... how to spread risk, but they were not able to track risk. "

OK, it's clear they knew how to spread the potential damage around, by splitting each piece of subprime crap (SPC) up into fragments, then bundling ones of different origin together to make a new composite.

They imagined this mean the risk went away. On this basis, they constructed and sold more and more SPC, increasing the total potential damage without limit, whilst imagining that only a fixed fraction would fail at any time.

Now, _If_ it were the case that each piece of SPC was uncorrelated with the rest, then this would perhaps work. But crap is crap, and the crap is in fact highly correlated, both by economic conditions and market psychology.

So if enough SPC goes bad to switch economic conditions and/or market psychology, it drags the rest with it.

But before it went bad, the total potential damage had been increasing far beyond what institutions could stand -- and because the risk of each SPC was highly correlated with the others, it mostly all went bad at the same time (i.e. quite recently) -- and hence kneecapped the institutions.

They clearly understood jack shit about spreading risk (because they didn't) and therefore whether they could track it or not is irrelevant.

"I'm not saying everyone is not going to go on a little bit of a diet"

WTF does that mean?

it occurs to me in my comment above I conflated two distinct concepts: to spead risk, and to reduce risk.

The institutions thought they could reduce risk by spreading it.

They did spread it, but they didn't reduce it.

But because they thought (acted as if) spreading risk = reducing risk, they continued increasing the potential damage (by constructing and selling moer SPC), whilst spreading it around in the assumption it would reduce it risk/damage.

But since SPC failure is correlated, neither risk or damage was reduced. They merely spread the ever-increasing potential damage around -- so they could bring everyone down at once.

Good comments and a good story. I think there is also a strong case to be made against the rating agencies that blessed a huge number of these these SPCs (subprime crap) with AAA ratings - on par with US govt bonds. Then, of course, the bond insurers who signed up to guarantee these pieces of crap - without having near the capitalization to handle even a tiny fraction of the potential defaults. With these two safety nets, investment banks, insurance companies, pension funds, and every other investor in the world now had the paperwork they needed to justify investing in these things.

The fundamental error, as Paul Kinsler points out above, is the idea that default risk in any individual underlying mortgage is unique to that mortgage. This is the uncorrelated risk assumption, which basically says "sure, some of these loans might go bad, some of them always do, but we know how many and we'll price that in. There isn't anything that points to big slews of these things all turning to crap at the same time." Except that the quality of the underlying loans was astoundingly bad. Many of these loans were no qualification, no down, no income verification, and allowed the mortgagee to pull out loads of cash. The fact the the quasi-governmental Freddie and Fannie were originating and/or buying these toxic loans, plus the high ratings and insurance coverage, made it OK for everyone else to jump in with both feet - and lose their asses (or actually, our asses, since we're bailing out everyone involved).

Palmisano is one of the few CEOs that actually seems to think, instead of floating to the top using just the Peter Principle.

A. Following Scott McNealy around like Mini-me