You are on page 1of 17

Immigration and Regional Development:

An Unskilled Country Experience in the Era of Mass


Migrations
Jorge Cariola

lez
Felipe Gonza

Pontificia Universidad Cat


olica de Chile

April 23th, 2012

Motivation
The Era of Mass Migrations: 55 million Europeans migrated to the

Americas between 1850 and 1913.


Main causes were (Hatton and Williamson, 1998):
Transport cost revolution.
Rise in population growth.
3 Labor scarcity in the New World
1
2

Chile as a relatively unskilled country receiving an inflow of

immigrants between 1850 and 1930.


Immigrants average literacy rate of 73% in 1865, higher than 16.7%

of Chileans Immigrants were relatively more skilled than Chileans.


Question of interest: What is the effect this had on the economy?

(we are particulary interested in human capital development). Which


channels are potentially important?

The Data

We coded the 1865, 1875, 1885, 1895, 1907 and 1920 censuses, with

information about 44 departments:


- Not easy to create geographic units comparable over time
(measurement error).
- We exclude all departments north of Copiap
o (Pacific War) and
Magallanes (unpopulated in our base year).
- Data restrictions: 1845 Census not available, no data at the
department level in 1835, few departments in 1813.
From these sources we obtained department-level information on

population, urban population, literacy rates and the total amount of


immigrants.
Lets look at some descriptive statistics...

Table: Immigrants in Chiles 44 departments

Year

Mean

Median

St.Dev.

Min

Max

Total

Over
Population

1865
1875
1885
1895
1907
1920

531
601
738
1,027
1,408
1,635

126
144
192
245
350
424

1,369
1,411
1,513
2,400
3,594
4,477

7
4
1
13
2
2

7,697
6,927
7,664
11,641
18,786
26,908

23,366
26,435
32,467
45,182
61,953
71,944

1.28%
1.27%
1.35%
1.78%
2.08%
2.11%

990

210

2,740

26,908

261,347

1.72%

Total

Ancud
Arauco
Casablanca
Castro
Caupolican
Coelemu
Combarbala
Concepcion
Constitucion
Copiapo
Coquimbo
Curico
Elqui
Freirina
Illapel
Itata
La Ligua
Laja
Linares
Llanquihue
Los Andes
Melipilla
Osorno
Ovalle
Parral
Petorca
Puchacay
Putaendo
Quillota
Quinchao
Rancagua
Rere
San Carlos
San Felipe
San Fernando
Santiago
Talca
Talcahuano
Union
Valdivia
Vallenar
Valparaiso
Victoria
Yungay

Number of Immigrants in 1920


10,000
20,000
30,000

Figure: Number of Immigrants in 1920

Table: Native Literacy Rate


Year

Mean

Median

St.Dev

Min

Max

1865
1875
1885
1895
1907
1920

16.7
20.2
23.6
25.2
35.4
44.9

14.3
17.7
22.4
23.5
33.8
43.5

6.6
7.3
7.1
9.3
9.3
9.5

9.0
8.7
13.6
7.9
21.9
29.8

33.3
40.0
46.0
52.5
60.8
71.2

Total

27.7

24.6

12.7

7.9

71.2

20

Native Literacy Rate


40
60

80

Figure: Native Literacy Rate

1865

1875

1885

1895

1907

1920

Table: Descriptive Statistics

Literacy

1865

16.7

531

41,271

11,828

(6.6)

(1,368)

(32,460)

(20,071)

1875
1885
1895
1907
1920

Total

Immigrants

Population

Urban
Population

Year

20.2

600

47,174

16,467

(7.3)

(1,410)

(39,016)

(26,974)

23.6

737

54,805

22,247

(7.1)

(1,512)

(52,106)

(33,366)

25.3

1,026

57,817

25,072

(9.2)

(2,400)

(63,815)

(45,396)

35.5

1,408

67,586

28,751

(9.3)

(3,594)

(81,278)

(58,609)

45.0

1,635

77,455

35,777

(9.4)

(4,476)

(104,942)

(83,131)

27.7

990

57,685

23,357

(12.7)

(2,739)

(67,515)

(49,566)

Theory: Towards a Unified Framework?

Do natives respond endogenously to immigration? A large flow of

skilled immigrants can affect both the labor and education markets.
Theory analyzes the impact in one or the other. But recently there

have been attempts to study the effect in both markets.


Eberhard (2011):
- Native workers differ in ther ability to accumulate human capital, and
maximize their stream of earnings deciding optimally the amount of
time to devote to human capital accumulation taken as given current
and future prices.
- Workers re-optimize their human capital decisions (education) after
immigration changes the relative price of human capital (labor market).

The Impact of Immigrants on Development

Fixed effects panel data estimation to control for common trends and

variables affecting literacy rates in each department that do not vary


across time:
log ydt

0
= t + d + log(Immigrants)dt + Xdt
+ dt

Robustness: falsification exercises, control variables, different trends.


We interpret these results as correlations. Then we try to identify a

potential causal impact.

Table: Correlation between Immigrants and Literacy Rates


Dependent variable: Log Literacy Rate

Log Immigrants

Not
Grouped
(6)

(1)

(2)

(3)

0.053**
(0.023)

0.052**
(0.022)
0.028
(0.071)

0.071***
(0.024)
0.013
(0.054)
0.015
(0.009)
-0.138**
(0.054)
0.011
(0.017)

0.071**
(0.027)
0.004
(0.062)
0.013
(0.008)
-0.107
(0.067)
0.011
(0.019)

0.081***
(0.025)
0.028
(0.061)
0.011
(0.008)
-0.118*
(0.062)
0.006
(0.017)

0.069*
(0.035)
0.044
(0.069)
0.004
(0.008)
0.006
(0.063)
-0.007
(0.020)

Yes
Yes
264
44
0.879

Yes
Yes
264
44
0.879

Yes
Yes
264
44
0.885

Yes
Yes
232
43
0.878

Yes
Yes
234
39
0.891

Yes
Yes
150
25
0.925

Log Literacy t 1
Log Urban Population
Log Population
Log public schools

Department fixed effects


Year fixed effects
Observations
Departments
R-squared

Without
big cities
(4)
(5)

Table: Robustness and Falsification Exercises


Dependent variable: Log Literacy Rate
All

Log Immigrants #1

Europeans
(3)
(4)

Latins
(5)
(6)

(1)

(2)

0.071***
(0.024)

0.031
(0.022)

0.082***
(0.023)

0.041**
(0.019)

0.060*
(0.031)

0.019
(0.017)

0.013
(0.054)
0.015
(0.009)
-0.138**
(0.054)
0.011
(0.017)

-0.032
(0.050)
0.019**
(0.008)
-0.059
(0.059)
0.013
(0.015)

0.203***
(0.059)
-0.005
(0.008)
-0.076*
(0.042)
0.005
(0.024)

0.078
(0.072)
0.004
(0.008)
-0.024
(0.041)
0.005
(0.019)

0.189**
(0.089)
-0.001
(0.009)
-0.126**
(0.057)
0.008
(0.027)

No
Yes
Yes
264
44
0.885

Yes
Yes
Yes
264
44
0.906

No
Yes
Yes
129
43
0.961

Yes
Yes
Yes
129
43
0.974

No
Yes
Yes
129
43
0.954

0.057
(0.081)
0.006
(0.008)
-0.031
(0.047)
0.007
(0.019)

0.074**
(0.029)
0.016
(0.026)
0.202***
(0.057)
-0.005
(0.008)
-0.093**
(0.043)
0.004
(0.025)

0.040*
(0.022)
0.000
(0.018)
0.078
(0.073)
0.003
(0.008)
-0.024
(0.045)
0.005
(0.019)

Yes
Yes
Yes
129
43
0.972

No
Yes
Yes
129
43
0.962

Yes
Yes
Yes
129
43
0.974

Log Immigrants #2
Log Literacy t 1
Log Urban Population
Log Population
Log Public Schools

Different trends
Department fixed effects
Year fixed effects
Observations
Departments
R2

Europeans
and Latins
(7)
(8)

Exploring Causality: IV approaches

Initial proportion of immigrants as instrument for effective allocation


(Altonji and Card, 1991; Cortes and Tessada, 2011).
Immigrants tend to settle where others from the same origin are.
Initial proportion of immigrants is related to subsequent arrivals but

not to changes in literacy.


X immigrants

od1865

immigrantso1865

immigrantsot

Distance to departments with immigration laws (broadly defined).


A series of immigration laws during the XIX century gave incentives to

settle in certain areas.


Departments located closed to these areas are more likely to receive an

inflow of immigrants (first stage). Exclusion restriction plausible with


fixed effects.

Table: Exploring Causality


Dependent variable: Log Literacy Rate
Initial Weight
Distance
Total
European
to Laws
(1)
(2)
(3)
(4)

Log Immigrants #1

0.302***
(0.046)

-0.074
(0.074)
-0.002
(0.022)
-0.276***
(0.063)
-0.020
(0.022)

0.279***
(0.074)
-0.103*
(0.056)
0.235***
(0.088)
-0.016
(0.017)
-0.002
(0.068)
-0.010
(0.033)

-0.109
(0.106)
-0.009
(0.030)
-0.332***
(0.121)
-0.032
(0.032)

0.351**
(0.149)
-0.145
(0.092)
0.246**
(0.118)
-0.021
(0.021)
0.030
(0.099)
-0.014
(0.041)

Yes
Yes
264
44
50.37

Yes
Yes
129
43
11.25

Yes
Yes
264
44
5.752

Yes
Yes
129
44
3.941

Log Immigrants #2
Log Literacy t 1
Log Urban Population
Log Population
Log Public Schools

Department fixed effects


Year fixed effects
Observations
Departments
F-test first stage

0.395**
(0.153)

Theory: What History Suggests...


1

Immigrants created new industries (e.g. beer), which rised the


demand for skill-workers (Bernedo, 1999):
Immigration

labor market

native decisions

This implies the following possible native re-optimization: (i) training


and industry relocation, (ii) leave the labor force and accumulate
human capital at school. (Altonji and Card, 1991)
2

Immigrants created schools, were professors and brought their


childrens:
Immigration

education market

native decisions

This implies that immigration changed relative prices of education


and, hence, promoted changes in native decisions to study.

Table: Channels linking immigration and human capital (IV)


Dependent variable is:
Specialization
(1)
(2)

Log Immigrants
Log Urban Population
Log Population
Log public schools constructed

Department fixed effects


Year fixed effects
Observations
Departments

Log Professors
(3)
(4)

Log Students
(5)
(6)

Log Agricultural
Workers
(7)
(8)

0.029
(0.020)
-0.004
(0.004)
-0.024
(0.017)
-0.002
(0.006)

0.008
(0.010)
-0.002
(0.004)
-0.013
(0.014)
0.000
(0.006)

0.281
(0.314)
-0.039
(0.074)
0.616***
(0.188)
0.051
(0.064)

0.112
(0.137)
-0.019
(0.057)
0.710***
(0.150)
0.067
(0.052)

0.016
(0.740)
-0.168
(0.187)
1.097**
(0.552)
0.146
(0.173)

0.473
(0.607)
-0.222
(0.212)
0.844*
(0.482)
0.102
(0.172)

0.428*
(0.258)
-0.074
(0.066)
0.637***
(0.207)
0.020
(0.056)

0.519*
(0.295)
-0.084
(0.074)
0.587**
(0.242)
0.012
(0.056)

Yes
Yes
220
44

Yes
Yes
220
44

Yes
Yes
220
44

Yes
Yes
220
44

Yes
Yes
220
44

Yes
Yes
220
44

Yes
Yes
220
44

Yes
Yes
220
44