Most Indicators Keep Rising, But Some at Decelerating Rates as Shutdowns Take Hold
Unemployment claims data are broken but show improvement; Journalists mis-read or distort several of this week's stats
Summary:
Data show the economic recovery is continuing through late November in almost every respect. Some specific data points (the Chicago and Dallas surveys, the payroll employment [before considering a distortion stemming from the end of census jobs]) that show continuing positive growth are doing so at a "moderating" or "cooling" rate. Other specific indicators are much stronger, including the latest weekly asterisked/disclaimered jobs claims and the Weekly Economic Index of which they form a part, plus all housing data.
In spite of all the good objective, quantitative news, the general sense that things are "grim" or "bleak" continues, or intensifies. We certainly see that in the way the journalists frame the numbers, in the commentary of politicians, and in the surveys of consumers that we considered last week, all of which is heightened—or maybe largely caused—by the way state and local policymakers are trying to limit the virus by taking steps to limit economic activity.
If you heard any news of economics stats last week, it was very likely the (un)employment news on Friday. But I begin this newsletter, as I have the last several, with the more comprehensive and super-timely measure of activity known as the Weekly Economic Index (WEI) from the Federal Reserve Bank of New York. It is a composite of 10 daily and weekly indicators scaled to show an estimate of the percent change of total national output (= national income) from one year earlier. That is, the latest number of -2.31 says that the Gross Domestic Product is now, as best we know, about that many percentage points lower than it was in early December 2019—abstracting away from the fact that GDP is measured only quarterly. In April, the economy bottomed out according to this measure at -11.45% below April 2019, as seen in the blue line in the graph.
I've been watching the week-to-week fluctuations in this blue line quite closely. Last week I wrote that "the index is still trending upward, but at a weaker pace than in Sep and Oct, and including a step backward a week ago. Specifically, the last 4 data points are -3.18, -2.96, -3.00, and -2.83, meaning that it's looking like 4th quarter GDP this year will be about 2.83% lower than the same quarter in 2019." A look at the newest blip in the blue line show that last week's news was good, contrary to the relative weakness seen a week ago. The "weaker pace" reversed itself to a stronger pace, such that I can conclude that as of now:
Averaged over the last four weeks or any interval longer than that, the recovery according to WEI seems to be proceeding at the same steepness we have seen on average since the April low.
(Such week-to-week blips are why the New York Fed focuses on the smoother13-week moving average, which also shows a virtually straight, unwavering upward-sloping line since the series lows.)
The NY Fed statisticians summarize their report with this single paragraph of Commentary:
The increase in the WEI for the week of November 28 is due to a decrease in initial unemployment insurance claims and rises in tax withholding, fuel sales, and rail traffic (relative to the same time last year), which more than offset a decline in electricity output.
I discuss the good news about unemployment claims immediately below; my newsletter two weeks ago ended with an exploration of the rail traffic data; and I don't have more to say about the other indicators mentioned. Last week's Commentary also cited a decrease in initial unemployment insurance claims. But if you have been noticing the headlines every recent Thursday about those, you've been hearing that they are up or stubbornly not falling, alarmingly. Consider this CNBC headline from Tuesday Dec 1:
Nine months after the health and economic crisis began, unemployment claims near record high
Then, three days later, reflecting on the most recent weekly report, CNBC's story was:
Jobless claims hit pandemic-era low as hiring continues even with rising Covid cases
Adding to the confusion, we have many reports of "tremendous amounts" and "staggering" fraud in the filing of claims for unemployment insurance. Last week I relayed several such headlines and opined that "the table of initial claims numbers needs to be marked with an asterisk." By Tuesday, I wrote on Facebook "I was more right than I even knew" in response to this news story:
U.S. Jobless-Claims Data to Come With Disclaimer on Accuracy
… for reasons partially described in these three revealing paragraphs (with active hot links to further information):
[I]f a person files for six weeks of benefits in a given week, for example, that’s typically counted as six separate people in the total, instead of one person. The GAO report said this method of counting is normal for the Labor Department, and before the pandemic provided a good proxy for the actual number of people claiming benefits.
But with unemployment skyrocketing in March and April and jobless-benefit programs multiplying during the current crisis, the method has led to consistent inaccuracies that have been evident for months.
For example, the number of continued claims submitted in the Pandemic Unemployment Assistance program through June 27 exceeded the number of people who had submitted an initial claim by nearly 20 million, a GAO analysis of 20 states’ data showed.
With all of these asterisks and disclaimers in mind—but not adjusted for—let's look at the last 5 months of stats for initial claims for unemployment insurance, as they are published by the Department of Labor (in thousands):
Here's where it gets interesting: the single paragraph of Commentary on last week's WEI page, for the week of Nov 21, said the same thing as this week's—that there had been a decrease in initial unemployment insurance claims. The reason I didn't quote that paragraph a week ago, as I almost always do, is because I couldn't reconcile that statement with the official stats (in the table above) showing an increase for that week. It's not a matter of there having been a revision, since the 787,000 was originally reported (you can see it in last week's newsletter) as 778,000—still an increase of 30K.
So what this means is that, in light of the "consistent inaccuracies that have been evident for months" cited above about the Department of Labor's numbers, the NY Fed must be doing their own internal adjustment of them and using it in their Weekly Economic Index. Unfortunately, this adjustment is not transparent to me, but it was enough to covert a +30K number from Labor into a "decrease" according to NY Fed. Consequently, I will no longer access nor report on the official DOL numbers, but instead just let the adjusted Weekly Economic Index speak for the unemployment claims part of the economy.
For what it's worth, here are the money quotes from the CNBC "claims hit pandemic-era low" article I cited above:
The jobs market has demonstrated resilience even in the face of the new wave of Covid-19 cases. Claims are off their peak of 6.9 million in late March but remain well above the pre-pandemic record.
Continuing claims also fell sharply, dropping 569,000 to 5.52 million. …
Ian Shepherdson, chief economist at Pantheon Macroeconomics, said in a note, “Initial claims likely will rebound strongly next week, probably rising above the 800K mark for the first time in eight weeks.”
Mr. Shepherdson's pessimistic outlook is surely informed by news such as this:
Los Angeles Mayor Eric Garcetti: "It's time to cancel everything."
… illustrated further by this disturbing video from the Pineapple Hill Saloon and Grill in Sherman Oaks (San Fernando Valley, part of Los Angeles). The video is best understood after noting that the mayor had just then forced the closure of outdoor dining establishments, and that any movie filming activity requires the approval/licensure/whatever of the city government.
So, we think we know that the unemployment news is:
good for right now, but feared to be turning much worse within days.
The presence of weekly data is making the monthly data appear less relevant, but on Friday we did get news of October's employment 'situation,' as the Bureau of Labor Statistics calls it. In spite of most of the spin you’ve received from journalists, the jobs news was mostly quite positive.
BLS or their bots put out a fairly lengthy write-up of the situation, from which I quote (selecting the details that are showing the most change one way or the other from the previous month):
Total nonfarm payroll employment rose by 245,000 in November, and the unemployment rate edged down to 6.7 percent, the U.S. Bureau of Labor Statistics reported today. These improvements in the labor market reflect the continued resumption of economic activity that had been curtailed due to the coronavirus (COVID-19) pandemic and efforts to contain it. However, the pace of improvement in the labor market has moderated in recent months. In November, notable job gains occurred in transportation and warehousing, professional and business services, and health care. Employment declined in government and retail trade.
In November, the unemployment rate edged down to 6.7 percent. The rate is down by 8.0 percentage points from its recent high in April but is 3.2 percentage points higher than it was in February. The number of unemployed persons, at 10.7 million, continued to trend down in November but is 4.9 million higher than in February.
Among the unemployed, the number of persons on temporary layoff decreased by 441,000 in November to 2.8 million. This measure is down considerably from the high of 18.1 million in April but is 2.0 million higher than its February level. The number of permanent job losers, at 3.7 million, was about unchanged in November but is 2.5 million higher than in February.
In November, the number of persons who usually work full time rose by 752,000 to 124.3 million, while the number of persons who usually work part time decreased by 779,000 to 25.4 million.
In November, the number of persons not in the labor force who currently want a job increased by 448,000 to 7.1 million; this measure is 2.2 million higher than in
February. These individuals were not counted as unemployed because they were not actively looking for work during the last 4 weeks or were unavailable to take a job.
In November, average hourly earnings for all employees on private nonfarm payrolls increased by 9 cents to $29.58. Average hourly earnings of private-sector production and nonsupervisory employees increased by 7 cents to $24.87.
Quite a mix of good and bad signs are in here, and the full write-up contains much more. One additional fact worth noting is that, after adjusting for the census workers whose known-all-along-to-be-temporary jobs have recently come to an end, the number of net new jobs added in November was actually 468,000, which brightens the story substantially.
Here is the graph of the unemployment rate for the last 45 years, showing all of the business cycles that most adults can remember:
One obvious story in this graph is that the recession of 2020 is of a whole different category than any previous one—a conclusion regular readers should have reached a long time ago. Both the crash of Feb-April and the subsequent recovery have been lightning fast, creating a V that is needle-sharp and very unlike the Nike-swoosh patterns of other business cycles.
The last time the unemployment rate was headed downward and reached the same level of 6.7% just announced for November, was in December 2013, the 55th month (!) of that recovery. I looked up the old news about that from early Jan 2014, and found something interesting. The number of net new jobs that month was +74,000. The article reports that this was disappointingly low, and that's true. We would normally hope for at least 150K or up above 200K, particularly when the unemployment rate is high enough to put urgency on getting it lower. The article further notes that the guesspectation of polled economist soothsayers back in 2013/14 had been 196,000, which matches what I just said.
Compare that to where we are now. "US Job Gains Stall In November As Covid-19 Cases Surge," is the emphasis. You've already seen the numbers above: net +245,000 jobs gained in Nov 2020, or +468,000 after adjusting for the census-related distortion. Do I even need to say it? That is not a “stall” in job gains.
The matching CNBC story said that guesspectations had been for +445,000, although they note the contradictory facts that "Some bond market pros had expected an even worse job picture in November, and the market appears to have latched onto any positives, including the wage data."
To summarize what CNBC says:
experts expected +445K net new jobs
except some expected less than +245K
so when the number came in at +245K, that's good news that makes the financial markets react positively [you've noticed the stock market rose (‘melt-up’ is the word for it) in November by 12%, which is more than year's worth of normal gains, and the best single month since 1987, and that includes another big, healthy rise on Friday, the day this unemployment news was released (at 8:30 a.m.)]
and when that number came in at +245K, it generates headlines calling it a "stall" and a "weakening trend," and has politicians calling it "grim" and potentially "very bleak."
It all gets even more curious in light of the news coverage from December 2013 I referred to a bit ago. Remember, the experts then dared to hope for only +196K net new jobs when the rate was right where it is now; but now in 2020 the experts/journalists are calling for +445K and are disappointed when it comes in at a mere +245K (or +468K, census-adjusted). Why such a higher bar for success now?
Given that several states and localities were imposing fairly stringent restrictions on the conduct of business during November, and the fact that COVID cases and scares were sharply rising, it seems to me that for the parts of the country that remain open could power the national economy to the numbers we did see … well, that seems to me like a pretty good sign of underlying strength. And it seemed/seems the same way to the financial markets as well.
You'll notice that over these approx. 12 weeks, I have referred to the financial markets only I think one time, glancingly, before the paragraph above. That's because the stock market ≠ the economy. But the markets do serve 2 functions that are relevant to my project here. First, stocks are a leading indicator—i.e., a predictor—of other, later economic activity; indeed the stock market is one of the items in the Conference Board's Index of Leading Economic Indicators. Second, the various financial markets are also basically a set of permanent, on-going betting markets (like PredictIt.org or the old Intrade.com), where changing opinions are reflected, instantly, as changing prices. What I said a paragraph ago is relates to the second: the unemployment news of Friday was either shrugged off as not nearly as important/dire as the journalists' headlines would have you believe, or was taken actually as good news, better than expected, as said in the quotation from the CNBC story.
But if you want to see "grim" and "very bleak," wait till the next month or two as more governors and mayors do more to shut the doors of commerce.
Other data dumps came out on Monday Nov 30. In the interest of saving time and space, I'll quote from a Reuters story headlined "U.S. housing, manufacturing data suggest economic recovery slowing."
Contracts to buy U.S. previously owned homes fell for a second straight month in October as an acute shortage of properties pushed up prices, though the housing market remains supported by record low mortgage rates.
The National Association of Realtors said its Pending Home Sales Index, based on contracts signed last month, decreased 1.1% to 128.9. Economists polled by Reuters had forecast pending home contracts, which become sales after a month or two, would rebound 1.0% in October.
Compared to a year ago, pending homes sales jumped 20.2% in October. The monthly decline in contracts suggests a slowdown in sales of existing home sales after they accelerated in October to their highest level since November 2005.
Robust demand for housing has outstripped supply, boosting home prices out of the reach of many first-time buyers, despite builders ramping up construction. The government reported this month that single-family homebuilding, the largest share of the housing market, raced to the highest level since April 2007.
Though homebuilder confidence is at historic highs, builders have complained about shortages of land and materials.
What this says is that the housing market is too good for its own good. People are so eager to buy homes, and have bought so many earlier this year, that people searching now can't find what they want, and homebuilders are too busy to keep up. Continuing from Reuters:
Other data on Monday showed activity at factories in the Midwest and Texas slowing this month, likely as a nationwide resurgence in new COVID-19 infections curbed new orders and disrupted production. The reports support expectations of a sharp slowdown in economic growth in the fourth quarter because of the raging coronavirus pandemic and depleted fiscal stimulus.
A report on Monday showed the Chicago Business Barometer dropped to a reading of 58.2 in November from 61.1 in October. A reading above 50 in the index indicates expansion in factory activity in the Chicago area. The survey’s new orders measure dropped for the first time since May, when the recovery from the pandemic started.
The moderation in factory activity was corroborated by another survey from the Dallas Federal Reserve showing its production index, a key measure of state manufacturing conditions in Texas, tumbled to a reading of 7.2 this month from 25.5 in October. Factories in the region reported a significant slowdown in new orders, and were less upbeat about the outlook.
The surveys, together with reports earlier this month showing a cooling in activity in factories in New York and the mid-Atlantic region, suggest national manufacturing moderated in November after accelerating in October.
Unfortunately this writing is imprecise to the point of being misleading. The precise and accurate part comes in the second paragraph: diffusion indices like the Chicago Barometer are built such that any number above 50 is good news, expansion. Numbers as high as 58 show very distinct, unambiguous expansion. Slowing from 61.1 to 58.2 is a shift from "super fast" to "very fast" but in no way signals a "drop" nor that anything "curbed." The third paragraph and again the final line of the excerpt say it defensibly correctly: it's a "moderation in activity." Growth, expansion, more moderate in Nov than in Oct.
The rest of that paragraph sounds very bad, but does not explain what "7.2" really means. Here's how the Dallas Fed's own press release explains the 25.5 and the 7.2: "Texas factory activity expanded in November for the sixth consecutive month, though at a markedly slower pace." The number 7.2 is the margin by which the percentage of survey respondents answered "doing better/looking good" above the percentage that answered "doing worse/looking bad." So, any positive number is still showing growth or optimism.
The statisticians collecting the data tell us "activity expanded," and the journalists turn that into "tumbled." Again it's a matter of super positive news a month ago "falling" to still positive but less-than-it-was news now. More journalists should brush up on their differential calculus to be reminded of the differences between decreases in actual values and decreases in rates of change. I'm finding that I'm tediously explaining this almost every week, and I wish I didn't have to.
It wasn’t only the financial markets that shrugged off the week’s dire-sounding journalistic headlines or rather interpreted the news as positive. So too did the ‘now-casters’ at the Federal Reserve Bank of Atlanta, whose model called GDPNow was featured in last week’s newsletter. On Tuesday Dec 1 they looked at the Chicago and Dallas surveys and the housing boom, ran it through their quantitative tools and figured it meant the fourth-quarter Gross Domestic Product was on track to grow a bit faster than they had thought a few days earlier. Those data nudged the nowcast up from 11.0 to 11.1% annual rate of growth. Then again on Friday, after digesting the weekly unemployment claims news and the (un)employment situation report that morning, GDPNow updated again to show another nudge up another notch to 11.2% Q4 growth. Hence my headline “most indicators keep rising” and my summary “the economic recovery is continuing” and in fact accelerating slightly.