Revisiting the Causality between Electricity Consumption and Economic Growth in South Africa: A Bootstrap Rolling-Window Approach

This paper revisits the causality relationship between electricity consumption and economic growth in South Africa for the period 1971-2009 using annual data and takes into consideration time variation in causal relationships using bootstrap rolling Granger non-causality tests. Full-sample Granger causality tests find absence of any causality between electricity consumption and economic growth. However, parameter stability tests indicate that there is instability in our VAR model and therefore findings from our full-sample Granger causality test cannot be relied upon. This motivates the use of bootstrap rolling-window estimation to investigate the electricity consumption-growth nexus which accounts for the time varying causal link between the two variables. The results indicate two sub- periods, 2002-2003 and 2005-2006, whereby electricity consumption had a causal effect on GDP. Apart for these brief sub periods, the results indicate no causality between the two series. However, these sub-sample periods correspond to significant economic events in the South African electricity market and the economy in general, indicating that the findings are not statistical artefacts, but correspond to real economic changes.