Nowcasting Chilean household consumption with electronic payment data

When economies are hit by relevant shocks, the need to be able to follow developments in real-time increases for policymakers and private agents alike. When an event of this type is underway, the situation can change dramatically in a matter of days. The COVID pandemic is only the latest example. Electronic payment data is available with virtually no time lag and could therefore contribute to increasing the speed at which assessments are made. This paper makes use of a novel database to track Chilean household consumption in real time during the pandemic and compares the results to those of standard nowcasting methods used at the Central Bank of Chile. The results suggest that payment data is most useful as the shocks occur, when traditional models may have a harder time interpreting the information. The gain in more stable times is less obvious. The results also show, as one might expect, that the relationship between this naturally-occurring data and the variable of interest can be affected by its own shocks. In this case at least, electronic payments showed sudden shifts in intensity that needed to be accounted for in order to produce the final forecasts. All in all, the models based on payment data appear to be a relevant addition to the forecasting toolkit.