A company with a Rs 300 crore floating loan at benchmark plus 1.5% buys a three-year 8% cap for 0.9% upfront. What does it pay each year if the benchmark runs at 7%, 9% or 10%?

At 7% the loan costs 8.8%, at 9% and at 10% it costs 9.8%, which is the ceiling. Below the 8% strike the cap pays nothing and the company pays benchmark plus 1.5% plus 0.3% a year of spread premium. Above it the cap refunds everything over 8%, so the cost is 8% plus 1.5% plus 0.3%. In rupees that is Rs 26.4 crore at 7% and Rs 29.4 crore at either 9% or 10%, against Rs 34.5 crore unhedged at 10%.

Think of a prepaid electricity plan where you pay the market tariff each month but the supplier refunds anything above Rs 8 a unit, in exchange for a joining fee. Your bill still falls when the tariff falls; it just cannot rise past the ceiling. The cap is that refund: each year the benchmark fixes above 8%, the seller pays Neelgagan the excess on Rs 300 crore, and in years it fixes below 8% the cap pays nothing and the company simply pays less interest. It does not touch the 1.5% margin, which the bank charges whatever the benchmark does, and it does not refund its own premium.

Add the three pieces that still have to be paid when the cap is working. The capped benchmark, 8%, plus the lender's margin, 1.5%, plus the premium spread over three years, 0.3%, is 9.8%, and no benchmark level can push the cost above it. At 10% the cap pays 2% of Rs 300 crore, Rs 6 crore, which is exactly the gap between the Rs 34.5 crore unhedged bill and the Rs 29.4 crore all-in. At 9% it pays Rs 3 crore. At 7% it pays nothing, and the company has paid Rs 0.9 crore that year for protection it did not use, which is what insurance costs in the years the house does not burn.

A swap to fixed would have removed the premium and the ceiling both, replacing them with one fixed rate; if that fixed rate were, say, 7.8%, the company would pay 9.3% all-in in every year, better than the cap at 9% and 10%, worse at 7%. The cap costs 0.3% a year more than doing nothing in every calm year, and that is the price of keeping the saving when rates fall; a borrower who is confident rates will rise should swap, and one who wants a ceiling but expects to benefit from falls should cap. Say the simplification out loud: spreading Rs 2.7 crore evenly ignores the interest on money paid three years early, so the true annual cost is a little above 0.3%, and the cap's payout timing, usually at each reset, has been treated as if it matched the loan's interest dates exactly.

A treasurer's closing line names the residual risks: the cap is written on the benchmark, so if the bank re-prices the 1.5% margin the cap does not help; the cap seller must be good for the payout in the year it matters; and if the loan is repaid early the cap is left over, with whatever value the market gives it then.