Tuesday, August 20, 2019

Wages and the Business Cycle 2


The issue of how real and nominal wages should behave over the business cycle at one point prompted a boom in business cycle models.  You have the gut feeling that real wages should be rising in an upswing and falling in a downswing, and there’s a bunch of literature – perhaps the preponderance – which at best doesn’t support the prediction of the sticky-nominal-wage models that real wages should be counter- cyclical: that they should rise in a downturn and fall in a recovery.  The question then was how to build a testable model which might be able to accommodate that view of the world.  One answer seemed to be real business cycle models.


The starting point again is the notion that the real wage has to equal the Marginal Physical Product of Labour (MPPL).  That falls out of the micro theory of the profit maximizing firm, and the simple fact is that if we don’t have some theoretical proposition like that, we have no basis at all for making theoretical predictions about the cyclical behavior of wages. 

The real business cycle strand of literature focused on things which could cause the MPPL curve to shift.  If we think about it for a bit, there are a fair number of possibilities.  An increase in the capital stock, for example, will shift the MPPL curve to the right.  There will be diminishing returns to this, so simply adding unending quantities of capital can’t be counted on to produce unending growth (as the Chinese government seems to be realizing) but giving labour more tools to work with will generally increase the productivity of labour.

So will giving labour better tools to work with: technological change will generally shift the MPPL curve to the right, and over the long run is probably a more important factor than the simple physical quantity of capital.  Realization of the importance of this factor had the effect of reviving interest in models of economic growth, and led to the development of models of endogenous growth. 

And its not just improvements in physical capital which matter – improvements in human capital will, too.  Those can come about through increased training and education, but also through experience.  When the baby boom group first entered the labour market, you had a large increase in the supply of inexperienced labour.  Inexperienced labour would tend to be less productive than experienced labour, and when you combine that fact with the magnitude of the increase in the quantity of labour, it’s easy enough to see how the wage of units of labour could be depressed even as total output was increasing.  

As the baby boom group acquired labour market experience, their productivity increased.  That factor tended to shift the MPPL curve out to the right even in the absence of any change in any other shift factors, and in the absence of a significant increase in the supply of labour there would be a tendency for real wages to increase.

These factors, though, tended to produce more in the way of trend effects than of cyclical effects.  The challenge was to develop models which could produce unanticipated upswings and downswings, while also yielding pro-cyclical real wages (real wages rising in an economic upswing and falling in a downswing).

Suppose that, contrary to the sticky-nominal-wage models, we assume that the labour market is always in equilibrium.  And suppose that we assume that the MPPL curve is subject to random shocks – i.e. that there are factors which we can’t observe which have a systematic effect on the physical productivity of labour and which change in an unpredictable manner (although it probably helps if there’s some cyclicality to the behaviour of the shocks).  Basically that there’s noise in the MPPL and that that noise is large relative to the level of the MPPL.  Large enough, at least, for its effects to matter.

In that case, a positive shock will shift the MPPL curve out to the right, at least for a while, and then a negative shock will shift it back to the left.

In order to get pro-cyclical behavior in real wages we need to assume that the labour supply curve is positively sloped and that the labour market is always in equilibrium (the s-n-w models make the same assumption about the slope of the labour supply curve, they just assume that the labour market is frequently in disequilibrium).  Then a positive shock to the MPPL curve will shift the labour demand/supply equilibrium to the right, and since the labour supply curve is positively sloped in the real wage both the real wage and the level of employment will rise, while a negative shock will shift the MPPL curve to the left and the equilibrium levels of the real wage and the level of employment will fall.  And since unemployment falls when employment rises, and unemployment rises when employment falls, this means that real wages will tend to be rising as unemployment falls and real wages will tend to be falling as unemployment rises, which tends to accord with most people’s gut feelings.

The main obstacles here are the nature of the factors which could shift the MPPL curve enough to produce business upswings and downswings cycles of realistic size, and the equilibrium nature of the model.

The s-n-w models assumed that the nominal wage didn’t adjust because labour wouldn’t accept a nominal wage cut (they’d be perfectly happy to accept a nominal wage increase, of course).  This was usually ascribed to things like generous unemployment benefits and increased union power.  In the real business cycle strand of the literature, labour allowed the real wage to move to its new equilibrium level even if that level was lower than previously (which meant that nominal wages would fall along with prices) but then decided whether they were willing to work at the new, lower real wage.  So the early real business cycle strand of the literature basically said that unemployment rose because workers withdrew labour in response to a fall in the real wage. 

That was the knock on those versions of the model.  To get procyclicality in real wages they needed to replace involuntary unemployment with optimal, voluntary unemployment.

No comments:

Post a Comment

Note: Only a member of this blog may post a comment.