Population Research and Policy Review

, Volume 31, Issue 1, pp 119–140

The Environmental Impact of Immigration: An Analysis of the Effects of Immigrant Concentration on Air Pollution Levels

Authors

    • Department of SociologyFurman University
  • Ben Feldmeyer
    • Department of SociologyThe University of Tennessee
Article

DOI: 10.1007/s11113-011-9216-3

Cite this article as:
Price, C.E. & Feldmeyer, B. Popul Res Policy Rev (2012) 31: 119. doi:10.1007/s11113-011-9216-3

Abstract

Despite growing interest in the impact of immigration on U.S. society, research has rarely examined the effects of immigration flows on the natural environment. The current study addresses this gap in research using data on 183 Metropolitan Statistical Areas drawn from the Environmental Protection Agency, the U.S. Census Bureau, and the National Oceanic and Atmospheric Administration to empirically assess the relationships between contemporary immigration and seven measures of air pollution. In doing so, we seek to (1) broaden knowledge about the social consequences of immigration to include its potential effects on the environment, (2) address competing theoretical perspectives about immigration-environment relationships (i.e., population pressure/social disorganization versus ecological footprint/community resource perspectives), and (3) extend knowledge about the predictors and sources of environmental harm within local communities. In contrast to popular opinion and population pressure positions, our research indicates that immigration does not contribute to local air pollution levels across any of the seven pollution measures examined.

Keywords

ImmigrationPollutionEnvironmentEcological footprint

The impact of immigration on local communities has garnered considerable attention and debate among both the general public and social scientists throughout the last century, particularly following periods of high immigration. Near the end of the twentieth century, the United States experienced dramatic increases in immigration that have been unrivaled since the European migration flows during the early 1900s. Specifically, between 2000 and 2007, the U.S. immigrant population rose by nearly 25% to include more than 38 million individuals (Pew Hispanic Center 2009). In response to these changes, questions about the social consequences of immigration for community well-being—including crime, employment, and other social conditions—have resurfaced in public and scholarly communities. However, there remains a surprising scarcity of social science research addressing the environmental impact of contemporary immigration to the United States.

Separately, immigration and the natural environment are politically charged issues that have received substantial public and scholarly consideration. Yet, researchers have rarely addressed questions about potential immigration-environmental relationships, such as: Do immigrants increase environmental pollution in local communities? Have environmental problems increased in response to rising immigration flows in recent decades? Theoretically, why (or why not) might we expect immigration to shape local environmental problems and what is the process through which such relationships might work?

Politically, identifying whether (and the degree to which) immigrants shape the natural environment has implications for efforts aimed at combating local environmental problems, identifying communities at risk for environmental harm, and determining the most effective resource allocation strategies to address potential sources of pollution. Addressing these questions also has critical implications for political debates concerning immigration and environmental policies, particularly in light of the increasingly salient concerns surrounding immigration and environmental issues and growing concerns that immigration could contribute to environmental harm. For example, despite a paucity of research on the subject, claims about immigration-pollution links and the “environmental threat” of immigration are commonplace in public and political discourse (Mann 1990; Population-Environment Balance 1992, p. 306; Simcox 1992; Zuckerman 1999; see reviews in Kraly 1995, 1998; Squalli 2009, 2010; Stevens 2010).

Substantively, assessing immigration-environment relationships contributes to knowledge across a wide range of disciplines interested in the impact of population movement on social well-being, including demography, environmental sociology, urban/rural sociology, public health, and related fields. Furthermore, addressing this relationship has important theoretical implications for environmental sociology and for demographic research on the effects of population movement, particularly considering that sociological theory provides alternative, competing perspectives on the potential impacts of immigration on the environment. Several mainstream theoretical traditions suggest that immigration may contribute to increased environmental problems at the local level by rapidly increasing local population pressures (e.g., increased size and density) (Bartlett and Lytwak 1995; Daily et al. 1995), or by socially disorganizing communities and inhibiting residents’ abilities to effectively organize to combat sources of environmental degradation and other social problems (Shaw and McKay 1942; see reviews in Lee et al. 2001; Martinez 2002; Stowell 2007). In contrast, some research suggests that immigration may have little or no effect on pollution levels (Squalli 2009, 2010), as immigrants typically have a smaller ecological footprint and create less pollution than U.S.-born residents (Bohon et al. 2008; Hunter 2000a; Neumayer 2006; White 2007). Additionally, several emerging community resource perspectives—including, the “Latino Paradox” (Sampson 2008; Sampson and Bean 2006) and “Immigrant Revitalization” perspectives (Desmond and Kubrin 2009; Martinez 2006)—suggest that rather than disorganizing neighborhoods, immigration may actually stabilize communities and enhance residents’ abilities to address social problems like environmental harm by providing protective resources and reinforcing social institutions and social support networks (Feldmeyer 2009; Light and Gold 2000; Portes and Rumbaut 2006; Portes and Stepick 1993; Sampson 2008).

Using data from the Environmental Protection Agency (EPA), U.S. Census Bureau, and National Oceanic and Atmospheric Administration (NOAA), our study builds on the foundational work of Squalli (2009, 2010) and extends prior immigration-environment research by providing a rare empirical assessment of immigration effects on multiple forms of air pollution. To our knowledge, this also marks the first such analysis to assess these relationships using Metropolitan Statistical Areas (MSAs) as a study unit. The current study extends prior research in several ways, by (1) broadening knowledge about the social consequences of immigration to include its potential effects on the environment, (2) addressing competing theoretical perspectives about immigration-environment relationships (i.e., population pressure/social disorganization perspectives versus ecological footprint/community resource perspectives), (3) extending knowledge about the structural sources of environmental harm across local communities, and (4) assessing the robustness and generalizability of prior work on immigration and air pollution using alternative locales and study units.

Immigration and Environmental Harm

There is no shortage of claims in public discourse, popular press, and political arenas suggesting that immigration may contribute to environmental harm (and other social problems) by placing increased pressure on local ecosystems and straining the environmental carrying capacity of local communities. For example, these popular perceptions and assumptions are well illustrated in the President’s Council on Sustainable Development Population and Consumption Task Force report (1996, p. 8), which concludes that “reducing immigration levels is a necessary part of population stability and the drive towards sustainability.” Furthermore, in response to these assumptions, we have seen the development of politically-oriented organizations (e.g., Numbers USA, Population-Environment Balance, Carrying Capacity Network, and Center for Immigration Studies) that advocate for reduced immigration and U.S. population growth as a method for preventing environmental harm (Beck 1996; Population-Environment Balance 1992).

Despite the urgency of this debate and increasing political attention to both immigration and environmental issues, extant research has provided few empirical assessments of immigration-environment links within the U.S. (for notable exceptions see Cramer 1998; Squalli 2009, 2010; also see review in Kraly 1995). Nevertheless, mainstream sociological theories and research offer several positions that have been used to justify the assumptions described above and explain why immigration to the U.S. could potentially increase local levels of environmental harm. First, population pressure arguments drawn from environmental sociology and demographic research suggest that immigration contributes to environmental problems by increasing local population growth, which places excess strain on the community infrastructure and environment. The origins of such assumptions stem from Thomas Malthus’s (1797) argument that population growth could eventually outpace food production. Since Malthus, population pressure concerns (i.e., Neo-Malthusian positions) have evolved to suggest immigration and other forms of population growth place demands on an area’s environmental carrying capacity, including its ability to meet the population’s demands for energy and water and to manage waste, pollution, land use, and other environmental concerns (Catton 1980; Daily and Ehrlich 1992; Ehrlich and Ehrlich 1992; Foster 2002; see Kraly 1995).

Because immigration has been one of the primary causes of U.S. population growth at the turn of the twenty-first century (along with natural population growth), population pressure positions suggest it may be a noteworthy source of environmental strains and harm (Bartlett and Lytwak 1995; Daily et al. 1995; Garling 1998). Nearly 7 million immigrants entered the United States between 2000 and 2007, which accounts for more than one-third of all U.S. population growth during this period (Pew Hispanic Center 2009). According to population pressure positions, community infrastructures may be unable to accommodate the increased consumption and waste (e.g., increased use of cars, consumption of water/food) that accompany such sizable population increases. In addition, these population pressure effects may be particularly salient due to the rapid rate at which immigrants have migrated to the U.S., leaving some communities and local environments little time to adapt.

Research also suggests that immigration may have particularly harmful environmental effects compared to other forms of population growth (i.e., natural population growth, domestic migration). Because immigrants tend to be highly geographically concentrated, immigration contributes to increased population density and urbanization (Garling 1998; Healey 2006)—two factors that are commonly linked to greater environment problems, such as increased carbon dioxide and vehicle emissions (Cole and Neumayer 2004).

Furthermore, theory and prior research offer several potential reasons why immigration could contribute to long-term environmental harm, even if it has little immediate impact on local pollution. For example, population pressure arguments note that contemporary immigrant populations tend to have higher fertility rates than the U.S.-born (Carter 2000; Heim and Austin 1996; Johnson and Lichter 2008), which may place additional pressure on local communities and environments over time and with increasingly large successive generations (Bartlett and Lytwak 1995; Garling 1998; Hall et al. 1994; see Kraly 1995). Immigration may also contribute to long-term environmental harm if foreign-born populations begin to adopt American lifestyles and resulting increased consumption and waste patterns (Bartlett and Lytwak 1995; Population-Environment Balance 1992; Hall et al. 1994; Hunter 2000a). In addition, because they serve as an attractive source of low-wage labor for many businesses, immigrant populations may inadvertently attract and support the development of industrial and manufacturing sectors of the labor market (e.g., meat processing, textile industries) that tend to contribute to pollution problems (Ilea 2009; Light and Gold 2000; Martinez 2002; Portes and Zhou 1992). Although this is likely to generate employment opportunities and bolster labor force participation within the community, it may also have the unintended consequence of strengthening industries that tend to produce high levels of nitrogen dioxide, ozone, sulfur dioxide, particulate matter, and other pollutants. Thus, as Muradian (2006, p. 209) summarizes, population pressure positions offer several potential links between immigration and pollution and generally suggest that “immigrants contribute directly to the degradation of the local environment by means of urban sprawl, congestion and pollution, waste generation, water consumption, land conversion, depletion of natural resources and biodiversity loss.”

Last, the theoretical assumptions of the Chicago school of sociology and social disorganization perspectives (at least the more traditional versions) imply that immigration could potentially inhibit community residents’ ability to effectively communicate and organize to address local social problems, such as crime, physical and social disorder, and environmental harm (Garling 1998; Shaw and McKay 1942; Thomas and Znaniecki 1927; see reviews in Kubrin and Weitzer 2003; Martinez and Lee 2000). These perspectives suggest that as new residents from abroad flow into communities, immigration contributes to population turnover, creates a greater mixing of people with different backgrounds and languages, and may increase the poverty level of the community by introducing a population with relatively fewer economic resources (see reviews in Feldmeyer 2009; Martinez 2002; Stowell 2007). These social conditions combined with the weaker political capital of some new immigrant groups (Light and Gold 2000; Portes and Rumbaut 2006; Steffensmeier and Demuth 2001) may weaken residents’ ability to address pollution problems or raise the social and political attention needed to insulate the community from potential sources of pollution. Thus, according to this perspective, immigration may disorganize or destabilize communities in ways that make them more susceptible to environmental harm, even if immigrants themselves do not produce high levels of pollution.

Neutral Effects of Immigration on the Environment

In contrast to the theories and perspectives described above, there are also several theoretically driven reasons why we might expect immigration to have no appreciable effect on environmental problems. First, the ecological footprint perspective suggests that although immigration contributes to local population growth, it may produce less environmental harm than other forms of local population growth (natural growth, domestic migration) because immigrants tend to have a smaller ecological footprint than U.S.-born residents. While U.S.-born residents have one of the largest ecological footprints worldwide (White 2007), foreign-born individuals tend to have lifestyles that are less demanding on the ecosystem, involve less consumption and waste, and generally produce less environmental harm. Immigrants tend to be less affluent and utilize fewer of the luxury items and technologies (e.g., SUVs, private golf courses, airlines/jets, consumer electronics) that are commonly linked to increased consumption, waste, and environmental stress (Dietz and Rosa 1997; Hynes 1999). Immigrants are also more likely to carpool, use public transportation, and live in smaller houses (Atiles and Bohon 2003; Blumenberg and Shiki 2008; Bohon et al. 2008), all of which result in a smaller ecological footprint. Moreover, compared to U.S.-born individuals, recent immigrants tend to express higher concern for the environment and are more likely to engage in environmentally-friendly behaviors, such as water conservation and reduced meat consumption (Hunter 2000a; Pfeffer and Stycos 2002). Thus, rather than assuming all forms of population growth (immigration), have the same effects on the environment, the ecological footprint positions suggest that more nuanced models of environmental influence (e.g. IPAT and STIRPAT,1 see Dietz and Rosa 1994, 1997; Ehrlich and Holdren 1971; York et al. 2003a, b) are needed that can consider the unique environmental impacts of the different types of population growth and that can account for the different characteristics and lifestyles of the foreign- versus U.S.-born (Cohen 2006; Hynes 1999; Preston 1996; Rees 1992; Wackernagel and Rees 1996).

Second, community resource positions—including the “Latino paradox” (Sampson 2008; Sampson and Bean 2006) and the “immigrant revitalization thesis” (Martinez 2002; see Desmond and Kubrin 2009; Ousey and Kubrin 2009)—suggest that rather than disorganizing communities, immigration may help to stabilize and solidify community organization to counter social problems like pollution and environmental harm. Drawing from research on immigration assimilation and ethnic economies, these positions suggest that immigration reinforces strong kinship bonds and social networks within foreign-born communities (Feldmeyer 2009, 2010; Light and Gold 2000; Stowell 2007). Immigration solidifies community cohesion and strengthens social capital and social support networks by reinforcing residents’ common heritage and culture (Lee et al. 2001; Martinez 2002; Portes and Rumbaut 2006; Velez 2006). Furthermore, research indicates that geographically concentrated immigrant populations tend to strengthen protective social institutions (such as the family, church, and labor market) that provide beneficial community resources and support (Light and Gold 2000; Martinez and Lee 1998, 2000; Portes and Rumbaut 2006). Thus, immigration may not destabilize communities at all, but may instead provide resources and foster greater organization and more effective coordination among residents for addressing environmental harms and other social problems.

Although there is a near absence of empirical research assessing immigration-environment relationships, the small body of existing literature is generally supportive of the ecological footprint/community resource positions and indicates that immigration has little to no impact on environmental harm. Specifically, several noteworthy studies conducted by Squalli (2009, 2010) report that geographic areas (counties and states) with higher proportions of foreign-born residents have lower levels of air pollution (Squalli 2009, 2010). In addition, Cramer (1998) found that although population growth (in general) increases emissions in California, the effects cannot be attributed specifically to immigration. Although these studies offer promising directions in this line of research, there is an urgent need for further analyses that assess potential links between immigration and environmental harm.

The Current Study, Data, and Methods

The current study extends research on the environmental consequences of immigration and builds in particular on the work of Squalli (2009, 2010) by examining the effects of contemporary immigration flows on levels of air pollution across 183 U.S. MSAs from 2000 to 2006. In doing so, this research contributes to our understandings of the social consequences of immigration on local communities and extends research identifying the structural predictors of environmental harm. This project addresses the competing hypotheses identified above concerning whether immigration contributes to increased air pollution (as suggested by population pressure and social disorganization perspectives) or whether immigration has minimal impact on local air pollution levels (as suggested by the ecological footprint and community resource positions). In addition, we extend prior research by providing one of the first empirical analyses that (1) simultaneously accounts for the independent effects of all three forms of population change (immigration, domestic migration, and natural population growth), (2) accounts for population change and immigration flows (rather than more static “percentage immigrant” measures commonly used in prior research), and (3) uses MSAs as an alternative unit of analysis for addressing immigration-pollution relationships.

Data

Data on immigration, air pollution, weather trends, and structural characteristics of MSAs are drawn from three sources. Our measures of air quality are drawn from the U.S. EPA Air Quality Trends Report, 2000–2006. Data on immigration, population change, and structural conditions are from 2000 to 2006 U.S. Census data (Estimates of Population Change—Summary File 2). Last, we gathered information on weather and temperature trends from the NOAA climate data. Data from the EPA, U.S. Census, and NOAA were merged and aggregated to the MSA-level—which serves as the unit of analysis—resulting in a sample of 183 MSAs across the United States.2

Dependent Variables

Our dependent variables include seven measures of air pollution drawn from 2006 EPA data. These include six measures of specific principal air pollutants: carbon monoxide (CO), nitrogen dioxide (NO2), ground-level ozone (O3), sulfur dioxide (SO2), all measured in parts per billion (ppb), and two measures of particulate matter (coarse particlesPM10; fine particlesPM2.5) measured in micrograms per cubic meter (μg/m3). In addition, we used principal components analysis to create an Air Pollution Index that combines information from the six specific measures to create a seventh composite measure of overall MSA air pollution levels.3 The EPA tracks concentration levels for each of these pollutants at selected monitoring (or trend) sites throughout the nation. Each collection site monitors different pollutants, and not all measurements are collected at each trend site. As a result, sample sizes for our multivariate models vary across types of pollutant and depend on the number of collection sites monitoring a specific pollutant, ranging from a sample size of only 52 MSAs for NO2 models to more than 150 MSAs for models of particulate matter (PM2.5) and the Air Pollution Index (see Table 4 for sample sizes for each pollutant).4 Table 1 provides a brief description of each air pollutant examined, including how the pollutant is measured, its primary sources, and its possible health effects on populations (for further details, see http://www.epa.gov/air/urbanair/6poll.html).
Table 1

Descriptions of principal air pollutants

Pollutant

Measurement

Sources

Effects

Carbon Monoxide

2nd Max, ppb

Vehicle exhaust

Cardiovascular and vision problems, high levels are poisonous

Nitrogen Dioxide

Annual Mean, ppb

Motor vehicles, electric utilities

Water quality deterioration, acid rain, visibility impairment

Ozone

2nd Max, ppb

Vehicle exhaust, electric utilities

Damages vegetation and ecosystems, asthma, chest pain

Sulfur Dioxide

Annual Mean, ppb

Fuel combustion, electric utilities

Breathing difficulty, heart disease, acid rain, visibility impairment

Particulate Matter (10)

2nd Max, μg/m3

Smokestacks, construction sites, power plants

Asthma, heart attacks, lung disease, changes nutrient levels in soil and water sources

Particulate Matter (2.5)

Weighted Annual Mean, μg/m3

Smokestacks, construction sites, power plants

Asthma, heart attacks, lung disease, changes nutrient levels in soil and water sources

Independent Variables

Our primary variable of interest is the MSA net immigration rate for the 2000–2006 period drawn from U.S. Census data on population change. Net immigration is measured as the net number of international migrants to an MSA (i.e., people arriving from another country minus people leaving for another country) divided by the MSA population. Although prior research has commonly relied on measures of immigration based on population composition at one point in time (e.g., % Foreign Born in 2000), a key strength of our measure is that it reflects a more dynamic process of immigration flows and contemporary population change (from 2000 to 2006). In light of the rapidly changing contemporary immigration flows to the US, capturing immigration change (rather than static measures of % immigrant) is particularly important and provides a noteworthy contribution to prior immigration-environment research.

Drawing from prior research and theory on population pressure and the environment, we also control for two other key measures of population change: (1) net domestic migration rate, measured as the net number of people moving into and out of an MSA (from/to other U.S. locales, during the 2000–2006 period) divided by the MSA population and (2) the natural population growth rate, measured as the number of births minus deaths in an MSA from 2000 to 2006 divided by the MSA total population. As we noted earlier, accounting for each of these three forms of population growth (immigration, domestic migration, and natural growth) reflects a key contribution of the current study. Specifically, this allows us to assess and compare the independent impact of immigration on pollution levels, net of the effects of natural growth and domestic migration.

In addition, we control for several structural and environmental conditions of MSAs that may be related to both immigration and MSA-levels of air pollution. These include: average annual temperature, measured in degrees Fahrenheit; average annual precipitation, measured in inches; total population based on year 2000 census population estimates; population density, as the number of people per 1,000 square meters of land; percent workers driving alone to work, as the percentage of the MSA civilian workforce that drives a car to work with no other passengers; percent workers in manufacturing,5 as the percentage of employed workers that are in manufacturing jobs; and percent population in poverty, measured as the percentage of the MSA population with incomes below the poverty line.

We use multivariate ordinary least squares regression (OLS) models to assess the effects of immigration on MSA air pollution levels (net of other factors). We estimated a series of seven models—one for each pollutant and the air pollution index—in which immigration and controls for population change, environmental, and structural conditions of MSAs were entered as predictors of each pollutant. We conducted preliminary tests to ensure that the models and variables did not present multicollinearity problems or violate other Gauss Markov assumptions (e.g., homoskedasticity, skewness). The independent variables were all correlated below r = 0.50 (excluding the relationship between population size and density, r = 0.62). Regression diagnostics revealed that variance inflation factor (VIF) scores were at acceptable limits (at or below 3.0) for all variables included in the models, indicating that multicollinearity was not a critical concern.6

Results

Table 2 displays descriptive statistics for the control variables and for measures of immigration and air pollution levels within MSAs. The results show that average MSA air pollution levels within the sample are well below the current EPA standards for acceptable levels for five of the six pollutants examined (ground-level ozone is at the national standard). In addition, standard deviation values shown in Table 2 indicate that these air pollution levels vary widely across MSAs, with some places having levels that fall far below EPA air quality standards and other places having pollution levels (i.e., ground-level ozone) that are more than 80% higher than EPA standards.7
Table 2

Means and standard deviations for independent and dependent variables

Variables

Mean

Std. Dev.

EPA Standard

Dependent

 Carbon monoxide

2.255

0.804

9.000

 Particulate matter 10

56.105

23.011

150.000

 Ground-level ozone

0.075

0.009

0.075

 Nitrogen dioxide

0.013

0.005

0.053

 Sulfur dioxide

0.004

0.002

0.030

 Particulate matter 2.5

11.504

2.768

15.000

 Air pollution index

0.000

1.000

 

Independent

 Immigration

15.846

16.350

 

 Domestic migration

22.347

67.221

 

 Natural population growth

40.530

31.956

 

 Average annual temperature

56.566

9.589

 

 Average annual precipitation

36.617

14.556

 

 Total MSA population

818,545.843

1,957,795.946

 

 Population density

0.113

0.106

 

 % Workers driving alone to work

79.193

4.438

 

 % Workers in manufacturing

13.983

6.734

 

 % Population in poverty

13.460

6.361

 

Note: Principal pollutants are measured in parts per billion (ppb) except particulate matter 10 and particulate matter 2.5 that are measured in micrograms per cubic meter (μg/m3)

The descriptive statistics also provide interesting information about patterns of migration and population change within MSAs during the 2000—2006 period. Table 2 indicates that the MSAs examined generally experienced positive population growth from a combination of immigration, domestic migration, and natural population growth (births minus deaths). Natural population growth and domestic migration were responsible for the greatest share of this population change, based on their mean MSA growth rates of 41/1,000 and 22/1,000 respectively. Immigration also contributed to positive population growth (mean = 16/1,000) but had a much smaller impact on MSA population change compared to natural growth and domestic migration rates. Although most MSAs experienced growth, standard deviation values shown in Table 2 indicate that population change from 2000 to 2006 varied widely across places. That is, some MSAs experienced natural population declines (more deaths than births) and population losses from domestic and international out-migration, while others saw dramatic growth and in-migration during this period.

Bivariate Results

Table 3 displays the bivariate relationships between measures of immigration, pollution, and structural and environmental conditions within MSAs, which provide an initial method for exploring links between immigration and air pollution. The bivariate findings indicate that immigration is associated with several forms of air pollution but not in consistent ways. Specifically, immigration is positively correlated with particulate matter 10 (r = 0.23, P < 0.05), indicating that this pollutant is more prevalent in MSAs with high immigration rates. In contrast, sulfur dioxide levels are significantly lower in MSAs with high immigration rates (r = −0.45, P < 0.01). Immigration rates are not significantly related to the air pollution index or levels of air pollution for the other four specific pollutants examined (carbon monoxide, nitrogen dioxide, particulate matter 2.5, or ground-level ozone). Thus, our bivariate findings provide mixed evidence concerning immigration-pollution relationships but generally support the ecological footprint and community resource positions, showing that immigration does not substantially contribute to increased air pollution for six of the seven pollution measures examined. In contrast, natural population growth is positively correlated with five out of seven pollution measures (carbon monoxide, particulate matter 10, ground-level ozone, nitrogen dioxide, and the air pollution index), signifying that some types of population growth are associated with increased urban air pollution.
Table 3

Bivariate correlation matrix

 

1

2

3

4

5

6

7

8

9

1. Carbon monoxide

1.000

        

2. Particulate matter 10

0.142

1.000

       

3. Ground-level ozone

0.156

0.321**

1.000

      

4. Nitrogen dioxide

0.340*

0.278

0.114

1.000

     

5. Sulfur dioxide

−0.163

−0.097

0.089

0.025

1.000

    

6. Particulate matter 2.5

0.086

0.125

0.509**

0.185

0.508**

1.000

   

7. Air pollution index

0.364 **

0.542**

0.773**

0.546**

0.412**

0.740**

1.000

  

8. Immigration

0.170

0.228*

0.117

0.227

−0.449**

−0.084

0.052

1.000

 

9. Domestic migration

0.147

−0.039

0.161

0.032

−0.100

−0.178*

−0.014

0.109

1.000

10. Natural population growth

0.265*

0.377**

0.189*

0.473**

−0.530**

−0.119

0.133*

0.448**

−0.039

11. Average annual temperature

−0.121

0.080

0.135

−0.138

−0.315**

0.077

0.022

0.210**

0.217 **

12. Average annual precipitation

−0.175

−0.360**

0.150

−0.304*

0.236*

0.460**

0.128

−0.327**

0.066

13. Total MSA population

0.121

0.014

0.227**

0.491**

−0.182

0.157*

0.164**

0.392**

−0.089

14. Population density

−0.015

−0.170

−0.010

0.180

−0.085

0.148

0.015

0.292**

−0.080

15. % Driving alone to work

−0.029

−0.132

0.146

−0.326*

0.387**

0.240**

0.095

−0.497**

0.011

16. % Workers in manufacturing

0.037

−0.043

−0.005

0.096

0.416**

0.399**

0.158**

−0.193**

−0.272**

17. % Population in poverty

0.070

0.431**

0.231**

−0.057

−0.127

0.205**

0.128*

0.248**

−0.146*

 

10

11

12

13

14

15

16

17

1. Carbon monoxide

        

2. Particulate matter 10

        

3. Ground-level ozone

        

4. Nitrogen dioxide

        

5. Sulfur dioxide

        

6. Particulate matter 2.5

        

7. Air pollution index

        

8. Immigration

        

9. Domestic migration

        

10. Natural population growth

1.000

       

11. Average annual temperature

0.044

1.000

      

12. Average annual precipitation

−0.439**

0.304**

1.000

     

13. Total MSA population

0.092

0.059

−0.077

1.000

    

14. Population density

0.027

0.127

0.165*

0.618**

1.000

   

15. % Driving alone to work

−0.437**

−0.108

0.365**

−0.390**

−0.328**

1.000

  

16. % Workers in manufacturing

−0.232**

−0.208**

0.213**

−0.055

0.038

0.424**

1.000

 

17. % Population in poverty

0.336**

0.440**

−0.029

−0.075

0.147 *

−0.337**

−0.150*

1.000

P < 0.05, ** P < 0.01

For further descriptive information on immigration-pollution links, Fig. 1 provides a graphic illustration of U.S. immigration and air pollution levels during recent decades. Similar to the bivariate results, these trends provide no indication that immigration has contributed to increased air pollution. Instead, the U.S. has experienced steady declines in air pollution levels, despite the fact that immigration rates to the U.S. have increased between 1990 and 2008, rising from a rate of 470 per 100,000 people in 1990 to 610 per 100,000 by 2008. Specifically, ground-level ozone decreased from nearly 20% above EPA recommended levels (in 1998) to slightly below the EPA national air quality standard during this period of high immigration. Similarly, average carbon monoxide levels decreased from 33% below EPA standards in 1990 to nearly 80% below the standards by 2008. However, these bivariate and descriptive findings provide only preliminary conclusions. We turn next to our multivariate models to provide more rigorous tests of immigration effects on pollution and to determine if these bivariate relationships persist when controlling for other structural conditions of MSAs.
https://static-content.springer.com/image/art%3A10.1007%2Fs11113-011-9216-3/MediaObjects/11113_2011_9216_Fig1_HTML.gif
Fig. 1

Immigration and air pollution trends, 1990–2008 1

1 Due to annual fluctuations in the data, we used 3-year moving averages to smooth the immigration trend, with 2-year averaged rates for the series endpoints in 1980/1981 and 2007/2008.

2 Values above 100% indicate pollutant is above EPA national standard. Values below 100% indicate pollutant is below EPA national standard

Multivariate Results

Table 4 displays the results of our multivariate OLS regression models predicting MSA-levels of air pollution for six specific pollutants and the air pollution index. Turning first to the control variables, Table 4 supports findings from prior literature and indicates that several characteristics of MSAs are associated with higher ambient levels of air pollution (see Cole and Neumayer 2004; U.S. Environmental Protection Agency 2008). Domestic migration into MSAs from other U.S. locales contributes to higher levels of carbon monoxide, ozone, and the air pollution index, net of other factors. Similarly, natural population growth contributes to increased ozone, nitrogen dioxide, sulfur dioxide, and air pollution index levels. MSAs with larger overall populations in general tend to have greater air pollution (i.e., ground-level ozone, nitrogen dioxide, particulate matter 2.5, and pollution index levels). Poverty rates are linked to higher air pollution levels for nearly all of the dependent measures (excluding nitrogen dioxide and sulfur dioxide). MSAs with larger shares of manufacturing jobs and greater percentages of residents that drive alone to work tend to have higher air pollution levels. Additionally, we find that higher average temperatures are linked to lower levels of air pollution, and precipitation rates contribute to higher levels of air pollution (except for particulate matter 10).
Table 4

OLS multivariate regression models of air pollution

 

Carbon monoxide

Partic. matter 10

Ozone

Nitrogen dioxide

Sulfur dioxide

Partic. matter 2.5

Air pollution index

b

β

b

β

b

β

b

β

b

β

b

β

b

β

Immigration

8.883

0.180

0.031

0.024

0.018

0.034

−0.103

−0.375*

−0.025

−0.178

−0.010

−0.059

−0.008

−0.136

Domestic migration

4.137

0.326*

0.016

0.044

0.029

0.221*

0.006

0.080

0.004

0.126

0.000

0.011

0.003

0.196**

Natural population growth

5.831

0.237

0.132

0.208

0.074

0.260*

0.079

0.568***

−0.036

−0.411**

0.001

0.011

0.007

0.243**

Average annual temperature

−44.359

−0.483*

0.077

0.036

−0.025

−0.027

−0.003

−0.005

−0.035

−0.156

−0.052

−0.183*

−0.017

−0.171*

Average annual precipitation

8.741

0.164

−0.479

−0.343*

0.120

0.198*

−0.041

−0.141

0.013

0.075

0.073

0.400***

0.009

0.137

Total MSA population

5.25E−05

0.141

8.41E−07

0.078

1.67E−06

0.353***

9.19E−07

0.480**

6.56E−08

0.062

4.63E−07

0.268***

2.86E−07

0.459***

Population density

1313.328

0.160

−0.920

−0.004

2.705

0.028

5.316

0.087

2.170

0.090

3.657

0.106

−0.443

−0.037

% Driving alone to work

−0.034

0.000

1.265

0.261

0.820

0.418***

−0.307

−0.248

−0.025

−0.046

0.084

0.131

0.054

0.239**

% Workers in manufacturing

9.805

0.057

0.364

0.084

0.012

0.009

0.244

0.253

0.121

0.288*

0.156

0.364***

0.041

0.274**

% Population in poverty

67.574

0.317*

1.630

0.313*

0.890

0.366***

−0.220

−0.177

0.120

0.176

0.277

0.432***

0.094

0.426***

Constant R-Square

2791.671

0.217

−65.756

0.297

−9.998

0.335

34.852

0.599*

5.221

0.415

−1.184

0.499

−5.811

0.346***

N

73

 

91

 

128

 

52

 

74

 

160

 

183

 

P < 0.05, ** P < 0.01, *** P < 0.001

We turn now to the central objective of our study—assessing the relationship between immigration and air pollution, net of controls for other forms of population growth and for structural and environmental characteristics of MSAs. Based on the multiple regression results shown in Table 4, we find that immigration generally has null or trivial effects on MSA levels of air pollution. Immigration is not significantly associated with five of the six specific pollutants examined (carbon monoxide, ground-level ozone, sulfur dioxide, and particulate matter 10 and 2.5) or with the air pollution index. In fact, we find only one significant effect of immigration on air pollution across all seven models—a negative effect on nitrogen dioxide (b = −0.103, P < 0.05), indicating that MSAs with higher immigration rates actually have lower nitrogen dioxide levels. In sum, we find no evidence that immigration contributes to increased air pollution as argued by the population pressure or social disorganization perspectives. Instead, our findings are more consistent with the ecological footprint and community resource perspectives, indicating that immigration has no notable impact (or possibly some pollution-reducing effects) on MSA air pollution levels.

Supplemental Analyses

In order to assess the validity and robustness of these results and to further exhaust the data, we conducted an extensive series of supplemental analyses. First, we estimated a series of reduced and full models for each pollutant to account for potential indirect effects of immigration on air pollution that might work through poverty levels, population size, and other control variables. To accomplish this, we regressed each pollutant on our immigration measure with a reduced set of controls (domestic migration, natural growth, average precipitation, and temperature) and then sequentially added the remaining controls to identify any changes in the immigration-pollution relationships. The results of this analysis closely mirrored the findings reported above and revealed no significant effects of immigration on air pollution in either the reduced or full models for any of the pollutants examined.8

Second, we replicated each of our full models using time-lagged dependent variables that measure change in air pollution levels from 2000 to 2006 (rather than the 2006 measures used in the earlier analysis). In general, time-lagged models are advantageous because they are somewhat better able to account for potential selection effects of immigration on pollution compared to cross-sectional models, which are unable to determine whether immigration shapes pollution levels of MSAs or whether immigrants are simply funneled into places with higher/lower pollution levels (for further discussion on the spatial association between environmental equity of risk and immigrant neighborhood concentration see Crowder and Downey 2010; Hunter 2000b). In addition, time-lagged models provide a more conservative assessment of effects on air pollution, and thus are even less likely to produce significance in immigration-pollution relationships compared to cross-sectional analyses. That is, immigration and other predictors reach significance in the time lagged models only if they are associated with significant upward or downward changes in baseline air pollution levels between 2000 and 2006. We estimated the time-lagged models by adding controls for air pollution levels in 2000 to each model predicting 2006 pollution levels. Findings from the supplemental models mirrored our earlier results and again showed no significant immigration-pollution relationships for any of the dependent measures examined.9

Last, we replicated our multivariate analyses using two alternative estimation procedures established in prior research on immigration and pollution as effective means for addressing potential bias due to heteroskedasticity and outliers: quantile regression and multivariate models with robust standard errors (Squalli 2009, 2010; also see Hamilton 2009). Findings from these alternative models were nearly identical to those reported in our main analysis, again indicating that immigration flows do not contribute to increased MSA levels of air pollution.

Discussion and Conclusions

This study addresses competing theoretical perspectives about immigration-environment relationships by providing one of the first examinations of the effects of immigration flows on local urban air pollution levels. The results of our primary and supplemental analyses clearly indicate that immigration does not contribute to increased air pollution, either directly or indirectly, for any of the pollutants examined. Six out of the seven models revealed no significant effect of international migration on air pollution (carbon monoxide, ozone, sulfur dioxide, particulate matter 10 and 2.5, and the pollution index). Additionally, one model linked immigration to lower nitrogen dioxide levels. These results remained consistent throughout supplemental analyses (testing for mediating and time-lagged effects and using alternative estimation procedures). Thus, in contrast to popular sentiment and arguments drawn from population pressure/social disorganization perspectives, immigration does not appear to have harmful effects on local air quality levels. Instead, local population growth resulting from immigration appears to be generally unrelated to air pollution levels across MSAs.

At the same time, other forms of population growth appear to substantially increase air pollution. Our results indicate that domestic migration is linked to higher levels of carbon monoxide, ground-level ozone, and the air pollution index. Similarly, natural population growth is related to higher rates of ground-level ozone, nitrogen dioxide, and the air pollution index. Thus, it appears that population growth from immigration does not have the same pollution-generating effects that accompany other forms of growth.

These findings hold noteworthy implications for competing theoretical arguments about the environmental impact of immigration. Our results conflict with arguments drawn from population pressure and social disorganization positions, which suggest that immigration contributes to environmental harm by increasing population pressures on local ecosystems and by destabilizing communities in ways that inhibit collective action for addressing social problems like pollution. Rather, the null effects of immigration on air pollution observed here are more consistent with the ecological footprint and community resource positions, which suggest that immigration may not be harmful to (and may even be supportive of) local environments and communities. As the ecological footprint perspective suggests and as illustrated here, the different forms of population growth are likely to have complex and possibly quite different impacts on local communities and ecosystems. Based on this position, it is not altogether surprising that population growth from immigration appears to be less taxing on the environment than domestic migration and natural population growth, perhaps due to differences in lifestyles of the foreign- and U.S.-born or the potential stabilizing effects of immigration.

These findings also hold important implications for research in environmental sociology, political ecology, and environmental demography by extending our understanding of the structural determinants of air pollution. Urbanization, suburbanization, and population pressure are all factors that have been found to exacerbate air pollution (Cole and Neumayer 2004; Cramer 1998, 2002; Dietz and Rosa 1997; Shi 2003). Yet outside of a handful of noteworthy studies (e.g., Cramer 1998; Squalli 2009, 2010), the proposed relationship between immigration and environmental harm in the U.S. has been largely theoretical. Our research addresses this empirical gap and extends current understandings of the environmental impact of population movement and growth by examining contemporary immigration flows and changes and statistically assessing their effects on MSA-levels of air pollution.

In addition, our results provide key political implications for law makers and social scientists interested in the social consequences of contemporary immigration flows and whether these demographic shifts are harmful to local communities. In terms of air pollution, our study clearly suggests that immigrants, at least in the short run, are not contributing to increased environmental harm, which raises serious questions about the validity of claims concerning the “environmental threat” of immigration and calls to reduce immigration in order to curtail environmental harm (Population-Environment Balance 1992, p. 306; see reviews in Kraly 1995, 1998; Stevens 2010). Although additional research that builds on this analysis is needed to more thoroughly explore immigration-environment relationships, our study suggests that critical commentary and policies calling for reduced immigration on the grounds that it harms the environment may not be in line with social realities.

To further probe immigration-environment relationships, future research should extend this analysis in several ways. Although we find no link between immigration and air pollution, researchers should build on this analysis and broaden the scope of this work to assess whether immigration is related to other forms of local environmental harm and stress, such as water and soil contamination, waste production, or energy consumption. Research is also needed that addresses the impact of immigration on rural environments and on environmental harm at a global level. For example, even if immigration were to put pressure on local U.S. environments, these local strains may be offset on a global level if outward migration leads to reduced population pressures and environmental harm in other countries (Ehrlich and Holdren 1971; see Kraly 1995; White 2007).

Further research is needed that moves beyond cross-sectional measures of immigration (percent foreign-born in a population) to assess how changes in immigration are related to environmental harm. Our analysis provides initial evidence on the environmental impact of short-term immigration flows (2000–2006). However, long-term longitudinal analyses are needed to assess any possible delayed effects of immigration on the environment that could emerge, for example through higher fertility rates of immigrants (Carter 2000; Johnson and Lichter 2008) or as immigrant families adopt more “Americanized” lifestyles and ecological footprints (Hunter 2000a). In light of our findings, there is also an urgent need for further research that assesses the independent environmental impacts of the different forms of population growth (immigration effects net of the effects of natural growth and domestic migration). Last, while it is beyond the scope of the current project, research is needed to identify the potentially different environmental impacts of immigration flows originating from developed countries with large ecological footprints (e.g., Australia, Canada, Western Europe) versus from countries with reduced consumption and waste patterns (e.g., Mexico, Central and South America) (White 2007). In sum, although there is an urgent need for further research that explores the different nuances of immigration-environment relationships identified above, our study offers an important primary step in extending current understandings about how immigration shapes—or in this case fails to shape—local levels of air pollution and environmental harm.

Footnotes
1

The IPAT or I = PAT equation proposed by Ehrlich and Holdren (1971) has been widely used among environmental and population social scientists to assess the impact of humans on the environment (or I). This impact is the product of the number of people (P), the amount of goods consumed per person (A), and the pollution generated by technology per good consumed (T). The STIRPAT model, built on the IPAT model, adds more sophistication and allows social scientists the ability to test hypotheses.

 
2

MSAs offer several advantages as study units for our analysis. First, they provide an alternative and relatively untapped spatial unit for assessing the robustness of immigration-pollution findings from prior research. Second, they provide larger sample sizes and greater statistical power for identifying potential immigration-pollution links compared to state-level analyses (see Squalli 2010). Last, MSAs provide greater internal uniformity or “spatial homogeneity” than more-highly aggregated study units (counties, states). As a result, they are less prone to statistical noise from internal heterogeneity and are more likely to capture community-level processes described in social disorganization and community resource positions (see Feldmeyer 2009; Feldmeyer and Steffensmeier 2009; Peterson and Krivo 2005).

 
3

Following standard procedures of social science research, we used principal components analysis to combine information from the six specific pollutants into our air pollution index with regression based factor scores. All six pollutants were loaded onto one factor (Eigenvalue = 1.671). Factor loadings for the specific pollutants are as follows: CO (0.310), NO2 (0.409), O3 (0.732), SO2 (0.307), PM10 (0.501), PM2.5 (0.725). Findings were nearly identical when pollutants were loaded onto two separate latent factors in supplemental analyses.

 
4

Our sample excludes MSAs in Puerto Rico. Several MSAs that recorded/reported no air quality information were also excluded from all analyses. It is unknown whether these MSAs (or MSAs that record only one or a few types of pollution) differ discernibly from those that report multiple air quality measures. Although we cannot fully address this caveat with the available data, we conducted several supplemental analyses to adjust for missing data using multiple imputation techniques and using dummy variables to control for the number of pollutants reported for each MSA. Results from the supplemental analysis were substantively similar to those reported here.

 
5

Although we controlled for percent workers in manufacturing, we did not control for overall economic growth of metropolitan areas. Economic growth is likely connected to increased air pollution (through an increased volume of cars and factories) as well as increased domestic and international migration (from job seekers). However, this omission seems unlikely to bias our main finding that immigration has null effects on air pollution. First, economic growth is closely associated with domestic migration levels, which we control for. Thus, adding measures of general economic growth in addition to our controls for employment sector, domestic migration, and total population growth is not likely to provide much further explanatory power and creates a substantially higher risk of multicollinearity. Second, given that our immigration effects on air pollution are null, it seems unlikely that an additional control for economic growth would drive them to significance (omitted variables pose a greater threat for Type 1 error, but are less likely to produce Type 2 errors). However, further attention to the potential connections between these concepts is warranted in future research.

 
6

In addition to collinearity tests, we conducted an extensive series of regression diagnostics and supplemental analyses to account for potential violations of Gauss–Markov assumptions. First, we looked at Cook’s D and DFFIT values to identify potential outliers. After identifying and removing outliers, we replicated all analyses. Findings from this analysis were nearly identical to those presented here. Second, we conducted Breusch-Pagan/Cook-Weisberg tests to assess heteroskedasticity. Based on these results, we performed several supplemental analyses using alternative estimation procedures commonly used in prior research (quantile regression and multivariate linear models with robust standard errors) to account for potential bias from outliers and heteroskedasticity (see Squalli 2009, 2010). Results of these models were substantively similar to those described here. We return to the results of these models in further detail in our discussion of findings.

 
7

Although ozone levels were greater than other pollutants, several MSAs also had PM2.5 and PM10 levels that exceeded EPA standards. Carbon monoxide, sulfur dioxide, and nitrogen dioxide were below EPA standards in all MSAs examined. It is also important to note that while MSAs with low air pollution levels tended to be low for all pollutants examined, MSAs often had high pollution levels for one or two pollutants but not all seven measures.

 
8

It is worth noting that immigration effects on nitrogen dioxide were negative but non-significant in reduced models and only reached significance in the full model with all controls.

 
9

Our presentation and discussion of findings focuses on the cross-sectional results over the time-lagged models for several reasons. First, the findings are remarkably similar using both methods. Second, as we noted earlier, the time-lagged models have much higher thresholds for finding significance compared to the cross-sectional models. Thus, the fact that immigration had consistent null effects on air pollution—even in the cross-sectional models where significance is easier to obtain—more clearly illustrates the overwhelming absence or “nullness” of immigration-pollution relationships.

 

Acknowledgments

The authors would like to thank R. Scott Frey, Stephanie A. Bohon, and Meghan E. Conley for their useful comments to earlier versions of this manuscript.

Copyright information

© Springer Science+Business Media B.V. 2011