Multiple-Days-Ahead Value-At-Risk and Expected Shortfall Forecasting for Stock Indices, Commodities and Exchange Rates: Inter-Day Versus Intra-Day Data

Abstract

Multiple-days-ahead value-at-risk and expected shortfall forecasting for stock indices, commodities and exchange rates: Inter-day versus Intra-day data

Time‐varying dynamics of expected shortfall in commodity futures markets

“…One of the best‐known methods of forecasting ES is to assume that the data follow the skewed distribution of Hansen (1994) which, in contrast to a normal distribution, can model the empirically relevant features of asymmetry and heavy tails. In a commodity context, Degiannakis and Potamia (2017) have used it to forecast VaR and ES of COMEX gold, silver, and copper futures and pointed out its theoretical merits in comparison to simple standard approaches.…”

Volatility specifications versus probability distributions in VaR forecasting

“…The preference for asymmetric distributions and APARCH/FGARCH volatility disappears in 10-day VaR forecasting obtained by filtered historical simulation, with all models showing an overestimation of risk that is less obvious for GARCH volatility specifications. Following a different simulation strategy, other authors have obtained an underestimation of risk (see Degiannakis and Potamia, 2017). Since the Basel Committee on Banking Supervision (2009) requires 10-day VaR predictions, a further analysis of the different performance of alternative VaR models and simulation strategies at the different horizons remains as a central issue for further research.…”

“…In spite of the Basel Committee on Banking Regulation (2009) switch to require 10-day VaR estimation, there is not much work yet exploring the performance of alternative VaR models. Degiannakis and Potamia (2017) and Degiannakis et al (2013) analyze a number of issues regarding 10-day VaR and expected shortfall forecasting. Degiannakis and Potamia (2017) conclude that the use of intra-day data does not lead to better risk estimates.…”

“…Degiannakis and Potamia (2017) and Degiannakis et al (2013) analyze a number of issues regarding 10-day VaR and expected shortfall forecasting. Degiannakis and Potamia (2017) conclude that the use of intra-day data does not lead to better risk estimates. They also obtain a preference for skewed distributions and a GARCH volatility specification, as well as a better forecasting performance at 97.5% than at 99% significance level.…”

The Financial Economics of White Precious Metals - A Survey

“…Working with daily data January 1993 and 2013, the author fits the data into three different non-linear long memory volatility models (FIGARCH, FIAPARCH, HYGARCH) and finds that for all three white metals, a Fractionally Integrated Asymmetric Power ARCH (FIAPARCH) model is best suited to capture their long memory, asymmetry and fat tails, outperforming the other models in predicting oneday-ahead VaR positions. Degiannakis and Potamia (2016) base their research on the recommendations of the Basel Committee on Banking Supervision and examine whether inter-day or intra-day model provide accurate predictions for reliable Valueat-Risk (VaR) and Expected Shortfall (ES) forecasts. Daily data for silver between the 3 rd of January 2000 and the 5 th of August 2015, indicates that a GARCH-skT model, relying on inter-day data, provides better results than a HAR-RV-skT model, as it satisifes most of the conditions implied in VaR and ES forecasting, but that it overall fails to provide accurate forecasts of the risk measures implied.…”