Revisiting the accuracy of inflation forecasts in Nigeria: the oil price –exchange rate perspectives

Motivated by the dual characteristic of the Nigeria economy as exporter and importer of oil, we extend the Tule et al. (2018) oil price –based predictive model to include the role of exchange rates in the predictability of inflation in Nigeria. Using the FQGLS estimator developed by Westerlund and Narayan (2012, 2015), we account for endogeneity, persistence and conditional heterosecedatcity effects in the forecasting analyses of inflation in Nigeria. We use both the single and pairwise measures to compare the forecast results of oil prices and/or exchange rates based augmented Phillips curve with the variant that contain them both simultaneously. The result suggests that augmenting the Phillips curve with oil prices and exchange rates in the same model set up is the more accurate to forecasting inflation in Nigeria relative to having them individually in a predictive model. More so, the augmented Phillips curve with oil prices and exchange rate consistently outperforms time series models such as ARMA and ARFIMA. However, we find that accounting for structural breaks will largely improve the accuracy of CMB-APC for forecasting inflation in Nigeria. We find the robustness of these findings evident for in-sample and out-of-sample forecasts.