Today marked yet another week of glacial progress in initial jobless claims, at levels worse than the worst weekly levels of the Great Recession.
On a non-seasonally adjusted basis, new jobless claims rose by 5,312 to 804,307. After seasonal adjustment (which is far less important than usual at this time), claims fell by 9,000 to 840,000, another new pandemic low. The 4-week moving average also declined by 13,250 to a new pandemic low of 857,000:
Here is a close-up of the last two months highlighting the glacial progress in initial claims during the past 7 weeks:
Here is my weekly update on the 2020 elections, based on State rather than national polling in the past 30 days, since that directly reflects what is likely to happen in the Electoral College. Remember that polls are really only nowcasts, not forecasts. They are snapshots of the present; there is no guarantee they will be identical or nearly identical in early November.
A special reminder: these polls were all taken before Trump’s COVID-19 diagnosis, and only a few were entirely after the debate.
The best news for Trump, and the worst news for Biden, is that despite an awful debate performance, Trump’s approval *rose* 0.7% over the past week, while his disapproval dropped 0.4%. He is now near the upper end of his normal range of approval going back 3.5 years:
This may be a reaction to his nomination of a Judge for the Supreme Court who is on record saying that both Roe v. Wade and the ACA should be overturned. It is way too soon for Trump’s COVID-19 diagnosis to be showing up in these polls.
661,000 million jobs gained. The gains since May total a little over half of the 22.1 million job losses in March and April. The alternate, and more volatile measure in the household report was 275,000 jobs gained, which factors into the unemployment and underemployment rates below.
U3 unemployment rate fell -0.5% from 8.4% to 7.9%, compared with the January low of 3.5%.
U6 underemployment rate fell -1.6% from 14.2% to 12.8%, compared with the January low of 6.9%.
[UPDATE: Correcting error] Those on temporary layoff decreased 1.523 million to 4.637 million.
Permanent job losers increased by 345,000 to 3.756 million.
July was revised upward by 27,000. August was also revised upward by 118,000 respectively, for a net gain of 145,000 jobs compared with previous reports.
Leading employment indicators of a slowdown or recession
I am still highlighting these because of their leading nature for the economy overall. These were positive:
the average manufacturing workweek rose 0.2hours from 40.0 hours to 40.2 hours. This is one of the 10 components of the LEI and will be a positive.
Manufacturing jobs rose by 66,000. Manufacturing has still lost -647,000 jobs in the past 7 months, or 5% of the total. A little over half of the total loss of 10.6% has been regained.
Construction jobs rose by 26,000. Even so, in the past 7 months -394,000 construction jobs have been lost, 5.2% of the total. About 1/3rd of the worst loss of 15.2% loss has been regained.
Residential construction jobs, which are even more leading, rose by 6,600. Even so, in the past 7 months there have still been -14,000 lost jobs, or about 1.7% of the total.
temporary jobs rose by 8,100. This is a *drastic* slowdown from the gains of the past few months, which typically were over 100,000. Since February, there have still been -463,800 jobs lost, or 15.8% of all temporary help jobs.
the number of people unemployed for 5 weeks or less rose by 271,000 to 2.552 million, compared with April’s total of 14.283 million.
Professional and business employment rose by 89,000, which is still -1.386 million, or about 6.4% below its February peak.
This morning’s third economic release is August construction spending. I track this because it is part of the important long leading housing sector. While it isn’t nearly as leading as housing permits or starts, it is the least volatile of all of the data.
And the news was good, in particular for private residential construction spending, which increased by 3.7%:
Residential construction spending is only -0.6% below its February peak.
This is more confirmatory evidence that the economic *wants* to expand, and quite smartly, in 2021, if the pandemic is brought under control
Another week of glacial progress in initial jobless claims, at levels worse than the worst weekly levels of the Great Recession.
On a non-seasonally adjusted basis, new jobless claims fell by -40,263 to 786,942. After seasonal adjustment (which is far less important than usual at this time), claims fell by -36,000 to 837,000, another new pandemic low. The 4 week moving average also declined by -11,750 to a new pandemic low of 867,250:
Continuing claims declined on a non-adjusted basis declined by -1,020,192 to 11,410,703. With the seasonal adjustment, they declined by -950,000 to 11,767,000. Both of these numbers are also new pandemic lows:
It’s a very slow most-of-the-week for economic data. Except for house prices, nothing much gets reported until Thursday, when a slew of data (personal income, spending, ISM manufacturing, construction, and jobless claims) all gets reported at once. And then Friday is the last jobs report before the election.
I’ll update the Coronavirus dashboard tomorrow or Wednesday. In the meantime, here are a few random notes for your Monday morning.
1. YouGov has now premiered their own election model, which gives Biden 351 Electoral votes as the median forecast. The model does not give probabilities.
This is a light week for economic data, so don’t be surprised if I take a break for a day or two.
In the meantime, I wanted to post a few graphs that I think give a good “macro” view of the Presidential election this year, and how it is different from 2016.
Aside from the fact that Trump now has a nearly 4 year record, and Presidential elections are usually a referendum on the incumbent President or party, the simple overriding fact is that Biden is viewed much more favorably, or perhaps much less *un*favorably, than Hillary Clinton was in 2016.
To refresh your memory, here are a couple of graphs showing how (dis)liked both Trump and Hillary Clinton were in 2016:
Both were historically more disliked than any other previous candidate of either party. Almost 50% of likely voters *strongly* disliked her, and majority did dislike her:
NOTE IMPORTANTLY: I am not making a value judgment that Hillary Clinton was dislike*able*. I am simply reporting that she was in fact historically disliked. When right in the middle of ongoing early voting, Comey insinuated that she might be a crook, that’s all it took to give Trump his inside straight win in the Electoral College.
As a result of the fact that *both* candidates were historically disliked, there were wide swings in voter intentions throughout 2016, depending on which candidates problems were dominating the front page. Here’s the static-y day by day view:
Note that there were several brief periods where Trump actually led.
Here’s a smoothed out view of the year:
And here is the forecast probability of winning:
As I said above, there were wide swings in the race.
Before I go further, let me add into the mix this graph of the average deviation in polls depending on the time to the election:
As I write this, we are 43 days from the election, meaning that there’s an average of about 2.3% error in polling compared with the actual result.
Now let’s compare that with 2020 so far. Here’s Nate Silver’s summary of national polling for the past 6 months:
This has been a remarkably steady race for months on end. Even when Trump got the most benefit from his brief “rally round the flag” moment in early April, he still trailed Biden by 3.5%, 44.1% to 47.6%. For the past month, Biden has polled very close to 50% in either direction consistently.
In short, there’s every reason to believe that the polling this year is closer to the ultimate result than has historically been the case. I’m not even sure the thermonuclear bomb of Justice Ginsburg’s death is going to have much of an effect on the polling; at most I suspect it will increase the population of likely voters.