US unemployment may blow up stress test parameters

The rise in unemployment to 8.9% in April from its low of 4.4% in 2007 is more rapid than all but one post-war recession, and is closing in on the “worst case” scenario envisioned in the US bank stress tests. Since stress test loss calculations were based on statistical analyses of historical data, if unemployment soars into uncharted territory there could be some nasty surprises.

The modest reduction in job losses in April from a revised 699,000 to 539,000 indicated that the recessionary decline was losing momentum, but not by much. The public sector created 72,000 jobs during the month – presumably the first effect of President Barack Obama’s “stimulus package” -- so the loss in private sector jobs was still over 600,000. That, combined with upward revisions in job losses for February and March showed that even if downward economic momentum is declining, it is still very substantial.

The unemployment rate has already risen by 4.5% since its 2007 low. That’s the second largest rise in a single post-war business cycle, after 1979-82’s 5.2%, and with the jobless rate climbing rapidly the surge seems likely to break that record soon. The “stress test” worst case assumption of a 10.4% peak in unemployment and 1982’s post-Great Depression peak of 10.8% are also by no means unattainable by next year.

Loss assumptions in the stress test were based off data from past recessions. Since credit cards did not exist during the Great Depression, and the housing finance system was different (home mortgages suffered a peak delinquency rate of 50%) the regression analysis will necessarily have used data from post-World War II recessions.

Since the depth (in terms of maximum unemployment) of this recession will soon be matched by only one other post-war recession, and its severity (in terms of unemployment’s percentage increase) may soon be unparalleled, the US economic position may move “off the chart” of past data. That would invalidate the statistical analyses, greatly increasing the likelihood of unpleasant surprises in bank loss rates.

The stress test has reassured markets about the US banking system. However that reassurance may have been false, resulting from faulty conclusions drawn from a downturn more severe than any we have experienced.