Dean Baker Completely Unaware How BLS Calculates Unemployment Rate

Holy crap, how did Dean Baker from the Center for Economic and Policy Research get a job writing about economics?

Check out this line of his report on the recent unemployment numbers:

The unemployment rate fell to 9.7 percent in May, primarily as a result of 411,000 temporary Census jobs.

Mr. Baker is apparently unaware of how we calculate the un2employment rate. Let’s help him out here.

First we take the number of people who have jobs. This is not the “non-farm payrolls” number, which increased by 431,000. It is the “Employed, 16 years and over” number which decreased 35,000 (from 139.455 million to 139.420 million). Mr. Baker seems to have those two numbers confused, so I thought I’d clarify.

Then we take the number of people who are looking for jobs but can’t find them. This is where we get the “unemployed” number, which decreased from 15.260 million to 14.973 million.

Then we add the employed number to the unemployed number and you get the Labor Force number. In order to calculate the unemployment rate, we divide the number of people unemployed from the labor force.

139,420,000 + 14,973,000 = 154,393,000 people in the labor force

14,973,000 / 154,393,000 = 9.7% Unemployment rate

Let’s try to prove Mr. Baker’s statement that the unemployment rate dropped to 9.7% “primarily” due to the 411,000 census jobs. We’ll subtract 411,000 from the “employed” number.

139,009,000 + 14,973,000 = 153,982,000 people in the labor force

14,973,000 / 153,982,000 = 9.7% Unemployment rate

We get the exact same unemployment rate with or without the census jobs. That is because unemployment rate dropped due to people leaving the labor force. And most of the people who left the labor force came from the “unemployed” category. Otherwise known as “discouraged workers”.

This is part of the reason that economic understanding is so dismal among the general public. An economic reporter should be able to get the simple facts right about a job report.


  1. Andy says:

    If we move 411,000 people from employed to unemployed then we get:
    139,009,000 + 15,384,000 = 154,393,000 people in the labor force
    15,384,000 / 154,393,000 = 10% Unemployment rate

    Didn’t some of the census workers come from the unemployed number?

  2. politicalmath says:

    But that assumes that the labor force is a sealed group in which people can only move from the “employed” column to the “unemployed” column and vice versa.

    In reality, the number of unemployed only decreased 287,000. Assuming that every single person who left the “unemployed” category got a census job, there are still 124,000 (411,000-287,000) who would have been drawn from another category.

    But then we have a displacement problem. If the above is the case, we would have to assume that several hundred thousand people lost their jobs and then decided not to look for other jobs (otherwise we would see them in the unemployed category).

    This doesn’t even begin to explain how we gained 411,000 temporary Census jobs and still saw the labor force AND the employment level decrease.

  3. Dean Baker says:

    Hi Andy,

    The principle here is that we have a common sample universe even if we have two different surveys of that universe. So, if we can identify an industry (it could be the Census, or it could be anything else) through the establishment survey that has created or eliminated a large number of jobs in a month, then we expect that to have a roughly corresponding impact on the household survey. Because of sampling error we know that the relationship will never be one to one. However, holding everything else constant, if we know that Census created 400,000 jobs in May, then we should expect the household survey to show 400,000 more people employed (ignoring multiple job holders) some of whom will come from the unemployed and some of whom will be new entrants to the labor force.

  4. Andrew E. says:

    PoliticalMath, thank you for explaining this. It irks me to no end when people (mainly the media) make economic claims without realizing how economists calculate measurements. In Macroland, the media, politicians and the public wax poetic about trade deficits, government deficits, taxation, savings and unemployment all the time, yet there are so many errors that it is almost impossible to correct. Some people in economics (such as mere mortal undergrads like me) think government policy is probably ineffective most of the time, and I debate with other people in economics about this; but we can’t even arrive there when the objective facts upon which all economists agree are completely misunderstood by the media, the government and the public. Here’s to math and economics education!