Nowcasting Rwanda’s Quarterly GDP Using Mixed-Frequency Methods

Abstract

The delay in publication of Rwanda’s gross domestic product figures has sparked the search for econometric tools to produce timely prediction of the current state of economic activity for timely policy making at the Ministry of Finance and Economic Planning. Now-casting is a useful econometric tool that utilizes readily available information contained in high-frequency indicators to forecast the current (now) quarter GDP. In this paper we apply now-casting techniques, namely bridge equations and a battery of mixed-frequency data sampling (MIDAS) regression models and mixed-frequency vector auto-regressive (VAR) — a relatively recent methodology — to obtain current-quarter and next quarter now-casts of Rwanda’s real GDP growth by exploiting information contained in monthly macroeconomic and financial indicators. We then compare forecasting abilities (accuracy) of these techniques out of sample. Key to our findings is that a combination of bridge equation and unrestricted MIDAS forecasts gives best accuracy for current quarter now-cast of GDP.