Rural to Urban: Migration

BAI Ying
Department of Economics
CUHK

February 24, 2025

Outlines

1 Rural to urban
I Dual model: Lewis 1954
I Rural-urban migration: Harris-Todaro 1970

2 Understanding migration 1: how to solve uncertainty?

I An extension of Harris-Todaro model
I Network and migration
I International migration: from Mexico to the US

3 Understanding migration 2: Self-selection

I Roy/Borjas model 1987
I Empirical tests
The structural viewpoint of economic growth

I The literature on economic growth might tempt you to view

economic development as a process that transforms all income and all
sectors of the economy in an even fashion.
I This is seldom the case
I Most often we have uneven growth → Growth that first proceeds by
benefiting some groups in society.
I More often than not, economic development entails the rapid growth
of some parts of the economy, while other parts are left behind.
I We must look at economies and development in a more
dis-aggregated form.
Rural and urban

I The most important structural feature of developing countries is the

distinction between the rural and the urban sector.
I As economic development proceeds, individuals move from rural to
urban
I Agriculture acts as a supplier of labor to industry.
I Agriculture acts as a supplier of food to industry
I These twin resources �food and labour�need to move together if
developments is to proceed.
Non-agricultural sector vs. agricultural sector

I 1. Non-agricultural: formal urban sector

I Firms that operate under accepted rules and regulations imposed by
governments.
I The workers often belong to a union.
I Firms are required to pay minimum wages and must conform to
certain standards of safety, rules of compensation for workers, pension
schemes, and the like.
I Firms pay taxes, may receive infrastructure facilities, and may have
access to foreign exchange quotas or the right to import certain
inputs.
Non-agricultural sector vs. agricultural sector

I 1. Non-agricultural: formal urban sector

I 2. Non-agricultural: informal urban sector
I The informal sector is characterized by a large number of small-scale
production and service activities that are individually or family owned
and use simple, labor intensive technology.
I The informal sector enterprises have lower productivity.
I The workers have low human capital
I Firms are unregulated and do not receive access to privileged
facilities.
I The informal sector usually does not adhere to norms of minimum
wages, retirement plans, or unemployment compensation.
Non-agricultural sector vs. agricultural sector

I 1. Non-agricultural: formal urban sector

I 2. Non-agricultural: informal urban sector
I 3. Agriculture
I In most cases agriculture is a giant informal sector in itself.
I The primary occupation in agriculture is farming.
I Production is organized in many ways:
I Family farms; Large owner-cultivations or capitalist farms; Tenant
farmers who lease land; Laborers who work for wages or a commission
on the land of others.

I The notion of risk and uncertainty is central to the concept of

agricultural organization in developing countries.
Some stylized facts

I Urban areas are growing rapidly

I Urban wages are higher than rural wages

I Urban unemployment / underemployment is high

What models can explain these patterns?

A Dual Economy

I Development often proceeds unevenly and results in a dual economy

consisting of a modern sector and a traditional sector.

I The modern sector typically differs from the traditional sector in that
it has:
I Higher value of output per worker
I Higher wages
I Lower returns to capital
I Higher capital intensity
I Persistent unemployment (especially in urban areas)
Dual model - Lewis 1954

I The first theoretical discussion of internal movements is due to Lewis

(1954).

I He describes developing countries as dual economies, i.e. economies

where two sectors coexist of different nature and functioning.

I A “traditional” sector, often equated to the agricultural sector, labor

intensive, using old production techniques, with a family-based
organization implying output sharing rather than wage payment.

I A “modern” sector, with opposite characteristics: industrial, intensive

in capital and using “new” technologies, organized on capitalist
principles, with wage payment.
Dualism

I A high-wage, capital-intensive industrial sector coexists with a

low-wage traditional sector.

I Dualism is a sign of markets working poorly (market failure case for

deviating from free trade).
Dual model - Lewis 1954, contd.

I The general idea is that the traditional sector supplies labor to the
modern one.

I Obviously, for this to be possible, it requires that the agricultural

sector with less workers can still produce sufficient food surplus to
feed the whole economy.

I Lewis�idea is that there exist a labor surplus in the traditional sector,

that can be removed at little or no cost in terms of lost output from
this sector. i.e. the marginal productivity of labor is nearly nil in this
sector.

I This might be possible in economies where there is a high population

pressure for available arable land.
Dual model - Lewis 1954, contd.
Surplus labor on the family farm
Dual model - Lewis 1954, contd.
Income sharing and surplus labor
I Observation that the wage is positive vs. the parallel observation that
the marginal product is 0
Dual model - Lewis 1954, contd.

I Because in the traditional sector, one is not paid at its marginal

productivity but by output or income sharing (i.e. at average
productivity), it is possible for workers to earn a living even in a
sector where marginal productivity of labor is nil.

I The concept of surplus labor can be extended:

I By considering the possibility that marginal product in the traditional

sector is lower than wage in the modern sector, but not 0.This leaves
a margin for efficiency gain by reallocation of labor among sectors.

I By distinguishing surplus laborer from surplus labor: remaining

workers adapt their labor supply.
Dual model - Lewis 1954, contd.
Disguised unemployment
I The difference between existing labor and the labor that sets
marginal product equal to the wage
Dual model - Lewis 1954, contd.

I Lewis uses this model to describe the development process:

I It proceeds by the transfer of surplus labor (along with surplus food

necessary to sustain modern workers) from the agriculture to industry.

I As long as there is surplus labor in agriculture, the marginal product

of labor does not increase when labor is withdrawn.

I When the entire surplus is soaked up by the modern sector, wages

start to rise in agriculture and dualism disappears.

I In this model, labor moves because of higher wages in the modern

sector that can attract workers.
Rural-urban migration: Harris-Todaro (1970)

I Model that is dedicated to explaining rural-urban migration.

I It assumes a formal urban sector with higher wage than the rural
sector.
I In the formal urban sector, wage is not only high, but also rigid
downward due to unionization, legal framework (minimum wage law
for example) or efficiency wage payment.
I In the rural sector, conversely, wage is fully flexible.
I Migration is viewed as a response to the wage gap between the two
sectors.
I Those who migrate and are not absorbed in the formal sector join
either the pool of the unemployed or the informal sector.
Harris-Todaro (1970), contd.

I Original focus: rural-urban migration in LDC

I Rural residents move to urban regions despite already high
unemployment there
I Puzzle: migration continues although it makes (some) rural migrants
worse off
I HT model: migrants motivated by expected returns
I Expected returns may be different from actually realized returns
Harris-Todaro (1970), contd.
Model setup:

I Two locations: Urban and Rural

I In the rural location �one sector: agriculture (A)
I In the urban location �two sectors: formal urban sector (F) and
informal urban sector (I)
I LA - labor in the agriculture sector
I LF - labor in the formal urban sector
I LI - labor in the informal urban sector
Harris-Todaro (1970), contd.
Demand curves with flexible wages:
I Wage: W∗ ; Agrariculture:L∗ ; Urban: L∗
A F
I Labor demand curve in formal urban sector: AB
I Labor demand curve in agricultural sector: CD
Harris-Todaro (1970), contd.
Minimum wage in the formal sector (urban)
I A floor on the formal wage: w̄
Harris-Todaro (1970), contd.

I The formal urban sector will hire no more than the amount L̄F
I Where do the remainder go?
I If all go to the agricultural sector → wA = w
I This cannot be an equilibrium: No unemployment but different wages,
people will move

I If both sectors pay the same wage, w̄ → unemployment if formal and

agriculture are the only two sectors.
I This cannot be an equilibrium: wages in agriculture are flexible and
unemployed people would go to agriculture and drive down the wage.

I They cannot be in agriculture → must be in urban sector but not in

formal urban → Informal urban sector
Harris-Todaro (1970), contd.

I Worker choose between remaining in the rural/agricultural sector and

a sure wage and moving to the urban area with a positive probability
of landing a job in the formal sector, but also with a positive
probability of ending up unemployed or working for a pittance in the
informal sector.
I The expected wage in the urban sector is neither w̄ norw but
somewhere in between given by: pw̄ + (1 − p)wI
I wI - urban informal wage, exogenous
I p - probability of finding a job in formal sector
I This is the expected wage that is compared to the wage in the
agricultural sector.
Harris-Todaro (1970), contd.

I Harris-Todaro sees the probability of getting a formal job as the

number of formal positions over the number of urban dwellers:
LF
I p= LF +LI

LI
I 1−p= LF +LI

I Because the fate of a potential migrant is not known, we must

consider the expected income from migration and compare it with the
actual income received in agriculture.

L̄F L̄I
I If pw̄ + (1 − p)wI = L̄F +L̄I
w̄ + w
L̄F +L̄I I
= wA
I we are at an equilibrium where no person wishes to migrate from one
location to the other.

I This is the Harris-Todaro equilibrium condition.

Harris-Todaro (1970), contd.
Equilibrium:
Government Policy I: urban job creation

I This simple model is enough to explain the observed puzzling fact

that policies aiming at reducing the size of the informal sector
through job creation in the formal one lead to an increase in
unemployed or informal employment (ex: Kenya, end of the 60’s,
when Harris and Todaro were working there).

I The increased number of jobs, increase p and hence increase the

expected urban earning, driving additional migration.
Government Policy I: urban job creation

I Due to migration, w′A > wA , so that p′ > p

I The fraction for formal jobs in the urban sector has therefore
increased, but since the size of the urban labor force has increased,
this is compatible with an increase in the absolute size of the informal
sector (or of the pool of unemployed).

L̄′F L̄′I
I In the new equilibrium: p′ wF + (1 − p′ )wI = L̄′F +L̄′I
w̄ + w̄
L̄′F +L̄′I I
= w′A
An extreme case: Lewis model - the wage in agriculture does not change a
lot
Government Policy II: to curb unemployment (or informal sector)

I If the policy objective is to curb unemployment in urban areas, what

are the policy options left?

I 1. Migration restrictions

I 2. Wage subsidies
Government Policy II: to curb unemployment (or informal sector)

1. Migration restrictions: LM
A stay in agriculture
Government Policy II: to curb unemployment (or informal sector)

2. Offering a subsidy to employers in the formal sector

Government Policy II: to curb unemployment (or informal sector)

3. A combination: migration restrictions and wage subsidy

Government Policy II: to curb unemployment (or informal sector)

4. Uniform wage subsidy

Extensions
I Harris-Todaro model tends to overestimate actual migration flows.
Various extensions have been proposed to make it more realistic:

I Introduce risk aversion: Risk averse workers will not migrate for an
expected wage equal to the certain earning in agriculture.

I Introduce heterogeneity in the probability of landing a job in the

formal sector

I ...

I Introduce migration costs.

I Note: General understanding on migration, not only rural-to-urban

migration
Outlines

1 Rural to urban
I Dual model: Lewis 1954
I Rural-urban migration: Harris-Todaro 1970

2 Understanding migration 1: how to solve uncertainty?

I An extension of Harris-Todaro model
I Network and migration
I International migration: from Mexico to the US

3 Understanding migration 2: Self-selection

I Roy/Borjas model 1987
I Empirical tests
An extension: risk and risk aversion
Risk and probability
I Take a coin toss, bet $100 on head

Probability Physical outcome Payoff of bet

1/2 Head $100
1/2 Tail $0

Payoff R and expected payoff E(R)

I A random variable R ∈ {0, 100}
I withp1 = p(R1 ) = p(R = 0) = 1/2, and
p2 = p(R2 ) = p(R = 100) = 1/2
I E(R) = 1
2 × 100 + 1
2 × 0 = 50
I E is expectation operator
Risk aversion, contd.
Risk and utility: payoffs yield utility
I An expected payment of $ 50 (?) A certain payment of $ 50

Probability Physical outcome Payoff of bet Utility

1/2 Head $100 U(M + 100)
1/2 Tail $0 U(M + 0)

I U(M) = ln(M) and M = 1000

I Expected utility:
E[U(M + R)] = 21 × ln(1000 + 100) + 1
2 × ln(1000 + 0) = 6.955
I The expected utility of a random variable R is lower than the utility
of the expected value of the random variable:
E[U(M + R)] = 6.955 < U[E(M + R)] = ln(1050) = 6.957
Risk aversion, contd.
Risk aversion, contd.
Certainty equivalent
I An expected payment of $ 50 = A certain payment of $ (?)

Probability Physical outcome Payoff of bet Utility

1/2 Head $100 U(M + 100)
1/2 Tail $0 U(M + 0)

I Certainty equivalent CE:

E[U(M + R)] = 6.955 = ln(1049) = U(M + CE)
I CE = 49, the agent is indifferent between the random variable R and
the certain payment of $ 49
Risk aversion, contd.
Risk premium
I π = E(R) − CE

Probability Physical outcome Payoff of bet Utility

1/2 Head $100 U(M + 100)
1/2 Tail $0 U(M + 0)

I π = 50 − 49 = 1
I Risk premium is small because lottery is relatively small as opposed
to baseline consumption
I M = 100, then CE = 41 and π = 9

I A positive risk premium means a decision maker is willing to pay for

eliminating the risk
I Such a decision maker is risk averse
Risk aversion, contd.
The Arrow‐Pratt measure of relative risk aversion :
′′
I RRA = − UU′ (M)
(M)
M

Probability Physical outcome Payoff of bet Utility

1/2 Head $100 U(M + 100)
1/2 Tail $0 U(M + 0)

I Risk premium is positive if utility is concave

Risk aversion and migration
LF LI
LF +LI w̄ + LF +LI wI = wA
I Assuming that all individuals were risk-neutral
I A risk-averse potential migrant?

How to solve uncertainty? Migrants, being newcomers to the destination

economy, are especially vulnerable to information and commitment
problems that prevent them from participating fully in the market.
Network and migration: Munshi 2003
What role do networks play in integrating migrants into the workforce?
I Setting: migrants from Mexico to the US; The same argument as
Munshi applies domestically as well
I Empirical problem: If we observe that people from a village always go
to the same place, it could be because they have common skills and
there is a demand for those skills in that location.
I Munshi�s solution:
I Use lagged rainfall shocks at the origin location to instrument for the
size of the network
I Use individual fixed effects for migrants to control for selection in
ability

I Data from the Mexican Migration Project

Network and migration: Munshi 2003
Fact 1: More new migrants (<= 3 years) when recent rainfall low. More
established migrants (> 3 years) when old rainfall is low.
Network and migration: Munshi 2003
Recent rainfall has no effect on employment in US. Lagged rainfall has
positive effect on US employment
I Also true when looks only at migrants who have arrived in past 1 or 2
years (column 3) though not much variation left after individual FE
removed
Network and migration: Munshi 2003
Large effects of network size on probability of employment
Outlines

1 Rural to urban
I Dual model: Lewis 1954
I Rural-urban migration: Harris-Todaro 1970

2 Understanding migration 1: how to solve uncertainty?

I An extension of Harris-Todaro model
I Network and migration
I International migration: from Mexico to the US

3 Understanding migration 2: Self-selection

I Roy/Borjas model 1987
I Empirical tests
Some basics of Probability Theory

I u ∼ N(0, 1)
I η ∼ N(0, 1)
2
√1 e− 2
η
I ϕ(·): Probability density function ϕ(η) = 2π
I Φ(·): Cumulative distribution function (CDF)
Self-selection: Borjas 1987

I Heterogenous labor (different level of ability/skill)

I Self-selection: conditional on their own ability, skill, ....
I Applied to: both internal and international migration
Borjas 1987
Borjas 1987: migration costs

I Migration cost is π, which is assumed to be the same for all

individuals
I Each work knows π, µ0 , µ1 , ε0 , ε1
I The correlation of home and destination country earnings is:
ρ = σσ001
σ1 where σ01 = cov(ε0 , ε1 )
I To implement the model we need to know ρ but not the individuals ε;
that is good as in a cross-section we can only observe ε0 or ε1
Borjas 1987: the migration decision
Borjas 1987: the migration decision
Borjas 1987: selection - positive sorting

Technical details
Borjas 1987: selection - negative sorting

Technical details
Borjas 1987: selection -‘refugee’ sorting

Technical details
Immigration flows from Mexico to the US
Chiquiar and Hanson 2005
Outflows from Mexico to the US are large
Chiquiar and Hanson 2005, contd.
Men who migrated to the US are more likely to come from the middle of
the education distribution:
Chiquiar and Hanson 2005, contd.
What would be the wage distribution of Mexican immigrants (US migrants) if
their skills were evaluated at Mexican returns based on education and age?
Conterfactual wage
Chiquiar and Hanson 2005, contd.
Explaning intermediate selection:
Chiquiar and Hanson 2005, contd.
Explaning intermediate selection:
Chiquiar and Hanson 2005, contd.
Comparing with the standard Borjas model:
Fernandez 2011
Fernandez 2011, contd.
Men
Fernandez 2011, contd.
Women
Fernandez 2011, contd.
Wage distribution of men
Fernandez 2011, contd.
Wage distribution of women
Fernandez 2011, contd.
Selection by education
Fernandez 2011, contd.
Summary of selection results
A summary

1 Rural to urban
I Dual model: Lewis 1954
I Rural-urban migration: Harris-Todaro 1970

2 Understanding migration 1: how to solve uncertainty?

I An extension of Harris-Todaro model
I Network and migration

3 Understanding migration 2: Self-selection

APPENDIX
Some basics of Probability Theory, contd
I u ∼ N(0, 1)
I η ∼ N(0, 1)
2
√1 e− 2
η
I ϕ(·): Probability density function ϕ(η) = 2π
I Φ(·): Cumulative distribution function (CDF)

ϕ(z)
I E(u|u > z) = 1−Φ(z)
ϕ(z)
I E(η|u > z) = ρηu 1−Φ(z) Note: ρηu is the correlation between u and η
Borjas 1987: selection
Borjas 1987: selection
Borjas 1987: selection
Borjas 1987: selection conditions
Borjas 1987: positive sorting
Borjas 1987: selection - positive sorting
Borjas 1987: negative sorting
Borjas 1987: selection - negative sorting
Borjas 1987: ‘refugee’ sorting
Borjas 1987: selection - ‘refugee’ sorting
Borjas 1987: a potential fourth case?

back