[OPE-L:7079] [OPE-L:577] Re: New Evidence on Sectoral Prices and Values
Allin Cottrell (cottrell@ricardo.ecn.wfu.edu)
Tue, 2 Mar 1999 23:26:15 -0500 (EST)
In response to Andrew's report on his simulations. I've tried
exploring the space of such simulations a bit further. The
results seem to depend on the variance of the value/"cost"
ratios, and also the statistical characteristics of the
simulated random variables (e.g. shape of distribution).
Depending on the settings, I can reproduce both the results I
reported here before, and those reported by Andrew.
The main conclusion I draw at this point is that if one takes,
e.g. the Shaikh specification of price-formation as null
hypothesis, then tests of the sort reported in Andrew's paper
have an uncontrolled probability of type 1 error (rejecting a
true null hypothesis), and that the probability seems to
approach 1 as the variance of value/scale-factor approaches
zero.
Allin Cottrell.