EDIT: turns out I'm wrong about the nuance of this error so adjusting the below text...
That's because this error isn't an error due to some analyst doing something wrong. What is happening is that people are being asked questions about their employment status and are answering in a way that gets them classified as "absent" instead of "unemployed". It's not really clear from the report if this is due to interviewer training or the way questions are being answers. Regardless, we call this kind of misattribution a `misclassification error`. To avoid introducing even worse errors into the data (namely, selection bias), the analyst has to just assume that the people who answered wrong, in fact answered correctly (this is called Intention to Treat Analysis [0]. What the BLS report is stating is that if they did treat those who said "absent" as "unemployed" you would see an extra 3% unemployed.
It’s weird to describe it as cooking the books when they are being completely transparent about the issue that is occurring. If they wanted to cook the books all they had to do was leave out the disclaimer.
When the U.S. government’s official jobs report for May came out on Friday, it included a note at the bottom saying there had been a major “error” indicating that the unemployment rate likely should be higher than the widely reported 13.3 percent rate.
The special note said that if this “misclassification error” had not occurred, the “overall unemployment rate would have been about 3 percentage points higher than reported,” meaning the unemployment rate would be about 16.3 percent for May.
...
Some took this as a sign that President Trump or one of his staffers may have tinkered with the data to make it look better, especially since most forecasters predicted the unemployment rate would be close to 20 percent in May, up from 14.7 percent in April. But economists and former BLS leaders from across the political strongly dismissed that idea.
...
Economists say the BLS was trying to be as transparent as possible about how hard it is to collect real-time data during a pandemic. The BLS admitted that some people who should have been classified as “temporarily unemployed” during the shutdown were instead misclassified as employed but “absent” from work for “other reasons.”
...
This problem started in March when there was a big jump in people claiming they were temporarily “absent” from work for “other reasons.” The BLS noticed this and flagged it right away. In March, the BLS said the unemployment rate likely should have been 5.4 percent, instead of the official 4.4 percent rate. In April, the BLS said the real unemployment rate was likely about 19.7 percent, not 14.7 percent.
Economists said the big takeaway is that it’s hard to collect real-time data during a pandemic and that while the unemployment rate remains high — likely more than 16 percent — it has declined a little from April.
...
Instead of focusing on possible Trump interference, many economists wish people would focus on the fact that 21 million Americans are currently unemployed and over 2 million have permanently lost their jobs.
The situation remains dire, they say, even after a few jobs returned in May as the economy reopened.
---
Given that the NASDAQ hit an all-time high today (!) based largely on the unemployment number being better than expected, one wonders how the market will react to the correction on Monday.
Market traders don't read the report, they just see the headline and get excited.
People ascribe way too much wisdom to stock traders. Remember back in March when everybody was buying buckets of stock in the wrong company called Zoom? Zoom the Chinese parts manufacturer's price went up 2000% before the SEC forced it to be delisted because that's the only way to stop the lemmings on Wall Street.
umm, people who move millions of dollars are quite sophisticated. I am sure the retail investors like me don't dissect the report, but we hardly move the market. Institutional investors aren't as sloppy as me.
Sometimes they are. But there are definitely companies that analyse the shit out of every piece of data they can get their hands on. I think part of the problem is that getting things "right" is just really really hard.
I am sure there are such investors, but our companies are already crossing a trillion dollar market cap, you need to have a few billion dollars to move market significantly. As the amount increases, so does the due dilligence.
While a “market trader” may not, the large banks have an army of analysts who spend all their time far beyond 40 hours of work reading everything they can about particular industries or market drivers. They know more about this stuff than you or I and that is why they make money ahead of anyone you know. A bunch of MBAs read that report cover to cover and delivered a summary to the trading exec before any of us even had time to flip to the first page.
The whole market smells suspect. It’s an orgy of Fed induced moral hazard where something so misleading could rally the market which had already recovered to near all time highs.
I suspect a lot of the surprise had to do with businesses recalling furloughed workers to essentially do nothing to quality for PPP loan forgiveness.
So in reality this likely is just showing PPP working as intended. The bigger question is if demand is still depressed on reopening if these companies will just have to lay off the employees anyway once the loan is forgiven.
the other thing is the initial wave of laid off workers was heavily in service and healthcare which were most impacted by lockdowns, so makes sense those would return on reopening. Bigger question is if there will be a second wave of layoffs for white collar workers due to persistently depressed demand even on reopening.
The relevant section from the actual report is arguably clearer than the WaPo article:
| However, there was also a large number of workers who were classified as employed but |
| absent from work. As was the case in March and April, household survey interviewers |
| were instructed to classify employed persons absent from work due to coronavirus- |
| related business closures as unemployed on temporary layoff. However, it is apparent |
| that not all such workers were so classified. BLS and the Census Bureau are |
| investigating why this misclassification error continues to occur and are taking |
| additional steps to address the issue. |
| |
| If the workers who were recorded as employed but absent from work due to "other |
| reasons" (over and above the number absent for other reasons in a typical May) had |
| been classified as unemployed on temporary layoff, the overall unemployment rate |
| would have been about 3 percentage points higher than reported (on a not seasonally |
| adjusted basis). However, according to usual practice, the data from the household |
| survey are accepted as recorded. To maintain data integrity, no ad hoc actions are |
| taken to reclassify survey responses.
18 comments
[ 1.9 ms ] story [ 54.7 ms ] threadSo was the April report, with the exact same error.
That's because this error isn't an error due to some analyst doing something wrong. What is happening is that people are being asked questions about their employment status and are answering in a way that gets them classified as "absent" instead of "unemployed". It's not really clear from the report if this is due to interviewer training or the way questions are being answers. Regardless, we call this kind of misattribution a `misclassification error`. To avoid introducing even worse errors into the data (namely, selection bias), the analyst has to just assume that the people who answered wrong, in fact answered correctly (this is called Intention to Treat Analysis [0]. What the BLS report is stating is that if they did treat those who said "absent" as "unemployed" you would see an extra 3% unemployed.
0. https://en.wikipedia.org/wiki/Intention-to-treat_analysis
Excerpting some key paragraphs:
---
When the U.S. government’s official jobs report for May came out on Friday, it included a note at the bottom saying there had been a major “error” indicating that the unemployment rate likely should be higher than the widely reported 13.3 percent rate.
The special note said that if this “misclassification error” had not occurred, the “overall unemployment rate would have been about 3 percentage points higher than reported,” meaning the unemployment rate would be about 16.3 percent for May.
...
Some took this as a sign that President Trump or one of his staffers may have tinkered with the data to make it look better, especially since most forecasters predicted the unemployment rate would be close to 20 percent in May, up from 14.7 percent in April. But economists and former BLS leaders from across the political strongly dismissed that idea.
...
Economists say the BLS was trying to be as transparent as possible about how hard it is to collect real-time data during a pandemic. The BLS admitted that some people who should have been classified as “temporarily unemployed” during the shutdown were instead misclassified as employed but “absent” from work for “other reasons.”
...
This problem started in March when there was a big jump in people claiming they were temporarily “absent” from work for “other reasons.” The BLS noticed this and flagged it right away. In March, the BLS said the unemployment rate likely should have been 5.4 percent, instead of the official 4.4 percent rate. In April, the BLS said the real unemployment rate was likely about 19.7 percent, not 14.7 percent.
Economists said the big takeaway is that it’s hard to collect real-time data during a pandemic and that while the unemployment rate remains high — likely more than 16 percent — it has declined a little from April.
...
Instead of focusing on possible Trump interference, many economists wish people would focus on the fact that 21 million Americans are currently unemployed and over 2 million have permanently lost their jobs.
The situation remains dire, they say, even after a few jobs returned in May as the economy reopened.
---
Given that the NASDAQ hit an all-time high today (!) based largely on the unemployment number being better than expected, one wonders how the market will react to the correction on Monday.
People ascribe way too much wisdom to stock traders. Remember back in March when everybody was buying buckets of stock in the wrong company called Zoom? Zoom the Chinese parts manufacturer's price went up 2000% before the SEC forced it to be delisted because that's the only way to stop the lemmings on Wall Street.
I suspect a lot of the surprise had to do with businesses recalling furloughed workers to essentially do nothing to quality for PPP loan forgiveness.
So in reality this likely is just showing PPP working as intended. The bigger question is if demand is still depressed on reopening if these companies will just have to lay off the employees anyway once the loan is forgiven.
the other thing is the initial wave of laid off workers was heavily in service and healthcare which were most impacted by lockdowns, so makes sense those would return on reopening. Bigger question is if there will be a second wave of layoffs for white collar workers due to persistently depressed demand even on reopening.
But there isn’t nearly that amount of work needed. So they are asking employees to report the high numbers anyway.
This effectively shifted unemployment insurance benefits to business benefits that are still sponsoring (at least in part) zero utility.
It’s a great smokescreen. And fwiw, it is probably a better Posture to have workers “ready to go” should there be something for them to do.
But make no mistake this is not business as usual. It is more like a shell game for stats and monetary distribution.
HCA for example has pledged to not lay off anyone.
| However, there was also a large number of workers who were classified as employed but | | absent from work. As was the case in March and April, household survey interviewers | | were instructed to classify employed persons absent from work due to coronavirus- | | related business closures as unemployed on temporary layoff. However, it is apparent | | that not all such workers were so classified. BLS and the Census Bureau are | | investigating why this misclassification error continues to occur and are taking | | additional steps to address the issue. | | | | If the workers who were recorded as employed but absent from work due to "other | | reasons" (over and above the number absent for other reasons in a typical May) had | | been classified as unemployed on temporary layoff, the overall unemployment rate | | would have been about 3 percentage points higher than reported (on a not seasonally | | adjusted basis). However, according to usual practice, the data from the household | | survey are accepted as recorded. To maintain data integrity, no ad hoc actions are | | taken to reclassify survey responses.
Link found in that WaPo article, so credit due for that: https://www.bls.gov/news.release/empsit.nr0.htm