South African headline and core inflation surprised to the downside yesterday. Our measure of common price changes across categories in the CPI basket has, however, not turned down to the same extent and remains higher than the Statistics SA measure that excludes food, non-alcoholic beverages, fuel and energy. Another core measure, the trimmed mean, has moved closer to the mid-point of the inflation target. Since 2012, the trimmed mean measure has had the highest correlation with headline inflation of these measures, followed by CPI-Common. Our measure suggests that there has been more broad-based inflation pressure since 2019 than implied by the core measure of underlying inflation.
Its worth noting that there is typically no ‘best’ measure of core inflation, with different measures having different strengths and weaknesses. For example, our CPI-Common measures includes administered taxes and price components that are often excluded by other measures. The appropriateness of this methodological choice depends on the role government-determined prices play in domestic inflation in South Africa. We argued in this paper that unlike in many advanced economies, government prices are a very important driver of domestically-determined inflation in South Africa (the paper also calculated a non-tradable inflation series as used by major central banks). This post provides a chart.
Judging which of these measures best represents the current extent of underlying inflation pressure and provides the best guidance to monetary policy requires an empirical assessment. Something analysts in South Africa do not seem to appreciate is that these core measures have very low correlation with one-year ahead headline inflation and one can do a much better job predicting inflation using seasonal factors and judgements about government-related inflation and indexed prices. In a future post, we will evaluate whether any of these core measures has historically been a good guide to persistent changes in inflation in South Africa.