Cash flow forecasting is always a sensitive point when talking to anyone in finance or treasury.
There have been many attempts to find a solution to the problem of accurate cash flow forecasting but there always exists one fundamental problem: no one ever got fired for missing a cash flow forecast. But the pressure on organisations to become more accurate is increasing and doesn’t look like going away any time soon.
But the basis of the problem has not changed that much over the years:
- Every business knows that there are a series of known costs that in the short term are fairly predictable: wages, salaries, taxation, debt repayments.
- But there are two major components of cash inflows that are very difficult to predict: accounts payable and account receivable.
It is probably more obvious that accounts receivable might be difficult to predict. Customers do not always pay on time. Often customers will pay late but early payment can be just as bad in destroying the prediction. Collectors often intuitively know which customers will pay early, late and on time but this information is rarely quantified into any predictive model.
It might be more surprising that accounts payable can be just as difficult to predict. It might seem simple for an accounts payable manager to predict what will be paid to suppliers only one week ahead. But these answers can be quite inaccurate. This can be due to delayed authorisation processes within the organisation but often it is as simple as the invoice has not yet been received from the supplier.
More than 75% of all paper invoices take more than 7 calendar days between being issued by the supplier and being received by the customer. At one client we found that it took 2 days longer for paper invoices to reach the shared service centre than if they had been sent to the incorrect plant address. Suppliers do this to save cost on postage or couriers and as a result can make short term payables forecasts very difficult to manage.
The one good thing that has happened in recent years is that software vendors have started to produce models that use accounts payable and accounts receivable transactional data to try and predict the exact payment and receipt dates of cash outflows and inflows. To date all these models have a basic flaw; they rely on system due dates to be accurate. The fact is, as we saw earlier, that the gap between due dates and actual payment date can be extremely erratic and these models make no attempt to try and predict the probability of on-time payment.
Working Capital Channel models
To address this problem, the Working Capital Channel have developed our own models to predict the actual payment dates of both payables and receivables. These models borrow a thought from inventory management models that try to predict demand variability. In this case we measure the variability around the due date of the actual payment date. Thus for each payment term and each supplier it is possible to predict, within a margin of error, when payments will be early, on-time and late. The models can then be used to produce a valid forecast error measurement. Simulations show that those companies that have very stable internal processes will have lower rates of forecast error than those that have processes with less predictable outcomes. The transactions can be examined to see what is causing the forecast error, action can be taken to make the process more robust and that will lower the forecast error. Results to date have shown that highly robust processes can produce forecast errors of less than 5%.
But this is a journey:
Many companies suffer from short term forecast errors of up to 25%. It normally is a process of continually examining forecast error results and tweaking processes to resolve these errors before more accurate results can be derived. These models have already proved to be a huge advance on any other current available short term cash flow model.
Cash Flow Forecasting tool
The every latest advance is a tool that has been built by a niche software provider, called DiscoverEdge. Their Cash Flow Forecaster uses the latest big data technology to quickly construct your cash flow forecast from your own ERP and non-ERP data. You can then simulate various eventualities in your cash flow forecast and predict exactly what the effect will be and when it will have its impact. The real beauty of this tool is that it is a fraction of the price of previous tools that have attempted and failed to do the same thing and it is extremely easy to implement.
Holy Grail of Cash Flow Forecasting
But the Holy Grail of cash flow forecasting would be to marry up the short term forecast of the next 8 to 12 weeks, with a longer term forecast of 12 to 18 months. Our short term models do not yet answer the long term question, but with the increasing power of transactional databases and the ability to start answering the short term questions the logical progression should be that models emerge in the coming years that start to address this problem. Technology on its own is unlikely to solve these problems without input from business users and experts in forecasting techniques.
We would be very happy to hear about your experiences!