Resumen

SAAVEDRA ESPINOSA, Alberto. Estimation of Market Risk Measures in Mexican Financial Time Series. Rev. mex. econ. finanz [online]. 2017, vol.12, n.4, pp.365-388. ISSN 2448-6795.  https://doi.org/10.21919/remef.v12i4.234.

The objectives of this work are to investigate whether: i) a GARCH model with Generalized Pareto Distribution (GPD) innovations, complemented with an EWMA volatility forecast in order to consider practical problems that might arise in GARCH applications that comprise long periods of time, appropriately estimate risk measures (VaR and Expected Shortfall) for Mexican financial series, at high confidence levels; ii) the estimates yielded by such model are better than those given by a GARCH with Gaussian or Student-t innovations. Our quality assessment and comparison between models consist of backtests of the risk measures estimates yielded by each method used in this paper. Our results show that: i) the methodology used in this paper appropriately estimates our two risk measures; ii) the GARCH-GPD model yields better results than the GARCH-Gaussian and GARCH-t-Student models. Our results are limited to one-day risk measures estimates. As far as we know, our results on the Expected Shortfall are the first of its kind for Mexican series. We conclude that the study achieved its objectives and there are important areas of opportunity for further studies.

Palabras llave : Risk Analysis; Value at Risk; Volatility Forecasting; GARCH; Extreme Value Theory; Market Risk; Expected Shortfall.