Where Are These Jobs Coming From?

|
Includes: DDM, DIA, DOG, DXD, EEH, EPS, EQL, FEX, FWDD, HUSV, IVV, IWL, IWM, JHML, JKD, OTPIX, PSQ, QID, QLD, QQEW, QQQ, QQQE, QQXT, RSP, RWM, RYARX, RYRSX, SCHX, SDOW, SDS, SFLA, SH, SMLL, SPDN, SPLX, SPUU, SPXE, SPXL, SPXN, SPXS, SPXT, SPXU, SPXV, SPY, SQQQ, SRTY, SSO, SYE, TNA, TQQQ, TWM, TZA, UDOW, UDPIX, UPRO, URTY, UWM, VFINX, VOO, VTWO, VV
by: Scott Sumner

As always, the jobs market remains a bit of a mystery. If you think you have an explanation for the recent jobs growth, I'm about to show you that you are wrong. (Notice I didn't say "US jobs market", which is already a clue that the mystery is even deeper than we imagine.)

Job growth has been running at around 200,000 per month, and the unemployment rate has fallen to 3.7% (lowest since the 1960s.) It's best to start with the accounting, which basically involves three factors: population growth, the labor force participation rate, and the unemployment rate. You can use prime age labor force participation, but that makes things more complicated and also misses the growth in older workers.

In the last few years, employment has been growing faster than predicted by growth in adult population, which can only mean that either adult labor force participation is rising, or the unemployment rate is falling. It's not labor force participation, which was falling and then leveled off some time around 2014 or 2015 (it's hard to be precise, as the data is noisy.)

Some people might start screaming that I'm ignoring the aging of the population. I'm not. It's true that the population is aging, and it's true that this means the leveling off of LFPC rate is actually a very good thing; it shows participation is rising among prime age workers. It might even show that Trump is the greatest president in history. But it does not explain the recent growth in employment. It simply suggests that we should be doing less bad than the aging alone would predict.

To fully explain the recent growth in employment (of 200,000/month) in an accounting sense, you need to look at the unemployment rate, which has fallen to shocking low levels:

A couple years ago, I expected employment growth to slow by now. The main reason I was wrong is that I expected the unemployment rate to level off in the mid-fours, and it instead fell to 3.7%. I don't have any special ability to forecast, I was just going with the conventional wisdom:

In September 2016, for example, the median forecast of Federal Reserve officials was that the unemployment rate would be 4.5 percent at the end of 2018; it now looks likely to be substantially lower.

I also expected a tad worse performance for the LFPR, and rising prime age participation is part of the story. But unemployment is the big mystery that needs to be explained.

There are two possible explanations for the very low unemployment-a fall in the natural rate, or a demand shock that pushes unemployment below the natural rate. I would not completely rule out the latter, but Neil Irwin of the NYT points to a problem with demand-side explanations:

The even better news is that the last time the jobless rate was this low, at the end of 1969, it was already fueling high inflation. Consumer prices rose 5.9 percent that year. Currently, that measure is 2.7 percent.

In fairness, that 2.7% figure is consistent with somewhat of a demand boost, thus it's not that different from the situation in 1966 (when inflation was about the same). But, on balance, I don't see much evidence that this is a demand-side issue, partly because the inflation figure includes recent oil price increases, and inflation forecasts continue to run at around 2%. In a couple years, we'll have more perspective on this issue, but right now, I'm going with the supply-side explanation, i.e. an unusual fall in the natural rate. Later, I'll provide international evidence for that view.

I've been racking my brain for reasons why the natural rate of unemployment should have fallen to perhaps the lowest levels in history (actual unemployment was below the natural rate during the Korean and Vietnam Wars), but it's not obvious what those are. In a recent Econlog blog post, I discussed the fact that the federal minimum wage was now so low as to be almost meaningless. But that affects only a small part of the labor force. You could point to Trump initiatives like a corporate tax cut that boosted RGDP growth. But the same occurred during the Reagan boom, and yet the unemployment rate never fell below 5%, even after very strong RGDP growth spurred by tax cuts and deregulation.

At this point, I look to other countries for assistance. Europe has a very different unemployment pattern than the US. During the post-WWII boom, they had an extremely low natural rate, relative to the US. During the 1980s and 1990s, their natural rate rose far above the US, even in Germany. It remains elevated in France and southern Europe, but has recently fallen in the UK:

And in Germany:

So, there is modest evidence that this phenomenon of falling natural rates is affecting other countries. Canada is a bit more ambiguous, with a higher natural rate than the US, but some evidence of a downward trend since the 1980s:

I don't have a good explanation for why Canada has a significantly higher natural rate than the US. It may have a bigger welfare state, but the same is true of the UK and Germany.

So, unemployment remains something of a mystery. By the end of the Obama administration, unemployment had already fallen below the lowest levels of the Reagan boom (to 4.6% on November 2016), despite slower growth and less business friendly regulation. So, while I would not rule out the importance of supply-side policies, especially the tax cuts, I think a portion of the story remains unexplained.

Note that Germany and the UK have seen their unemployment rates fall dramatically to the 3.5% to 4% range, despite tight fiscal policies. So, Keynesians should be just as confused as supply-siders. Fiscal policy has also gotten tighter under Abe, and Japan's unemployment has fallen to 2.5% (although they've traditionally had very low unemployment.)

Before commenting, think about how your explanation fits the international pattern, and also the US time series going back to the 1940s. It's harder than you think. By the way, we won't be able to figure out the trend rate of growth in RGDP until we can observe a year of two of RGDP growth with a stable unemployment rate. The wait continues...

If you want an even longer time series, there's a UK graph going back to 1760, when unemployment was 3.63%. (Love that precision!)