Pacific Premium II

The discipline of watching is attribution. A file that has improved is not a file that has recovered — it is a file where something happened, and the year's work is to name what. Recoveries draw on three accounts: structural change, where the business itself now runs differently; purchased change, where a number was bought from someone at a price recorded somewhere else; and deferred change, where the improvement is simply a cost that hasn't been paid yet. Twelve months ago the bank spent a hard conversation buying this set of numbers. Before it rewards them, it should establish which account they were drawn on.

Go through them. The term-loan record and the inventory discipline are structural — a hired credit controller, declined contracts, a shorter cycle; behaviour changed, and the balance sheet followed. Half the DSO improvement is structural too. The other nine days were purchased: a 2% prompt-payment discount to a customer worth thirty percent of revenue costs about US$0.26m a year, quietly repaid out of the very margin line the strip is celebrating. That is not a sin — it is a price, and arguably a fair one — but it means the receivables book didn't speed up; the company is paying its biggest customer to make it look as if it did. And the fifth cell is deferred: capex at 0.36× depreciation means the recovery's third funder, after the customers and the margin, is the machinery — aging on the factory floor, appearing in no covenant, presenting its own invoice two or three years from now.

Option A's flaw is not that it is wrong about the year — it is that it is early, and it un-asks the question Issue 01 asked. Twelve clean months is necessary evidence; it is not sufficient. Restore the full US$3.5m now and the first US$0.7m of new headroom will be absorbed by the deferred capex the moment the MD stops deferring it — the line goes back to funding a gap, which is precisely the pattern the restructure existed to break. C makes the opposite error: it reads the discount and the capex freeze as concealment. They are neither. They are the normal financing of a repair year — acceptable, even sensible, provided they have an end date. Tightening against them would punish the borrower for the mechanics of doing what the bank asked.

The senior banker's read is B: keep the structure, and turn graduation from a favour into an instrument. Three tests, written into the review: DSO at or below 66 with the discount effect stripped out, held for two consecutive quarters — proof the cycle is shorter, not subsidised; capex back above 0.8× depreciation — proof the recovery has stopped borrowing from the machines; term loan past its halfway mark — proof the old imbalance is more than half retired. Meet them, and the line restores itself without another committee. The difference between this and A is not caution. It is authorship: the MD walks out with the restoration in his own hands, on his own numbers — which is exactly where a recovering borrower's incentives belong. A recovery is a hypothesis until it has survived one full cycle without borrowing from the next one.