UPDATE: In two places below, I implied that the formula for unemployment rate had been altered over the years (true enough) and that the motive was to “sweeten” the employment news for political benefit. Dean Baker, whom I trust and quote throughout this article, offers the following comment via email.
He begins by addressing the difference between the way unemployment was measured during the Great Depression and the way it’s measured today, then comments on the state of the U.S. statistical agencies in general. Baker writes:
The issue in the 1930s is that the government workers on projects like the WPA and CCC were counted as unemployed. This would not be true today.
I think the BLS [Bureau of Labor Statistics] has mostly been pretty kosher in dealing with the data. The statistical agencies in the U.S. have been largely free of political interference and that is something that we should be thankful for.
There are problems with the data, which people like me bitch about all the time, but they are often inherent in the data (some people don’t answer surveys or don’t answer the way we would like) or the result of inertia. Rarely can I identify situations in which BLS or one of the other statistical agencies fudged data to make a politician happy.
I trust Baker’s judgement as well as his political insight; he sees both politics and economics clearly, so I offer this as an offset to my cynicism. I’ll try to find the sources that gave me the impression I’d formed. In the meantime, consider the above as more definitive.
The September 2013 jobs report is out, and the news isn’t good. Even though the unemployment rate is down, the number paints a false picture. Unemployment is down because people are leaving the workforce as fast as others are joining it. (For a more detailed justification for that last sentence, jump to here.) In other words, we’re going nowhere.
From Dean Baker at CEPR.org:
The unemployment rate edged down to 7.2 percent in September, the lowest level since November of 2008. The Labor Department’s establishment survey showed a gain of 148,000 jobs. … In spite of the September drop in unemployment, the employment-to-population rate (EPOP) remained unchanged at 58.6 percent.
Unless my math is bad, there’s only one way to read those two facts — in a month that saw nearly 150,000 join the work force, enough people left it to keep the total employment rate unchanged.
Evidence? Here’s Baker on the September employment-to-population ratio (my emphasis and paragraphing):
This continues the pattern that we have seen throughout the recovery as the unemployment rate falls mainly because workers leave the labor market.
The unemployment rate is now down by 2.8 percentage points from its 10.0 percent peak in October of 2009. However, the EPOP is up just 0.4 percentage points from its low point in June of 2011. Over the last year the EPOP actually edged down by 0.1 percentage point, while the unemployment rate dropped by 0.6 percentage points.
This drop in labor force participation is now occurring at an equal pace among men and women, with the participation of both dropping 0.5 percentage points in the last year.
More fascinating details here.
Looks like we’re going backward to me, at least a little, in this so-called “recovery.” Note the second bolded sentence. Over the last year, while the so-called “unemployment rate” is down, the more accurate “employment rate” is also down. Despite what the unemployment rate tells you, the number of new workers is not keeping up with population growth.
Here’s what that looks like historically:
Shorter employment story — we lost more workers in September than we needed to grow the employed population. Jobs growth is flat. And in the last year we’ve gone somewhat backward overall.
As you can see from the data above, someone’s painting a false picture, and it’s the so-called “unemployment” rate. Here’s why. Both the Unemployment Rate and the Employment-to-Population Rate are ratios, but they measure different things using different data sets.
In fact, the only thing they have in common is “total workers” — the total number actually employed. Here are the formulas.
Unemployment rate. According to Wikipedia:
“The unemployment rate is a measure of the prevalence of unemployment and it is calculated as a percentage by dividing the number of unemployed individuals by all individuals currently in the labor force.”
“Labor force” can have a variety of meanings. In the U.S. it includes only those working plus those “actively” looking for work. Thus the formula is:
Total “labor force” — Total workers
Total “labor force”
The trick is in the definition of “labor force.” No government wants to report a higher unemployment rate than it has to. Since the numerator (the top number) will always capture all of the employed (the “total workers” part), the key if you’re into fooling people is to make the bottom number (the denominator) smaller. That will shrink the whole fraction. (If you like, test it and see.)
(Of course, the real key to a lower unemployment rate is to provide more good work for more people. But hey, you go to war with the statistics you have, not the ones you want.)
Hidden unemployment. The tendency to redefine “labor force” as a smaller number for political gain produces “hidden unemployment“:
Hidden, or covered, unemployment is the unemployment of potential workers that is not reflected in official unemployment statistics, due to the way the statistics are collected. In many countries only those who have no work but are actively looking for work (and/or qualifying for social security benefits) are counted as unemployed. Those who have given up looking for work (and sometimes those who are on Government “retraining” programs) are not officially counted among the unemployed, even though they are not employed. …
The statistic also does not count the “underemployed“ – those working fewer hours than they would prefer or in a job that doesn’t make good use of their capabilities. …
As CNN puts it:
The unemployment rate only includes jobless people who have searched for work in the last four weeks. It skips over those who left the labor force entirely because they retired, went back to school, or simply gave up on finding a job.
To get a more accurate picture, we need to look at actual employment, not so-called “unemployment.”
Employment-to-population ratio. Comparing total workers to total adults in a population gives a more accurate picture, since no government will understate the worker numbers, and population numbers are much less fiddleable than “labor force”. Here’s the definition:
[The employment-to-population ratio] is a statistical ratio that measures the proportion of the country’s working-age population (ages 15 to 64 in most OECD countries) that is employed. This includes people that have stopped looking for work.
The formula is therefore:
Total “civilian non-institutional” working-age adults
How this applies to this month’s jobs report
This kind of math comparison can hurt your head — there’s little in common between these ratios — so let’s walk through it to see what happened in September. What we say here will, in general, be true of every jobs report — that is, the movement in the ratios will reflect the same dynamics.
1. The only thing the two ratios have in common is “total workers”.
2. When unemployment goes down, it means that either “total workers” went up, or “labor force” went down, or both. The opposite is true when unemployment goes up. (Crossing a boundary where “labor force” is being redefined brings its own special problems. That didn’t happen here.)
3. When employment (EPOP) goes down, it means that population rose faster than total workers (and the reverse when EPOP goes up).
4. When EPOP stays flat, it means that total workers changed at the same rate as population. Both either went up together (as here) or both went down together.
Combining these concepts, in September 2013 the number of “total workers” as a percentage of “labor force” went down, but the same number as a percentage of population stayed the same.
So the real change in September is the change in “labor force” (in the unemployment equation) versus the change in population (in the EPOP equation). Population went up in September and “total workers” went up by the same percentage. To get a smaller unemployment ratio, “labor force” went down.
In other words, we went nowhere.
To follow or send links: @Gaius_Publius