2008: How AAA Became Junk

From 2859 on 31 October 2007 to 1269 on 9 March 2009, the NASDAQ Composite fell 55.6%.

The crisis did not begin in equities. It began in something that sounded extremely safe: bonds rated AAA.

The chain

To understand what happened, follow a single mortgage as it becomes a AAA bond:

1. Origination. A lender lends to a homebuyer, including subprime borrowers with weak credit.
2. Pooling. Thousands of mortgages are bundled into a pool and sold to an investment bank.
3. Tranching. The pool’s cash flows are sliced into priority levels. The senior tranche is paid first and absorbs losses last.
4. Rating. Because it is paid first, the senior tranche receives AAA — the same grade as US Treasuries.
5. Re-packaging. Mezzanine tranches from many deals are bundled into CDOs and tranched again, so the middle of a subprime pool produces yet more AAA.

Every step is defensible in isolation. The problem is that the whole chain rested on one assumption.

The assumption

Tranche safety depends on one thing: these mortgages will not default at the same time.

That has some logic to it. A default in California and a default in Florida look like independent events; unemployment in one state should not drive the other. Mix enough regions together and the aggregate default rate should be stable and predictable.

That is exactly what the rating models computed. And the historical data they used to estimate default correlation came from a period of continuously rising national house prices.

While prices rise, the assumption holds: a borrower who cannot pay can sell the house, clear the loan, and perhaps pocket a gain. Rising prices masked credit quality.

When national house prices began to fall, those apparently independent mortgages revealed the factor they all shared.

Correlations go to 1

This is the sentence worth carrying away: in a crisis, correlations go to 1.

Assets that are independent in normal times move together under stress, because a common driver sits behind them — here, national house prices; in LTCM’s year, the liquidity premium.

So:

Prices fall → defaults rise → junior tranches wiped out → losses reach the “safe” AAA tranches → institutions holding AAA (many banks and money market funds) write down assets → forced deleveraging and asset sales → prices fall further → …

Diversification did not eliminate the risk. It hid the risk inside a correlation assumption. When the assumption broke, all of it returned at once.

15 September 2008

Lehman Brothers filed for bankruptcy.

Until that day the market broadly believed in “too big to fail” — that any systemically important institution would be backstopped. Bear Stearns being arranged into a sale in March had reinforced the expectation.

Lehman was not rescued. What that day destroyed was not Lehman itself but the consensus about who gets rescued.

Every institution simultaneously began doubting every counterparty. Interbank lending stopped, the commercial paper market froze, money market funds faced runs. This was no longer an asset price problem — the plumbing of the financial system had blocked.

The NASDAQ still stood at 2180 the day Lehman failed. The real collapse came over the following six months: another 42% by March 2009.

What transfers

One: a rating or a label is not the risk itself

AAA was a model output, not a fact. When the model’s key input — default correlation — was estimated from an unusual historical period, its output was only valid inside that period.

Whenever you see a “safe” label, the questions are: who assigned it, on what assumptions, and under what conditions do those assumptions fail?

Two: diversification depends on correlation stability, and correlation is least stable under stress

If ten holdings sit on one driver, you own one thing ten times, not ten things. Real diversification requires different drivers, not different names.

Three: leverage decides whether you earn the right to be proved correct

House prices did eventually stabilise and recover, and many written-down assets were not ultimately worthless. Highly leveraged institutions never found out — margin calls and capital requirements forced them out along the way.

Which is exactly LTCM again: whether a judgement is correct and whether you survive to collect on it are two independent questions.

This is a historical review written for risk education. It is not investment advice and makes no inference about any current market conditions.