You are on page 1of 16

The Lewis Dual Sector Model

The first and the most important dual sector model is the model proposed
by Sir Arthur Lewis in 1954. He proposed that the economy of an LDC
is made up of two distinct sectors, a modern sector and a traditional
sector. These sectors are called different things by different people:

Modern sector Traditional sector


Urban Rural
Industrial Agricultural
Advanced Backwards
Capitalist Subsistence

These terms do not have the same meaning within the context of the
Lewis model but they do imply the idea of duality or two different
sectors. For example, it is possible for the traditional sector to be located
in the urban areas. The informal sector is effectively traditional in the
production sense. Some agricultural activities may use sophisticated
capital (plantations).

Assumptions of the model

A1. There are two sectors, traditional and modern.


A2. There is unlimited supply of labour at a fixed wage in the traditional
sector.
A3. Accumulation of fixed capital occurs only in the modern or capitalist
sector. The traditional sector uses labour and land.
A4. The capitalists save all the profits and reinvest it. Workers do not
save.

1
The two sectors are different with respect to the production function they
use. The modern sector has a production function of the form Q = Q(L,K)
and the traditional sector T = T(L, land) where land is the fixed input.

The model

Q = Q(K,L) (1)

is the modern sector production function. All firms maximise profit,


which is the value of output less labour costs (the wage bill), therefore:

P = Q(L,K) – WL (2)

W = wage per worker. Firms hire labour until the marginal product of
labour, MPL equals the wage. This means

¶Q
=W (3)
¶L

Why? To show this we can use the condition of profit maximisation as


labour is increased by 1 unit.

¶P ¶Q
= .1 - W = 0 for a maximum.
¶L ¶L

¶Q
Therefore =W.
¶L

If the supply of labour is perfectly elastic at a fixed wage due to surplus


labour then

¶Q
=W.
¶L

All profits are saved and then re-invested \

2
dK !
P=S=I= =K (4)
dt

This implies that all investment adds to capital stock, i.e. for simplicity
there is no depreciation.

In the traditional sector T = T(L,land) where T = traditional sector output.


In this sector members working on a farm share the farm income. When
we have income sharing then the income going to each member is the
average product of labour, hence

T
WT = = average product of labour where, WT is the earnings of each
L
traditional sector worker.

Lewis assumed that the modern sector wage is higher than the traditional
sector wage so W = WT (1+q) where q is the exogenous wage differential.
He took q to be around 30%.

Assume that in the modern sector output is determined by constant


returns to scale. This means Q(lK, lL ) = lQ(K, L ) and that K L ratio

¶Q ¶Q
determines = (MPL ) and that = (MPK ) . Given this, once we assume
¶L ¶K

a particular short run position, we can solve the model.

¶Q
Assume K = K0. Given equation 3, = W Þ L 0 is the demand for
¶L
K0
labour. Given that we know K0 and now L0 we know therefore we
L0

¶Q ¶Q
know and .
¶L ¶K

3
Now we can use Euler’s theorem. This theorem states that, given CRS,
the value of output is equal to the sum of the returns to the two factors K
and L.

Hence:
¶Q ¶Q
Q= .L + .K
¶L ¶K

We know

P = Q(K,L) – WL

so
ì ¶Q ¶Q ü
P = í .L + .K ý - WL
î ¶L ¶K þ

but
¶Q
.L = WL
¶L

thus
¶Q
P= .K
¶K
which is the return to the capitalists. Divide both sides by K and we get
P ¶Q
the rate of profit =
K ¶K

So the rate of profit = marginal product of capital, as long as we have


surplus labour.

Total output in the economy is Y=Q+T


! Q
Y ! æ Q ö T! æ T ö
The rate of growth is = ç ÷+ ç ÷
Y QèYø TèYø
! Q
Y ! T!
\ = (a ) + (1 - a )
Y Q T

4
Q T
where a is and 1- a =
Y Y
¶T
We know that T! = 0 since = 0 or MPL = 0 in the traditional sector.
¶L
!
Y !
Q
\ = a , i.e. growth rate of the economy is proportional to growth rate
Y Q

of the capitalist sector.

!
Q !
Y
Given constant returns to scale is constant but a increases leading to
Q Y

increasing.

P P Q P
The profit share = . = .a
Y Q Y Q

P P P K P K
is fixed since = . and = MPK which is fixed and is fixed
Q Q K Q K Q

because of fixed input ratio.

P
This implies that as a ­ ® ­ so we have an economy with a rising rate
Y
P
of growth and a rising saving rate æç ö÷ as the traditional sector is
èYø

absorbed into the modern sector. The share of profit in the economy is
also rising.

Implications
a) The growth of the economy is directly related to the growth of the
modern (capitalist) sector which is caused by capitalist savings and
hence investment. Since capitalist savings rises over time growth
also rises over time.
b) There is a redistribution of income in favour of the capitalist sector,
hence inequality rises in the early stages of development, the
Kuznets hypothesis.

5
c) Since capitalists pay a wage W which is W = (1 + q)WT , i.e. higher than
the opportunity cost of labour, there may be a case for protecting
(tariff for example) this sector (see the lecture on trade which shows
why tariff protection may be beneficial, but then shows other
methods are better).

The above diagram is from Todaro and Smith’s textbook. Whilst Ray also
shows the Lewis model with diagrams, this is the best illustration of the
model. The upper left part is the modern sector, exhibiting a standard

6
neoclassical production function. The upper right hand side is the
traditional sector production function, subject to diminishing return. The
lower left side is the modern sector labour market. Because of the
assumption of surplus labour supply is perfectly elastic at the going wage,
WM. The lower right part shows the traditional sector labour diagram,
which implies that, given surplus labour, marginal product of labour is
zero, and income sharing implies that the wage of the worker is the
average product, WA. Wage in the modern sector is exogenously higher
than the traditional sector wage, which provides incentive to migrate.

The explanation is as follows: Start from level of capital, KM1, and


production function TPM1. In the labour market, lower diagram, demand
for labour = supply of labour gives rise to L1 employment. The area under
the demand curve shows rewards to factors of production. OWMFL1 is the
wage bill and WMFD1 is profit. Capitalists invest all this profit (this is
again a simplification), hence create more capital, so capital stock rises to
KM2, production function shifts up, and because of higher capital marginal
product of labour increases so labour demand increases and hence
employment rises to L2.

Profit rises to D2GWM, and the wage bill rises to OWMGL2. However,
profit rises more proportionally (see also mathematical proof) than the
wage bill. The process starts over again until labour demand hits the kink
(see the solid black upward slopping part of the labour supply). At this
point the economy is developed, and surplus labour no longer exists.
Growth in the modern sector now implies that labour demand intersects
labour supply at a higher wage, hence the situation is now just like a
normal neoclassical solution, implying the wage in both sectors has to
rise. Note that now profit rises less and less proportionally and the wage
bill more and more proportionally, hence the Lewis model is effectively

7
providing the theoretical background to the Kuznets inverted U
hypothesis.

Criticisms
1. Assumption of MPL = 0 (or even low MP of labour). This
assumes disguised unemployment. Many researchers have shown
that during peak labour demand (during harvesting or planting
seasons etc), all workers are employed in the traditional sector. If
then labour moves to the modern sector T will fall and the terms
of trade between the two sectors has to rise in favour of the
Pt
traditional sector, i.e. ­ . This implies that W has to rise so
Pm

that capitalist sector workers can buy traditional sector output as


before, i.e. food.

2. Empirical research shows that migration may result or be


associated with a rise in modern sector wage. The first to point
this out was Mabro, who showed that the urban wage in Egypt
rose by 12% in 1976 as migration was occurring.

3. The model assumes that all the absorbed labour, i.e. migrants, are
fully employed in the capitalist sector. Empirical research shows
widespread unemployment in urban areas of LDCs.

4. The model ignores depreciation of K, international trade and


population growth.

5. Economic development takes place via the absorption of labor


from the subsistence/traditional sector where opportunity costs of
labor are very low. However, if there are positive opportunity

8
costs, e.g. loss of crops in times of peak harvesting season, labour
transfer will reduce agricultural output.

6. The transfer of unskilled workers from agriculture to industry is


regarded as almost smooth and costless, but this does not occur in
practice because industry requires different types of labour. The
problem can be solved by investment in education and skill
formation, but the process is neither smooth nor inexpensive.

7. Profits (P) or capitalists’ savings is assumed to be reinvested in


an identical technology, but it is possible that reinvestment may
be in a labour-saving form \ MPL0 ® MPL1 raises P but not
employment.

Wage

SL

WM

MPL0 MPL1
MPL0
L1 L

For more criticism see Basu. There are extensions of this model, for
example the Fei-Ranis model, but candidates are not required to know
these models.

Fields (1980) proposes an interesting model of the dual economy and


looks at development, poverty and inequality. He defines a social welfare
function that the developing country attempts to maximise, which is a
function of income, poverty and inequality. He then analyses three

9
effects: modern sector enrichment (wages rising in the modern sector);
modern sector enlargement (employment rising in the modern sector);
and traditional sector enrichment (wages rising in the traditional sector).
This study is an interesting way of looking at Kuznets inverted U effect
and the circumstances that may give rise to it and relates the labour
market to inequality and poverty. The model is presented here so
candidates are not required to read Fields’ book.

Before the model is presented we briefly look at some concepts of


poverty and inequality. Inequality can be measured by the Gini
coefficient, and represented by the Lorenz curve. Assume z is the poverty
line in money terms. The poverty profiles can be shown in the following
two diagrams. The diagram on the left assumes a linear poverty profile
and the one on the right shows a non-linear one. Assume the poorest
member of the economy earns Y0 and the richest earn YN and there are N
individuals in the economy. The simplest measure of poverty is the
headcount ratio, which is q/N. This is the total number of poor, q, divided
by the total number of people in the economy.

The horizontal axis shows individuals ordered from poorest to richest.

Y Y
YN YN

z z
Yi
Y0 Y0

0 i q N q N

10
Consider individual i. This individual is poor because her income is
below the poverty line. Another way to represent her poverty is the
income gap method. She has an absolute income gap of z-Yi. In other
words, if we give this individual z-Yi, she will become non-poor. The
proportional income gap is given by (z-Yi)/z. How much does the
government need to spend in order to eliminate poverty? In the left hand
case (1/2)(z-Yi)(0q) and in the right hand diagram this is given by:

We can now present the Fields’ model. Assume we have a social welfare
function of the following form:
W=f(Y,I) fY>0 fI<0
where Y is total income and I is an indicator of inequality in its
distribution. Income distribution is said to have '"improved" or
"worsened" according to Lorenz domination (i.e., whether one Lorenz
curve lies wholly above or below a previous one or according to one or
more measures of relative inequality, such as the income share of the
poorest 40% (S) or the Gini coefficient (G)). We also assume that W is
negatively related to poverty as follows:
W = h(P) h'<0
Usual measures of poverty are the number of individuals or families
whose incomes are below the poverty line or the gap between the poverty
line and the average among the poor. Aother poverty measure is the Sen
index (= H [I+ (1 - I)Gp], where H is the head count of the poor, I is the
average income shortfall of the poor, and Gp is the Gini coefficient of
income inequality among the poor).

11
W can be represented as W=W(Y,I,P). We can introduce implications of
the dualistic development model into this function. Assume total income
(Y) is given by
Y = Ym + Yt = Wmfm + Wtft
where t is the traditional sector and m the modern sector. We assume that
Wm > z >Wt, where z is the poverty line. This implies that workers in the
traditional sector are poor and workers in the modern sector are non-poor.
fm and ft are the share of labour in the modern sector and in the traditional
sector respectively; the total economically active population is fm + ft = 1.

The inequality function is assumed to take the following form:


I = I(Wm,fm, Wt, ft)

The poverty index (P) depends on the wage in the traditional sector
and/or the share of the population in that sector:
P = P(Wt,ft)

Substituting these functions in the overall welfare function we get:

W =W[Y, I(Wm,fm, Wt, ft), P(Wt,ft)]

W =W[Wmfm + Wtft, I(Wm,fm, Wt, ft), P(Wt,ft)]

We assume (dW/dY)>0, (dW/dI)<0 and (dW/dP)<0 .

With respect to the dualistic economy we additionally make the following


assumption:
(dI/tdWt)<0 that is if the traditional sector wage rises, inequality will fall,
(dI/dWm)>0, that is inequality will rise if the modern sector wage rises,

12
and (dI/dft) = - (dI/dfm)>0, that is inequality rises if ft fraction of labour
force in the traditional sector rises, and that of the modern sector falls.
With respect to poverty we assume that (dP/dft)>0, and (dP/dWt)<0, that
is if the traditional sector wage falls, or the fraction of the labour force in
the traditional sector rises, poverty will rise.

To find what is the effect of economic growth we can do the following:


ΔY = Y2 -Y1 = (Y2 m + Y2t) - (Y1 m + Y1t)
If we substitute the relevant terms we can decompose ΔY as:

ΔY=(f2 m - f1m)(W1m - W1 t)+(W2m – W1m)f1m +(W2m - W1 m)(f2m -f1m)+(W2t –


W1t)f2t

The first term is called modern sector enlargement effect (change in the
number of people in the modern sector multiplied by the difference
between the wage of the modern sector and that of the traditional sector),
the second term is modern sector enrichment (change in income within
the modern sector), the third term is the interaction between modern
sector enlargement and enrichment (change in income in the modern
sector multiplied by change in the number of workers in that sector), and
the fourth term is traditional sector enrichment (change in income in the
traditional sector multiplied by the number of workers in that sector).

1. Traditional Sector Enrichment

This is the case when income in the traditional sector rises because
for the same employment, wages rise in this sector (Wt2 > Wt1). The
effect on inequality can be shown in the following diagram:

13
% of Income

Wt2
Wt1

ft % of population

ft remains the same but Wt rises so traditional sector enrichment


growth reduces poverty and inequality (red Lorenz curve shifts to
blue Lorenz curve).

2. Modern Sector Enrichment

In this case again ft remains the same but now Wm rises. This
obviously will not change poverty since poverty is not a function of
modern sector wages, but raises Y. The effect on inequality is
shown in the following diagram:

% of Income

Wm1
Wm2

ft % of population

14
For unchanged employment, the rise in modern sector wages
(Wm2> Wm1) shifts the blue Lorenz curve down to the red Lorenz
curve). Hence, modern sector enrichment growth results in higher
Y, more inequality and no change in poverty.

3. Modern Sector Enlargement


This is the case when modern and traditional sector wages remain
the same, but modern sector gets larger. Obviously, this means less
poverty because workers move from the poorer sector into the
richer sector. Income will also rise. The important point here is
now ft2<ft1 and fm2>fm1. The effect on inequality is shown in the
following diagram:

% of Income

ft2 ft1 % of population

fm2
fm1

However, what happened to inequality is not clear since the Lorenz


curves cross the result is ambiguous. Those who are still in the traditional
sector have the same income, but these incomes are a smaller fraction of a
larger total so the new Lorenz curve (red) is below the old one (blue) at
low levels of income. Each worker in the modern sector gets the same

15
absolute income but the share of the richest fm1% is now smaller, so in
that portion the Lorenz curve lies above the old one.

The formula for the Gini coefficient in our dualistic model is:

G = 1 – {[Wt + (Wm-Wt)(fm)2]/[Wt + (Wm-Wt)fm]}

This is a quadratic function. By inspection, G = 0 when fm = 0 and when


fm = 1, and G > 0 if 0 <fm < 1. Thus, the Gini coefficient follows an
inverted-U path. This supports the Kuznets inverted-U hypothesis, which
states that as growth occurs inequality first rises, and then falls.

References:
Ray, D., Development Economics, 1998, Chapter 10.
Further reading:
Basu, Kaushik, Analytical Development Economics, 1997, Chapter 7.
Fields, G.S., (1980) Poverty, Inequality and Development, Cambridge
University Press.

16

You might also like