Janet Yellen has joined the unemployment truthers:
“We have an unemployment rate that, if properly measured in some sense, is really close to 10 percent,” Ms. Yellen said on CNBC Thursday. A week earlier, Mr. Powell cited the same figure in a speech about lingering labor market damage.
Come on. We have U1, U2, U3, U4, U5, and U6. We have the labor force participation rate. We have the employment to population ratio. We have both those things for prime age workers only. That's nearly a dozen different ways of measuring "unemployment," and now we have yet another. Here is Fed chair Jerome Powell explaining the unemployment measure that Yellen was talking about:
All told, nearly 5 million people say the pandemic prevented them from looking for work in January. In addition, the Bureau of Labor Statistics reports that many unemployed individuals have been misclassified as employed. Correcting this misclassification and counting those who have left the labor force since last February as unemployed would boost the unemployment rate to close to 10 percent in January.
Well, sure it would. And I could invent a new measure that would boost it even further. But neither Powell's metric nor mine would represent "properly measured" any more than the others. It all depends on what you want to know.
In any case, a raw number shorn of context doesn't really tell us anything. What we really want to know is how much higher (or lower) it is than normal levels in the past. This is fairly easy to see for measures like U3 (the normal headline rate) and U6 (unemployed plus marginally attached):
As you can see, these two track each other pretty well, with U6 a fairly steady 4-5 points above the headline rate. This tells us that if we improve the labor market, both measures will probably reach their previous lows at about the same time. They're both telling us roughly the same thing.
But Powell cheats. He only shows his new metric from the beginning of the pandemic:
Sure, his made-up measure is at 10% now. But what is its natural low level over the past 20 years or so? How much lower should we expect it to drop? You can't simply take a metric that's artificially designed to match the headline rate during a single month and then declare this the "right" way to measure things.
That's especially true for this metric. The "misclassification error" has dropped to almost nothing, and the labor force participation rate always drops during a recession. It dropped a fraction of a point in 1990-91; a little more than a point in 2001-04; more than 3 points in 2008-15; and two points during our current recession. It dropped faster and more dramatically this time around, just like every other economic indicator, but it's still not outside the range of normal for a recession of this size.
As usual, maybe there's something I'm missing here. But my take is that any of the normal measures of unemployment are perfectly suitable right now. What's more, we know exactly what we have to do to push the unemployment rate down: get everyone vaccinated and open up the economy completely. Then we can see where we're really at.
Will it turn out that some of the new unemployment is permanent? I wouldn't be surprised if it is. Ever since 2000 it's been normal for the participation rate to fall during a recession and then never make up all of its lost ground. This may be due to an uptake in automation or it might be something else. But whatever it is, it's not something brand new.
POSTSCRIPT: It may seem like I devoted a lot of words to something not all that important. Probably. I just find this kind of thing annoying, regardless of whether it comes from the Secretary of the Treasury or some nutball on the Fox Business Channel.
I, too, may be missing something It’s well known that I’m not a numbers guy but Kevin seems to be saying that workers who previously had jobs but are now unemployed because of the pandemic should not, strictly speaking, be counted as “unemployed” because they’re maybe going to get their jobs back eventually.
Yelled is saying that 10 percent of the workers don’t have jobs. This seems to be the correct figure unless you apply what can only be described as a highly idiosyncratic definition of who counts as unemployed.
No. It's not about whether they became unemployed due to the pandemic. It's about whether they have stopped looking for work due to the pandemic. If you stop looking for work you are no longer in the labor force. This is the normal way of looking at unemployment.
Powell (and Yellen) are trying to say that the reason someone left the labor force matters and if the reason is the pandemic then we should still count them in the labor force. This seems to me to be created to allow them to pump up their progress, as soon as the pandemic is over those people will either join the labor force and get jobs, or Powell will kick them out of his measure because there's no special circumstance for them anymore and they'll show their "new" measure as having gone from 10% down to 4% even though by the normal measure it went from 6% down to 4%.
That’s clearly not right either. If you became unemployed because of the pandemic and have temporarily stopped looking for work, perhaps because you feel it’s unsafe or because your particular industry (food service) has been hammered and nobody’s hiring line cooks because places are closed, you are clearly still as much a part of the work force as if you trudged from closed restaurant to closed restaurant all day everyday. It seems to me that describing that person as unemployed seems much more rational and reflective of reality than the method Kevin’s using.
All this concern about precise values of unemployment stems firstly from the disproven dogmas that economists use to relate unemployment to inflation (the Phillips curve and the NAIRU); and secondly to the disproven dogma that the Fed controls inflation by adjusting federal funds rate. The Maestros and the economists who love them think that by watching unemployment they will know when to raise interest rates to head off the dreaded inflation. But both of these types of relationship were disproven in the 70's, as there was high inflation combined with high unemployment (stagflation); and high inflation despite record high levels of federal funds. They were further disproven in the late 90's and up to 2020 as unemployment (by any measure) went very low without causing inflation; and since 2009 as record low levels of federal funds (and even negative rates in Europe and Japan) did not cause inflation - or increased investment for that matter.
The fact is that central banks do not control economies by adjusting interest rates. The belief that they do prevents governments from taking the actions that really do affect economies, in terms of tax rates, regulations and fiscal actions.
Uh, no. Just no. The problem is the Fed controls not fiscal, but monetary policy. And Congress has long since shirked if not outright abandoned its responsibilities in this regard.
Oh, c'mon Kevin. This was clearly an 'optics' thing, not an 'analysis' thing. You can't criticize Democrats as bad at messaging when they stick to purely factual/analytic content and then complain when they serve one up for the cameras.
One tell for those not in the know is the suspicious roundness of the figure, btw. Note that what's being tossed out isn't 10.5 or 11 or even 9. It's a round 10.
Well, I'll be doing my bit to help with the UE problem. As soon as I get the keys to the new place, Ima be hiring a metric shitton* of experts in a variety of fields to get the old place up and running.
*Not to be confused with a shitload, which is totally different.
This whole post is way off.
Look at LFPR for 25-54 over the past 3 recessions. You'll see at most a 0.5ppt decline in each.
Look at the decline in the current one. You'll see a 3ppt decline.
This one is different, and it's in part a measurement problem. Every labor economist has been agonizing about this since the beginning, but Kevin is wading through without even understanding what everyone is worried about.
Well, given that the vast bulk of the adjustments to the line that would put it close to 10% is literally "labor force decline since February 2020"... that's pretty self-explanatory. If you care about "what's the long term low over the last 20 years or so", then all you care about is eliminating the effect that gets the headline rate to the red line: misclassification error. It looks to be about 1% or so, using the Mk. 1 Eyeball, and it's probably pretty steady.
So even taking into account your heartburn over this, it would still be about 9% instead of about 10%. What's a little rounding among friends? The larger point (that overall employment is declining) still stands.
Considering it's headline rate plus (for the most part) one other thing that's easily measured and we have numbers on, this isn't any more made-up than any other simple sum. Calling it "made-up" is being unfair, especially since the underlying point is true.
In the top graph what is the most recent approximate value of U6? To me it looks like it's above 10%. If so, new measure of unemployment was needed to boost the rate to 10%.
If this were a table instead of a chart I would know for sure and I would also have a date for that most recent value.
Where is all this annoyance coming from? There is no doubt that the U-3 has severe problems as a measure of real unemployment. The U-6 comes closer to reality, but even it understates the difference between what employment numbers would look like in a strong economy and a weak one. Beyond that, the pandemic totally screws up the numbers because of actions -- like fear of going to work -- that have nothing to do with whether a person wants to be employed or not.
If you want a real measure of unemployment, you start with the employment rate -- the employment-to-population ratio. After five decades of steady growth to almost 65% in 2000, it never recovered fully even before the 2007-11 collapse, when it dropped 8 percentage points below the 2000 level. That represents close to 20 million people who would work if they could find a decent job -- including many who are not considered part of the labor force because they have reluctantly made some life choice that prevents them from being ready to accept a job immediately. Baby Boomer retirements explains some of that employment rate decline, but a lot less than some would like to claim to pretend the labor market is stronger than it is. Other than that, as well, there is little reason to believe that the employment rate would not have continued to rise after 2000, since it has in other advanced economies during the the two decades following.
The relationship between U-3 and U-6 is not 4 or 5 percentage points. U-3 fluctuates from between 50% and 60% of U-6, and U-6 was frequently about 7 points above the U-3 during the Great Recession. Because they necessarily rely on artificial constructs -- like what constitutes actively seeking employment -- these measures, as valuable as they are for comparison purposes over time, do not do a good job of representing reality as it is experienced by the population. Looking at changes in the employment to population ratio comes a lot closer to that.
The 2000 - 2011 drop in the e/pop was 7 points, not 8.