The US Unemployment Rate Revisited
At the beginning of this month, Bob Higgs posted an excellent discussion, Two Views of the Labor Market in the Deepening Recession. There was some confusion, however, in the comments section over whether U.S. government statistics counted military personnel as part of the labor force or counted just the civilian labor force. Bob quite correctly insisted that currently the Bureau of Labor Statistics (BLS) only reports the civilian labor and the civilian unemployment rate. But I (along with one of the commenters) was aware that back in the 1980s the total unemployment rate, which did include military personnel, was also reported. So I asked one of my students, Justin Rietz, to do some digging for me. Here is what he found out.
Up until 1983 the government measured only civilian employment and unemployment. This made complete sense as long as the U.S. had a military draft. A large portion of the armed forces was thereby analogous to the prison population and logically excluded from the labor force for the same reason. But with the ending of conscription, it made more sense to treat military personnel like regular employees responding to wage and benefit offers. This is probably why the National Commission on Employment and Unemployment Statistics, in its report of 1979, recommended compiling total employment and unemployment, including resident military personnel, as well. This new series was initiated in January 1983 but lasted only a decade. Because of certain statistical ambiguities, the BLS effectively discontinued labor force measures that included the armed forces in June 1994. For those who want to know the details, what follows are the answers that Steve Hipple of the BLS provided to Justin's questions:
Why were the resident armed forces no longer included? There were 3 main reasons:
1) Dept of Defense data on resident armed forces were problematic. For example, many of the troops involved in Desert Storm were still classified as residing in the US because their official station was not changed from their US post. To complicate matters further, each service classified its troops differently, which is, of course, a problem for us. In short, we had problems with inconsistent data.
2) As far as we know, only the New York Times among our data users ever used the total rate, rather than the civilian rate;
3) We don't have any of the demographic detail on the armed forces that would allow us to publish the type of detail on a total basis that we do on a civilian basis. So one could not compare a group's rate (say, black teens) with the total rate, because one would be on a civilian basis and the other on a total basis.
Has anyone complained about BLS excluding them from the counts? We have received no complaints about the lack of inclusion of the armed forces in the rate since we discontinued the series.
How much would including the resident Armed Forces affect unemployment rates? The unemployment rate for the total labor force (which was always published in addition to, and not in place of, the civilian rate), was usually about one-tenth of one percentage point lower than the civilian rate (occasionally it was two-tenths lower). That's because adding the Armed Forces did not change the number of unemployed (numerator), since persons in the Armed Forces are all employed, by definition. But the labor force (denominator) grew slightly; thus, the total unemployment rate was always fractionally lower than the civilian rate. For instance, in 1983, the total unemployment rate (including the Armed Forces) was 9.5 percent; the civilian unemployment rate was 9.6 percent. In 1993, the total unemployment rate was 6.7 percent; the civilian unemployment rate was 6.8 percent.
What about military reservists? Individuals who are on active duty are not included in the CPS [Current Population Survey] universe (the civilian noninstitutional population) or in any of the labor force measures. Those serving in the Armed Forces Reserves are within the scope of the CPS. That is, they are considered civilians, so they are interviewed and their responses go into the official labor force estimates. However, if their unit is called to active duty, they are then considered to be in the Armed Forces and thus outside the scope of the survey. The time reservists spend in initial or ongoing training, or their annual 2-week (or longer) duty, is not considered active duty. Only if their unit is called up by military or Presidential order are they considered to be on active duty.
Reservists are treated differently in the CES [Current Employment Statistics], which is a measure of jobs, not people. According to CES definitions, if a reservist is on active duty the ENTIRE pay period of the 12th, they are NOT counted as employed even if: 1. the employer is making up the difference between their military pay and their normal pay, 2. they receive no pay but still receive benefits, or 3. they receive their normal pay. If a reservist works even one day during the pay period, they are counted as employed. CES data exclude the military. The sample 790 forms available under the technical notes section at http://www.bls.gov/ces/home.htm show that 'armed forces personnel on active duty the entire pay period' are excluded from All Employee counts.
Bill Woolsey - 7/1/2009
Always great information.
Jonathan Dresner - 6/25/2009
I find it more than a little absurd that the Department of Defense and the Bureau of Labor Statistics between them can't come up with a consistent method of counting people, leaving me to wonder who benefits from our failure to accurately record the location and labor status of hundreds of thousands of service members.