Value-at-risk and expected shortfall: a dual long memory framework

In this article, we use the dual long memory properties to assess the value-at-risk and expected shortfall for the Argentinean stock market under both short and long daily trading positions. We attempt to show whether considering for long memory properties in both the returns and volatility, volatility asymmetry and fat-tails could provide more accurate value-at-risk's and expected shortfall's estimations. For this purpose, the joint ARFIMA-FIGARCH, ARFIMA-HYGARCH and ARFIMA-FIAPARCH models are applied to the MERVAL stock price index under normal, student-t and skewed student-t distributed innovations. We show that the skewed student-t-ARFIMA-FIAPARCH model performs better in predicting the in-sample and out-of-sample one-step ahead value-at-risk and expected shortfall for both short and long trading positions.