Three Empirical Essays on Interventionism and the Economic Impact of International Financial Integration in Africa

Nitish Gobin

Advisor: Carlos D Ramirez, PhD, Department of Economics

Committee Members: Peter J. Boettke, Lawrence White

Buchanan Hall, #D002
November 17, 2023, 11:00 AM to 01:00 PM


Chapter one studies the impact of International Financial Integration (IFI) on economic growth in Africa by using panel vector autoregression modeling on 40 African states from 2000-2021. IFI is the cross-border holdings of financial assets and liabilities. Africa’s share in global trade in financial assets and liabilities is low despite its high marginal productivity of capital (MPK), hence illustrating the Lucas paradox. The impulse response function depicts that the lagged one values of financial integration have significant positive effects on contemporaneous economic growth. I also show evidence that Granger causality is unidirectional and runs from IFI to economic growth. In addition, this paper also sheds light on the Lucas paradox by revealing that institutional factors explain the low IFI in Africa. Improvements in Economic Freedom Index, Human Development Index, and national income contribute significantly to the enhancement of IFI in Africa. African policymakers should improve both the economic freedom and human development indicators since it takes less than 2 years for their impulses to manifest on IFI, which in turn impact real GDP by 2% annually within two years. Moreover, robustness tests show that Economic Freedom Index and Human Development Index amplify the IFI effect on economic growth.

Chapter two investigates the unintended consequences of government intervention and the optimal point for 54 African countries through a panel data regression analysis from 2001-2021. The empirical findings of this study reveal that direct government intervention, proxied by general government final consumption expenditure as a percentage of GDP and central government debt as a percentage of GDP significantly and adversely affect Real GDP per capita in Africa. A 10% increase in government expenditure as a percentage of GDP causes Real GDP per capita to fall by 2.8%. Moreover, a 10% rise in central government debt as a percentage of GDP results in a 0.23% reduction in Real GDP per capita. Introducing a quadratic variable of government expenditure as a percentage of GDP further elucidates these dynamics: initially, at lower levels of government spending to GDP, a positive effect on growth is observed until government expenditure as a percentage of GDP reaches 11.4%, at which point growth is maximized at 2.3%. However, beyond this optimal threshold, economic growth begins to decline, eventually becoming negative once government expenditure as a percentage of GDP surpasses 23.8% and hence confirming the inverted U-Shaped effect of government expenditure on economic growth.

Chapter three investigates the applicability of Wagner's Law of "expanding state expenditures" to Mauritius using time series regression and Vector Autoregression models using data from 1976 to 2022. Wagner's Law posits that economic growth causes growth in public expenditure. The first statistically significant evidence of the applicability of Wagner’s Law to Mauritius stems from the bivariate regression of real government expenditure on real GDP where the income elasticity of demand for real government expenditure is elastic (1.02). The time series regression results also give strong statistical evidence that the Wagner’s Law applies to Mauritius where a 10% rise in the natural log of Real GDP per capita causes the natural log of real government expenditure per capita to rise by 10.3. Liberal democracy index is negatively correlated with the natural log of real government expenditure per capita, implying that government is incentivized to increase its activities when democracy declines. Under a more dynamic setting, the VAR(1) model confirms that the Wagner’s Law is forecast to persist in the short to long-run. Granger causality test confirms that unidirectional causality runs from real GDP per capita to per capita government expenditure. When the lag one value of the natural log of Real GDP per capita rises by 10%, the contemporaneous value of natural log of real government expenditure per capita rises by 4.3% and is statistically significant. The Impulse Response Functions analysis show that a one standard deviation shock to natural log of Real GDP per capita positively impacts natural log of real government expenditure per capita, with an immediate rise of 0.38% in year 1, reaching a peak of 1.1% in year 3, after which it declines marginally to stabilize around 0.7% in the medium to long-run. These findings hold significant implications for policymakers, offering insights into government expenditure policy formulation in Mauritius.