Can agricultural commodity prices predict Nigeria's inflation?

This study examines the predictability of agricultural commodity prices in Nigeria's inflation forecast. It considers twelve major agricultural commodities and their predictability are evaluated singly and jointly for both food and headline inflation using various methodological approaches. Our preliminary analysis suggests that the predictor series exhibit some salient features such as persistence, endogeneity and conditional heteroscedasticity (high volatility) effects. Also, the parameters of the agricultural commodity-based inflation model tend to shift over short periods based on the results of the Bai and Perron (2003) test. Thus, we employ the Westerlund and Narayan (2015) estimator which accounts for these salient features. Also, we control for the time-varying parameters of the inflation model using the breaks obtained from the Bai and Perron (2003) test as well as the recursive (expanded) window approach. Our results show that agricultural commodities individually predict both headline and food inflation better than the random walk model which is the benchmark model for forecasting inflation in the literature. Thereafter, we construct a composite index for all the agricultural commodities using the principal components approach and the performance of the predictive model significantly improves relative to the individual agricultural commodities particularly when food inflation is considered. These results are consistent for both in-sample and out-of-sample forecasts and are robust to alternative measures of forecast performance, multiple forecast horizons and different data frequencies.