Last entry was on Nov 30, 2007. Now that I have gotten myself settled, and a rhythm in life has returned -- I should probably be in a position to post more frequently.
At the heart of the present crises that we are three factors: (i) default fears (ii) draining of liquidity (iii) industrial downturn. What has most gotten attention in popular press is default fears -- with default in household mortgages that has captured popular imagination. It is important to thus get briefly acquainted on how mortgage defaults (MD) are understood in most pricing or analytics.
MD is defined, in general, as when both of the following events occur: (i) unable to pay 3 consecutive monthly payments (ii) the equity in the home diminishes to leave no prospects of refinancing. Further, MD rates have a following shape. At first they rise, and then really spike up and then they decline -- graphed over years of mortgage existence.
Typical mortgages made out in the subprime market have been adjustable rate mortgages (ARMs) -- where the coupon is fixed for a while and then begins to reset periodically. Even the 30 year ARM in the regular markets are "5 x 1" (fixed for 5 years, and resets annually). In contrast, the 30 year ARM in a subprime markets are "2 x 28" (fixed for 2 years, resets semiannually for the next 28 years). Typically this reset is based of some pre-agreed index that tracks the cost of money, conditional on borrower quality, across the economy. So, if the generic interest rates begin to rise in the economy -- associated payable rates by the mortgage borrower are reset at a higher rate. What makes things worse is that most of these ARMs have an early "interest only" feature. Resultantly, as the "interest only" period finishes off -- the mortgage borrower has to start paying off for the principal component as well.
One of the critical issues every mortgage lender worries about is the prepayment of the lent amount. Typically, most of the pricing is done with some sort of prepayment model. The prepayment is a function of (i) available refinancing (ii) housing turnover (iii) curtailments (paying more every month than obligated) (iv) defaults. Typical modeling efforts of prepayment are applicable to a tranche of loans. As opposed to modeling prepayment for each loan.
Another layer of complication is to wonder about how to model the extent of loss that occurs as a mortgage defaults.
Now imagine writing securitized tranches that are sold which contain slices of mortgage backed securities -- i.e., CDOs on MBSs. It is not surprising then that we end up having this situation.
In essence, things get hairy pretty fast...