Unemployment Stats Attracting More Skeptics

As the economy weakens, some of the authoritative media outlets are beginning to question some government statistical releases. We noted over the weekend (“Surprise! It’s Rough Out There in the Job Market“) that a New York Times seemed caught aback by weakness in the labor market. It had, like most of the media, focused on the unemployment releases, which have considerable shortcomings, particularly in excluding the large and growing pool of discouraged workers.

The Times’ David Leonhardt, in “Unemployed, and Skewing the Picture,” does a nice job of remedying this oversight, focusing on the issue of how reported unemployment and other measures of labor force participation have diverged in recent years, and looks at possible explanations. He also starts with a telling anecdote, that the inventor of unemployment measure set out to prove that unemployment was lower than it was widely perceived to be at the time. He’s had an unusually long-lived success.

Leonhardt also gives an observation that those of us railing in the statistical wilderness have made for some time: the lack of labor bargaining power is inconsistent with a robust job market.

From the New York Times:

In 1878, Carroll D. Wright set out to do something that nobody in the United States had apparently ever done before. He tried to count the number of unemployed.

As is the case today, the 1870s were a time of economic anxiety, with a financial crisis — the panic of 1873 — having spread into the broader economy. But Wright, then the chief of the Massachusetts Bureau of the Statistics of Labor, thought there weren’t nearly as many people out of work as commonly believed. He lamented the “industrial hypochondria” then making the rounds, and to combat it, he created the first survey of unemployment.

The survey asked town assessors to estimate the number of local people out of work. Wright, however, added a crucial qualification. He wanted the assessors to count only adult men who “really want employment,” according to the historian Alexander Keyssar. By doing this, Wright said he understood that he was excluding a large number of men who would have liked to work if they could have found a job that paid as much as they had been earning before.

Just as Wright hoped, his results were encouraging. Officially, there were only 22,000 unemployed in Massachusetts, less than one-tenth as many as one widely circulated (and patently wrong) guess had suggested. Wright announced that his “intelligent canvas” had proven the “croakers” wrong.

From Massachusetts, he went to Washington, where he served as the inaugural director of the federal government’s Bureau of Labor Statistics and later as the head of the United States Census. His method for counting — and not counting — the unemployed became the basis for Census tallies of the jobless and, eventually, for the monthly employment report put out by the Bureau of Labor Statistics. Wright is the father of the modern unemployment rate.

This Friday, the government will release the latest employment report….Whatever the survey ends up showing, however, you can be sure of one thing: Politicians will be quick to point out that joblessness remains low by historical standards. “Five percent is still a low unemployment rate,” Ed Lazear, the chairman of President Bush’s Council of Economic Advisers, said recently. “It’s below the average for the last three decades.”…..

Statistically, all this is true enough. But it’s also deeply misleading.

Over the last few decades, there has been an enormous increase in the number of people who fall into the no man’s land of the labor market that Carroll Wright created 130 years ago. These people are not employed, but they also don’t fit the government’s definition of the unemployed — those who “do not have a job, have actively looked for work in the prior four weeks, and are currently available for work.”

Consider this: the average unemployment rate in this decade, just above 5 percent, has been lower than in any decade since the 1960s. Yet the percentage of prime-age men (those 25 to 54 years old) who are not working has been higher than in any decade since World War II. In January, almost 13 percent of prime-age men did not hold a job, up from 11 percent in 1998, 11 percent in 1988, 9 percent in 1978 and just 6 percent in 1968.

Even prime-age women, who flooded into the work force in the 1970s and 1980s, aren’t working at quite the same rate they were when this decade began. About 27 percent of them don’t hold a job today, up from 25 percent in early 2000.

There are only two possible explanations for this bizarre combination of a falling employment rate and a falling unemployment rate. The first is that there has been a big increase in the number of people not working purely by their own choice. You can think of them as the self-unemployed. They include retirees, as well as stay-at-home parents, people caring for aging parents and others doing unpaid work.

If growth in this group were the reason for the confusing statistics, we wouldn’t need to worry. It would be perfectly fair to say that unemployment was historically low.

The second possible explanation — a jump in the number of people who aren’t working, who aren’t actively looking but who would, in fact, like to find a good job — is less comforting. It also appears to be the more accurate explanation.

Various studies have shown that the new nonemployed are not mainly dot-com millionaires or stay-at-home dads. (Men who have dropped out of the labor force actually do less housework on average than working women, according to Harley Frazis and Jay Stewart of the Bureau of Labor Statistics.)

Instead, these nonemployed workers tend to be those who have been left behind by the economic changes of the last generation. Their jobs have been replaced by technology or have gone overseas, and they can no longer find work that pays as well. West Virginia, a mining state, is a great example. It may have a record-low unemployment rate, but it has also had an enormous rise in the number of out-of-work men.

These nonemployed remain a distinct minority of the population. But the growth in their numbers is one reason that overall wage growth has been so weak lately. With such a large pool of people who aren’t employed — but willing to work for the right price — those who do have jobs find themselves with less bargaining power. Since 2003, total compensation, including the value of health insurance and other benefits, has failed to keep pace with inflation for most workers, according to Jared Bernstein of the Economic Policy Institute.

I’m not suggesting that the government change its definition of the unemployment rate after all these years. (The government has tried to come up with various alternate measures of joblessness, which are broader but not especially useful.) I’m also not suggesting that the Bureau of Labor Statistics somehow cooks the books. Both Republican and Democratic economists praise the bureau as a model of professional nonpartisanship.

Yet there is no doubt that the unemployment rate is a less telling measure than it once was. It’s simply no longer the best barometer of the country’s economic health. A truer picture can be found elsewhere, by looking at compensation growth, for instance, or to changes in the percentage of the employed.

No less than Tom Nardone — who, as the economist overseeing the unemployment survey, might reasonably be considered the Carroll Wright of today — made a similar point to me the other day.

“Just saying the unemployment rate is 5 percent, without any other context, really doesn’t tell you much,” Mr. Nardone said. “It’s far more complicated than that.”

Print Friendly, PDF & Email


  1. Jojo

    Wouldn’t it make more sense to track EMPLOYMENT rather than UNEMPLOYMENT?

    We should then be able to determine the number of those not working through simple arithmetic, without being bamboozled by government manipulations.

    I would expect that the government and corporate bigwigs would fight such a proposal. Facing the true numbers of those out-of-work would not fit well into all the hogwash about how good the economy is.

  2. Yves Smith

    Anon of 4:23 AM,

    Thanks, Just added it to links. I am interested in free speech and Web censorship.

  3. Ivan Kitov

    As almost any economic variable unemployment has two sides – political and theoretical. Politically, unemployed people, if many, give rise to negative emotions and are usually used in order to punish current power. So, democrats now should raise questions about underestimated unemployment. No news here. It was, is, and will be forever.

    Economics as a theory needs a consistent definition. Actually , as in hard sciences it should not be even some absolute measure of true unemployment (employment, inflaiton, etc.) but should give a constant portion of this true value. For example, we were always sure that definition of mass is absolute, but now it looks like we did not count in some bigger portion called darl matter. This omitted mass does not change our equations , however. Same is with unemployment. If one consistently measured a constant portion of true unemployment it is enought for theoretical considration and search for actual links between measured economic variables.
    In that sense, the discussion about “hidden” unemployment or biased measurements is less appropriate. We just need same definition over time – not randomly changing from miles per hour to kph.
    Using consistent definition and measurements one can reveal real links between economic variable. As an example: there is a (lagged linear) trade-off between unemployment and the change rate of labor force in the USA. This figure
    illustrates the link. As the figure shows, unemployment in the USA will be decreasing the next five years, with some possible real and measurement related monthly flucs, however.

    Inflation is also on a downward trend.

  4. Jojo

    @Ivan Kitov said – Please explain why unemployment will decline over the next 5 years in the USA.

  5. Ivan Kitov


    If you take a look at the Figure I mentioned in my previous post you will find a relationship linking unemployment and the change rate of labor force level, i.e. employment + unemployment:

    UE(t)= 0.021 + 2.2dLF(t-5)/LF(t-5)

    This relationship is a linear lagged function with 5 years lag of unemployment behind labor force change.
    The change rate of labor force has been decreasing since 2000 (see bls.gov). Thus, the unemployment has been also decreasing since 2005.
    Labor force started to increase at a higher rate last year. So, in 5 years we will observe an unemployment pick up. This will happen if the relationship, which has been valid from 1960, holds in the future.
    I do not see any reason this relationship to fail.

    P.S. It is not enough place and time for details. But you can find my papers on unemployment (and inflation) prediction for Germany, France, Australia, Canada, Japan, Austria. Everywhere similar linear relationships hold.

Comments are closed.