Invited Speakers Participants Programme Programme Committee Organization Sponsors

abstracts

VAR COMPUTATIONS UNDER VARIOUS VAR MODELS AND STRESS TESTING

Suat TEKER (İstanbul Technical University)
Barıs AKÇAY (Oyak Bank)

Financial markets and institutions have operated in fastly changing environment in recent years. There could be over 100 crises sited in the economic literature in last 20 years. This makes five financial crises per year on the average. Therefore, the importance of risk management for financial institutions keep increasing at an increasing rate. Lately, the tendency of banks about risk management is proactive. As a result of this tendency, banks work on identfying all kins of risks likely to affect the banks financial positions, and measure and manage financial risks by applying various parametric VAR models. The crises oftenly occurred in recent years have led financial players to be precausious. Then, the need for questioning stress testing of employed parametric models has come across. Stress testing attempts to identify the weakest points of a portfolio. Stress testing is applied for banks' portfolio by simulating the real crises happened in the near past and/or likely senarios.

This empricial research will examine the effects of stress testings on banks capital adequacy ratios and the amount of computed market risk. The publicly available financial data of the four largest Turkish Banks (Garanti,Iş, Akbank and YapiKredi bank) as of december 2001, will be used. For all banks an hypothetical portfolio of marketable securities, derivative securities and short exchange positions will be constructed. Next, the market risk of the hypothetical portfolio will be computed by using historical simulation, monte-carlo, variance-covariance and standard methods. The results will be stress tested for two Turkish financial crises (November 2000, and February 2001). We believe that this emprical study will be one of the front liner academic studies in the area since there is a few study available about stress testing applications in the literature.

Please click to see the full paper