Saturday, April 9, 2022

Weekly Indicators for April 4 - 8 at Seeking Alpha


 - by New Deal democrat

My Weekly Indicators post is up at Seeking Alpha.

After several weeks of tightening and then inversion, the yield curve in the US Treasury market un-inverted in a big way this past week - via higher long term rates which drove mortgage rates above 5%, which will have a decidedly negative effect on housing.

As usual, clicking over and reading should be educational for you, and will help me pay for dinner out if the weather is nice.

Friday, April 8, 2022

In the Paul Krugman vs. Nate Silver cage match, I’m on Team Nate


 - by New Deal democrat

Prof. Paul Krugman and Nate Silver got into a dust-up earlier this week about why so many voters seem to have soured on the Democrats. So that you don’t have to go digging through all the Twitter detritus, Alternet has a good write-up copying all the relevant tweets. (Apparently the two have been at odds at least since 2014, when Five Thirty Eight left The NY Times and went to ESPN; I do not know nor do I care who dissed whom first back then).

The dust-up started when Nate wrote:

Krugman responded:

Since how the average American is doing in the economy is something I’ve been obsessively following for about 20 years, and I’ve also spent a fair amount of time parsing what economic indicators appear to best forecast at least Presidential votes, I think I’m qualified to weigh in.

Sorry, Professor, I’m on Team Nate on this.

Yes, Nate Silver shouldn’t have said “some folks on here” without being able to back it up with at least one specific (Note: I’m not going to go digging for that either, but my recollection is that I have read tweets from partisans to the effect that “Actually the Economy is Great;” I think one such partisan may have been Dana Houle). But the *substance* of Nate’s criticism - that voters do not think the economy is going so great - is in my opinion on target. Krugman “straw man” criticism is actually something of a “straw man” itself, failing to come to grips with the issue.

First of all, I should point out that, as to Presidential elections, Silver himself did a regression on some 50 economic indicators to attempt to figure out which ones best predicted *Presidential* outcomes. But to my knowledge no such systematic review has taken place with regard to Congressional and State elections. So take the Presidential formula with a hefty grain of salt.

If this were a Presidential election year, the decline in the unemployment rate would be good news for Biden, because the unemployment rate is one of the best predictors of incumbent vs. insurgent Presidential vote. I was also able to determine that real retail sales were also a very good indicator. James Surowiecki has calculated that changes in real disposable personal income during the election year is also predictive. Finally, economists at the London School of Economics have calculated that the Index of Leading Economic Indicators through the first quarter of a Presidential election year work well.

The news with respect to the above statistics is mixed. The unemployment rate is clearly down, the index of leading indicators is still positive but has decelerated sharply through February, and real retail sales per capita and real disposable personal income are down.

Perhaps more importantly, as shown in the graph below, since last May (when the big stimulus effect first wore off), on a per capita basis, real personal income is down -1.6%, real *disposable* personal income is down -2.6%, real average nonsupervisory wages are down -1.9%, and real retail sales per capita are only up 0.1%:

These are averages; in other words, while median measures would be better if we had them, taking the vast mass of people together, about half of all of them have sustained *real* economic losses in the past 9 months since the stimulus wore off, and they are reining in their spending. 

By contrast, only 1.8% of the population in the entire age bracket from 16 through 65 who were not employed as of last May has gotten a job:

In other words, 1.8% have benefitted from less unemployment, while somewhere near 50% (again, we don’t have median measures so this is the best we can do) have sustained losses. On top of that, since some stimulus benefits have ended (the $3000 child care credit comes to mind), this has created a hole in lots of families’ wallets, and the evidence shows they are blaming Democrats (thank you, Manchin and Sinema).

The difference between how “the economy” is doing, vs. how average families are doing, is best shown by comparing real *aggregate* payrolls, up 3.0% since last May, vs. real  *average* wages:

In the aggregate, even after inflation payrolls are up 3.0%. But the average worker’s hourly wage is down -0.8%. The economy has been booming; the average worker’s wallet, not so much.

So I side with Nate Silver.

Thursday, April 7, 2022

Benchmark revisions, oh my! Four week average of jobless claims makes all-time 55 year series low


 - by New Deal democrat

The DoL made revisions to the last five years of jobless claims, in particular revising the seasonal adjustments, and the differences are eye-popping.

Last week initial claims (blue) were reported at 202,000. With the revisions, they are now 171,000! This week they declined -5,000 from that revised figure to 166,000, tying the revised number from two weeks ago. The 4 week average (red) for last week, originally reported at 208,500, became 178,000, and declined a further -8,000 to 170,000 this week. On the other hand, continuing claims (gold, right scale), originally reported last week at 1,307,000, were revised to 1,506,000, and rose 17,000 this week to 1,523,000:

But that’s not the biggest story. Not only are the initial and 4 week averages pandemic lows, with the revisions initial claims last week is the lowest for the entire series going back almost 60 years with the exception of one week - November 30, 1968, which saw only 162,000 layoffs. The 4 week average was even better: this week’s number is the all-time series low, including all of the 1960s back to the series inception in 1966, as shown in the below graph which shows the current week’s numbers normed to 0:

And remember, the US population in the 1960s was only half of what it is now, so as a rate, this is by far the fewest % of layoffs in the population ever recorded. By the way, the revisions also indicate that layoffs were even lower than we thought - significantly below 200,000 per week - in 2018 and 2019.

For completeness’ sake, continuing claims were lower throughout the 1960s:

Is it any wonder that wages have been increasing as sharply as they have?

Wednesday, April 6, 2022

Scenes from the March jobs report; and the Great Resignation as “Take This Job and Shove It!”


 - by New Deal democrat

It’s been a little while since I took a more in-depth look at the jobs market, so let’s take a look.

As I wrote last Friday, we are at historic lows in both the unemployment and underemployment rates. In the graphs below, the current values of each are normed to zero for easy comparison:

Historically few people are involuntarily unemployed.

So, how close are we to “full employment,” and what sectors haven’t caught up yet?

As I wrote Friday, as of March we are only 1.579 million jobs short of where we were in February 2020 just before the pandemic hit. But since the population, and in particular the working age population has changed since then, below I show the total labor force participation rate (that is, the % of the US population that is either employed or unemployed, vs. not in the labor force at all), and also the prime age (25-54 years old) labor force participation rate:

The total population LFPR is 62.4%, vs. 63.4% in February 2020, while the prime age LFPR is 82.5%. The former has declined sharply since 15 years ago, and has continued to decline, for a very simple reason: Boomers are old! When we focus on the prime age rate, that is only 05% short of its value as of February 2020. Since there are about 176 million Americans in that prime age demographic, that’s roughly a 0.9 million shortfall in that age group. If we figure there’s about another 60 million adults below the age of 65, the total is also probably around 1.5 million:

Similarly, when we look at the series “Not in Labor Force, Want a Job Now,” which is a pretty self-explanatory statistic, that is 5.737 million, also about 1.5 million above its best level just before the Great Recession:

In other words, all three ways of looking at the remaining jobs shortfall tell us that about 1.5 million more jobs need to be added to get to full pre-pandemic employment. We’ve been adding over 500,000 jobs a month, so if this continues for just 3 more months, we will have full employment.

Turning the second question, the BLS itself has helpfully prepared the below graph of the gains and losses of various employment sectors since the pandemic started:

The usual suspects of leisure and hospitality (including food and drink), government, and education are still there, and manufacturing is still lagging.

The above tells us the nominal shortfalls. What about the % of job losses? That’s what the below graph shows:

Surprisingly, on a %age basis, it’s mining and logging that has sustained the biggest losses, down -12.5%. Leisure and hospitality comes in next, down -8.7%. Government is down -3.1%, education is down -1.9%, wholesale trade is down -1.8%, and manufacturing is down -1.0%.

Finally, on the theme of “somebody is wrong on the internet,” Kevin Drum writes that “The Great Resignation was neither great nor a resignation,” citing work by the San Francisco Fed that “suggest that the high rate of quits is simply a reflection of the rapid pace of overall labor market recovery.”

The San Francisco Fed makes an interesting case, citing historical manufacturing data. But my bone to pick is with Drum, who claims that “the quit rate [from the JOLTS report] merely rebounded to its old Great Recovery trendline . . .[or] merely ... only very slightly above trend,” producing the below graph in support:

I submit that Drum’s choice of the bottom of the Great Recession as the starting point for his “trend” is performing some Olympic-level heavy lifting. By contrast, when we graph all the way back to the beginning of the statistic in 2000, here’s what it looks like:

The quits rate very suddenly jumped 25% higher than it had ever been in the previous 20 years (3.0%/2.4%=1.25). By contrast, hires, which had previously in Y2K been as high as 4.3%, and 4.1% in 2019, only increased to 0.2% above that to 4.5%.

So, hires barely budged, but quits jumped to all-time new highs by a very wide margin.

A better comparison is with job openings, which like quits soared to all-time highs as well, almost simultaneously with quits:

Drum’s claim is almost preposterous. The only difference is that the Great Resignation is less “I’m going to contemplate my navel” than “Take this job and shove it! [because I’ve found another one that pays way better].”

Tuesday, April 5, 2022

Coronavirus dashboard for April 5: BA.2 likely only causes a ripple; and on track for record low daily deaths


 - by New Deal democrat

This is a good time to look at the impact - or, better speaking, the lack thereof - of the BA.2 Omicron variant in the US.

Nationwide the 7 day average was 28,961 yesterday:

This is the lowest since last July, and lower than all but about 3.5 months since the end of March 2020. Only late spring 2020 and from mid-May through mid-July in 2021 were lower. In other words, on a national level, there is no BA.2 wave whatsoever yet.

The above graph comes from The NY Times, because the State of Kentucky decided to make an enormous data dump yesterday (like 190,000 cases and 2900 deaths!) yesterday, which is ruining the comparisons on my typical data source, 91-Divoc.

Omitting the South (and so KY), here are the regional breakdowns in cases, as well as NY, NJ, and MA, 3 of the States in the Northeast where there *has* been something of a BA.2 wave:

Note that for the Northeast as a whole, and NY in particular, cases are beginning to level off after 3 weeks. I mention this, because the BA.2 wave in Europe has set the pattern for BA.2 impacts. Below are cases in the 5 big countries in Europe, as well as the EU as a whole:

The BA.2 waves started between the end of February through the first week in March. In every country but Germany, they have already peaked. And Germany is misleading, because they had a data dump 6 days ago. Without it, cases there would only be where they were 2 weeks ago. In other words, in Europe BA.2 waves, where they occurred, have only lasted 2.5-3.5 weeks.

Further, in Europe, the BA.2 waves generally peaked once the variant reached the level of 80%-90% of all cases. With that in mind, here is today’s update from the CDC of variance prevalence in the US, showing that BA.2 is no 72% of all case, having risen about 15% in each of the past two weeks:

And here their map of the regional breakdown:

BA.2 makes up 75% of all cases on the West Coast (where there has been no discernible wave whatsoever) and 84% in the Northeast (New England + NY and NJ).

The Northeast is going to hit 90% or more BA.2 prevalence within a week. Above I mentioned that cases in the Northeast look like they are beginning to level off. I expect a peak to be obvious there within a week to 10 days. The rest of the country is probably about one or two weeks behind that.

Turning to hospitalization admissions, they are at their lowest level ever during the pandemic, averaging 10,744/day during the past week:

And finally, here are deaths, currently averaging 604:

This is the lowest level since last August, and lower than all but 4 months of the past two years.

In conclusion, it now looks nearly certain that, on a national level, at worst there is going to be just a BA.2 ripple, and maybe just a long tail of a very slow decline from 30,000 cases daily.

Further, while deaths have declined at a far slower rate than cases, because cases are down over 96% from their Omicron high, and at peak there were only 2600 deaths a day from Omicron, a 96% decline in deaths would take us down to only about 100 per day. And because victims are typically hospitalized before they succumb, and hospitalizations are at new record lows, this further supports the hypothesis that we are perhaps only a month away from a new record low in daily deaths from COVID.

Monday, April 4, 2022

Manufacturing positive, but no longer red hot; inflation-adjusted construction spending is flat


 - by New Deal democrat

[Programming note: I’ll discuss some of the innards of the March jobs report in further detail in the next day or two, as well as update the Coronavirus dashboard, with particular emphasis on the likely trajectory of BA.2. Since most States don’t bother any more to report on the weekends, there’s no point in doing that today.]

In addition to the jobs report, Friday gave us updates on manufacturing and construction.

The ISM manufacturing index, and especially its new orders subindex, is an important short leading indicator for the production sector. While the index remained positive, its more leading new orders component stumbled.

In March the index declined from 58.6 to 57.1, and the new orders subindex declined from 61.7 to 53.8, the lowest since 2020. Since the breakeven point between expansion and contraction is 50, these remain positive, (note the graph below does not contain the latest numbers):

This forecasts a continued expansion on the production side of the economy through summer.

Meanwhile, construction spending for February rose 0.5% in nominal terms for overall spending including all types of construction, while the leading residential sector also rose 1.1%, both thus making new highs:

Adjusting for price changes in construction materials, which for the first time in six months actually declined, by -1.2%, “real” construction spending rose 1.7% m/m - also the first increase since last August. In absolute terms, “real” construction spending has declined sharply - by -17.8% - since its peak in November 2020,  while “real” residential construction spending has declined -12.6% since its post-recession peak in January of last year:


While total construction spending has declined by more than the -10.4% it did before the Great Recession, the decline in residential construction spending, while increasingly substantial, remains nowhere near the -40.1% decline it suffered before the end of 2007.

There were some positive revisions to the past several months which helped out these comparisons this month, and reinforced that while down, they are not recessionary at this point. But with mortgage rates rising to nearly 5%, I would expect these metrics to deteriorate further later this year, after new home permits, starts, and sales.