Titolo della tesi: Empirical Essays on the Determinants of Economic Growth
This doctoral thesis consists of three disjointed essays, all of which belong to the macro-area of research on economic growth, in particular with empirical applications. Studying the determinants of economic growth is essential not only in times of crisis, in order to target public policies and investments in the most profitable sectors in terms of economic and social added value.
In Chapter 2, we ask which are the most robust determinants of growth in Europe, among 70 potential (economic, demographic, institutional, etc.) explanatory variables. The sample collects data for the years 2002 to 2019, as the 2020 observation is somewhat distorted by the Covid-19 pandemic. In this experiment, we use a Bayesian Model Averaging model, estimated several times by changing both model priors (2) and coefficient priors (5), and compare which variables are selected in the 10 different specifications. Our results support neoclassical growth theories, as the initial level of GDP per capita and savings are robust determinants of growth. Other robust determinants include the share of manufacturing in GDP, demography, public accounts, wage and labor contract regulation, and fixed capital accumulation.
In addition, in Chapter 3, we perform a meta-analysis on the relationship between public debt (government and foreign) and economic growth, coding 422 observations from 32 studies estimating cross-sectional or panel regressions. This relationship has been at the centre of political and academic debate following the 2008-2011 crisis. With the work presented here, we intend to systematically review and analyze the empirical literature that addresses this issue. Concerning the results, the average size of the estimated effect is negative: using the partial correlation coefficient, it averages -0.2. We also find that heterogeneity is substantial and influenced mainly by within-studies variability. The moderators that allow to slightly mitigate it and that influence the estimate of the effect size concern the publication status, the journal ranking, the variables used as proxies, the level of wealth and development of the countries considered, the sample size, the region, and the estimation method. Finally, publication bias arises, both as regards the direction of the estimated effect size, and the statistical significance of the results presented.
To conclude, in Chapter 4 we ask how and how much government debt impacts the innovation rate, which is one of the main stimulus variables for economic growth, according to the literature. Furthermore, we take into account the government debt (calculated as the amount of debt held at the central bank) in this link. We use a panel threshold regression model on a sample of 15 industrialized countries from 2000 to 2019. Our results show a strong non-linearity, in the sense that an increase in debt above a certain threshold has a negative impact on the innovation rate, while below, it has a positive effect. Debt monetization contributes positively to innovation if it is below the "debt turning point", while it becomes detrimental for debt-to-GDP ratios above the threshold. The same inverted U-shaped relationship is found between the monetization rate and the innovation rate. Finally, going to investigate the main transmission channels in the debt-innovation relationship, we find that the main ones are the number of researchers and market capitalization.