‹ What Are Derivatives? Lesson 15 of 16
Contents Lesson 15 of 16

5 min read · professional

How did insurance on mortgages nearly end the financial system?

The 2008 crisis had many causes, and derivatives were one link in a long chain. This lesson describes that link factually — what the instruments were, what was done with them, and which specific decisions turned a housing downturn into a solvency crisis at the centre of the system.

The chain, step by step

Mortgages were pooled. Thousands of home loans were bundled into mortgage-backed securities, so that investors received a share of the pool's payments rather than exposure to any single borrower.

The pools were tranched. Those securities were sliced into collateralised debt obligations with a priority order: the senior tranche gets paid first and absorbs losses last; the junior tranche absorbs losses first. Senior tranches were rated AAA, on the argument that mortgage defaults in Florida, Nevada and Ohio were largely independent events, so it was implausible for enough of them to fail simultaneously to reach the senior slice.

Protection was written on the tranches. A credit default swap is a bilateral contract in which one side pays a periodic fee and the other pays out if a specified credit defaults. Economically it is insurance. Legally it was not, and three consequences followed: no requirement to hold reserves against expected claims; no requirement to have an insurable interest, so protection could be bought on debt one did not own; and no insurance regulator supervising the aggregate book.

AIG

AIG Financial Products, a unit of roughly a few hundred people in London and Connecticut inside one of the world's largest insurers, sold "super-senior" protection on a portfolio whose notional reached roughly half a trillion dollars, of which around $78 billion referenced multi-sector CDOs with substantial subprime mortgage content.

It held essentially no reserves against that book. The reasoning was internally consistent: these were the most senior claims on diversified pools of loans, rated AAA, and modelled as having a negligible probability of loss. The premium income was therefore treated as close to free.

The contracts contained collateral triggers. AIG had to post collateral if the market value of the referenced assets fell, and post considerably more if AIG's own credit rating was downgraded. That is a liquidity obligation that activates precisely when everything is going wrong.

In 2007–2008 US house prices fell nationally, the CDO tranches were marked down sharply, and collateral calls arrived in the tens of billions. AIG was downgraded on 15 September 2008 — the day Lehman filed — which triggered the remaining calls at once. It could not pay. The US government's support package for AIG ultimately reached approximately $182 billion.

What actually caused it

  1. A correlation assumption that was wrong for the scenario that mattered. Regional housing markets are largely independent in normal times. In a national credit cycle driven by common lending standards and a common interest-rate environment, they are one market. The AAA rating was an output of the independence assumption, not of a stress test.
  2. Writing insurance without reserves. Insurance regulation requires reserves for a reason. A CDS produced the same economic exposure while sitting outside that requirement.
  3. Opacity. The contracts were bilateral, unreported and uncleared. No regulator, and no counterparty, could see the aggregate. Nobody knew who was exposed to whom, which is why the system froze even where exposure was small — an unknown exposure is treated as a large one.
  4. A rating used as a substitute for analysis. Institutions bought AAA tranches without modelling the underlying loans, because the letters were treated as the analysis.
  5. Collateral triggers creating a spiral. Falling marks demanded cash; raising cash required selling; selling lowered marks. The trigger tied to AIG's own rating turned a solvency worry into a liquidity failure within days.

The instrument was not the problem

A credit default swap is a legitimate and useful risk-transfer contract: a bank holding concentrated exposure to one borrower can lay part of it off, exactly as the farmer laid off wheat. CDS contracts are traded today — standardised and reported to trade repositories, with index CDS largely centrally cleared and single-name clearing still partial, and with sovereign naked CDS restricted in the EU.

What failed in 2008 was concentration (one unit holding an entire market's tail risk), absent reserves, and invisibility. Reread the causes of Barings and LTCM and the same two words appear: leverage and opacity. Three completely different instruments; three failures of the same two things.

Try it now

  1. Put two series side by side across 2005–2010. The first is a US house price index, before inflation, quarterly, with 2010 = 100. Seven of its quarters are in the table below: find the quarter it peaked and the quarter it stopped falling. The second is the S&P 500, charted at full length under it. Measure its 2007 peak to its 2009 trough, then count how many quarters earlier the housing series turned.
Live API response: der2 us house prices 2005 2010
Interactive line chart: GSPC.INDX (MAX)
  1. Look up the definition of a "super-senior" tranche and write down, in one sentence, what has to happen to the underlying pool before it takes a loss. Then ask what assumption makes that sentence comforting.
  2. List which of the five causes above were properties of the contract, and which were properties of how it was used and supervised. The split is the lesson of this entire unit.