A dynamic factor model for forecasting macroeconomic variables in South Africa

Abstract

This paper uses the Dynamic Factor Model (DFM) framework, which accommodates a large cross-section of macroeconomic time series for forecasting the per capita growth rate, inflation, and the nominal shortterm interest rate for the South African economy. The DFM used in this study contains 267 quarterly series observed over the period 1980Q1-2006Q4. The results, based on the RMSEs of one- to four-quarters-ahead out of sample forecasts over 2001Q1 to 2006Q4, indicate the DFM outperforms the NKDSGE in forecasting per capita growth, inflation and the nominal short-term interest rate. Moreover, the DFM performs no worse compared to the VARs, both classical and Bayesian, indicating the blessing of dimensionality.