Finance-growth nexus, Financial development, Economic growth, Semiparametric methods, Time-varying estimates, Long-panel.

C14, E44, N20, O16

This paper studies the finance-growth nexus in historical perspective. We employ a panel data model with interactive fixed effects and time-varying coefficients for a sample of advanced economies since the late 19th century. The model considers flexible specifications of heterogeneity and accounts for global common shocks that have likely shaped the finance-growth nexus over time. We present three main sets of results. First, our empirical analysis shows that the relationship between finance and growth is time-varying. Using our benchmark model, we estimate the time-varying slope coefficient of financial development and show that the finance-growth nexus has secularly evolved, thus challenging the mainstream assumption of a uniform association over time. Second, by accounting for global common shocks and their heterogeneous impact, we challenge the dominant narrative suggesting a consistently positive contemporaneous relationship between financial development and economic growth. Third, differences emerge when we distinguish between Schumpeterian finance (bank credit growth) and a more speculative type of finance (stock market growth). While both exhibit time-varying behaviors, the empirical evidence points to a substantially stronger and positive association between bank credit growth and economic growth, as opposed to stock markets, which tend to display a weaker or even negative relationship. Our results remain robust when we account for a range of alternative specifications and potential sources of variation.