Causality clubs: grouping countries with cluster causality detection

Abstract

This paper explores the heterogeneity of causal structures of economic growth among countries by proposing a two-step procedure. First, we apply a causal discovery technique to uncover the underlying causal structure for each country. Second, we employ hierarchical clustering over these estimates to identify groups of similar countries in terms of their causal relationships. We obtain five 'causality clubs', each one associated with a different structure of causal determinations of the growth process. We find that the usual associations between income or geographical location and the nature of economic growth processes may not always hold true.