Central Bank Forecasting of Cash Demand with AI

Artificial Intelligence (AI) is a hot topic and there are a number of conferences focusing on how central banks could, and should, be using it. Jelena Stapf, Head of Division Branch operations, IT Applications and Automation, Bundesbank (BBk), presented how the BBk’s cash demand is using it at the ATMIA conference.

Many planning decisions of the Bundesbank rely on net cash demand forecast based on data from its regional branches.

Inevitably cash demand is uncertain and accordingly needs to be forecast. This presentation considered the BBk’s range of forecast methods and how it is incorporating AI.

The BBk’s starting point is a manual forecast (seasonal naïve) which is its benchmark for other work. It creates longer term forecasts for production requirements based on (structural) time series models and uses machine learning models for optimising the replenishment of ATMs. This creates a cash demand forecast for regional branches that incorporates recycling. The BBk plans replenishment to an optimal level down to branch level.