[PEN-L:1203] Return on R&D - Cobb-Douglas macro production function

["Rosser Jr, John Barkley" <rosserjb@xxxxxxx>]

     I would say that the criticisms are at two levels.
The simple-minded level says that, well, this is a
restricted mathematical form that is convenient and was
invented to justify/explain an aggregate constant breakdown
between wages and profits in the US in the early part of
the twentieth century.  This critique then suggests, oh
goody, let's use more generalized production function forms
such as the transcendental logarithmic that allow for
economies of scale, interactions between factors, and all
sorts of other fun stuff (just stepped out of a seminar on
internal adjustment costs where somebody did exactly this
leap, from C-D to translog).
     The other is more fundamental and says that any
production function estimated on value of capital stock at
any aggregated level is bogus.  This is what one gets from
the Cambridge capital theory debate and critique literature.
Barkley Rosser
On Thu, 3 Dec 1998 22:37:11 +0100 (MET) Anders Ekeland
<anders.ekeland@xxxxxxxxx> wrote:

>
> I am in the process of writing a small and non-mathematical survey paper on
> returns to R&D and a natural starting point was the the newly published
> collection of articles by Zvi Griliches "R&D and productivity".
>
> Reading it I felt a strong need for a detailed exposition/critique of the
> Cobb-Douglas production function. Something like Anwar Shaiks paper
> published in the Ed. Nell anthology in honor of Joan Robinson. Only more
> detailed and preferably adressing the work of Zvi Griliches and Co. Not that
> ZG is uncritical, he becomes progressively more and more aware of the
> limitations of the C-D approach.
>
> But still I feel sceptical to the use of C-D for various reasons:
> -the heterogeniety of firms
>
> - the time dimension of R&D projects
>
> - the very skewed distribution of success (most small firms die - how is
> that accounted for)
>
> - what about investment cycles (everybody is now investing heavily in R&D -
> substantial benefits are years ahead).
>
> For short: I have a lot of naive questions about the use of C-D in general
> and also for this particular purpose. Is C-D really a valid approach to the
> measuring the return on R&D?
>
> All hints are welcome.
>
> Regards
> Anders Ekeland
>
>
>
>
>
>

--
Rosser Jr, John Barkley
rosserjb@xxxxxxx