Can Monetary and Money-Market Indicators Improve Inflation Forecasts in Nigeria? Evidence from Econometric and Machine-Learning Models, 2006-2026

Abstract: This paper asks whether the monetary policy rate, the Treasury bill rate and the interbank call rate help to forecast Nigerian headline inflation one month ahead. Using monthly Central Bank of Nigeria data from 2006 to April 2026, I compare a random walk, an AR(3), an autoregressive distributed-lag (ARDL) model with money-market indicators, a ridge regression and two random forests, one fitted to the level of inflation and one to its monthly change. All models are re-estimated each month and evaluated on forecasts for January 2022 to April 2026, a period that includes the steepest inflation surge in two decades and the rebasing of the consumer price index in January 2025. The ARDL model is the most accurate, with a root mean squared error of 0.844 percentage points against 0.972 for the random walk and 0.894 for the AR(3), and its gains are statistically significant (Diebold–Mariano p = 0.043 against the random walk; Clark–West p < 0.001 against the AR(3)). It is also the best model in each sub-period, and its advantage over the AR(3) survives rolling estimation windows and a three-month horizon, although its advantage over the random walk does not. A random forest fitted to inflation levels is the worst model by a wide margin (RMSE 3.78) because it cannot predict values above those in its training data, which never exceeded 18.7 per cent before 2022. Fitting the same forest to monthly changes removes this problem and brings its accuracy close to that of the AR(3). The money-market variables are jointly significant in-sample over the full period (F = 2.87, p = 0.016) but not before 2022, suggesting that their predictive content emerged during the recent high-inflation episode. The rebasing month itself produces a forecast error of about seven points in every model and is best treated as a measurement break rather than a forecasting failure.

Keywords: inflation forecasting; monetary policy; money-market rates; random forest; forecast evaluation; Nigeria

JEL classification: E31; E37; E52; C53