Model-free connectedness measures

Abstract

This paper introduces a model-free connectedness approach by constraining the model-based connectedness approach of Diebold and Yilmaz (2012). By doing so, we demonstrate how the adoption of this benchmark model is relevant for statistical, theoretical, and practical purposes as well as illuminate analytical relations that have previously been observed only empirically. Moreover, we investigate eight publicly available connectedness datasets and illustrate the differences between model-free and model-based connectedness measures. Finally, robust unconditional connectedness measures are provided.