ECB Paper Introduces Structural Quantile VAR Model for Stress Testing

The paper develops the general quantile structural vector autoregressive framework. It provides the links with standard OLS structural VAR, derives the asymptotic distributions, and shows how to do forecasting with quantile structural VAR. Finally, it estimates the quantile VAR model for the euro area and performs a stress testing exercise and the counterfactual analysis before Lehman’s bankruptcy, before presenting the conclusions. Unlike standard VAR, which models only the average interaction of the endogenous variables, quantile VAR models their interaction at any quantile. The authors show how to estimate and forecast multivariate quantiles within a recursive structural system. The model is estimated using real and financial variables. The dynamic properties of the system change across quantiles. This is relevant for stress testing exercises, whose goal is to forecast the tail behavior of the economy when hit by large financial and real shocks.

Unlike OLS VAR, quantile VAR models each quantile of the distribution. This provides the natural modeling environment to design particular stress scenarios and test the impact that they have on the economy. A stress scenario is just a sequence of tail quantile shocks, which can be chosen arbitrarily by the policy maker or calibrated to mimic previous crisis episodes. The authors find the presence of strong asymmetries in the transmission of financial shocks in the euro area, with negative financial shocks being particularly harmful when coupled with negative real shocks. By modeling the average interaction between the random variables, OLS VAR models miss most of these detrimental interactions.

Related Link: Working Paper (PDF)

Keywords: Europe, EU, Banking, Insurance, Stress Testing, Modeling, Quantile VAR Model, ECB