How to Measure Unemployment? Is there a new scale?

My friend Anne from college asked me to comment on an interview on the NPR program On the Media. The interview was with Kevin Phillips, whose claim to fame is running Richard Nixon’s presidential campaign. His expertise appears to be in politics, journalism, and economic issues. In the interview Phillips claims that government economic statistics such as the unemployment rate and GDP are being manipulated. I thought I would take a look at some of Mr. Phillips claims.

Let’s start with this quote from the interview
“Well, let me start with unemployment. It's always a question of what the workforce is and how you define people who aren't quite in it. And this might sound like it's fairly simple, but it's not remotely. Government unemployment measurements run from the U-1 to the U-3 and up to the U-6. …….
The U-3 is the number that they generally report. The U-6 includes a lot more people who maybe they're looking for a job, maybe they're not. There's some larger explanation of why they're not working. And the U-6 has unemployment about twice as high as the U-3.”

Phillips is making this sound more complicated then it is. If you want to measure the numbver of unemployed people you might want to take out the people who are not looking for the job (as U-3 does). Basically the difference between U-3, U-4, U-5, and U-6 is how you want to count people who work part time or may no longer be actively looking for a job. (Long article here that explains the difference)

U-3 is the unemployment rate you hear in the news, which has been around 5.5% this does not include people not actively looking or people working part time. U-6 includes just about everyone and it is about 9%. Paul Krugman a few weeks ago suggested we might want to look more at U-6, as “This measure shows what most people sense: the labor market has gotten a lot worse over the past year, not just in the last few months.”

However, as commenter at Krugman’s blog and Econ blogger Karl Smith points out the two measures are highly correlated (just look at this graph)

In the end I don’t think the Bureau of Labor Statistics is trying to game the system. They are using the traditional unemployment definition (U-3) so it is in some sense easier to compare across time. Although it is worth noting that the measure of unemployment must change overtime. For example the reported measure only included household heads until 1978, so in many cases women were excluded.

A final thought a nice parallel can be drawn to baseball statistics. In some sense the homeruns Babe Ruth hit were similar to the ones Hank Aaron hit or hit by Ryan Zimmerman and Magglio Ordonez . But we have to adjust for context of those homeruns. Unemployment data is useful as long as you keep that in mind, and the difference between the two measures in terms of their change over time does not seem to be all that big.