Over the past month, I , citing Feldstein, have said that if one looks at available information on monthly GDP, available from estimates of MacroAdvisers, that output declined within the first quarter of the year, even though as standardly reported GDP was higher in QI overall than it had been in the last quarter of 2007. But, as it turns out, there is some ambiguity to the question.
The estimates do show GDP falling in February, by a hefty 10.1% anualized. But the numbers for January and March are up. To net out the three months, one must split hairs. The positive numbers for January plus March are just slightly greater in absolute value than February’s negative 0.9 (monthly). So the net is up? Not necessarily.
We are trying to figure out the change within the quarter, from beginning to end. Technically, that means from January 1 to March 30. But of course even Macroadvisors doesn’t report daily or weekly estimates. Estimated total real GDP in the month of March was just slightly above total real GDP in the month of December. So again the net is up? The most precise measure of the change between January 1 to March 30 is the change between the December-January average and the March-April average. That is a tiny negative number: GDP fell by an estimated $28 billion within the first quarter (in year-2000 $). And April is so flat as to be essentially zero.
I think I am sorry I brought the subject up.
It would in any case be a mistake to make much of these numbers. The reason the Commerce Department’s Bureau of Economic Analysis doesn’t report monthly numbers is that the data are so unreliable, and subject to revision. For anyone who needs some sort of estimate of monthly GDP, as we do on the NBER Business Cycle Dating Committee as an input into our thinking, this is what we have to go on. But one sees here yet another illustration as to why the BCDC waits a long time, until all the data are in, before declaring a recession.
*** Comments can be posted at http://www.rgemonitor.com/us-monitor/bio/660/jeffrey_frankel . ***
I’m not certain that I understand your post. Macroadisors indicates that monthly GDP peaked in January 2008 and e-forecasting indicates that monthly GDP peaked in March 2008.
Is the NBER business cycle dating committee using macroadisors or e-forecasting or both data for monthly GDP cycle peaks and troughs?
No private firm has an official status in the NBER Committee’s deliberations. But we look at Macroadvisors’ monthly GDP series, along with other things.
— JF
Thank you for the response. The Macroadvisor monthly data linked upthread is now presented through April not February: hence my initial uncertainty.
FYI, Menzie Chinn at EconBrowser has been consistently logging the five key variables used by the NBER BCDC to determine cycle peaks and troughs.
From Chinn’s series, real retail/wholesale sales peaked September 2007, monthly real GDP (macroadvisors only) peaked January 2008, industrial production peaked January 2008, private nonfarm payroll employment peaked December 2007 and personal income less government transfers is basically flat since November 2007.
perhaps I am stupid, but why is the comparison to Q407 the salient one? Seems like you would compare to Q107 to remove seasonality to see if the economy is really growing much year/year. A Q/Q view could make you look like one quarter is bad while others are extremely good as christmas hits/back to school/etc…
[...] Frankel revisits recession. After Mankiw, even Frankel switches off the comments [...]
[...] necessarily register a negative change q/q even if there is a 3 month changes in monthly GDP; see Jeff Frankel’s discussion.) The first principal component peaks earlier than both the GDP measures, in October of 2007, [...]