Jamie Dimon Says They Didn't Model Massive House Price Collapse

Kevin Drum is shocked to find Jamie Dimon admitting that they weren't modeling a total collapse in house prices.  I'm shocked to find that Kevin is shocked.  That's pretty much the standard explanation--at least, a partial one--for why lenders became willing to take on so much risk.  Massive house price depreciation had pretty much dropped out of their models, which mostly focused on prepayment risk.

This is not quite as crazy as it sounds. For one thing, Kevin has truncated the quote a little bit; the version I read has Dimon saying "We didn't stress test housing prices going down by 40%." America had not had a sustained national decline in residential housing prices since the Great Depression.  So while local banks might need to model the risk of substantial price depreciation, banks glomming together national pools of mortgages figured this wasn't such a big problem--as long as you didn't think we were going to have another Great Depression.  And most regulators, commentators, economists, bankers, and ordinary folks thought we weren't going to have another Great Depression.

Indeed, we didn't.  It turned out to be a sufficient, but not necessary condition for a collapse in housing prices.

Even if they had put housing price implosion in their models, where would they have gotten the data to fine-tune their models?  It's not enough to say, "We should model a broad national decline in house prices;" you need some values for how many people will default when house prices fall.  The last time we had such a national collapse, mortgages were relatively short term debt instruments that didn't self-amortize.  We've had local bubbles since in places like New York and California.  But New York is definitely a bad model--it's a city mostly of renters in which co-ops frequently demand down payments of 25-50%, or even all cash.  California might have been better, but unlike a lot of places, it's a non-recourse state.  And so on. How well could Dimon have hoped to build a nationwide model off of a few local jurisdictions?

That's not to excuse the bankers for not trying; some allowance for the risk of a broad price decline would have been better than none.  But I'm not sure that it would have done much to alter their lending habits.  Going on historical data, the risk of a huge price drop within the average lifetime of a mortgage (which is less than ten years), would normally have been very small, and would have shown up in any probability-weighted model as a fairly trivial adjustment compared to the large risk that the mortgages in the pool would be refinanced. By the time it was obvious that the risk of a broadly falling market was very great, the bubble was about to pop of its own accord.  Indeed, even without such a model, Dimon pulled out of subprime, because he didn't need a spreadsheet to tell him that it was going to turn into a disaster.