature, types and reasons of default correlation, but also build a series of mathematical or econometric models to measure the default correlation, so as to provide theoretical guidance for economic entities to prevent credit risk. However, studies at present have ignored the fact that the actual default correlation is not only influenced by the observable factors (such as GDP, CPI, interest rates and other macroeconomic variables as well as the company’s specific characteristics), but is usually associated with many unobservable variables. Especially in the industrial sectors with high uncertainty, the determinant of default correlation is often the variables impossible to be quantified or estimated.Concerning on the above problem, the paper introduced an effective frailty, which stands for the set of all unobserved variables, to estimate the default correlation between companies more accurately. The steps of the research are listed as follows:to begin with, summarizing the main observed risk factors affecting default intensity based on the existing literatures; then, introducing a frailty to the traditional default intensity model, in other word, constructing model of default correlation based on shared frailty; finally, confirming the frailty effect in real economic cycle, which severely influences the intensities of all companies through the empirical study.The main contribution of this paper is:constructing a type model of default correlation based on shared frailty after identifying the observed risk factors influencing default intensity, and demonstrating it is helpful to improve the prediction accuracy of default correlation through empirical research.