Does the use of big data variables improve the prediction of Monetary Policy decisions? The case of Mexico

Abstract

We analyzed the predictive power of a market attention variable, generated using big data, for Banco de Mexico’s (Mexican central bank, hereby “Banxico”) monetary policy decisions. The novelty of this paper relies on the lack of previous research that incorporates a nonconventional variable that uses big data analysis in monetary policy research. We used a binary probit approach and contrasted different models to identify whether the proposed variable improved the prediction. Our general results show there is significant evidence that the variable improves the prediction, as it helps reduce information criteria and it stays significant across the different models. We consider that further research is necessary to determine the scope of big data in monetary policy analysis prediction.