Earlier this week, I received an email from the Real Estate editor of an (unnamed) publication. They wrote:
“Just wanted to point out something I read in a Wall St. Journal article that made me think of you immediately. (Home Gauge Climbs Amid Bargains)
“April pending home sales—meaning signed sales contracts—rose 6.3% from March to a level of 88.2, the highest in six months, the National Association of Realtors said. A reading of 100 is equal to the average level of sales activity in 2001. Pending sales typically lead actual sales by a month or two, but may overstate strength because of sales cancellations. Existing-home sales data for May will be released June 26.”
There’s no question in my mind, that the part in bold above was inserted as a result of your persistent efforts to point out how skewed sales data reports are. At the very least, I believe your repeated exhortations on the matter have contributed to more accurate reporting of home sales data on a national level.”
Thank you, JW. Your kind words helped crystallize some of my thoughts on a related “misreportage” — the way the US FinPress reports the Unemployment Rate.
U3 is the “official unemployment rate” according to the BLS website. Due to this, it is the current measure of Unemployment that gets focused upon by most media, and therefore the public. It has, over the years, slowly excluded many of the factors that USED to go into how the US reported unemployment. Hence, there has been a gradual decrease in the Unemployment rate that has occurred regardless of what was happening in the Jobs market.
U3 is now comprised in a way that merely repeating it without a slew of caveats borders on fraud.
U6, on the other hand, is the broadest measure of Unemployment: It includes those people counted by U3, plus marginally attached workers (not looking, but want and are available for a job and have looked for work sometime in the recent past), as well as Persons employed part time for economic reasons (they want and are available for full-time work but have had to settle for a part-time schedule).
To be honest, I do not know what the true Unemployment rate actually is; I believe it is considerably higher than U3 (by 100s of basis points), but I suspect it might be lower than U6.
Here is a modest proposal for all of the poor scribes (like me) who slog through the monthly NFP report:
For the sake of more accurately describing the conditions in the labor
market, let’s begin reporting two measures of Unemployment: U3 as well
as U6.
Its been pretty obvious for sometime that the Financial Media are doing a disservice to their readers by only reporting U3, given how dramatically it understates Unemployment. Indeed, consumer sentiment reports are at deep negative levels that only occur when Unemployment is much than what U3 has been saying. It is painfully obvious that U3 does not paint an accurate view of the Employment situation.
Its way past time to fix that.
Here’s the experiment I propose: Let’s start reporting both, with appropriate descriptions of each. Report U3, add U6, provide monthly and year over year changes. Let the reader see the full picture, via BLS data.
Here is how I would have reported the Unemployment portion of April’s 2008 NFP:
“The number of unemployed persons was little changed in April. U-3, the official unemployment rate as a percent of the civilian labor force, was 4.5%. Itis only modestly elevated from one year earlier, when it was 4.3%.
U-6, the broadest measure of total unemployed*, was also little changed at 8.2%. It is up from 7.9% one year ago.
_______
* U6 includes U3, plus discouraged workers, those working part time who want a full time position, plus marginally attached workers. It is the broadest measure of Unemployment.”
That presents a much more complete and accurate picture of what the employment situation is actually like. This doesn’t require any math, or massaging of the numbers — its just passing along data directly from BLS in a more inclusive and accurate way.
The Bureau of Labor Statistics, to its credit, provides lots more data on inflation,
unemployment, GDP , etc. than ever gets reported in the mass media. Few people seem to want to slog through the numbers
to get at the details of the economic picture — and that’s a shame.
The Financial press can provide a more accurate picture of what’s going on with a slight modification of how they present the BLS releases. Its long overdue.
Let’s take the data and statistics out of the hands of the politicians and spinmeisters. As a nation, we would be better off for it.
~~~
Jobless Claims today out at 8:30am today.
Alternative Measures of Labor Underutilization
Sources:
Table A-12
http://www.bls.gov/news.release/pdf/empsit.pdf
BLS introduces new range of alternative unemployment measures
John E. Bregger , Steven E. Haugen
BLS, October 1995
http://www.bls.gov/opub/mlr/1995/10/art3full.pdf
Measuring Unemployment in the Nineties
Janet L. Norwood and Judith M. Tanur
The Public Opinion Quarterly, Vol. 58, No. 2 (Summer, 1994
http://www.jstor.org/pss/2749542
Historic Annual Employment (Note: footnotes in each year track major changes)
BLS, 1942 to date
http://www.bls.gov/cps/cpsaat1.pdf
HISTORICAL COMPARABILITY Household Data
Employment and Earnings, February 2006
http://www.bls.gov/cps/eetech_methods.pdf
NUMBERS and STATISTICS can be very misleading — sometimes on purpose.
I would LOVE to see truth in reporting…
Barry, how about creating a more realistic inflation measure. Everyone uses the CPI (that measures everything BUT inflation) only because it is what the government produces. Why rely on those who want to manipulate the numbers?
I suggest you (only because it is your field and your well connected) devise a more realistic inflation gauge (or find an existing one) and start encouraging everyone to use it. The more people who use it, the more it will be reported.
Thank you for the great site!
Barry….you are just a cockeyed optimist….never gonna happen because it’s not in the interest of politicos, ceos and media to tell people the truth….bread & circuses is what the people want and the previously cited will accomodate…I leve in the Great american Toxic Dump (otherwise known as New Jersey) and here Gov Corzine has been attempting to deal with years of profligate spending and a refusal by politicians in dealing with fiscal responsibility & his public negatives register at 52%….don’t tell me the truth just make sure I get lower taxes & keep those public services readily available when I need them
“U3 … has, over the years, slowly excluded many of the factors that USED to go into how the US reported unemployment.”
Not only that, but also it’s been an international trend. Remember the bad old, pre-restructuring days of the early 1980s, when we used to speak of ‘Eurosclerosis’ and the hopeless long-term unemployment generated by the low-growth, structurally-rigid welfare states? Most of Europe had unemployment rates ranging from the high single digits up to high teens; I seem to recall Spain going over 20% at one point. Holland had a permanent cadre of unemployed ‘artistes’ whose finger paintings were purchased with public funds and warehoused. And so forth.
Has there been a European job miracle in the past couple of decades? Not really. Instead, just as in the U.S., they modified the unemployment definition to exclude structural unemployment, and bingo! Out pops a respectable, mid-single digits unemployment rate.
I’ll admit, such a restricted definition might be more cyclically sensitive, for policy management purposes. But it gives a misleading glow of health to economies whose job creation and real wage growth are mediocre at best.
Good post and good suggestion.
Good idea Barry. Subtracting u-6 from 100% to gives a much better reflection of the true percentage who are fully employed. Too much we only look at unemployment statistics. We should be focusing on employment statistics. U-3 is incomplete in this respect.
i’ve often heard that U6 isnt reported because any change in unemployment almost always comes in U3; that is, the % of marginally attached, or part time for economic reasons, workers, never really changes, and instead its people that would fall into U3 that go up or down depending on the economic status of the country. that could be an indication of just how bad its been (and for how long), or simply statistical fudging. Either way, more information is more power, and adding the line or two of copy in press reports makes a world of difference.
congrats on #1 blog. keep it up!
I bet reporting the U6 will become commonplace when it comes time to hang a crummy economy on the Democrats — a year from now, everyone will know the unemployment rate is “nearly 9%”.
Lets hope this takes off BR. For the past year I’ve been inserting the U6 number in the comments section of another blog whenever they make an entry on the monthly employment report but I’ve felt it really was just pissing in the wond.
But, things can change – Over 2 years ago, I complained bitterly and persistently to a few news organizations when the government montly housing sales were reported without reference to the margin of error, which usually DWARFED the actual number – e.g. a sales increase of 1.2% with a margin of error of say +/- 5%. It did seem to pay off – people started mentioning the error and or mentioning the YoY number where the margin of error was lower than the actual value.
More power to you.
-K
Prison Population and Unemployment Data.
I’m a big fan of your efforts to untangle these statistical series because the problems you point to in the series also indicate “problems” in other areas.
I hope you will find this addition to the unemployment discussion useful.
In Europe the prison population is included with the unemployed. People over there used to point it out to offset US claims about how lower unemployment numbers reflect the stronger US economic fundamentals.
In the U. S. I don’t think there ever has been an SIC employment code or sub-code for “Prisoners” or “Prison Labor”.
Are prisoners employed, or unemployed?
Prisoners do work: license plates, landscaping and maintenance , agriculture, many other things. Does that make them part of the work force as well as part of the population of labor force age?
Which statistical pigeon hole should they go in?
Should they be in the labor force numbers? If so, should they be counted among the employed or unemployed, or divided between the two?
There are more than 2 million of them.
It would make quite a difference if the prison population was included among the unemployed.
Not to dump on the party, but the definition of initial definition of U6 and the definition incorporated in your asterisk’d footnote don’t seem to match:
Initial U6 definition:
U6, on the other hand, is the broadest measure of Unemployment: It includes those people counted by U3, plus marginally attached workers (not looking, but want and are available for a job and have looked for work sometime in the recent past), as well as Persons employed part time for economic reasons (they want and are available for full-time work but have had to settle for a part-time schedule).
Footnote definition:
* U6 includes U3, plus marginally attached workers, those working part time who want a full time position, plus marginally attached workers. It is the broadest measure of Unemployment.”
“Department of Redundant Redundancies. How may I direct your call?”
PS – congratulations on the kudos from P&I
~~~
BR: That should read “discouraged” — I’ll fix above.
U3 is the “official unemployment rate” according to the BLS website. Due to this, it is the current measure of Unemployment that gets focused upon by most media, and therefore the public. It has, over the years, slowly excluded many of the factors that USED to go into how the US reported unemployment.
Like what? Name something and when it was changed.
Great idea, Barry.
I have one question: What do you base this statement on: “U3 … has, over the years, slowly excluded many of the factors that USED to go into how the US reported unemployment.”
As far as I know, the government hasn’t changed the basic definition of unemployment since the 1940s. It’s always been based on whether an individual has worked for pay or profit, and if not, whether he or she is looking for work.
The BLS says that “Since 1967, to be classified as unemployed, an individual has to have actively looked for work within the last four weeks.”
I agree that reporting on discouraged workers and those forced to work part time paints a more complete picture of the job market, but i don’t think it’s the case that current unemployment rates aren’t comparable by definition to rates in the 1940s, 1950s and later.
Citation, please.
~~~
BR: We’ve addressed this before, but since Rex asked:
The BLS puts all the info out there — but there have been many gradual changes over the year that has served to lower U3.
JFK’s admin created a category of unemployed called “discouraged workers” — unemployed and given up looking. Then in 1994, under Clinton, the Bureau of Labor Statistics redefined the work force to include only that small percentage of “discouraged workers” who had been seeking work for less than a year.
1996: Clinton dropped one sixth of BLS sampling, with a disproportionate number of the dropped households were in the inner cities. (Urban dwelling African-Americans have a higher unemployment rate than the nation at large).
(1983): BLS: Persons in the Armed Forces stationed in the United States will be included in the national labor force and employment totals and thus in the base for the overall unemployment rate. By adding millions of soldiers, sailors and marines to the Labor Pool, unemployment rate was driven down.
(Year?) “Part-Time for economic reasons” used to be considered Unemployed — now, they are in a different category (“Part-Time for economic reasons”) and don’t count in U3
There’s more, but thats off the top of my head . . .
Merrill Lynch’s David Rosenberg hinted at this when he noted that we have never had Sentiment levels this negative when we were not in a recession, and with Unemployment Rate this low.
So either the Public is merely unhappy about other things (first we heard it was Iraq, now we hear its Gasoline) or the Unemployment Rate is failing to capture the full picture.
http://www.bls.gov/opub/mlr/1995/10/art3full.pdf
Glad someone else noticed that, Rex.
You should always be suspicious when someone makes such a nebulous statement containing the words “over the years” and “many of the factors.”
Much as the thought will make readers here cringe, something that bears pointing out is that there are a number of questions even U6 doesn’t answer.
How many people are employed, but outside of the trades/professions in which they’re trained?
To go more macro, how many people could find themselves holding FT jobs if single-payer health insurance were a reality?
How are these stratospheric gas prices affecting employment trends?
“U3 is the “official unemployment rate” …. It has, over the years, slowly excluded many of the factors that USED to go into how the US reported unemployment.”
BR: As much as I appreciate your input when it comes to murky gov statistics, unsubstantiated accusations are not helping. It would be helpful though if you would update this post by giving convincing evidence about the above “factors”.
I concur with Rex and Steve on this one.
It’s a fine idea for economic clarity. But as a member of “the media” who reports this stuff I see it as a pot-kettle and who’s blacker problem. The U3 get’s reported because that’s what people in your business trade on. There’s a widespread impression over here that actual information doesn’t matter, only a tradable number.
With that title, I thought you were going to propose eating unemployed people until the real rate came down to the reported rate.
Let’s leave the minimum amount of weasel room and focus on the “employed-to-population” ratio.
This minimizes the number of variables that the government can lie about.
Barry, you are mostly wrong about your assertions on how BLS has changed the definition of the unemployment rate. Let’s go over them:
1. BLS under JFK created a category of unemployed called discouraged workers.
INCORRECT. Discouraged workers predates the 1960s as a concept. What BLS was criticized for in the 1960s was that some discouraged workers were counted as unemployed based on information they volunteered in the monthly household survey. Most discouraged workers were not counted as part of U5 (the old U3). When BLS piloted a new methodology to treat discouraged workers consistently as not being part of the labor force (and therefore not part of the unemployed), they found it changed the unemployment rate by at most one tenth of a percent, within the margin of error.
2. Under Clinton, BLS redefined the work force to include only that small percentage of “discouraged workers” who had been seeking work for less than a year.
INCORRECT. First of all, “work force” is not a precise term. Let’s use “labor force”, which is the sum of the employed and the unemployed. Discouraged workers have never (conceptually) been included as part of the U-3 unemployment rate. In the mid-1990s, BLS defined discouraged workers to be those who were NOT looking for work (and therefore not in the labor force and not part of the unemployment rate) BUT had looked for work, or been employed, sometime in the past year. Left out of the definition were people not in the labor force, who reported that they would like a job but had given up looking, and who had not been employed or done any jobseeking activity in the past year. This change had NO EFFECT on U-3. What BLS tried to do was to bring some kind of objective measure to differentiate discouraged workers from, for lack of a better term, slackers. Whether you agree that their definition is a good one, it is, at least, an objective measure.
3. (1983): BLS: Persons in the Armed Forces stationed in the United States will be included in the national labor force and employment totals and thus in the base for the overall unemployment rate. By adding millions of soldiers, sailors and marines to the Labor Pool, unemployment rate was driven down.
INCORRECT. BLS started publishing two unemployment rates, U5 (the old U3) and U5A (or something like that), which was slightly lower because it added in resident armed forces to the unemployed. Nobody paid any attention to the armed forces rate, once they stopped laughing. BLS dropped the alternative a number of years later. U5 (the old U3) was never affected by this temporary alternative.
(Year?) “Part-Time for economic reasons” used to be considered Unemployed — now, they are in a different category (“Part-Time for economic reasons”) and don’t count in U3.
INCORRECT. U3 and its predecessor U5 have never included part-time for economic reasons.
You owe me a beer.
And in answer to Chris’s point above about prisoners, BLS measures the civilian noninstitutionalized labor force and unemployment rate. Folks in institutions like prison aren’t counted. It would be nice if we quit locking up so many people of color, but that’s an issue for another blog.
I love it. Accurate information is worth more than gold. You may not be able to keep people from reacting irrationally to information, but surely reacting irrationally to correct information would be an improvement.
Scott,
You ask for details, I provide them, and your response is to claim “Incorrect” but you provide no sources.
Here are my sources — please provide yours.
Measuring Unemployment in the Nineties
http://www.jstor.org/pss/2749542
Historic Annual Employment (see footnotes each year for changes)
http://www.bls.gov/cps/cpsaat1.pdf
HISTORICAL COMPARABILITY Household Data
http://www.bls.gov/cps/eetech_methods.pdf
Fair enough.
1. I’ve already sent you a BLS memo that responded to Kevin Phillips’ charges. The memo describes precisely how the change in methodology in 1967 which clarified their treatment of discouraged workers had only a tiny affect on the unemployment rate. The memo also showed how they treated resident armed forces for ten years or so by producing an alternative unemployment rate. However, the civilian unemployment rate was not affected.
2. The memo you cite at http://www.bls.gov/cps/eetech_methods.pdf does indeed document changes in 1994 in the definition of discouraged workers and part-time employed for economic reasons. What you don’t seem to understand is that neither of these changes affected who was counted as employed, and who was counted as unemployed.
The household survey sorts people into one of three big buckets: employed, unemployed, and not in the labor force. The labor force is the sum of the first two buckets. The unemployment rate is the ratio of the unemployed bucket to the labor force bucket.
Discouraged workers are a subset of those not in the labor force. Changing the definition of discouraged worker didn’t move anybody from one bucket to another, it just changed the sub-buckets in the third bucket.
Part-time workers for economic reasons are a subset of the employed bucket. Changing the definition didn’t move anybody into or out of the employment bucket.
U-3 was not affected. U-6 was affected, however. I do not know if BLS ever went back and reestimated the historical U-6, it looks like they start the series in 1994 on their website.
My point remains, is that U-3 is pretty consistent over time. I don’t know enough about the oversampling of inner cities–I don’t know if when the sample was enlarged again a couple of years ago whether the oversampling was resumed. I would probably agree with you that there was some impact on U-3, I just don’t know how much. The decennial Census has undoubtedly undercounted inner city poor, along with Indians on reservations. Since the household survey is based on the Census, there is probably undercounting of unemployed.
“Indeed, consumer sentiment reports are at deep negative levels that only occur when Unemployment is much than what U3 has been saying.”
So to solve this dilemma we should compare today’s U6 to the U3 of yesteryear? Consumer spending is also at odds with sentiment – do we need to “fix” the spending data too? Doesn’t it seem much more likely that consumer sentiment is the flawed data series?
I don’t believe any of the several U3 changes you described resulted in a recognizable step-function-like change in the data, therefore each of these changes could only have introduced small biases. There’s no way these small biases could add up to the 420 basis point difference between U3 and U6.
The JFK and Reagan era changes can’t be used to explain why U3 is well below ’92 and ’03 peaks.
It just doesn’t make sense that U3 is really that biased and frankly I find ScottB’s argument compelling that none of these factors changed U3 in any measurable way.
“So either the Public is merely unhappy about other things (first we heard it was Iraq, now we hear its Gasoline) or the Unemployment Rate is failing to capture the full picture.”
Isn’t it much more likely that a poll of consumers’ subjective sentiment isn’t comparable over the decades? In today’s Internet age if a butterfly flaps its wings in Portland people read about it on their Blackberries in Topeka. It is said that we are more politically polarized than at any point in history. We have the most unpopular president in modern history. Any one of these factors could make it irrelevant to compare sentiment across decades.