How Big Must a Bubble Be to be Dangerous

Robert Waldmann

This is a brief follow up on my post on measuring bubbles. The amount lost due to sub-prime mortgages in default is tiny compared to the damage done. How large can the ratio of damage to losses be ?

My answer is it can be infinite — that it is possible for irrational investing to destroy the financial system, even if the financial system looses nothing on average when reality bites.

Let’s imagine a case in which big money center banks secretly bet each other tens of billions on the flip of the coin. Average gains and losses must be zero. However, if counter-parties don’t know who bet on heads and who bet on tails, the financial system will seize up as the solvency of all the gambling banks is in doubt.

In the real world, the coin flips were called CDSs. It was known that investment banks and hedge funds had huge positions in CDSs, but it wasn’t known who was long and who was short. When the price of CDSs written on CDOs made up of MBSs shot up there were winners and losers. However, unless and until a firm went bankrupt it was unclear who the winners were (after bankruptcy it is clear that the winners are lawyers and everyone else loses on average).

That’s enough to disrupt finance and cause a huge worldwide recession. I think that’s what did it. Not the average loss but the immense variance of losses across banks.

It’s true that banks lost on average. Partly the lost to some hedge funds. This hurts the system because hedge funds don’t provide financial services. However, I think the banks could have failed in their role in the economy by making stupid investments even if their losses had added up to zero.