The determinants of economic growth have long interested economists. A number of variables have been found to be significant, among them the private investment rate, human capital investment rates, the political stability of a country and others. An important sub-category of such determinants is policy variables. Specifically, two such variables are government consumption expenditure and the inflation rate.
In this paper we will employ an endogenous growth model as we investigate the effects of policy on per capita GDP. We allow for the possibility of non-linearities in the relationship between government consumption expenditure and the inflation rate and GDP.
Cross-sectional studies of the determinants of economic growth find the impacts of both government consumption expenditure and the inflation rate to be negative, as shown in Table 1. A distinguishing feature of these studies is that the policy variables enter the specification linearly. Either of the feasible signs on the policy variables implies a corner solution that seems implausible. Complete reliance on private markets is challenged at least by the literature surrounding the impact of human capital on economic growth. Complete nationalization of the economy is difficult to justify on efficiency grounds. The implied interpretation of the policy variables in growth studies is that they capture piecewise linearity. A better solution, therefore, would be to recognize the likely nonlinearities explicitly. It is with this task that the present paper is concerned.
The idea is that for relatively low levels of government consumption spending and inflation, the impact on the growth rate may be positive but as the ratio of government consumption spending to GDP and the inflation rate increase they begin to have negative effects on GDP. Time series estimations of this hypothesis show that this may indeed be the case for South Africa.
This is the only known study of its kind to undertake such an investigation. While the South African literature is peppered with comments and thoughts on the role of policy with respect to economic growth there have been no empirical investigations. Further, there is no known study worldwide that examines the possibility of a non-linear impact of policy on growth.
The paper draws from both the theoretical literature on growth as well as the international empirical findings. The following section provides a brief summary of the literature. In Section 3, we extend the finding of Barro (1990) that government consumption expenditure has an optimal level beyond which it begins to reduce per capita consumption to show that it also has an optimal level with respect to the growth rate of output. We also present a brief analysis of the effects of inflation on growth.
We then allow for the possibility that policy may have an indirect effect as well as a direct effect on growth via its effects on investment. We show that government policy 3 can affect investment, which, as shown by Levine and Renelt (1992), is one of the most robust determinants of growth.
Section 4 outlines the econometric methodology to be used, specifically the Johansen estimation approach as well as the autoregressive threshold effects methodology. We provide an outline of the models to be estimated. In Section 5 we discuss the data to be used. Section 6 outlines the univariate time series characteristics of the data and reports the empirical results. Using appropriate time series estimation techniques the empirical findings show that policy does indeed have a significant direct effect on per capita GDP in that higher levels of government spending and inflation reduce GDP. An examination of the possibility of the existence of an optimal level of government consumption spending and inflation show that there could indeed exist such threshold levels for both variables. The final estimation suggests that it is insufficient to examine only the direct effects of policy. It is necessary to examine the indirect impacts too. Section 7 concludes the paper.