By the way, anyone with a basic stats degree can understand that mortgage defaults are not independent. It’s naive and incorrect to assume that large banks don’t have multi varient stress test models, but Rubin just ignored the results.

Citi Insider

Correlation is a bitch…

I have a friend who works for an investment research company. She is a mathematician and was involved in some of math models for these securities. I asked her how they could’ve thought that mortgage failures were independent events when once real estate prices start dropping defaults lead to more defaults. She said “well, now we know that they are correlated, but at the time everything pointed to them being independent…”.

Of course as a CEO he should’ve known better, and if you make a stupid decision on the job, you should be accountable for it.

Their problem with this logic was that it only works when each of the events is independent. But under certain conditions – falling real estate market, resetting interest rates – multiple defaults are likely to occur simultaneously. The defaults cause foreclosure that causes further decline in prices that causes more defaults.

Another problem, not related with the model was the SEC decision of 2006 to apply mark-to-market rule to CDOs. This caused greatly inflated value of CDOs (and hence banks’ assets) during good times, but it caused the decline in market value of CDOs during bad times. As the value of CDOs declined – often to the point well below the potential value of profits generated by the majority of mortgages that still worked – so did the total value of banks’ assets (on paper). Since mark-to-market requires banks to put more money in reserve to compensate for the losses in assets’ values, the fear of future losses caused banks to sell CDOs causing further decline in their value even if they’d have preferred to keep them and collect interest on still-working mortgages. As a result, CDOs are now selling at pennies on a dollar even though the vast majorities of mortgages don’t default. Remember, most CDOs don’t contain just sub-prime, there are prime mortgages there as well, but as nobody can trust AAA rating of these securities, all of them are selling for rock-bottom prices. A bank may have a lot of cash, but it has to keep it in reserve because it has to keep certain ratio of assets to loans. It also needs to report declines in resale value of CDOs as “loss” every quarter causing bad earnings reports and decline in confidence.

Also, there are credit default swaps which act as “insurance” only with no requirement for insurer’s having adequate capital to cover losses. Also, anybody can buy credit default swaps on assets held by any bank so it’s a little like 10 people being able to buy insurance on your house. These 10 people then would profit if something happens to this house, but without fear for the loss of the house. So when a lot of people buy credit default swaps against mortgage-backed securities held by, for example, Citibank, the value of mortgage-backed securities drops as well. This gives lots of ammunition for “bear attacks” on bank stocks – short, buy a lot of default swaps, continue shorting. Something facilitated by another “brilliant” SEC decision made in 2007 – elimination of the uptick rule. Of course, when they try to collect, the insurer may not have the money, so this would cause problems for insurers as well.