Subject: [OPE-L:1894] Re: productivity increases and rising real wages
From: Patrick L. Mason (pmason@garnet.acns.fsu.edu)
Date: Sat Dec 11 1999 - 18:40:00 EST
Mike:
First, I'm glad you responded. I thought my postings had fallen into
bottomless pit of nonresponse. I don't think we have any major
disagreements. But, your questions help to clear up positive points of
misunderstanding. Please excuse any typos below.
Second, you write:
I. In #1852 Patrick indicated "the job is the unit of analysis" and
proceeded to list as determinants of the ability to pay-- "competitive
characteristics of the firm, characteristics of the labour force". This
appears to be a formula suited for cross-sectional analysis and not for
exploring the effect of increasing productivity and its relation to real
wages over time. Further, if the focus is on individual firms, what is the
relation to the macro-level? To what extent does this set out key variables
as determined at the level of competition rather than at the level of
capital as a whole (a point about which Fred has repeatedly cautioned)?
Yes, when I list competitive characteristics of the firm and
characteristics of the labor force as determinants of ability to pay, I am
primarily concerned with cross-sectional analysis. In particular, I am
concerned with intra- and interindustry wage differentials.
You are also correct - and it is helpful for the discussion - to point that
there is a relationship between the aggregate or macro level of analysis
and micro issue of wage differentials.
>From a theoretical perspective, we need a way of making micro and macro
analysis consistent with each other, a unified theory as they say in
physics. What of the geniuses of Marx's analysis is that value theory
provides a unified theory, a way of consistently aggregating up from micro
units (the job when we are discussing the labor market) to capital as a
whole. The simplistic or popularized version of Marx's wage that I proposed:
Wage = f(ability to pay, ability to make pay)
Works quite well at the aggregate level. You are correct to point out that
divisions within the working class reduce the bargaining power of labor.
This is true at both the aggregate and micro levels. Indeed, it is the key
to the theory of discrimination that I point forward in two CJE articles.
Three, you further write:
If necessary labour time is determined at the level of the whole, though,
so is productivity, class struggle and the real wage. So, one question I
have for Patrick is that, given that your argument still appears to focus
on the individual industry (and that I suspect you have already done
empirical work on this), how do you move back from the level of the whole
to that of the individual firm and industry? Do your variables take the
form of sectoral deviations from the average (productivity, degree of class
struggle, etc) for the whole? Does the question of the value composition of
capital enter into this (directly or directly) at the industry level? Also,
what is your variable representing class struggle?
Marx's value theory allows movement between the macro and the micro.
Indeed, I strongly believe that it is most appropriate to start with the
analysis of capital as a whole to derive the general relationship between
labor and capital, that is, to derive the aggregate relationship between
productivity, class struggle, the real wage, and joblessness. My last reply
to Ajit discussed to issues. Also, Botwinick's book does this in great deal
and he sticks with Marxian value categories. Moreover, Botwinick makes an
excellent transition from value analysis to the monetary form of value
categories and their importance in wage theory.
Empirically, I have no data on the value composition of capital by
industry. Most of the data sets that I worked were put together by human
capital theorists. So, it's almost impossible to do Marxian wage analysis.
This will change very shortly. There is a government project by John Abowd,
a neoclassical economists, to make employer and employee characteristics
and labor market outcomes. This may become the dataset of choice for
Marxian economist interested in doing empirical work. I did publish a paper
that examine the impact on capital intensity on industry wage rates. {See
Patrick L. Mason. (1994). "An empirical derivation of the industry wage
equation." Journal of Quantitative Economics, 10(1) (January):155-170}. I
recognize that capital intensity is not the same thing as the value
composition of capital, but you gotta work with whatever data is available
until you get better data.
In my CJE 1999 that emphasized the importance of class struggle in relation
to racial inequality, I didn't attempt to message class struggle. Rather,
the empirical strategy was to examine whether such factors as the
race-gender composition of job would have the effects on individual wages
that one would predict in a Marxian wage framework. So, empirically, the
variables one uses to measure class struggle really depends on what
available. Sometimes nothing is available, then one is confined to
measuring the predicted effects of class struggle.
Four, you write:
Clearly, "the bargaining power of labor" (ie., "ability to make pay")--
both at the level of whole and firm-- depends on much more than nonmarket
opportunities. I've proposed the "degree of separation among workers" (or,
its opposite, the degree of unity) as a variable meant to capture the
balance of class forces and wonder whether you see any possibility of
introducing this empirically. (Eg., to what extent could measures like
degree of unionisation, wage dispersion, size of bargaining unit, etc be
introduced as proxies?)
We are in agreement. Both the quantity and the quality of worker
organization determined the ability to make pay. As I said before, I used
this insight to develop my theory of persistent discrimination. The
specific variables employed in an empirical study depends on both the
endogenous variable under analysis and what's available in the dataset and
the degree of one's imagination.
Five, you write:
What concerns me here is the place assigned to productivity in the
determination of the real wage (di). If we grant that an increase in
productivity raises the upper bound for wages, *why assume a link between
(the limit to) what capital is able to pay and what it does pay*? Isn't the
willingness of capital to grant increased wages already captured by the
variable for class struggle? Linking the wage to what capital is able to
pay as you do would seem to be a way by which the neoclassical nexus
between productivity and the wage slips in--- an alien intrusion that Ajit
has been particularly concerned to warn against. In short, I would argue
that a Marxian version (once we leave Capital's fiction of the given real
wage) of the determination of that standard of necessity depends only on
class struggle (current and past); and indeed in your own numerical example
it is only the "sufficient bargaining power" of workers which permits them
to secure increased real wages as the unit values of wage goods fall.
My argument is simple. During a given time period, say during a given
business cycle of 8 - 10 years, technological change and productivity
growth may have no impact on the composition of average consumption bundle.
But, within an 8 - 10 year period it is entirely possible that
technological change will lower the labor content of several components of
the consumption bundle. This is not neoclassical in the least. In the
neoclassical analysis productivity changes automatically raises wages. I'm
not ascertaining that at all, only that technological lowers the value of
labor power. Over a longer time period, say a generation, technological
changes alters the consumption bundle because it alters the socially
required commodities. For example, I once had the extreme displeasure of
trying to search for a job without owning a telephone or even having a
neighbor who had a telephone. This severely hurt my job search. Telephones
in modern america are socially necessary. This will be increasing true as
digital telephones, which are just portable computers, handle data
transmission and other information technology stuff. Technological change
even affects the capacity to wage class struggle. For example, the
existence of OPE-L allows Marxists in Western Canada, Southern US, and
Europe to simultaneously communicate and thus considerably reduce social
and intellectual isolation and transmit ideas, knowledge, etc.
Sixth, you quote me and write:
"There are also two additional factors which enter into the determination
of the value of labor value of labor power: (1) the cost of training the
laborer; and, (2) the labor market participation of labor of women and
children, which "makes a great difference in the cost of maintaining the
family of the laborer, and in the value of the labor-power of the adult
male." (Capital, volume I, page 569).
VLP = Slidi + ljtj + g, where tj represents the commodities and living
labor used to train workers and g represents the extent of labor market
participation of women and children."
Aren't these factors, however, already captured in Slidi? Also, I assume
that ljtj refers to the private cost of training, no?
Yes, this is correct. These comments were taken out lecture notes so they
are partly out of context.
Peace, patrick l mason
This archive was generated by hypermail 2a24 : Sun Dec 12 1999 - 15:45:04 EST