Modeling National Financial Cycles with Strong and Weak Cross-Sectional Dependence

Although financial market fluctuations that move together are often interpreted as evidence of a global financial cycle, driven by macroeconomic and financial shocks, they may also result from spillover effects generated by local financial linkages and interdependencies among countries. This study is among the first to jointly test the impact of global common factors and local spatial spillover effects by integrating both mechanisms into a unified dynamic spatial panel data model. The global common factors are determined not only by means of a latent dynamic factor derived from a large panel of monthly data on financial variables, but also by means of cross-sectional averages of the dependent variable. Local spatial spillover effects are determined by means of spatial lags determined by neighboring countries through different metrics of proximity. Using quarterly data on national financial cycles in credit and equity markets in 25 countries over the period 1997-2019, along with four distance matrices and two control variables, we find that the dependence between countries in national financial cycles stems from a combination of all the proposed model elements. In addition, geographic distance appears to be the best performing representation of the local spillover effects between national credit cycles and economic distance between national equity cycles.