Saturday, April 25, 2009

Stress Tests

The full White Paper can be found here.

Here is the Fed's (which itself is a misnomer as the Fed is not carrying out the tests, only supervising and commenting on the results thereof) "methodology":

The firms were asked to project their credit losses and revenues for the two years 2009 and 2010, including the level of reserves that would be needed at the end of 2010 to cover expected losses in 2011, under two alternative economic scenarios. The baseline scenario reflected the consensus expectation in February 2009 among professional forecasters on the depth and duration of the recession, while the more adverse scenario was designed to characterize a recession that is longer and more severe than the consensus expectation. The firms were also asked to provide supporting documentation for their projected losses and resources, including information on projected income and expenses by major category, domestic and international portfolio characteristics, forecast methods, and important assumptions.

And for their macroeconomic projections (empahsis added):

The baseline assumptions for real GDP growth and the unemployment rate for 2009 and 2010 were assumed to be equal to the average of the projections published by Consensus Forecasts, the Blue Chip survey, and the Survey of Professional Forecasters. The projections were based on forecasts available in February 2009 just before the commencement of the SCAP. The baseline scenario was intended to represent a consensus view about the depth and duration of the recession. The supervisors developed an alternative “more adverse” scenario to reflect the possibility that the economy could turn out to be appreciably weaker than expected under the baseline outlook. By design, the path of the U.S. economy in this alternative more adverse scenario reflects a deeper and longer recession than in the baseline. However, the more adverse alternative is not, and is not intended to be a “worst case” scenario. To be most useful, stress tests should reflect conditions that are severe but plausible.3
The assumptions for house prices in the baseline economic outlook are consistent with the path that was implied by futures prices for the Case‐Shiller 10‐City Composite index in late February and the average response to a special question on house prices in the Blue Chip survey. For the more adverse scenario, house prices are assumed to be about 10 percent lower at the end of 2010 relative to their

3 The “more adverse” scenario was constructed from the historical track record of private forecasters as well as their current assessments of uncertainty. In particular, based on the historical accuracy of Blue Chip forecasts made since the late 1970s, the likelihood that the average unemployment rate in 2010 could be at least as high as in the alternative more adverse scenario is roughly 10 percent. In addition, the subjective probability assessments provided by participants in the January Consensus Forecasts survey and the February Survey of Professional Forecasters imply a roughly 15 percent chance that real GDP growth could be at least as low, and unemployment at least as high, as assumed in the more adverse scenario.

Since the announcement of the SCAP in late February, the economy has deteriorated somewhat and professional forecasters have revised their outlooks for GDP growth and the unemployment rate in 2009 and 2010. New information on house prices suggests that the market’s expectation for house price declines is similar to what was anticipated in February. A large share of projected losses at banks are expected to be related to house prices, and the specified path for house prices in the more adverse scenario still represents a severe level of stress. Although the likelihood that unemployment could average 10.3 percent in 2010 is now higher than had been anticipated when the scenarios were specified, that outcome still exceeds a more recent consensus projection by professional forecasters for an average unemployment rate of 9.3 percent in 2010.

Footnote number 3 is important as is shows the hand of the Fed and provides all the information I need to evaluate the "rigorousness" of these stress tests:  Blue Chip forecasts since the 1970s? Using historical data during historical hinge-points like this is next to useless, and the "10 percent chance" is not taken from any probability distribution that could be useful.  Note also the time period "Since the 1970s" (which itself is vague).  Even if history was a useful guide, arbitrarily lopping off decades of data points is most unhlepful.   

In other words, the White Paper, and perhaps the tests themselves, are only meant as a public relations exercise and not a truth-finding mechansim.  Now (warning, rhetorical question ahead), why would the Fed need to conduct such an exercise if everything was copacetic? 

No comments: