A0442
Title: Spillover and quantile-spillover indexes: Simulation-based evidences
Authors: Massimiliano Caporin - University of Padova (Italy) [presenting]
Giovanni Bonaccolto - University of Enna Kore (Italy)
Jawad Shahzad - Montpellier Business School (France)
Abstract: By resorting to simulations, it is shown that the spillover indexes, estimated from a vector auto-regressive (VAR) model or from a quantile-VAR, might sensibly differ from their true value. The origin of these distortions is first explained and then introduce a methodology for improving the evaluation of spillover indexes. Furthermore, a methodology is proposed, based on simulations, for computing a confidence interval for the spillover and the quantile-spillover indexes. Finally, an approach is introduced for decomposing the total index in the contributions coming from the model dynamic, from the residuals covariance, and from a remainder interaction term. Empirical analyses of different data sets support the findings and exemplify the use and interpretation of proposals.