November 04, 2023
Understanding the economy is often like putting together a jigsaw puzzle. You try to put together the pieces in a way where they fit together and give a clear picture. Unfortunately, the data do not always cooperate with this effort.
That is very much the case with the October jobs report released yesterday. On the one hand, we can look to the 150,000 job growth reported in the establishment survey and say that we had another month of very solid, albeit slowing, job growth. The figure is especially impressive when we add in the roughly 30,000 workers who were not counted because of the UAW strike. These workers will show up in the November data, now that the strike has ended.
However, we get a very different picture from the household survey. This showed another 0.1 percentage point rise in the unemployment rate to 3.9 percent. While this is still very low by historical measures, it is an increase of 0.5 percentage points from the April level. Furthermore, the household survey showed an actual drop in employment, with the number of people reporting that they are employed down by 348,000 from the September level.
This divergence continues a pattern since April. Over the last six months, the establishment survey showed a gain of 1,234,000 jobs. The household survey showed an increase in the number of people employed of just 191,000.
The two surveys are often out-of-line, that has been especially the case in the pandemic recovery. Many of us were very troubled by the discrepancy that was reported last year. In the eleven months from January 2022 to December, 2022 the establishment survey showed the economy created 4,430,000 jobs. The household survey showed employment had increased by just 2,120,000, creating a gap of more than 2,300,000 jobs.
This gap was hugely reduced when the Bureau of Labor Statistics (BLS) introduced new population controls in January, based on Census data, which added 954,000 to the employment figure. Job growth in the establishment survey was also revised down by around 300,000 in the annual benchmarking to state unemployment insurance filings. That still left a substantial gap, but considerably less than had previously been reported. (Some of this gap is due to differences in definitions. There was a fall in self-employment in 2022. This would lower employment in the household survey, but it would not show up in the establishment data.)
When the surveys do diverge, to my view it is always better to go with the establishment survey. It has a much larger sample and far higher response rate. It surveys 651,000 establishments every month. By contrast the household survey only covers 60,000 households.
The response rate to the household survey has fallen sharply over the last three decades and it is now just over 70 percent. This raises serious issues of non-response bias, since there is reason to believe that people who are not employed are less likely to respond to the survey.
The response rate to the establishment survey has also fallen somewhat. The advance report, which is the basis for the employment report for the month, only has a response rate around 65 percent. However, BLS continues to collect responses for two subsequent months. The response rate by the third month is over 93 percent. Given its size and high response rate, at least by the third month, the establishment survey should be a much more reliable gage of the labor market.
We can also look to try to reconcile what we see in these surveys with other data on the economy. We saw truly extraordinary productivity growth in the second and third quarters, 3.6 percent and 4.7 percent, respectively. The productivity data are highly erratic, and also subject to large revisions, but it is definitional that if employment grew less than reported in the establishment survey, productivity growth would have been even higher than reported.
We also have other labor market surveys, notably the Job Openings and Labor Turnover Survey (JOLTS). This survey measures job openings listed by companies, as well as hiring, firing, and quits. The JOLTS data have shown some weakening over the last nine months, but this is consistent with the sort of slowing in job growth we have seen in the establishment data. It is not consistent with the virtual halting of employment growth implied by the household data. Private measures, like the Indeed data on new hires and listings, is also consistent with gradual slowing in a still strong labor market, rather than the grim story in the household data.
The household survey also regularly gives us anomalies that clearly did not actually happen in the economy. In the October report, the data on employment rates for college educated workers implied that the number of college-educate people in the country increased by 1.1 million from September to October.[1] The Biden administration has tried to push policies that will make it easier for people to go to college, but I doubt that it will be taking credit for this one-month jump in the number of college grads.
What Does the Establishment Survey Tell Us?
If we can largely dismiss the grim picture from the household survey, we then have to ask what does the establishment survey tell us about the state of the economy? It is generally a good story, but there are some grounds for concern.
The rate of employment growth is clearly slowing, but 180,000 new jobs is hardly cause for concern. The Fed had been raising interest rates with the goal of slowing the rate of job growth. The argument is that with unemployment below 4.0 percent, there are not very many people still looking to be employed, and the number of new entrants to the labor market is not over 200,000 a month. Therefore, if we had continued to see the rapid job growth of 2022 and the first half of 2023, we would see substantial upward pressure on wages, which would be inflationary.[2] From this standpoint, the slower rate of job growth is just what the Fed was looking for, and should be the basis for it easing up on interest rates, or at least not raising them further.
In this vein, the establishment data also showed that the rate of growth of the hourly wage has slowed sharply. The annual rate over the last three months is just 3.2 percent. With inflation slowing to rates below 3.0 percent, this will still allow for real wage growth, but should not cause concerns about inflationary pressure in the economy. Wage growth by this measure averaged 3.4 percent in 2018-2019, when inflation was at the Fed’s 2.0 percent target.
All of this looks very good from the standpoint of a sustainable recovery. Strong job growth, combined with modest growth in real wages, should allow for consumption growth to remain healthy. And, consumption is by far the largest component of GDP.
However, there is some basis for concern in the establishment survey. The job growth for October was heavily concentrated in a small number of sectors. Health care added 58,400 jobs, and the larger category health care and social assistance added 77,200 jobs, more than half of the employment growth reported for the month. (It’s still 43 percent of job growth if we adjust for the UAW strike.) The government sector, mostly state and local governments, added 51,000 jobs in October. This doesn’t leave much room for job growth in other sectors.
Retail, which accounts for almost 10.0 percent of payroll employment, added just 700 jobs. Restaurants, which have added an average of 31,000 jobs a month over the last year, lost 7,500 jobs in October. The transportation and warehousing sector, which is responsible for moving stuff around, lost 12,100 jobs, mostly in the category of warehousing and storage. And, the financial sector lost 2,000 jobs, driven by a drop in credit intermediation (e.g. mortgage issuance) of 10,000 jobs.
These are all somewhat worrying signs, since a recovery that is driven by a small number of sectors may not have long to live. In keeping with this concern, the one-month employment diffusion index, which measures the percentage of industries adding workers, fell from 61.4 in September to 52.0, the lowest level since the lockdowns. So, there is some ground for concern.
However, we can find some positives looking by industry. Construction, which is historically the most cyclically sensitive sector, along with manufacturing, added a healthy 23,000 jobs in October. And manufacturing employment would have been pretty much flat, had it not been for the impact of the UAW strike.
Also, the information sector lost 9,000 jobs in October, primarily due to a decline in employment of 5,400 in the motion picture industry. This is largely due to the ongoing strike by the Screen Actors Guild, which has caused most movie production to grind to a halt. Presumably this strike will end at some point and we will see a jump in employment in the sector.
Overall Picture: Things Look Good, but We Need to Worry
I suppose we should always be worried about the possibility of bad events on the horizon. For example, it is certainly possible to envision scenarios in which the wars in Ukraine and the Middle East expand in ways that have large economic impacts, in addition to an enormous human toll. High long-term interest rates have both put a huge cramp on the housing market (existing home sales are down more than 30 percent) and have created the basis for more of the financial instability that we saw with the collapse of Silicon Valley Bank. A Fed rate cut, or even a signal that one is on the horizon, could be a huge help here.
But for now, most aspects of the economy are looking pretty good. There have been few times in the last half-century where we could tell a better story about the state of the economy.
[1] The September data showed an employment-to-population ratio for college grads of 71.9 percent, with employment of 62,907,000. This implies a population of college grads of 87,492,000. The October data showed the employment-to-population ratio had fallen to 71.3 percent, but the number of employed college grads had increased to 63,172,000. That implies a total population of college grads of 88,600,000, a gain of more than 1.1 million from September.
[2] There are two arguments on wages and inflation that should be distinguished. One is that rapid wage growth caused the jump in inflation we saw in 2021 and 2022. This can be easily dismissed, since prices outpaced wages, at least through the first half of 2022. However, there is a second argument that needs to be taken seriously. If we sustain a rapid rate of wage growth going forward (wages had been growing at more than a 6.0 percent annual rate at the start of 2022), it will lead to inflation. Wages can outpace prices in line with productivity growth, and we can have some period where wages rise at the expense of profits, reversing some of the increase in the profit share since the pandemic and in prior years. However, wages cannot persistently outpace profits by an amount in excess of productivity growth, without leading to serious problems with inflation.
Comments