I think of two other imperfect measures - the Dow Jones Industrial Average, and the relative temperatures (Windchill, Head Index).

The S&P 500 is a better index than the Dow. There are other indexes that are even more accurate. Yet the average layman 'gets' the Dow much better. Almost no one knows that the Dow is just a few dozen companies, and that the actual composition of the index changes all the time. Likewise, no one actually knows how windchill and heat index are calculated.

Yet, everyone can compare those things to past experience and 'get' what is going on. Holy crap, the Dow is up 400 points? Wow, it's going to feel like 105 tomorrow? Unemployment is 9%? Something is wrong.

If we suddenly switched to U6 as the 'official' measure, it would take a long time for people to adjust. Is 10% U6 good or bad? I have no idea without looking it up, even though I know what U6 is.

Right now it seems like the fed is working off of the higher unemployment numbers and semi-ignoring U3. I believe that is a good thing. But it is confusing to many.

U3 is the official measure because it matched the International Labour Organization definition[1]. Most countries use the same definition.

Measurement is always hard, and always requires caveats. There are good arguments both for including discouraged workers (hidden unemployment) and part-time workers who would prefer full time work (underemployment) in unemployment statistics. There are also good arguments against it, mainly around the fact that gathering those numbers tends to be more error-prone.

[1] http://en.wikipedia.org/wiki/Unemployment#United_States_Bure...

So I don't have any specific knowledge on this, but the intuitive answer is a combination of:

a) The Department of Labor puts out these measures in a boring, dry document. The news media then picks which measures to report on that will grab headlines.

b) Politicians and public officials latch onto the numbers that allow them to paint government and policies in the best (or worst) possible light.

Both of these lead to simplification of the source data. And if you are trying to paint it positively, you'd probably pick the narrowest, lowest unemployment number.

Not necessarily disagreeing with your intuition here, but that doesn't appear to be what happens - pretty much everybody reports on that same U3 measure. Certainly there are articles and discussions about other measures as well (for instance, the OP), but I never seem to see them in the "mainstream".

I'm not sure how or why that became the standard, but it's the standard per the International Labour Organization. The U6 rate only started being recorded in 1994.

U3 data goes back to 1890. There's always someone, usually in an opposing political camp, who says "it's misleading" when it goes down, or when it goes up. Both the left and the right have done this sort of handwaving over the years, but having a stable statistic that goes back many decades is useful even if you want to look at others for a more complete picture about what's going on. On some planet having U6 go from above 17% to just over 11% from 2009 to the present counts as "failed recovery" but I'm not quite sure how. You could argue that there's a bit of a gap there, like around 1.5% if you look at recent history, and maybe even argue that it's structural, i.e. that information technology is causing full-time work to go away, etc. but that's a technical argument to have, which would require looking at a lot of data and not just trying to score points by cherry-picking.

>On some planet having U6 go from above 17% to just over 11% from 2009 to the present counts as "failed recovery" but I'm not quite sure how.

Agreed. I argue this point all the time, actually. I think it's interesting to go back through and look at how unemployment fairs in comparison to the political party of the president during that time. Looking at the U3 rate for the past 50 years, unemployment has only gone down once under a republican president (Reagan) and Carter is the only democrat in the past 50 years who didn't see unemployment drop from the time he took office to the time that he left office.

Ah, this seems like the key answer to my question - it's the only one that's been tracked for generations, and we often want to compare across that time scale.

Since the U3 and U6 more or less follow the same trend., it makes sense to keep using the same reported measure for continuity. People know that 8% "umemployment" (aka U3) is bad. If you suddenly start reporting U6, people will think a 8% U6 is bad, when it is rather good.

If you want to use it to measure or model something else, you might have a reason to pick a different measurement.