Tactics for approaching cash optimisation in bank branches

Banks, through their branches, offer cash transactions as a key service, so any lack of available cash is a critical issue, since it affects the prestige of the entire bank not just of that branch in particular. This paper aims to establish a vault policy, that lets the staff know how to manage the branch's money efficiently, by handling orders to the central vaults with realistic assumptions and easy‐to‐implement criteria. We present a model based on dynamic programming principles, to represent the problem and generate an input for the vault policy. Other inputs for the policy are some parameters set according to a branch's demand for a cash transactions approach, from the perspective that the best fit is between frequentist and Bayesian. With these inputs and a definition of some 'intuitive' rules we can implement and assess the vault policy. The results for 30 branches are presented.