BLS | "Phony Numbers" on Their Way?

"Phony Numbers" Yesterday? Tomorrow? 

Jan. 23, 2017–The Trump Team attacked the BLS in 2016 for issuing misleadingly favorable numbers on unemployment. Trump called them "phony numbers".

In the second-last paragraph of his Op-Ed today, Paul Krugman wonders whether the Trump White House will be leaning on the BLS to do exactly what they were accused of under Obama.

The payroll job numbers are fairly hard to manipulate. They are based on filings by employers for unemployment insurance. If states with Republican governors wanted to influence the numbers they might be able to do this for a single month–for example by changing deadlines or filing procedures. But it is hard to see how they could do this easily over several months or years.

Meanwhile, blue-state governors–and some red-state governors–would be on the lookout for misuse of their data. Economists and statisticians are often alert when numbers that don't make sense and in the tradition of whistle-blowers they sometimes they speak up regardless of the consequences.

The household employment numbers are based on a sample of households surveyed by the Census Bureau, which reports to the Secretary of Commerce. When I was actively working with the numbers the Census work was paid for by a contract with the Commissioner of Labor Statistics, who reports to the Secretary of Labor; I have not heard of any change in this procedure.

One has to have faith that any secret changes in collecting these numbers, which have long been subject to scrutiny, would be difficult because someone would soon speak out in private or public.

JOBS | Two Job Numbers for Trump to Beat (Updated Jan. 6, 2017)

Trump visits Obama at the White House. 

It is understood among the cognoscenti that the Democrats managed the economy better for most Americans than the Republicans in the new millennium.

The problem for the Democrats in the 2016 election was voters who switched parties between 2012 and 2016. They were predominantly non-college-educated, from the least-well-educated states.

They picked up on the plain-language appeals and promises of candidate Donald Trump. The Democratic arguments were pitched to their college-educated base.

Perhaps the Democratic appeal to voters should have been based on a smaller number of performance indicators. President-elect Trump has promised to increase jobs. Two broad indicators–as opposed to anecdotal evidence from individual companies–might suffice to measure the success of his administration.

We can even use the BLS November job numbers [Jan. 6, 2017: see below for update, which changes little] for clear and independent benchmarks against which the economic performance of our presidents can properly be measured. The final report on the Obama Administration will be issued on January 6, subject to revision, but the December numbers, based on November reporting (and prior months for their seasonal adjustment), are not likely to change much from November.

Two Key Indicators

Here are the two key long-term numbers against which Obama's economy can be compared with the Bush economy of eight years ago and the Trump economy of two and four years hence:

1. Unemployment =  a rate of 4.6 percent in November

This is arrived at by dividing the number of unemployed, 7.4 million in November, of whom 1.9 million have been unemployed for 27 or more weeks, by the labor force (employed + unemployed), 159.5 million.

This compares with 7.3 percent in December 2004, the last month of the George W. Bush administration. (The numbers are easily found by punching into Google the two words  Unemployment and FRED. This will take you to the super-user-friendly St. Louis Fed database, God bless them.) That is a reduction of 2.7 percentage points. This compares with an increase of 3.4 percentage points during the Bush 43 administration and a decrease of 3.5 percent during the Clinton administration:

• Clinton vs. Bush 43: Better by 3.5 - (-3.4) = 6.9 percentage points.
• Obama vs. Bush 43: Better by 2.7 - (-3.4) = 6.1 percentage points.

2. Employment-population ratio = a rate of 59.7 percent in November.

This number is arrived at by dividing the number of civilian noninstitutional employed, 152.1 million, by the civilian noninstitutional population, 254.5 million.

This number is solid for long-term comparisons because it is not affected by answers to the unemployment survey. The labor force participation rate is dependent on the unemployment rate in the definition of the labor force. The employment-population ratio is not affected by any long-term change in the definitions of the unemployed or in the conduct of the monthly surveys of the labor force.

In December 1992 when President Clinton came to office, the employment-population rate had been falling and was at 61.4 percent. It rose during his administration to 64.3 percent, an increase of 2.9 percentage points. Under G. W. Bush, the rate fell by 3.3 percentage points to 61.0 percent. Under Obama the rate fell further to 59.7 percent, a drop of 1.3 percentage points. So here is the record of this measure:

• Clinton vs. Bush 43: Better by [2.9 - (-3.3)] = 6.2 percentage points.
• Obama vs. Bush 43: Better by [3.3 - 1.3] = 2.0 percentage points.

On both measures, the last two Democratic administrations outperformed the Bush 43 administration, by a lot.

Update, Jan. 6, 2017

Here are the final numbers for November, which arguably should still be the baseline, seasonally adjusted, if the incoming President wants to take credit for changed expectations in December (or his opponents do). Here also are the December seasonally adjusted numbers. Either way, from here on, it is President Donald Trump's baby–TrumpCare, TrumpEconomy and all.

JOBS | "Nonemployment"

Data before the 1960s show the importance of women
 entering the workforce in far greater numbers. Fewer men
have worked since then. The departure from the workforce
 of many women since 2000 is a puzzle.

The New York Times series by its Upshot staff on "nonemployment" is another reminder of what a great newspaper can do to put in front of the public the data and research that a democratic society needs to make decisions about public policy.

I have posted at least four times on the subject in the last year and a half.

The term "nonemployed" appears to have been formally introduced in 1997 by two University of Chicago economists as a broader category of nonworking people than the unemployed. They aren't working but are self-described in surveys by the Census Bureau for the Bureau of Labor Statistics as currently available for work and actively looking within the four weeks prior to the survey.

Two articles in the Times series on nonemployment have appeared. The first, by David Leonhardt, introduces the topic and stresses "the Decline of Work" as the theme. The central number he puts in front of us is that back in the 1960s only 5 percent of men in their prime working years of 25 to 54 did not work. Today, the number has more than tripled, to 16 percent. Is this a terribly worrisome number?

• Up till 2000, the relaxed response might have been: "Sure, relatively fewer men are working, but a far higher percentage of women are working, so that's the reason." It is natural that some men of prime work age might stay home to look after their elderly parents or children.
• But, since 2000, the share of women who are working has also been declining. So it is fair to say that since 2000 "the Decline of Work" applies to the population as a whole. A team composed of the Times, the Kaiser Family Foundation and CBS has set about trying to find explanations of the decline through polling.

The second report in the series shows highlights from their findings, as reported by Gregor ("driven by data") Aisch and Josh Katz as well as David Leonhardt. For example:

• Of men aged 25-54, 64 percent would like to have a job. 
• But only 45 percent have been actively looking in the last year (a looser definition than that of the BLS, which wants to know if they have been looking in the prior four weeks).
• In other words, about 30 percent of those nonemployed who say they would like to have a job have not looked for work within the prior year.
• Why not? Well one reason is that one-third of the nonemployed have been convicted of a crime. Nearly half of the nonemployed say they suffer from a disability or from general health problems and 43 percent say that not being employed in itself is bad for their mental health.

An amazingly useful chart shows the nonemployed status in every county in the United States, shaded to show where the nonworking status is most prevalent. The numbers are more complete than the unemployment numbers of the BLS, which by definition covers only the "civilian noninstitutional" population. The BLS therefore excludes the military plus institutionalized populations such as those who are incarcerated, presumably because of the difficulties inherent in conducting random-sample surveys of such populations.

This chart should be pondered by every elected official in the United States.

The research brims with public-policy implications. It supports bipartisan efforts to reduce the sky-high U.S. incarceration rates and to lower the high barriers to re-entry in the job market resulting from licensing requirements that are not related to job performance. I believe it supports programs for more apprenticeships among younger workers and Kurzarbeit-type programs to keep older workers work-ready during a recession.

Surely not by coincidence, the Times editorializes in favor of Mayor de Blasio's plan to try to keep the mentally ill who do not pose a risk to others out of jail because 40 percent of the 11,000 people in jail in New York City are mentally ill - an increase from a few years ago when the figure was 25 percent. Many of the inmates were arrested and convicted of low-level crimes such as not paying a fare or trespassing. These mentally ill inmates are expensive to incarcerate because they stay twice as long as  other inmates since they find it harder to obtain bail.

The Times is making an important contribution to public policy formulation for our recovery from the Great Recession. As the economy picks up we will need workers and the sooner we identify the reasons why people are not working and address them, the better prepared we will be to put them to work.

However, I wish that the Times would help the public understand how the concept of nonemployment fits together with something that has been measured and reported regularly since the Full Employment Act was passed after World War II, namely the civilian employment-population ratio. This indicator, which is presented at the top of this post, pretty fairly presents the extent of the nonemployment problem over time, although it omits the institutional populations that bulk large in some parts of the country.

JOBS | November Growth by Industry

Overall, the news is good. But well-paid government jobs at all levels declined
between 2010 and 2014. The fastest-growth private replacement jobs have
been lower-paid jobs in hospitality, retail and administrative/waste services.

The headline news about the November jobs report is that 321,000 new payroll jobs were created, seasonally adjusted, and the unemployment rate was steady.

Also, hourly wages rose 0.4 percent, double the expected rate.

A troublesome aspect of the recovery (see chart at left) is that wages have stagnated because jobs have been growing in lower-paid industry sectors–retail, hospitality (especially food services and drinking places), administrative and waste services.

So an interesting place to look in the monthly BLS job numbers is the industry-sector breakdown.

This information is in Summary Table B, which is provided below in slightly simplified format with comments attached to each sector. Interesting sectors in November were nondurable goods manufacturing, retail trade, information services (includes a lot of tech companies), financial activities (heavily concentrated in NYC), professional and business services, health care, leisure and hospitality and other services.

Sen. Randal ("Rand") Paul will be pleased that growth in government jobs is slow, but the loss in government jobs has been a drag on the economy, especially since government workers are paid above-average wages.

ESTABLISHMENT DATA Summary Table B.  EMPLOYMENT BY SELECTED INDUSTRY
(Over-the-month change, in thousands, seasonally adjusted)
Category	Nov. 2013	Sept. 2014	Oct. 2014(p)	Nov. 2014(p)	Comment, Nov. 2014
TOTAL NONFARM (Private+Govt)	274	271	243	321	Above expected
TOTAL PRIVATE (Goods+Services)	272	249	236	314
Goods-producing	68	36	28	48	Pickup from October
Mining and logging	1	6	1	0
Construction	32	18	7	20	Healthy growth
Manufacturing	35	12	20	28	Picking up?
Of which: Durable goods(1)	19	11	18	17	Steady
[Of which: Motor vehicles, parts]	4.7	1.7	2.0	3.0	Recovery?
Of which: Nondurable goods	16	1	2	11	Interesting
Private service-providing(1)	204	213	208	266	Very positive
Wholesale trade	16.8	2.9	6.1	2.5	Reduced growth
Retail trade	22.3	39.9	34.2	50.2	Great, but is some growth at expense of December?
Transportation and warehousing	32.4	7.0	15.3	16.7	Continuing growth
Information	1	3	-5	4	Turnaround
Financial activities	-4	14	6	20	Good news for NYC
Professional, business services(1)	73	66	52	86	Big jump
Of which: Temp services	36.6	23.2	19.5	22.7	Continuing growth
Education and health services(1)	25	35	37	38	Mostly health care
Of which: Health, social care	24.4	24.8	31.5	37.2	Obamacare?
Leisure and hospitality	37	47	55	32	Fast growth, slowing down?
Other services	-1	0	7	15	Interesting
TOTAL GOVERNMENT	2	22	7	7	Constrained growth

BLS | Boy Cries "Wolf" Again on Unemployment Data

Funny... but unfair. (Washington Post cartoon, 2011)

Onto her 23, 2014–A New York Post columnist yesterday wrote that Los Angeles, "the city known as America’s story factory", is "making up Census data".

This is a reprise of a challenged report last year  emanating from Census worker Julius Buckmon that the BLS had been "faking" the unemployment numbers in Philadelphia in 2012 to make President Obama look good for his reelection.

The columnist, John Crudele, says the new whistleblower reports that Census workers in the Los Angeles region have been manipulating economic data and that she is now coming forward because she “applauds” whistleblowers in Denver and Philadelphia.

Crudele's report last year was in support of former GE CEO Jack Welch, a self-identified Republican who publicly criticized the BLS in September 2012 when the U.S. unemployment rate dropped to 7.8 percent from 8.1 percent. Welch tweeted:

Unbelievable jobs numbers..these Chicago guys will do anything..can't debate so change numbers.

His tweet was widely critiqued by people who noted GE's own reputation for earnings manipulation. Sample conclusions:

• In the Atlantic, Jordan Weissman argues that poor survey data wouldn't make much difference. He says the unemployment data historically mirror the payroll-jobs data, which are more broadly based. 

• The Census Bureau’s Office of Inspector General looked into Buckmon's report and concluded:

Our investigation did not find any evidence to support allegations that supervisors in the Philadelphia Regional Office manipulated, or attempted to manipulate, the unemployment rate prior to the 2012 presidential election.

• Commerce Inspector General Todd Zinser found wrongdoing on the part of Buckmon, but not his supervisors.

The reason for public concern is that the Census Bureau conducts a survey of about 60,000 households each month to find out who is working and who is unemployed. These surveys are turned over to the Bureau of Labor Statistics, to compute and estimate the monthly unemployment rates, which are usually reported on the first Friday of every month and are closely watched by Wall Street and politicians contemplating reelection.

The LA whistleblower brings to four the number of regions where survey problems are alleged by Crudele - Philadelphia, Chicago, Denver and now LA. (No problems have recently surfaced in the other two regions, New York and Atlanta.)  Each region is divided into Field Service Areas (FSAs) with 10-15 field reps reporting to one supervisor. The BLS requires a 90 percent success rate for interviews to be included in the Current Population Survey. The LA whistleblower told The New York Post:

There are some FSAs that month after month had a 100 percent response rate. That alone should raise flags! I can understand an occasional 100 percent response rate but you have to raise an eyebrow when some FSAs have 100 percent every month.

At any rate, the unemployment rate was just reported as having fallen in September to 5.9 percent, below the benchmark 6 percent rate - historically considered the rate below which inflationary pressures are supposed to start. So the decline in 2012 was certainly not off the trend line.

JOBS | Strong Growth, Jobless Rate Falls to 5.9% (Comment)

Jobs Grow Faster in September,
Unemployment Below 6%

Summary. The BLS jobs report for September [link is to the archived BLS News Release] supports the growing consensus that the economy is breaking out of its slow-growth mode. The payroll jobs increase is a healthy 248,000 - and would be 317,000 if we added in the BLS's upward revisions in the July and August job numbers. Unemployment fell to 5.9 percent, falling for the first month since 2009 below 6 percent.

The September numbers and retroactive revisions bring the BLS reports more in line with economic forecasts of faster recovery for the August payroll jobs and unemployment rates.

Payroll Jobs. Nonfarm payroll jobs grew by 248,000 in September, with notable increases in retail trade, health care service, and professional and business services. This is well above the 213,000 per month average gain over the previous 12 months. The September increase is after combined upward revisions of the July and August numbers by 69,000 jobs, so the picture has improved by 317,000 jobs since June.

The good news extends to individual sectors such as professional and business services, which added 81,000 jobs in September, well above the 56,000-per-month average gain in the previous 12 months. The professional-services gains were concentrated among temp and other employment services, management consulting, and architectural and engineering firms - but law-firm jobs fell by 5,000.

These increases were on top of upward revisions in job numbers for July (by 31,000 jobs) and August (by 38,000 jobs).

Unemployment.  The September unemployment rate declined by 0.2 of a percentage point to 5.9
percent, the first time it fell below 6 percent since 2009, in mid-recession. The drop in the unemployment rate resulted from a decrease in the number of unemployed persons by 329,000 to 9.3 million. The September unemployment rate was 1.3 percentage points below the number in September 2013. The number of unemployed persons fell by 1.9 million.

September unemployment rates declined from August, seasonally adjusted, for

• adult men, 5.3 percent
• whites, 5.1 percent
• Hispanics, 6.9 percent. 

The rates changed little, seasonally adjusted, for

• adult women, 5.5 percent
• teenagers, 20.0 percent
• blacks, 11.0 percent

The rate for Asians was 4.3 percent, not seasonally adjusted, little changed from a year earlier.

In September the number of discouraged workers fell to 698,000, a decline of 154,000 from a year earlier.

White House Blog on Job Numbers, Jason Furman . Data on August economy in New York City and the NYC Metro Area.

Comment

Saturday, Oct. 4 - The New York Times this morning, with two reporters'  names at the top and two more at the end of the story, notes the higher jobs and lower unemployment, and, looking for a new angle, adds in the headline: "Sunny News May Not Bolster Democrats."

GOP-leaning John Crudele at the NY Post, by contrast, argued that the good numbers will help Democratic candidates "because issue No. 1 with Americans is the economy. In addition to the unexpectedly strong job growth, the unemployment rate broke through the 6 handle."

Let's look at the NY Times case for their sceptical take on the good news. It is stated in para. 2 and is supported starting in para. 11.

Paragraph

• 1. The economic news is good.
• 2. But the news "appeared to be too little, too late to bolster the prospects of Democratic candidates facing voters".
• 3. Weak support for the statement in para. 2.
• 4-10. More good news, not supporting para. 2.  at all.
• 11. "[V]oters' opinions on the economy tend to lag considerably behind the actual numbers. [The new data] could help lift President Obama's party [but not until] in 2016." Doesn't that depend on how well Democratic candidates get out the word? Surely it doesn't take two years for data on jobs to percolate down to voters in the age of social media (see paras. 18-19).
• 12(a). "The employment rate ... of women, young people and black voters - did not improve in September." The actual BLS language is that it "changed little" in September. The tables accompanying the story show that unemployment among teenagers was down by 1.3 percentage points from September 2013 but rose 0.4 percent from August. But among blacks the rate was down by two full percentage points below September 2013 and was down 0.4 of a percentage point below August 2014 - still high, but both numbers are improving.
• 12(b). "Nor did the numbers of people employed part time because they could not find full-time work." The number shown in the Times table is 7.1 million people in this status, down 2.4 percent from August and down 10.2 percent from September 2013. Both numbers are improving.
• 13. "The actual percentage of working-age people with jobs - 59 percent - has not changed for four months ... the lowest level since 1978." This number doesn't change much, relative to the unemployment rate, by its nature. The unemployment rate is based on a monthly telephone poll of a relatively small percentage of households. The employment to population ratio is an aggregate number divided by a population estimate. A major reason it increased after 1978 is women were entering, and staying in, the work force in much greater numbers. Maybe we are seeing a long-term shift in how and where people work, and their preference for leisure and spending time at home - not just discouraged workers.
• 14-15. Rep. Kevin Brady (R-TX), Chairman of the Joint Economic Committee, says Obama has held back recovery.
• 16. "Hourly earnings are stuck in the mud, down slightly in September." But the table immediately adjacent shows hourly earnings grew 0.2 percent in September over August and 2.2 percent over September 2013.
• 17. "Economists said the slow pace of wage gains despite the sharp drop in the unemployment rate was a continuing puzzle..." Unemployment may just not be low enough to drive higher wages, and other factors are at work. But the Phillips Curve, the tradeoff between unemployment and inflation documented decades ago by LSE Professor A. W. Phillips, still lingers in the thinking of central bankers everywhere because no one has found a better core model.
• 18-19. Pollster Peter Hart says that despite the speed of communication made possible by social media, opinions still take along time to change. "What you're telling me is the water in my basement has dropped a third, but what I'm telling you, I've got water in my basement." However, an unemployment rate below 6 percent is not water in the basement. Historically, the 6 percent unemployment rate was the level below which inflationary pressures were supposed to start.
• 20. Polling by Hart and Republican William McInturff showed in August that four out of 10 Americans say they have been directly harmed by the recession starting in 2008. Independents were much more likely (51 percent) to report that a member of their household lost a job than Democrats or Republicans (38 or 39 percent).
• 21-26. More good news relating to the economy, and statements from Democrats that their economic policies are working.
• 27. Concern about 4.8 million workers "missing from the job force". See my comment in para. 13.
• 28-32. The rest of the story raises the question of the impact of the good job numbers on Fed interest-rate policy. The Fed did not expect 5.9 percent unemployment before the end of 2014, at the earliest. The President of the traditionally hawkish St. Louis Fed, James B. Bullard,  is quoted as saying that the Fed ought to start raising interest rates before mid-2015.  However, the consensus is that Janet Yellen's Fed is cautious about over-reacting to good economic news.

I have followed up on this post after a comment from MercatorNet, here: http://cityeconomist.blogspot.com/2014/10/demography-destiny-and-employment.html.

JOBS | Disappointing in August

No breakout in job creation.

Nonfarm payroll jobs increased by 142K in August, a break in a six-month series of 200K+ job increases, the BLS has just reported.

The predictions were for a 220-230K increase.

Both the August unemployment rate of 6.1 percent and the number of unemployed persons, 9.6 million, were little changed from July.

Some commentators were hoping for a 6.0 percent unemployment rate, or even sub-6.0 percent.

However, to look at the bright or hopeful side:

• The August unemployment rate was down by 1.1 percentage points from August 2013.
• The number of unemployed persons was reduced by 1.7 million from August 2013.
• As I said earlier this week, the post-Labor Day employment figures will reflect the back-to-work scene better, as well as the net effect of educational hiring.
• The Fed meanwhile gets no encouragement to hike interest rates quickly.
• Eyes will be on the next release on October 3.

Comment

October 3 - The September job numbers - more reliable than August because the dust settles in the educational arena - were revised upward for both July and August, narrowing the gap between the predictions and the BLS-reported numbers. The BLS report for September also showed the unemployment rate down to 5.9 percent, in line with some predictions for the August numbers. The BLS seems to be catching up to the forecasters.