Macroeconomic Predictions using Payments Data and Machine Learning

Predicting the economy’s short-term dynamics—a vital input to economic agents’ decision-making process—is often done using lagged indicators in the linear models. This is typically sufficient during normal times, but it could be inadequate during the crisis periods such as COVID-19. In this paper, we demonstrate: (a) the payments systems data which captures a variety of economic transactions can assist in estimating the state of the economy in real time, and (b) the machine learning can provide a set of econometric tools to effectively handle wide variety in the payments data and to capture sudden and large effects of the crisis. Furthermore, we mitigate interpretability and overfitting challenges of machine learning models by using the Shapley value-based approach to quantify marginal contribution of each predictor, and devising a novel cross-validation strategy tailored for the macroeconomic prediction models.