Posted by: Doug Henwood | August 23, 2019

Fresh audio product

Just added to my radio archive (click on date for link):

August 22, 2019 Brian Hioe updates us on the Hong Kong protests • Kavita Krishnan on India’s crackdown in Kashmir

Posted by: Doug Henwood | August 17, 2019

Fresh audio product

Added this to my radio archive the other day but had to run off before announcing it. Click on date for link.

August 15, 2019 Andrew Sernatinger on the recent DSA convention (see article here; mine is here) • Margaret Kimberley on Elijah Cummings, the black misleadership class, Trump, white supremacy, and gun violence

[back after summer vacation and KPFA fundraising break]

Posted by: Doug Henwood | August 3, 2019

Fresh audio product

Just added to my radio archive (click on date for link):

August 1, 2019 Derek Seidman of Litlle Sis on the sleazy ties between Big Enviro and Big Carbon • Marshall Auerback on what happened with all those tax cuts for corporations that already have more money than they know what to do with

Posted by: Doug Henwood | July 18, 2019

Fresh audio product

Just posted to my radio archive (click on date for link):

July 18, 2019 Bernat Tort on Puerto Rico’s economic and political crisis • Sahan Karatasli on Turkey’s economic and political crisis

Posted by: Doug Henwood | July 12, 2019

Fresh audio product

Just added to my radio archive (click on date for link):

July 11, 2019 [back after holiday break] David Adler, policy director at DiEM25, on Syriza’s loss in Greece (Guardian article here) • Joel Whitney, author of this article, on Pablo Neruda and his flight from the CIA

Posted by: Doug Henwood | June 28, 2019

Fresh audio product

Just added to my radio archive (click on date for link):

June 27, 2019 Natalie Shure, author of this article, on what the skyrocketing price of insulin is doing to diabetics • Aaron Medlin, co-author of this paper, on young adults in the job market • Ken Silverstein, proprietor of Washington Babylon, on the reeking worlds of DC journalism and think-tankery

Posted by: Doug Henwood | June 20, 2019

Fresh audio product

Just added to my radio archive (click on date for link):

June 20, 2019 Brian Hioe on the Hong Kong mass demonstrations • Ashley Sanders on climate grief

Posted by: Doug Henwood | June 16, 2019

Fresh audio product

Just added to my radio archive (click on date for link):

June 13, 2019 Rune Møller Stahl on the Danish elections, which the left won but partly by going anti-immigrant • Heidi Matthews, author of this article, on Canada’s genocidal treatment of its indigenous people

Posted by: Doug Henwood | June 6, 2019

Fresh audio product

Just added to my radio archive (click on date for link):

June 6, 2019 Neda Bolourchi on life and politics in Iran • Sean Guillory on Russophobia


Posted by: Doug Henwood | May 30, 2019

Fresh audio product

Just added to my radio archive (click on date for link):

May 30, 2019 Tim Shorrock (Nation page here) on the U.S. conflict with North Korea • Vijay Prashad, director of the Tricontinental Institute, on Modi’s victory in India

Posted by: Doug Henwood | May 24, 2019

Fresh audio product

Just added to my radio archive (click on date for link):

May 23, 2019 Jeff Sparrow on the right’s electoral victory in Australia (article here) • Andrew Cockburn, author of this article, on the spectacular waste in US military spending

Posted by: Doug Henwood | May 17, 2019

Fresh audio product

Just added to my radio archive (click on date for link):

May 16, 2019 [back after KPFA fundraising hiatus] Eric Blanc, author of Red State Revolt, on the teachers’ strikes • Catherine Kaiman, environmental lawyer, on community-based reparations (paper here)

Posted by: Doug Henwood | May 7, 2019

Responding to Rasmus’s response

Jack Rasmus is out with a response to my critique of his analysis of the April U.S. employment numbers. Enlightening Rasmus looks to be a hopeless case, but since there are may be some onlookers who wonder what’s up, here are a few comments. As with yesterday’s post, his comments are quoted and italicized (though the formatting doesn’t show up on an iPhone unless you choose the desktop version—sorry!).

What is significant is that Henwood thinks the CES (Current Employment Survey) is more important and accurate than the CPS (Current Population Survey).

I never said one was more important than the other; they tell you different things. What I did say was that on a monthly basis, the CES (payroll or employer survey) requires smaller moves than the CPS (the household survey) to be statistically significant, since it’s from a much larger sample. And it’s not me who came up with that—it’s the compilers (see question 1 in the FAQs):

“The household survey and establishment survey both produce sample-based estimates
of employment, and both have strengths and limitations. The establishment survey
employment series has a smaller margin of error on the measurement of month-to-
month change than the household survey because of its much larger sample size. An
over-the-month employment change of about 100,000 is statistically significant in
the establishment survey, while the threshold for a statistically significant change
in the household survey is about 500,000.”

But the CES is not really a survey…

It is a survey. See above.

But the CPS is not just a “household survey”; it is also a survey of employment conditions of millions of smaller businesses through the survey of worker households. In fact, it can be argued that, in surveying 110,000 individuals each month, and then rotating and adding more households throughout the year, (roughly doubling the number contacted) the CPS in fact reflects a much larger body of business hiring, layoffs, and thus total employment, than does the CES.

The CES is a survey of jobs; the CPS, of people. They tell you different things. The CES is best for measuring employment, hours, and earnings overall and by industry. The CPS is great for measuring employment and unemployment by age, race, sex, educational attainment, and other demographic characteristics.

Henwood further argues that the CES 142,000 is more accurate because it is checked against the unemployment insurance system. But unemployment insurance has nothing to do with the numbers of employed or unemployed.

The UI system has a lot to do with the numbers of employed. It has less to do with the unemployed; less than a third the officially unemployed are drawing benefits. But the UI system covers almost everyone in formal employment. The quarterly comparisons are part of the Quarterly Census of Employment and Wages (QCEW). Here’s what it’s about:

“The program originated in the 1930s, and was known as the ES-202 program until 2003 when the current QCEW name was adopted. The primary economic product is the tabulation of employment and wages of establishments which report to the Unemployment Insurance (UI) programs of the United States. Employment covered by these UI programs represents about 97% of all wage and salary civilian employment in the country.”

Henwood is confused about the CES and CPS in another important way. There are more than 9 million businesses in the US economy. The 60,000/110,000 CPS survey is a statistically significant survey of employment in those 9 million. The comparison therefore should be 142,000 businesses vs. 9 million businesses. Henwood thus erroneously compares a population (CES 142,000) to a sample (60,000), when the comparison should be a business population (CES 142,000) to a business population (CPS 9 million businesses).

Rasmus’s point here eludes me. Perhaps a few numbers would help him sort things out. There are about 10 million employing establishments in the U.S. A sample of 142,000 of them is 1.4% of the total. There are 128 million households in the U.S.; a sample of 60,000 of them is 0.05% of the total. The CES’s sample is 28 times the share of its total universe as the CPS’s.

But Henwood would have readers believe the CES, with 142,000 businesses, and the 263,000 jobs created last month in that group, is all that matters.

Again, I never said one was more important than the other. In fact, three of the four graphs I included in yesterday’s post were drawn from the CPS.

Like the business press and government politicians, to believe Henwood we should take the 263,000 as the final word of the state of the US job market and forget all the rest.

I never said that either. April’s number is going to be revised next month and again the month after, and yet again next January with the annual benchmarking exercise, in which the monthly survey results are revised in accordance with the near-complete coverage of the employment universe. Revisions are generally small, but the monthly release is not the “final word.”

The BLS publishes a household employment series that’s adjusted to match the payroll concept (mainly by subtracting agricultural and family workers and the self-employed). It’s much more volatile than the establishment count from month to month, but over the long term they tell the same story. Here’s a graph of the two; they’ve only been publishing the adjusted household series since 1994, which is why the graph begins in 1995. Note that while the two lines generally move in tandem, the adjusted household one is far bouncier—even on an annual basis, which smooths out some of the monthly volatility.

Empl yty estab & adj HH

Instead we get jobs created by large businesses (CES) and unemployment from the 9 million population of all businesses. This problem leads to often conflicting data reported by the two sources, CES and CPS.

Everything in these two sentences is wrong. It’s not a matter of opinion or interpretation—they’re just wrong.

Henwood further assumes the role of slavish apologist of government stats by defending the U-3 unemployment rate as the best and final word on the state of the US labor market. He does refer to the U-6 unemployment rate, but unquestionably accepts the government’s current (and chronic) low estimates for the U-6.

Where did I say the U-3 rate is “best and final word”? And how does Rasmus know the U-6 figures are “chronically” low? Does he have a problem with the definition or the method of collecting the raw numbers? Given how little he seems to know about the construction of the employment stats, I’m curious how he knows how they’re wrong.

The U-6 picks up ‘involuntary part time’ employment. (U-4 and U-5 reflect what’s called ‘marginally attached’ and ‘discouraged’. These latter numbers too are grossly underestimated in the official stats).

Again, how does he know? The CPS survey asks people who aren’t working if they’d like to work. Depending on their answers, they’re counted as “discouraged” or “marginally attached.” Here are the official definitions of those two categories (see note to table A-15): “Persons marginally attached to the labor force are those who currently are neither working nor looking for work but indicate that they want and are available for a job and have looked for work sometime in the past 12 months. Discouraged workers, a subset of the marginally attached, have given a job-market related reason for not currently looking for work.” That job-market reason is typically “there’s no work out there.” It doesn’t seem too strict to me to exclude people who haven’t looked for a job for over a year from these categories; others might disagree.

Henwood disputes my claim that the U-3 is essentially an estimate of ‘full time’ jobs and says “No, it refers to work of any kind, not just full-time”. But if that were true, why add on ‘part time’ as the U-6 category separately? If there were part time unemployed in the U-3 and part time in the U-6 there would be likely ‘double counting’ of part time unemployed.

Again, everything here is wrong. People who are working part-time who want full-time work are counted as employed because they are employed—just not as much as they’d like to be. Adding unwilling part-timers to the U-6 measure is a way of accounting for their partial unemployment.

How do you square the 108,000 full time jobs created in the CES with the 191,000 full time jobs lost in the CPS, Doug? What’s your explanation?

I explained that yesterday, and here I am explaining it again.

Given the CPS number showing full time job decline (191,000), and the otherwise CPS rise in part time jobs last month (155,000), in my prior article I suspected that there are more workers taking on second and third jobs. Henwood pooh poohs this and trusts the government numbers on ‘multiple job holders’ showing little change. Once again, trust the government numbers!

Where do you think the inaccuracy arises? The survey of 60,000 is poorly designed? Government economists can’t add? They’re just making shit up because they’re the gummint? The affinity of this line of thinking to right-wing know-nothingism is impressive.

Henwood provides charts that show that Temp jobs (almost always part time) have not been changing for at least the past two decades. As he says, temp jobs have been steady as a percent of the total work force for the past two decades, peaking at 2% of total jobs. “It’s barely changed for five years.” Sure, Doug. No one’s been hiring attempts except through agencies.

Again, not what I said. What I said: “This graph only shows employment arranged through temp firms. For other forms of contingent labor, like contract and on-call work, see here, here, and here. Spoiler: it’s nowhere near as big a deal as the likes of Rasmus would have you believe.”

The Labor Dept. has been covering up the growth of temp jobs since the 1990s. It produced three one-off reports, then George Bush stopped it. Too volatile.

Is Rasmus referring to the contingent worker survey? The latest version was released in June 2018 after a long delay. I wrote it up here. Despite claiming it hasn’t been produced in decades, Rasmus then cites it, but not very accurately.

How the government purposely underestimates labor stats that are embarrassing to it was clearly revealed, yet again, last year in its report on ‘precarious jobs’ (meaning temp, part time, gig, otherwise contingent, etc.). I and others have dissected that official report which claimed the gig economy was insignificant. But it turned out what the report defined as ‘gig’ was only full time uber/lyft drivers. Drivers as second and third jobbers were left out.

Completely untrue. Quoting the report on “electronically mediated employment”: “Electronically mediated workers were more likely than workers overall to work part time…. By industry, workers whose main job was in transportation and utilities were the most likely to have done electronically mediated work, with 5 percent of workers in this industry having done such work. Those employed in professional and business services, information, and other services on their main job were also more likely to do electronically mediated work, at 3 percent, 2 percent, and 2 percent, respectively.”

And from the FAQ: “This work can be done either in person or entirely online. The work could be done as a main job, a second job, or additional work for pay. For example, some people use their own cars to transport others from place to place, having obtained customers through a mobile app (such as Uber or Lyft) that also facilitates payment. Others do household chores or yardwork after finding clients through a website (such as TaskRabbit or Handy). Those who do this work entirely online could take surveys or add descriptive keywords to photos or documents through a platform (such as Amazon Mechanical Turk or Clickworker).”

As far as my suggestion that the April jobs numbers may reflect hiring of census workers, it is true the government to date has not indicated how many hired.

It has indicated how many it has hired, on a page I linked to in yesterday’s post. In case Rasmus missed it, here it is again: 1,000 in May 2018, who were laid off in July, and another 1,000 in February 2019, who are still on the job.

Henwood does give a backhanded concession to me that maybe my point of the 646,000 ‘Not in the Labor Force’ reported number indicates something is going on with the government data underestimating the total actually unemployed by having left the labor force in recent years.

Here’s what I said, after analyzing the persistently low participation rate: “That translates into a shortfall of about 2.5 million jobs, which is equal to the past year’s job growth. So there are more people on the sidelines than a 3.6% unemployment rate suggests, which is almost certainly a lingering effect of the Great Recession, whose damage persists more than a decade after its onset.”

Henwood accepts as accurate the Labor Dept’s estimate of discouraged workers (U-5) as accurate. In November 2007 just prior to the great recession the discouraged workers category represented only .2 of 1% of the labor force. Given the 10 million increase in the labor force since then, it is today still .2 of 1%. Can it be true that the percentage of discouraged workers has not risen at all in the intervening years–given the impact of the great recession, lagging economic recovery for years, and the fact of 7 million have dropped out of the labor force? It makes no sense that there should not be a corresponding increase in the percent of discouraged workers given the changed conditions of the last decade. The government data must be underestimating the discouraged worker category of unemployed (defined as out of work but having given up looking for the past year).

What a mess this is. The share of the population categorized as “discouraged” was around 0.16% just before the Great Recession. It peaked at 0.55% in 2010 and was 0.18% in April 2019. (See below.) Recent levels are low compared to that miserable slump, but are still slightly higher than they were in 2007, before the recession began, or 2000–2001, the strongest job market in recent decades.

The BLS estimates these things by asking lots of people lots of questions. Rasmus, by contrast, just has some feelings.

Also, they have to have looked for work at some time in the past year, not “having given up looking for the past year.”

Discouraged workers

You help them maintain the fiction that the economy is doing great, that jobs are plentiful and well-paid, and we’re all better off than we think.

What I actually said: “An awful lot of Americans have a hard time getting by. Many are working dull, even dangerous, jobs, for stagnant pay and disappearing benefits. There’s strong support for that story. But unless you want to discredit yourself, there’s no point in making a lot of less-than-half-baked claims.” Put this one back in the oven, Jack.

Posted by: Doug Henwood | May 6, 2019

Misreading the latest jobs numbers

Z Communications’ resident statistician Jack Rasmus is out with some fresh disinformation about the economic news. It’s been a while since I took his nonsense apart, so this seems like a good opportunity, since his latest looks to be making the rounds.

The problems start in the first paragraph (Rasmus is in quoted italics, my comments in Roman.)

The just released report on April jobs on first appearance, heavily reported by the media, shows a record low 3.6% unemployment rate and another month of 263,000 new jobs created. But there are two official US Labor dept. jobs reports, and the second shows a jobs market much weaker than the selective, ‘cherry picked’ indicators on unemployment and jobs creation noted above that are typically featured by the press.

There’s actually one monthly report , “The Employment Situation,” though it’s based on two surveys. One survey is of employers, the other of households. The first, also known as the establishment or payroll survey (details here), covers “142,000 businesses and government agencies representing approximately 689,000 worksites.” The other, known as the household survey (details here), covers some 60,000 households. Rasmus doesn’t seem to know that the “second” survey, which he apparently views as more accurate, is the source of the unemployment stats. There’s no cherry-picking involved here.

Since the establishment survey is much larger than its household counterpart, it’s far more accurate. To get a statistically significant change in employment in the establishment survey, you need a move of 100,000 or more; for the household survey, you need 500,000. Numbers from the household survey tend to be far more volatile than the establishment numbers, so one month moves in the household stats are fairly meaningless.

While the Current Establishment Survey (CES) Report (covering large businesses) shows 263,000 jobs created last month, the Current Population Survey (CPS) second Labor Dept. report (that covers smaller businesses) shows 155,000 of these jobs were involuntary part time.

The CES, another name for the establishment survey, doesn’t cover just large businesses. It is a survey, though a very large one, and therefore is subject to error—it might miss new firms, or rapidly growing ones. As a check on that, every three months it’s compared with records from the unemployment insurance system, which covers almost everyone in formal employment. It’s rarely off by more than a tenth of a percentage point or two.

The CPS, another name for the household survey, doesn’t cover small businesses—as its name suggests it covers households. In other words, it asks people if they worked last month, if they wanted to but didn’t, what they did, and questions like that. And yes, the number of people working part-time for economic reasons (the official name for the category) rose by 155,000. Over the last year, however, that number is down by 255,000. Involuntary part-time work accounts for 3% of employment (see graph below). That’s way down from the recession peak of close to 7%, but a bit above its lows in the strong job market of the late 1990s/early 2000s. According to the economist Danny Blanchflower, that stickiness may be a reason that wages aren’t rising smartly despite a low headline unemployment rate—the job market isn’t as tight as it looks. But it’s not as weak as Rasmus wants to believe either.

Part-time econ

This high proportion (155,000 of 263,000) suggests the job creation number is likely second and third jobs being created. Nor does it reflect actual new workers being newly employed. The number is for new jobs, not newly employed workers. Moreover, it’s mostly part time and temp or low paid jobs, likely workers taking on second and third jobs.

It’s bad form to make month-to-month comparisons between the two surveys, because the household version is quite volatile, and even the establishment count is not free of statistical noise. In any case, the popular claim that many people are holding down multiple jobs has no basis in fact. Here’s a history of multiple jobholding; it’s down from 1990s levels, and has been pretty flat since the job market bottomed out in 2010.

Multiple jobholders

Even more contradictory, the second CPS report shows that full time work jobs actually declined last month by 191,000. (And the month before, March, by an even more 228,000 full time jobs decline).

See above re: volatility of the household survey from month to month. Over the year ending in April, full-time employment was up almost 2 million; part-time for all reasons, both voluntary and involuntary, is down 188,000.

The much hyped 3.6% unemployment (U-3) rate for April refers only to full time jobs (35 hrs. or more worked in a week).

No, the U-3 rate—the headline rate that’s most commonly cited—refers to work of any kind, not just full-time. A common, and justified, criticism of that rate is that you only have to have worked an hour in the previous month to be counted as employed, and to be counted as unemployed you have to have actively looked for work. There’s a broader definition of unemployment, the so-called U-6 rate, that accounts for people who want full-time work but can only find part-time and for those who’ve given up the job search as hopeless (“discouraged” workers). That was 7.3% in April, a bit higher than the 6.8% low in late 2000, and a lot higher than the U-3 rate.

Here’s a graph of the U-3 and U-6 rates, along with the old U-7 rate (a predecessor to the U-6 rate which the U-6 replaced in January 1994) and the share of the employed working part-time for economic reasons. The overall picture is pretty good, but not the best ever, as the Tweeter-in-Chief claims.

Unemployment rates

Looked at another way, employers may be converting full time to part time and temp work, as 191,000 full time jobs disappear and 155,000 part time jobs increase.

A repetition of the same pair of bad numbers joined to another canard: the proliferation of temp work. Here’s temporary employment as a share of the total. It rose steadily from the early 1980s through the late 1990s, and has been more or less flat ever since, declining around the 2001 and 2008 recessions, and rising in expansions to a peak of around 2%. It’s barely changed for five years.

Temp work

This graph only shows employment arranged through temp firms. For other forms of contingent labor, like contract and on-call work, see here, here, and here. Spoiler: it’s nowhere near as big a deal as the likes of Rasmus would have you believe.

And there’s a further problem with the part time jobs being created: It also appears that the 155,000 part time jobs created last month may be heavily weighted with the government hiring part timers to start the work on the 2020 census–typically hiring of which starts in April of the preceding year of the census. (Check out the Labor Dept. numbers preceding the prior 2010 census, for April 2009, for the same development a decade ago).

No, they’re not yet hiring temporary Census workers in quantity. They hired 1,000 in February, and none since. That will change in the coming months, but they’ll be noted in the monthly statistical release.

The April selective numbers of 263,000 jobs and 3.6% unemployment rate is further questionable by yet another statistic by the Labor Dept.: 

Rasmus is the one doing the selection of numbers here.

It is contradicted by a surge of 646,000 in April in the category, ‘Not in the Labor Force’, reported each month. That 646,000 suggests large numbers of workers are dropping out of the labor force (a technicality that actually also lowers the U-3 unemployment rate). ‘Not in the Labor Force’ for March, the previous month Report, revealed an increase of an additional 350,000 added to ‘Not in the Labor Force’ totals. In other words, a million–or at least a large percentage of a million–workers have left the labor force.

This is closer to an accurate point, though, again, a monthly move of 646,000 is not conclusive in itself. Over the year ending in April, the labor force (which consists of the employed plus those actively looking for work, that is, the U-3 unemployed) grew by 919,000.

Where Rasmus approaches making a real point, though one that eludes him because of a lack of skill and knowledge, is that a smaller share of the population is working or looking for work than in past expansions. Here’s a graph of the labor force participation rate (the share of the adult population either working or looking for work) and the employment/population ratio (the share of the adult population that is working for pay) over time.


The participation rate fell gently in the early 2000s, then hard in the 2008 recession and its aftermath, and has barely risen since. The employment/population rate fell hard during the recession and, despite a sustained rise, hasn’t returned to its 2007, pre-recession level. (And that peak was below 2000’s all-time high.) Part of this can be explained by the aging population—but not all. If the population had the same age structure as it did in 2000—before the mass of baby boomers started becoming eligible for AARP membership—the employment/pop ratio would be about 3 percentage points higher, closer to 64% than April’s 61% (rounding to the nearest whole numbers). That would still be short of the April 2000 record high of 65%, but 1 percentage point below, not 4. That translates into a shortfall of about 2.5 million jobs, which is equal to the past year’s job growth. So there are more people on the sidelines than a 3.6% unemployment rate suggests, which is almost certainly a lingering effect of the Great Recession, whose damage persists more than a decade after its onset.

Another point: these employment stats don’t cover people in prison, which we have in scandalously great quantity. To put that into perspective, were the 2 million people behind bars all in the labor market and unemployed—which wouldn’t be the case, but we’re just testing things out here—the unemployment rate would be 4.8%, not 3.6%.

Back to Rasmus. An awful lot of Americans have a hard time getting by. Many are working dull, even dangerous, jobs, for stagnant pay and disappearing benefits. There’s strong support for that story. But unless you want to discredit yourself, there’s no point in making a lot of less-than-half-baked claims.

Posted by: Doug Henwood | May 2, 2019

Fresh audio product

Just added to my radio archive (click on date for link):

May 2, 2019 David Palumbo-Liu on the culture of Stanford University, and why it wants to shut its press • Natasha Lennard, author of Being Numerous, on protest, rights, the state, social media, privacy, individuality…

« Newer Posts - Older Posts »


%d bloggers like this: