Accurate long-term cash flow forecasting may be treasury’s white whale, and while the hunt is certainly far from over, assistant treasurers at recent NeuGroup meeting heard about a solution for producing medium-term forecasts.
Lack of balance. At the meeting, one AT told the group he is having issues balancing two standard methods for cash forecasting in the short- and long-term.
- The short-term system forecasts two weeks in advance for AP purposes, based on a direct model of “cash positioning” that analyzes upcoming payments and receipts. It has a high degree of accuracy.
- But his yearly long-term forecast, a top-down approach based on high-level revenue and expense forecasts from FP&A, has a higher margin of error.
The power of a hybrid. In response, another AT shared that his company’s treasury and shared service center teams worked together to build a hybrid between the two models, a tool that generates far more accurate cash flow forecasts over the coming six to 12 weeks.
- He said the tool works by “looking into the system of AR and AP for what you see within your current terms.” Think of that as old school.
- It then “uses some AI and machine learning techniques to figure out historically where the rest of that period is going,” and makes extrapolations about the next three-month period. New school.
Bridging the gap. The tool helps to bridge a gap the member’s company had when “planning around how much we need for AP, share repurchases, and other outflows for the whole quarter, not just the next week.”
- “If you’re managing an investment portfolio that you’re using for liquidity you need to know your position on the curve in relation to your cash needs,” the member said.
- Another member who uses a similar medium-term forecasting model said that this method is typically far more accurate over a single quarter than the one-year forecast.
