I recall a conversation sometime in May 2001, when on the staff of the CEA, where the topic was how all the indicators were all pointing toward avoiding a recession. Indeed, using the oft-cited rule of thumb of two-consecutive quarters of negative GDP growth, there was no 2001 recession (although there was by this criterion a period of about 2 years, 2002-04, when there was). Consider this graph of GDP, by different vintages:
Figure 1: GDP in billions Ch. 1996$ SAAR. NBER defined peak and trough dates at dashed lines. 6/29/22 GDP calculated dividing nominal GDP by GDP deflator rescaled to 1996=100. Source: BEA via ALFRED, NBER, and author’s calculations.
The first reading of GDP at then end of April 2001 allowed people to breathe a sigh of relief – q/q annualized growth was around 2%:
Figure 2: Quarter-on-quarter annualized real GDP growth. NBER defined peak and trough dates at dashed lines. Source: BEA via ALFRED, NBER, and author’s calculations.
That positive reading remained true through the second and third releases. The advance reading for 2001Q2 was also positive. This remained true through the second and third releases, and only in Q3 did the advance reading go negative, and remain so. However, the advance Q4 reading (1/30/2002) was positive — so there was no two-consecutive-quarter string of negative readings.
Only with the advance 2002Q2 release — incorporating the annual revision — did we get consecutive negative readings, in this case three quarters, starting Q1 running through Q3. The annual revisions adjust the preceding 5 years of GDP estimates (see here).
However, every five years, the BEA undertakes comprehensive revisions, incorporating the latest data. This once again changed the contours of GDP – as shown in Figure 1 and Figure 2. As the most recent series show, there are no consecutive quarters of negative growth. And yet, the NBER declared a recession running from (peak-to-trough) March 2001-November 2001, 2001Q-2001Q4 (recession announcement November 26, 2001, expansion announcement, July 17, 2003).
For a previous discussion of GDP revisions, see here, and here. None of the foregoing should be taken to mean that one can’t use GDP to determine recession dates. In fact, Jim Hamilton’s real-time indicator of recession (see this post for extended discussion) does not rely upon the most recent quarter’s data, but the next to most recent.