Why falling unemployment doesn’t always mean rising employment

Last Friday, the Bureau of Labor Statistics reported that the unemployment rate for the United States fell from 9 percent in October to 8.6 percent in November and that nonfarm employment rose by 120,000 workers.  This sounds sensible: unemployment fell and employment rose.

But consider this: On Nov. 17, the Minnesota Department of Employment and Economic Development announced that Minnesota’s unemployment rate fell from 6.9 percent in September to 6.4 percent in October while employment decreased by 6,100 jobs.

Doesn’t falling unemployment meaning rising employment? Not necessarily. To be a critical user of economic data, it pays to spend some time getting acquainted with the concepts behind the headlines. 

Two surveys: household and establishment
The Bureau of Labor Statistics conducts two surveys each month, one of households and one of business establishments. The household survey, formally known as the Current Population Survey (CPS), is the source of our data on the unemployment rate. The business establishment survey — the Current Employment Statistics (CES) survey — provides data on payroll employment.

The two surveys generally tell us similar stories about the state of the labor market, as they did in last Friday’s national report. Sometimes, however, they give mixed signals, as happened for Minnesota in October. At least two possibilities are at work.

First, the household survey measures employment as the number of people who reported working for pay; it doesn’t differentiate between self-employed and payroll employees. Thus, for example, self-employment could be rising while payroll employment is falling or vice versa.

Second, the establishment survey is slow to catch new businesses. That is, there might be many new companies being created that employ workers, but the sample of firms in the survey isn’t capturing this. The BLS tries to fix this by regularly replacing firms in the sample but there is probably some slippage.

Understanding the unemployment rate
Another reason why the unemployment rate and the employment numbers might contradict one another involves the data from the household survey itself. We need to get some concepts straight, so stick with me for a moment.

The place to start is the pool of potential workers. In the United States, this is defined as anyone 16 years old and up who is not in an institution such as a nursing home, in prison or on active duty in the armed forces. Economists never come up with cool names for things, so we call this the civilian, non-institutionalized, working-age population. I’ll just call it the working-age population.

Next, the survey breaks down the working-age population into two groups: the labor force and those “not in the labor force.” (See what I mean about names?) People not in the labor force include retirees and anyone else who is not seeking paid employment.

The labor force is then divided into the employed (i.e. those who work for pay) and the unemployed. According to the household survey, individuals are unemployed “if they do not have a job, have actively looked for work in the prior 4 weeks, and are currently available for work.”

Whew. Now, we can actually calculate the unemployment rate. Divide the unemployed by the labor force and voilà, you’ve got it.

Now, here is the problem. Suppose someone has been unemployed for a long time and quits looking for work. (Economists call them discouraged workers.) This person no longer counts as unemployed but rather as “not in the labor force.”  This means that both the unemployment number and the labor force number go down and, since the number of unemployed is usually quite a bit smaller than the labor force, this number will fall faster and the unemployment rate will decrease.

The Bureau of Labor Statistics is well aware of this difficulty and therefore constructs an alternative measure of unemployment that includes discouraged workers; it’s called U-6.  (There’s that naming problem again…)  For example, the U-6 unemployment rate was 15.6 percent in November as opposed to the standard rate of 8.6 percent. This is a difference worth noting, as it involves almost 11 million people who are not counted in the standard rate but show up in U-6.

My labor market dashboard
Rather than focusing only on the unemployment rate or on payroll employment, I monitor four labor market indicators every month:

  • The unemployment rate.
  • The U-6 unemployment rate.
  • The labor force participation rate. This is the ratio of the labor force to the working-age population and provides an indication of whether people are joining or leaving the labor force.
  • The employment-population ratio, that is, the number of people employed divided by the working-age population. I prefer this number to the unemployment rate because it avoids the discouraged worker problem.

I do this for both the national numbers and the state-level data. Here’s my national dashboard for the period since the recession began:

Source: Bureau of Labor Statistics
Source: Bureau of Labor Statistics

All of the data to make your own dashboard are available at the Bureau of Labor Statistics website. If you don’t like drawing your own graphs, you can go to the St. Louis Federal Reserve’s FRED database  and it will draw the graphs for you.

The advantage of using all of these indicators is that I have a more nuanced view of the labor market. In particular, it keeps me from focusing on the ups and downs of the unemployment rate and forces me to see that not only is this a severe slump in the labor market but one that has shown no improvement since the summer of 2009.

So, next time you read about the unemployment rate and the employment numbers, dig a little deeper ask yourself, what do you see in the data.

You can also learn about all our free newsletter options.

Comments (3)

  1. Submitted by Noel Martinson on 12/07/2011 - 01:29 pm.

    Great post. I find the LPFR and Employment-population ratios a particularly insightful view of overall employment trends. Thank you.

  2. Submitted by Dale Hoogeveen on 12/08/2011 - 09:31 am.

    Do any of your figures include the black market labor force paid in cash with no benefits or taxes calculated and/or under the table? By its very nature that is not going to be accurately reported to any survey.

    That needs to be considered in the total picture, because by necessity quite a few of those discouraged persons who have officially quit looking have been forced to tap into that for any kind of income at all.

    That has the potential to be the elephant in the corner in unemployment figures just like it is in employment of illegal workers.

  3. Submitted by Paul Udstrand on 12/09/2011 - 08:42 am.

    Thank you Prof. Jonston, this rally nice and concise treatment. Perhaps you could answer a couple questions I honestly cannot answer for myself? I’ve tried researching this and I just can’t sort it out. I think would be interesting to know how the currently unemployment rates compare to the great depression, and to other countries, primarily Europe. The problem is that our government keeps changing the way it calculates unemployment, I think Reagan changed it twice, and Clinton changed it yet again- with the net effect of lowering the unemployment rate.

    So here are my questions:

    1)What method were they using in the 1930s, were they using something closer to the U6 measurement?

    2)How do they calculate unemployment in Europe? And which one of our measurements compares?

    3) Even in the US, when we compare this recession to the one in the 1980s, what measures are being used? Like I said, they changed the way they calculate it, so when we compare the pre-change 1980s rates with todays, do economists adjust for that?

Leave a Reply