The 90-day mismatch

An exporter runs on 90-day trade terms. Its revolving credit facility tests covenants quarterly, on calendar dates. Two clocks, wound at different times—and they are almost never in the same conversation at the same moment.

Walk the cycle once. In January the exporter draws on the facility to fund raw material purchases for the season. Through February the goods are produced and shipped; the receivables are booked but nowhere near due. The covenant test lands in March: utilisation at its peak, receipts not yet arrived, and the leverage ratio prints at its worst point in the entire cycle. In April and May the cash comes in, the facility pays down, and the ratio looks pristine—but nobody tests in April.

Now the two readings. The bank's credit team sees a covenant breach, or a near-miss, on a facility that was presumably sized correctly—so either the business is deteriorating or the client is over-trading. The client sees a bank that has financed three of these cycles already and still panics at the same point in every one of them. Each side finds the other's reaction unreasonable, and each side's reaction is exactly what its own information predicts. The covenant result is a fact; the cycle explanation is a story; and in a credit file, facts outrank stories. Unless someone reframes it, the paper version wins—at the cost of a little more goodwill each quarter.

The underlying error was made months earlier, at the drafting table: the facility was structured for a generic borrower with a smooth cash cycle, not for this borrower with this cycle. That is the structural failure mode described in the five dimensions—the conversation defaulting to the most familiar structure rather than the most appropriate one. The fix is correspondingly unglamorous. Test on the dates that mean something: after the receipts land, not before; or replace the point-in-time ratio with a measure that respects the cycle, a rolling average or a seasonal utilisation schedule. One adjustment to covenant timing routinely resolves what had been escalating as a creditworthiness problem.

The judgment question—and the reason this note is not simply "banks should relax"—is telling a benign mismatch from real deterioration, because both print the same bad ratio in March. Three checks separate them. Does the cycle repeat? A timing problem looks the same every year; deterioration gets worse each year at the same calendar point. Are the receivables behind the drawn position of the same quality as last cycle—same buyers, same terms, same ageing—or is the March position now propped up by slower, more concentrated names? And does the cash actually arrive in April, in full, the way the ninety-day habit says it should if nothing heroic is assumed? A client who can answer all three from their own records has a structure problem, and it is the bank's to fix. A client who cannot has handed the bank a genuine early warning—on a facility whose covenant dates were, for once, doing their job.

Before repricing, before declining, before another round of reassurance in either direction: ask whether the covenant date is the problem.

In the Banking Judgment Lab this note becomes a covenant-calendar exercise, worked with numbers month by month.