Paul C wrote [#4047] > Why would the distribution of the log of the organic compositions be > gaussian? > > Does not matter if you take industries or firms. > > occ = c/v > > in log form = log c - log v > > if c and v are random variables, log occ is then in the form > of a sum of random variables, and thus will have a Gaussian distribution > > Is this by some version of the central limit theorem? Doesn't one need a sum of rather more variables than two to get this? It seems a very strong result: if true, any random variable which is the ratio or product of two others must have a log-normal distribution -- including, among others, not only the rate of profit and the rate of surplus value, but also Farjoun and Machover's psi (specific price). F&M claim that the first has a gamma distribution, the second is "almost" degenerate (tho' I suppose this doesn't rule out an "almost-degenerate" log normal), and the third normal (in fact, they use the central limit theorem to get this last). Paul, were you writing in haste here? Julian >
This archive was generated by hypermail 2b29 : Tue Oct 31 2000 - 00:00:09 EST