The Reseach Of Educing Bank's Economic Capital Based On Extreme Value Theory

A sticking point of credit risk management is to educe bank's economic capitalwhich defences the unexpected loss by measuring risk. Now the mainly credit riskmodels are based on the whole credit loss distribution. But because of the "leaningsummit and fat tail" of this distribution, it is difficult to fit the whole credit lossdistribution and calculate economic capital. This paper, based on Extreme ValueTheory(EVT), pays attention to the tails of credit loss distribution, combines the tailscurve fitting with Monte Carlo Simulation, and aims at finding out a measure ofeducing economic capital accurately.The first step of educing economic capital based on EVT is to gain credit lossdistribution by Monte Carlo Simulation. Because of the samping error, this distribu-tion that is made by simulation isn't accurate. But by simulation this area can bereserved very well, and the simulation date will be the basic of fitting credit lossdistribution and educing economic capital.The second step is by using EVT to fit the tail of credit loss distribution. The hardpoint is calculating the model parameter and choosing the proper threshold. Thispaper uses MATLAB7.0 software to fit the credit loss distribution and calculate theparameter, and takes the CHI-SQUARE value as a standard to judge how well theseries fits the reference distribution. So the Pareto distribution function of the nettingsurpassing the threshold is getting.In the end, the bank's confidence absorbing credit loss is firstly gained, which iscorresponded with the expectable bank's credit level. Based on the added probabilityof credit loss at the "best" threshold in the simulation data, the paper calculates creditportfolio VaR at the confidence, then gains the economic capital which bank shouldretain. Comparing this "CaR" with the "CaR′ " by simulation, the paper analyses theinfluence of the the samping error and elicits the rationality of educing economiccapital based on EVT.