Forecasting Annual Inflation Using Weekly Money Supply

Forecasting inflation may be challenging, especially when inflation is high. Over the past decades, many developing countries have faced, and some are currently facing high inflation. For these countries, it is challenging to have predictive inflation accuracy. This paper presents a mixed-data sampling (MIDAS) method to model and forecast inflation in Suriname. We use the weekly money supply from the central bank's balance sheet as an explanatory variable. We apply this method for forecasting inflation for Suriname, where average inflation in the 1990s exceeded 350%. The results of the MIDAS models with annual inflation data show large variations in forecasts. Some of these models include money supply as an explanatory variable. We assess the models' forecasting performance based on the forecast error. The MIDAS models lead to a substantial improvement in forecast accuracy, also for the years with high inflation. We show that our method is particularly relevant for forecasting high inflation rates.