Karvanya Auto Loans securitises a Rs 1,000 crore pool into senior (85%), mezzanine (10%) and equity (5%) tranches, with excess spread of 2% a year. Cumulative losses reach 7%. Which tranches lose, how does excess spread change the answer, and what credit risks would you evaluate?

Without excess spread, losses of Rs 70 crore wipe out the Rs 50 crore equity and take Rs 20 crore, 20%, from the mezzanine. With Rs 40 crore of excess spread absorbing losses first, equity loses Rs 30 crore and the mezzanine and senior are untouched. The mezzanine now survives cumulative losses up to 9% and the senior up to 19%, as long as the spread is actually trapped and earned before losses arrive.

Picture a building flooding from the ground floor. The basement fills first, then the ground floor, and the top floor stays dry until everything below is under water. In a securitisationSelling a pool of loans to a separate trust that funds itself by issuing tranches of notes repaid from the cash flows of the pool. losses climb from the bottom: excess spread first, then equity, then mezzanine, and the senior tranche only after everything beneath it is gone. Cash flows the other way, paying the senior first. That opposition is the whole design.

Excess spread is earned on the balance still outstanding, not on the original pool, and the pool shrinks every year. 2% on average balances of Rs 875, 625, 375 and 125 crore is Rs 40 crore over the life, not the Rs 80 crore you get from 2% times Rs 1,000 crore times four years. If borrowers prepay faster, the balance shrinks sooner and the cushion gets smaller still.

Because the Rs 40 crore is earned over four years, and losses may not wait. If most defaults arrive in the first year, when only about Rs 17.5 crore of spread has been earned, the losses cut into equity and mezzanine before the rest of the cushion exists. And if the structure releases spread to the originator each month instead of trapping it in a reserve, it is gone for good. So read the documents for where the excess spread goes, and look at the default timing curve, not just the cumulative figure.

Four families, each with one question. Asset risk: how past pools of this originator defaulted and recovered, vintage by vintage, and how used car prices affect recoveries. Structural risk: the size and trapping of excess spread and any reserve account. Counterparty risk: whether collections could be mixed up with Karvanya's own cash if it failed, and who takes over servicing. Legal risk: whether the sale to the trust is a true sale that Karvanya's creditors cannot claw back. A tranche is only as good as the weakest of the four.