Most small lenders I've seen get away with one LGD for the whole book. The better ones split it by LVR band, say 25% for loans under 80% LVR. Either way it gets rolled forward each year and nobody looks too hard at it.
A band tells you how much you lose, not where. On a secured loan the money comes back in 3 parts. First is what the borrower pays after default but before you take the car. Second is what the car sells for. Third is whatever you get after the sale, from chasing the shortfall or selling it to a debt buyer.
My PhD was on something similar, using US mortgage data. Each of the 3 parts behaved differently and had different drivers. In some cases modelling them separately gave a better answer than one blended rate. The paper is here.
Here are 2 lenders in the same band, both losing 25%.
Lender A gets almost nothing before the repo. B gets good payments early but almost nothing after the sale.
Problem is the split tells you where to look, not why. A's gap could be slow early collections, or a customer base that just can't pay once they fall behind. B's thin tail could be weak chasing, or borrowers with nothing left to chase. If it's your process, fix the process. If it's your customers, that's a pricing and who-you-lend-to question, and for B probably a case for selling the shortfall sooner instead of paying staff to chase it.
On one blended rate you cant even see the question. Either way the test is net benefit: whether more effort there acutally changes what comes back, after its cost.
Worth pulling apart on your own book. Happy to have a free call if you want to talk it through.
General information only, not advice on your accounts. Roro doesn't arrange or introduce finance and doesn't make credit decisions.