Download as pdf or txt
Download as pdf or txt
You are on page 1of 8

The "Wage-Productivity" Theory of

Underemployment: A Refinement'
The concept of disguised unemployment or underemployment has drawn considerable
attention in the economic development literature for almost a decade. The efforts to

Downloaded from http://restud.oxfordjournals.org/ at University of Western Australia on August 26, 2015


reconcile a zero or " low" marginal productivity of labor with a higher wage rate have
followed two directions. First, the approach usually associated with W. Arthur Lewis'
work bases the phenomenon of underemployment on a postulated dualism that exists
between the industrial and the agricultural or subsistence sector of an underdeveloped
economy.' Second, the theoretical approach first suggested by Harvey Leibenstein pos-
tulates a certain relationship between the wage rate and productivity and complements this
relationship with certain assumptions on institutions."
The purpose of this article is to refine some aspects of the "wage-productivity"
theory of underemployment. We will argue that the analysis as it stands supplies some
of the necessary but not all the sufficient conditions for the theory to be correct. Accord-
ingly, this approach to underemployment may be confirmed or rejected by theoretical or
empirical evidence that has yet to come. In the last section of this paper we will argue
that even if the wage-productivity relationship be confirmed by future research, still the
conclusion to which the theory leads is not, thereby, logically necessary.

Leibenstein's analysis rests upon two postulates." The first is " the often-neglected
relationship between the wage level or income level and productivity". Income is related
to productivity through the nutritive (chiefly caloric) value of food intake. A higher
wage level enables the worker to increase the caloric content of his diet and this in turn
will cause an increase to the amount of effort (or " units of work ") that he supplies.

I am grateful to Professors Robert E. Baldwin, Peter O. Steiner and Mr. Van Mablekos for discuss-
ing earlier drafts of this paper with me,
W. Arthur Lewis, "Economic Development with Unlimited Supplies of Labour ", The Manchester
School of Economic and Social Studies, Vol. XXII, No.2 (May 1954), pp, 139-191; the same, " Unlimited
Labour: Further Notes ", ibid, Vol. XXVI, No.1 (January 1958), pp. 1-32; William J. Barber, "Disguised
Unemployment in Underdeveloped Economies ", Oxford Economic Papers, Vol. 13, No.1 (February 1961),
pp. 103-115; Gustav Ranis and John C. H. Fei, "A Theory of Economic Development ", American
Economic Review, Vol. LI, No.4 (September 1961), pp. 533-565. Cf. also Pan A Yotopoulos, "The Elusive
Test of Disguised Unemployment: .John Lossing Buck's Data .., The Indian Journal of Economics, Vol.
XLII, No. 164 (July 1961), pp. 27-35.
2 Harvey Leibenstein, "The Theory of Underemployment in Backward Economies", Journal of
Political Economy, Vol. LXV, No.2 (April 1957), pp. 91-103; the same, "Underemployment in Backward
Economies, Some Additional Notes ", ibid, Vol. LXVI, No.3 (June 1958), pp. 256-258. Cf. also, the same,
Economic Backwardness and Economic Growth (New York: John Wiley and Sons, 1957), Chapter 6;
Harry T. Oshima, "Underemployment in Backward Economies, An Empirical Comment ", Journal of
Political Economy, Vol. LXVI, No.3 (June 1958), pp. 259-264; Harvey Leibenstein, " Reply", ibid, p. 264;
Hannan Ezekiel, "An Application of Leibenstein's Theory of Underemployment ", Journal of Political
Economy, Vol. LXVIII, No.5 (October 1960),pp. 511-517; Dipak Mazumdar," The Marginal Productivity
Theory of Wages and Disguised Unemployment ", The Review of Economic Studies, Vol. XXVI (3), No. 71
(June 1959), pp. 190-197.
3 This section draws mainly upon pp, 62-76 of Economic Backwardness and Economic Growth which
present a much improved version of pp. 94-103 of the original article. All references hereinafter are to the
book.
S9
60 REVIEW OF ECONOMIC STUDIES

This first postulate is employed in order to derive the" optimum employment revenue
curve OR" (figure 6-6). Let us start from a "low wage level equilibrium position" in
the sense that this wage is equal to the marginal product of labor and clears the market.
This position is associated with a certain (" low") net revenue for the landlords (net
revenue OL in figure 6-6). By invoking the wage-productivity postulate we come to the
conclusion that the landlords may improve their net revenue position. By paying a higher
wage rate they engender an upward shift in the average and marginal productivity curve
of labor due to the increased amount of " units of work" forthcoming after the improve-
ment in the workers' diet. They now reduce employment to the point where the marginal
product of labor equals the higher wage rate. Their net revenue position improves so long
as the marginal receipts that result from selling the higher output at a given price offset

Downloaded from http://restud.oxfordjournals.org/ at University of Western Australia on August 26, 2015


the marginal outlays that result from the increased wage bill. Obviously, beyond a certain
point (point N in figure 6-6) this is no more the case and the landlords by paying an even
higher wage rate would only worsen their net revenue position.
To this point, Leibenstein's analysis has not touched yet upon the problem of disguised
unemployment. The higher wage rate is still equal to the marginal product of labor which
is increased because of the increase in the" units of work" and the decrease in the quantity
oflabor employed. In order to explain the possibility of a wage rate higher than the marginal
product of labor Leibenstein introduces his second postulate. Assume that the higher
wage rate cannot stick with less than full employment because " competition among the
visibly unemployed depresses wage rates toward the zero level ".1 Under such condition
the wage-productivity relationship will fail to operate. Lest this happen, the landlords
are prepared to pay the higher wage rate to the full employment work force and thus
pay some workers a wage higher than their marginal product. In this way the" full em-
ployment revenue curve FR" is obtained (figure 6-6). It is obvious that at the "low
wage level equilibrium" which we started with (point L in figure 6-6) the FR and OR
curves coincide since both were derived by landlords offering full employment. For
higher wage rates the FR curve will rise but always inside the OR curve; for in the deriv-
ation of the OR curve the marginal increase in the wage bill was less because the higher
wage rate was attended by a decrease of the quantity of labor employed. Again, after the
FR curve reaches a maximum (M in figure 6-6) it will drop back-i.e., landlords would
only worsen their net revenue position by offering full employment at an even higher wage
rate. This occurs because the marginal outlay of paying an increased wage bill offsets the
marginal receipts of selling the higher output at a given price.
Here is where the weak point in Leibenstein's theory lies. Will the net revenue which
is obtained by offering full employment and a higher than the initial" low equilibrium
wage" be greater than the net revenue obtained by offering full employment at just the
" low equilibrium wage"? Or, in terms of Leibenstein's graphs, does FR reach a maximum
above L and wage rate W 2 ? (figure 6-6). If this does not occur then the choice of the land-
lords is again restricted to two alternatives: either pay the optimum wage W 4 and hire a
less than full employment quantity oflabor that will net you revenue ON; or, if you cannot
prevent competition that would drop the wage below w4 , employ the whole supply of labor
at the "low equilibrium wage" W 2 and be satisfied with net revenue OL. In this event,
in other words, Leibenstein's analysis breaks down and we are back where we started from:
in order to maximize profit pay a wage rate equal to the marginal product of labor.
Leibenstein only in his later work recognizes this possibility but seems to gloss over it:
It should be clear that the full employment revenue curve need not be at a maxi-
mum above the wage W2 where the labor supply and the optimum employment are
equal. Although it is true that at wages above W2 the full employment revenue must
always be less than the optimum employment revenue . . . it may nevertheless be
1 Harvey Leibenstein, op. cit., p. 70.
THE" WAGE-PRODUCTIVITY'" THEORY OF UNDEREMPLOYMENT 61

that the full employment revenue is higher than what it was at a lower wage. The
reason for this possibility is (that) ... above the wage w2 , the work units per man
increase by a greater proportion than the increase in wages, and this effect may,
up to some point, be more significant than the depressing effect of the greater wages
bill on net revenue.'

II

Downloaded from http://restud.oxfordjournals.org/ at University of Western Australia on August 26, 2015


We will proceed now to examine the conditions that are necessary and sufficient for
the full employment revenue to be increasing beyond the initial " low equilibrium wage
rate" and thus for curve FR to have the shape that would satisfy Leibenstein's theory.
The relationship between the full employment revenue and the optimum employment
revenue curves will become more clear if we use a simplified production function to
which we add the relationship between productivity and wages that Leibenstein pos-
tulated. This will be shown by using algebra and it will be illustrated by employing a
three-dimensional production mountain.
Assume a simple production relationship that expresses productivity (P) as a function
of employment (X). Within the range meaningful for economic theory we know that
output is an increasing at a decreasing rate function of employment.f In symbols,
P =j(X)

dP d 2P
where dX> 0 (1); and dX2 < 0 (2).
Now, Leibenstein's contribution consists in adding a new relationship and making
productivity also an indirect function of the wage level (W), as follows." Productivity
depends upon the amount of effort or "units of work" (E) that the worker will con-
tribute.! which in turn is related to his caloric intake (C);5 the relationship between income
(= wages) and nutrition then closes the circuit,"

The relationship between the wage level and productivity displays the following
properties: Up to a certain point, productivity is an increasing 7 at an increasing rate 8
function of caloric intake. This is the range meaningful for Leibenstein's analysis. Beyond
this range, however, productivity may be a decreasing, or perhaps still an increasing but
at a decreasing rate function of caloric intake."

1 Ibid, p. 74; cf. also Dipak Mazumdar, op. cit., p. 195.


2 R. G. D. Allen, Mathematical Analysis for Economists (London: Macmillan and Company, Ltd.,
1953), p. 288.
3 Op. cit., p. 62.
4 Idem
5 Ibid, p. 64.
6 Ibid, p. 63.
7 " • • • up to some point the effective work units are increased as wages are increased ", ibid, p. 66;
cf. also ibid, pp. 63, 64.
8 " • • • if at the outset an employer pays a wage that is sufficient only to enable his workers to obtain
a diet of no more than 2,140 calories, the employer can approximately double his effective work force by
paying a wage that would enable the workers to purchase a diet yielding 2,460 calories", ibid, p. 64.
II Ibid, p, 73.
62 REVIEW OF ECONOMIC STUDIES

In symbols,
P = peE) = e(C) = w(W).
By making the simplifying assumption that productivity is a single function of the wage
rate, we can express the marginal output with respect to wages in the range meaningful
for Leibenstein's theory as follows:
dP d 2P
dW> 0 (3); and dW2> 0 (4).

By combining the simplifiedproduction function with Leibenstein's postulated relation-

Downloaded from http://restud.oxfordjournals.org/ at University of Western Australia on August 26, 2015


ship, we can express productivity as a function of two variables, employment and wage rate.
We can now formulate the necessary and sufficient conditions for Leibenstein's theory to
hold true.

P & TW

FIGURE 1
THE" WAGE-PRODUCTIVITY" THEORY OF UNDEREMPLOYMENT 63

Brief analysis shows that the FR curve will reach a maximum above L if, but only if,

x' dw < J:' [Fw'+dW - F..] dx

where x* is full employment, w* is the initial" low equilibrium employment wage rate"
and the F's the marginal product curves corresponding to different w's. Thus, the left
side of the inequality represents marginal outlay resulting from payment of a higher than
the initial "low equilibrium wage rate" to the full employment labor force, while the
right side represents marginal receipts resulting from the increased productivity of the
full employment labor force because a higher than the initial "low equilibrium wage
rate" is paid. Only if the net addition to the marginal product (evaluated at a certain

Downloaded from http://restud.oxfordjournals.org/ at University of Western Australia on August 26, 2015


price) is greater than the addition to the wage bill-both evaluated for x*-will the maxi-
mum of curve FR be as Leibenstein suggests.' Thus, it seems that far more important for
his theory than relationships (3) and (4), which he diligently analyzes, is the relationship

dr Fwdx
between 0 and dw, which he completely overlooks. It would be nice if one had
dw
some compelling theoretical or empirical evidence on this relationship, but such evidence
is hard to come by.
The presentation will be more elegant if we use a three-dimensional diagram. In
figure I, total output (P) and total wage bill (TW) is plotted as a function of two variables:
employment (X), from a certain (small) level x to full employment x* is shown along the
one axis; while wage rates (W), from the initial " low equilibrium wage level" w* to wage
W 4 is shown along the other axis. The total wage plane OARe that cuts the productivity
mountain describes the combinations of different employment and wage levels that cor-
respond to it.
For a given wage rate from w* to w4 , the productivity plane PX describes contours
that increase first at an increasing and then at a decreasing rate, according to properties
(1) and (2) above. For a given level of employment from x to x*, the productivity plane
PW describes contours that increase at an increasing rate (and only eventually at a de-
creasing rate), consistent with the properties (3) and (4) above.
Leibenstein's main concern is with varying wage rates. By slicing figure 1 along the
Waxis at different wage levels we get a profile of total product and total wage curves as
in figure 2. The slope of these curves can be explained on the basis of Leibenstein's postu-
lated relationship between the wage rate and productivity.'
From figures 1 and 2 we can derive the different net revenue possibilities arising for
the landlords at alternative contours of the total product mountain. For each quantity
of labor employed and for each wage rate, the net revenue is shown by the distance between

1 This can be illustrated in terms of Leibenstein's figure 6-4. The increase in the wage bill x*dw is
equal to the rectangle formed by e2 w 2 w a projected to SS. The marginal receipts are equal to the difference
between the area bound by the MP a and that bound by the MP 2 curve.
2 In figure 2, the higher the wage rate, the steeper the total product curve. This property follows
from the higher average productivity of the work force that Leibenstein postulated after an increase in
the amount of" units of work" per man with a higher wage rate (op. cit., p. 67). An increase in the average
product means that the ratio of the j-intercept to the x-intercept of the total product curve increases;
and for the same quantity of labor it means that the total product curve shifts upward. The second charac-
teristic of figure 2 is that the higher the wage rate the higher and to the left is the maximum point of the
total product curve. This is a result of Leibenstein's assumption that the marginal productivity curve for
a higher wage rate starts at a higher level than the one for a lower wage and below a point the former falls
more rapidly than the latter (ibid, p. 68). A higher maximum of the marginal product curve means that
the inflection point of the total product curve moves to the left for higher wages. The need for the marginal
product curves to intersect implies that the top of the total product curve moves also to the left for higher
wages.
64 REVIEW OF ECONOMIC STUDIES

P & TW

S'

Downloaded from http://restud.oxfordjournals.org/ at University of Western Australia on August 26, 2015


o e :=x*
2
x
FIGURE 2

the total wage bill and the total product. There is a different level of employment that
maximizes net revenue under the total product curve that corresponds to each wage rate.
Assume that the wage rate we observe in the market is the initial low equilibrium
H

wage level" w*. The total wage bill corresponding to different levels of employment
can be read off figure 2 (TW2 ) or along the TW, X plain in figure 1 (OA). By definition
of the initial" low equilibrium wage", the graphs have been drawn so that the full employ-
ment quantity of labor x* yields the maximum net revenue fg. For slices of the total
product mountain corresponding to higher than the equilibrium wage rate (e.g. w3 , w4 ) ,
the net revenue is necessarily maximized (cd, ab) at lower levels of employment (Oe3' Oe4)'
Leibenstein's optimum employment revenue curve (OR in figure 6-6) is the locus of all
such possible maxima. It is obvious that under the circumstances landlords would prefer
the solution under curve TP 4 with net revenue ab and quantity of labor employed Oe4 •
Leibenstein's argument is that the landlords can move on the PW plane only along
the slice corresponding to full employment x*. According to his second postulate the land-
lords cannot take advantage of the higher productivity engendered by an improved wage
rate, unless they first rule out a competitive decrease in wages by employing the whole
labor force. The full employment revenue can then be read as fg, jk, lm off figures 1 and 2.
Curve FR (in figure 6-6) describes the locus of all such full employment revenues.
THE" WAGE-PRODUCTIVITY" THEORY OF UNDEREMPLOYMENT 65

By the construction of figures 1 and 2, jk is greater thanfg and so FR curve reaches


a maximum above point L (figure 6-6). There is no guarantee, however, that this fortuity
will indeed occur. In our model it does so only because we postulated that the wage
plane OABC is sufficient to educe the energy that will prompt the productivity surface we
have drawn. The relationship between changes in wage rates and changes in productivity
still remains obscure.
The wage-productivity relationship postulates that productivity increases with an
increase in wage rates. The important question, however, remains unanswered: which
is the particular wage increase that will induce a certain increase in productivity? Consider,
for example, the case that an alternative wage plane OAB'C' in figure 1 (line TW'a in

Downloaded from http://restud.oxfordjournals.org/ at University of Western Australia on August 26, 2015


figure 2) is necessary to reach the productivity surface of our graphs. In this event j'k is
less thanfg and curve FR reaches its maximum at the intial " low equilibrium wage rate"
and then it bends back. If this occurs, the full employment revenue curve becomes com-
pletely meaningless. We are back to the established conclusion of economic theory: in
order to maximize profit pay a wage rate equal to the marginal product of labor, whatever
the marginal productivity curve may be. And of course, one maximizes profit by picking
the marginal productivity curve that would do so-TP4 in our case!

III

Professor Leibenstein has based his analysis of underemployment upon a postulated


relationship between wage rates and productivity. This relationship has been defined only
in terms of the first and second derivative of product with respect to a change in wages.
What he dwells upon is the shape of the PW surface. Our analysis suggests that the con-
ditions he supplies are necessary but not sufficient to establish the " wage-productivity"
theory of disguised unemployment. What is far more important is the quantification of
the change in product that follows every particular change in wages-evaluated at different
employment levels between zero and full employment.' It is conceivable that a certain
relationship between wages and productivity exists. At the present state of theoretical
and empirical evidence, however, the theory suffers from the indeterminateness of the
wage level connected with each particular increase in productivity.
Now let us assume that Professor Leibenstein's analysis is destined to be vindicated
by empirical evidence that awaits marshalling. Still his conclusions are not necessarily
justified. He concludes, namely, given the peculiarities of the underdeveloped world, it
would be consistent with profit maximizing behavior if landlords went out of their way to
provide full employment at higher than the initial "low equilibrium wage level". The
reason for this is presented as the postulate that competition of the unemployed will bring
the wage down, so decreasing the productivity of the labor force and possibly the net
revenue of the employers.
Without adequate data we cannot determine whether landlords do in fact provide
full employment or not. But assume that they do and that the underdeveloped world is
indeed as Leibenstein portrays it. Still one must explain why landlords do not go straight
to the point of maximum revenue, rather than stopping at a kind of a second-best position
on the full employment revenue curve. Certainly, it is not because landlords could not
possibly eliminate competition in the labor market that might drive the wages to a level
lower than the wage-productivity postulate requires. If this were to their advantage, they
X.

1 i.e. the relationship between


d
J
0
dw
Fedx
and dw in terms of our notation.
66 REVIEW OF ECONOMIC STUDIES

could absolutely refuse to pay any worker less than the minimum wage necessary for
maximum efficiency-i.e., they could operate on curve TP4 no matter what the full employ-
ment wage rate would have been.
The real issue, however, is not how to institutionalize a wage rate higher than the
initial" low equilibrium level". In our opinion, it rather is how to circumvent the income-
sharing characteristics of the extended family system.' For example, a custom which is
widespread in underdeveloped countries now calls for the employer to supply one or two
meals during the long work day of the peak season agricultural activities. 2 Such an in-
stitution may purport to circumvent the family allegiance that might induce the employed
to share their income with the unemployed and so enable the employer to take full advantage

Downloaded from http://restud.oxfordjournals.org/ at University of Western Australia on August 26, 2015


of the increased productivity connected with an improved nutritive intake.
University of Wisconsin, PAN A YOTOPOULOS.
Milwaukee.

1 This Leibenstein acknowledges in the subsequent discussion on his article. cr." Some Additional
Notes ", op, cit., pp. 257-258..
2 cr. Harry T. Oshima, op, cit., p, 262.

You might also like