Project CEMAPRE internal
|Title||Evaluation of the firm-level VaR and ES forecasting performance of parametric volatility time series models|
|Participants||Rui Louro, Nuno Sobreira (Principal Investigator)|
|Summary||The objective of this project is to run a forecasting competition of different parametric volatility|
time series models to estimate Value-at-Risk (VaR) and Expected Shortfall (ES). We also want to
bring new insights about the methods used throughout this exercise.
For these purposes, a number of models from the GARCH class are used with different distribution
functions for the innovations, in particular, Normal, Student-t and Generalized Error Distribution
(GED) and corresponding skewed versions.
We analyze the performance of these different models to forecast 1% and 5% VaR and ES for 1-day,
5-days and 10-days horizons of the PSI20 individual firms and of the PSI20 index itself, although we
intend to extend this dataset. The VaRs and ESs are compared with backtesting procedures based on a
number of statistical tests and loss functions.
An indirect approach is thus applied where volatility forecasts produced by different models are
evaluated and compared according to the results of a specific application of interest. This
methodology avoids the difficulties of the direct approach as regards to the ex post unobservability
of the forecasted variable.
The final results are analyzed in several dimensions.