Sticky wage model test: The results are in
By Scott Sumner
In October 2009, the US unemployment rate peaked at 10%. By January 2012, it was down to 8.3%. But in January 2013 it was still 8.0%. (And “austerity” was about to be implemented.) It was during this period that a lot of doubts were raised about the sticky-wage model (perhaps with some justification.) Other factors such as sharply higher minimum wage rates and 99 week unemployment insurance probably also played a role.
I saw one persistent line of criticism, and two alternative hypotheses. The persistent criticism (especially from the right) was that it didn’t seem plausible that wages would be sticky for so long. You’d think that by the time 3 or 4 years had gone by, wages would have adjusted. I responded by pointing to the zero lower bound problem on nominal wage growth (money illusion) and also that NGDP growth was unusually low in the recovery, so the US was continually being hit by (modest) negative demand shocks. We were seeing 4% NGDP growth, not the 11% of the first 6 quarters of the Reagan recovery. Money was effectively “tight”.
The two alternative hypotheses were as follows:
1. We are moving toward a permanent condition of higher unemployment, as technology makes many low-skilled workers unemployable. Especially low skilled men who (culturally) aren’t suited for office work.
2. The Obama administration was creating all sorts of disincentives to employment, such as Obamacare.
In contrast, I claimed that sticky wages were still the main problem (while guesstimating that perhaps 0.5 percentage points of unemployment was due to extended UI benefits.) I suggested that there was a lot of ruin in a nation, and that employment soared under LBJ’s highly interventionist big government “Great Society” programs (although arguably their long run effects were more negative.)
Over business cycle frequencies, I still think it’s NGDP growth and sticky wages that drives the unemployment rate. (I call this the “musical chairs model“). What can we say today about the sticky wage model debate of 2012-13? Here’s what we can say:
Fed’s estimate of natural rate of unemployment = 5.0% to 5.2%
Actual (U3) unemployment rate in August, 2015 = 5.1%
1. No, you cannot point to hidden unemployment or part timers, or U6, as the sticky wage skeptics were specifically claiming that U3 was high due to factors other that sticky wages.
2. Yes, unemployment could have fallen for other reasons–I believe Casey Mulligan has looked at some real factors that might explain movements in unemployment.
But suppose unemployment was still at 8%, despite 3 more years of 4% NGDP growth. Even I would have had to throw in the towel. After all, I consistently predicted that unemployment would continue falling if we had even halfway decent NGDP growth. And we know from Europe (post-1970s) that it’s quite possible for unemployment to rise sharply and then get stuck at 8% or 10%. So here is one point I’d insist on:
Suppose both sides of the debate over theory A believe that outcome B would tend to discredit that theory. And suppose opponents of theory A think outcome B is quite plausible (but not certain.) Then if outcome B does not occur, that provides one significant data point in favor of theory A.
Here’s an analogy. I’ve predicted 6% Chinese growth. If it comes in at 0%, then I’m wrong. Full stop. If it comes in at 6.2% or 5.9%, I still might be wrong, as the Chinese government might be faking the data. But it’s beyond dispute that 6.2% is better for my prediction than 0%.
The sticky wage model successfully dodged a bullet:
Nothing in life is so exhilarating as to be shot at without result.