We focus on the debate of the "disappearing" empirical link between finance and growth and address the issue of the features of financial development that ensure stable and long-lasting economic growth. We build the Financial Resilience Index, a composite measure intended to capture the characteristics of openness, financial structure and stability of financial systems. We then analyse the impact of financial resilience on economic growth for the EU28 over 2005-2017, employing the GMM technique estimator to address the endogeneity concerns of the finance-growth nexus. We compare our result with those obtained using standard financial development indicators. In a second step, we employ the Phillips and Sul (2007) methodology to identify convergence clubs for financial resilience across the EU and analyse if belonging to the high or low resilience clubs impacts on the relationship between financial development and economic growth.
We shall contribute to the literature on economic growth and European integration analysing the asymmetries in financial systems. Our research is the first attempt to benchmark financial systems and assess their converging patterns with respect to financial resilience.
This research is innovative with respect to the current literature in several ways.
- A) We shall build a new index to measure financial resilience, namely the FRINDEX; this will allow us to benchmark financial systems with respect to their resilience;
- B) We shall use the FRINDEX in growth regressions to estimate the finance-growth nexus from a broader perspective.
- C) We shall look at the convergence patterns od financial systems within the EU and identify clusters of high/low financial resilience. This will then allow us to analyse the implications of belonging to high/low financial resilience clubs for the finance-growth nexus.
As to point A), the approach followed for the construction of the Financial Resilience Index is analogous to the one used to build the IMF Financial Development Index (Sahay et al., 2015 and Svirydzenka, 2016): we aggregate the variables of interest into five sub-indices (internationalisation, markets vs institutions, stability, the ratio of long-to-total liabilities, equity-to-debt ratio) using the weights obtained from the Principal Component Analysis. This procedure is performed again and the sub-indices are aggregated into the final index.
As to point B), The Dynamic System GMM estimator (Arellano and Bover, 1995; Blundell and Bond,1998) allows overcoming problems of heteroskedasticity, serial correlation and endogeneity of the explanatory variables. It is useful when the number of individuals is greater than the number of time periods; the dependent variable is dynamic and depends on its own past values; the independent variables are not strictly exogenous and are correlated with past and current realizations of the error; there are arbitrarily distributed fixed individual effects; there are heteroskedasticity and autocorrelation within panels.
As to point C), we shall employ the PS technique. There are several appealing features of this methodology. First, it lies on the concept of sigma convergence, i.e., the reduction of disparities through time. Second, it allows considering heterogeneity both across countries and over time, i.e. with the possibility of transitional divergence. Third, the identification of clubs is endogenous, avoiding a-priori grouping of countries. Fourth, it has no requirements on the stationarity of the series. The PS methodology is increasingly employed to assess panel (subgroups) convergence in real per-capita incomes among groups of countries/regions (Bartkowska and Riedl, 2012; Lyncker and Thoennessn, 2017 and Monfort et al., 2013; Borsi and Metiu, 2015, Cutrini, 2019) and in financial development (Apergis et al., 2012), stock market indices (Apergis et al., 2014), retail banking (Rughoo and Sarantis, 2014) and asset returns (Caporale et al. 2015).