July 29, 2022 | 13:22
What Recessions Want
What Recessions Want
U.S. real GDP contracted 0.9% annualized in 2022Q2 after contracting 1.6% in Q1. This satisfied the popular definition of a recession, which is two consecutive quarters of negative growth, one that’s often mentioned in the media. However, the task of dating business cycles, and thus officially calling recessions, falls on the National Bureau of Economic Research (NBER). It’s up to a committee composed of eight esteemed academic economists to determined when “a significant decline in economic activity that is spread across the economy and that lasts more than a few months” is occurring or has occurred.
There are three criteria for this determination; a recession requires depth (must show a “significant decline”), diffusion (must be “spread across the economy”) and duration (must last “more than a few months”). This official recession definition and the NBER’s determination methodology are more robust than the simple ‘two consecutive quarters of negative growth’ rule. The popular rule would not have flagged the start of the recessions in 1973/75 and 2007/09 along with the entire recessions in 1960/61 and 2001 (Chart 1).
The NBER measures the length of economic contractions and expansions in months. For example, there have been 11 recessions in the post-WWII period (since 1950) (Table 1), and they’ve averaged around 10 months. All of them satisfied the depth, diffusion and duration (‘3 Ds’) criteria. Note that the pandemic recession (2020) lasted just two months. This was the shortest recession in history (since 1874), which clearly came up short on the duration criterion. For the NBER, the extreme depth and diffusion of the downturn overwhelmed its brevity.
With monthly business cycle dating, the NBER uses a set of monthly economic indicators to identify turning points. The set includes real personal income less transfer receipts, payroll employment, real personal consumption expenditures (PCE), real business sales (manufacturing, wholesale and retail), household-surveyed employment and industrial production. While all six elements matter, the NBER acknowledges that the first two, personal income and payrolls, carry more weight than the others.
Why not just use GDP?
First, from a technical consideration, recession dating is done on a monthly basis and GDP data are quarterly. However, GDP is used to translate the monthly turning points into quarterly versions, along with quarterly averages of the monthly indicators. Most monthly peaks and troughs translate to the quarters in which they occur. Exceptions can take place when the descent from the peak or the climb from the trough is so steep that it flips the whole quarter from a turning point to a first full quarter of contraction or expansion. For example, while February 2020 was the peak of the last cycle, the quarterly peak was 2019Q4 because March 2020 was abysmal.
A potential problem with GDP is that it’s possible to get a negative overall result without there being widespread weakness in the economy, if there’s a large-enough downturn in a specific region, sector or industry. This would give a false signal of recession. Relying on an array of broad economic indicators avoids this problem.
Another issue with using GDP as a sole recession signal is that all economic data are susceptible to revisions. Second quarter GDP growth of -0.9%, which is an advance figure based on incomplete and thus estimated inputs, is scheduled for potential revision at the end of August and again at the end of September. And, then again, as part of next year’s annual benchmark revisions. This year, in September, the annual revisions will cover five years worth of GDP data up to 2022Q1. This is where statistical issues like chronic weakness in first quarter GDP gets addressed (…hmm). It’s also where the competing expenditure and income approaches to GDP can be reconciled.
The early-released GDP data are based on expenditures, while the income method (called gross domestic income or GDI, not to be confused) is released with the first revision. Theoretically, they should generate comparable growth outcomes and trends because one person’s spending is another person’s income. In Q1, real GDP was -1.6% annualized or +1.6% y/y, while GDI was +1.8% annualized or +4.7% y/y. The current discrepancy between the levels of GDP and GDI is the largest on record (since 1947). Note that the NBER uses the average of real GDP and real GDI in its quarterly business cycle dating, which was unchanged in Q1 (Chart 2). Consequently, recent GDP is primed for some major upward revisions come September. There’s a good chance that the first part of the ‘two consecutive quarters of negative growth’ could be revised away. And who knows what will happen to the second part amid its first revision at the end of August.
So, is it recession?
While the popular definition says “yes”, the NBER’s definition says, “not yet”. And, while the cycle dating committee tends to wait for revisions to work their way through before officially sanctioning any turning point, we can track what they’re considering to assess whether we’re likely in a recession. Among the monthly economic indicators mentioned above, we’re looking for net declines that continue for at least a few months (there’s an adage about three months making a trend), which would indicate that they’ve probably peaked.
Payroll employment expanded by 372k in June, running in the 365k-to-400k range for the past four months (Chart 3). In the previous four-month period, average payroll growth was above 600k, so the trend is slowing. But the recent readings remain historically strong; monthly percent changes are more than 50% above their long-run average. If you need one fact to point to as to why we’re not in a recession, this would be it. Indeed, Fed Chair Powell pointed to it often in Wednesday’s presser. However, household-surveyed employment dropped in June for the second time in three months (May’s level is the recent peak so far). Even Powell acknowledged that this could be a harbinger of further slowdown in payrolls. The next set of jobs data will be released on August 5.
Real personal income excluding transfers decreased 0.3% in June, after increasing 0.2% in May (Chart 4). May’s level is the recent peak. Meanwhile, real PCE rose 0.1% in June after falling 0.3% in May (peak so far). Real goods spending also inched up 0.1%, after peaking last year as consumers are shifting back to services and also reflecting supply constraints.
Industrial production dropped 0.2% in June after inching up only 0.1% in May (peak so far) (Chart 5), with the heavy-weight manufacturing sector down at least 0.5% in the past two months. Meanwhile, real business sales dropped 0.9% in May after rising 0.2% in April, the third decline in the past four months for a cumulative -2.7%. In the pre-pandemic period, this is the largest four-month decline since the Great Recession. January’s level is the recent peak, making this the indicator with the strongest recession signal.
Taking stock and giving more weight to personal income and payrolls, the economy was still likely expanding in May, albeit in an ebbing fashion. However, more ‘knife-edge’ performance appears to have emerged in June, but it’s still too early to determine recession. That said, the risk of falling into one is now being amplified.