Optimized ATM Management Saves Millions

2004-Mar-13 | With the help of neural networks, Siemens has developed a forecast tool that banks can use to precisely plan when to fill ATMs and how much cash to put in them. The system can lead to millions in additional profits for banks. The Intelligent Cash Management system uses the actual demand for cash as a basis for precisely calculating the amount of money to be put into ATMs. This reduces a bank's unproductive capital to a minimum, and its customers are no longer left standing in front of empty ATMs.

Depending on their locations, ATMs are generally filled with up to €400,000 in cash. So a bank that has about 1,500 cash machines is left holding anywhere between €45 million and €60 million in unproductive capital, because tens of thousands of euros in each ATM are never withdrawn. Such high cash reserves represent lost interest income, and they also generate additional insurance costs. In Germany alone there are approximately 50,500 ATMs. If each of them were filled with just €20,000 less cash, the amount of additional funds available would total one billion euros. And even if the money were invested conservatively, it would generate interest income of €50 million per year. The engineers at Siemens Corporate Technology in Munich have a good 15 years of experience with neural networks and forecasting for dynamic systems. Intelligent Cash Management is based on the Simulation and Development Environment for Neural Networks (SENN), which is designed for complex economic and industrial data analyses.

Cash turnover at ATMs can vary significantly depending on the location, day, and time of the year. For example, an ATM in a downtown area during a winter clearance sale is used more frequently than one in a rural community on a holiday. The system software uses withdrawal data provided by the bank to calculate the optimal amount of cash to be put into each ATM. The cash in the machine thus corresponds as precisely as possible to actual demand. (IN 2004.03.3)