Meaningless?
Even CBS dis agrees with you, this stuff is really simple
he drop in the unemployment rate comes with an asterisk: while there was a 278,000 gain in employment, t
here was a concurrent labor force decline of 315,000 from October. It would be far preferable for the unemployment rate to drop because the economy is creating over 200,000 per month consistently, rather than due to would-be employees leaving the work force, either because they're retiring or they're simply too discouraged to keep looking for a job. If some of those people resume their job searches, we could see the unemployment rate tick up next month.
November Unemployment: Why the big drop? - CBS News
The civilian labor force dropped another 300,000 this month while the not-in force grew 500,000
These numbers have everything to do with the UE rate changing, in fact there having more of an impact than actual job growth is, twice as much latley
If 50% of the numbers had went into the labor force sense November than has we would have had huge swing in UE% rate the wrong way
Changing Demographics cannot explain all if this in my opinion
Nothing in the article disagrees with my saying that absolute numbers don't mean much. Every value in the article is referenced to something, even "Total Number of Unemployed: 13.3 million (from 13.9 million)" That is a comparison of one level to another. It has some traction, though it is still not quite as meaningful because it doesn't reference how a .6 mil drop compares to the typical change.
I didn't say that the unemployment rate doesn't go down because the labor force decreases. I just said that 6.1 million, as an absolute level, doesn't mean much. That is why I went after a lower boundary, 4 million, so that it is comparable to something. And I see that, oddly enough, it was as high back in the beginning when they started collecting WAJN data. I have no clue what it means, but at least it's a reference point.
So yes, as an absolute number, it's meaningless. As I quoted the entire post, I didn't leave out any comparison value. And none was given. I don't recall it being compared to anything as I scanned the thread.
A drop has some meaning because it's not an absolute number. Even then, it may not be very meaning full for a number of reasons.
It would be preferable "for the unemployment rate to drop because the economy is creating over 200,000 per month consistently". Unfortunately, the economy is not consistent. That is why we can only look at trends, not absolute months or data points.
Have you looked at the month to month or quarter to quarter GDP data? The month to month CPI data? The day to day stock market prices? Why would we expect employment data to have any less variability in it? (That's not a rhetorical question. Some things have physical limitations, some things don't. What, one might ask, is reasonably less constrained by actual physics and what is purely psychological?)
BTW, the CBS articles point is meaningless because she was discussing November and the first consideration coming into the winter is seasonal variability. It usually falls coming into the winter. That the author didn't mention this obvious fact doesn't speak well of her. But then, she's not a statistician and probably hasn't studies statistics, sampling, or surveys in her financial education.
And, her point is well taken, but she really didn't address why it dropped. She made a good point in general, but wasn't exactly committal to addressing why the labor force dropped. It would be a much better point if she was making it in May.
You're making a better point applying her comment to April then she was in December. It generally goes up coming into the summer, so a drop coming into the summer is a bit more questionable.
Yeah, if "some of those people resume their job searches, we could see the unemployment rate tick up next month." Absolutely.
But, a single month change doesn't mean much. The labor force and employment varies from month to month, up and down. Since 1979, the March to April change in the labor force has fallen by as much as -.44% and risen by as much as -0.36%. There is random variability. This March to April change was -0.27%. That is well within the range of March to April changes.
WAJN does attempt to gauge the reason that they aren't looking. But, in further examination, it includes things that we really aren't considering for our interest, like "Reasons other than discouragement". It is reasonable to take those out first.
And it doesn't include a category for "I retired" or "I retired but am still looking for work." Unless the CPS surveyed those people that left the workforce and asked them if they retired, there is no way to conclude that they left the workforce because they retired. We really can't use the number of people retiring with the CPS data, not in any direct sense. Not without very detailed understanding of the two survey methods.
I sure would like it if the CPS included a category for retired and not working. I saw some mention that the labor force is 16 to retirement age of 65, but I haven't confirmed that and it is in conflict with other information. That the CBS author put "discouraged" and "retired" into the same sentence makes her suspect.
In general, estimates from one survey cannot be reliably subtracted from estimates of another survey. The general trends can be compared. But the estimates, like how many people have retired and how many people have are in the labor force, cannot be subtracted.
They are estimates, they have sampling error. Combining them combines the sampling error. The data and estimates within a survey are internally consistent. They can be subtracted and added because they are tied together in each questionnaire, each data point. Estimates between different surveys are not guaranteed to be consistent.
And the groups that are being counted in two separate surveys are not necessary mutually exclusive. I think that was someone's point earlier, that "retired" doesn't mean not working. I would expect quite a few people that had expected to retire are not and that there are others that lost a job and decided to just go ahead and retire. There is just no way to know what the retirement numbers mean in terms of the CPS.
We get a lot more traction out of looking at the subcategories under WAJN then we do trying to compare the numbers from completely separate surveys. At the very least, seeing as the WAJN data is part of the CPS, that is where we should start.
Even then, we need something to gauge it against, like an average and the variability.
Personally, I don't put much stock in a single month's unemployment rate. It just doesn't mean much, up or down. It's the trend that is important.
Look, it's like having a crappy speedometer that bounces all over the place. A snapshot doesn't tell us if we are doing 45, 50, or 55 mph.
You and I are talking apples and oranges
Job creation and the UE rate at this time are 2 different things
I mean no dis respect to your effort. The reason the UE is going down has allot more to do with people leaving the work force than jobs being created
This is what CBS was saying even thought there spin (or not) made it seen retirement and not people giving up is the reason
The not-in column is growing by this month 5 times what the job creation numbers did
(+-)
This has everything to do with the UE going from 9% to 8% as we find it will below in November
Not directed at you, but if one keeps repeating a 1/2 truth it will become the truth
Saying there is job creation is true (although it is anemic), saying the UE rate is falling is not even a 1/2 truth
People in general only care about that 5-6-7-8-9-10% number
Kudos to CBS at least explaining the why once any-way
Except CBS didn't explain it. All they did was suggest it. I saw no hard facts that demonstrated that this was the case in November.
And, in examination of the past four months, April is the only month that LF dropped. Suddenly, the reason for the increase in UE is because retirees have left the LF in droves in April?
To be clear, the convo is convoluted and difficult to follow as it runs through pages of thread as it bounces back and forth about this "how many people are retiring" thing and if it can be used. It's hard enough to keep track of when you're in the convo. I am not willing to study the entire thing, being sure I know who is on what side of it all. Rather, I will capture the general sense of the idea then just address them objectively, as a concept as a whole.
Still, we are not talking "apples and oranges". I see the convo talking about an apple while I am talking about the entire crate of apples. If, after two years of collecting apples, the last apple happens to be green and not red, it doesn't make the entire box of apples green. I haven't examined every apple, to see how many are green, but in the over view, the color of the crate of apples is basically red, not green.
If there is a question as to whether a particular number is confounded by some underlying issue, the way to address it is to go back to the original CPS levels, not try and cram in some estimate of the number of retired folk from a totally unrelated survey. All that does is take one survey and estimates, that has inherent error, and add another survey with inherent error. Going that route when a similar answer can be obtained by looking more closely at all the CPS data is, frankly, just dumb.
Here is the basic levels for the past four months. CPOPNSA is Civilian Non-institutionalized Population, not seasonally adjusted. LF is labor force. EmpLevel is employment level.
Date CPOPNSA LFNSA. EmpLevelNSA
4/1. 242784. 153905. 141995
3/1. 242604. 154316. 141412
2/1. 242435. 154114. 140684
1/1. 242269. 153485. 139944
The entire concept is a bit of a mute point.
The employment level has gone up. It may be that it isn't great, but it went up. Still, we expect that it will, we are coming into summer. And it isn't all that great of an increase In the overview, I'd consider it to be just flat until it shows a real trend.
The problem with the other direction is it assumes that the only change in the numbers is due to those that left (or entered). The fact of the matter is, it could be ten million people left and another ten million entered. It could be that 20 mill left and 20 mill entered.
Just as well, there is natural variability and seasonal variability. Unless the change is significantly greater then the variability, it doesn't men much.
The only meaning that these numbers have is if the change is particularly large from one month to the next, or in terms of the general trend.
The reason you see "apples and oranges" is because I am going back to the basics, so that there is some reasonable and correct footing as opposed to jumping into some questionable method without examining the underlying issues.
And in doing so, I find that the basic numbers don't support the premise that it's all about retirees. The level of employment increased. Right there, it suddenly isn't so simple as just subtracting some estimate of the number of retirees that decided to not take jobs.
Unemployment is just one of dozens of statistics. And under normal circumstances, it's okay.
Here is a more complete set of statistic. (date sorted differently, sorry).
Date.......LFtoCPOP...EmpToLF.... UnempToLF.... EmpToCpop.... UnempToCpop
1/1/2012... 63.35%.... 91.18%.... 8.82%......... 57.76%....... 5.59%
2/1/2012... 63.57%.... 91.29%.... 8.71%......... 58.03%....... 5.54%
3/1/2012... 63.61%.... 91.64%.... 8.36%......... 58.29%....... 5.32%
4/1/2012... 63.39%.... 92.26%.... 7.74%......... 58.49%....... 4.91%
LF to CPOP went down. Emp to LF went up. Unemp to LF went down. Emp to CPOP went up. Unemp to CPOP went down. LF Level went down. CPOP went up. Emp went up.
Sure, LF went down. But just because one way that it might transpire is this retiree concept doesn't mean it did. That is a big "if". And the fact that other statistics went in the "right" direction, along with just employment, kills the retiree concept as being a singular and simple explanation. Maybe it's part of it. So what, it's always part of it. April isn't some how different.
So people left the labor force for permanent retirement. That's going to affect employment to labor force as much as it affects unemployment to labor force. And in the same manner. And they both didn't move in the same direction.
(I'm still trying to figure out how LF going down is suppose to simply decrease unemployment. I don't care to reread the ten pages of posts. But, in simple terms, a lower denominator makes a bigger number. Retirees would have to be all unemployed and leave LF and Unemployed at the same time so that unemployed level fell more then LF level to get the effect. Right?)
There is simply no way to determine what retirees did to the numbers unless the CPS included those questions in the survey.
It would be great to figure out exactly how much of what did what. I love the idea.
Complaining that the unemployment number, as reported, is a 1/2 truth then picking just the LF level to prove it is a 1/5th truth.
And the fact of the matter is that if all you have time for, like a typical manager, then the unemployment rate is "good enough". Every manner of engineer and statistician has found that managers and the general public want one number. And they have been continuously frustrated by the fact that they tried to explain it earlier, then been berated later when the "complication" became important.
I cannot count the number of times I have examined some statistics or rule of thumb, only to find out that it is a bit more complicated then I otherwise believed.
It would be great if there was one single number that could capture it all. That is what professionals do, attempt to find some singular number that will explain it all, all of the time. Sometimes there is, but it is so complicated that no one wants to use it. Sometimes there simply isn't. I've been looking for one that takes out the underlying issues. I haven't found it. And I suspect that there isn't one. That is why the BSL publishes so many. It all depends on what we want to know.
The fact of the matter is that there is an average increase, variance both general and seasonal. The fact of the matter is that each of them suffers from some underlying use if, not understood, may lead to a mis-interpretation.
I cannot count the number of times I have thought some statistic always meant something, only to discover that it wasn't exactly that. It was close, but not always under all conditions. This is why, especially of late, I have gone back to the fundamentals. But that I have to doesn't mean someone is "lying". That I discovered that the number doesn't mean exactly what I thought doesn't mean someone is "lying". It means I have learned something.
Of the five statistics and three levels that I posted above, the only one that is at odds is LF and LF to CPOP. Those are the basic stats. I am not surprised that some stats are a bit off, given the conditions. I am sure that there are busy little beavers at the BLS taking a few moments out of the day to revisit the issue.
But I see no reason to go after some other marginally related survey, itself an estimate, without examining the full set of CPS data and statistics first. If there is something to be had, it's more likely going to be in the data that is already consistent as a whole. If it's not, then it's worth going after something more complicated.
And the retiree concept is not "just that simple". If it was, I wouldn't have to be reading ten pages of posts. If it was "just that simple", I'd have already posted proof that it is the retirees that account for the change in LF. I may yet, if I find it or someone presents something that gets close enough to actually being provable. We may even find that, should the structure of the economy change and become apparent, the BLS will start asking that very question.
Until then, unfortunately, it isn't 'just that simple'. It's "intuitive", it's "possible", but it isn't, "just that simple."