A Study On The Interest Risk Of Commercial Banks And A Modified Model Based On Extreme Value Theory

Since the 1980s, financial market has developed quickly along with economy globalization and financial liberalization. The risks of financial instruments have been increasingly complex and the financial market become more fluctuating than ever before. Caught in the center of the whirlpool, financial instruments—the medium and the main player of financial market—witnessed operational plight stemming from interest rate risk. Credit risk, market risk and operational risk are the most influential ones for commercial banks. Among all the financial risks, interest rate risk is particularly important because all the economic values of assets are faced with it and it could lead to other risks.Interest rate risk management is a complex process which involves four stages, namely, risk identification, risk measurement, risk management policy &implementation, and risk control. Risk measurement techniques are the premise for risk management. This paper centers on the study of risk measure, and tries to improve the previous risk measure methods.The paper introduces such interest rate measurement methods as Sensitive Gap method, Macaulay Duration method, OAS method and VaR method. It points out that currently, commercial banks in China are faced with not only re-pricing risk, basis risk,embedded option risk and yield curve risk, but also other special interest rate risks resulting from the marketization reform of interest rate. Therefore, we need to use advanced management skills to improve interest rate risk management. The paper uses Sensitive Gap method and VaR method to study the overall interest rate risks of commercial banks and draws its results.The study finds out that value-at-risk (VaR) method, which is widely recognized and used by the financial instruments, does not take the tail probabilities of asset return distributions intro consideration. Since financial sequence usually does not suit normal distribution assumption, it does not workable to adopt traditional measurement methods. This paper uses extreme value theory to improve the original model and enhance its e preciseness of estimation. To test the effectiveness of the model, the paper compares two extreme value methods (extreme distribution and POT) with normal distribution method, analyzes the week turnout rate of LIBOR, 1986-2002 and 2003-2007. The test proves that extreme value method is more precise than the original model based on the normal distribution assumption.