Measuring market risk using extreme value theory

Abstract

The adoption of Basel II standards by the Bangko Sentral ng Pilipinas initiates financial institutions to develop value-at-risk (VaR)models to measure market risk. In this paper, two VaR models are considered using the peaks-over-threshold (POT) approach of the extreme value theory (EVT): (1) static EVT model, which is the straightforward application of pot to the bond benchmark rates; and (2) dynamic evt model, which applies pot to the residuals of the fitted AR-GARCH model. The results are compared with traditional VaR methods such as RiskMetrics and AR-GARCH-type models. The relative size, accuracy, and efficiency of the models are assessed using mean relative bias, backtesting, likelihood ratio tests, loss function, mean relative scaled bias, and computation of market risk charge. Findings show that the dynamic EVT model can capture market risk conservatively, accurately, and efficiently. It is also practical to use because it has the potential to lower a bank’s capital requirements. Comparing the two EVT models, the dynamic model is better than static as the former can address some issues in risk measurement and effectively capture market risks.

Classification-JEL: C01, C22, G12