Using payments data to nowcast macroeconomic variables during the onset of Covid-19

Abstract

The spread of COVID-19 has caused large-scale loss of life and economic damage. This pandemic has had a swift effect on the macroeconomy, posing a new and different shock to the Canadian economy. Governments have responded in many ways, including through public health measures, fiscal stimulus and monetary policy. Policy-makers seek to understand the current state of the economy for their COVID-19 support to be effective. However, major economic indicators are released with a substantial delay. This problem is usually dealt with by using a linear model of past economic variables. But this is not the best approach presently, given the large and nonlinear effects of COVID-19 on the economy. In this paper, we develop a model to predict the current state of the economy—known as nowcasting—using retail payments system data and machine learning. The Canadian retail payments data aid in understanding the current state of the economy because they include many types of transactions and are available daily. These data features are ideal for macroeconomic nowcasting during a crisis. The flexibility of machine learning can help capture the large and nonlinear effects of the COVID-19 shock. We find that our model—compared with a benchmark model—has a significant increase in prediction accuracy, measured as a 15 to 45 percent reduction in forecast error.