Measuring spillovers and connectedness in gretl

This article introduces a new gretl package for computing connectedness measures, as proposed by Diebold and Yilmaz (2009) and extended by Diebold and Yilmaz (2012; 2014, hereafter DY). The h-step ahead connectedness indices, as defined by DY, are based on the variance decomposition, derived from the estimation of a Vector Autoregressive (VAR) model. We provide gretl functions for computing static and dynamic connectedness indices. Additionally, we introduce a bootstrap-based technique for detecting statistically significant changes in connectedness, following Greenwood-Nimmo et al. (2024). Finally, we test our procedure by replicating the global stock market returns analysis of Diebold and Yilmaz (2009).