Not Quite a Recession

Today’s report contained some good news. The main reason that the final GDP number was weaker than predicted was the big drawdown in inventories. Without that negative contribution from inventories, real final sales grew at a robust 3.8% annual rate. Housing subtracted 0.6% from the annual real GDP growth rate, though that’s actually the smallest negative contribution we’ve seen in a year. Remember that new home construction has to be worse (on a seasonally adjusted basis) than it was the previous quarter in order to make a negative contribution to the GDP growth rate. Bigger exports and smaller imports each boosted the GDP growth rate by over 1%. Complain as you like about the Fed permitting such a big slide in the dollar, but at least it’s having the intended effect on aggregate demand through the international channel. Without the gains on net exports, real GDP would have actually fallen, making one worry about recent indications of a global economic slowdown. Real personal consumption spending also grew, though less than I had been expecting, given the presumed stimulus from the tax rebate.

The BEA also released its annual revision of earlier GDP numbers. Most noteworthy here is the adjustment to 2007:Q4 growth, which was initially reported at +0.6%, and has been revised down to -0.2%. One rule of thumb sometimes used is that two quarters of falling real GDP is characterized as a recession, according to which the -0.2% for 2007:Q4 and anemic +0.9% for 2008:Q1 do not quite qualify. I have developed an algorithm (described here and in more detail here) that looks at the full history of GDP data to refine that rule a little, based on a simple pattern-recognition procedure. Given the revisions in the data (and the great usefulness of having more than one quarter’s data to identify the most recent trend), I only use this to make a call as to where the economy was in the previous quarter. With the latest release of the 2008:Q2 GDP numbers, I’ve now calculated our recession indicator index for 2008:Q1, which turns out to be 38.4%. Based on a historical analysis of the algorithm, I would not declare a recession to have begun unless the indicator rises above 66%. So my current assessment is that a U.S. recession had not yet started as of 2008:Q1.

The plotted value for each date is based solely on information as it would have been publicly available and reported as of one quarter after the indicated date. Shaded regions represent dates of NBER recessions, which were not used in any way in constructing the index, and which were sometimes not reported until two years after the date.