A0151
Title: Semi-parametric financial tail risk forecasting
Authors: Richard Gerlach - University of Sydney (Australia) [presenting]
Chao Wang - The University of Sydney (Australia)
Abstract: The finding of a class of loss functions that are elicitable for the well-known financial tail risk measures Value at Risk and Expected Shortfall (ES), considered jointly, has allowed some recent advances in the field of semi-parametric tail risk modelling and forecasting, as well as in the formal assessment of those ES forecasts. The aim is to present some of these developments and build upon them through the incorporation of realized measures, the addition of measurement equations and through allowing separate dynamics for the ES equation. Evidence is shown that a Bayesian approach to estimation and forecasting can yield favourable results and an application to financial market returns illustrates that the recent developments, especially when realized measures are included in the models, can generate improvements in the accuracy of forecasts of financial tail risk.
