Article 22 — How Capital Allocation Decisions Create Cash Flow Risk
The investment committee had approved 4 initiatives in Q1. Each had been evaluated individually on its strategic merit and expected return. Each had a comp...
Get started
The forecast had been built with genuine rigor.
Bottom-up revenue projections. Segmented collection assumptions. Expense timing based on actual payment schedules. The CFO had spent 3 days building it and had presented it to the board with confidence in the methodology. The assumptions were defensible. The model was well-constructed. The number at the bottom was credible.
4 weeks later the actual cash position was $1.4M below the forecast. The variance was not explained by a single event. It was the accumulated effect of 23 smaller variances, each individually within an acceptable range, that had all moved in the same direction simultaneously. The forecast had not been wrong in its methodology. It had been wrong in its assumption that the individual variances would be random rather than correlated, and that assumption had been built into every cash flow forecast the business had produced for the previous 4 years.
Cash flow forecasts are systematically biased toward optimism not because the people building them are careless but because of specific structural features of how the forecasts are constructed and what the forecasters are incentivized to produce.
Revenue assumptions in cash flow forecasts typically reflect the operating plan rather than a conservative cash receipt estimate. The operating plan is built to motivate the organization toward a target. A forecast that assumes full achievement of the operating plan is building organizational aspiration into a financial planning instrument. When the operating plan is not fully achieved, the cash flow forecast built on it is wrong by construction.
Collection timing assumptions typically reflect contractual terms rather than actual behavior. A forecast that assumes customers will pay in 30 days when the actual average collection is 47 days is systematically understating the time cash will be in the receivables cycle. If the 47-day average is itself subject to variability, the actual range of collection timing is wider still, and the forecast reflects neither the average nor the distribution.
Expense timing assumptions frequently reflect planned payment dates rather than actual payment behavior. Discretionary expenses that could slip are modeled as occurring on schedule. Supplier payments that could be managed are modeled as being made at the standard cycle. The forecast is built for the world where everything happens as planned, and that world is not the world the business actually operates in.
“Every individual assumption in our cash flow forecast was reasonable. The problem was that they were all optimistic, and when we were wrong we were wrong in the same direction every time. A good forecast has errors that go both ways.”
The most technically significant driver of forecast miss is the correlation between individual forecast errors. A cash flow forecast that models 50 individual cash flow items assumes implicitly that the errors on each item are independent. If revenue comes in below forecast, the probability of expenses also coming in below forecast is not affected. Each item misses or hits independently.
In practice, cash flow components are correlated. A period of revenue underperformance is also likely to be a period of collection delays, as the commercial conditions that affect revenue also affect customer payment behavior. It is also likely to be a period of inventory build, as procurement based on the revenue plan continues even as demand comes in softer. And it is likely to be a period where the business defers discretionary spending to conserve cash, which can shift some expenses into future periods and others out entirely.
When the correlations are negative, forecast errors offset each other and the forecast is more accurate than the individual component accuracy would suggest. When the correlations are positive, errors amplify each other and the forecast miss is larger than any individual component error would explain. Cash flow components in a soft operating period are typically positively correlated, which is why forecast misses tend to be larger in magnitude than the average component accuracy predicts.
How cash flow forecasting connects to working capital management is a structural relationship that most businesses have not examined explicitly. The forecast governs working capital decisions. The forecast is systematically wrong in soft periods. The working capital decisions made on the basis of the forecast are therefore made on optimistic assumptions precisely when conservatism is most warranted.
Improving cash flow forecast accuracy requires addressing the structural features that produce systematic bias rather than refining the individual component assumptions that are already being carefully constructed.
Using scenario-based forecasting rather than single-point forecasting addresses the correlation problem by modeling the full distribution of outcomes rather than the expected outcome. A forecast that shows the expected case, a downside case where revenue comes in 15% below plan and collection timing extends by 10 days, and a stress case where both conditions occur simultaneously gives the leadership team a working capital decision framework that is robust to the range of likely outcomes.
“Moving from a single point forecast to a 3-scenario model changed our working capital decisions immediately. We stopped making decisions based on the expected case and started making them based on what the downside case required us to hold. The cash position became more stable because our decisions were sized for the risk we were actually carrying.”
Using actual behavioral data rather than contractual assumptions for collection timing addresses the systematic collection timing bias. The forecast that assumes 30-day collection because the terms say 30 days will be wrong in every period. The forecast that assumes 47-day collection because the data shows 47 days will be right on average and will be more useful for working capital planning as a result.
Building the forecast at a weekly rather than monthly granularity addresses the timing correlation problem. Monthly forecasts obscure the within-month timing of inflows and outflows that can produce cash gaps even in months where the monthly total is on plan. Weekly forecasting surfaces those gaps with enough lead time to address them.
This Article Is Part of a Larger Series
Access the complete working Capital Series
The investment committee had approved 4 initiatives in Q1. Each had been evaluated individually on its strategic merit and expected return. Each had a comp...
Get started
The business had been profitable for 6 consecutive years. Not marginally profitable. Consistently profitable, with EBITDA margins that compared favorably t...
Get started