We model and estimate the conditional variance (volatility), in order to calculate VaR by means of the heteroskedastic models GARCH(1,1), RiskMetrics, EGARCH(1,1) and GJR-GARCH(1,1), both under Gaussian and t-Student conditional distributions. Existence of structural breaks has been tested by using the Bai Perron (2009) test. Structural breaks in financial series concur to leptokurtic distributions of the returns, and also explain the increase of persistence in the volatility models. In order to evaluate the performance of the various methods to calculate Value-at-Risk, backtesting procedures are applied

--

Time series, GARCH, VaR