The forecasting of economic activity is of interest to governments, policymakers, researchers, investors and indeed all participants in the economy. While the South African Reserve Bank (SARB) publishes a number of formal indicators of economic activity, and researchers have managed to create substantial macroeconomic forecasting models, these indicators and models are usually quite complex and not always readily accessible. There is therefore ample scope for simplicity in the forecasting of economic activity. Indeed, a few well-chosen variables can perhaps be suitable indicators of future economic activity, without the need for complicated analyses or modelling of which to draw conclusions.
We construct a simple index of growth in real currency in circulation in the South African economy. The intuition is that increased currency in circulation is likely to be spent in the economy in the near future, therefore pointing towards a possible increase in economic activity. Traditionally, analyses of economic activity are concerned with the business cycle, which is formally published by the SARB. The formal business cycle, however, is a complicated exercise which incorporates in excess of 100 economic variables. We argue that analyses of the formal business cycle can be enhanced by other, complementary, measures of economic activity or output. We therefore deviate from the traditional view and propose that economic activity be measured by the output gap. To this end, we construct a dummy variable taking the value of Zt = 1 if the output gap is negative (i.e. output below trend), and Zt = 0 for a positive output gap. We subsequently utilise a probit framework, as popularised by Estrella and Hardouvelis (1991), to test how accurately our currency index can forecast these binary outcomes. In addition, we test the forecasting capability of the money stock (M3) in the same framework as a robustness check. Our models are estimated over the sample period 1980 – 2012.
It is found that the growth rate of real currency in circulation can predict economic activity up to 6 months ahead, whereas the money supply can predict economic activity up to 9 months ahead. Some valuable breakpoints can also be determined in the growth rates of currency and the money supply. If real currency in circulation grows by 1% or less, the probability of a slowdown occurring 6 months later is at least 60%. If real M3 contracts by 1% of more, the probability of a slowdown occurring 9 months later is 70%.
The relative simplicity of calculating the growth rates of monetary aggregates enables interested parties to form a forecast virtually instantaneously. Along with other well-known indicators, such as the yield spread or the SARB’s leading indicator, these variables can be valuable in supplementing any binary forecasting analysis.