Paul Krugman does not think highly of Austrian or liquidationist theories of the business cycle. He recently remade an argument which is hard to refute “the whole notion falls apart when you ask why, say, a housing boom — which requires shifting resources into housing — doesn’t produce the same kind of unemployment as a housing bust that shifts resources out of housing.”
I think the answer is simple — labor market rents. A bit more after the jump.
Well except not for this recession. I think there is one key additional assumption on which which Schumpeter relied without knowing what he was doing (this is also a case for using models and not just stories). The assumption is that there are labor market rents so the same worker will receive a higher wage in one sector than in another (and it’s not a compensating differential).
IN Schumpeter’s story, the two sectors are capital goods production and everything else (agriculture, consumption goods production and services). The key assumption that he needs is that workers would really rather work in capital goods production, because they get higher wages there. There is strong empirical evidence that this is true.
In Schumpeter’s story as written there is a bubble with over-investment, the build up of an excessive capital stock then a recession with low investment until the aggregate capital labor ratio is brought back down due to depreciation, population growth (Solow might add exogenous labor augmenting technological progress).
Your question is “Why is prolonged unemployment required to move someone from capital goods production to other and not vice versa ?” My answer is that workers in other sectors are eager to get high wage jobs producing capital goods so there is always, no matter what, at any unemployment rate a queue of workers in line for those jobs. Thus the minimum unemployment rate ever seen is plenty to support any flow into capital goods production.
However, to get people to flow the other way takes a huge push — high unemployment. I’m pretty sure I can write down a model that works in an hour. Note I said write down not type up.
Clock starts 9:00 pm EST October 6 2009.
update:Now 9:51 pm EST October 6 2009. I haven’t been watching the clock. I claim that I have a model written down. It is not solved (no closed form solution for numerical solution).
update 2: In the cold grey light of morning, my 1 hour model looks pretty dumb. I have modified it so it isn’t totally dumb and might actually simulate it numerically. However I did run a bit over the hour as it is now 2:14 AM EST. I did sleep most of that time.