A recent primer on labor force indicators evoked many questions and comments from readers. Some scathingly asserted that Bureau of Labor Statistics data are either idiotic or politically manipulated. Others posed thoughtful questions. Grouping these about major topics and addressing the issues raised may be useful for other readers.
Start by recognizing that labor market indicators such as the unemployment rate, the labor force participation rate and the number of nonfarm jobs all are good-faith, best estimates made with limited time and money. They are statistical inferences from carefully selected samples, but they are not a perfect measure of the whole universe of the population or employers. In this regard, they are no different from our indicators of output, price levels or international flows of goods and money.
As noted last week, the unemployment rate is calculated from questions asked of a sample of roughly 60,000 households out of a total of 127 million. So inferences about the rate for the nation as a whole, for all states and many metropolitan statistical areas derive from responses given by less than one-20th of 1% of the population. As one learns in introductory statistics, increasing sample size as a fraction of the population of interest increases the confidence warranted by the inference made. But this involves diminishing returns in accuracy and querying larger samples takes more resources, both time and money.
Also understand that any indicator ultimately must include somewhat arbitrary definitions. How many hours of work in the “reference week” constitute employment? Ten? One? Or even 15 minutes? What constitutes actively seeking employment? Going to a job service office? Actually turning in job applications to employers? Scanning an employment listing website? Asking friends and former colleagues if they know who is hiring? Thinking about your dream job while playing Grand Theft Auto in your parents’ basement at age 35?
Should employers surveyed include college interns? Paid or unpaid? Who reports temps — the business using one days, weeks or months at a time? Or the agency supplying and paying these workers? How about contract janitors? Or a hotel that calls in occasional wait staff whenever there is a large convention?
What about people who have jobs, but are not at them when surveyed by the Bureau of Labor Statistics? In the 1950s, accounting for workers out on strike but who would return to their usual jobs was a big issue. But what about new mothers out on family leave for three months? Or a manager in alcohol rehab? Are they employed or out of the labor force? Does it matter if they are paid or unpaid, getting benefits or not?
The second important thing to remember is that population, institutions and culture all change over time. What was the best assumption to make or most useful tabulation of responses in the 1960s may cause erroneous inferences in 2020. How do you account for the gig economy? What about consultants working from home?
We divide everyone past their 16th birthday into three categories: employed, unemployed or out of the labor force. But we first exclude the military and people in institutions. How does this affect the numbers?
In 1945 there were about 13 million people in uniform out of a population of 139 million. In 1968, at the post-WWII peak and the height of the Vietnam war, the military numbered 3.5 million out of 200 million. At the low military service strengths in 2000, it was 1.4 million out of 282 million. Today there are just slightly fewer from a population of 328 million.
Add those serving in the military to both the numerator and denominator of the age 16-plus equation and you change the number of jobs, the unemployment rate and the labor force participation rate. But these might tell you less about how the economy as a whole really is faring. It also would make our indicators more comparable to other industrialized nations such as Japan that have far smaller militaries relative to population.
You have free articles remaining.
Ditto for people in prisons. Depending on the source, we have between 650 and 740 people per 100,000 population in prison. The U.S., Russia and China together account for half of all the incarcerated persons in the world. The United Kingdom has about 140, France 104, Germany 75 and Japan 41 per 100,000. Adding our current 2.1 million prisoners to the denominator of the labor equation is simple addition. But how many of these would have jobs if our incarceration rate was the same as, say, France? How many would be unemployed and how many out of the labor force?
When we deinstitutionalized mental health treatment decades ago, we pushed hundreds of thousands of people into the “civilian, non-institutionalized population.” Again, how many of these are employed or unemployed? How many are simply out of the labor force? Are those on the street, or any others who are homeless for any reason, ever going to fall into the 60,000 households in the Current Population Survey?
The aging of the U.S. population is one of the primary challenges of our time, as it is for many other countries, rich and poor. The proportion of Americans out of the labor force simply because of retirement is growing and will continue to grow. So, is the declining labor force participation rate the fault of George W. Bush, Obama or Trump? Is it a sign of a sick labor market? Or does it show how well off we are that people can retire? And what if inadequate Social Security or other pensions force people to work as greeters a big-box store? What does this tell us?
In the 1960s, it was clear who was institutionalized in a nursing home and who wasn’t. Now, with home health aides and block nurses allowing aged to stay in their own homes, institutionalization rates of elderly are falling. Easy cataract operations and joint replacements fix woes that might have driven our grandparents “into the home.”
How do you categorize those in assisted living? Does it matter if the facility adjoins a memory care unit or is freestanding? What about senior condos with extra in-house services? If a couple goes together to a facility that provide wide services to a spouse with Alzheimer’s or advanced Parkinson’s, but the other spouse lives a near-normal life in an apartment differing little from any senior condo, how do you classify either person?
Changing roles of women in the economy was the biggest factor in the rise of the labor force participation rate in the last third of the 20th century. The traditional pattern for women was a high rate just out of high school or college, but then a precipitous drop with marriage and childbearing followed only by a partial rise in their 40s or 50s, after all children were launched. Now there are slight, short drops for childbearing. These occur later in life as the average age of first marriage has increased. And somewhere around age 50, the labor force participation rate for women passes that of men and stays higher into old age. Simply looking at this rate as an economic indicator without understanding the deep underlying cultural forces affecting it over time is a waste of effort.
There are many more issues. One hears pundits and politicians, including candidates for president, assert that “the real unemployment rate is 35 or 40%,” even though this mostly represents people out of the labor force by choice, for example for child-rearing or retirement or full-time high school and college students. Understand that this is ignorance or malign deception, even when uttered by a candidate.
One also hears that “the real unemployment rate is two or three times as high” as the official one because it does not include “discouraged workers” or those getting only 15 hours a week when they want 40, or are Uber drivers with a Ph.D.s in rocket science. All of these categories are tabulated by the Bureau of Labor Statistics and are listed every month. Most are in the U-6 table of the Labor Situation publication. The “headline” unemployment rate is the primary one for good reasons; but both rates, and others in between, are published and are consistent over time.
Yes, labor force indicators are not perfect measures of the work world. But when compiled thoughtfully and tabulated consistently over time, they do give vital information about the state of workers and how that changes. Far more attention, expertise and experience go into their compilation than people realize. The fact that those tabulating the “household survey” cannot even talk to those compiling the “establishment survey” until after the numbers are released is an indicator of the effort that goes into publishing unbiased information.
The idea that some government cabal dictates these statistics to advance an agenda is popular among some on both the political left and right — especially when the survey results are inconvenient for an alternate agenda — but it is entirely moonshine.