(Un)employment & Manufacturing Data Extremely strong
only partially offset by rival employment report and deceleration in services growth
Election week 2020 featured pairs of reports on manufacturing activity and on (un)employment plus one other measure of activity in the services sector. In spite of contradictory evidence within the pairs, the general picture for the week remains one of continued recovery from the COVID crash. The news on (un)employment was, on balance, particularly favorable.
First, though, I want to revisit the composite indicator that I introduced in last week's (delayed) newsletter. The Federal Reserve Bank of New York's Weekly Economic Index (WEI), described below, is the blue line in this graph:
This week's version appears identical to the naked eye, but it does have one new data point showing that both the primary blue WEI line and its lagged 13-week moving average line have continued to rise for an additional week. Last week's terminal blue point lay 3.52% below par; this week's is -3.12%. If the same rate of improvement were to be continued for 10 more weeks, we would be back to pre-COVID norms. (Not a prediction, just an extrapolation.) The low point in the series was -11.45% in late April.
Because of the comprehensiveness of its broad array of 10 indicators (which I listed last week) and the weekly timeliness with which it is released, I plan to begin my run-down each week with a check of this Weekly Economic Index. I'll highlight any deviation from the rising trend, and I'll at least mention any continuation of it. For visual consistency and comparability, I'll continue to use July 1, 2019 as my left margin.
The NY Fed provides a single brief paragraph of commentary about its findings, which I'll also relay to you. For this week it reads:
The increase in the WEI for the week of October 31 was due to a decrease in initial unemployment insurance claims and increases in tax withholding and rail traffic (relative to the same time last year), which more than offset decreases in electricity output and fuel sales.
The one sector of the economy I have not devoted major attention to in the previous seven weeks of this project is manufacturing. Two weeks from now I'll highlight the industrial production and capacity utilization figures that come out on Nov 17. This week I turn to two back-to-back measures of manufacturing activity. In chronological order, the first is Monday's Institute for Supply Management's PMI (Purchasing Managers' Index) for manufacturing in October. ISM, not being a government organization, puts together a whiz-bang press release that I am delighted to find does a lot of my work for me. It reports the raw numbers, of course, but also contextualizes them and compiles quotations from and links to all the media reports and interpretations you would want to find. I really encourage you to access the preceding link and read the whole thing, as a substitute for me typing hundreds of words conveying the same thing.
The news was extremely positive. The CNBC.com video report announcing the news begins with the word "Wow" and concludes "on the ISMs, it's zoom, zoom, zoom." CNBC's URL and headline say "best number since August," but that is actually August of 2018. For a PSI, any number above 50 indicates expansion, any number below means contraction; the October number was 59.3, up from 55.4, and also well higher than expectations. The PSI aggregate number breaks down separate readings on orders to factories, prices paid, and employment—all of which were also >50 and taken as exciting news by the CNBC reporter and others.
Some other excerpts from media sources, with the links found inside the ismworld.org link provided above:
The international bank ING's 'Think' service wrote: "this suggests ongoing healthy gains for orders and output in coming months. This in turn should be good news for ongoing employment growth — the employment component rose to its highest level since June 2019."
Bloomberg suggested the wider implications of this news: "The figures follow a report last week that pointed to firmer consumer spending and income gains in September and suggest economic growth could be stronger than forecast this quarter."
Then, one day later, the official U.S. Department of Commerce / Census Bureau reported their reading of manufacturing activity for September. The full name of the data dump is "manufacturers' shipments, inventories and orders," and consists of 7 sets of data series, each broken down into about 70 specific industries. So, there's a lot here and a lot of it is inherently boring and some is inherently ambiguous. For example, if inventories are up, that could mean either that goods are going unsold on shelves, indicating weak consumer demand; or it could mean that factories that had been shuttered are now re-opening, and all the supplies, parts, and half-finished goods on the factory floor are amounting to higher totals this month than last. It could mean both of those things at once. In fact, inventories were up in September and in 3 of the 4 preceding months, but in the most recent 2 months, the increase was "virtually unchanged," in the words of the press release. Nothing interesting to see here.
Except for the one of 7 data sets that does usually deserve (and attract) attention: new orders. By nature these are a leading indicator: if they are up this month, the actual production (and consequent employment) for them increase some months in the future. September's new orders increased for both the long-lasting durable goods sector (1.9%, un-annualized) and also for non-durable goods (0.3%). The aggregate of 1.1% represents a "solid growth rate" that extends the streak of winning months to five. The dollar value of these orders still lies 6.9% (non-durable) to 7.5% (durable) below their December pre-COVID levels.
The press release also includes this graphic:
… showing that the horrible declines of March and April were mostly offset by three super-strong months in May-July plus now two additional more muted/ordinary gains.
Then the next day, Wednesday, the ISM people were back with their report on the services sector for October, to complement Monday's report on manufacturing. Again for services, they provide a very helpful "roundup" of commentary, interpretations, and press clippings, which I recommend you read for the full picture. The news for services was mixed. The headline number of 56.6 indicates a fifth month of growth, since, again, any reading above 50.0 is positive. But 56.6 is down from 57.8 in September, and was below the expectors' expectations. The Roundup quotes an ISM representative Anthony Nieves saying, "The services sector is doing well." Then, restating the obvious:
The biggest factors limiting services jobs growth, Nieves said, are such businesses as hotels, restaurants and theaters being unable to fully reopen, as well as shortage of workers for some construction trades. “And with the COVID-19 situation varying by municipality and state, it’s a mixed bag right now,” he said.
Many news reports emphasized the negative. The same CNBC reporter who had gushed "Wow … zoom, zoom, zoom" regarding manufacturing on Monday said about services on Wednesday: "we do see that this is a step back." From the Associated Press: "Risks are to the downside from new restrictions and closures that will weigh on demand and activity going forward."
One sub-indicator within the ISM report concerns employment. That index "expanded for a second straight month, but at 50.1 percent, the growth was by a hair."
ISM employment data point is a minor player compared to the other two October (un)employment releases that came out on Wednesday and Friday. The first is the report from ADP, which I described to you at some length a month ago. It was broadly negative, in spite of showing a gain of 365,000 net jobs throughout the economy in one month, and slight upward revisions to the previous two months. In normal/good times, 365K is a strong number of net new jobs (as you can see in the next graph). During the COVID recovery it is considered tepid, and expectations had been for more, around 600K. ADP's graph looks like this, with the literally off-the-charts losses for April and the gains for May and June truncated, to allow the other months to appear as visible.
ADP's boss said in the press release, "Although the pace is slower, we’ve seen employment gains across all industries and sizes." Their next graph backs that up:
Notice that the biggest gains by far are in the services industries, which makes sense on two levels. First, these are the sectors that we know were knocked flat by COVID, so they have the most to gain in the recovery. Second, services also dominate our largely post-industrial economy, so just on a proportional basis, service jobs must grow by large numbers just to match their share of the economy. It seems to me significant, though, to stress that sectors such as "Leisure & hospitality" are recovering quite vigorously, and I wonder if that comes as a surprise to the general public, who might not be getting that message from the news and from their largely home-bound experience.
As previewed at the top of this newsletter, the relatively weak ADP employment report was offset or contradicted by Friday's official Bureau of Labor Statistics (BLS) news about the unemployment rate and the net payroll jobs growth. This was the most important, most wrongly estimated, and most optimistic/positive news of the week. It showed that we gained a net 638,000 nonfarm jobs in October, 108,000 better than expectations and much higher than ADP's number that we just saw. (The two have been diverging quite a lot this year, but not systematically; for example, in the previous month, ADP showed higher job gains than did the BLS; but both are showing sustained and generally strong gains.)
This companion piece to CNBC's write-up provides the rest of this week's graphics:
You'll notice right away that on this point ADP and BLS agree: The COVID recovery is now reaching the hard-hit Leisure & Hospitality sector, including bars and restaurants, which (net) gained 192,000 of those jobs. [ADP's numbers are only for the private sector—i.e., excluding the government from their numbers, so there is no contradiction between BLS' big loss in that category and ADP's silence on the matter.]
Apart from net new jobs, BLS calculates the unemployment rate. [They actually calculate six different rates, using different definitions of 'unemployed'; I'll have to clarify this matter in a future newsletter, since it often comes up when people accuse the government of lying about unemployment. There are also issues about labor force participation to consider.] The normal, headline figure for unemployment fell by a full percentage point from 7.9% to 6.9%. That's five times the improvement that the experts had been expecting.
The graph shows that we are now back to the same unemployment level as in 2014, five years after the end of the Great Recession (March 2009).
And those +638,000 net new jobs in October puts us back to the same number of jobs that powered the economy in 2015. Given population growth over that time, I'd adjust the comparison to say we're about on par with 2014 (instead of 2015) in this way of looking at it, too. Notice that the Great Recession of 2008-09 dug a job hole that took about 5 years to dig out of, but see how the steepness of the recovery in 2020 is in a whole different league—we’ve seen about six years worth of 2010s jobs gains in just six months. In round numbers, that’s about 12 million of the 22 million COVID job losses, regained.
Digging into the BLS report even further, the good news compounds to levels I'm literally having trouble believing. I have to keep re-checking my understanding of the numbers, but I'm confirming that CNBC gets it right. Referring to the BLS household survey that generates the unemployment rate (separate from the separate BLS payroll survey that generates the headline +638K job growth number):
The survey of households showed an even stronger level of job growth, with the total employment level rising by 2.24 million and the employment-to-population ratio increasing by 0.8 percentage points to 57.4%. The household survey also showed a decline of 1.52 million in the total unemployed level and a drop of 541,000 in those considered not in the labor force.
October’s gains would have been even better were it not for the loss of 147,000 Census workers that contributed to an overall fall of 268,000 in government jobs. In all, private job creation came to 906,000, better than September’s 892,000.
Those reporting that they had been out of work because their employer had lost business or closed during the pandemic also dropped considerably, falling to 15.1 million from 19.4 million in September.
Summarizing it all:
Although offset some by the mixed, less rosy ADP report, the BLS news on (un)employment is overwhelmingly positive. Similarly, the ISM's PMI measure of manufacturing activity was fantastic and the Commerce Department's was also positive, but these were partially offset by a "step back" ISM PMI report for services (which still showed positive but decelerating levels of activity). All in all, another set of good news for all of us with a stake in getting the economy quickly back to a recognizable level of health.