Does the price of oil help predict inflation in South Africa? Historical evidence using a frequency domain approach

Abstract

The association between oil prices and inflation has remained an intriguing issue for media, academic as well as policy enquiry. Against this backdrop, we perform the frequency-domain causality test to investigate whether the growth rate of oil prices has predictive content for inflation in South Africa. As a preliminary step in our analysis, given that we use a long historical data set spanning from 1922:M01 to 2013:M07, we investigate the possibility of structural breaks in the inflation equation. We detect three breaks which define four regimes. We then perform the frequency-domain test on the full-sample, as well as, the four identified regime-specific sub-samples. We find evidence of the growth rate of oil prices to have predictive content for South African inflation based on the full-sample, as well as, two of the four regime-specific sub-samples. Given that the frequency-domain test allows us to decompose the causality across different time horizons, results also suggest that cycles of predictability of South African inflation emanating from the the growth in international oil prices could last for much longer durations for periods preceding the adoption of an inflation-targeting framework for monetary policy in South Africa.