[OPE-L:2223] Re: Re: Statistical regularities

["clyder" <wpc@xxxxxxxxxxxxx>]

 > > F&M argue that
> > market power, collusion, etc. undoubtedly reduce the number of degrees
of
> > freedom in the system (increase the amount of co-ordination among
agents),
> > but in practice not by enough to significantly reduce the number of
> degrees
> > of freedom.
>
> I would appreciate a bit of an account of what is 'enough', 'not enough'
and
> 'significant' here.

I would say that F&M's argument starts to operate once the system
has a few tens of degrees of freedom.

Perhaps Julian and Allin might have some estimate here.


At the scale at which one can test the hypotheses readily, i/o tables
with a scale of about 100 industries, the price/labour ratio does apear to
have the statistical properties that they predict. By the time you reach
this scale we are clearly in the stochastic rather than deterministic
domain. At the scale of 3 industries at which Steedman works, the
number of degrees of freedom are too small for stochastic effects
to operate. Somewhere in between I reckon there will be a phase
change.

I am speaking from memory here, but if I recall correctly Steedman
published some empirical work on the economy of the Australian
state of Victoria in the Cambridge Journal a couple of years back
where the scale of the I/O tables was somewhat smaller than
those used in most of the literature. He obtained results that
were somewhat out of line from those found in the rest of the
literature. It is possible that the smaller sample of industries
used meant that stochastic effects were weaker at the scale he
worked.