Determinants of county migrant regularization policymaking in the United States: Understanding temporal and spatial realities

Abstract

While immigration policymaking has traditionally been the sole prerogative of nation states, research has documented increased instances of migration policymaking at sub-national levels across migrant-receiving societies. This paper examines the temporally and spatially distinctive dynamics that underscore the adoption of these policies at the county level in the United States. The study considers the implementation of migrant labor market regularizations (LRs) for the time period 2004–2014. LRs are defined as discrete arenas of policymaking at the sub-national level that affect aspects of migrant workers’ status in labor markets and include laws and ordinances related to: anti-solicitation, language access, local enforcement of federal immigration law, and employment verification. Utilizing a multilevel event histories model, we analyze data from a unique dataset of over 5000 LR policies across 2959 counties in the United States, and address two research questions: (1) What are the social, economic, and political factors that influence the adoption of LRs by counties and municipalities in the United States; and (2) do policy adoption trends that occurred during 2004–2014 indicate a unique type of diffusion pattern? We find that the adoption of LRs by county governments are influenced by the racialization of immigration discourse and by policy behaviors at the municipal and state government levels, while economic characteristics of the local labor market and perceived ethnic competition from migrants have little direct impact on the probability of policy adoption.

Keywords

Immigrant workers socioeconomic integration socioeconomic change immigrant regularization labor markets

Migrant regularization—the conditional provision of legal status to irregular migrants in a host country—has become a contentious area of policy debate in post-industrial countries (Duman, 2014; Papdemetriou, 2004: Visser, 2017). In recent decades, growth in the demand for cheap labor pools, few legal pathways available for low-skilled migration, and an inability of nation states to maintain effective border controls have influenced growth in the number of undocumented migrants residing in migrant-receiving societies (Dauvergne, 2008; Engbersen et al., 2001). The presence of a growing undocumented migrant population presents challenges to a nation state’s capacity to effectively govern due to the everyday social practices of this population. Such practices render undocumented migrants simultaneously visible (through interactions with state actors, consumption of public resources, and involvement in communities), yet legally invisible within the regulatory gaze of the state due to their undocumented status (Bibler Coutin, 2005; Visser, 2017). Existing research on migrant regularization has focused on the conditions under which regularization programs are implemented in migrant-receiving societies, as well as the legal and institutional aspects of regularization policies (Apap et al., 2000). Scholarship has also sought to explore the efficacy of regularization programs and the role of social movements in influencing their adoption at the nation state level (Levinson, 2005; Papademetriou, 2004). At the same time, an emerging area of research has documented the increased level of policy activity related to immigration implemented by local and regional governments in migrant-receiving societies (Papademetriou, 2004; Walker and Leitner, 2011).

Part of a broader discovery of the localization of policymaking, this research observes that local governments—rather than national governments—are becoming more active players in managing migrant integration, service provision, and immigration policing and control (Coleman, 2007; Levison, 2005; Papdemetriou, 2004; Varsanyi, 2008; Visser, 2017). Such local policy actions seek to regulate socioeconomic opportunities afforded to migrants in ways that depart from or seek to strengthen the existing laws and policies on immigration within specific physical geographies. Visser (2017) notes that these include a growing number of local migrant labor market regularizations (LRs), which seek to influence the economic citizenship of irregular migrants. Such LRs promote or restrict the labor market and economic integration of irregular migrants, and are reflective of a growing political determination to influence migrant socioeconomic opportunities within specific local contexts. This emerging body of work has articulated new ways in which the devolution of immigration policy has pushed the border inward and rescaled conventional political geographies that regulate immigration, particularly in the United States (Coleman 2007; Varsanyi, 2008).

Yet, while existing research has explored the growth of local immigration policymaking and its potential to influence the social integration of immigrants in receiving countries, factors that influence the adoption and diffusion of local migrant regularizations remain understudied. Such a gap in the knowledge base is significant, given that scholarship has argued that local immigration policies are both an outcome of trends in policy devolution and of grassroots responses to undocumented migrants (Papademetriou, 2004; Varsanyi, 2008; Wells, 2004). Indeed, debates surrounding immigration combine with unique local processes of social positioning (i.e., racialization and criminalization) to embed migrants at the margins of prevailing state, market, and social structures in receiving societies (Goldring and Landolt, 2013). As a result, understanding why local governments are active in local immigrant policymaking activities requires controlling for complex multi-scalar social, economic, and political processes that shape migrant economic opportunities as well as their unique temporal and spatial contexts.

This article provides an important contribution to the literature by examining factors that influence the adoption of migrant regularization policies at sub-national levels of government in the United States. Specifically, the article seeks to consider the social, political, economic, and spatial and temporal factors that influence the adoption of local migrant LRs¹ by county-level governments in the United States. An analysis of this nature is important for scholarly and policy debates for at least two reasons. First, existing studies on local immigration policies have considered demographic and socioeconomic factors that influence the adoption of such policies (Walker and Leitner, 2011), but have broadly overlooked the spatial and temporal realities that influence the adoption of local immigration policies. Articulating these realities can help identify and understand those factors that consistently influence the passage of LR policies over time.

Second, a focus on county-level policymaking helps bridge a geographical gap in the US literature. Previous studies in the literature have considered select municipal, state, and national policies,² but analyses have yet to be conducted at the county level and on a national scale. This is a significant gap in the literature, given that most of the prominent local immigration policies in the United States have been implemented at the county level (Farrish and Hollman, 2017; Walker and Lichtner, 2011). County governments are the largest political subdivision within individual states in the United States and function primarily to administer state laws. In many parts of the country, counties are the most relevant sub-national political units and serve as an important intermediary between municipalities and states. Given these realities, examining county-level policies can promote a more nuanced understanding of the multi-scalar nature of migrant regularizations (Visser, 2017). The goal of this study, therefore, is to help build theory related to the temporal and spatial factors that underscore the adoption of local migrant LRs, and the variegated nature of migrant regularization in contemporary migrant-receiving states.

We begin by reviewing the literature on migrant regularization and the localization of migration policy in the United States to advance a theoretical and analytical framework upon which our analysis is based. We then use data from the United States Local Migrant Regularizations Database (Visser, 2017) to consider the social, political, economic, spatial, and temporal factors that underscore the adoption of local migrant LRs across US counties. We undertake a multilevel discrete-time event history model to explore the likelihood that counties adopt LRs and the changes in the likelihood of policy adoption over the time period 2004–2014. We find LR policy adoption by county governments is influenced by a unique racialization of LR policymaking linked to the presence of large Latino populations and by policy behavior at the municipal and state government levels. We also find that economic and labor market conditions have little direct impact on the probability of a county government adopting an LR. We conclude by discussing the implications our findings present for scholarship and policy discussions surrounding migrant regularizations in the United States specifically, and post-industrial societies more generally.

Local migrant labor market regularizations and the United States context

Migrant-receiving societies have generally employed three types of policy strategies for dealing with irregular migration: deportation, toleration, or regularization (Duman, 2014). Since 2000, the majority of migrant-receiving nation states have adopted some form of regularization policy (Levinson, 2005). Regularization implies the conditional provision of legal status to irregular migrants in a host country and regularization policies are generally dependent upon conditions of employment, although some are based on humanitarian premises. Regularization policies generally fall under the purview of national governments, and existing research on migrant regularization focuses almost exclusively on national-level policies such as quota or visa programs. However, an emerging body of research has also documented increased instances of migration policymaking at sub-national levels of government across migrant-receiving nation states (i.e., municipalities)—most notably in the United States (Varsanyi, 2008; Visser, 2017; Walker and Leitner, 2011; Wells, 2004).

Increased activity by sub-national governments in the United States on issues of immigration has occurred within the context of changes in labor market demographics and federal immigration laws. Over the last 30 years, the US labor market has experienced dramatic demographic change. In 2014, the Bureau of Labor Statistics estimated that 16.4% of the US labor force was foreign-born—including legally admitted migrants, refugees, temporary residents, and undocumented migrants (Cordero‐Guzmán, 2015; Visser and Meléndez, 2015). These demographic changes in the labor market have given rise to increased tensions surrounding the presence—or perceived presence—of irregular migrants and their impact on economic conditions in local labor markets—particularly as employment security has fallen for US-born workers (Kallenberg, 2011; Peck and Theodore, 2012; Visser, 2017: Visser and Cordero-Guzmán, 2015).

This contentious discourse has been further influenced by the inability of the federal government to pass comprehensive immigration reform, and has shifted the burden of responding to irregular migration onto local, county, and state governments. This shift in immigration governance is conceptualized as the combined result of the multilayered structure of the American nation state, the diffusion of powers between various levels of governments, and continued devolution of social responsibilities under the Welfare Reform Act (Varsanyi, 2008; Visser, 2017; Walker and Leitner, 2011). Such realities have created an environment of devolved policy governance, where the federal government retains the right to regulate the entry and exit of migrants in the United States but there is an increased role for local governments in enforcing these federal statues (through cooperation and coordination with local authorities). This devolution of governance has also occurred alongside the devolution of social and civil rights, as social policy decisions are now increasingly made at sub-national levels (Brenner and Theodore, 2002). Together, these processes of devolution have generated the creation of new policy domains that seek to regulate migrant place-making, socioeconomic activities, and civil rights. These policy domains include “inclusive” policies such as “sanctuary” initiatives that stipulate local authorities will not check residents’ immigration status or resolutions in support of the rights of undocumented residents. These also include “exclusionary” policies that enforce housing codes to target overcrowding by immigrant workers, place restrictions on unlicensed day labor markets, or fine businesses that employ workers without proof of legal residence (Rodriguez, 2008; Varsanyi, 2008). The rise of these policy domains has resulted in a variegated landscape of immigration policy governance in which socioeconomic opportunities and rights afforded to migrant communities often vary from locality to locality within the same state, region, and across the nation (Walker and Leitner, 2011; Wells, 2004).

Local migrant LRs are a particular policy domain within this increasingly variegated landscape of immigration governance. Visser (2017) defines LRs as policies which seek to regulate economic and employment opportunities in ways that depart from or strengthen existing national laws and policies on the economic activity of immigrants within specific geographic areas. In this view LRs are policies that aim to manage the economic citizenship of irregular migrants by promoting or restricting the integration of irregular migrants into the labor market and, thus, “regularize” the migrant labor market. These include policies aimed at influencing opportunities for irregular migrants to obtain identification cards, to work legally as independent contractors, as well as to receive employment and training materials in their primary language. LRs are a particularly active policy domain for county-level governments in the United States and since 2005, 2959 out of the total 3007 counties in the country have implemented some type of LR (Visser, 2017).

Research surrounding factors that influence the implementation of local immigration policies broadly in the United States have identified a variety of reasons for why local governments engage in immigration policymaking. Many scholars conceptualize the passing of immigration-related policies by sub-national governments as an outcome of the grassroots response to the presence of a growing number of immigrants (Ellis and Goodwin-White, 2006; Wells, 2004). This body of work links the localization of immigration politics directly to the geography of immigration—suggesting that local immigration policies are enacted where there are large immigrant populations. Growth in the population of immigrants in a particular area has also been argued to influence the types of local and state policies (i.e., whether integrative immigrant measures or exclusionary measures are adopted), correlating that increased growth in immigrant populations can lead to the passage of anti-immigrant measures—resembling the contact and threat hypotheses of sociocultural contact (McClain et al., 2006; Pettigrew and Tropp, 2006). Other studies have argued that the rate of growth in the immigrant population is what matters—not the concentration of migrants (Esbenshade, 2007). However, as Walker and Leitner (2011) note, these analyses have generally failed to account for how specific characteristics of areas may influence the type of policy response enacted, such as socioeconomic characteristics or attitudes of local residents.

The socioeconomic characteristics and cultural attitudes of particular localities (e.g., political persuasion, educational attainment, violent crime) have been suggested as factors that influence the immigration policymaking activities of sub-national governments. In the United States, political persuasion is generally understood in the context of Democrat versus Republican, with Republican or “conservative” areas more active on issues of immigration control and enforcement as they tend to favor local and state rights over federal government intervention (Ramakrishnan and Wong, 2007; Steil and Vasi, 2014; Walker and Leitner, 2011). In addition, popular commentary in the United States claims a link between immigration and increased violent crime. Such commentary has become common in local policy debates—despite immigrant criminality and victimization research which has demonstrated immigrants typically engage in less crime than their native-born counterparts (Sohoni and Sohoni, 2014; Zatz and Smith, 2012) and research which has suggested that at the aggregate level immigration does not increase crime rates (Reid et al., 2005).

Moreover, studies have found that the level of educational attainment by residents in a given locality may influence the passing of immigration policy (Walker and Leitner, 2011; Wilkes et al., 2008). This literature has made a strong determination that holders of university degrees are less likely to have negative perceptions of immigrants (Haubert and Fussell, 2006), whereas those individuals without college degrees tend to view immigration unfavorably (Chandler and Tsai, 2001; Pantoja, 2006; Wilkes et al., 2008). Research suggests that the relationship between education and attitudes toward immigration is rooted in the notion of economic competition, and argues that more educated workers are less likely to face competition from immigrant workers, while lower-educated workers are more likely to oppose immigration due to a reality that they are more likely to have to compete with low-skilled migration flows (Kessler, 2001; Mayda, 2006). Yet, research has found no significant relationship between personal economic circumstances and attitudes toward immigration.

Visser (2017) suggests that LRs are likely related to real or perceived “ethnic competition” in the labor market. In this sense, limited job opportunities in local labor markets breed conflict over real or perceived growth of undocumented immigrants in the labor market. Recent research related to new immigrant-gateway literature further suggests there is a particular relationship between larger Latino/Hispanic populations in local areas and the passing of local immigration policies. For example, Shohoni and Green (2008) find that the passing of 287(g) agreements is more prevalent in areas where the rate of growth of the Hispanic population outpaces that of the United States as a whole. Scholarship in ethnic and racial studies has further found that, in the United States, political and social discourses surrounding immigration often link undocumented immigration with the Latino/Hispanic population due to the proximity of Latin America to the US border and the history of irregular migrant labor forces in agriculture industries and sectors (Chavez, 2013; Mehan, 1997). Research also suggests that economic competition at the bottom of the labor market is increasing due to a general tendency of labor markets in the United States to trend toward job polarization and income inequality. Latino and immigrant workers are overwhelmingly concentrated in low-wage jobs and sectors of the labor market, and many US-born workers who have been impacted by the loss of jobs in the middle of the wage and skill structure are experiencing increasing employment insecurity, which brings them into real or imagined competition with these workers for limited jobs in the labor market (Visser, 2017; Visser and Meléndez, 2015). These co-occurring realities are important, given that studies suggest that even the perception of increased economic insecurity due to immigration influences public opinion about immigration policies, regardless of whether actual economic or employment competition is occurring (Pantoja, 2006; Wilkes et al., 2008).

In a study of immigration policy activities of 174 municipalities in the United States, Walker and Leitner (2011) highlight how geography matters in the formulation of local immigration policymaking activities. Their analysis sought to understand how local immigration policy activity is the result of the unique social/power relations surrounding issues of race, as well as notions of nation and place, and to account for the impact of these processes by controlling for spatial autocorrelation. Yet, their study considered immigration policies of only 174 (primarily urban) municipalities that were enacted prior to 2009, while local immigration policymaking by sub-national governments in the United States increased substantially after 2008. In addition, the analysis does not account for the temporal realities that influence local contexts, nor does it account for the impact of the complex multi-scalar immigration policy governance structure in the United States. As such, while Walker and Leitner’s study highlights the importance of geography in understanding the adoption of immigration policies, the nature of the study makes it difficult to fully understand the scalar and temporal realities that influence sub-national governments’ engagement in immigration policymaking and how these may mitigate or augment local contexts. In contrast, as elaborated below, our analyses controls for both temporal effects and the multi-scalar policy governance structure that surrounds local immigration policy in the United States.

Accounting for the multi-scalar immigration policy governance structure in the United States is particularly important. Legal and political institutional scholars tend to characterize growth of local immigration policies within the notion of a “politics of scale” that results from the struggle between different tiers of government over rights and responsibilities related to irregular migration (Ellis and Goodwin-White, 2006; Varsanyi, 2008). In this sense, local and state governments implement immigration policies in response to policy activities by other similar or lower-/higher-level jurisdictions. The suggestion that policy activity by one political jurisdiction may influence policy activities of another has been widely confirmed by policy diffusion studies. This work has shown that, even when controlling for demographic and socioeconomic characteristics, the adoption of a policy by a specific political jurisdiction is largely influenced by policy experiences or activities of other jurisdictions (Berry and Baybeck, 2005; Givel and Glantz, 2001; Shipan and Volden, 2008). For example, policy adoption may be undertaken as a preemptive measure by higher-level jurisdictions to disallow lower-level government actions that are contrary to state or federal law; or policies may be passed as defensive mechanisms to prevent potential negative spillover effects of policies enacted by political jurisdictions of similar levels.

Such a “politics of scale” may be particularly important in understanding the adoption of LRs. Visser (2017) notes that LRs are best understood as a mode of social regulation in the migrant labor market and are generated within the intersection of the localization of immigration policy and the local production–reproduction dialect of the local economy. In this sense, the migrant labor market, like any labor market, is created through unique spatial and scalar processes that result in its institutional embeddedness and geographic variation. Peck (1996) has emphasized that labor markets are formed by and experienced within distinct local production–reproduction dialects that link processes of production and social reproduction to regulatory dialects (i.e., laws and policies) that are developed, implemented, and experienced locally. These processes are situated and constructed as responses to broader international and national contexts, which can lead to uneven development (Herod, 2012; Massey, 1984). Therefore, policies enacted to regulate economic activities for migrant workers in other political jurisdictions and at other scales of government produce different effects and give rise to different policy behaviors in different places. For example, a county may adopt an LR as a way to circumvent federal or state laws that could impact immigrant workers in their local economy, or a county may adopt an LR that will disallow any municipality from enacting an LR contrary to state or federal law out of concern regarding spillover effects from cities working in contrast to federal or state law (such as an influx/exodus of migrant workers to the county or cuts to state and federal funding). The literatures surrounding the localization of immigration policymaking suggests multiple factors that may influence whether or not localities may enact an LR. However, the empirical record remains largely qualitative, descriptive, or limited to a small number of jurisdictions. Rigorously assessing factors that influence the passing of LR policies by local governments while also accounting for the temporal realities and the multi-scalar context in which such policies are enacted can offer insight into why sub-national governments engage in passing local migrant LRs. Such an analysis provides a strong contribution to policy and scholarly debates and discussions.

Data and methods

This study examines when, and under what conditions, county governments in the United States adopt a local migrant LR by testing the validity of the hypotheses that emerge from the literature discussed above, and the effect of policy diffusion processes on the adoption of an LR by a county government. In undertaking this analysis we use data from the United States Local Migrant Regularizations Database (USLMRD) constructed by Visser (2017). The USLMRD includes information on four types of LR policies enacted at the local, county, and state levels across all 50 states and the District of Columbia for the time period 1983–2015. The types of LRs included in the dataset are: (1) language access laws; (2) anti-solicitation ordinances; (3) local enforcement of federal immigration laws; and (4) employment verification laws (see Appendix 1 for a detailed description of each type of ordinance). LR ordinances included in the USLMRD were identified through a systematic process beginning with information from a database created by the National Immigration Law Center of local laws limiting immigration enforcement, and adding to this list through an extensive archival research process. The original database was supplemented with information on policy ordinances collected by UNIDOS (formerly the National Council of La Raza) and LatinoJustice, as well as an extensive search of municipal code databases including municode.com, generalcode.com, codepublishing.com, and amlegal.com.

For each ordinance included in the dataset, information on the year the ordinance was adopted and/or repealed, and the nature of the ordinance and geographic area impacted are recorded. Each ordinance is also categorized as either “pro-migrant labor market regularization” or “anti-migrant labor market regularization” based on an understanding of the policy as favorable or unfavorable from the perspective of migrants’ ability to seek employment, enjoy legal protections, or experience mobility out of informal employment. The USMLRD includes information on 5528 ordinances in 3067 cities, counties, and states across the United States and is updated annually. For the purposes of this analysis, we restrict the analysis to policies passed by county governments, and include policies enacted between 2004 and 2014 to capture the changes that accompanied the increased lawmaking that occurred after 2006 related to migration regularizations—a period that remains relatively unexamined in the literature.

Data were organized into a long-format table in which each entry is a county-year with the policy state and explanatory variable for each county in a given year. Counties contribute to the table until they adopt an LR. Counties that never adopt a policy during the study period are said to be right-censored and have county-year entries for the entire period under study. Counties were excluded if they had passed an LR before the study period and did not have any policy activity to repeal or pass any new additional LR during the period under study. We only modeled results for unfavorable policies, as this included all but three LRs passed by counties during the period. As such, we are explicitly examining factors that influence the adoption of negative LRs at the county level. Of the 27,587 county-years included in the study, 2255 were excluded from the regression analysis due to missing data. The counties excluded were primarily counties with small populations, where demographic data (upon which the explanatory variables in the specified model were constructed) were unavailable.

To examine factors that influence the passing of LRs by county governments, we employ a multilevel discrete-time event history approach. Event histories (also known as survival analyses, duration analysis, or hazard modeling) describes a suite of methods originally developed within health research to analyze variation in mortality rates. In recent years, these methods have been expanded to examine a variety of health, social, and political events (Hosmer et al., 2008; Singer and Willett, 2003). While a simple linear/ordered regression model could be fitted to understand the types of factors that influence the passing of ordinances in the United States, controlling for place, such an approach would tell us nothing about the timing of the implementation of LRs nor how the factors that influence the adoption of an LR change over time—particularly given that many of the factors that have been suggested to impact the passing of LRs vary across time (e.g., population rate, unemployment, crime).

In this approach, the response variable is the hazard rate, which is a measure of the probability or odds of an event occurring in the preceding time period. Probability is calculated through binomial regressions and we use a logit function (although other links are possible). In event history analysis time can be measured continuously or in discrete intervals, and explanatory variables can be either fixed across the study period or be dynamic over time. One limitation of the event history analysis approach is the assumption of spatial independence between units. We address this limitation through the use of a multilevel structure in the model to examine the effect of a county being located in a particular state on the hazard rate. Multilevel models, also known as mixed-effect or random-effect models, allow for an analysis attentive to both between-group variation and within-group dependence (Artelaris, 2015; Austin, 2017; Kulu and Billari, 2006).

We use a two-level hierarchical model with counties as the unit (or lower-level units), and states as the group (or higher-level units). The model is specified as:

p t i = Pr (y t i = 1 | y t - 1, i = 0)

where the dependent variable (pti) is the hazard rate for each county (i) in a given year (t), and where (Pr) is the probability of a county government adopting an LR (y = 1) or not adopting an LR (y = 0). A logit link function was fit for a binary response model and was specified as:

log (p t i / 1 - p t i) = β 0 D t + β X t i

In the model, β0Dt is a vector of functions where D is defined as a series of dummy variables for each year, and β0 is the intercept. Xti is the vector of explanatory variables for each county in a given year, and β is the vector of coefficients.

To address within-group dependence we include the addition of random effects to the model, specified as:

log (p t i / 1 - p t i) = β 0 D t + β X t i j + (u 0 j + e i j)

Here, we use a two-level hierarchical model with counties (i) nested in states (j). The random parts of the function are the between-group variations (u0j) and the unexplained variations between hazard rates for counties within their states (eij). We further calculated annual hazard rates for each state, and calculated an annual Moran’s I for the state hazard rate. Moran’s I is a measure of spatial autocorrelation or spatial dependence that identifies patterns of clustering or dispersion in spatial data.

In the estimated model, the dependent variable of interest in the outcome equation is the probability of a county passing a negative LR. The model controls for factors thought to influence policy activity on LRs by county governments, including demographic (rates and concentrations of foreign-born, undocumented migration, and of the Latino/Hispanic populations); socioeconomic factors (violent crime rate, industrial base of regional economy, poverty, and unemployment); political factors (voting and politics of representation); and the LR policymaking of higher and lower government jurisdictions that may influence the passing of LRs by county governments. Explanatory variables were tested for covariance, and when levels were above an absolute value of 0.7 only the variable with the stronger relationship to the response variable was included in the model. Only the covariance of percentage Republican and percentage Democrat voter was this high (0.99). As a result, we included only the percentage Republican voter in the model. Explanatory variables were centered around their mean to increase interpretation and reduce collinearity, which prevents convergence of the multilevel model (Bell and Jones, 2015). The regression was calculated in the R statistical environment using the MCMCglmm package. Rather than maximum likelihood, MCMCglmm uses Markov chain Monte Carlo (MCMC) simulated priors to calculate a set of posterior estimates associated to the hazard rate.

Results and discussion

To understand variation and policy diffusion and adoption patterns of LRs by county governments in the United States, we calculate the annual hazard and survival rates to consider any geographic or temporal trends in the adoption of LRs by county governments.

Figure 1 shows the change in hazard rates over the time period of study, as well as demonstrates the odds change over time. As shown in Figure 1, there is a sharp increase in the adoption of LRs at the county level over the time period 2008–2012, with a sharp decrease in the adoption of LRs after 2012. The sharp increase during 2008–2012 may be due to the adoption of policies related to implementing the Secure Communities program by the federal government. The Secure Communities program formed a partnership among federal, state, and local law enforcement agencies to integrate databases and partnerships with local and state jailers to build domestic deportation capacity (United States Immigration and Customs Enforcement Office, 2014). The program was piloted in 14 jurisdictions in 2008 at the end of the George W. Bush Administration, but substantially expanded to 1210 jurisdictions under the Obama Administration in 2011. In 2012, a federal mandate expanded the program to all 3141 jurisdictions (state, county, and local jails and prisons) and further promoted the adoption of policies related to immigration at the municipal level, including those pertaining to sanctuary cities, English-language ordinances, and E-verify laws (Varsanyi, 2008; Visser, 2017). Given that most immigration policies enacted at the municipal (city) level during this time were adopted to counter or negate the effect of the federal mandate, it could be likely that county governments became active to mediate instances in which municipalities were seeking to enhance or counter state policies enacted at the time.

Figure 1.

National hazard rate (probability of any county adopting a policy in a given year) and survival rate (rate of counties that have not passed a policy).

While the descriptive analysis shows evidence of a distinct trend in policy adoption, there are also unique temporal and geographic shifts present in the intensity of LR adoption over the time period of study. Figure 2 maps the hazard rates in each year of the study at the state scale, and Figure 3 maps the Moran’s I z-score for each year. As shown in Figure 2, the probability of counties adopting LRs are close to zero in the first years of the time period. However, as the study period continues, states in the southeast, west, northeast, and finally the center of the country exhibit increased hazard rates, indicating growth in the adoption of LRs by county governments. Finally, by 2014 all but 14 states had all of their counties right-censored, meaning all counties had passed some type of LR. Figure 3 maps the Moran’s I z-score for each year, which indicate the significance of the spatial patterns. As observed in Figure 3, significant clusters of hazard move throughout the country across the time period of study, illustrating a unique geographic trend to the policy adoption of LRs.

Figure 2.

Hazard rate (probability of any county adopting a policy in a given year) for each state across the study period.

Figure 3.

Z-score local Moran’s I estimate, significant low or high hazard rate clusters.

To understand those factors that influence the adoption of LRs over the years of the study, we employ a multilevel discrete-time event history approach across the time period 2004–2014. We started by fitting a series of models to test the predictive power of each group of variables to understand their broad impacts. We compared the AIC (Akaike information criterion) and BIC (Bayesian information criterion), both relative measures of the goodness of model fit for a given dataset, where smaller numbers indicate a better fit. We found that the social factors were the strongest predictors (full model AIC: 8544.5, BIC: 8764.272; social AIC: 8804.1, BIC: 8926.15; economic AIC: 9428.3, BIC: 9558.493; political AIC: 8793.2, BIC: 8939.736). To understand relationships over the years of the study and across the country we employ a multilevel discrete-time event history approach across the time period 2004–2014. Table 1 shows the shift in values across the time period. As shown in Table 1, most of the explanatory factors are largely dynamic, indicating that given that these variables change over time, the multilevel discrete-time event history approach is appropriate for more accurately assessing those factors that influence the adoption of LRs at the county level. Table 1 shows the shift in values across the time period of study.

Table 1.

Changes in values of independent variables in analysis across the time period of the study.

Variable	Min.	Max.	Mean	First quartile	Median	Third quartile
2004
Total population	67	9,519,338	88,103	11,400	25,271	61,274
Population that is Latino (%)	0.00	99.69	8.36	0.86	1.81	5.56
Population that is in poverty (%)	6.41	89.66	35.41	27.57	34.25	41.35
Population with higher education (%)	4.37	61.41	14.96	10.30	13.32	17.44
Population that is foreign-born (%)	0.00	50.94	3.41	0.88	1.70	3.86
Total unemployed population (%)	0.00	41.42	6.19	3.99	5.37	7.06
Civilian unemployed population (%)	0.00	41.67	6.21	4.01	5.39	7.09
Workers in agricultural sector (%)	0.00	27.09	2.25	0.67	1.49	2.84
Workers in construction sector (%)	1.34	24.14	6.96	5.30	6.50	8.17
Workers in manufacturing sector (%)	0.00	36.01	11.01	6.42	10.18	14.86
Median household income ($)	9192	82578	34723	29149	33401	39073
Number of Latino elected public officials (state)	1	2075	260	2	7	64
Votes for Democratic candidate (%)	7.10	89.18	38.72	30.11	38.30	46.49
Votes for Republican candidate (%)	9.34	91.97	60.24	52.37	60.75	68.83
Violent crime rate (per 100k)	0	1587	136	50	107	188
Type of state policy	0	2	0	0	0	0
2014
Total population	73	1,259,828	46,445	9,529	17,137	34,462
Population that is Latino (%)	0.37	99.96	16.72	1.75	2.69	6.45
Population that is in poverty (%)	15.06	89.34	43.02	31.35	38.22	49.64
Population with higher education (%)	5.55	49.57	17.34	13.13	16.26	19.50
Population that is foreign-born (%)	0.00	32.38	2.76	0.90	1.65	2.83
Total unemployed population (%)	0.00	34.05	8.91	4.65	7.01	10.90
Civilian unemployed population (%)	0.00	34.18	9.04	4.76	7.14	11.08
Workers in agricultural sector (%)	0.00	15.61	2.47	0.90	2.00	3.26
Workers in construction sector (%)	1.19	18.96	6.31	4.75	5.87	7.23
Workers in manufacturing sector (%)	0.00	21.50	8.93	5.65	8.78	11.75
Median household income ($)	10550	100584	41167	33558	42430	49198
Number of Latino elected public officials (state)	1	2542	36	2	3	3
Votes for Democratic candidate (%)	0.55	93.39	40.30	28.98	39.67	49.48
Votes for Republican candidate (%)	5.98	96.15	57.22	48.61	58.19	67.61
Violent crime rate (per 100k)	0	587	118	46	95	182
Type of state policy	0	2	2	1	2	2

Table 2 presents the results of the multilevel discrete-time event history approach. The results of the regression analysis provide estimates of the relationship between the estimated hazard and the characteristics of the counties across all time periods. The coefficient column reports the relationship between a change in the odds of a county passing a negative LR and a change in the explanatory variable. The coefficients for each year are estimates of the global temporal effect on the odds. The table also reports the lower and upper bound estimates for the 95% confidence interval. We are interested in testing the set of interrelated hypotheses described above, which have emerged from the literature related to factors that influence the passing of immigration policy generally, and LRs specifically in the United States.

Table 2.

Results of the multilevel discrete-time event history approach assessing factors that influence the passage of LRs at the county level in the United States (2004–2014).

	Post mean	Lower bound 95% confidence interval	Upper bound 95% confidence interval	Eff. samp.	pMCMC
(Intercept)	−12.06	−13.79	−10.56	18.34	0.0004
timet2	−0.22	−0.69	0.49	3.90	0.5058
timet3	0.91	−0.13	2.13	1.40	0.1742
timet4	1.39	0.20	2.33	1.98	0.0027
timet5	2.28	0.94	3.53	1.69	0.0004
timet6	4.38	2.75	5.58	1.83	0.0004
timet7	8.92	7.14	10.23	1.91	0.0004
timet8	12.56	10.66	13.92	1.84	0.0004
timet9	16.40	14.36	17.82	2.10	0.0004
timet10	15.61	13.70	17.00	2.41	0.0004
timet11	1.73	0.84	2.76	8.95	0.0004
Total Population	0.50	0.45	0.55	59.64	0.0004
Foreign Born Pop	−2.52	−3.63	−1.45	105.58	0.0004
Undocumented	−0.09	−0.20	0.01	109.60	0.1147
Latino Population	1.27	0.80	1.74	141.20	0.0004
Percent of Econ Agriculture	4.00	2.34	5.79	134.23	0.0004
Percent of Econ Const	1.28	−0.06	2.96	116.65	0.0827
Percent of Econ Manu	−1.21	−2.11	−0.30	93.62	0.0107
Population in Poverty	0.77	0.27	1.32	113.94	0.0009
Violent Crime	0.00	0.00	0.00	99.56	0.1538
Higher Education	−1.36	−2.20	−0.55	93.10	0.0027
Republican Vote	−0.40	−0.76	−0.09	107.23	0.0204
StatePolicyType_factormix	−6.03	−8.48	−2.32	3.65	0.0004
StatePolicyType_factornone	−1.26	−2.42	2.33	2.86	0.2862
StatePolicyType_factorunfav	0.27	−0.84	3.26	3.18	0.7982
place_fav_cent	−1.01	−1.86	−0.21	125.72	0.0151
place_unfav_cent	1.50	0.56	2.37	101.37	0.0004
Latino Elected Officials	−0.48	−0.59	−0.39	98.77	0.0004
Unemployment	−5.40	−6.80	−3.97	80.88	0.0004

A substantial amount of scholarship related to the localization of policymaking suggests that population demographics and demographic change patterns influence the passing of LRs—particularly in relation to the population of immigrants or foreign-born populations in an area (Cain et al., 2000; Ellis and Goodwin-White, 2006; Pettigrew and Tropp, 2006; Steil and Vasi, 2014; Walker and Leitner, 2011; Wells, 2004). As shown in Table 2, the mean posterior coefficient for the total population variable is positive, which suggests that the likelihood of adopting an LR by a county government increases as the general population increases. However, as reported in Table 2, both percentage of foreign-born population in a county and the percentage of undocumented workers actually reduces the odds of a county adopting an LR. Yet, it is important to note that this negative result occurs at a lower level of confidence for the percentage of undocumented residents variable than for the percentage of foreign-born residents variable. This finding contradicts previous literature that argues that level of foreign-born population is a significant factor that encourages local governments to enact anti-immigrant initiatives (Esbenshade, 2007; Singer et al., 2009; Walker and Leitner, 2011). In contrast, our findings suggest that even with changes in rates over time, the foreign-born population is not a significant contributing factor influencing whether or not a county government passes an LR.

Given recent case study literature that has suggested there is a unique conflation in US public discourse with undocumented migration and Latinos (Chavez, 2013; Mehan, 1997), we include a variable that controls for the percentage Latino population in each county. Including this variable alongside the percentage of foreign-born and percentage of undocumented population allows us to understand to what extent there may be a “racialization” aspect to LR policy adoption. As shown in Table 2, the percentage Latino population in a county is positively related to the hazard, where a 1% increase in the Latino population is related to an increase in the mean hazard ratio by an estimated factor of 3.56.³ This positive influence on the hazard rate, when considered alongside the findings of the total foreign-born population and undocumented immigrant population, suggests that there is a distinct racialization to LR policy adoption. Specifically, these policies tend to target Latinos—a point that has been suggested by case study research undertaken in specific local contexts. Our analysis supports these findings at the aggregate level, and suggests that it is not necessarily the presence or growth of the foreign-born or undocumented immigrants that influence the adoption of LRs, but more specifically the growth of the general Latino population that significantly increases the likelihood of county governments in the United States passing LRs.

The results in Table 2 also show limited relationships between economic factors and the implementation of LRs. Research has suggested that LRs are passed when there is perceived “ethnic competition” from migrants to US-born workers in the labor market (Pantoja, 2006; Shohoni and Green, 2008; Visser, 2017; Wilkes et al., 2008). Of the variables included to control for economic and labor market factors in the model, only the percentage of the labor force in the agricultural sector, unemployment, and poverty were significant, at an estimated p-value at or below 0.001. A 1% increase in the share of agricultural labor was related to an increase in the odds of adopting an LR by a factor of 54.74. However, given the large confidence interval (lower bound 10.36 and upper bound 3272.32), the results suggest that there is a great deal of variety in the effect of agricultural workers. For example, in counties where agriculture (which in large part depends on undocumented and low-skill immigrant labor) may be a primary base of the county’s economy, it could be likely that these counties may be less likely to pass (exclusionary) LRs—given that this workforce is essential to the functioning of the region’s economy. This could explain the very large confidence interval reported for this variable. Interestingly, the percentage of a county’s labor force that is in construction and manufacturing is not significant, despite these sectors having seen a growing share of low-skill immigrant employment over the last two decades—for example, in manufacturing industries of rural areas (e.g., meat-processing plants), and in construction industries in urban areas (e.g., residential construction).

In addition, as reported in Table 2, there was a positive relationship between poverty rates and probability of passing an LR. For county-years where poverty levels were higher, the probability increased by a factor of 2.16 (lower bound 1.31, upper bound 3.75) when poverty was one unit higher. The results also suggest that higher rates of unemployment decrease the probability of a county passing an LR, but only by a very small factor of 0.005 (lower bound 0.001, upper bound 0.019 confidence interval estimates). Together, the results of the poverty variable and the unemployment variable are suggestive of the “ethnic competition” hypothesis in relation to passing LRs (Visser, 2017), but indicate that this “ethnic competition” is not solely located in the labor market context, but other socioeconomic contexts as well.

In the US literature, political persuasion is generally understood in the context of Democrat versus Republican, with Republican or “conservative” areas more active on issues of immigration control and enforcement as they tend to favor local and state rights over federal government intervention (Ramakrishnan and Wong, 2007; Walker and Leitner, 2011). In terms of political persuasion, the results suggest that the percentage of Republican voters in a county is not significant and the range of the confidence interval suggests an uncertain relationship between political persuasion and the adoption of LRs. Such a result does not directly counter previous research, which has suggested the influence of ideological conservatism on the passing of LRs, but does indicate that there may be mitigating political behaviors or factors that influence the impact of ideological conservatism on the passing of LRs at the county level. In terms of political behavior, the number of Latino elected officials is found to be negatively correlated with an increased probability of enacting an LR. Specifically, an increase by one elected Latino official reduces the odds of a county passing an LR by a factor of 0.62 (lower bound 0.56, upper bound 0.68).

In addition, violent crime does not show up as significant in the model. This suggests that although popular discourse in the US claims a distinct link between immigration and increased violent crime—particularly in local policy deliberations—over time this discourse does not seem to contribute significantly to the passing of LRs at the county level. In addition, the percentage of individuals in a county with a Bachelor’s degree or higher is negatively associated with the passing of LRs at the county level. This result supports existing studies which suggests that educational attainment levels by residents in a given area may influence immigration policy (Walker and Leitner, 2011; Wilkes et al., 2008), and specifically that areas with larger populations of university-educated individuals are likely to create more favorable contexts for immigrants (Chandler and Tsai, 2001; Pantoja, 2006).

Moreover, research related to the localization of immigration policymaking generally, and for LRs specifically, continues to point to the “politics of scale” surrounding LRs in the United States which results from the struggle between different tiers of government over rights and responsibilities related to irregular migration (Ellis and Goodwin-White, 2006; Varsanyi, 2008). We included three variables to control for the impact that policy behaviors by municipal and state governments had on the adoption of LRs by county governments in the United States. As shown in Table 2, the hazard of passing a policy was not significantly related to a state passing an LR—except in those cases where states had passed a mix of favorable and unfavorable policies. In this sense, the policy environment may have required county governments to become more engaged in mitigating diverse policy mandates. However, the effect of this variable is somewhat negligible, with an average estimate of 0.002 (with a lower bound 95% confidence interval estimate of 0.0002 and an upper bound of 0.099). On the other hand, the odds of a county passing an LR increased in relationship to the share of municipalities in the county that had passed unfavorable LRs by a factor of 4.47 (lower bound 1.75, upper bound 10.71), while the odds decreased with the share of municipalities that had passed favorable LRs. This result is demonstrative of the role that county governments play as an intermediary between municipal and state governments, and suggests that county policy behavior is possibly influenced more by the political behavior and policy adoptions of municipal governments than by state governments.

Figure 4.

Variation in the hazard rate (probability of any county adopting a policy in a given year) for each US state.

However, as the random portion of the model (Figure 4) suggests, there are specific state effects on counties passing an LR. As shown in Figure 4, there are significant differences in the state effect on county policy behavior related to LRs. Florida, Texas, Arizona, New Mexico, and California had the strongest positive effect (suggesting that counties are more engaged in enacting LRs). In contrast, Iowa, Montana, Alaska, Arkansas, and Alabama have the strongest negative effect, which suggests that if a county is located in these states, the average probability of the county government passing an LR is reduced.

Discussion and implications

This study explored factors that influence the adoption of immigration-related policies by sub-national governments in the United States. Using the case of LRs, the study examined factors that influence the adoption of LRs by county-level governments in the United States in the period 2004–2014. A multilevel discrete-time event history analysis model was used to examine the impact that factors identified as significant in the literature had on the probability of county governments’ adoption of an LR. The results indicate that county governments are particularly active in implementing LRs and that while population demographics, economic characteristics, and policy diffusion processes have all been suggested to influence the probability of sub-national governments engaging in immigration policymaking, there are unique nuances to these factors that may have been masked in previous studies. However, it is important to stress that caution must be applied to generalizing the findings of the analysis beyond the county level. First, data are limited to specific types of local immigration policies that sub-national governments have enacted (LRs), and the analysis includes a time period of study in which the federal government was particularly active in relation to immigration deportation and enforcement. In addition, the analysis also includes a period of unique economic decline (the Great Recession) and conflict theory suggests that actions aimed at maintaining status or privilege—such as enacting exclusionary policies—are heightened during times of resource scarcity (Weber, 1978).

Results of the analysis suggest that while previous research has suggested that demographic changes in the foreign-born population and economic conditions are important factors that influence the passage of local immigration laws by sub-national governments, when the analysis controls for the dynamic changes in these factors across time, unique nuances emerge. Perhaps most interesting is that rather than a growth in the foreign-born population (including undocumented migrant residents), the relationship between LR adoption and the share of the foreign-born population is negative. Rather, over time the adoption of LR policies at the county level is strongly related to large Latino populations. The positively correlated relationship between growth in Latino population and adoption of LR policies is the strongest and most consistent relationship in the estimated model. Such a result contradicts previous studies (Ellis and Goodwin-White, 2006; Steil and Vasi 2012; Wells, 2004), and suggests that when the percentage of the population that are Latino is controlled for, alongside the foreign-born population, there is evidence of a distinct racialization of LR policies. Such a result provides evidence (at the aggregate level) for the argument made by case study research that LRs are enacted largely in response to growing or large Latino presence in a given area (see, for example, Chavez, 2013). However, it is important to note that Latino political representation (measured here by number of Latino elected officials) lowers the probability of an LR being adopted.

The results of our analysis further indicate that there is a limited direct role of economic factors in influencing the passage of LRs. In particular, there appears to be very little impact of rates of employment in industries with the most direct (or perceived direct) competition (manufacturing, construction, and agriculture). Theoretically, when taken together these results provide support for the theory of the racialization of immigration and may indicate that social factors and sociopolitical discourse at the county level are perhaps more important than economic factors in influencing the passage of LRs at the local level.

Finally, the results of our analysis indicate that geographic location and policy activity of municipal and state governments also influence the policy adoption of LRs by county governments. County governments enacted overwhelmingly unfavorable or exclusionary LR policies during the time period of study and the results of our analysis suggest a unique relationship to policy activity by other actors at other scales of government. The shift in the hazard rate over time largely follows federal trends, with the largest increase in policymaking occurring in 2009—likely in response to the Secure Communities initiative by the federal government. However, the relationship between county LR policy adoptions to the adoption of LRs by state governments is varied. In this sense there appears to be a unique geography to the state–county interaction as the effect of state policymaking was highest for Florida and states along the United States–Mexico border. In addition, the results further indicate a type of tandem LR policy adoption occurring between municipalities and counties over the time period of study and suggests that county LR policy adoption may be influenced more by municipal and local-level policy activities rather than state-level policy activities. Given this, the primary channel for the diffusion of LR policymaking at the county level appears to be local to the county. However, more research is needed to better understand the policy diffusion mechanisms and systems of policy transfer influencing these trends.

What is clear from the analysis presented here is that the adoption of LR policies by sub-national governments demands more careful attention and further investigation. Particularly relevant to this study (and the US case) is the question of whether state and municipal LR policy adoption is influenced by similar factors that influence county government and whether or not there are scalar-specific effects. While previous research on immigration policy broadly has highlighted the policy mechanisms and actors that influence the mobilization of local immigration ordinances, this body of work has underemphasized the ways LRs are embedded in political, economic, and social contexts and discourses that surround immigration. Of particular interest is to what extent the effect of the broad racialization of LR policy adoption is also observed when examining state and municipal governments’ adoption of LRs and whether or not there are unique temporal or spatial trends to this reality. In this sense, there is also a need to understand the ways in which LRs as policy decisions are interpreted and represented in local contexts, and how these specific policy decisions shape broader immigrant regularization policy deliberations and decisions. At the same time, it would be interesting to consider other types of policymaking behaviors, beyond LRs, that also target immigrant communities in migrant-receiving societies, such as overcrowding housing ordinances, which are targeted at addressing housing strategies used by migrants to help promote integration into receiving countries and communities (Simpson, 2017). Although these issues are among the many that warrant further investigation, insights from this study lead to the conclusion that county governments are particularly active in immigration policymaking in the United States, and that sociopolitical factors may be more influential in driving these decisions than the real impact of migrants on socioeconomic conditions. Such a reality has enormous implications for scholarship and policy debates and discussions related to the geography of immigration regularization, migrant socioeconomic integration, and socioeconomic stratification in the United States.

Footnotes

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

Notes

Appendix 1: Definitions of types of migrant labor market regularizations ( Visser,2017 )

References

Apap

De Bruycker

Schmitter

(2000) Regularisation of illegal aliens in the European Union: Summary report of a comparative study. European Journal of Migration & Law 2: 263–308.

Artelaris

(2015) Local versus regime convergence regression models: A comparison of two approaches. GeoJournal 80: 263–277.

Austin

(2017) A tutorial on multilevel survival analysis: Methods, models and applications. International Statistical Review 85(2): 185–203.

Bell

Jones

(2015) Explaining fixed-effects: Random effects modeling of time-series cross-sectional and panel data. Political Science Research and Methods 3(1): 133–153.

Berry

Baybeck

(2005) Using geographic information systems to study interstate competition. American Political Science Review 99(4): 505–519.

Brenner

Theodore

(2002) Cities and the geographies of “actually existing neoliberalism.” Antipode 34(3): 349–379.

Chandler

Tsai

(2001) Social factors influencing immigration attitudes: an analysis of data from the General Social Survey. The Social Science Journal 38(2): 177–188.

Chavez

(2013) The Latino Threat: Constructing Immigrants, Citizens, and the Nation. Stanford, CA: Stanford University Press.

Coleman

(2007) Immigration geopolitics beyond the Mexico–US border. Antipode 39(1): 54–76.

10.

Cordero‐Guzmán

(2015) Worker centers, worker center networks, and the promise of protections for low‐wage workers. WorkingUSA. 18: 31–57.

11.

Coutin

(2005) Contesting criminality: Illegal immigration and the spatialization of legality. Theoretical Criminology 9(1): 5–33.

12.

Dauvergne

(2008) Making People Illegal: What Globalization Means for Migration and Law. New York: Cambridge University Press.

13.

Duman

(2014) Reducing the fog? Immigrant regularization and the State. Politics & Policy 42: 187–220.

14.

Ellis

Goodwin‐White

(2006) 1.5 generation internal migration in the US: dispersion from states of immigration? International Migration Review 40(4): 899–926.

15.

Engbersen

Guiraudon

Joppke

(2001) The unanticipated consequences of Panopticon Europe. In: Guiraudon V and Joppke C (eds) Controlling a New Migration World. London: Routledge, pp.222–246.

16.

Esbenshade

(2007) Division and Dislocation: Regulating Immigration Through Local Housing Ordinances. Washington, DC: Immigration Policy Center, American Immigration Law Foundation.

17.

Givel

Glantz

(2001) Tobacco lobby political influence on US state legislatures in the 1990s. Tobacco Control 10(2): 124–134.

18.

Goldring

Landolt

(2013) Caught in the Work -Citizenship Matrix: the Lasting Effects of Precarious Legal Status on Work for Toronto Immigrants. In: Migration, work and citizenship in the new global order. Routledge, pp. 85–102.

19.

Haubert

Fussell

(2006) Explaining pro‐immigrant sentiment in the US: Social class, cosmopolitanism, and perceptions of immigrants. International Migration Review 40(3): 489–507.

20.

Herod

(2012) Workers as geographical actors. Labor History 53: 335–353.

21.

Hosmer

Lemeshow

, and May

(2008) Applied Survival Analysis. 2nd ed. Hoboken, NJ: Wiley.

22.

Kessler

(2001) Immigration, Economic Insecurity, and the “Ambivalent” American Public. La Jolle, CA: Center for Comparative Immigration Studies.

23.

Kulu

Billari

(2006) Migration to urban and rural destinations in post-Soviet Estonia: A multilevel event-history analysis. Environment and Planning A 38: 749–764.

24.

Levinson

(2005) The Regularisation of Unauthorized Migrants: Literature Survey and Country Case Studies. Oxford: Centre on Migration.

25.

Massey

(1984) Spatial Divisions of Labor: Social Structures and the Geography of Production. New York: Routledge.

26.

Mayda

(2006) Who is against immigration? A cross-country investigation of individual attitudes toward immigrants. The Review of Economics and Statistics 88(3): 510–530.

27.

McClain

Carter

DeFrancesco Soto

et al . (2006) Racial distancing in a southern city: Latino immigrants’ views of black Americans. The Journal of Politics 68(3): 571–558.

28.

Mehan

(1997) The discourse of the illegal immigration debate: A case study in the politics of representation. Discourse & Society 8(2): 249–270.

29.

Pantoja

(2006) Against the tide? Core American values and attitudes toward US immigration policy in the mid-1990s. Journal of Ethnic and Migration Studies 32(3): 515–531.

30.

Papademetriou

(2004) Observations on Regularization and the Labor Market Performance of Unauthorized and Regularized Immigrants. Hamburg: Migration Research Group.

31.

Peck

(1996) Work-Place: The Social Regulation of Labor Markets. London: Guilford Press.

32.

Peck

Theodore

(2012) Politicizing contingent work: Countering neoliberal labor market regulation … from the bottom up? South Atlantic Quarterly 111(4): 741–761.

33.

Pettigrew

Tropp

(2006) A meta-analytic test of intergroup contact theory. Journal of Personality and Social Psychology 90(5): 751–759.

34.

Ramakrishnan

Wong

(2007) Immigration policies go local: The varying responses of local governments to undocumented immigration. Unpublished paper, University of California, Riverside, CA.

35.

Reid

Weiss

Adelman

et al . (2005) The immigration–crime relationship: Evidence across US metropolitan areas. Social Science Research 34(4): 757–780.

36.

Rodriguez

(2008) The significance of the local in immigration regulation. Michigan Law Review 228: 567–642.

37.

Shipan

Volden

(2008) The mechanisms of policy diffusion. American Journal of Political Science 52(4): 840–857.

38.

Simpson

(2017) Spatial patterns of international migrant resident settlement and incorporation in Winnipeg Manitoba. Population, Space and Place 23(7). DOI: 10.1002/psp.2060

39.

Singer

Willett

(2003) Applied Longitudinal Data Analysis: Modeling Change and Event Occurrence. Oxford: Oxford University Press.

40.

Singer

Critchley

, and Preuschoff

(2009) A common role of insula in feelings, empathy and uncertainty. Trends in Cognitive Sciences 13(8): 334–340.

41.

Sohoni

(2014) Perceptions of immigrant criminality: crime and social boundaries. The Sociological Quarterly 55(1): 49–71.

42.

Steil

Vasi

(2014) The new immigration contestation: Social movements and local immigration policy making in the United States, 2000–2011. American Journal of Sociology 119: 1104–1155.

43.

United States Immigration and Customs Enforcement Office (2014) Delegation of Immigration Authority Section 287(g) Immigration and Nationality Act. Available at: https://www.ice.gov/factsheets/287g

44.

Varsanyi

(2008) Immigration policing through the back door: City ordinances, the “right to the city,” and the exclusion of undocumented day laborers. Urban Geography 29: 29–52.

45.

Visser

(2017). Reshaping migrant labor market geographies: Local regularizations and the informal economy. Population, Space, and Place. DOI: 10.1002/psp.2025

46.

Visser

Cordero-Guzmán

(2015) Low wage workers and organizing. WorkingUSA 18(1): 1–6.

47.

Visser

Meléndez

(2015) Working in the new low wage economy: Understanding participation in low wage employment in the recessionary era. WorkingUSA 18(1): 7–29.

48.

Walker

Leitner

(2011) The variegated landscape of local immigration policies in the United States. Urban Geography 32: 156–178.

49.

Weber

(1978) Basic sociological terms. Economy and society 1: 3–62.

50.

Wells

(2004) The grassroots reconfiguration of US immigration policy. International Migration Review 38: 1308–1347.

51.

Wilkes

Guppy

Farris

(2008) “No thanks, we’re full”: Individual characteristics, national context, and changing attitudes toward immigration. International Migration Review 42(2): 302–329.

52.

Zatz

Smith

(2012) Immigration, crime, and victimization: Rhetoric and reality. Annual Review of Law and Social Science 8: 141–159.