Nowcasting Inflation at Quantiles: Causality from Commodities

This paper proposes a non-parametric test for Granger causality in quantiles to detect causality from a high-frequency driver to a low-frequency target. In an economic application, we examine Granger causality between inflation, as a low-frequency macroeconomic variable, and a selection of commodity futures, including gold, oil, and corn, as high-frequency financial variables. We find that logarithmic returns on given commodity futures are a prima facie cause of inflation at the lower quantiles of the distribution and marginally around the median. In the context of a nowcasting exercise, we find that incorporating commodity futures in the model with a polynomial function enhances short-term forecasting accuracy, leveraging timely data for more precise nowcasting of inflationary trends.