Pigouvian Exploitation of Labor

Author(s): Joseph Persky and Herbert Tsang

Source: The Review of Economics and Statistics, Vol. 56, No. 1 (Feb., 1974), pp. 52-57
Published by: The MIT Press
Stable URL: https://www.jstor.org/stable/1927526
Accessed: 02-03-2020 15:27 UTC

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide
range of content in a trusted digital archive. We use information technology and tools to increase productivity and
facilitate new forms of scholarship. For more information about JSTOR, please contact [email protected].

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
https://about.jstor.org/terms

The MIT Press is collaborating with JSTOR to digitize, preserve and extend access to The
Review of Economics and Statistics

This content downloaded from 137.50.152.224 on Mon, 02 Mar 2020 15:27:31 UTC
All use subject to https://about.jstor.org/terms
 PIGOUVIAN EXPLOITATION OF LABOR

Joseph Persky and Herbert Tsang*

Introduction and econometric tools for estimating production

functions has made the quantification of the
T HE concept of exploitation, while primarily
concept quite feasible. While most studies of
associated with Marxian economics, has a
production have been oriented toward questions
long if somewhat neglected history within the
of growth, technological change and input sub-
neoclassical tradition. Pigou and other neo-
stitution, there is no reason why these same
classicists concerned with the welfare implica-
techniques cannot be used for exploring theories
tions of their theory found it natural to define
of the income distribution and exploitation. In-
exploitation of labor as the difference between
deed, two recent research efforts, those of
labor's marginal product and its real wage.'
Thurow, and Hildebrand and Liu, have derived
Whether or not the normative connotation of
marginal products from the partial derivatives
the term, exploitation, is deserved, the Pigouvian
of empirically estimated production functions.4
definition opens important questions for positive
While the first of these studies is based on time
economics. In particular, the level of exploita-
series data for the entire United States economy
tion can be interpreted as a measure of the rela-
and the second on cross-sectional data for man-
tive bargaining power of labor and employers
ufacturing industries, both suggest that labor
in an imperfect market economy. In this con-
is paid substantially less than its marginal prod-
text, Pigouvian exploitation provides a useful
uct.5
way of introducing and testing many of the in-
Given the existence of Pigouvian exploitation,
teresting insights of bargaining theory in a more
the natural question is what factors determine
rigorous framework.2
its extent. The purpose of this paper is to ex-
Despite its theoretical importance, Pigouvian
plore this question at both the macro and micro
exploitation has received little empirical atten-
levels. Specifically, we concentrate on testing
tion.3 However, the development of reliable data
several well-known (if somewhat intuitive)
arguments drawn from labor economics. The
Received for publication December 18, 1972. Revision ac- emphasis on the level of exploitation, rather
cepted for publication May 28, 1973.
than on the real wage rate, allows for the sup-
* We would like to thank the referee and our colleagues
of the Department of Economics, University of Alabama in position that in a market economy, wages are
Birmingham for many helpful comments. limited by marginal productivity. Hopefully
1A. C. Pigou, (1952) pp. 555-560 and pp. 883-884. To this effort will demonstrate the importance and
be sure, this definition has been subject to a good deal of
criticism on the grounds that such a divergence may not feasibility of research concerning Pigouvian ex-
be a deliberate result of employer actions. For a discussion ploitation.
of this minor controversy see A. M. Cartter (1959), pp. 65-
70. Moreover Pigou, himself, was careful to distinguish the 4 Lester Thurow, (1968), G. H. Hildebrand and T. C. Liu
level of exploitation from the level of unfairness, i.e., the (1965). Both Thurow and Hildebrand and Liu used variants
divergence between factor payments and marginal produc- of the Cobb-Douglas production function estimated directly
tivities which would hold in a perfectly competitive economyfrom data concerning labor and capital. Thus, they avoid the
in equilibrium.
generally indefensible approach which uses wage data to
2 The authors feel that the normative interpretation estimate
of production functions on the assumption that mar-
exploitation is important. The common content of principlesginal products are equated to wage rates by an invisible
courses and the long standing interest in refuting the dis-hand.
tributional implications of the labor theory of value, belie 5 Thurow finds that the rate of Pigouvian exploitation,
the pure positivist position. For perhaps the strongest defined as the ratio of marginal product to wage rate, has
normative statement concerning marginalist analysis see fallen from about 177% in 1930 to about 159% in 1965, al-
John Bates Clark (1899). For a statement of the more wide- though the rate of decline has been sporadic. Hildebrand
spread positive approach see Schumpeter (1954), pp. 868- and Liu compute rates of exploitation of production workers
870 and pp. 883-884. to vary between 171% in the stone, clay and glass indus-
3 This lack of attention is probably due to the fact thattries and 87% in the transportation equipment industry. The
marginal productivity, unlike the average labor productivity
latter is the only industry with a rate of exploitation below
of Marxian economics, eludes easy quantification. Indeed,100%. It should be noted that Hildebrand and Liu also
some neoclassicists may have regarded the Pigouvian defini-found substantially higher rates of exploitation among non-
tion as desirable precisely because of these difficulties. production workers.

[ 52 ]

This content downloaded from 137.50.152.224 on Mon, 02 Mar 2020 15:27:31 UTC
All use subject to https://about.jstor.org/terms
 PIGOUVIAN EXPLOITATION OF LABOR 53

A Macro Model FIGURE 1.- THE "CONTRACT ZONE" IN THE

MACRO LABOR MARKET
We start our exploration at the broadest level
with a simplified model of the macro labor mar-
ket as illustrated in figure 1. Here, "full employ-
ment" (Ef) is defined by the intersection of the
aggregate supply curve of labor (S) and that
factor's marginal product curve (MPP(K*)).
At any point in time, the actual level of employ-
ment (E*) is determined by the aggregate de-
mand generated in the economy and the existing
capital stock (K*). It should be noted that we
assume that aggregate demand is independent
of the income distribution. Given these assump-
tions, the "effective demand" for labor is a dis-
continuous curve (ABE*), which falls to zero
at a point determined by aggregate demand.
Hence, there is a broad range of wage indeter-
minancy or a "contract zone" at this level of
employment. In Figure 1 this is illustrated by because they are poor and ignorant or because
the segment (BC). In this context, our problem they are women who do not expect to continue
becomes one of explaining the divergence (or in industry after marriage or for any other
as in the following empirical work, the ratio) reason - are unorganized there are grounds for
between marginal productivity and the real fearing that exploitation will often occur."7
wage rate, given K* and the level of aggregate Unfortunately, there are serious questions
demand. While this picture is an oversimplifica- concerning the formalization of this argument
tion of the macro-economy, it is helpful in re- at the macro-level. In particular our simple
casting Pigouvian exploitation in terms of "con- model does not allow us to consider separately
tract zones" and bargaining power as discussed the effects of unionization on the non-unionized
by labor economists.6 sectors of the economy. While the traditional
Perhaps the most widely mentioned influence neoclassical analysis of this problem suggests
on labor's bargaining power, and hence on ex- that union gains may lead to a reduction in non-
ploitation, is the extent of unionization in the union wages, this conclusion is based on the
labor force. As Pigou suggested: assumption that neither sector evinces exploita-
"Even when the gap (between marginal tion as defined here. A meaningful analysis of
products and the supply curve) is large the these relative wage and exploitation effects
occurrence of exploitation is not certain, and in would require a highly sophisticated model of
occupations where the workpeople have been labor market interactions. Such a model is
able to organize themselves into strong Trade clearly beyond the scope of this paper. More-
Unions, supported by reserve funds and bar- over, whatever this type of analysis would pre-
gaining for their wage rates as single collec- dict, it should be kept in mind that a strong
tive wholes, it is not even probable. But in labor movement sets the tone of labor-manage-
occupations in which the workpeople - whether ment relations and by its political activities has
because they are widely scattered in space, or important effects on wage settlements through-
out the economy. Thus, while we do not deny
6 It should be noted that the "ideal" measure of the rela-
the possibility of some negative externalities for
tive bargaining strength of labor and capital would be the
ratio of (BC/WC) rather than (BE*/WE*) which is used non-unionized labor, we expect the fraction of
in the empirical work which follows. However, such a the civilian labor force which is organized to be
measure would require a knowledge of the underlying supply
negatively related to the general rate of exploita-
curve of labor. Since this curve is itself somewhat fuzzy in
an exploitation situation and since in any event we are un- tion.
able to quantify it, we feel justified in using the measure
proposed here. 7 Pigou, (1952), p. 559.

This content downloaded from 137.50.152.224 on Mon, 02 Mar 2020 15:27:31 UTC
All use subject to https://about.jstor.org/terms
 54 THE REVIEW OF ECONOMICS AND STATISTICS

A second obvious hypothesis is that labor will 1942-1945 to explore the implications of such
bargain more aggressively and more success- governmental intervention in the economy. A
fully the lower the level of unemployment. priori, it is difficult to hypothesize the direction
While we would be reluctant to advance a of this effect. However, given the bias of govern-
theory of the reserved army of the unemployed,
mental agencies toward the status quo, we would
it is reasonable to assume that the probability guess that exploitation is higher under controls,
of obtaining alternative employment should in- than in an unregulated economy with the same
fluence both the wage demands of workers and employment and price conditions.
their willingness to enforce those demands
through strikes, slowdowns, etc. Clearly in a Empirical Results
tight labor market we would hypothesize that
While we recognize that all of these variables
workers have greater bargaining power.
should be viewed as part of a larger dynamic
A third potentially important determinant of
model of the economy, for the present explora-
exploitation is the rate of inflation. Whether it
tory work it seemed reasonable to use a simple
results from demand pull or cost push, inflation
one-equation format. Equation 1 is meant to
clearly reduces the magnitude of the wage
embody the hypotheses discussed above:
gains obtained by workers. All other factors be-
ing equal, we would expect that the ratio of (MPP/W)t -a+ bM+ cU + dP+ fK+gC.
marginal productivity to real wages is likely to (1)
increase both during and immediately after
Here, (MPP/W) is the ratio (in percentage
periods of high inflation. It should be noted
terms) of the niarginal productivity of labor to
that this argument does not imply that exploita- the level of real compensation of employees in
tion must increase during inflationary periods.
year t. The empirical measure of Pigouvian ex-
Since inflation is usually associated with rising
ploitation is drawn from Thurow's study of dis-
levels of employment, the rate of exploitation
equilibrium in the United States economy.
may well fall because of the effects of the latter
Thurow first estimated a Cobb-Douglas aggre-
variable.
gate production function for the private econ-
A fourth factor to be considered is the rate omy with the capital-labor ratio, time and the
of growth of the capital stock at any given time. unemployment rate as the independent vari-
Capital growth may arise from employers' ef- ables.8 Taking the derivative of real output with
forts to exploit technical progress or may be respect to man-hours of labor and multiplying
directed toward meeting increases in aggregate
by 2000, Thurow calculates the real marginal
demand. In either case, the marginal product of
product of an "average" worker for each year
labor will rise in the short run. While we would
1930-1965. Dividing this figure by the actual
expect labor's wages to lag behind rising mar-
employee compensation per 2000 man-hours (in
ginal products in any case, this lag is probably
the same constant dollars) gives the ratio of
most pronounced for increases due to technical
(MPP/W) .9
progress embodied in investment. Moreover,
when the purpose of investment is to substitute .8 The actual equation, estimated by Thurow takes the
capital for labor, the demands of workers will following form:
ln(Y/K) = .6048 -.0269U2 + .01167t + .83041n(L/K).
often be tempered by the threat of automation.
Here Y is the private GNP in constant 1958 dollars, L is
Thus, while high levels of investment are often labor man-hours with embodied technical progress of 1%
associated with periods of high labor demand per year, K is the capital stock with embodied technical
progress of 4% per year in investment (constant 1958 dol-
(as are high rates of inflation) once we have
lars), U is the unemployment rate and t is time. The capital
controlled for the unemployment rate, we hy- stock data are taken from the Survey of Current Business,
pothesize that the rate of capital growth will beFebruary, 1967 and are based on the gross stocks, constant
cost 2 series. The labor stock data are from the 1967 Eco-
positively related to exploitation.
nomic Report of the President and are based on tle labor
Finally, we are interested in the effect of wage force series. The choice of the rates of embodied technical
and price controls on the level of exploitation. progress was based on the overall fit of the production func-
tion. See Thurow (1968) for further details.
Since our sample data include the World War 9 The compensation data used by Thurow were taken
II years, we introduced a dummy variable for from the Survey of Current Business. Our dependent vari-

This content downloaded from 137.50.152.224 on Mon, 02 Mar 2020 15:27:31 UTC
All use subject to https://about.jstor.org/terms
 PIGOUVIAN EXPLOITATION OF LABOR 55

The independent variables are: M, the per- includes the unemployment rate as an explana-
centage of the civilian labor force who were tory variable. Thurow finds that low unemploy-
union members in year t - 1; U, the rate of ment is associated with higher productivity.
unemployment in the civilian labor force in While wages may rise in such periods, they may
year t; P, the average annual rate of inflation not rise much faster than productivity itself.
between year t - 2 and year t; K, the percent- Hence the relatively small influence of unem-
age increase in the capital stock between year ployment on exploitation may be the result of
t - 1 and year t and C, a dummy variable for these offsetting effects of aggregate demand.
the controlled years, 1942-1945.1o Alternatively, it may be that only in the im-
The expected signs of the coefficients, based mediate proximity of full employment is the
on the theoretical arguments presented above, unemployment rate an important factor in bar-
are that unionization (M) should reduce the gaining. Since many of the years with truly low
rate of exploitation, while unemployment (U), unemployment included in the sample occurred
inflation (P), capital expansion (K), and gov- during periods of wage and price controls, it is
ernment controls (C) should all tend to increase difficult to test this hypothesis. Clearly further
the level of exploitation. work on this question would be desirable.
Equation 1 was estimated for the period As hypothesized, inflation seems to increase
1930-1965, with the following results: the rate of exploitation. The price change vari-
able suggests that, ceteris paribus, workers re-
(MPP/W)t 176.0 - 0.89M + 0.46U
capture only half of their reduced purchasing
(0.15) (0.20)
+ 0.52P + 0.40K + 9.20C R2 = 0.86 power after an inflationary period. Here, it
(0.17) (0.17) (3.00) D.W.= 1.23 would be interesting to explore the lag struc-
ture associated with the inflation variable to
The figures in parentheses are the standard
determine whether such losses are subsequently
errors of the coefficients. All the coefficients are
made up. The use of a two-year average in the
significant in the expected directions.
inflation rate was justified on these grounds,
One of the most interesting findings is that but this lag actually increased the importance of
an increase in the extent -of unionization of 1the% inflation variable. A more extended lag
is associated with a decrease of exploitation of structure, however, might reduce this coefficient.
about 0.9 %. This would suggest that the level The expansion of the capital stock also tends
of 1965 exploitation (159%) could be substan-
to increase exploitation. It should be noted that
tially reduced by further organization of labor.
if there are in fact lags. in recouping marginal
On the other hand, the unemployment rate productivity increases, they must apply largely
and hence the level of aggregate demand doesn't
to those embodied in capital. This conclusion
seem to have an impressive effect on the rate of
is based on estimated equations which include
exploitation. To some extent this may be the
the rate of increase in marginal productivity as
result of Thurow's production function which
an independent variable. These equations pro-
duce no evidence of a significant overall relation
able is actually the inverse of the measurement reported by
Thurow, but this hardly affects the equations. See Thurow between the growth rate of marginal productiv-
(1968) p. 25 for the original data. ity and the rate of exploitation.
10 The total United States union membership before 1957
Finally, it is interesting that during World
is taken from the Historical Statistics of the U.S. (1960), p.
98 and is based on the estimates of the Bureau of Labor War II, the rate of exploitation was substan-
Statistics. Values since 1957 are taken directly from the tially higher than would be expected given the
Bureau of Labor Statistics (1970), p. 67. This figure is
level of the independent variables during the
divided by the civilian (not total) labor force as reported
in the Economic Report of the President (1972), p. 247. period. We are tempted to ascribe this fact to
Finally the capital stock figures are based on the Survey of the wage policies of the controlled economy.
Current Business, (February, 1967). Following Thurow we
used the gross stocks constant cost 2 series as the base for
This hunch is supported by estimates of equa-
calculations which incorporate a 4% rate of embodied tech- tion (1) which excluded the war years from the
nical progress in capital. We assumed a lifetime of capital of sample. These estimates produced coefficients
fifty years. See Thurow (1968) for the general approach. It
should be noted that the Survey's net stock series gave much
strikingly similar to those shown for the
the same results. complete sample period. Moreover, when the

This content downloaded from 137.50.152.224 on Mon, 02 Mar 2020 15:27:31 UTC
All use subject to https://about.jstor.org/terms
56 THE REVIEW OF ECONOMICS AND STATISTICS

equation was estimated without the dummy bilateral monopoly suggests that the presence
variables, but with the World War II years, of trade unions should tend to offset this type
only the level of unionization and the rate of of monopsonistic power. Following Pigou again,
inflation were still significant. Thus, the ab- we would hypothesize that industries with
sence of the control variable would confuse the strong unions would generally show lower rates
interpretation of the unemployment variable. It of exploitation than those with a less organized
should be noted however, that controls in this labor force.
period undoubtedly did keep down the rate of To test these two predictions in a cross-sec-
price increases. Since inflation leads to increased tional setting we used Hildebrand and Liu's
exploitation the net effect of the controls is in 1957 estimates of (MPP/W) for production
some doubt. Still, even if prices had risen at a workers in ten two-digit manufacturing indus-
tries."2 Obviously the extremely small sample
rate of 25% a year, the equation would predict
lower levels of exploitation than those actually involved makes any literal reading of our results
experienced under controls. extremely dangerous. Thus, we emphasize the
tentative nature of what follows.
Monopolistic and Monopsonistic Equation (2) is an attempt to include the im-
Exploitation portant effects of both monopolistic and monop-
Up to this point we have viewed exploitation sonistic exploitation:
only as a macro-economic phenomenon. How-
(MPP/W)i -- a + bM + cR. (2)
ever, the neoclassical approach to exploitation
has most often concentrated on the micro-level. Here M is the percentage of production work-
As suggested above the theoretical literature in ers in industry i who were unionized in 1954 as
this area is quite rich. What follows is an ad- estimated by Lewis.'3 Several measures of con-
mittedly crude attempt to illustrate the possibil- gcentration were used for the variable R." How-
ities of empirically testing the micro-theory of ever, none of these entered the equation with a
exploitation. In particular, we concentrate on significant coefficient. This negative result may
the two most widely discussed aspects of this be due to the fact that in an oligopoly situation
theory: monopolistic exploitation and monop- a system of market shares or "kinked demand"
sonistic exploitation. curves seriously complicates the interpretation
Monopolistic exploitation originates in the of marginal revenue product. In particular, a
incentive for monopolies to maximize profit by "kink" will create a range of wage indetermi-
equating the wage rate to marginal revenue nacy in the derived labor demand curve of an
product rather than to marginal product. For individual firm. Given this implication of oligo-
empirical purposes it seems plausible to assume poly theory it is not obvious that the degree of
that the more concentrated an industry (and the concentration will necessarily be related to the
more conscious firms are of their influence on extent of exploitation. A related explanation of
price) the greater will be the divergence be- our result is that we have failed to take account
tween actual wages and marginal productivity. of the underlying elasticity of each industry's
Hence, we expect that industries with higher product demand curve. It these curves are
concentration ratios should show higher levels
of Pigouvian exploitation., 12 These ten industries were food and kindred, apparel,
paper, chemicals, petroleum, stone, clay and glass, fabricated
Monopsonistic exploitation, on the other metals, machinery excluding electrical, electrical machinery
hand, relates to a firm's power to control mar- and transportation equipment. For other two-digit indus-
ginal resource costs. Labor economists have tries, Hildebrand and Liu do not distinguish between pro-
duction and nonproduction workers. The original data can
increasingly treated corporations and other
be ad-
found in Hildebrand and Liu, (1965), p. 124.
ministrative units as internal labor markets in 13 H. G. Lewis, (1963), p. 274.
14 Two concentration measures were tried: average four-
which employees have little mobility and hence
firm product concentration ratio developed by Leonard
employers have substantial monopsonistic Weiss for 1954 and average four-firm concentration ratio
powers."1 In such a situation the theory of for 1958 as developed by the Subcommittee on Antitrust
and Monopoly, Senate Committee on the judiciary. Both
11 See Peter Doeringer and Michael Piore (1971) for a full sets of data were drawn from Norman R. Collins and Lee
presentation of internal labor market theory. E. Preston (1970) p. 30 and p. 53.

This content downloaded from 137.50.152.224 on Mon, 02 Mar 2020 15:27:31 UTC
All use subject to https://about.jstor.org/terms
 PIGOUVIAN EXPLOITATION OF LABOR 57

highly elastic, even pure monopolies would show REFERENCES

little tendency toward exploitation. Cartter, A. M., Theory of Wages and Employment,
The effect of unionization on exploitation (Homewood, Ill.: Richard D. Irwin, 1959).
across industries does seem to be important. Clark, J. B., The Distribution of Wealth, (New York:
Macmillan Company, 1899).
This result is suggested in the simple linear re-
Collins, N. R., and L. E. Preston, Concentration and
gression presented below:
Price-Cost Margins in Manufacturing Industries,
(MPP/W), 232. - 2.10M. R2 0.58 (Berkeley: University of California Press, 1970).
(0.64) Directory of National and International Labor Unions
in 'the U.S., 1969, Bureau of Labor Statistics, U.S.
Despite the low overall explanatory power of Department of Labor (Washington: Government
the equation, it is encouraging that unionization Printing Office, 1970).
at the industry level provides a micro basis for Doeringer, P. B., and M. J. Piori, Internal Labor Mar-
interpreting the aggregate impact found at the kets and Manpower Analysis, (Lexington: D. C.
Heath, 1971).
macro level.
Economic Report of the President 1972, (Washington:
Government Printing Office, 1972).
Conclusions
Hildebrand, G. H., and T. C. Liu, Manufacturing Pro-
In summary, our research suggests that the duction Functions in the United States, 1957,
level of exploitation in the economy is signifi- (Ithaca: Cornell University Press, 1965).
cantly affected by union power both over time Historical Statistics of the U.S., U.S. Bureau of Census,
(Washington: Government Printing Office, 1960).
and cross-sectionally. Moreover, unemploy-
Lewis, H. G., Unionism and Relative Wages in the
ment, inflation, growth of the capital stock and
United States (Chicago: University of Chicago
government controls tend to increase exploita- Press, 1963).
tion in the macro labor market. We have hardly Pigou, A. C., The Economics of Welfare, fourth edition
offered a full explanation of Pigouvian exploita- (London: Macmillan Company, 1952).
tion in the American economy. However, it is Schumpeter, J. A., History of Economic Analysis, (New
York: Oxford University Press, 1954).
interesting that many of the same factors em-
Survey of Current Business, vol. 47, no. 2 (Feb. 1967),
phasized in the descriptive work of bargaining
p. 24.
theory (or for that matter, in the general Marx- Thurow, L., "Disequilibrium and the Marginal Produc-
ist literature) should prove useful in the anal- tivity of Capital and Labor," this REVIEW, XLX,
ysis of this neoclassical concept. no. 1 (Feb. 1968), 23-31.

This content downloaded from 137.50.152.224 on Mon, 02 Mar 2020 15:27:31 UTC
All use subject to https://about.jstor.org/terms