A conditional heteroscedastic VaR approach with alternative distributions

Objective: The purpose of this paper is to explored different distributions in conditional Value at Risk (VaR) modeling as an option the Mexican market. Methodology: We estimate a GARCH model under Gaussian, Normal Inverse Skew Generalized t and Stable distribution assumption, then we implement predicting one-day ahead VaR finally examine performance among four models during period high volatility. Results: backtesting result confirms that stable-VaR approach outperforms other VaR’s prediction 99% confidence level. Limitations: Although widely used risk measure not coherent measure, for reason, natural extension our work should be expected shortfall may produce insights. Conclusions: Our findings reveal consider some empirical characteristic financial returns such leptokurtic, volatility clustering asymmetry improve capacity. This finding important search more robust approaches estimates. Recepción: 09/08/2018 Aceptación: 31/10/2019