Nowcasting and the Use of Big Data in Short-Term Macroeconomic Forecasting: A Critical Review

Abstract

This paper provides a discussion of the use of Big Data for economic forecasting and a critical review of recent empirical studies drawing on Big Data sources, including those using internet search, social media and financial transactions related data. A broad conclusion is that whilst Big Data sources may provide new and unique insights into high frequency macroeconomic activities, their uses for macroeconomic forecasting are relatively limited and have met with varying degrees of success. Specific issues arise from the limitations of these data sets, the qualitative nature of the information they incorporate and the empirical testing frameworks used. The most successful applications appear to be those which seek to embed this class of information within a coherent economic framework, as opposed to a naive black box statistical approach. This suggests that future work using Big Data should focus on improving the quality and accessibility of the relevant data sets and in developing more appropriate economic modelling frameworks for their future use.