Saturday, September 11, 2021

Weekly Indicators for September 6 - 10 at Seeking Alpha


 - by New Deal democrat

My Weekly Indicators post is up at Seeking Alpha.

The big surprise of the past several months has been how little effect the Delta wave has had on the data, and in particular consumer data. 

As usual, clicking over and reading hopefully is rewarding to you, and a tiny bit rewarding to me as well.

Friday, September 10, 2021

Coronavirus dashboard for September 10: was Labor Day indeed the peak of the Delta wave?


 - by New Deal democrat

I have been saying for some time that the Delta wave would probably peak around Labor Day. It’s not certain yet, but it is looking increasingly likely to have been the case.

The Delta wave struck in both the US and Israel at almost the same time, with almost the same vaccination profiles. Here’s what cases per capita (bold lines) and deaths per capita (dotted lines) look like for each: 

Cases in both countries appear to have peaked in the last week (repeating the pattern in the Delta waves in India, the Netherlands, and the UK). Deaths have either stabilized or (more likely) are still slowly increasing.

Data in the US was affected by the Labor Day weekend. Here is the daily count of cases for the last 4 weeks in the US:

Note the difference in last Monday and Tuesday, the 6th and 7th, compared with the prior three weeks. The weekly average of cases is down from one week ago because the two days combined did not have as many cases reported as in the previous week. So we won’t have a real measure until next Monday’s and Tuesday’s cases are reported for comparison.

Some further good news is that there is a clear uptrend in only 10 States, of which 9 are relatively small and/or rural. In the remaining State, Pennsylvania, the uptrend is quite small:

Once again, we will have to wait for next week to see if this was just an artifact of Labor Day or holds up.

But if we are at or have just passed the peak of the Delta wave, now is a good time for a quick review of my forecasting of the Delta wave, which began back in June, while national case counts were still declining, here and here, when I said

As the “Delta” variant becomes more widespread in the next 4 to 8 weeks, it will be a real challenge for the relatively unvaccinated States

And further that:

If Missouri’s rate continues to rise (and, as we’ll see below, there is every reason to expect that to happen), then COVID is going to burn through Missouri’s (and other similarly situated States’) population in the next few months like a forest fire exploding uphill.
If these States’ trends continue - and there is no reason to think their populations are going to change their minds about vaccines, masking, or social distancing at this late date - then they will be in the thick of a “Delta wave” in about 2 to 4 weeks, with many other States in the Deep South and interior West close behind

By early July, right as the upturn in cases was beginning, I wrote:

The bad news is that the “delta wave” is spreading, and we should expect a real outbreak on the order of last summer’s by early August. The *relatively* “good” news is that the death rate is likely not to be nearly so bad, if the experience in the UK is any guide.

And also

“I expect the situation for all of the above States [AR, MO, NV, and FL], except possibly Arizona, to change considerably for the worse before the end of this month.”


And also:

“If the US follows the same course as the UK, 1 month from now the US will have about 1000 deaths per day.”

And also:

“So, brace yourselves. Cases have nearly tripled in the US over the past 2 weeks. Deaths are likely to increase to nearly 1000/day over the next 2 to 4 weeks.”

By the end of July, I was looking for when the peak would occur:

“It seems pretty clear that Delta burns through the dry tinder very fast - on the order of 9 to 12 weeks from onset to peak, based on the experience of the UK and India, respectively. The US is 7 weeks past its trough in cases, so it is a fair hypothesis that the Delta wave will reach its peak at some point in the next 2 to 5 weeks - roughly at some point in the second half of August or early September.”


And again in early August:

“from prior trough to peak took 12 weeks in India, and 9 weeks in the UK. During that time, in India *confirmed* cases rose 35x. In the UK it was 25x. Seven weeks in, the US has seen an 8x rise in cases.”

[Note: last week, the US had seen a 14-fold increase in cases from the end of June trough.]

And also here. And then two weeks ago:

“So if the pattern continues, it looks like the Delta wave is about 1 week from peaking - I.e., right about and maybe a little before Labor Day.”

Since no critical self-analysis would be correct without acknowledging my misses, one area where I was wrong, at least early on, was based on the seropositivity report out of India that during the Delta wave over 50% of the population had been infected, based on seropositivity samples.

So in the late July note I referenced above, I wrote:

“If we use the UK and India as our range, when it comes the peak in the US will be about 280,000 to 400,000 cases/day!

“If I am correct that Delta is going to infect at least 50% of the US’s unvaccinated population over about the next month, then the remainder of this month and September are going to be brutal. But if about 70% of US adults are fully vaccinated by then, and over half of the rest have antibodies due to recent Delta infection, then with over 85% of US adults immune either the easy or the hard way, by late this autumn there may an actual return to near-normal life.”

Based upon the same estimate that 50%+ of the population had been infected in India, I also initially specter cases to rise right into October.

This was wildly high (at least so far!). Trevor Bedford, who wrote in July that he expected Delta to infect about 35 million people before the wave was over, looks to have been closer to the mark. Since the end of June there have been a little over 7 million *confirmed* new cases, with at least as many so mild or asymptomatic as to never have been confirmed by testing. If the wave recedes as quickly as it hit (a *very* optimistic scenario), that would total about 28 to 30 million cases in total.

So where do we go from here? One month ago I wrote:

“Once Delta burns through the dry tinder nationwide, which is looking more and more to happen sometime around Labor Day, just what % of all Americans have actually been infected by COVID becomes determinative in what is likely to happen next.”

Since early July about 20 million people, or 7% of the US population, have become fully vaccinated. If that rate continues over the next 2 months, then 60% of the total US population will have been fully vaccinated. If Delta infects a total of 30 million since its onset, in addition to the estimated 20% previously infected, that gives us a total of 30% of Americans with antibody resistance of some efficacy. If those 30% are randomly allocated between vaccinated and unvaccinated, this gives us a total of about 72% of Americans with either vaccine or natural resistance as of the end of October. Not enough for herd immunity, but certainly enough to limit the prevalence of the next wave.

Additionally, with the Biden Administration, and some States and localities taking tougher measures to mandate vaccinations - and here I am particularly thinking of the LA School District’s mandatory vaccination policy, which I suspect will spread like wildfire to many other large urban school districts - we are likely to get a boost of at least a few more % of vaccinations among the populace.

So, with full understanding that there are lots of unknowns, my best guess as to the near future is:

1. We are at or have just passed the peak of the Delta wave.

2. Cases will decline, but at a rate much slower than at which they rose in July and August, as school openings, autumn sports, and increased indoor gatherings in the North give Delta some new dry tinder to burn through.

3. Another late autumn and winter wave looks likely, but not nearly as bad as last winter’s, and probably not as bad as this summer’s Delta wave; and it will remain a wave very much concentrated on the vaccine holdouts who manage to dodge the various mandates that are likely to increase sharply.

Thursday, September 9, 2021

Jobless claims blow away the Delta wave (but beware Labor Day seasonality)


 - by New Deal democrat

This morning’s initial jobless claims report makes it shockingly evident that the Delta wave has had no appreciable effect on at least the “firing” side of the jobs market (vs. the “hiring” side, where it might have).

Initial claims declined 35,000 to 310,000, and the 4 week average also declined 16,750 to 339,500, both yet more pandemic lows:

By way of reference, it took almost 5 years into the last expansion - until spring 2014 - for initial claims to be this low.

Continuing claims declined 22,000 to 2,783,000, also another pandemic low:

In the last expansion, this number was first seen in early 2014 as well.

These are, to put it bluntly, normal expansion numbers.

With one caveat: Labor Day seasonality may have driven some of the decline. The below graph shows the % by which the seasonally adjusted number has exceeded or fallen short of the unadjusted number, beginning in July 2020:

During the period from late July through October, seasonality adjusts the raw number higher. But as you can see from the call-out, the Labor Day week gets the least upward adjustment. That pattern was only accentuated this year. So take this week’s report with a grain of salt, and see if the big reduction is maintained or reversed next week.

Wednesday, September 8, 2021

July JOLTS report shows market still out of equilibrium, no additional hiring from early termination of benefits


 - by New Deal democrat

This morning’s JOLTS report for July is particularly important, because July was the first full month after a number of GOP-controlled States terminated enhanced unemployment benefits, on the theory that they were excessive and were coddling idle workers. Thus we should be seeing a big drop in unfilled job openings, as those people were incentivized to rush out and accept new employment.

It didn’t happen.

Job openings increased roughly 750,000 to yet another new all-time record of 10.934 million (blue in the graph below). Meanwhile actual hiring *decreased* by about 250,000 (red):

Here are the month over month percentage changes for each of those metrics:

Voluntary quits also rose, and are higher than any other prior month except this past April, as to which they were only 15,000 lower:

The record number of people voluntarily quitting their jobs (meaning they are not eligible for unemployment benefits) is testimony to the record robustness of the jobs market, and is even more impressive given the cutoff of emergency benefits by many States.

Meanwhile, layoffs and discharges (violet, right scale) rose by 105,000 from their record low levels of May and June, while total separations (light blue, left scale) rose by about 150,000 to a level that was typical for late in the last two expansions:

This is a market that continues to be out of equilibrium and is searching for a new one, as hires are well-above typical rates in the last two expansions for the second month in a row.

To return to the issue of job openings, here is this month’s update of a graph by Wolf Richter, showing that continued unemployment claims have declined in the aggregate in States that have cut off pandemic unemployment benefits vs. those that retained them:

But here is another, showing that there has been no enhanced level of actual new hiring in those States that terminated emergency benefits early vs. those that didn’t:

In other words, the termination of benefits was effective in reducing the unemployment compensation rolls, but apparently not effective at all in actually generating new employment. That job openings have only increased since those States terminated benefits suggests that other factors, such as the inability to find child care, or concerns about the safety of jobs on offer, are essential factors.

A great deal continues to depend on the course of the Delta wave, and in particular whether the wave is peaking right now and will sharply recede, or simply migrate north with the colder weather (thus increased indoor gatherings in those areas). A new equilibrium in the jobs market will be reached a lot sooner in the former case than in the latter.

Tuesday, September 7, 2021

The unemployment rate is not *uniquely* overestimating the “true” employment situation


 - by New Deal democrat

Bill McBride a/k/a Calculated Risk put up an entry over the weekend positing that the employment situation is worse than the unemployment rate indicates.

He basis this on the expectation that the overall labor force was expected to grow by 100,000 a month in 2020 and this year, whereas as of last month there were a little more than 2.9 million less people employed compared with just before the pandemic. This shortfall, he calculates, amounts to an “adjusted” unemployment rate of 7.9% vs. the official 5.2%.

This type of calculation is similar to many that were floating around for almost the entire duration of the last expansion - that the employment situation was far worse than the “official” headlines. 

My take is a completely different one. I start out with the proposition that Of Course the unemployment rate underestimates “true” unemployment. That’s why we have the U6 underemployment rate, that most notably counts people only working part time because they cannot find full time work. And even beyond that, we have the people who aren’t unemployed and aren’t even in the labor force, but tell survey takers that the “want a job now,” series NILFWJN that is posted in every monthly employment report.

In short, my question is, is there any *unique* way that the unemployment rate is undercounting now vs. any other month? 

My answer is “No.”

Let me start with a graph of all three of the above metrics: U3 (the official unemployment rate, blue), U6 (the underemployment rate, red), and those Not in the Labor Force who Want a Job Now (gold, right scale):

The “official” unemployment rate now, 5.2%, is the same as where it was in summer 2015, and spring 2005 in the earlier expansion. The underemployment rate, at 8.8%, is where it was in March 2017, and in autumn 2005 in the earlier expansion. Finally, NILFWJN as a percentage of the labor force is equal to where it was in September 2017, and slightly above its worst level of 2004 in the earlier expansion.

In other words, all three of these metrics now are at about levels they were at in the middle of the last two expansions. Nothing terribly unique about that.

Beyond that, let’s take a look at the prime age labor force, aged 25-54 years. This is the group that has, generally speaking, finished their eduction, but is still too young to retire, and is the group Paul Krugman honed in on when discussion employment shortfalls during the last expansion.

Let’s start with the Labor Force Participation Rate for this group:

Note that in the 1980s this ratio was rising strongly, as more and more women entered the labor market. That dynamic had pretty much hit peak in the 1990s. Since then, there has been a nearly persistent slow decline.

At its peak in 2007 and 2020 just before the pandemic, 83% of people aged 25-54 were in the jobs market, whether they were employed or unemployed. As of August, this ratio was 81.8%, roughly a 1.2% shortfall.

Now here is the employment-population ratio for this same group:

In three of the last four expansions, this peaked at 81%. As of last month it was 78%, a 3% shortfall.

The difference between the two metrics is that those who are unemployed are still in the labor force, but not employed. Hence they are included in the first metric but not the second.

In other words, the unemployment rate among the prime age workforce is about 1.8% above its best levels in the last several expansions. In the 2002-07 expansion, that was 4.4%. In the 2009-20 expansion, that was 3.5%. The current unemployment rate of 5.2% is 1.7% above the latter, and only 0.8% above the former. At most that suggests that the current unemployment rate is about 1% lower than it would otherwise be.

This conclusion is buttressed when we take into account the actual population of the prime wage labor force (red):

Just before the pandemic, it was 125.9 million. Last month it was 126.1 million, only a little over a 0.1% increase - nowhere near the 1.8 million increase suggested in Bill McBride’s analysis. In fact, when we use the entire working age population estimates covering ages 16 through 64 (blue isn’t the graph above) , there has been nearly a 500,000 decrease!

In conclusion, while certainly the official unemployment rate undercounts those marginally attached to the labor force, or not even in the labor force who want a job, there is no reason to believe that it is *uniquely* undercounting the unemployment rate now vs. any other time in the past 30 years.