CMStatistics 2017: Start Registration
View Submission - CFE
A1691
Title: Forecasting systemic risk Authors:  Massimiliano Caporin - University of Padova (Italy)
Laura Garcia-Jorcano - UCM (Spain)
Juan-Angel Jimenez-Martin - Complutense of Madrid (Spain) [presenting]
Abstract: The Conditional Value at Risk (CoVaR), the VaR of the financial system conditional on an institution being in financial distress, has been previously modified by proposing a change in the definition: from the maximum loss of the system conditional on the financial institution being in its VaR, to the financial institution being at most at its VaR. We extend this methodology. The first objective is to evaluate whether the multivariate GARCH specification can be relevant for forecasting CoVaR. Additionally to the DCC model used in GE, we use the BEKK model and the Orthogonal GARCH (OGARCH) model. The second novelty is about the distributional assumption on the error. We use Filtered Historical Simulation (FHS) that has emerged as one of the best tools for calculating risk measures as VaR and consequently a robust alternative as a procedure to forecast CoVaR. The third contribution is to produce 1-day and 10-day ahead CoVaR forecasts. Finally, in order to assess the validity of the systemic risk measure forecast we run, not only the traditional statistical based unconditional and conditional coverage tests, but we also use a variety of loss function applied by banks and regulators to evaluate VaR model performances.