Evaluating Value-at-Risk in BIST Using Copula Approach

Abstract

Modelling dependence structure between variables is commonly investigated in literature. A large variety of methods have been improved in recent times. One of the methods is the copula which models accurately dependency regardless of marginal distributions. In this paper Value-at-Risk (VaR) is computed using the copulas. It is assumed that the dependency does not vary through time since small time interval is used. The study composes of two steps. In the first step the best fitted copula is determined by ML (Maximum likelihood). In the second step equal-weighted portfolio analysis is performed by joint distribution function obtained from the copula and maximum possible losses of the portfolio are evaluated. The main challenge in portfolio analysis is that joint distribution function for stocks cannot be correctly constructed by considering dependence structure among them. We obtain joint distribution function for stocks using the copula approach that has been commonly used in recent times. After the best fitted copula is determined using the criterions such as AIC (Akaike information criterion) and SBC (Schwarz