Q3 US GDP Mystery, Jobs Report

By in
Q3 US GDP Mystery, Jobs Report

Three “Data” items today:

#1: We’ll begin with a “Data” item that, ironically, has no data: the New York Fed’s automated Nowcast model of current-quarter US GDP growth. We got the following email from the NY Fed this morning:

“Nowcasting Report: Suspension Notification

The uncertainty around the pandemic and the consequent volatility in the data have posed a number of challenges to the Nowcast model. Therefore, we have decided to suspend the publication of the Nowcast while we continue to work on methodological improvements to better address these challenges.”

This is the equivalent of a company backing out of speaking at a conference, so we got to wondering about why the NY Fed chose today to suspend publication. Our decades on Wall Street has taught us that there are no coincidences. Also, recall that when we reviewed the NY Fed Nowcast model last week we mentioned that Friday’s Jobs Report would certainly have an outsized impact on its Q3 estimate. This stood at just 3.8 percent at the time.

The Atlanta Fed’s GDPNow model is not yet updated for Friday’s Jobs Report, but even without that data the trend in Q3 US GDP is not our friend. The chart below shows that the latest readout calls for just +3.7 percent Q3 GDP growth, down from an estimate of 5.1 pct at the start of September. Again, this drop was not due to the weak employment report. We won’t know what effect that had on the GDPNow model until Friday September 10th.

Takeaway: it looks very much like the Fed’s own algorithmic GDP models are pointing to a zero growth Q3 for the US economy. As bad as that sounds, there are 2 positives. First, a good chunk of that is due to supply chain constraints (think automobile production). Second, US corporate earnings growth has managed to beat the last 2 quarters even though GDP growth either merely met expectations (Q1) or even missed (Q2). It will be interesting to see if the Atlanta Fed also stops releasing its GDPNow estimates; theirs is the better-known model and the market may well take that as a bad sign.

#2: The piece of any Q3 GDP miss that WILL concern equity markets: how much is due to a retrenchment in US consumer confidence and therefore spending? The delta variant has taken a toll on Q3 in a variety of ways, ranging from back-to-school shopping to travel, as we have been chronicling in recent weeks.

The Google Trends chart below shows pandemic-related search volumes since January 2020. The first peak on the left (late March 2020) was the initial wave. The second peak in the middle (November 2020) was the second wave plus the news of a vaccine. The third peak on the right was 30 days ago and corresponds to the delta variant wave. Search volumes here were only 12 percent below the November 2020 peak and, as you can see, did not decline much during August.

Takeaway: the rapid increase in pandemic-related searches from the end of Q2 (index of 26, as noted above) to sticky near-peak current volumes tells a story of why Q3 has been so sluggish. When we last looked at this data about 10 days ago it looked as if American public concern was beginning to moderate. Now we see it is holding in at very high levels, a sign that investors should not expect much from this month’s economic data.

#3: By now you know the basics of Friday’s Jobs Report (“very disappointing” was the common headline), so let’s put the data in some mathematical and historical context to show why the US equity market did not respond more forcefully to the miss either last week or today.

First up, here is the monthly growth in nonfarm payrolls over the last 12 months, with August 2021’s +235,000 highlighted in the rightmost data point and the July 2021 high water mark of 1.1 million just before it. If your first reaction is “hmm … that seems like a pretty volatile data series …”, we (and the math) would agree.

  • Mean monthly job growth over the last year has been +503,000.
  • The standard deviation is +/- 381,000, or more than half the mean.
  • A 1-sigma band is therefore +884,000 to +122,000.

Takeaway: no surprise, but US job growth has been – and will likely continue to be – volatile so we should take any one reading (like August, but also July’s) with a grain of salt. The market clearly does.

Next, let’s look at labor force participation (LFP) for 25 – 54-year-old Americans. While overall LFP (red line, right axis in the chart below) has gone nowhere for a year, LFP for working-aged Americans (blue line, left axis) has been at post-pandemic highs (82 pct) for 2 months straight and is just 1 point shy of pre-pandemic levels.

Takeaway: check out the gap between the red (total LFP) and blue (25–54-year-old LFP) lines that starts to form in early 2021 and continues to widen even now – this is predominantly due to retirements. That’s an important point when considering the ongoing US labor market shortage. There is still some slack in the 25 – 54-year-old worker cohort relative to pre-pandemic times, but even after that closes total US LFP may never get back to pre-pandemic levels.

Finally, last month’s wage growth got some attention, so let’s put that in some historical context. The chart below shows average hourly earnings for production and nonsupervisory employees back to 1970. (This is the BLS wage data series with by far the longest history). We purposefully made this a very long run data series so you could see how wage inflation trended in the 1970s (6-9 percent, leftmost part of the chart) and be able to compare that to today (choppy in the last year, 4.7 pct in August).

Takeaway: we remain very cautious about any comparison that implies current US wage growth is like the inflationary 1970s. As with our analysis of the headline job growth data, our preferred approach is to acknowledge the data is and will continue to be very choppy.