NOTES ON THE SOCIAL OPPORTUNITY COST OF LABOR

(SHADOW WAGE RATE)

On compensating differentials

Labor is a location-specific input. The SWR may vary across regions and provinces, rural vs.
urban. Labor is also unique from other inputs. As I sell my labor services, it matters to me who is
buying it, for I have to interact with the buyer. With material inputs, transactions are more
detached, especially in an exchange where I don’t even know who is buying what I am selling..

With labor markets, therefore, both physical and psychic considerations matter and lead to so-
called compensating differentials. For example, in rural areas, the reservation wage or voluntary
supply price of a worker to his own family’s farm may be lower than his supply price to his
neighbor’s farm as a hired worker, because he may have some psychic disutility being ordered
around by somebody other than a family member. His supply price to working in a rural public
works project may be even higher, as there may be an additional physical disutility to working
more strenuously in the public works project. (Studies have shown that you burn more calories
with the latter than with regular farm work.)

Another example: a friend earns P5,000/day in his job in Makati. He joins our project at
P2,000/day. What is the cost to the economy of his joining our project? As he leaves his Makati
job, national output drops by P5,000/day, which is his marginal product in Makati (or his
employers would not have paid him that much). Some economists argue that this must then be
his SWR. But then, given the fact that he voluntarily gives up that job and accepts our job, we
may infer that he gains at least P5,000 – P2,000 = P3,000/day in psychic income working with
us. His gain of P3,000/day should be deducted from the economy’s loss of P5,000/day, leaving
SWR = P2,000/day, the price that our project pays him. In short, the voluntary supply price or
reservation wage to our project is our estimate of SWR.

We should keep this in mind in our subsequent discussions of rural versus urban, or agricultural
versus industrial wages. Differences in wages and earnings across areas or occupations or
sectors may reflect compensating differentials, some of them psychic, some of them physical.

Neoclassical:
W, MPL

Ws

MPL
Labor
L* = Le

In the neoclassical paradigm where there are no distortions, the market wage reflects the value
of the marginal product of labor (or simply MPL). When we hire a worker in our project, output in
the rest of the economy drops by what that worker was contributing to the economy, which is
how much he was getting elsewhere. Therefore, SWR = MPL = Ws = market wage.
In the graph above, the supply of labor is perfectly elastic at Ws (the “subsistence” wage). The
demand for labor is given by MPL. Labor market equilibrium is at the intersection of demand
(MPL) and the perfectly elastic supply (Ws). Equilibrium employment is at Le.

When our project hires a worker, that worker will come from agriculture, where he is contributing
MPL = market wage. Therefore, SWR = MPL = Ws = market wage.

W. Arthur Lewis (Economic Development with Unlimited Supplies of Labor, 1954):

W, MPL

Ws

SWR MPL APL

In agriculture, because of cultural and social norms, a large and expanding population, and
typically small farms, the farmer-operator employs workers (own children, relatives, even hired
labor) beyond the optimum point L* where Ws = MPL. He absorbs workers up to Le, where Ws
= APL. (He cannot absorb workers beyond this point, for then, the average product per worker
will fall below the subsistence wage and workers will be undernourished.)

As before, when our project hires a worker, that worker will come from agriculture, where he is
contributing MPL = market wage. Therefore, SWR = MPL < APL = Ws = market wage.

T.W. Schultz (Transforming Traditional Agriculture, 1964)

Schultz argued that farmers could not be as “irrational” as Lewis viewed them. The empirical
evidence from many developing countries (China, India, Egypt, the Philippines) pointed to profit-
maximizing behavior by farmers, i.e., farmers employ workers up to Ws = MPL. Therefore, in
(rural) agriculture, SWR = MPL = Ws = market wage.

Schultz believes that the urban labor market is distorted by factors like minimum wage laws. But
when our project creates a job in the city or in industry paying the high minimum wage, the job
created is ultimately filled in by a migrant from the rural area. Output in the rural area thus drops
by the MPL in agriculture. SWR for industry = MPL in agriculture = marker wage in agriculture,
even if our project pays a higher financial wage.

Harberger and Todaro (independently of each other)

Both Harberger and Todaro (H&T) follow Schultz in considering that there are no distortions in
rural labor markets; i.e., market wage = marginal product. However, they ask: Why do rural
workers continue to migrate to urban areas despite high open urban unemployment rates? Why
do rural-urban wage differentials despite rural-urban migration?
They model the decision-making process of the prospective migrant as follows: migrate, if your
expected wage in the city We is higher than your wage in the barrio Wr. But We is the urban
wage Wu weighted by the probability p of getting the job there: We = pWu + (1 – p)0 (assuming
that you have zero earning if you are unemployed, and the whole urban labor market is “formal”
or “protected” or “organized”). H&T believe, like Schultz, that urban labor markets are distorted
by higher-than-market-clearing minimum wages. In other words, Wu is fixed, and is higher than
Wr, which behaves like a constant because of the large (Lewis’s unlimited) supply of labor.

The next problem is how to model p. At its simplest, p = Lu/(Lu + U), where Lu = employment in
the protected sector and U = the (voluntarily) unemployed. Lu is exogenous; it is not affected by
the number of migrants. Lu + U is the total urban labor force. Urban labor market equilibrium is
reached when We = pWu = Wr, or when p = Wr/Wu.

Consider now our project creating G jobs in the city, paying Wu per worker. The initial effect is
for p’ = (Lu + G)/(Lu + U) > p. the new expected wage is We’ > Wr. This triggers rural-to-urban
migration, until p is back to Wr/Wu. This means that the urban labor force has to increase by
more than G; the number of migrants coming into the city will have to be higher than the number
of jobs that our project created. In fact, the number of migrants = G(Wu/Wr).

The urban SWR, as in Schultz, will equal output forgone in the rural area because of the job we
created in the city. But, with Schultz, only one worker leaves the barrio for a new city job, so that
his urban SWR = rural MPL = Wr. With H&T, urban SWR = rural MPL x migrants per job created
in the city. With equilibrium p = Wr/Wu, H&T’s urban SWR = Wr x (Wu/Wr) = Wu!!! Meanwhile,
H&T’s rural SWR = Wr, as with Schultz.

Implications on development paradigms: for Lewis, SWR < Wr or Ws meant that labor to be
used for industrialization efforts hardly costs the economy anything, for moving labor out of
agriculture into industry does not cause any drop in agricultural output. For Schultz, rural SWR =
urban SWR = Wr implied that more balanced growth for agriculture and industry should be
pursued. For H&T, it is cheaper to create a job in the barrio than in the city: urban SWR = Wu,
while rural SWR = Wr.
Harberger’s India case

Harberger also has an India case where, in the city, side-by-side with a protected or formal
sector is a large protected or informal sector.

W
St
Sn
Wp

Wn

Dn

Labor
Lt
 Ln U Lp

In the graph above, Lp is the (fixed) number of workers hired by the protected sector at wage
Wp, assumed to be allocated randomly among the larger number of workers Lt willing to work at
wage Wp. After Lp is deducted from Lt, the curve Sn becomes the schedule of supply prices (or
reservation wages) of workers to the unprotected or informal sector. With Dn as the unprotected
sedctor’s demand for labor, the equilibrium is one where Ln workers are with the unprotected
sector getting wage Wn, Lp workers are with the protected sector getting the fixed Wp > Wn,
and U is the number of people who are unemployed with respect to the protected sector but out
of the labor force with respect to the unprotected sector. There U are the people who are not
working but say they are looking for work, only for a high wage above Wn.

Now comes our project hiring G workers at wage Wp. This will tilt Sn to the left, causing Wn to
rise slightly. But the important changes to note are that our labor needs G will be filled in by
some workers coming from the unprotected sector and by others coming from the pool of
“unemployed,” in the respective proportions Ln/(Ln + U) and U/(Ln + U). For those coming from
the pool Ln, the opportunity cost is Wn. For those from the unemployed pool, it is ½ (Wn + Wp).
Thus, SWR = Wn x Ln/(Ln + U) + ½ (Wn + Wp) x U/(Ln + U).

The discussions above apply mainly to unskilled labor, where a dualistic labor market (protected
vs. unprotected, formal vs. informal, organized vs. unorganized, modern vs. traditional) may
exist. For higher skills, labor markets are observed to have hardly any distortions.

R.P. Alonzo