Why LTV ratios aren’t always a good default predictor
John Carney responds to my post about his anti-CRA crusade this morning:
That deal has a ridiculous LTV. You go wrong by calculating only the bank loan, and not the total financing. A slight drop in home values puts this buyer underwater, which hugely increases odds of default. LTV is meaningless for predicting defaults unless you are considering all the debt that goes into the financing.
This is simply not true. LTV is useful for predicting defaults because it gives an indication of how much home prices need to fall before the homeowner is underwater. But LTV is not a default predictor on its own — it can only be useful when combined with a second key variable, which is the ratio of mortgage payments to prevailing rents.
During the housing boom, nearly all houses were more expensive to buy than to rent: your mortgage payments would be higher than it would cost to rent the same place. (Or, to put it another way, if you immediately rented the place out, you’d be losing money every month, although you might be making money on a mark-to-market basis if the value of the home was rising quickly.)
When it’s cheaper to rent than to buy, a huge amount of work is done by the homeowner’s equity: it’s the main economic incentive to keep current on a non-recourse mortgage. On the other hand, if it’s cheaper to buy than to rent, the amount of equity that a homeowner has is pretty much irrelevant. If the borrower defaults and loses his home, his monthly costs go up rather than down — in Carney’s example, from $550 a month to $750 a month.
Paying $550 a month and owning your own home is clearly superior in every way to defaulting on your mortgage, ruining your credit, and renting for $750 a month instead. As a result, when a homebuyer is saving money every month a result of buying his home, LTV is pretty much useless as a default predictor.
I’m on the board of a financial institution which helps underserved poor New Yorkers buy homes — in our case, normally apartments in HDFC (Housing Development Fund Corporation) buildings which have severe restrictions on resale and which often cost about $20,000. It’s hardly surprising that our default rate on those loans is basically zero: our borrowers might have high LTVs, but they’re managing to live in Manhattan for a few hundred dollars a month, and you can’t beat that deal.
Carney’s example was along similar lines: it was all about helping first-time homeowners into low-income housing. Trying to extrapolate from that to the speculative subprime bubble is ludicrious — not least because most subprime borrowers were not low-income at all.
Carney says that I would “make a terrible mortgage lender” — actually, no. Look at the default history on mortgages issued by CDCUs around the country, and you’ll find it’s very good indeed. Partly because all these loans are scrupulously underwritten, and are generally kept on the lender’s books. There’s just no comparison to the toxic assets being churned out by subprime originators during the boom, and I can’t for the life of me work out why Carney insists on trying to draw one.