Oops! Something went wrong while submitting the form.
The coronavirus pandemic has created problems for traditional cash management forecasting tools. Banks have reported ‘turning off’ their software-based tools in favor of manual overrides. In a fast-changing environment, forecasting needs to be dynamic, considerate of the challenges the supply chain is experiencing and more individually site focused than ever.
CMS Analytics’ outsourced cash management has achieved industry leading results for our clients since our inception, however our philosophy and solution design were never stronger than through the pandemic and the continued impacts of it.
How did we do it?
Our Pandemic Started Early
While our major client bases of the US, Canada and the UK would most likely concur the pandemic started in March 2020, at CMS we had begun to consider the implications of the coronavirus from the turn of the year. With CMS’ cash management solution being utilized in over 30 countries around the world, we were able to take lessons from our Asia Pacific territories for what the consequences may be. When coronavirus cases and measures began to be impactful in the UK, US and Canada, we were prepared.
Our Solution is Dynamic
Software solutions typically utilize one algorithm across ATM and branch networks which show vastly different characteristics. This can lead to underperformance either on a residual or availability basis – too much or too little cash. This weakness was exacerbated through the pandemic as demand volatility increased significantly (Figure 1), individual site requirements became far more specific and software solutions got caught out on both lockdowns and reopenings:
Due to lockdowns, sites’ demand could fall to zero if the building they were in was inaccessible or the usual footfall or traffic wasn’t in place. However, the recurring nature of software solution’s forecasting meant that demand for an ATM could still be forecasted at tens of thousands of dollars, even though no customers could get to it!
Upon reopening, many sites saw a completely new demand profile, yet the forecast algorithm was still using historic data on a rolling basis. This meant that it couldn’t catch-up and sites were being under- or over-forecast depending on their new profile; the historic data of which had become even more meaningless due to the lockdown period.
At CMS, we forecast to the lowest possible level, usually denomination within a site. Furthermore, we have a roster of forecasting techniques which we use on a per-site basis, so your network will be forecast by many different algorithms. This allows us to build a bottom-up approach to the forecast for improved accuracy and a more realistic picture of customer demand.
Figure 1: Demand Volatility by Weekday Pre-Pandemic to Pandemic
Our Solution Adapted to the Specifics of the Pandemic
It wasn’t just lockdowns and reopenings that our solution can adapt to. Along with the normal challenges such as weather events and holidays, we were also able to include the impact of wider economic factors:
Panic Spikes: As cases started to rise and governments acted, the initial impact that we identified was large panic spikes in demand. These had to be accounted for in current forecasts and excluded from future ones, and the severity of the impact determined on a site-by-site basis.
State, Provincial or Regional Measures: Sites were impacted by local measures which meant cash demand and operational differences between regions. The reopening also brought the same challenges, as regions eased restrictions at different times and to varying extents.
Stimulus: Many stimulus measures across the world brought about a greater supply of cash. The magnitude of these, as well as the demographics of the region, significantly impacted the cash demanded from financial institutions and utilized at retail businesses. The utilization at retailers was affected in the pandemic when some retailers didn’t accept cash – this also had to be factored into forecasts and our retail clients’ optimization.
Changing Consumer Behavior: Following the reopening, there have been changes to consumer behavior that need to be accounted for in forecasting. For example, higher average transaction values were being withdrawn from ATMs impacting both the forecast and operators’ costs. We optimize every ATM every day to ensure these changing behaviors did not impact our clients as significantly as others.
Supply Chain: The cash supply chain went through tremendous turmoil throughout the pandemic. Armored transport companies had to navigate shortages in staff, circulation challenges with coin, restrictions around access and, in certain regions, public disturbances. Considering these in the forecasting and overall cash management process was essential for a bank to stay optimized.
Figure 2: Country Specific Challenges through the Pandemic for CMS' Main Markets
We Leverage our Solution with Human Intelligence
The dynamism of our solution is enhanced by the involvement of human intelligence and decision making. We have a team of mathematicians, data scientists and economists who build, monitor and continuously improve our models to ensure the best outcomes. When a site’s demand is flagged as showing an anomalous profile, our team can intervene, understand the root cause of the problem and make a decision for the mediation of the site. This results in greater availability overall with potential future cash-outs being seen at the earliest possible point in time and supply chain challenges adapted to.
Furthermore, our team solves unique challenges for our clients such as ATM multi-denomination optimization and re-scheduling of armored transport. The systemic shock of the pandemic and the changing customer needs coming out of it resulted in several such challenges being experienced by banks in particular. By working with our clients, they were able to adjust strategies, achieve cost savings despite the pandemic and were able to adapt to new demand profiles thus ensuring a high customer experience.