Obviously, the main problem with high unemployment right now is the joblessness for those looking for work and those who’ve given up for lack of opportunity, along with the absence of wage pressures for the millions of active workers who depend on tight labor market for the bargaining power they otherwise lack.
But there’s another big problem, one that’s more technical in nature: what is the unemployment rate?
If your answer is 6.7% as per the Bureau of Labor Statistics, you’re low-balling it. Because of the sharp decline in the labor force in recent years, that number provides a downwardly biased take on labor market slack. Estimating the magnitude of that bias happens to very important, because the tautness of the labor market as measured by the unemployment rate is a key policy input both at the Federal Reserve and Congress.
For example, the Fed has signaled that they view 6.5% as the jobless rate at which they’ll start thinking about raising the federal funds rate (the key interest rate they set to raise or lower the cost of borrowing; it’s been near zero for years). Well, if you believe the BLS number, they’re almost there. But if you think the official rate is biased down, the Fed is still some ways away from that target. (To be clear, none of this a criticism of the BLS, one of my personal favorite agencies.)
It’s as if your speedometer is off kilter such that when you’re driving 40 mph it says 60 mph. Under those conditions, you’d be likely to put on the brakes to slow down before you really wanted to.
A tiny bit of arithmetic will help here. Typically, the unemployment gap is thought off as u – u*, where u is the actual rate (6.7% right now) and u* is the lowest unemployment rate the economy can handle without triggering spiraling price growth, or the NAIRU (non-accelerating inflationary rate of unemployment). If you believe the CBO’s NAIRU of 5.5%, that gap is a not-all-that-big 1.2 percentage points (ppts) right now: 6.7-5.5=1.2
But something’s missing: a term to reflect the gap in labor force participation. That is, there’s considerably more slack in the job market than u – u* reveals, because a bunch of potential workers have given up looking for work and are thus not counted in ‘u.’ Another way of saying this is that u is biased downward. If that bias amounts to, e.g., 2% of the labor force, then u – u* should really be 8.7-5.5, or 3.2 ppts.
In other words, these days the proper slack equation is u + lfg – u* where lfg is the labor force gap (or, using the 2 ppts I just mentioned, 6.7+2-5.5).
Now, there’s a cottage industry among economists trying to guesstimate lfg so as to adjust that equation and get a more accurate bead on labor market slack. The labor force participation rate, as I noted after last jobs report, is down about three-and-a-half ppts off of its pre-recession peak, but demography is at work here, as opposed to weak demand, as aging boomers leave the labor force for retirement. So there’s lots of argumentation as to how much of the decline to assign to cyclical slack and how much would have happened anyway. Or to perhaps put the question more intuitively: how many of those ppts would come back if the job market really strengthened?
I’m not going to spend the time to go through all of the estimates, but I think this is a fair summary: most analysts have assigned most of the drop to the cycle. A few go the other way, arguing it’s mostly demography, and it’s tricky, because lfg is a moving target (in fact, for most of the data history, the labor force isn’t very cyclical at all). I thought this paper takes a particularly rigorous look at the question and they argue that the decline is pretty much all cycle, at least through the beginning of last year. But other careful work disagrees. (Brad Plumer usefully reviews the issues here.)
So let’s be conservative and say that 2/3 of the decline in the labor force rate is “fixable,” i.e., at full employment it would eventually be reversed.
Then the unemployment gap is: 6.7 + (2/3*3.4) – 5.5 or about 3.5 ppts (that 3.4 is the drop in the labor force rate using quarterly data since its pre-recession peak). And you will very quickly notice that 3.5>1.2, the latter being the naïve gap as per u – u*.
One last thing. I wouldn’t be so quick to plug 5.5 into all those equations above. As Dean and I argue, the NAIRU is probably biased up as well. If you use, say, 4.5 instead that’s another percentage point of slack.
Fed chair Janet Yellen (still getting used to writing that!) knows this stuff but it’s still worth making a lot of noise about. It’s certainly not well understood by those members of Congress arguing that the decline in the jobless rate means they don’t need to extend UI.
The real unemployment rate is a good bit higher than 6.7%–forget the speedometer: we’re not going 60…we’re going 40. So keep that foot off the brake for now.
Update: I just noticed that the great labor economist Heidi Shierholz posted a piece by the same name about the same thing about two years ago. So, a) sorry to crib your catchy title, Heidi, and b) this problem ain’t new.