>The neoclassical's violate their own marginal productivity theory every time
>they calculate the productivity of machines from the price of those
>machines. Hedonic pricing makes an even bigger mess of the whole problem.
>In the neoclassical marginal productivity theory the price of capital must
>be derived from its productivity not vice-a -versa. So simply calculating
>productivity by multiplying a nominal figure by some imputed coefficient
>derived from a guess about the relationship between processing power and
>efficiency seems like a boondoggle to me.
Have you read any of the "productivity revolution" papers (e.g. Jorgenson & Stiroh)? What an amazing pile of assumptions on top of imputations on top of estimates. It all relies on a growth accounting framework, which requires the assumption that rates of return on investment are equal throughout the economy, and can be observed with some precision. Since real investment in computers increased in the U.S. in the late 1990s, and since we "know" what the economy-wide rate of return is, we can conclude what the contribution of computer investment is to overall growth. Change an assumption or two and it all falls apart. It strikes me as delusional - the stuff that gets Ted Winslow going. I go into all this at some length in my book.
>It seems to me that it is better to simply stick with nominal values
>discounted for inflation.
But what's inflation? The economy-wide rate of inflation is the sum of specific sectoral and commodity rates of inflation, and if you're going to compute those, you've got to get into the quality issue somehow. I don't think the hedonic technique is flawed at the conceptual level, but the execution leaves lots to be desired (for some of the reasons you mention in your GIS example).
Doug