Bureau of Labor Statistics Commissioner Keith Hall discusses unemployment data

By Frank Ahrens
Washington Post Staff Writer
Thursday, May 6, 2010; A17

Each month, the Labor Department's Bureau of Labor Statistics releases its estimate of U.S. unemployment.

And the debate begins again.

At 8:30 a.m. on Friday, the Labor Department plans to release the April unemployment rate. We'll learn whether the continuing -- if unsteady -- economic recovery has eaten into the 9.7 percent rate reported in March. Unemployment is down from its peak of 10.1 percent in October, but it has remained stubbornly high.

The government has several classifications for people who are not working full time, and that's where the debate starts. The number that will get all the attention Friday is the official unemployment rate. To be counted as officially unemployed, a person must be out of work and have looked for work in the previous four weeks. That's the 9.7 percent number. Or, in Bureau of Labor Statistics parlance, the U-3 figure.

Critics say this number is inaccurate because it doesn't include two key categories: people who want to work full time but can only find part-time work, and those who have grown so discouraged, they've given up looking for work.

If you add those two categories on top of the official unemployment rate, the number you got for March was an eye-popping 16.9 percent. That's the U-6 number. It's down from its recent high of 17.3 percent in December. Troublingly, however, it has increased over the past two months, even as the official rate has stabilized.

The man in charge of gathering these numbers and crunching them is Keith Hall, commissioner of the Bureau of Labor Statistics. Hall, 53, is an economist and a federal government lifer. Prior to taking over BLS in January 2008, Hall worked at Treasury, Commerce, the International Trade Commission and was chief economist for the White House Council of Economic Advisers. So he knows his numbers.

Q. The BLS surveys a rotating sample of 60,000 households each month to help arrive at its unemployment figures. Why is that method used?

We want to get some consistency with the sample and we want something that's comparable over time. It's the same measure we had when the survey began back in the 1940s.

Do you think the U-3 number paints an accurate picture of unemployment in the United States?

The U-3 official number does capture people who are actively looking, but it's really helpful and important to add in discouraged workers so it gives a fuller picture. I get a little frustrated about the limitations of the U-3 because it doesn't include discouraged workers and part-time workers. People quote our numbers and tell us what we know we're missing, that's in the U-6. You need to look at all of those things to get a clearer picture.

What would you fix about the way you count the unemployed, if you could?

The surveys are really designed to get as accurate as possible a national number, but the sample sizes are not big enough to give us state and metro-area numbers. Those are much less accurate; that's a matter of sample size. If we had more money to do a bigger survey, it wouldn't affect the national numbers, but it would affect regional, state and metro-area unemployment.

The survey methodology was changed in 1994. Why?

It was somewhat of an overhaul. It's a much stronger computer-based system. We feel like we got an improvement in the data quality. Also, we got an increasing amount of monthly data. We didn't collect [discouraged worker] data monthly, only quarterly. We couldn't get a broader measure of underutilized workers.

Frequently, we hear that U.S. unemployment peaked at 25 percent in 1933, during the Great Depression. How reliable do you think that number is and does it compare to the U-3 or U-6 number?

We didn't have any official data back then. It was really the Great Depression that made it clear we had an inadequate statistical system. Our data starts shortly after 1929. On the 25 percent number, nobody knows for sure, but trying to take a guess backward, I would say the Great Depression rate mimicked the U-3 number.

So we would need the official unemployment rate [now at 9.7 percent] to reach 25 percent to reproduce unemployment in the Great Depression.


Has BLS received political pressure to change its methodology to make the unemployment number look lower than it actually is?

The answer is no. One of the things that pleasantly surprised me is the amount of insulation we've been given. We have no political appointees; I'm the closest thing. I'm a presidential appointee with a fixed term. We provide the president with the data the night before it's released. Nobody else sees it except for the morning of the data release. And even though I will brief the secretary of labor at 8 a.m., I lock her in a room -- I actually get to do that -- and go over the data. Nobody is able to leave the room.

There is a story people remember from back during the Nixon administration. It was in the Nixon tapes where he was rather upset with BLS. He felt BLS was not treating him fairly and made the effort to actually remove some people.

© 2010 The Washington Post Company