Re: [OPE-L] Sraffa and the productivity paradox

From: Paul Cockshott (wpc@DCS.GLA.AC.UK)
Date: Tue Sep 20 2005 - 05:52:22 EDT

I have seen neo-classical ecnomists using this argument which may be
right. What seems to me to be interesting is the ambiguity of the
definitions of productivity and the attempts to compute the 
marginal efficiency of computer capital using a Cobb-Douglas function.

This model differs from Sraffa's in that it is non-linear and 
causality operates in the reverse direction. 
Sraffa says that production of 1 ton of iron  uses up
0.4 ton of iron and 6 qrtr wheat. 
The neo-classical model says that if
we  put in quantities K_1, K_2,L of inputs, 
then we will produce Y of output.
Perhaps most significantly, the 
Sraffian model measures all inputs and outputs in
physical terms whereas the neo-classical 
model measures them in money terms.
From the Sraffian point of view the 
measurement of capital in money 
is a serious flaw since the valuation 
of commodities depends upon the 
distribution of income between labour 
and capital. One can thus not hope
to measure the productivity of aggregations 
of capital goods since the
valuation of these aggregations is itself 
a function of the class distribution
of income.

From a classical standpoint the notion of 
productivity measured in money
terms was ill-defined. The only context 
in which one could define productivity was
as the inverse of labour values, an increase 
in productivity was then equivalent
to a fall in the labour required to make goods. 
Sraffa added to this concept the idea of the 
productivity of the basic sector measured in terms of its own
inputs. One could in principle measure R for 
different years and see if it has gone up after 
the introduction of computer technology. Since there
were many other technical changes at the same time, 
it would be hard to say whether such an increase 
in R might have stemmed from computer technology or from
other innovations.

Beyond this point though, the concept of 
the basic sector may provide another
reason why productivity gains due to 
computers are so hard to discover.
Since computers are largely used in 
non-basic sectors, Sraffian theory predicts
that they will leave R unchanged.

Of course investingation R is not the same thing
as investigating productivity in a neo-classical sense,
but it is the nearest analogue.

-----Original Message-----
From: OPE-L [mailto:OPE-L@SUS.CSUCHICO.EDU] On Behalf Of Michael
Sent: 19 September 2005 18:19
Subject: Re: [OPE-L] Sraffa and the productivity paradox

Paul David uses the example of electricity to make the case that the
productivity boost
comes with a lag.

On Mon, Sep 19, 2005 at 12:32:45PM -0400, Gerald_A_Levy@MSN.COM wrote:
> >   The productivity paradox that I refer to is the observation
> > made by Solow, and Roach that computers do not seem to
> > have made a significant measurable contribution to productivity.
> <snip, JL>
> Hi, Paul C:
> This is an interesting issue, but one that is hard to address
> abstractly.  Putting aside the issue of how productivity is
> measured in standard theory, the answer to the "productivity paradox"
> might not be found at the aggregate level.  If one were, however,
> to consider why productivity might not have increased in
> individual branches of production and sectors after the
> introduction of  specific computer technologies, then one might
> come up with a number of explanations.  E.g. the reason that
> productivity (as conventionally measured) hasn't increased by the
> amount anticipated after computers were widely diffused as
> means of production in offices is quite different from the reasons
> why productivity hasn't been increased in many cases following
> the adoption of industrial robotics in assembly-based forms of
> manufacture.  Thus, while this might seem to be a 'macro' issue,
> the answers might be found only on the 'micro' level.
> In solidarity, Jerry

Michael Perelman
Economics Department
California State University
Chico, CA 95929

Tel. 530-898-5321
E-Mail michael at

This archive was generated by hypermail 2.1.5 : Wed Sep 21 2005 - 00:00:02 EDT