Cash forecasting: predicting the money before it moves

Cash forecasting is the cash management service that predicts the corporate cash position, days and weeks ahead, by combining the known balances, the scheduled payments, the expected receipts, and the seasonal patterns, so the treasurer can decide today whether to borrow, invest, or leave the money where it is. A treasurer who knows on Monday that the group will be EUR 5 million short on Friday can arrange borrowing on Monday, at a good rate, while a treasurer who discovers the shortfall on Friday must borrow urgently, at a worse rate, or miss a payment. Source: Bank for International Settlements, payment system glossaries; ISO 20022 camt.052 and camt.053 for balance reporting. Why forecasting exists is that a corporate cash position changes every day, and the treasurer must manage the change, not just report it. A group receives EUR 10 million from customers, pays EUR 8 million to suppliers, pays EUR 2 million in salaries, and repays EUR 1 million of debt, all in the same week, and the net is a deficit of EUR 1 million, but the timing matters: if the receipts arrive on Friday and the salaries are due on Wednesday, the group is short on Wednesday even though it is in surplus on Friday. Forecasting answers by showing the position day by day, so the treasurer can see the Wednesday shortfall on Monday, and can borrow EUR 1 million for two days, or delay a supplier payment, or bring a receipt forward, and the decision is made with time to spare. The parties are the treasurer, who owns the forecast, the subsidiaries, which provide their expected payments and receipts, the bank, which provides the balances and the reporting, and the systems, which combine the data into a forecast. The forecast inputs are the current balances, the scheduled payments from the enterprise resource planning system, the expected receipts from the sales ledger, and the historical patterns, such as the fact that customers usually pay five days late, or that December is always a high-spending month. Each input has an owner, and a forecast is only as good as its inputs, so the BA must name the owner of each input, because a forecast built on stale inputs is worse than no forecast, because it gives false confidence. How forecasting works is shown by a worked example with real numbers. A group starts Monday with EUR 4,000,000, expects receipts of EUR 2,000,000 on Tuesday, EUR 3,000,000 on Thursday and EUR 1,000,000 on Friday, and must pay salaries of EUR 2,500,000 on Wednesday, suppliers of EUR 3,000,000 on Thursday, and debt of EUR 1,000,000 on Friday. The forecast shows Monday EUR 4,000,000, Tuesday EUR 6,000,000, Wednesday EUR 3,500,000, Thursday EUR 3,500,000, and Friday EUR 2,500,000, so the group is never short. If the Thursday receipt of EUR 3,000,000 slips to Monday, the forecast shows Thursday EUR 500,000 and Friday a deficit of EUR 500,000, so the treasurer arranges a EUR 500,000 overdraft on Wednesday, before the shortfall happens, and the salaries are paid on time. On success the forecast is accurate, the treasurer borrows less, invests more, and never misses a payment, and the forecast accuracy is measured, usually as the difference between the forecast and the actual, which should narrow over time as the inputs improve. On failure the forecast is wrong, because a receipt slipped, a payment was larger than expected, or a subsidiary did not report its payments, and the treasurer borrows too much, or too little, or invests money that is needed the next day, and the cost is interest paid unnecessarily, or an overdraft fee, or a missed payment. The recovery is a forecast that is updated daily, with actuals replacing forecasts as the week progresses, and a variance report that shows which inputs were wrong, so the inputs can be fixed. The data in a forecast is the opening balances, the expected receipts and payments by day, the value dates, the currencies, and the historical accuracy. The value date is the field that matters most, because a receipt with a value date of Friday is not available on Wednesday, and a forecast that treats it as available on Wednesday will show a surplus that does not exist. A BA must specify the forecast horizon, usually 13 weeks, the inputs and their owners, the update frequency, the variance reporting, and the distinction between a committed payment and an expected payment, because a forecast that treats an expected receipt as certain will be wrong when the receipt slips. Edge cases show the depth. A forecast in multiple currencies must convert each currency to the base currency at a forecast rate, and the rate must be stated, because a forecast that uses the spot rate for a payment due in three months will be wrong if the rate moves. A forecast that includes a subsidiary in a country with exchange controls must not assume the cash can be moved, because the cash may be trapped, and the forecast must show it separately. A forecast for a seasonal business, such as a retailer at Christmas, must account for the seasonal pattern, or it will understate the December receipts and overstate the January payments. The effect on the bank is a service that helps the treasurer, earns fees, and deepens the relationship, because a treasurer who forecasts with one bank is unlikely to move to another. The effect on the customer is less borrowing, more investment income, and no missed payments, which is measurable in interest saved and in supplier relationships kept. The effect on operations is a daily forecast run, with inputs collected from the subsidiaries, actuals replacing forecasts, and a variance report that shows which inputs were wrong and why. What a BA, PO or SME must understand is that forecasting is a prediction, and its value is the decision it enables, not the report it produces. The requirement must name the horizon, the inputs and their owners, the update frequency, the variance reporting and the value-date handling. The test is to take one week, forecast it on Monday, compare the forecast with the actual on Friday, and state the variance and its cause, because if the variance cannot be explained, the forecast cannot be trusted. Further reading: Bank for International Settlements, payment system glossaries; ISO 20022 camt.052 and camt.053. Last reviewed 7 November 2026.