Forecasting takedown

It’s a wonder that anyone has any faith in forecasting anymore. The failure of ratings agencies to see the storm brewing in subprime and economists to fully grasp the vulnerability of the financial system should be making cynics out of all of us. Paul Krugman devoted 8 page screens over at the New York Times explaining what went wrong with economists. I must admit, I stopped reading after this line:

As I see it, the economics profession went astray because economists, as a group, mistook beauty, clad in impressive-looking mathematics, for truth.

Bank of America Merrill Lynch analysts devoted a portion of their research to a glitch they see in forecasting corporate default rates. For high-yield bond and loan investors, default rates are a key component in trying to figure out whether it's time to snap up that incredibly risky CCC debt or exit quickly. If they're looking at S&P and Moody's forecasts it could be both.

Moody's forecasts defaults to peak at 12.6% in November 2009, and then
improve rapidly over the next 9 months, reaching 4.3% by next year, a level which
is below historical average of 4.8%. This puts the two major rating agencies on
the opposite ends of default forecasting spectrum. S&P is forecasting a 13.9%
default rate in a year from now.

That differential highlights the critical role of liquidity – the ability to refinance
maturing debt – in forecasting defaults. In our assessment, Moody's historically
had ignored such a variable and as a result during the credit bubble years
consistently overstated default forecasts. Partially as a result of such errors, in
August 2007 they switched their modeling methodology to incorporate implicitly a
liquidity factor. Incorporating liquidity factors into default models is very difficult as
no precise measure of liquidity exists. Unfortunately from a modeling perspective,
Moody's chose to use credit spreads.

While the inclusion of spreads lends to better explanation of defaults in sample,
such an inclusion raises logical flaws when the output of the default forecast is to
be used to forecast spreads. That is because a forecast of spreads is required to
forecast the Moody's default rate. In the above forecast of 4.3%, Moody's
forecasts spreads at 578 (vs. roughly 900 currently).

Circular thinking at its best.