When not to hedge, and the policy that replaces guessing

A hedge that did everything right

The season before LKM-9126, Ardessa took PO LKM-8804 from Larkmoor: 30,000 units at USD 11.40, value USD 342,000, payment due 20 March.

On 12 September the treasurer covered it in full. Rhosden's spot was 47.60, and its forward for value 20 March, 189 days out, was 49.26. He sold 342,000 dollars forward at that rate. Treasury policy at the time said cover confirmed orders in full, and he followed it.

On 6 December the second fit sample failed on the shoulder, in one of the three colours. The block was corrected and the other two colours ran. But the third colour could not make the delivery window, and Larkmoor cut it: 9,000 units, USD 102,600.

Nobody did anything wrong. The technologist caught a real fault at the right stage. The buyer cut the colour rather than accept a bad fit. And the treasurer then had to close the surplus, because a forward with no order behind it is not a hedge. It is a currency bet the factory entered by accident, and leaving it open would have been the real error.

The sela had weakened sharply in early December. Spot on 6 December was 49.90, and Rhosden's forward for 20 March, now 104 days out, was 50.87. Ardessa had sold 102,600 dollars at 49.26 and had to buy them back at 50.87.

Cost of the close-out: 165,186 sela.

By the middle of January the sela had given the whole move back. Left alone, the position would have been fine. But it could not be left alone. That is the point: the loss was crystallised by the order cut, not by the market. A currency does not move in a straight line, and an over-hedge is a position you are forced to close at a moment you did not choose.

The cure is a ratio, and it comes out of your own files

The treasurer's mistake, if it was one, was hedging 100% of a quantity that was 100% certain in the system and less than that in reality. So he did the only sensible thing. He went and looked.

The average cut is 6.5%, and the worst is 30%. Hedge at 70% on confirmation and you can never be over-hedged on this history: 70% of 342,000 is 239,400 dollars, which is exactly what was left after the 30% cut. Hedge at 90% and two of these eight orders would have needed a close-out.

Two honest warnings, because the tidiness above is arithmetic, not foresight. Eight orders is a small sample. And the ratio protects against cuts up to 30%, and not one unit further. Also, 70% leaves a third of the exposure open. That is a deliberate decision to carry currency risk, not the absence of one.

The step that makes it work is topping up. Raise the cover as the order becomes more certain: at fabric commitment, at cutting, at final quantity confirmation. The uncovered share then shrinks as the risk of a cut shrinks.

Three cases where not hedging is the right answer

Hedging is not free, and it is not always proportionate.

The exposure is smaller than the trouble. Ardessa's development programme with a new customer is USD 18,000 of samples and first production. After netting off the imported fabric, the net exposure is about USD 7,400. A move as large as the one that hit LKM-9126 — 6.3% — would cost USD 466. Rhosden's minimum forward ticket is USD 25,000, and it charges a documentation fee on every contract. There is nothing here worth covering. Covering it would cost more in attention than the whole risk is worth.

The natural hedge already did the work. A factory whose dollar purchases and dollar sales are close in size and close in date has very little left to cover. Adding a forward on top uses credit line and produces paperwork for a leftover.

The cash flow is not certain enough for the instrument. Forecast business must never be covered with a forward, for exactly the reason LKM-8804 shows. If you want to protect a forecast, that is what an option is for. It lapses quietly.

There is a fourth case, and it is a judgement rather than a rule. A forward uses the same bank credit line as your working-capital facility. If you are near the limit of that line — and course 13.4 is about the days when you are — the right answer may genuinely be to carry the currency risk and keep the borrowing capacity. Say so out loud when you decide it, so that it is a decision and not an omission.

Act as a commercial manager in a garment export factory who has been burned by an exchange rate before. I need two things: a currency clause I can put in front of a buyer, and a quoting rate for the coming season. Facts: my functional currency is [HOME CURRENCY]; I invoice in [INVOICE CURRENCY]. Today's spot is [RATE]. Over the last [NUMBER] months the rate has moved between [RATE] and [RATE]. My cost base per unit is [AMOUNT] in the invoice currency for imported materials, and [AMOUNT] in my home currency for everything else. My target margin is [PERCENT]. My payment terms are [TERMS], and my quotations are valid [NUMBER] days. My main competitors are in [COUNTRIES]. Do the following. First, work out what share of my cost base is exposed, and tell me exactly how much of any devaluation would ever reach me, as one division I can repeat each season. Second, draft a currency adjustment clause in plain contract English: a band, a sharing rule, a named rate source, a time of day, and — most importantly — the DATE it measures on, with an explanation of why the payment date and not the invoice date. Third, propose a quoting rate for the season, show what it does to my quoted price in both absolute and percentage terms, and state honestly what share of a historically typical move it would absorb. Fourth, tell me what that quoting rate costs me competitively against the countries I named, and frame the decision as how much business I am willing to lose in order not to lose a margin. Fifth, give me the four lines of a one-page currency policy I could take to my board. Do not tell me where the rate is going.

The one-page policy

Almost every currency argument inside a factory is really an argument about a policy nobody ever wrote. Here is what belongs on the page. None of it needs a treasury department.

1. The functional currency, stated in one line. Everything else follows from it, and lesson 3 shows what happens when it is left vague.
2. The quoting rate: who sets it, and when it is reviewed. Ardessa now sets a rate for the season deliberately below spot: 44.50, against a January spot of 46.00. The local cost line is then costed at USD 5.1236 rather than USD 4.9565, and the quotation becomes USD 12.02 rather than 11.85. That is 1.4% dearer. On this order, that buffer would have absorbed 51.7% of the move that actually happened. It is not free: at 12.02 against Vantorn's 11.00, Ardessa loses more orders. A quoting rate is a decision about how much business you are willing to lose in order not to lose a margin. It should be taken by the person who owns both.
3. The hedge ratio and its step-ups, set from your own order-cut history, as above.
4. Which instruments are allowed, and which are forbidden. A workable rule: nothing that can lose more than the exposure it covers, and no forwards against forecast business.
5. Who may deal and who confirms — and that they are two different people.
6. That the currency result is reported on its own line, separately from the operating result.
7. That the hedge paperwork is done on the day the deal is done. IFRS 9 requires a hedging relationship to be formally designated and documented at the start. There is no way to decide afterwards that a contract was a hedge. If the document is written in March for a deal done in February, the accounting treatment is simply not available, whatever the commercial reality was.

What to take away

A forward is an obligation. So hedge the quantity you are confident of, not the quantity on the purchase order, and set that confidence from your own history of cuts. Not hedging is legitimate when the exposure is small, already matched, or not certain enough for the instrument. It is not legitimate when it is simply what happened. And write the policy down. The alternative is that every order becomes a fresh argument, decided by whoever in the room feels most strongly about a currency none of them can predict.