Title: Extreme returns in the Russian stock market: Unexpected tails in risk measures
Authors: Oleg Lebedev - Innopolis University (Russia) [presenting]
Andrei Ankudinov - Innopolis University (Russia)
Abstract: The purpose is to investigate extreme fluctuations in the Russian financial market using extreme value theory methods. For that, we: (i) determine the probability distribution that most accurately describes the behavior of extreme returns in the Russian stock market and further present the analysis of the dynamics of its parameters over time; (ii) test the reliability of dynamic risk-management models based on the fitted distributions. A highly volatile Russian market provides a natural environment for the analysis of distributional properties of extreme changes in financial returns. At the same time, the high degree of heavy-tailedness and heterogeneity of observations dealt with require applications of robust statistical methods. The results show that the extreme returns parameters lie between the less heavy-tailed GEV distribution and the more heavy-tailed GLO distribution. However, when we move deeper into the tails of distributions, it is the GLO distribution that appears to be more suitable since other distributions fail to accurately estimate the probability of extreme returns. Accordingly, the GLO distribution appears to be more adequate for integration into risk-management models for the Russian stock market. Applications of less heavy-tailed distributions would result in systematic underestimation of the largest market slumps.