RBC and u* (for 30 Aug)
David Altig, Angry Bear, and Max, among others, have been discussing this paper by Robert Hall, in which he discusses neoclassical economics and Kydland [Thank you, Tom] and Prescott's RBC (real business cycle) model in loving detail.
Hall notes that the neoclassical model is an economy in which "production functions, consumption demand functions and labor supply functions" are "embedded in markets that clear."
Now, if the markets clear--that is, if all transactions make economic sense to all participants, given their information at the time--why would the same not be true of the "natural" rate of unemployment (u*)? It would seem that the neoclassical model implies that any unemployment rate above zero would be a sign that potential GDP (y*) has been lost.
What am I missing?
Hall notes that the neoclassical model is an economy in which "production functions, consumption demand functions and labor supply functions" are "embedded in markets that clear."
Now, if the markets clear--that is, if all transactions make economic sense to all participants, given their information at the time--why would the same not be true of the "natural" rate of unemployment (u*)? It would seem that the neoclassical model implies that any unemployment rate above zero would be a sign that potential GDP (y*) has been lost.
What am I missing?
posted by Ken Houghton at 11:55
3 Comments:
Typo alert: it's Kydland.
At the risk of betraying how long it's been since I've thought seriously about RBC models, I'd think that (among other things) some combination of voluntary (temporary) quits and search in the labor market model would provide u*>0 w/o introducing what would really constitute a market failure.
Thanks. (Thought I copied that directly from the JSTOR ref...)
If, as the model professes, all market transactions clear (i.e., are acceptable to both sides based on the known information), then why would one assume that labour transactions are any different?
Certainly, there will be a transition period for a positive but small number of people at any given time. But that would be predicated on the idea that (for instance) a person working in Madison, WI, found a preferable position in, say, Cambridge, MA. (Again, I am assuming that the known information is not asymmetric, since the model appears to do so--Hall appears to credit "variance" to informational asymmetries.)
I'm also defining unemployed to exclude those who cannot work, so I grant a possible state transition for the firm--but that also would not affect the unemployment rate (which would remain naturally zero).
There are plenty of mathematical reasons that the natural unemployment rate cannot be zero; what I'm trying to find is a conceptual reason in the model (as presented by Hall, and possibly misinterpreted by me) that the natural rate of unemployment should not be assumed to be zero.
Aha! I just took a look at two of Hall's cited papers. this working paper posits a constant separation rate with an endogenous probability of rehiring in the current (vs. next) quarter. His "Employment Fluctuations with Equilibrium Wage Stickiness" (AER, March '05) posits an equilibrium wage model taking "the standard view of matching friction," also with a fixed hazard of separation and an explicit job search "technology" producing probability that an unemployed worker is matched with a job in any given period.
The upshot is that it probably is necessary to introduce labor market friction to get u*>0, but that's what modern RBC people evidently do to get around the problem (mentioned by Hall) that K&P needed unrealistic labor supply elasticity assumptions.
Post a Comment
<< Home