A Cointegration and Causality Test on Government Expenditure –Economic Growth Nexus: Empirical Evidence from a South African Province

This paper examines the longstanding debate on government expenditure–economic growth nexus based on two opposing theories (i.e., Wagner’s Law and Keynesian theory) focusing on the Free State (FS) province in South Africa using both bivariate and trivariate models estimated for the period 2001:Q1 to 2014:Q4. Time-series on population per capita is used as a third variable in the multivariate to obviate the ‘omitted variable’ problem and erroneous conclusion on the existing causative links among variables. The long-run relations and causal links were tested employing the Autoregressive Distributed Lag (ARDL)–bounds testing approach and Toda-Yamamoto causality test. Our results show a bi-directional causal link between total provincial government expenditure (PGE) and economic growth (real GDP) in the short run, while the real GDP Granger-causes PGE in the long-run. We also find a unidirectional causal flow from population growth to both GDP and PGE in the short-run, suggesting that demographic factor plays a key role in explaining a rise in total PGE. Other evidence reveals a long-run income elasticity ranging from 0.99 to 1.2% suggesting that a 1% rise in real GDP in the FS province can cause PGE to rise by 0.99 – 1.2%. The result on ‘voracity effect’ shows that external shock could lead to an increase of about 0.49 – 2.12% in total PGE, in the short-run. These findings provide concrete support for the adopted stringent fiscal consolidation stance by the provincial government to ensure efficient allocation of limited available fiscal resources, curb wasteful expenditure and reduce government size. Albeit, the bi-directional causal link between real GDP and PGE requires policymakers to prudently balance both current and future public expenditures in order not to crowd-out output and labour productivity.