4.1 Descriptive Analysis and Ranking Tables
Before turning to our regression results, we examine the occupational distribution of native-born, immigrant, and Mexican immigrant workers across the nine occupational categories we used to develop the occupational diversity index. As shown in Table 1, native-born workers are heavily concentrated in three occupational groups in both 2000 and 2010: sales and office, professional, technical and protective services, and management. These three occupational groups, which tend to be higher paying, employ 67.5% of the native-born workforce in 2010. In addition, a greater proportion of native-born residents are working in low wage service jobs in 2010 (13.3%) than in 2000 (11.6%). A very small proportion of native-born individuals work in agriculture and military occupations, 0.3% and 0.2%, respectively.
When immigrant workers are considered, they are less concentrated in higher wage occupations than native-born workers. In 2010, there were 18.6% of immigrants working in sales and office, 18.6% in professional, technical and protective services, and 10.5% in management, for a total of 47.7%. Compared to native-born workers, immigrant workers are also more concentrated in low-wage services: construction, other blue collar, and production/manufacturing than native-born workers. These jobs account for 47% of the immigrant workforce in 2010. These statistics show that immigrant workers appear to be bifurcated into high-wage and low-wage jobs. Like native-born workers, a greater proportion of immigrant workers are working in low-wage service occupations in 2010 than in 2000. This signals the changing economic structure over time.
We also compare Mexican immigrants, the largest immigrant group in the U.S., to immigrants overall. We find that there are stark differences in their occupational makeup, as compared to all immigrants. Mexican workers have low representation in management and professional, technical, and protective service occupations. A large number of Mexican workers, 73.9%, are concentrated in low wage services, construction, other blue collar, and production/manufacturing.
When we compare the occupational distribution of native-born, immigrants, and Mexican immigrants, there are clear differences, with a greater proportion of native-born workers concentrated in higher wage, higher status jobs and Mexican immigrants are more concentrated in lower wage, lower status jobs. Yet, when we consider the diversity index, all groups have high levels of occupational diversity. Immigrant workers have the highest occupational diversity score in 2000 at 0.85. Mexican immigrant workers have the next highest at 0.83, and native-born workers have the lowest at 0.81. These trends remain similar in 2010. Taken together, these statistics reveal that the overall occupational diversity, as measured by the diversity index, of the three groups is similar, but how each group is distributed across occupational categories varies.
[TABLE 1 ABOUT HERE]
Next, we rank metropolitan areas by the top ten most diverse and least diverse occupationally for immigrants, using our two measures: the diversity index and the index of specialization. Metropolitan areas that were ranked highest with the diversity index include six California metropolitan areas: Modesto, Stockton, Chico, Oxnard-Ventura, Santa Barbara-Santa Maria, and Santa Cruz-Watsonville (see Table 2). There were two metropolitan areas from Texas, Beaumont-Port Arthur and McAllen-Edinburg-Mission, that are also ranked high on occupation diversity of immigrants. In general, these regions have traditionally been magnets for immigrants, particularly Mexicans. They have also had strong agricultural and manufacturing bases and tend to be smaller in population than other metropolitan areas in our study. Boise-City-Nampa, Idaho and Lakeland, Florida round out the top ten most occupationally diverse metropolitan areas for immigrants. The range in the diversity index value for these metropolitan areas ranges from a high of 0.877 to a low of 0.859, therefore, the differences in level of occupational diversity is small among these metropolitan areas.
Looking at the ten metropolitan areas with the least occupational diversity for immigrants, they seem to be metropolitan areas that have older economies, such as Pittsburgh, Pennsylvania and Atlantic City, New Jersey, or are smaller in size. Four of the metropolitan areas could be considered college towns: Champaign-Urbana, Illinois, Gainsville, Florida, Ann Arbor, Michigan, and Madison, Wisconsin. Whereas the most diverse regions appear to be magnets for Mexican immigrants, metropolitan areas that have less occupational diversity seem to attract immigrants from a broader range of countries.
Since the diversity index is the spread of immigrants across the nine occupational groups without reference to any other group, we also calculate the index of specialization to compare the occupational distribution of immigrants in each metropolitan area relative to their distribution in the U.S economy as a whole. The rankings using the index of specialization are somewhat different than our rankings using the diversity index. Examining the top ten regions for occupation diversity of immigrants, we find that larger metropolitan areas, and those with strong high-tech and service economies tend to rank high. As shown in Table 2, two populous California regions, San Diego-Carlsbad-San Marcos and Sacramento-Roseville, rank highest for occupational diversity for immigrants. There is much more regional variation using the index of specialization as compared to the diversity index, with three metropolitan areas in the West, three in the South, two in the Northeast, and two in the Midwest.
According to the index of specialization, the regions that have the least occupational diversity for immigrants appear to be smaller places with more specialized economies. There are six metropolitan areas that are the same as those found in the diversity index ranking for least diverse area. The places that make it on this list and are different than on the diversity index list for least diverse include: Yakima, Washington, Bakersfield, California, Fayetteville-Springdale, Arkansas/Missouri and Merced, California. In general, the places on this list appear to have more specialized economies.
[TABLE 2 ABOUT HERE]
4.2 Explaining occupational diversity across regions
Beyond the rankings tables, we ran a set of OLS regression models to explore which characteristics are associated with greater occupational diversity of immigrants that take the general following form:
As indicated by equation 3, the dependent variable is the diversity index for each metropolitan area i, and is predicted by five sets of variables. is a vector of variables that measure the human capital of immigrants and include: the share of immigrants with a BA or higher, the share of a region’s immigrants who immigrated in the 1980s or 1990s (our proxy for labor market experience), and the degree of linguistic isolation or language ability. The terms and are vectors of variables that measure the degree to which immigrants are spatially concentrated or cut off from regional job opportunities and include variables such as the spatial dissimilarity index of the foreign-born and the suburban share of jobs. Since the occupational diversity of all workers in a region, much less immigrant workers, is shaped by the structure of regional labor demand, we include a set of control variables that account for the industrial structure of each region. Lastly, the vector includes variables that proxy for the context of reception, such as the presence of pro- or anti- immigrant policies in the central city of each region and the number of immigrant service NGOs per capita.
We conduct this analysis for the occupational diversity index for all immigrants, Mexican immigrants and also for the differential diversity index for immigrants versus native-born (Table 3). In specifying the models, we attempted to balance the competing needs of including variables from each of our conceptual factors while maintaining as parsimonious a model as possible with decent explanatory power.
[TABLE 3 ABOUT HERE]
Table 3 lists the results of our regressions predicting occupational diversity of immigrants. Column one contains the results for all immigrants, while column two is for Mexican immigrants separately. The measure of human capital (% of immigrants with a BA or higher) is associated with lower occupational diversity and is significant for all immigrants and the differential. While this result is contrary to individual level theories regarding immigrant integration, at the regional level, this makes sense because as more immigrants earn a higher degree of education, they tend to be more concentrated in the two higher skill occupational categories (management and professional services), which is an indicator of labor market specialization. Furthermore, a greater share of educated immigrants makes immigrants’ occupational diversity more similar to native-borns. Interestingly, this variable is not significant for Mexican immigrants, which is consistent with the literature on racialization in the labor market for Mexicans. The degree of linguistic isolation also appears to reduce labor market diversity for immigrants, but not for Mexican immigrants.
Our findings also indicate that when immigrants are more spatially segregated, they are less diverse across occupations, as indicated by the negative and significant coefficient for the spatial dissimilarity index among immigrants. Recall that the dissimilarity index (calculated at the census tract level) measures the share of immigrants that would have to move to another neighborhood to even out the distribution of foreign-born and native-born individuals within a region. This finding is consistent with the existing literature on the spatial isolation of immigrants.
When we examine the results for our variables that attempt to capture the context of reception for immigrants, we find that generally none of the policy variables were significant, except for Mexican immigrants and the number of immigrant service NGOs per capita. The fact that regions with more immigrant service NGOs have less occupational diversity among Mexican immigrants is somewhat of a puzzling finding. One reason why this may be the case is that these organizations form in response to the fact that Mexican immigrants are concentrated in a relatively narrow set of low-wage jobs that may require more social and support services.
Interestingly, our primary measure of ethnic enclaves—the share of the regional immigrant population residing in census tracts that are 30 percent or more foreign-born—is not significant but spatial segregation of foreign-born from native-born is significant. Among all immigrants, higher levels of spatial segregation are associated with lower occupational diversity.
Our indicator for spatial mismatch was insignificant for all three models, although the two indicators for regional industrial structure are highly significant for all immigrants. Specifically, the industry diversity index (calculated in a parallel manner as our occupational diversity index) is positive and significant for models 1 and 3. This suggests that more diverse economies in terms of labor demand are associated with greater occupational diversity for immigrants. It is critical that we control for the effect of industrial structure of the region since the availability of jobs in different industries shapes the ability of workers to be employed in a broader range of occupations. Therefore, we view industrial structure as a critical control factor. Notably, this variable is insignificant for Mexican immigrants, again perhaps, indicating that Mexicans face additional barriers within the labor market that prevent them from taking advantage of a wider array of jobs within the regional economy. In addition, the share of regional jobs in manufacturing is negatively associated with occupational diversity.
Among the other variables that control for the economic structure of regions, the overall median household income level was positive and significant. This indicates that richer regions tend to have greater occupation diversity among immigrants. The only other variable that is significantly associated with occupational diversity is the measure of income inequality (the 90/10 household income ratio). Regions that are more unequal are slightly more diverse.
Overall, the model predicting occupation diversity among Mexican immigrants has low predictive value, with an adjusted R2 equal to 0.1325. The other two models have much higher R2 values, with model 1 (all immigrants) equal to 0.4424 and model 3 (the differential between immigrants and native-born) equal to 0.4681. This suggests that our none of the factors (eg. human capital, context of reception, ethnic enclaves, spatial mismatch, industrial structure, and regional controls) strongly explain the occupation diversity of Mexican immigrants and that there may be some unexplained factors that are difficult to capture in a quantitative model, such as the racialization of Mexican immigrants in the labor market,that might be shaping labor market outcomes for this population.
Our models predicting the occupational index of specialization have higher predictive values with R2 value for the all immigrant model equal to 0.537 and R2 value equal to 0.322 for Mexican immigrants (See Table 4). The results for our models that explain variation in the occupational index of specialization are similar to the results for the occupational diversity index. As a reminder, the signs on each coefficient have the opposite interpretation. Therefore, as the index of specialization increases, occupational diversity decreases. Again, the share of workers with a BA or higher is associated with less diversity, while good English language ability of immigrants seems to increase diversity.
[TABLE 4 ABOUT HERE]
Unlike the diversity index regressions, for the index of specialization, we do find evidence that spatial mismatch and enclave effects matter. For all immigrants, the degree to which regional jobs are spread out in suburban locations, rather than concentrated in the central city tends to reduce occupational diversity (positive sign and significant at the 5% level). While immigrants are increasingly bypassing the central city for emerging suburban immigrant hubs, the majority of immigrants are still concentrated in the older central city areas, which tend to have better transit access and lower housing costs. Interestingly, in regions that have a higher share of the Mexican immigrant population living in ethnic enclaves, labor market diversity is reduced. This suggests that Mexicans face dual barriers in terms of labor market and residential segregation. This is also consistent with the close ethnic ties and social networks built up within ethnic enclaves (e.g. strong ties) that could serve to exacerbate labor market segmentation.
The story with industry structure is also broadly similar, with the percent manufacturing negatively associated with diversity, and the index of specialization negative and significant. As regions specialize in other sectors for their primary export base (e.g. high-tech services, FIRE), rather than manufacturing, they tend to generate job opportunities across a wider spectrum of occupations and thus create a more diverse labor demand structure. In other words, regions with high-tech or service-based economies produce a broad based labor demand for all workers, including immigrant workers. The fact that these variables are not significant for Mexican immigrants again is suggestive of labor market segmentation within the labor market for Mexican immigrants.
4.3 Explaining economic resilience
Next, we turn to our models that explore the relationship between occupational diversity and economic resilience. We analyze the impact of diversity on two distinct measures of economic resilience; the change in unemployment and the net change in real wage income among immigrants between 2000 and 2010. A comparison across this time period offers a comparison with a base year that was the pinnacle of the overall labor market in terms of unemployment (2000) to 2010, which is just after the Great Recession that ended in 2009. Our diversity index variables are measured in the year 2000. Table 5 contains the results of models that test the relationship between our occupational diversity measures and economic resilience among immigrants. In each of the models below we include the same set of regional controls that appear in tables 3 and 4 for population, age, homeownership, inequality, and base-year unemployment and income. These variables are included to control for the broad economic factors that differentiate the economic characteristics of U.S. metropolitan areas. In addition to these controls, we also include a measure of overall economic performance—the change in unemployment rate for all workers—so that we can better isolate the impact of our occupational diversity measures on the change in unemployment or income among immigrants. We also include a measure of changing labor supply—percent change in total immigrant population between 2000 and 2010—as well as an indicator of the nature of labor demand (percent of regional employment in finance, insurance, and real estate (FIRE) sectors.
[TABLE 5 ABOUT HERE]
Based on the results presented in Table 5, we find that greater occupational diversity is moderately associated with smaller changes in unemployment and higher real wage income changes among immigrants. However, the results are significant only for the occupational diversity index, rather than the index of specialization. This indicates that regions where immigrants are more broadly spread out across the labor market are more resilient to the economic shock posed by the Great Recession. This is essentially a ‘portfolio argument’ at work here. To the degree that immigrants are not concentrated in any one sector (e.g. construction) that faces a macroeconomic shock, immigrant workers as a whole will not see as large a spike in unemployment. Interestingly, regions with higher rates of homeownership had a lower change in real income for immigrants, which may be reflective of the housing foreclosure crisis during the Great Recession.
While the literature on immigrant integration often pitches competing theories, our study of metropolitan areas with the largest immigrant populations suggests that there is support for combining different schools of thought on immigrant economic integration. Using two different measures for immigrant economic integration (operationalized as occupational diversity in the metropolitan labor market), we found that the most robust factors affecting occupational diversity for all immigrants, regardless of how we operationalized occupation diversity, include: human capital and industrial structure. In particular, English language ability (or isolation) are associated with immigrants’ labor market outcomes, thus lending support to the classic assimilation theory. It is not surprising that the distribution of jobs across industries is also highly predictive of whether immigrants are able to enter into a more varied set of occupations.
To some extent, there was some support for all of our hypotheses, but it was largely dependent on the outcome variable of choice. When the diversity index was the dependent variable, spatial segregation was associated with less occupational diversity for all immigrants. When the index of specialization was the dependent variable, having pro-immigrant policy stance, which is an indicator of the local context of reception, was associated with more occupation diversity. In addition, the spatial mismatch variable, suburban share of regional jobs was related to a decrease in occupational diversity.
For Mexican-immigrants, regardless of the outcome variable used, the predictive power of our statistical models was much lower than for all immigrants, suggesting that there are unobserved factors that may better explain labor market integration for Mexican immigrants. In addition, the continual stream of Mexican immigration and the high numbers of undocumented immigrants may also contribute to Mexican immigrants being employed in occupational niches, rather than being distributed throughout the labor market. These factors could explain why more human capital did not translate into greater occupational diversity for Mexican immigrants, but did for all immigrants. These results for Mexican immigrants are consistent with research that shows that Mexican immigrants do not realize the same occupational gains from human capital accumulation as other immigrants (Alba and Nee, 2003; Portes and Rumbaut, 2001; Haller, Portes, and Lynch, 2011). Furthermore, for Mexican immigrants, it is not the spatial distance from whites (dissimilarity index) that affects occupational diversity, but living in ethnic enclaves that does. Thus, living in ethnic enclaves for Mexican immigrants may be inhibiting their ability for labor market advancement. This supports Granovetter’s thesis regarding the need to have “weak ties” with individuals that can offer opportunities to a wider range of occupations.
Are metropolitan areas with greater immigrant economic integration more resilient to economic shocks? Our analyses using the occupational diversity index as the independent variable of interest indicates that greater occupational diversity does buffer immigrants within metropolitan regions from more pronounced effects of the Great Recession. Unemployment level change for immigrants is less dramatic in metropolitan areas with greater labor market integration. Furthermore, real wage income growth is higher in metropolitan areas with more economic integration. However, we did not find support for this with the other independent variable of interest: Index of Specialization. Thus, our findings are mixed.
While our findings are statistically significant, it is important to note the limitations of this analysis in drawing causal conclusions on the direct relationship between labor market diversity and overall economic integration of immigrants or regional resilience in a larger sense. There are other critical factors which shape opportunity which we are unable to measure, including multi-generational measures of integration such as homeownership and wealth transfers, as well as more specific measures of skill acquisition and labor market advancement. Moreover, our measures of resilience (change in unemployment and real wages) are rather coarse measures of resilience and do not explicitly measure the capacity to bounce back from economic shocks as Benner and Pastor (2012) do. Nonetheless, we argue that this exploratory analysis is useful in shedding light on the links between labor market diversity and the economic well-being of immigrants and the importance of the metropolitan scale for both further research and policy interventions.
Policy Implications and Future Research Directions
These findings have implications for current immigration policy reform, showing that policies that seek to recreate a segmented labor force will be unwise since occupational diversity may reduce the impact of a sectoral specific shock. Our current federal immigration policies offer preferential treatment to high-skilled (e.g. software engineers) and low-skilled or agricultural workers, thereby limiting the range of occupations that are allowed to migrate to the U.S. Policymakers should provide legal avenues for workers from more wide ranging types of occupations.
At the regional level, local government actions towards adopting immigrant-friendly policies can work to economically integrate immigrants. Also, having a greater industrial mix provides more opportunities for immigrants to be more diversely distributed across occupations. Overconcentration in one industry, such as manufacturing, however, can result in less occupational diversity for immigrants.
The importance of spatial dynamics, including the significance of racial segregation and the regional share of suburban jobs highlights the importance of policies that work to develop more economic, racial/ethnic, and land-use integration. Housing and land use policies, such as mixed-income, mixed-use, and inclusionary housing, for example, that are directed at integrating low-income families may result in greater spatial integration for immigrants, thus leading to economic integration.
Looking forward, the findings in this paper can serve as the basis for case selection for future qualitative analysis that seeks to drill down to the explanatory factors that enhance or exacerbate economic integration and labor market mobility of new immigrants. Also, the differences between the explanatory factors associated with occupation diversity for Mexican-immigrants and all immigrants deserve more examination. Furthermore, findings from this study offer ways policy-makers, elected officials, and street-level bureaucrats who would like to facilitate immigrant integration can do so through human capital accumulation, desegregating immigrants, and developing a more welcoming context.