I just left a CMO/MBS/ABS presentation by some guys from JPMorgan. There was the obligatory banter on whether or not there is a housing bubble, etc. But assuming there is a bubble, and it bursts, what happens when the CMO/MBS market realizes that homeowner defaults are highly correlated in the tails? Do they model how negative changes in economic fundamentals might cause the correlations of credit losses from individual investors to move toward unity? Presumably, your super senior AAA rated tranche could be a BBB or worse tranche in a matter of months or weeks, and in an environment where BBB minus AAA spreads would be higher than they are now. It seemed in the credit discussions no one was thinking about this possibility.I'm not saying it's going to happen, but I am skiddish about an LTCM kind of situation in this market stemming from banks, corporations, and broker dealers facing large losses on their mtge portfolios if we get a severe downturn in the US real estate market.

My understanding is that house prices do not crash like a stock market. The bubble will deflate itself slowly over years. The most recent US example is 1988-1994 or so. The market never collapsed. The effects on CMOs, ABS and their CDOs I think will be gradual too, not like the highly correlated defaults of the High-yield CDOs. As for ABS, they are only 40 % Home equity loans, the rest are Autos and credit cards that are not rate-sensitive.

Just thought I'd bring this up again. I originally posted this 2 years ago, and since I'm an arrogant ass, I figured I'd do an "I told you so."

Nice call bskilton81. I've been heavily involved in the subprime post mortem at one of the worst affected firms, and can't believe the sheer incompetence of apparently everyone involved in the MBS/ABS market. MBS/ABS are, it turns out, massively levered multi-name credit derivatives. So one might expect that they would be priced & risk managed using some fancy levy distribution-copula model, carefully calibrated to market prices. In fact, they weren't doing this. Nor were they using a simple delta hedged-lognormal-Black Scholes type model either. They weren't even using some 1960's, pre-Black Scholes, non-risk neutral statistical model of the market. No. Apparently they were using a deterministic model of the underlyings to price these derivatives! Of course, I don't have to explain to anyone on this forum why that is a monumentally stupid thing to do.Are there any quants working in the MBS/ABS field who can correct me on this? Is it possible that this whole field of derivative pricing has ignored quant fin techniques completely (don't see any chapter on MBS pricing in Hull, or Wilmott, or Joshi, etc)? Is this the main reason for the mess this market is now in?

QuoteOriginally posted by: bskilton81Just thought I'd bring this up again. I originally posted this 2 years ago, and since I'm an arrogant ass, I figured I'd do an "I told you so."Nice - 1 for 1. Now that you have accurately predicted the housing bust, what are your predictions for the new market? We will revisit them 2 years from now.

The question is: did he short the market?

QuoteOriginally posted by: AlanSNo. Apparently they were using a deterministic model of the underlyings to price these derivatives! Of course, I don't have to explain to anyone on this forum why that is a monumentally stupid thing to do.Is this the main reason for the mess this market is now in?No, I don't think the use of deterministic model is the cause of problem. In fact, I met people who did the simple calculation along the line of home price depreciation (HPA) can fall to -5% => conditional default rate (CDR) can rise to x% => fit to deterministic model (like Intex) and come up with the conclusion that every ABS CDO tranche of under AA should go underwater.... Fancy numerical technique is not a viable solution if the model is badly specified... Let's look at how people model prime mortgage products before moving on to subprime. For prime agency originated mortgage products (CMOs/ MBS), the modelling procedure is well established. Prepayment/ contraction risk are the main concern for CMOs. Credit risk is not that large (prime borrower + mortgage insurance from Fannie Mae/ Freddie Mac). CMO modelling can (and are often) quite fancy. First, you need to model the interest rate path with a horizon of up to 30-40 years. You can use as advanced interest rate model as you like. Then you need to model the relationship between prepayment and the interest rate. After then, you would come up with the cash flow for the underlying mortgage pool. At that point you can distribute the cash according to the waterfall and price the CMO for one possible interest rate path. Since this is path dependent, you need to run Monte Carlo for this, calculate the OAS.... I know people who employed similarly "sophisticated" OAS based solution for subprime but they still blew up. Default risk is the key here. Correlation, prepayment, interest rate modelling are all "noise" here. Since many fixed income people (portfolio managers, risk officers and senior managements) are quite comfortable about prime mortgages, they failed to re-examine the risk characteristics of subprime mortgages when loading their book with such product. The moral hazard of the whole mortgage origination/ securitization process and the failure of the rating agency aggravates the problem.QuoteIs it possible that this whole field of derivative pricing has ignored quant fin techniques completely (don't see any chapter on MBS pricing in Hull, or Wilmott, or Joshi, etc)? Hull, Wilmott, Joshi are good general books. But they can't cover everything. We need to go for more specialised books for this.

Quote .... Fancy numerical technique is not a viable solution if the model is badly specified... Completely agree that misspecified models were the core problem and that the ABS CDO traders, and their quants and risk managers, failed to examine the basic premise of their approach.But using Intex and even quite a pessimistic HPA assumption (e.g. -21% over 4 years) still priced the high-AAA ABS CDO tranches at around 95, compared to 30 where they are now valued. Compare this to the most senior tranches of a corporate CDO - if we used a deterministic macro-economic model to price these, they too would trade at par for anything outside of a severe depression scenario. But of course no one on a CDO desk would ever do that - they'd specify a distribution of future default rates and compute expected tranche payoff, as one would for any other derivative. It would then be clear that they have gamma risk and correlation risk, which no-one in the ABS business seems to have heard of.Quote Hull, Wilmott, Joshi are good general books. But they can't cover everything. We need to go for more specialised books for this. I can't think of any class of derivatives which don't even rate a mention in Hull. I wasn't trying to blame the books for the problem, just illustrating that securitization of retail debt appears to have ignored, and been ignored by, almost everyone in QF.