There Goes the Neighborhood? People’s Attitudes
and the Effects of Immigration on Australia
Mathias Sinning
ANU, RWI and IZA
Matthias Vorell
RWI
13 April 2012
Australian National University (ANU)
Mathias Sinning
13 April 2012
1 / 24
Motivation
• Immigration policies rely on potential effects of immigration and
public sentiment regarding immigration
• Very little is known about the relationship between actual and
perceived immigration effects
Aims of the paper
• Estimate effects of immigration
• Compare immigration effects to people’s attitudes
• Provide empirical evidence on Australia
Australian National University (ANU)
Mathias Sinning
13 April 2012
2 / 24
Australia
• The economic literature has mainly focused on the U.S. and Europe
• Less is known about Australia, although immigration history and
policies were very different from many other countries
• About 27% of the Australian population is foreign-born
• Immigrants to Australia are relatively skilled
Australian National University (ANU)
Mathias Sinning
13 April 2012
3 / 24
Effects of immigration
• The economic migration literature has typically found relatively small
or no effects of immigration on labor market outcomes
• Many studies have used regional variation in immigration and
employed instrumental variables or data of historically unique events
to address the problem of non-random location choices
Attitudes towards immigrants
• Attitudes towards immigration are influenced by economic factors and
non-economic factors
Australian National University (ANU)
Mathias Sinning
13 April 2012
4 / 24
Data sources
• Australian Election Study (AES): 1996, 2001
•
•
•
•
Data on voting behavior and public opinion
Nationally representative sample of voters (voting is compulsory)
Each survey includes about 2,000-3,000 individuals
We restrict our sample to Australian-born persons aged 18 years
and above
• Survey participants were asked
(1) whether immigrants are generally good for Australia’s economy
(2) whether immigrants take jobs away from native-born workers
(3) whether immigrants increase the crime rate
• Electoral divisions of respondents may be combined with
SLA-level Census data (after conversion to postal areas)
Australian National University (ANU)
Mathias Sinning
13 April 2012
5 / 24
Data sources
• Regional-level data from Australian Censuses: 1996, 2001, 2006
• Statistical Local Areas (SLAs) are the smallest geographical unit
in the data
• On average, about 13,000 people live in each of the 1,500 SLAs
• Relevant variables:
• Local unemployment rate
• Median individual income
• Share of migrants
• Neighborhood controls: Population density, median age,
occupational and educational distribution
• We aggregate the 1,313 SLAs in our sample to 131 functional
economic regions (FERs) and 39 Census regions
Australian National University (ANU)
Mathias Sinning
13 April 2012
6 / 24
Data sources
• Crime statistics: 1996, 2001, 2006 (Cornaglia and Leigh, 2011)
• Collected by the governments of six states and the Australian
•
•
•
•
Capital Territory
Northern Territory is not included
Regional-level data: Local Government Areas (LGAs)
Number of observations: 1996: 276, 2001: 499, 2006: 495
We aggregate the LGAs in our sample to FERs and Census
regions
Australian National University (ANU)
Mathias Sinning
13 April 2012
7 / 24
Data sources
• We use the annual total number of crimes per 1,000 persons in our
analysis:
•
•
•
•
•
•
•
•
•
Homicide and related offences
Sexual assault and related offences
Abduction and related offences
Robbery, extortion and related offences
Burglary (including intent)
Theft and related offences
Deception and related offences
Illicit drug crime
Weapons and explosives offences
Australian National University (ANU)
Mathias Sinning
13 April 2012
8 / 24
Figure 1: Proportions of respondents in favour of increasing, maintaining or reducing
current immigration flows to their countries, 2003
Reduce
Same
Increase
100%
90%
80%
70%
60%
50%
40%
30%
20%
10%
0%
Note: Percentages do not take account of non-responses. Weighted data.
Source: International Social Survey Programme 2003 / OECD (2010): International Migration Outlook.
Australian National University (ANU)
Mathias Sinning
13 April 2012
9 / 24
Table 1: Attitudes towards Immigrants
1996
Survey Year
2001
Total
Immigrants take jobs from Australians
Strongly agree
Agree
Neither agree nor disagree
Disagree
Strongly disagree
16.31
25.83
29.13
22.72
6.02
13.55
26.26
28.90
26.02
5.28
14.60
26.09
28.98
24.76
5.56
Immigrants good for economy
Strongly agree
Agree
Neither agree nor disagree
Disagree
Strongly disagree
4.47
41.36
29.51
19.03
5.63
4.44
44.96
32.85
13.31
4.44
4.45
43.59
31.58
15.49
4.89
Immigrants increase the crime rate
Strongly agree
Agree
Neither agree nor disagree
Disagree
Strongly disagree
20.78
31.65
26.80
16.12
4.66
18.11
32.61
28.30
17.15
3.84
19.13
32.25
27.72
16.75
4.15
515
834
1,349
Observations
Source: Australian Election Study.
Australian National University (ANU)
Mathias Sinning
13 April 2012
10 / 24
Table 2: Attitudes: Sample Statistics
1996
Australian Election Study
Age
Female
Unemployed
Married
Below High School
High School Only
Diploma/Trade Qualification
University
Income: up to AUD20,000
Income: AUD20,000-40,000
Income: AUD40,000-70,000
Income: AUD70,000+
Australian Census
Immigrant Share
Population Size (in 1,000)
Median Weekly Income
Unemployment Rate
Median Age
Certificate or Below
Diploma and Advanced Diploma
Bachelor
Graduate and Postgraduate
Observations
2001
Mean
SD
Mean
SD
42.7
0.524
0.037
0.718
0.309
0.120
0.241
0.330
0.159
0.390
0.291
0.159
14.8
0.500
0.189
0.450
0.462
0.326
0.428
0.471
0.366
0.488
0.455
0.366
46.5
0.534
0.044
0.700
0.369
0.133
0.197
0.301
0.203
0.284
0.247
0.182
15.9
0.499
0.206
0.458
0.483
0.340
0.398
0.459
0.402
0.451
0.432
0.386
26.5
342.6
48.05
0.045
34.4
0.477
0.204
0.236
0.083
16.1
191.4
38.68
0.011
2.2
0.123
0.016
0.087
0.037
25.5
348.3
62.12
0.006
36.2
0.497
0.170
0.252
0.081
15.4
201.6
47.54
0.002
2.6
0.127
0.014
0.084
0.040
515
834
Source: Australian Election Study and Australian Census of Population and Housing.
Australian National University (ANU)
Mathias Sinning
13 April 2012
11 / 24
Table 3: Sample Statistics, Income and Unemployment Sample
1996
SLAs
Median Weekly Income
Unemployment Rate
Share of Immigrants
Median Age
Population Size (in 1,000)
Certificate or Below
Diploma and Advanced Diploma
Bachelor
Graduate and Postgraduate
Observations
2001
2006
Mean
SD
Mean
SD
Mean
SD
309.71
0.089
0.173
33.9
12.94
0.485
0.203
0.233
0.079
94.46
0.040
0.101
4.1
17.87
0.138
0.036
0.095
0.049
390.46
0.072
0.175
35.8
13.39
0.495
0.170
0.254
0.082
115.57
0.033
0.103
4.3
18.26
0.143
0.027
0.092
0.053
497.43
0.048
0.181
37.8
13.95
0.476
0.173
0.262
0.088
156.92
0.023
0.107
4.8
18.77
0.150
0.027
0.095
0.058
1,313
1,313
1,313
Source: Australian Census.
Australian National University (ANU)
Mathias Sinning
13 April 2012
12 / 24
Table 4: Sample Statistics, Crime Sample
1996
LGAs
Crimes/1,000 persons
Median Weekly Income
Unemployment Rate
Share of Immigrants
Median Age
Population Size (in 1,000)
Certificate or Below
Diploma and Advanced Diploma
Bachelor
Graduate and Postgraduate
Observations
2001
2006
Mean
SD
Mean
SD
Mean
SD
0.978
282.26
0.093
0.136
34.6
34.7
0.522
0.207
0.204
0.068
1.682
76.76
0.038
0.111
3.2
65.0
0.113
0.031
0.078
0.036
1.654
368.12
0.069
0.151
36.2
32.1
0.527
0.170
0.235
0.068
3.117
113.06
0.030
0.111
3.9
58.1
0.124
0.030
0.082
0.038
1.404
465.50
0.050
0.158
38.8
33.9
0.512
0.174
0.242
0.073
3.120
148.00
0.021
0.113
4.1
61.9
0.133
0.026
0.083
0.045
276
499
495
Source: Australian Census and State Level Data on Offenses.
Australian National University (ANU)
Mathias Sinning
13 April 2012
13 / 24
Empirical Strategy: Attitudes
• We use the following model to estimate immigration effects:
Aijt
=
β0 + β1 Sjt + Xijt β2 + Zjt β3 + θj + λt + εijt .
• Aijt : Attitude measure of individual i in region j at time t,
• Sjt : regional share of immigrants,
• Xijt : socioeconomic characteristics,
• Zjt : region-specific characteristics,
• θj : regional fixed effects,
• λt : time fixed effects,
⇒ β1 captures changes in outcomes due to changes in the share of
immigrants if E (εijt |Sjt ) = 0.
Australian National University (ANU)
Mathias Sinning
13 April 2012
14 / 24
Empirical Strategy: Economic and Social Outcomes
• We estimate the model:
log(yjt )
= γ0 + γ1 log(Skt ) + Zkt γ2 + δk + φt + νkt .
• yjt : outcome measure in region k at time t,
• Skt : regional share of immigrants,
• Zjt : region-specific characteristics,
• δk : regional fixed effects,
• φt : time fixed effects,
⇒ γ1 captures changes in outcomes due to changes in the share of
immigrants if E (νkt | log(Skt )) = 0.
Australian National University (ANU)
Mathias Sinning
13 April 2012
15 / 24
Empirical Strategy
• We employ an IV strategy to obtain an unbiased estimate of the
immigration effect
• Historic settlement patterns are used as an instrument for
immigration:
⇒ We use the 1996 Census to construct the regional share of
immigrants who arrived before 1981 and who did not move over
the last 5 years (between 1991 and 1996)
• The identification strategy relies on the assumption that our outcome
variables are only correlated with historic settlement patterns of
immigrants through their relation with the current regional
distribution of immigrants
• Australia experienced considerable structural changes during the
1980s and 1990s (Productivity Commission, 1998)
Australian National University (ANU)
Mathias Sinning
13 April 2012
16 / 24
0
.1
Share of Immigrants
.2
.3
.4
.5
Figure 2: Relationship between Share of Immigrants and IV
0
Australian National University (ANU)
.05
.1
Instrument
Mathias Sinning
.15
.2
13 April 2012
17 / 24
Table 5: Immigration Effects on Attitudes
OLS
IV
(1)
(2)
(3)
(4)
(5)
0.000
(0.002)
0.004
(0.004)
-0.009
(0.021)
-0.000
(0.002)
265.0
0.891
0.003
(0.004)
144.4
0.714
0.000
(0.002)
0.002
(0.004)
0.010
(0.017)
0.000
(0.002)
265.0
0.891
-0.001
(0.004)
144.4
0.714
0.004∗∗∗
(0.002)
0.009∗∗∗
(0.003)
-0.004
(0.020)
0.004∗∗∗
(0.002)
265.0
0.891
0.011∗∗∗
(0.003)
144.4
0.714
Yes
No
No
Yes
Yes
No
Yes
Yes
Yes
Yes
No
No
Yes
Yes
No
Electoral divisions
Jobs
F (first stage)
Shea Partial R2
Economy
F (first stage)
Shea Partial R2
Crime
F (first stage)
Shea Partial R2
Socioeconomic characteristics
Regional control variables
Region fixed effects
Coefficients on immigrant share. All models include the population size and a time indicator as control variables.
Number of observations: 1,349.
Australian National University (ANU)
Mathias Sinning
13 April 2012
18 / 24
Table 6: Immigration Effects: SLA and LGA level
OLS
IV
(1)
(2)
(3)
(4)
(5)
0.011
(0.020)
0.120∗∗∗
(0.025)
-0.063
(0.074)
3,939
3,939
3,939
-0.184∗∗∗
(0.031)
1034.0
0.347
3,939
-0.059
(0.037)
738.9
0.268
3,939
0.213∗∗∗
(0.010)
0.166∗∗∗
(0.010)
0.103∗∗∗
(0.023)
3,939
3,939
3,939
0.251∗∗∗
(0.017)
1034.0
0.347
3,939
0.102∗∗∗
(0.012)
790.8
0.281
3,939
-0.464∗∗∗
(0.098)
-0.562∗∗∗
(0.108)
0.009
(0.153)
1,270
1,270
1,270
-0.354∗∗∗
(0.127)
571.2
0.477
1,270
-0.493∗∗∗
(0.139)
373.5
0.426
1,270
No
No
Yes
No
Yes
Yes
No
No
Yes
No
Unemployment (SLA)
Share of Immigrants
F (first stage)
Shea Partial R2
Observations
Income (SLA)
Share of Immigrants
F (first stage)
Shea Partial R2
Observations
Crime (LGA)
Share of Immigrants
F (first stage)
Shea Partial R2
Observations
Regional control variables
Region fixed effects
Coefficients on log of immigrant share. All models include the population size and time indicators as control variables.
Australian National University (ANU)
Mathias Sinning
13 April 2012
19 / 24
Table 7: Immigration Effects: FER level
OLS
IV
(1)
(2)
(3)
(4)
(5)
-0.127∗∗∗
(0.046)
-0.027
(0.047)
-0.166
(0.231)
390
390
390
-0.248∗∗∗
(0.080)
100.5
0.412
390
-0.102
(0.078)
112.5
0.360
390
0.247∗∗∗
(0.036)
0.186∗∗∗
(0.029)
0.167∗
(0.101)
390
390
390
0.183∗∗∗
(0.051)
100.5
0.412
390
0.096∗∗∗
(0.035)
113.4
0.359
390
-0.441∗∗
(0.194)
-0.643∗∗∗
(0.243)
-0.777∗∗
(0.384)
F (first stage)
Shea Partial R2
Observations
291
291
291
-0.293
(0.262)
139.8
0.507
291
-0.455
(0.312)
94.9
0.430
291
Regional control variables
Region fixed effects
No
No
Yes
No
Yes
Yes
No
No
Yes
No
Unemployment (FER)
Share of Immigrants
F (first stage)
Shea Partial R2
Observations
Income (FER)
Share of Immigrants
F (first stage)
Shea Partial R2
Observations
Crime (FER)
Share of Immigrants
Coefficients on log of immigrant share. All models include the population size and time indicators as control variables.
Australian National University (ANU)
Mathias Sinning
13 April 2012
20 / 24
Table 8a: Immigration Effects on Unemployment
OLS
IV
(1)
(2)
(3)
(4)
(5)
-0.122∗
(0.070)
0.069
(0.072)
0.452∗∗
(0.196)
-0.111
(0.076)
346.3
0.852
117
0.057
(0.078)
332.6
0.800
117
-0.092
(0.079)
413.3
0.878
117
0.059
(0.072)
417.2
0.846
117
-0.011
(0.082)
244.7
0.749
117
Yes
No
Census regions
Share of Immigrants
F (first stage)
Shea Partial R2
Observations
Share of Low-Skilled Immigrants
F (first stage)
Shea Partial R2
Observations
117
117
117
-0.076
(0.075)
0.099
(0.069)
0.370∗
(0.205)
117
117
117
-0.164∗∗∗
(0.061)
0.013
(0.070)
0.118
(0.108)
F (first stage)
Shea Partial R2
Observations
117
117
117
-0.158∗∗
(0.070)
272.0
0.816
117
Regional control variables
Region fixed effects
No
No
Yes
No
Yes
Yes
No
No
Share of High-Skilled Immigrants
Coefficients on log of immigrant share. All models include the population size and time indicators as control variables.
Australian National University (ANU)
Mathias Sinning
13 April 2012
21 / 24
Table 8b: Immigration Effects on Income
OLS
IV
(1)
(2)
(3)
(4)
(5)
0.227∗∗∗
(0.038)
0.110∗∗∗
(0.023)
0.084
(0.053)
0.194∗∗∗
(0.038)
346.3
0.852
117
0.058∗∗
(0.025)
309.0
0.799
117
0.173∗∗∗
(0.043)
413.3
0.878
117
0.041∗
(0.024)
390.6
0.849
117
0.098∗∗∗
(0.025)
229.3
0.750
117
Yes
No
Census regions
Share of Immigrants
F (first stage)
Shea Partial R2
Observations
Share of Low-Skilled Immigrants
F (first stage)
Shea Partial R2
Observations
117
117
117
0.186∗∗∗
(0.045)
0.077∗∗∗
(0.023)
0.060
(0.062)
117
117
117
0.251∗∗∗
(0.031)
0.138∗∗∗
(0.025)
0.023
(0.029)
F (first stage)
Shea Partial R2
Observations
117
117
117
0.227∗∗∗
(0.033)
272.0
0.816
117
Regional control variables
Region fixed effects
No
No
Yes
No
Yes
Yes
No
No
Share of High-Skilled Immigrants
Coefficients on log of immigrant share. All models include the population size and time indicators as control variables.
Australian National University (ANU)
Mathias Sinning
13 April 2012
22 / 24
Table 8c: Immigration Effects on Crime
OLS
IV
(1)
(2)
(3)
(4)
(5)
-1.504∗∗∗
(0.312)
-1.586∗∗∗
(0.546)
-0.856
(1.202)
-1.196∗∗∗
(0.335)
224.5
0.820
100
-0.940
(0.579)
257.6
0.807
100
-1.063∗∗∗
(0.310)
334.9
0.870
100
-0.760
(0.505)
309.9
0.867
100
-1.234∗∗
(0.611)
110.2
0.733
100
Yes
No
Census regions
Share of Immigrants
F (first stage)
Shea Partial R2
Observations
Share of Low-Skilled Immigrants
F (first stage)
Shea Partial R2
Observations
100
100
100
-1.348∗∗∗
(0.290)
-1.225∗∗
(0.523)
0.108
(0.325)
100
100
100
-1.416∗∗∗
(0.310)
-1.661∗∗∗
(0.474)
0.288
(0.449)
F (first stage)
Shea Partial R2
Observations
100
100
100
-1.314∗∗∗
(0.359)
166.3
0.789
100
Regional control variables
Region fixed effects
No
No
Yes
No
Yes
Yes
No
No
Share of High-Skilled Immigrants
Coefficients on log of immigrant share. All models include the population size and time indicators as control variables.
Australian National University (ANU)
Mathias Sinning
13 April 2012
23 / 24
Conclusions
• There are no adverse effects of immigration on unemployment,
income and crime levels
• The economic effects are in line with people’s expectations
• Many Australians overestimate the effect of immigration on
crime
Australian National University (ANU)
Mathias Sinning
13 April 2012
24 / 24