Are wages sticky?
New Keynesian models are adaptations of New Classical models that use the mechanism of sticky wages — wages that don't adjust in a timely manner — to explain why the job market doesn’t clear.
This graph, via Roger Farmer suggests that in the real world wages aren’t very sticky at all:
As Farmer argues: “the evidence from the Great Depression is that wages and prices are remarkably flexible. During the first six years of the Great Depression, nominal wages and nominal prices fell by thirty percent.”
A similar story was found in Britain during the recent Great Recession. In a study by Michael Elsby, Donggyun Shin and Gary Solon entitled “Wage adjustment in the Great Recession” they found that the proportion of workers in Britain experiencing negative nominal wage changes went from a low of 5% from 1979-1980 to a high of 24% in 2009-2010. To state that more clearly, almost a quarter of UK workers saw their take-home pay cut in 2009-2010.
And a more detailed discussion can be found in Truman Bewley's Why Wages Don't Fall During A Recession.
As David Glasner notes, the sticky wage assumptions seems to be an almighty fudge, and an inversion of Keynes’ ideas: “The association between sticky wages and Keynes is a rather startling, and altogether unfounded, inversion of what Keynes actually wrote in the General Theory, heaping scorn on what he called the “classical” doctrine that cyclical (or in Keynesian terminology “involuntary”) unemployment could be attributed to the failure of nominal wages to fall in response to a reduction in aggregate demand. Keynes never stopped insisting that the key defining characteristic of “involuntary” unemployment is that a nominal-wage reduction would not reduce “involuntary” unemployment. The very definition of involuntary unemployment is that it can only be eliminated by an increase in the price level, but not by a reduction in nominal wages.”
So neither new, nor Keynesian (nor holy, nor Roman, nor an empire).
So why do New Keynesian models rely upon this “classical” mechanism to produce a simulacrum of labour market disequilibrium? The answer may be that this is a convenient mathematical modelling technique. And ultimately, if there is labour market disequilibrium in the real world it is better to have a model that allows for the possibility of labour market disequilibrium. But if the mechanism of labour market disequilibrium used by New Keynesians to explain the data is an empirically unfounded one, then New Keynesianism has a big weakness.
Why? In more simple terms, imagine we are modelling the black death. Clearly we know the symptoms of the illness — buboes, fever, etc that lead to death. So our model must be consistent with that empirical finding. But it is not enough just to have a model consistent with the symptoms of the illness (or unemployment) if the theoretical mechanism explaining the illness is unsatisfactory. If our posited mechanism — let’s say that our model imagines that the bubonic plague is caused by a magical space rabbit wishing doom upon the human race — has no empirical support, we must find another explanation which does have empirical support. If you do not explain the mechanism (in the case of the black death, bacteria) causing the illness, it is much harder to be able to find an effective treatment.
Roger Farmer argues: “High and persistent unemployment is a problem. But it has nothing to do with inflexible wages or sticky prices. Both Classical AND new-Keynesian models are broken. It's time to think outside the box!”. I agree with him.
Farmer’s work draws on Keynes’ notion of animal spirits to explain unemployment. That is probably a good starting point.