September 03, 2008
A Confidence of Crisis
"At least I mean what I say -- that's the same thing, you know, [as I say what I mean]" Alice in Wonderland.
In his semi-annual testimony to Congress in July 2007, the Federal Reserve chairman Ben Bernanke, said "some estimates are in the order of between $50 billion and $100 billion of losses associated with subprime credit problems". Note, he did not say that he personally had arrived at this estimate but did not subsequently disavow or clarify the numbers.
Obviously Dr. Bernanke meant what he said, but did he say what he meant?
What nice round numbers these are! $50 billion to $100 billion. And this was in the now-long days when $100 billion genuinely was a lot of money. So obviously $100 billion was a worst case - wasn't it?
Put another way, the chairman estimated losses as a result of subprime at $75 billion plus/minus $25 billion or 33%. Ignoring for the moment that we (and Dr. Bernanke) know the distribution of credit losses is not normally distributed, the upper limit ($100 billion) is only 1 standard deviation from the mean.
Hold on a minute, regulators (including the Fed) require that banks, under Basel II, set aside capital to cover 99.9% of credit losses or over 3 standard deviations from the mean! What is going on?
In August 2008, the Bank for International Settlements reported* that "an aggregate $503 billion of assets [has been] written down by banks and brokerages since the start of the credit crisis in 2007 [with more to come]". Remember that this truly huge number covers only losses in the financial markets and does not include losses due to subprime in the rest of the economy, which Dr. Bernanke was referring to.
Could the half a trillion dollars, of financial markets losses, possibly be within the 3 standard deviations of confidence in the estimate alluded to by Dr. Bernanke?
Who better to know than the chairman of the Federal Reserve, who has access to dozens of economists, years of historical data, untold giga-hertz of computing power to run gigantic simulations and access to the latest unpublished information on mortgage default rates? Surely the Fed could do better than +/-33%? Moreover, regulators expect a bank holding a single CDO to estimate potential losses to a much higher confidence level, without the benefit of such overwhelming inside knowledge and expertise.
Why did Dr. Bernanke not say something like, "assuming a 'fat tailed' distribution of credit losses, some estimates are in the order of $1 trillion of losses to a 99.9% confidence interval"?
There are two reasons why he may not have said something like that.
First, they may just not have crunched the numbers, which is highly unlikely given the professionalism of the Fed's economists but if true would be a dereliction of duty.
The other reason is that Dr. Bernanke is definitely no fool and knows that making any comment containing such huge numbers would spook the natives. The media would run with the "big number", despite his caveats, and he would forever be branded as the "man who created the great 2007 stock market crash" - definitely a career-limiting move.
Bernanke faced one of the greatest dilemmas facing all risk managers - how do I get across the potential impact of a particular risk to an audience who doesn't understand what I am saying?
It is extremely difficult to communicate estimates of risk.
There are multiple assumptions behind all estimates and these assumptions are themselves often difficult to explain to non-specialists. The models that crank out the estimates are complex and based on theories that require a PH.D to fully comprehend. The data behind the models is also suspect, often combining apples with oranges (prime with subprime) to give an unappetizing mess, but it is the best we have.
But that is what risk managers under the regulations promulgated by Dr. Bernanke and his fellow central bankers have to do, day in - day out. Risk managers have to be very careful to say precisely what they mean, but this is often a thankless task, as the listeners only hear what they want to hear.
Maybe next time, when beating up on a luckless risk manager in some financial institution, regulators may reflect on Dr. Bernanke's dilemma and back off a bit.
Or maybe that is more Alice in Wonderland. Curiouser and Curiouser!
* BIS Quarterly Review September 2008
Posted by pjmcconnell at 05:30 AM
| Comments (0)
