Forecasting Inflation in Developing Economies: The Case of Nigeria, 1986-1998

In this article, we have sought to establish whether monetary aggregates have useful information for forecasting inflation, other than that provided by inflation itself. We have approached the problem in two ways. First, we conducted forecasting experiments, using Mean Absolute Percentage Errors (MAPEs). We then evaluated whether each monetary variable improved the forecasts of a simple AR (1) model of inflation. From the study, we found that the MAPEs for all the variables were less than that of the benchmark AR (1) model. The forecasting experiments showed that, over the whole sample period, most of the variables examined served as important information variables for price movements. We found that the Treasury bill rate, domestic debt and M2 provide the most important information about price movements. Treasury bill rate provided the best information, since it has the lowest MAPE. Conversely, the least important variables were the deposit rate, dollar exchange rate and M1. M2 provides more information about inflation than M1 in the sample period. We also estimated an inflation equation and determined alternately whether M2 enter the equation significantly. We found that M2 is not significant. Exchange rate at level, and contemporaneous value of the domestic debt, are significant in the model. The results obtained are robust across the two methods used and we concluded that although the monetary variables contained some information about inflation, exchange rate and domestic debt may be more useful in predicting inflation in Nigeria. A number of policy implications emerge from the study.