The Mismeasurement of Unemployment: Why it Matters

The official unemployment rate is a bad measure of the labor market in a down economy — we should think about using existing alternatives and devising some new ones.

The monthly unemployment reports from the United States Department of Labor tell us that the official unemployment rate remains stubbornly high at around 9.6%. While analysts speculate about why unemployment is high and when the rate may start to decrease, I want to ask a far more basic question, what is unemployment? Like many seemingly basic questions, the answer turns out to be more technical and complicated than it first appears.

Consider a few examples…

Person A. “I lost my job at a factory last spring, and I have been working ten hours a week at a call center and looking for something more full time.”

Person B: “After I got laid off I sent around a few resumes and asked my friends about jobs, but nobody is hiring, so I decided to stay at home to look after my daughter.”

Person C: “I was working at a hotel cleaning rooms, but business has been bad lately, so the manager told me to take unpaid time off and come back when there was more work.”

Broadly speaking, the unemployment rate is defined by the number of people that are looking for work, available for work, and not currently working, divided by the total workforce.  Person A, who is working part-time but cannot find full time work would be classified as employed. Person B, who became discouraged by the lack of job prospects and ceased looking for work, would be counted as out of the workforce. Person C, who is temporarily out of work, would normally be counted as unemployed. Worth noting: the Census Bureau often reports a seasonally adjusted unemployment rate, which applies an adjustment factor to the unemployment rate to reflect month-to-month cyclical trends in the labor market (less farm work in the winter, more hotel cleaning jobs in the tourist season).

The Unemployment Rate is a Poor Measure in a Bad Economy

As the above examples illustrate, the unemployment rate masks several negative labor market trends including underemployment and discouraged workers. Because the unemployment rate is also a snapshot indicator, it cannot distinguish between short duration and longer-term (“structural”) unemployment. The unemployment rate therefore tends to dramatically minimize the true underutilization of the workforce.

This underrepresentation of labor market conditions is likely to be more pronounced during a bad economy. First, more jobs become part-time or temporary during economic downturns. Second, many people become discouraged and drop out of the labor force. Factoring in workers that are “marginally attached” or discouraged – the so-called broad measure of unemployment computed by the Labor Department makes the picture seem substantially worse. Here is a dramatic graphical illustration of how the official unemployment rate, broad unemployment, and another alternative unemployment measure have varied over time. Although the three data series fluctuate together, the rise in broad unemployment is much greater than the rise in conventional unemployment during the recent down economy.

What Might We Learn from Alternative Measures of Unemployment?

One major reason for paying attention to alternative measures of unemployment is that job loss and job creation are highly variable phenomena not only over time, but also across geographic areas and demographic groups. The unemployment statistic is thus likely to be a better measure of labor force participation for some groups than for others – for example, following job loss, white, college-educated professionals may stay in the labor force longer than black, low-educated blue collar workers. This patterning could largely be explained by local labor market conditions – workers may be more likely to drop out of the labor market in areas with high and sustained unemployment, such as Detroit, Michigan. The result is that unemployment measures mask the severity of unemployment conditions more among populations where, for various reasons, people stop looking for work or take temporary work to make ends meet.

In a 2006 paper, analysts from the Bureau of Labor statistics found that recessions had very different impacts on different demographic groups. For example, women increased their labor force participation during recessions by 2 points, while the rate edged downward for males. Labor force participation also decreased more rapidly for people over age 55. I have not yet seen any systematic analyses for the current recession, but please let me know if there are any available. (I am similarly interested in the experience in the UK and Europe).

Expanding the Unemployment Measure

There are a few ideas for how to provide more nuanced, and potentially more accurate measures of unemployment. One is to increase awareness and understanding among policymakers of the already-existing alternative measures of unemployment calculated by the Labor Department, such as U-5 and U-6 unemployment. I also think it’s also helpful to devise measures of labor market potential. A simple measure of this kind is the number of job-seekers divided by the number of job postings (there are already surveys that collect measures of hiring patterns among employers). Finally, surveys could do a better job of systematically measuring labor market discouragement. It is worth understanding why certain workers are more likely to leave the labor market – can they not find work at all? Or rather, there are jobs, but not at their desired wages? Do they think that retraining might help them more than temporary employment? How much does transportation and child care problems affect decisions to drop out of the labor force? All of these questions could be addressed with subsamples of major surveys such as the Current Population Survey, and they would greatly improve our ability to respond to unemployment in real time.

About Brendan Saloner

I am a postdoctoral fellow at the University of Pennsylvania in the Robert Wood Johnson Health and Society Scholars Program. I completed a PhD in health policy at Harvard in 2012. My current research focuses on children's health, public programs, racial/ethnic disparities, and mental health. I am also interested in justice and health care.
This entry was posted in Articles and tagged , . Bookmark the permalink.

11 Responses to The Mismeasurement of Unemployment: Why it Matters

  1. John says:

    Nice review of the issue!

    This isn’t really about definition, but ‘m really curious about the prospect of using Internet-generated data to get a better picture of unemployment. Google has been doing this with the flu and with inflation (they created a Google Price Index)—employment seems like an obvious next step.

    Potential data sources:
    Search logs (e.g., people search for “Unemployment insurance”or “Job openings”)
    LinkedIn (status changes) (resumes posted etc.)
    Sign-ups for online labor markets like oDesk and Elance
    Twitter stream data (already used for

    Obviously there would be a need for calibration (and I can think of all sorts of potential problems with each data source and terms of coverage, strategic misreporting, identity etc.), but it would be hard to beat in terms of speed and cost and could augment more traditional measures.

  2. Wow, this is such a good idea it blows me away. You are Mr. Online Labor Markets, so I’m assuming you have a sense of how to do this in real time. I’m a little bit skeptical that you can back out a reliable unemployment statistic using online indicators (especially when you miss the non-negligible share of people that are not on the web), but there’s a real potential to use the internet to understand what the job search process looks like in different places and in different times. There’s also a huge potential to do some social networks analysis looking at, for example, how unemployment changes across a LinkedIn network. You should add this to your millions of other projects…

  3. Douglas Scott says:

    This sounds very similar to the debate that unfolded in South Africa a couple of years ago when it was revealed that South Africa had an official unemployment rate well above 20% and an actual unemployment rate that was possibly above 40%. I forget the number because it was a long time ago and so the debate is quite hard to track now. It turned into quite an ugly political battle with the government insisting on only using the lower official figures and opposition parties and civil society lobbying for government to also take into account the higher figure. Some thing the government could not do as it would led to a loss of face.

    By the way, I took John’s advice and did a quick and dirty check on Google Trends for the phrases “Unemployment insurance” and “Job openings” in the United States.
    Which you can see here:
    and here:

    For “Unemployment insurance” the graph shows a significant jump in people using the phrase in the second half of 2008 with it peaking in early 2009 then dropping off to pre-2008 levels today. The second phrase is more interesting with constant dips towards the end of the year followed by dramatic peaks immediately after wards just before the begining of the following year and then stabilizing again. Dont know if it means anything meaningful but it is fun to look at.

  4. Doug,

    Thanks so much for both comments. South Africa is in a league completely of its own with respect to unemployment, and you are right, it’s a huge embarrassment to the ANC. I should dig up the controversy you are talking about. Two things seem to make a big measurement difference in that context (and to a lesser extent in other developing countries too): how to classify workers in the informal labor market (small time vendors, car guards, etc.) and what the relevant denominator (there is no reliable estimate of how many people are in the workforce in South Africa partly because of the immigrant population).

    Those google trends charts are fascinating!! I have never worked with that, but it’s an excellent way to get a sense of how people are dealing with the economy. Maybe we should write a future post just based on google trends charts.



  5. Paul says:

    The google/online trend suggestion is really cool. Of course, it will be impossible to separate the online searches made by job seekers from those made by, journalists, researchers, etc. who are interested in those who are interested in work.

  6. Fortunately (?) only a small segment of academics seem to have much interest in studying unemployment in the recession, and most of the American news media is following Lindsay Lohan’s experiences in rehab.

  7. Paul says:

    Academia saves the day again!

  8. Pingback: Highlights so far… | Inequalities

  9. Ernie says:

    The standard unemployment rate is only a mismeasure if you expect it to measure what it was never intended to measure, like standard of living or ‘economic wellbeing’ or the like. It’s supposed to measure ‘labour market slack’, or the reserve army of labour. As you mention, the USBLS, to its credit, does report ‘alternative measures of labor underutilization’. You may find this post of interest:

    I confess I’m pretty dubious about the online sources John mentioned. While some of these services seem to place people in labouring and trade jobs, I harbour a strong suspicion that online services are not the principal method of advertising or finding such work. If that’s the case, then examining such sources could only reveal the tip of the iceberg. It may indeed be possible to establish a correlation between trends identifiable from online sources and other, more robust, sources, and whether the online indicators lead or lag standard labour force data, etc., but I think that is work that remains to be done.

  10. Pingback: Unemployment Disparities in Three Pictures | Inequalities

  11. Pingback: Available Metrics of America’s Divided Economies #2 | Musings on Maps

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.