This thesis is about modelling and forecasting stock market risk in two emerging East European economies: Bulgaria and Romania. The potential for stock market growth in both countries and their integration into the European Union have triggered an increase in investments by foreign financial institutions. However, due to high return volatility and low liquidity, investments in those markets are considered high-risk. In order to prevent unexpected losses, which may cause financial difficulties or even insolvency, financial institutions operating there need to develop accurate models for the estimation of market risk exposure.

In the present study, backtesting techniques (Kupiec’s test, 1995, and Christoffersen’s Markov test, 1998) are employed to evaluate the performance of several Value-at-risk (VaR) methodologies in both markets. In particular, parametric approaches based on Normal and Student-t conditional distributions (with time-varying and constant variance), and on the Gene