Revisiting the forecasting accuracy of Phillips curve: The role of oil price

Highlights: • The forecasting accuracy of the traditional-demand side Phillips curve model is revisited. • The inclusion of the supply side-oil-based factor in the Phillips curve model is proposed. • The augmented version outperforms the traditional both in in-sample and out-of-sample. • The augmented version also outperforms the autoregressive model. • The results are robust to different countries, measures of inflation and forecast horizons. - Abstract: In this paper, we propose a revision to the traditional (demand side) Phillips curve to capture the supply (cost-push) side of inflation. We adopt the Westerlund and Narayan [WN] (2015) approach which accounts for persistence, endogeneity and conditional heteroscedasticity effects in the predictive regression model. In addition, following the approach of Salisu and Isah (2018), we extend the oil-based bivariate framework of WN (2015) to a multi-predictor set-up in order to augment the traditional Phillips curve-based inflation model with the proposed cost-push factor. Using the OECD countries, we demonstrate that the forecast performance of the traditional Phillips curve tends to improve when it is augmented with oil price both for the in-sample and out-of-sample forecasts. Contrary to the prominent findings in the literature, the augmented Phillips curve model outperforms the first order autoregressive model. Our results are robust to alternative measures of inflation rate and different forecast horizons.