The Occupational Attainment of Natives and Immigrants: A Cross-Cohort Analysis∗ Hugh Cassidy† July 10, 2015

Abstract This paper investigates the occupational attainment of natives and immigrants in the United States, where occupations are characterized by a vector of task usages (analytical, interactive, and manual) that describe the activities performed on the job. Immigrants on average perform fewer analytical and interactive tasks and more manual tasks than natives, and these differences are larger for women than men. The average task usage gaps between natives and immigrants have widened significantly since 1970. Lower English language proficiency, living in a larger ethnic or language enclave, and originating from a larger conational group increase the task usage gaps. While earlier immigrant cohorts show some degree of task usage assimilation, newer cohorts have experienced a significant slowdown in their occupational assimilation rates. These results support findings of slower earnings assimilation of recent cohorts.

Keywords: Tasks, Immigration, Assimilation. JEL Classification: J60, F22, J24, J62.

∗

I would like to thank participants at the 49th Annual Conference of the Canadian Economics Association, and the Fourth Society of Labor Economists / European Association of Labour Economists World Meeting for helpful comments and suggestions. All code to generate the results and an online appendix are available for download from my website: https://sites.google.com/site/hughcassidy. † Kansas State University. E-mail: [email protected].

1

1

Introduction

Historically, the earnings of immigrants in the U.S. have to tended to assimilate to natives with time since migration.1 Recently, however, Borjas (2015) finds that not only has the initial labor market performance of immigrants declined with newer cohorts compared to prior cohorts, but that the economic assimilation rates of recent cohorts are also markedly slower than past cohorts. Comparing immigrants arriving between 1965-69 to those arriving between 1985-89, initial age-adjusted earnings are 24% and 33% lower than natives, respectively. In terms of growth rates, the 1965-69 cohort experienced an 11% point reduction in their earnings gap in their first ten years since migration, while the 1985-89 experienced only a 7% point reduction. The 1995-99 cohort, on the other hand, experienced essentially no earnings assimilation in their first ten years post-migration. These findings are striking, in that they imply a major change has occurred in the immigrant experience in the U.S. The importance of these results calls for replication and robustness checks to be performed. As a result, in this paper I explore the occupational attainment of immigrants compared to natives, how immigrant occupational attainment varies by time since migration, and whether there has been a change in either pattern over time. The motivation for examining these issues is, in part, to test whether the slowdown in immigrant assimilation observed in Borjas (2015) extends to labor market conditions other than earnings. By focusing on occupational attainment instead of on earnings, I explore a different (though closely related) dimension of the labor market experience of U.S. immigrants. If, indeed, there has been a slowdown in overall economic assimilation of immigrants in the U.S., evidence of this slowdown may be visible by examining occupational attainment. I use data from the Occupational Information Network (O*NET) to derive analytical, interactive, and manual task usage levels for each occupation. This approach allows for a succinct and interpretable analysis of a large number of occupations. Past work such as Green (1999) have relied instead on a small number of occupations to investigate occupational as1

See Chiswick (1978) and Borjas (1985).

2

similation, which ignores the large variation of job characteristics within broad occupational groups. Also, interpreting results can be cumbersome when more than a very small number of occupation groups are included in the analysis. Worker data are taken from the U.S. Census. I use the 1970, 1980, 1990, and 2000 decennial Census samples, as well as the three-year American Community Survey (ACS) samples for 2005-2007 and 2009-2010, which I henceforth refer to as the 2006 and 2010 samples, respectively. I perform all of the estimates separately for men and women. I find that immigrants perform fewer analytical and interactive tasks, and more manual tasks, than natives. Furthermore, the differences in tasks between native and immigrants are higher for women than men. These gaps have risen significantly since 1970 for all three tasks for both men and women. Male natives have experienced an increase in analytical and interactive tasks and a decrease in manual tasks, while male immigrants have experienced little change in task usage during the period considered. Both native and immigrant women have experienced increases in analytical and interactive tasks and a decrease in manual tasks over time, though the changes for native women have outpaced those of immigrant women, leading to a widening of the task usage gaps for women as well. The immigrant/native task gaps tend to narrow with time since migration as immigrants assimilate occupationally to natives. However, the rate of assimilation has slowed for more recent immigrant cohorts. In fact, newer cohorts appear to be diverging from natives in terms of occupational attainment. These results hold for both men and women. This finding supports the results from Borjas (2015) that newer immigrant cohorts are assimilating to natives in terms of earnings at a slower rate than past cohorts. Declining occupational assimilation rates would have important effects on overall economic assimilation of immigrants. If, for example, newer immigrant cohorts are less occupationally mobile than past cohorts, they may be stuck in occupations where they are poorly matched, leading to slower earnings growth and thus slower assimilation. Alternatively, a slow rate of occupational assimilation may signal that newer immigrant cohorts are not gaining analytical and interactive skills at

3

the same rate of past cohorts, and so earnings growth has suffered. Borjas (2015) points to a slowdown in English language proficiency accumulation as a potentially important factor in the slowdown in earnings assimilation by newer cohorts. I find that higher English language proficiency does diminish the task gaps, with more fluent English speakers performing more analytical and interactive tasks and fewer manual tasks. Incorporating language proficiency helps to explain a portion of the overall occupational assimilation, which is evidence that improved language proficiency is an important channel through which immigrants occupationally assimilate. Controlling for language proficiency also helps to explain a portion of the slowdown in occupational assimilation, which supports the evidence in Borjas (2015) of the importance of language in understanding the immigrant assimilation slowdown. In addition to language proficiency, Borjas (2015) argues that an increase in the effective size of conational groups in the U.S. can help to explain the slowdown in assimilation of recent cohorts. I investigate this by first separating immigrants into those originating from “large” sending countries, i.e. those such as Mexico, China, etc. that are the largest source of immigrants to the U.S., versus “small” sending countries. While immigrants originating from the large countries have significantly higher task usages gaps overall, both exhibit a slowdown in occupational assimilation for more recent cohorts. I further investigate this idea by controlling for the size of an individual’s ethnic and language enclave, as measured by the fraction of the individual’s state that originate from the same country and the fraction that speak the same (non-English) language. I find that living in a larger language enclave corresponds to lower analytical and interactive task usage, and increases manual task usage. However, controlling for enclave size can explain only a small portion of the occupational assimilation slowdown. Thus, while a large part of the slowdown remains unexplained, controlling for language proficiency and enclave size can help to explain a sizable portion. This paper is organized as follows. In Section 2, I discuss the existing literature. In Section 3, I describe both the O*NET and Census data used in the analysis. I also describe

4

the procedure for calculating the task usages. I begin the main analysis in Section 4, where I compare the task usages of natives versus immigrants across years. The central results of the paper regarding occupational assimilation are shown in Section 5. I then extend the analysis to include language and enclave size in Section 6. I conclude the paper in Section 7.

2

Existing Literature

This study is closely related to Peri and Sparber (2009), which also combines several Census decennial samples with O*NET task usage variables. The focus of that paper is the effect of immigration on the task specialization of natives. They argue that since low-educated immigrants have a comparative advantage in manual over communication tasks due to lower English language ability compared to low-educated natives, native workers respond to immigration by substituting toward communication tasks. As a result, immigration has little to no impact on native earnings since low-educated natives and immigrants are not fully substitutable. They do not focus on immigrant assimilation, either in earnings or occupationally. They do, however, provide some evidence of the task assimilation process in Figure 2 of their paper. This figure shows the relative communication to manual task usage of recent immigrants (less than ten years in the U.S.), old immigrants (more than ten years in the U.S.), and natives. They show that recent immigrants have a lower relative communication to manual task usage than older immigrants , which is evidence of assimilation to native task usage with time since migration. Though their empirical findings are consistent with what I find, my paper explores in more detail the occupational assimilation process. In addition, I consider a longer sample by including the 2010 three-year American Community Survey (ACS). Adser`a and Ferrer (2014a) discuss the role of the linguistic proximity of an immigrant’s native language to either English or French on the occupational assimilation of immigrants in Canada. Similar to my paper, they use the O*NET to assign a vector of skills to each

5

occupation (analytical, social, fine motor, physical strength, and visual). They find that immigrants with low linguistic proximity (i.e. first languages that are linguistically dissimilar to English or French) have lower levels of analytical and social skill usages, and higher levels of strength, than natives, while those with high linguistic proximity resemble natives quite closely.2 These results echo the findings in my paper, where I find that English language proficiency has a strong positive effect on analytical and interactive task usages, and a negative impact on manual task usage. Adser`a and Ferrer (2014b) investigate whether female immigrants in Canada behave as secondary workers, by merging O*NET occupational characteristics (strength and analytical) with Canadian Census data. They find that these immigrants use fewer analytical tasks and more strength tasks than comparable native women. They find little evidence of assimilation in task usages overall, but university-educated women exhibit some analytical task assimilation and significant physical task assimilation. In terms of economic content, my paper is closely related to Borjas (2015), which investigates the earnings assimilation of immigrants across cohorts. Borjas (2015) finds that newer cohorts not only have lower earnings shortly after entry to the U.S. than older cohorts compared to natives, but also exhibit slower assimilation rates with time since migration. However, while that paper does control for within- versus across-occupation earnings growth, it does not discuss the occupational assimilation process of immigrants. I view my paper as, in part, an independent assessment of the Borjas (2015) result of an earnings assimilation slowdown, where I examine whether there exists evidence of this slowdown from an occupational perspective. Green (1999) explores immigrants to Canada during the 1980s, and examines their occupational assimilation to natives. He aggregates occupations into a small number of groups, and investigates how the proportion of immigrants in those occupation groups compares to natives across cohorts and with time since migration. He finds that immigrants tend to work 2

The lack of direct language proficiency measures in the Canadian Census, either French or English, motivates their use of linguistic distance.

6

in more skilled occupations than natives, but that this relationship has declined with newer cohorts. My study is similar to Green (1999), with the major innovation being the characterization of occupations by task usages instead of aggregating occupations into a small number of categories. Goldman, Sweetman, and Warman (2011), using the Longitudinal Survey of Immigrants to Canada (LSIC), finds that immigrants tend to move to higher skilled occupations following migration, with the proportion of males in high skilled occupations increasing from 41.2% six months after migration to 60.5% four years after migration. Their definition of occupational skill, however, is uni-dimensional, while I allow for a vector of three occupational characteristics. Finally, two papers that explicitly examine the relationship between immigrant task usage and years since migration using panel data are Imai, Stacey, and Warman (2014), which also uses the LSIC, and Lessem and Sanders (2014), which uses the New Immigrant Survey. Both of these papers utilize a unique feature of the data sets, which is an immigrant’s premigration occupation. This information is used to assign pre-migration task usage, which is then compared to the immigrant’s current occupation in the destination country (Canada and U.S., respectively) to calculate the immigrant’s “task gap”. Imai, Stacey, and Warman (2014) finds that immigrants tend to use more fine motor, visual, and physical strength tasks, and lower interpersonal and analytical, shortly after migration than prior to migration, but that this gap closes with time since migration. Lessem and Sanders (2014) finds a similar result, with immigrants increasing their cognitive task usage and decreasing their manual task usage with time since migration. These changes in task usage over time support the idea that immigrants assimilate occupationally, and that this assimilation involves increased use of cognitive-related tasks, and decreased use of manual-related tasks. Both of these studies use data that follows immigrants during the 2000s for only up to five years. Thus, while they provide insights into the assimilation process of immigrants, they do so for only a brief period of time and for a narrow immigrant arrival window. My study, on the other hand, compares

7

the assimilation rates across a number of immigrant cohorts over several decades, allowing me to determine whether there have been changes in occupational assimilation patterns over time.

3

Data

3.1

U.S. Census

Worker data are taken from the U.S. Census Integrated Public Use Microdata Series (Ruggels et al. 2010).3 For years 1970-2000, I use the decennial Census, while for 2010 I use the threeyear pooled ACS which covers years 2009-2011.4 I consider both men and women as separate samples, ages 25-64, who are employed and have a valid occupational code assigned to them and who are not in the military. Immigrants are defined as those who are either non-citizens or naturalized citizens. For immigrants, I require that they arrived in the U.S. age 18 or older. While the Census occ1990 occupation code is available across all samples I consider, it is not balanced, meaning some occupations that appear in, for example, the 1990 sample do not appear in the 2000 or 2010 samples. To facilitate comparison across the years, I follow Autor and Dorn (2013) and generate a consistent occupation code similar to variable occ1990. I use the same crosswalks employed by Autor and Dorn (2013) to create a new occupation variable, occ1990alt, based off of the occ variable provided in each year. This new variable is consistent across samples, and is balanced, so all occupations appear in all sample years. Additional details of this procedure are discussed in the Appendix.5 A notable trend in the sample has been the steady increase in the years of education for both immigrants and natives. However, the increase in years of education has been greater 3

Data is available for download at: https://usa.ipums.org/usa/. The 1970 data is taken from the 1% State, Metro, and Neighborhood Form 1 samples. The 1980, 1990, and 2000 data represent the 5% random samples. The 2010 data is pooled from three 1% samples from 2009-2011. 5 Additional details and code are available upon request. 4

8

for natives than for immigrants. Between 1970 and 2010, male natives experienced a 2.2 year increase in years of education, while male immigrants saw only a 1.2 year increase. For women, these values are 2.5 and 2.2 years. Thus, the education gap between natives and immigrants has expanded in the past several decades in my sample.

3.2

O*NET

While the U.S. Census contains occupational information about both natives and immigrants, it lacks information regarding the types of activities performed on the job. Previous studies of immigrants, including Green (1999) and Cohen-Goldner and Eckstein (2010), have used broad occupational classifications. However, this method ignores the fact that a substantial amount of variation exists within these broad classifications in terms of activities performed on the job. Instead, I turn to occupational task usage to characterize jobs. As the Census includes no task information, however, another data set is needed to assign task usages on the job. For this purpose, I use the U.S. Department of Labor’s O*NET database.6 This data set contains information regarding a large number of characteristics required to perform an occupation. Examples include mathematical reasoning, oral comprehension, and finger dexterity. In this paper, as in Peri and Sparber (2009), I focus on the abilities survey component of the O*NET. In calculating the occupational task usage vector, I follow closely the approach used in Peri and Sparber (2009). I rescale each characteristic to represent the percentile of the 2010 and 2011 ACS samples whose occupation requires lower levels of that characteristic. The characteristics are then grouped into three task bundles representing analytical, interactive, and manual task usages. These groupings are shown in Table 1, and are the same groupings used in Peri and Sparber (2009), with the exception that I break apart their cognitive task group into analytical and interactive groups. I take the mean over the characteristics to 6

This data is available for download at: http://www.onetonline.org/.

9

arrive at the final analytical, interactive, and manual task usages. Additional description of both this procedure as well as how the occupational mapping is performed are available in Appendix A. Task usages are by construction between zero and one, since they are an average over percentiles. One of the potential issues with the task-based literature is that there are many ways to calculate task usage vectors, including different choices over which occupational characteristics to include, as well as different methods of combining these characteristics to produce a final occupational task usage. This issue is discussed in Autor (2013), who recommends that empiricists use, in addition to their own purpose-built task measures, existing “off-the-shelf” measures when conducting their analyses. This practice helps to discipline the researcher, and minimize the potential for task measures to be constructed in such a way that simply helps to support a hypothesis. It is in this spirit that I use in my primary analysis the task groupings and aggregation method from Peri and Sparber (2009). However, as a robustness check, I also recreate all tables and figures presented below using instead the analytical, interactive, and manual task usage groupings used in Imai, Stacey, and Warman (2015). I also calculate these alternative task usages using the principal component analysis method instead of averaging over the characteristics. All of these results are available as an online appendix. My results are qualitatively identical when this alternative task measure approach is used, which provides reassurance that the results in this paper are not driven by subjective choices of task grouping and aggregation technique.

4

Task Usages: Natives versus Immigrants

In this section, I compare the analytical, interactive, and manual task usages of immigrants to natives across the Census years. The first set of results, in Tables 2 and 3 for men and women, respectively, shows the mean task usages of immigrants and natives, as well as the

10

difference in the means, for each survey year.7 Several results are worth noting. First, immigrants across all years, for both men and women, on average have lower analytical and interactive task usages, and higher manual task usage, than natives. The second noteworthy finding is that for all three tasks, and again for both men and women, the gaps between immigrants and natives have widened over time. For example, in 1970, the analytical task usage gap between male natives and immigrants was -0.028, while in 2000 the gap was more than double this value, and in 2010 the gap increased further to -0.091. Similar patterns are present for the interactive and manual tasks. Figures 1 and 2 show these task usage divergences for men and women, respectively. I also find that immigrant and native women are more dissimilar in terms of tasks performed than immigrant and native men. In each survey year and for each task, the gap for women exceeds the gap for men. To test whether the task usage trends are statistically significant, I run a series of regressions with task usage as the dependent variable. I find that these trends are all statistically significant at the 1% level, except for manual tasks for women which does not show a statistically significant trend.8 What can explain this large divergence between natives and immigrants in tasks performed? As noted above, while the average number of years of schooling has increased for both natives and immigrants, the increase has been more significant for natives than for immigrants. As education is strongly related to occupational task usage,9 this change may help to explain at least part of this pattern. In order to compare immigrants to similar natives, I impute the predicted task usage of an immigrant given their age, education, and survey year. To do this, I run a series of 7

These summary statistics are weighted, where the weights are adjusted by the number of weeks each respondent has worked in the previous year. 8 In these task regressions, I control for survey year, an immigrant dummy variable and a variable that equals zero for natives, and equals the difference between the survey year and 1970 for immigrants, and I cluster standard errors at the year of immigration level. The coefficient on this variable is negative for analytical and interactive tasks, and positive for manual tasks, indicating a statistically significant declining trend for analytical and interactive, and increasing trend for manual, tasks for both men and women. 9 Analytical and interactive task usages increase significantly with years of education for both natives and immigrants, while manual task usage declines. Details available upon request.

11

regressions, separately for men and women, where the dependent variable is task usage of native workers in a given survey year, and the independent variables are age, age squared, age cubed, all interacted with education group dummies, as well as education group dummies introduced separately. This specification allows for education to have both a direct impact on task usage, as well as allow the age profiles of task usage to vary by education group. Coefficients from each of these regressions are used to predict task usages of immigrants in a given survey year, based on their age and education.10 I refer to the difference between an immigrant’s actual task usage and their predicted task usage as their task usage gap.11 The mean task usage gaps for immigrants by year are shown in Table 4 for men and Table 5 for women. First, note that the task usage gaps persist even when controlling for age and education. Between 1970 and 2000, these gaps have remained largely unchanged for men, while for women they have expanded across all three tasks. Between 2000 and 2010 there was a substantial increase in all three task usage gaps for both men and women. Due primarily to this large increase in the 2000s, between 1970 and 2010, the task usage gaps between natives and immigrants controlling for age and education have expanded for both sexes. Borjas (2015) notes that the period between 2000 and 2010 was characterized by significantly slower earnings assimilation of immigrants. One possible explanation for these observations may be that the Great Recession is influencing the results. If the recession disproportionately affected immigrants, then it may help to explain the large occupational divergence between natives and immigrants observed during the 2000s. To address this question, I turn to the 2006 sample, since the years covered pre-date the onset of the recessionary period.12 I find that the task usage gaps for men are actually largest in 2006, and narrow between 2006 and 2010. For women, I observe expansions in the task usages gaps between the 2000, 2006, and 2010 samples. Thus, it appears that the rapid divergence of task usage 10

Results of these regressions are not shown but are available upon request. Peri and Sparber (2009) use a similar approach, where they clean task usages of the effects of demographic characteristics, and instead use predicted task usages in their analysis. 12 Tables are available for download from my website. 11

12

between natives and immigrants during the 2000s was not the result of the Great Recession.

5

Occupational Assimilation

In this section, I explore the occupational assimilation process of immigrants. I further partition the samples into immigrant cohorts, based on their year of arrival into the United States. I then investigate how the task usages of immigrants vary with years since migration (YSM), and how the relationships between task usages and YSM have changed between older and newer cohorts. This analysis is motivated partly by the results in Borjas (2015), which provides evidence that newer immigrant cohorts are not only performing more poorly compared to prior cohorts in terms of earnings shortly after arrival to the U.S., but are experiencing slower earnings assimilation as well. Since occupational assimilation may be an important dimension of the overall economic assimilation process, it is natural to examine how the rate of occupational assimilation may have changed across cohorts. Furthermore, evidence of a slowdown in occupational assimilation would provide important confirmation of the results in Borjas (2015) from a source other than earnings. The first two cohort groups are those immigrants who arrived before 1950 and those who arrived between 1950-1959. The other cohorts are divided into five-year periods, starting with 1960-64 onward. This gives me a total of 12 cohorts. For ease of exposition, I start by considering only the 1965-69, 1975-79, 1985-89, 1995-1999, and 2005-2009 cohorts. I use only these groups as they allow me to look at task usage shortly following migration, since they arrived within five years of a Census survey. I track these groups over time: shortly after migration (0-5 years), 10-15 years after migration, and 20-25 years after migration. I focus on the predicted task usages of immigrants described above, that account for age, education, and year effects. The results are shown in columns (2) to (6) of Table 6 for men and Table 7 for women, and I include the average across the five cohorts considered in column (1).

13

Examining column (1), which includes all five cohorts, we can see that both analytical and interactive tasks exhibit convergence overall, while the pattern for manual task usage is less clear. Shortly after migration, male immigrants’ analytical task usage is 0.057 below the task usage of their comparable native. This gap shrinks to 0.048 below natives 10-15 years after migration, and further shrinks to 0.033 below natives 20-25 years after migration. These values for the interactive task gaps are 0.074, 0.067, and 0.050 below natives, respectively. The manual task usage gap initially expands from 0.020 to 0.025 above natives, then shrinks to 0.018 above natives. For women, the patterns are the same, though the gaps are higher than those for men at all years since migration groups. These patterns demonstrate that immigrants undergo occupational assimilation with time since migration. Examining individual cohorts, several interesting findings emerge. First, the analytical task usage gap of new arrivals grew between the 1965-69 and 1985-89 cohorts for both men and women, from -0.044 to -0.055 for men, and from -0.039 to -0.075 for women. As analytical task usage is positively related to earnings, this result is consistent with Borjas (2015) which finds that the 1985-1989 immigrant cohort had an especially high initial earnings gap with natives. This analytical task gap remained nearly constant between the 1985-89 and 199599 cohort, but expanded again between the 1995-99 and 2005-09 cohorts. The interactive and manual task usage gaps expanded for both men and women between the 1985-89 and 2005-2009 cohorts. Turning to the patterns in assimilation rates, I find more evidence supporting the results in Borjas (2015). For analytical and interactive tasks for men and women, both the 1965-69 and the 1975-79 cohorts exhibit convergence over 20-25 years since migration. The 198589 cohort, on the other hand, shows convergence in the first 10-15 years since migration in analytical tasks and to a lesser extent interactive tasks for both men and women, and manual tasks for women only, but little change in the subsequent decade. The 1995-99 cohort actually diverged from natives in all three tasks in the first 10-15 years since migration for both men and women. In summary, I find evidence that more recent cohorts have exhibited

14

slower overall occupational assimilation to the average task usage of natives in a given survey year controlling for age and education, and that this result holds for both men and women. Figures 3, 4, and 5 show the gaps for analytical, interactive, and manual for men by cohort and years since migration, while Figures 7, 8, and 9 show the same for women. To ease the interpretation of the results, I combine the task usage gaps into a single task difference index. For each immigrant worker, the index is calculated as follows:

\it ) − (manit − m T Dit = (anait − ana [it ) + (interit − inter \ anit )

where T Dit is the task difference index for worker i in period t, anait −ana [it is the gap between immigrant i’s actual analytical task usage and their predicted analytical task usage, based on their age and education, in period t, and similarly for the interactive and manual tasks. Note that, since immigrants tend to use more manual tasks than natives, and manual task usages are correlated with lower earnings, I subtract the manual task gap when forming the index.13 I show the mean task difference index by cohort and years since migration in Figures 6 and 10 for men and women, respectively. As discussed above, we see the earlier immigrant cohorts converging towards natives in their task usages, while later cohorts, starting with the 1985-89 cohort, actually diverge from natives with time since migration. Furthermore, each cohort following the 1975-79 cohort exhibits a higher initial task difference index shortly after migration. These results are consistent with Borjas (2015), where later cohorts arrive with lower earnings and experience slower earnings assimilation compared to earlier cohorts. These results are driven primarily by events in the 2000s, which is where the task usage divergence mostly takes place. As I investigate above, a potential cause for these findings may be the Great Recession, which may have disproportionately impacted immigrants compared 13

There are a number of methods of calculating this index that are possible. Past work that calculates task distances, including Poletaev and Robinson (2008) and Gathmann and Sch¨onberg (2010), have focused on the transferability of human capital across occupations, in which circumstance either an increase or decrease in task usage means a worker has “moved away” from their prior task, so a Euclidean distance measure is appropriate. In my context, however, I am interested in how far away immigrants are from natives in terms of their tasks, where there is a natural ordering of task importance, based on their correlations with earnings: higher analytical and interactive tasks, and lower manual tasks.

15

to natives. I address this concern in the same manner as above by replacing the 2010 sample with the 2006 sample, and recreating the tables described above.14 Using the 2006 Census sample, we observe the same slowdown in occupational assimilation, and in fact for all three tasks, the gap between natives and immigrants is larger when 2006 is used instead of 2010. Thus, the Great Recession cannot explain the slowdown in occupational asismilation. One of the most notable features of the recent U.S. immigration experience has been the rapid rise of immigrants from Mexico. However, recreating the task assimilation tables separately for immigrants originating from Mexico versus elsewhere, I find that the occupational assimilation slowdown is present for both groups, for men and women.15 While Mexican immigrants have large task usage gaps than non-Mexican immigrants, the large increase in Mexican immigrants in recent decades cannot explain the observed occupational slowdown. The preceding results demonstrate: first, the presence of occupational assimilation for analytical and interactive tasks; and second, that the rate of occupational assimilation has slowed for more recent cohorts, with a notable slowdown during the 2000s. Both results are found for men as well as women. Thus far, these results are based on summary statistics only. To test whether the observed occupational assimilation, as well as the slowdown in occupational assimilation, are statistically significant, I perform a series of regressions where the dependent variable is task usage. I include both natives and immigrants in the estimations. I include observations from the primary sample, which consists of the decennial surveys from 1970 to 2000, and the 2010 survey based on the ACS. The results from these regressions are shown in Table 8 for men and Table 9 for women.16 The dependent variable is analytical in column (1), interactive in column (2), and manual in column (3). The primary independent variables of interest are years since migration and 14

Tables available upon request. Due to the change in years, I adjust the cohorts used in the analysis. Instead of the 2005-09 cohort, I use the 2000-04 cohort, but leave the other cohorts the same. In addition, for the 1985-89 cohort, the third year range is no longer 20-25 years since migration but 15-20 years, while for the 1995-99 cohort, 10-15 years since migration is now 5-10 years. 15 Tables available upon request. 16 Since my aggregated cohorts contain only 12 levels, standard errors clustered at the cohort level may be too small, leading to over-rejection of null hypotheses. Thus, I cluster standard errors at the year of arrival instead, which expands the number of cluster levels to 62.

16

years since migration interacted with immigrant cohort. Note that I divide the years since migration as well as the interactions between years since migration and the cohorts by ten to make the results clearer. I show only variables related to years since migration, though all specifications include controls for survey year, cohort, age, age squared, age cubed, and five education dummies, and I interact the education and age effects by survey year.17 I omit the 1965-69 cohort as the reference group for the years since migration and cohort interaction, which is the cohort after which Borjas (2015) shows earnings assimilation rates appear to be declining. I also aggregate the final two cohorts into a single, 2000-2009 cohort. The regression results are, qualitatively, very similar for men and women. The summary statistics results suggest that the task usage gap between natives and immigrants declines with years since migration, but that this rate of decline has slowed for more recent cohorts. First, note that for both men and women, the years since migration coefficient is positive for analytical and interactive task usages, and negative for manual task usage, and is statistically significant in all models. Since immigrants on average start with lower analytical and interactive task usages and higher manual task usage than natives, these results imply that the gap closes with time since migration. Second, the effect of years since migration varies substantially by cohort. Consider, for example, the interaction between the 1995-99 cohort and years since migration for analytical tasks in column (1) for men, which has a value of -0.024 and is statistically significant at the 0.1% level. This result says that the change in analytical task usage after an additional ten years since migration for the 1995-99 cohort is 0.024 less than the change for the 1965-69 cohort. Given this value, and considering that the coefficient on years since migration (divided by ten) is 0.015, years since migration actually has a negative impact on analytical task usage for this cohort. In fact, as we move away from the 1965-69 cohort, the impact of years since migration declines with almost every successive cohort for all three tasks for both men and women. 17

I also performed estimations where I controlled for state of residence interacted with survey year and birthplace, and the results were qualitatively similar to those presented here. These results, as well as the full estimation results with all independent variables, are available upon request.

17

The overall effect of years since migration on task usages is similar between men and women for analytical and interactive tasks, while women seem to assimilate more rapidly than men in terms of manual tasks. This more rapid manual task usage assimilation for women may correspond to the significantly larger initial manual task usage gap between native and immigrant women providing a greater scope for change with years since migration. The rate of the slowdown seems to vary by gender, with men experiencing a more rapid decline than women. For instance, considering the interaction between YSM and the 1995-1999 cohort for the manual task usage between men and women, we find values of 0.027 for men and 0.014 for women, indicating that the slowdown has been more significant for men than women. Furthermore, while the interaction terms for men are statistically significant for almost all tasks and cohorts, for women many of the interactions are insignificant. In fact, no slowdown in analytical task assimilation is observed for women until the 1990-1995 cohort. Thus, while there is strong evidence of a slowdown in occupational assimilation across all three tasks for both men and women, the size of the slowdown appears to the smaller for women than for men. The regression results support the summary statistics that a slowdown in occupational assimilation, across all three task usages, has occurred for more recent immigrant cohorts. In the following section, I investigate potential causes of this phenomenon.

6

Explaining the Slowdown in Occupational Assimilation

Borjas (2015) provides two (related) explanations for the apparent slowdown in earnings assimilation for more recent cohorts. First, fewer immigrants in more recent cohorts are becoming proficient in English in the years following arrival to the U.S. Since language

18

proficiency is known to be strongly linked to the economic success of immigrants,18 a failure to become proficient can help explain poor earnings growth. Second, belonging to a large national origin or language group may reduce an immigrant’s incentive to acquire English proficiency and other country-specific forms of human capital, as they have a large “audience” to communicate with and have less need to interact with those outside of their particular group. Thus, growth in the size of an immigrant’s conational or language group in the U.S. may slow assimilation rates. In fact, Borjas (2015) estimates that approximately a third of the slowdown in earnings assimilation of more recent cohorts can be attributed to an increase in the size of conational groups. I begin investigating these hypotheses by re-creating the task usage assimilation tables presented earlier, where now I separate immigrants who originate from countries that are major sources of immigrants to the U.S. (“large” countries) from those who originate from all other countries (“small” countries). The countries categorized as large are: Mexico, El Salvador, Guatemala, Cuba, Dominican Republic, China, Korea, Philippines, Vietnam, and India.19 In 1970, 26.1% of immigrants in my sample (including men and women) originated from one of these countries, while this number increased to 59.8% in 2010. The results for the small and large countries for men are shown in Tables 10 and 11, and for women are shown in Tables 12 and 13. Several interesting patterns emerge. First, for both men and women, immigrants originating from larger sending countries show much larger task usage gaps for all cohorts and times since migration. For men, within each country of origin group, we see an inconsistent pattern in the size of the initial task usage gap, aside from a large expansion between the 1995-99 and 2005-09 cohorts observed for both groups. Thus, it appears that, aside from the 1995-99 to 2005-09 cohort expansion, we can attribute the overall increasing initial task usage gap for men to a compositional change away from small country of origin toward large country of origin. For women, both small 18

See Chiswick and Miller (2010), Bleakley and Chin (2004), and Dustmann and van Soest (2001), among many others. 19 These countries match those shown in Borjas (2015), Table 6.

19

and large country of origin immigrants have experienced increases in their initial task usage gaps. For both men and women, however, both country of origin groups have experienced occupational assimilation for earlier cohorts, which slowed and actually reversed for later cohorts. Thus, while the change in the proportion of immigrants from the large sending countries can explain a portion of the initial task usage gap expansion of earlier cohorts, it cannot explain the reversal in occupational assimilation that we observe. I continue investigating the cause of the occupational assimilation slowdown by extending the regression model described above. First, I include a control for the respondent’s proficiency in speaking English, which takes values only English, very well, well, and not well/does not speak English. As language proficiency information is unavailable in the 1970 Census, I omit that survey from the estimations. I include two measures for the size of an immigrant’s enclave.20 I calculate the fraction of individuals in the same state originating from the same source country as the respondent, as well as the fraction who speak the same language, in the current survey year. These variables are denoted as country share and language share, respectively. I set the share of conationals to zero for natives, and I set the share of language speakers to zero for those who report speaking English at home or who originate from largely English-speaking countries.21 The mean fraction of conationals for immigrants equals 0.030 for men and 0.024 for women, while the mean fraction of the state with the same first language for non-English speaking immigrants originating from a non-English-speaking country equals 0.091 for men and 0.076 for women. All other variables included in the previous estimations are included in these estimations, though I display only the years since migration variables, language proficiency controls, and enclave measures.22 Results from these estimations are shown in Tables 14 and 15 for men and women, 20

Here, I differ from Borjas (2015), whose results considered the effective size of a national group, and whose regression specification included as an observation unit an immigrant cohort, defined by age at arrival, source country, and year of arrival. 21 These include Canada, Bermuda, Belize-British Honduras, Jamaica, Antigua-Barbuda, Bahamas, Barbados, Dominica, Grenada, St. Kitts-Nevis, St. Vincent, Trinidad and Tobago, Guyana/British Guiana, the United Kingdom, Ireland, Northern Ireland, Liberia, South Africa, Australia, and New Zealand. 22 Full estimation results are available upon request.

20

respectively. As before, the dependent variables are analytical, interactive, and manual task usages.23 For comparison, I repeat the main analysis from the previous regressions since the 1970 survey is no longer included. These results are shown in columns (1), (4), and (7) for analytical, interactive, and manual tasks, respectively. All of the results from the full sample regressions continue to hold when the 1970 Census is omitted. Columns (2), (5), and (8) add the language proficiency controls, where the omitted category is only English. Higher English language proficiency is positively related to analytical and interactive task usage, and negatively related to manual task usage. For all three tasks, the language coefficients are larger in absolute value for women than for men. For example, the coefficient for speaking English “Not Well” in the interactive task estimations shown in column (5) equals -0.061 for men and -0.104 for women. Controlling for language proficiency reduces (in absolute value) the coefficients on most of the years since migration variables.24 The differences are statistically significant at the 0.1% level for the non-interacted YSM variable for both men and women. For example, the years since migration (divided by ten) coefficient for the interactive task usage for men declines from 0.015 to 0.011 with the inclusion of language proficiency. These results are consistent with an improvement in language proficiency being an important channel through which immigrants assimilate occupationally. If a failure of more recent immigrants to become proficient in English at the same rates as past cohorts can explain part of the decline in occupational assimilation, then we would anticipate a direct control for language ability to reduce the size of the YSM and cohort interaction terms, which is what we observe. For instance, the negative interaction between YSM and the 1985-89 cohort in the interactive estimation for men declines in absolute value from -0.020 to -0.017. For men, all of the YSM and cohort interactions are statistically 23

As with the previous regressions, I estimated versions where I control for state of residence interacted with year and birthplace, and the results are qualitatively the same. 24 The exceptions are the YSM and 1975-1979 and 1980-1985 cohort interactions for analytical for women, which show small increases in absolute value when language is controlled for. However, these variables are not significantly different from zero.

21

significantly different between the specification with and without language proficiency controls for all three tasks at the 1% level, except for the 1970-75 cohort interactions which are significantly different at the 5% level, and the 2000-2009 cohort interactions which are significantly different at the 10% level for analytical and interactive, and not significantly different for manual. For women, the decline in the (non-interacted) YSM coefficient is significant at the 1% level for all three tasks. Nearly all of the YSM and cohort interactions are also statistically significantly different at the 1% level between the two models, especially for later cohorts with the exception of the 2000-2009 cohort.25 Thus, these results are consistent both with language proficiency acquisition being an important channel of occupational assimilation, as well as a slowdown in language proficiency acquisition being at least partly the cause of the slowdown in occupational assimilation of more recent cohorts. I then add the ethnic and language enclave measures in addition to the English language proficiency controls, with the results shown in columns (3), (6), and (9). Having a larger share of state residents from the same source country decreases interactive task usage for men and women and increases analytical and manual task usages for women. Having a larger share of the state that speaks the same language reduces analytical task usage for both men and women, reduces interactive task usage for women, and increases manual task usage for men and decreases manual task usage for women. Overall, these results suggest that living in either a larger ethnic or linguistic enclave enlarges the task usage gap between natives and immigrants, with ethnic enclave and analytical task usage and language enclave and manual task usage for women being the only exceptions. Controlling for enclave sizes actually increases (in absolute terms) the effect of years since migration on all three task usages for men and women. This change is likely due to the positive relationship between years since migration and the size of both the linguistic and ethnic enclave size of an immigrant.26 The YSM and cohort interaction terms decline 25

A full set of statistical significance tests between the models are available online. This is partly due to the increase in immigrants overall, but the relationship between years since migration and both measures of enclave size persist when survey year is controlled for. Results are available upon request. 26

22

in some cases, though the changes in magnitude are minor. This suggests that the large increases in both language and ethnic enclave sizes observed over the past several decades can explain only a small portion of the decline in occupational assimilation. Language proficiency, therefore, dominates enclave size in explaining the occupational assimilation slowdown of recent immigrants.

7

Conclusion

In this paper, I compare native and immigrant occupational attainment for both men and women, where occupations are characterized by the levels of analytical, interactive, and manual tasks performed. Immigrants on average perform fewer analytical and interactive tasks and more manual tasks than natives, and this gap has grown significantly in the past few decades. Even controlling for age and education, the gap remains, has expanded consistently over time for women, and saw a large expansion between 2000 and 2010 for both men and women. Earlier immigrant cohorts tend to assimilate occupationally to natives with time since migration. However, newer immigrant cohorts have shown not only significantly slower occupational assimilation rates compared to previous cohorts, but in some cases divergence from natives in terms of occupational attainment. Controlling for English language proficiency can explain some of the observed relationship between time since migration and task usage, implying that improvement in proficiency may be an important dimension of the occupational assimilation process. Also, controlling for English proficiency helps to explain a portion of the occupational assimilation slowdown experience by more recent cohorts. However, much of the slowdown remains unexplained. My results support those found in recent work by Borjas (2015), which argues that newer immigrant cohorts have both higher initial earnings gaps compared to natives than past cohorts, and also show significantly slower earnings growth with time since migration. I

23

find that immigrants are falling behind natives in the occupational tasks that have not only historically been more highly rewarded (analytical and interactive), but that have also seen large increases in economic returns over the past several decades.27 While it is beyond the scope of this paper to assess the direct relationship between the observed slowdown in occupational assimilation and the slowdown in earnings assimilation found in Borjas (2015), the direction of change is nonetheless consistent. Thus, the results presented here provide additional evidence that, not only have recent immigrant cohorts performed more poorly economically immediately following arrival compared to past cohorts, but their economic assimilation rates have also slowed, if not stopped altogether.

27

See Scotese (2012).

24

.1 Task Usage Gap −.05 0 .05 −.1 −.15 1970

1980

1990 Census year

Analytical

2000

Interactive

2010 Manual

−.15

−.1

Task Usage Gap −.05 0 .05

.1

Figure 1: Task Usage Gaps by Year, Men Only

1970

1980

1990 Census year

Analytical

Interactive

2000

2010 Manual

Figure 2: Task Usage Gaps by Year, Women Only

25

0 Task Usage Gap −.06 −.04 −.02 −.08 −.1 0−5

10−15 Years Since Migration 1965−1969 1995−1999

1975−1979 2005−2009

20−25 1985−1989

Task Usage Gap −.14 −.12 −.1 −.08 −.06 −.04 −.02

0

Figure 3: Analytical Task Assimilation by Cohort, Men Only

0−5

10−15 Years Since Migration 1965−1969 1995−1999

1975−1979 2005−2009

20−25 1985−1989

Figure 4: Interactive Task Assimilation by Cohort, Men Only

26

.1 .08 Task Usage Gap .04 .06 .02 0 0−5

10−15 Years Since Migration 1965−1969 1995−1999

1975−1979 2005−2009

20−25 1985−1989

−.2

Task Difference Index −.16 −.12 −.08 −.04

0

Figure 5: Manual Task Assimilation by Cohort, Men Only

0−5

10−15 Years Since Migration 1965−1969 1995−1999

1975−1979 2005−2009

20−25 1985−1989

Figure 6: Task Difference Index by Cohort, Men Only

27

0 Task Usage Gap −.06 −.04 −.02 −.08 −.1 0−5

10−15 Years Since Migration 1965−1969 1995−1999

1975−1979 2005−2009

20−25 1985−1989

Task Usage Gap −.14 −.12 −.1 −.08 −.06 −.04 −.02

0

Figure 7: Analytical Task Assimilation by Cohort, Women Only

0−5

10−15 Years Since Migration 1965−1969 1995−1999

1975−1979 2005−2009

20−25 1985−1989

Figure 8: Interactive Task Assimilation by Cohort, Women Only

28

.1 .08 Task Usage Gap .04 .06 .02 0 0−5

10−15 Years Since Migration 1965−1969 1995−1999

1975−1979 2005−2009

20−25 1985−1989

Task Difference Index −.4 −.36−.32−.28−.24 −.2 −.16−.12−.08−.04 0

Figure 9: Manual Task Assimilation by Cohort, Women Only

0−5

10−15 Years Since Migration 1965−1969 1995−1999

1975−1979 2005−2009

20−25 1985−1989

Figure 10: Task Difference Index by Cohort, Women Only

29

Task Bundle Analytical

Interactive Manual

Table 1: Task Bundle Definitions O*NET Variables Fluency of Ideas, Originality, Problem Sensitivity, Deductive Reasoning, Inductive Reasoning, Information Ordering, Category Flexibility, Mathematical Reasoning, Number Facility, Memorization, Speed of Closure, Flexibility of Closure Oral Comprehension, Written Comprehension, Oral Expression, Written Expression Arm-Hand Steadiness, Manual Dexterity, Finger Dexterity, Control Precision, Multilimb Coordination, Response Orientation, Rate Control, Reaction Time, Wrist-Finger Speed, Speed of Limb Movement, Static Strength, Explosive Strength, Dynamic Strength, Trunk Strength, Stamina, Extent Flexibility, Dynamic Flexibility, Gross Body Coordination, Gross Body Equilibrium

30

Table 2: Task Usage by Year, Overall, Men Only (1) (2) (3) (4) (5) 1970 1980 1990 2000 2010 Mean Mean Mean Mean Mean Analytical Immigrants Natives Difference Interactive Immigrants Natives Difference Manual Immigrants Natives Difference Observations

0.430 0.458 -0.028

0.439 0.481 -0.042

0.431 0.484 -0.053

0.424 0.492 -0.069

0.408 0.499 -0.091

0.371 0.417 -0.046

0.379 0.436 -0.057

0.371 0.436 -0.065

0.359 0.449 -0.090

0.346 0.464 -0.118

0.590 0.580 0.009

0.590 0.562 0.028

0.585 0.554 0.032

0.590 0.540 0.050

0.600 0.526 0.074

998283

1948490

2319385

2655924

1555253

Source: 1970, 1980, 1990, 2000, and 2010 IPUMS Census.

31

Table 3: Task Usage by Year, (1) (2) 1970 1980 Mean Mean Analytical Immigrants Natives Difference Interactive Immigrants Natives Difference Manual Immigrants Natives Difference Observations

Overall, Women Only (3) (4) (5) 1990 2000 2010 Mean Mean Mean

0.346 0.387 -0.041

0.375 0.437 -0.062

0.402 0.475 -0.073

0.418 0.498 -0.080

0.414 0.518 -0.103

0.359 0.445 -0.086

0.378 0.494 -0.116

0.400 0.521 -0.121

0.413 0.548 -0.135

0.410 0.567 -0.156

0.526 0.449 0.076

0.525 0.415 0.110

0.505 0.400 0.105

0.501 0.385 0.115

0.500 0.383 0.117

646823

1465350

1991227

2375274

1445782

Source: 1970, 1980, 1990, 2000, and 2010 IPUMS Census.

32

Table 4: Immigrant Task Usage Gap by Year, Men Only (1) (2) (3) (4) (5) 1970 1980 1990 2000 2010 mean mean mean mean mean Analytical Interactive Manual

-0.036 -0.057 0.015

-0.033 -0.049 0.016

-0.034 -0.046 0.009

-0.040 -0.056 0.015

-0.059 -0.080 0.036

Observations

35844

101869

160144

285055

187671

Note: Task usage gaps are differences between actual and predicted task usages. Source: 1970, 1980, 1990, 2000, and 2010 IPUMS Census.

33

Table 5: Immigrant Task Usage Gap (1) (2) (3) 1970 1980 1990 mean mean mean

by Year, Women Only (4) (5) 2000 2010 mean mean

Analytical Interactive Manual

-0.031 -0.045 -0.067 -0.087 0.053 0.075

-0.047 -0.085 0.069

-0.050 -0.092 0.077

-0.069 -0.113 0.085

Observations

26921

121792

204334

152263

79205

Note: Task usage gaps are differences between actual and predicted task usages. Source: 1970, 1980, 1990, 2000, and 2010 IPUMS Census.

34

Table 6: Immigrant Task Usage Gaps, by Cohort and Time Since Migration, Men Only (1) (2) (3) (4) (5) (6) All 65 75 85 95 05 mean mean mean mean mean mean Analytical 0-5 years 10-15 years 20-25 years Interactive 0-5 years 10-15 years 20-25 years Manual 0-5 years 10-15 years 20-25 years Observations

-0.057 -0.048 -0.033

-0.044 -0.042 -0.030 -0.029 -0.008 -0.020

-0.055 -0.046 -0.051

-0.053 -0.062 .

-0.069 . .

-0.074 -0.067 -0.050

-0.065 -0.053 -0.053 -0.044 -0.031 -0.036

-0.060 -0.058 -0.065

-0.073 -0.086 .

-0.092 . .

0.020 0.025 0.018

0.019 0.017 0.007

0.013 0.006 0.006

0.013 0.018 0.030

0.016 0.041 .

0.029 . .

377642

44657

86999

118519

98930

28537

Note: Task usage gaps are differences between actual and predicted task usages. Note 2: Each column represents an immigrant cohort in five-year groups. Source: 1970, 1980, 1990, 2000, and 2010 IPUMS Census.

35

Table 7: Immigrant Task Usage Gaps, by Cohort and Time Since Migration, Women Only (1) (2) (3) (4) (5) (6) All 65 75 85 95 05 mean mean mean mean mean mean Analytical 0-5 years 10-15 years 20-25 years Interactive 0-5 years 10-15 years 20-25 years Manual 0-5 years 10-15 years 20-25 years Observations

-0.076 -0.059 -0.042

-0.039 -0.062 -0.075 -0.072 -0.038 -0.043 -0.054 -0.075 -0.023 -0.033 -0.054 .

-0.094 . .

-0.119 -0.103 -0.082

-0.089 -0.106 -0.109 -0.117 -0.085 -0.085 -0.096 -0.121 -0.060 -0.076 -0.094 .

-0.138 . .

0.086 0.078 0.072

0.072 0.076 0.058

0.085 0.065 0.067

0.086 0.079 0.080

0.084 0.083 .

0.094 . .

271004

37904

62504

84170

66300

20126

Note: Task usage gaps are differences between actual and predicted task usages. Note 2: Each column represents an immigrant cohort in five-year groups. Source: 1970, 1980, 1990, 2000, and 2010 IPUMS Census.

36

Table 8: OLS Regressions, Task Usages, Men Only (1) (2) (3) Analytical Interactive Manual 0.015∗∗∗ (17.98)

0.014∗∗∗ (15.86)

-0.003∗∗∗ (-5.77)

YSM*50/10

0.002 (1.03)

0.004∗∗ (2.38)

-0.000 (-0.24)

YSM*70/10

-0.004∗∗ (-2.49)

-0.007∗∗∗ (-4.14)

0.003∗∗∗ (3.23)

YSM*75/10

-0.009∗∗∗ (-6.79)

-0.013∗∗∗ (-9.11)

0.007∗∗∗ (9.09)

YSM*80/10

-0.012∗∗∗ (-8.27)

-0.016∗∗∗ (-11.55)

0.010∗∗∗ (8.66)

YSM*85/10

-0.015∗∗∗ (-4.10)

-0.019∗∗∗ (-5.57)

0.013∗∗∗ (4.20)

YSM*90/10

-0.018∗∗∗ (-8.68)

-0.019∗∗∗ (-8.17)

0.014∗∗∗ (5.18)

YSM*95/10

-0.024∗∗∗ (-7.20)

-0.026∗∗∗ (-6.32)

0.027∗∗∗ (6.93)

YSM*00/10

-0.035∗∗ (-2.03)

-0.045∗∗∗ (-2.79)

0.033 (1.45)

Observations R2

9477335 0.279

9477335 0.353

9477335 0.321

YSM/10

Notes: Dependent variables are analytical (1), interactive (2), and manual (3) task usages. All regressions include a constant term, survey year dummies, cohort dummies, age, age squared, age cubed, and five education dummies, and the education and age effects are interacted with survey year. “YSM” refers to the years since migration of an immigrant. Standard errors clustered at the immigrant year of arrival level. Estimation includes 1970, 1980, 1990, 2000, and 2010 Census samples, where 2010 sample is composed of the three-year 2009-2011 ACS sample. Sample consists of natives and immigrants. t statistics in parentheses. ∗ p < 0.10, ∗∗ p < 0.05, ∗∗∗ p < 0.01

37

Table 9: OLS Regressions, Task Usages, Women Only (1) (2) (3) Analytical Interactive Manual YSM/10

0.010∗∗∗ (13.41)

0.017∗∗∗ (18.92)

-0.009∗∗∗ (-7.81)

YSM*50/10

-0.001 (-0.46)

0.001 (0.65)

0.000 (0.34)

YSM*70/10

0.002∗ (1.69)

0.000 (0.41)

-0.004∗∗∗ (-3.53)

YSM*75/10

0.003∗∗ (2.23)

-0.003∗∗∗ (-7.21)

0.004∗∗∗ (2.82)

YSM*80/10

0.002 (0.92)

-0.006∗∗ (-2.27)

0.008∗∗∗ (4.04)

YSM*85/10

-0.002 (-0.64)

-0.011∗∗∗ (-3.35)

0.009∗∗∗ (7.97)

YSM*90/10

-0.012∗∗∗ (-5.72)

-0.016∗∗∗ (-6.17)

0.011∗∗∗ (7.95)

YSM*95/10

-0.011 (-1.54)

-0.020∗∗ (-2.19)

0.014∗∗ (2.07)

YSM*00/10

-0.024 (-1.36)

-0.046∗∗ (-2.16)

0.038∗ (1.73)

Observations R2

7924456 0.265

7924456 0.309

7924456 0.115

Notes: Dependent variables are analytical (1), interactive (2), and manual (3) task usages. All regressions include a constant term, survey year dummies, cohort dummies, age, age squared, age cubed, and five education dummies, and the education and age effects are interacted with survey year. “YSM” refers to the years since migration of an immigrant. Standard errors clustered at the immigrant year of arrival level. Estimation includes 1970, 1980, 1990, 2000, and 2010 Census samples, where 2010 sample is composed of the three-year 2009-2011 ACS sample. Sample consists of natives and immigrants. t statistics in parentheses. ∗ p < 0.10, ∗∗ p < 0.05, ∗∗∗ p < 0.01

38

Table 10: Immigrant Task Usage Gaps, by Cohort and Time Since Migration, Small Countries, Men Only (1) (2) (3) (4) (5) (6) All 65 75 85 95 05 mean mean mean mean mean mean Analytical 0-5 years 10-15 years 20-25 years Interactive 0-5 years 10-15 years 20-25 years Manual 0-5 years 10-15 years 20-25 years Observations

-0.050 -0.048 -0.033

-0.025 -0.009 -0.026 -0.025 -0.021 -0.010 -0.024 -0.042 0.003 -0.001 -0.030 .

-0.048 . .

-0.065 -0.067 -0.050

-0.050 -0.023 -0.037 -0.047 -0.043 -0.029 -0.040 -0.065 -0.020 -0.023 -0.045 .

-0.061 . .

0.015 0.025 0.018

0.008 -0.009 -0.001 0.011 -0.009 0.004 -0.001 -0.011 0.014

0.000 0.026 .

0.014 . .

26423

40689

12135

361240

39878

48926

Note: Task usage gaps are differences between actual and predicted task usages. Note 2: Each column represents an immigrant cohort in five-year groups. Note 3: Sample contains immigrants from smaller sending countries. Source: 1970, 1980, 1990, 2000, and 2010 IPUMS Census.

39

Table 11: Immigrant Task Usage Gaps, by Cohort and Time Since Migration, Large Countries, Men Only (1) (2) (3) (4) (5) (6) All 65 75 85 95 05 mean mean mean mean mean mean Analytical 0-5 years 10-15 years 20-25 years Interactive 0-5 years 10-15 years 20-25 years Manual 0-5 years 10-15 years 20-25 years Observations

-0.058 -0.048 -0.033

-0.081 -0.079 -0.089 -0.075 -0.042 -0.044 -0.060 -0.073 -0.021 -0.033 -0.064 .

-0.083 . .

-0.076 -0.067 -0.050

-0.095 -0.088 -0.087 -0.095 -0.068 -0.056 -0.070 -0.099 -0.045 -0.047 -0.077 .

-0.111 . .

0.021 0.025 0.018

0.040 0.026 0.017

0.037 0.019 0.018

0.030 0.027 0.040

0.029 0.050 .

0.039 . .

365507

18234

47121

69593

58241

16402

Note: Task usage gaps are differences between actual and predicted task usages. Note 2: Each column represents an immigrant cohort in five-year groups. Note 3: Sample contains immigrants from larger sending countries. Source: 1970, 1980, 1990, 2000, and 2010 IPUMS Census.

40

Table 12: Immigrant Task Usage Gaps, by Cohort and Time Since Migration, Small Countries, Women Only (1) (2) (3) (4) (5) (6) All 65 75 85 95 05 mean mean mean mean mean mean Analytical 0-5 years 10-15 years 20-25 years Interactive 0-5 years 10-15 years 20-25 years Manual 0-5 years 10-15 years 20-25 years Observations

-0.070 -0.059 -0.042

-0.028 -0.042 -0.054 -0.051 -0.028 -0.028 -0.030 -0.053 -0.016 -0.013 -0.027 .

-0.083 . .

-0.110 -0.103 -0.082

-0.066 -0.068 -0.076 -0.084 -0.065 -0.060 -0.062 -0.087 -0.043 -0.045 -0.059 .

-0.112 . .

0.082 0.078 0.072

0.056 0.064 0.046

0.051 0.036 0.039

0.059 0.057 0.062

0.063 0.069 .

0.078 . .

260020

23257

29192

37667

31085

9142

Note: Task usage gaps are differences between actual and predicted task usages. Note 2: Each column represents an immigrant cohort in five-year groups. Note 3: Sample contains immigrants from smaller sending countries. Source: 1970, 1980, 1990, 2000, and 2010 IPUMS Census.

41

Table 13: Immigrant Task Usage Gaps, by Cohort and Time Since Migration, Large Countries, Women Only (1) (2) (3) (4) (5) (6) All 65 75 85 95 05 mean mean mean mean mean mean Analytical 0-5 years 10-15 years 20-25 years Interactive 0-5 years 10-15 years 20-25 years Manual 0-5 years 10-15 years 20-25 years Observations

-0.075 -0.059 -0.042

-0.060 -0.083 -0.097 -0.094 -0.055 -0.058 -0.073 -0.092 -0.034 -0.050 -0.073 .

-0.102 . .

-0.120 -0.103 -0.082

-0.134 -0.145 -0.145 -0.150 -0.116 -0.108 -0.123 -0.148 -0.086 -0.101 -0.119 .

-0.158 . .

0.088 0.078 0.072

0.104 0.094 0.076

0.120 0.091 0.091

0.114 0.096 0.093

0.104 0.093 .

0.106 . .

261862

14647

33312

46503

35215

10984

Note: Task usage gaps are differences between actual and predicted task usages. Note 2: Each column represents an immigrant cohort in five-year groups. Note 3: Sample contains immigrants from larger sending countries. Source: 1970, 1980, 1990, 2000, and 2010 IPUMS Census.

42

43 -0.029∗∗ (-2.13) -0.003∗∗∗ (-3.27) -0.045∗∗∗ (-5.83) -0.066∗∗∗

0.002 (0.46) -0.004∗∗ (-2.24) -0.008∗∗∗ (-4.97) -0.010∗∗∗ (-5.74) -0.012∗∗∗ (-3.46) -0.015∗∗∗ (-6.76) -0.021∗∗∗ (-6.42) -0.031∗∗ (-2.10) -0.004∗∗ (-2.06) -0.051∗∗∗ (-5.50) -0.074∗∗∗

0.002 (0.72) -0.005∗∗ (-2.53) -0.010∗∗∗ (-6.07) -0.013∗∗∗ (-7.25) -0.016∗∗∗ (-4.16) -0.018∗∗∗ (-8.11) -0.025∗∗∗ (-7.13) -0.036∗∗ (-2.08)

YSM*50/10

YSM*70/10

YSM*75/10

YSM*80/10

YSM*85/10

YSM*90/10

YSM*95/10

YSM*00/10

Very Well

Well

Not Well

-0.020∗∗∗ (-5.86)

-0.015∗∗∗ (-6.77)

-0.011∗∗∗ (-3.16)

-0.009∗∗∗ (-5.68)

-0.007∗∗∗ (-5.11)

-0.004∗∗ (-1.99)

0.001 (0.38)

0.012∗∗∗ (9.64)

0.010∗∗∗ (8.21)

0.016∗∗∗ (12.53)

YSM/10

(3)

(2)

(1)

Analytical

-0.047∗∗∗ (-2.87)

-0.027∗∗∗ (-6.43)

-0.020∗∗∗ (-7.95)

-0.020∗∗∗ (-5.67)

-0.017∗∗∗ (-9.76)

-0.014∗∗∗ (-8.00)

-0.008∗∗∗ (-4.14)

0.001 (0.32)

0.015∗∗∗ (10.66)

(4)

-0.061∗∗∗

-0.045∗∗∗ (-3.61)

-0.001 (-0.17)

-0.042∗∗∗ (-3.02)

-0.024∗∗∗ (-6.06)

-0.017∗∗∗ (-7.23)

-0.017∗∗∗ (-5.18)

-0.014∗∗∗ (-9.58)

-0.012∗∗∗ (-7.79)

-0.007∗∗∗ (-3.91)

-0.000 (-0.07)

0.011∗∗∗ (8.43)

(5)

Interactive

-0.058∗∗∗

-0.043∗∗∗ (-3.73)

-0.001 (-0.17)

-0.042∗∗∗ (-3.05)

-0.024∗∗∗ (-5.96)

-0.018∗∗∗ (-7.32)

-0.017∗∗∗ (-5.14)

-0.014∗∗∗ (-9.92)

-0.012∗∗∗ (-7.95)

-0.007∗∗∗ (-3.74)

-0.001 (-0.28)

0.012∗∗∗ (9.75)

(6)

0.035 (1.56)

0.030∗∗∗ (7.23)

0.016∗∗∗ (5.62)

0.016∗∗∗ (4.66)

0.012∗∗∗ (7.53)

0.010∗∗∗ (6.76)

0.006∗∗∗ (3.73)

-0.003 (-1.53)

-0.006∗∗∗ (-4.16)

(7)

0.046∗∗∗

0.031∗∗∗ (3.09)

-0.005∗ (-1.87)

0.032 (1.54)

0.027∗∗∗ (6.98)

0.014∗∗∗ (4.98)

0.013∗∗∗ (4.11)

0.010∗∗∗ (6.14)

0.008∗∗∗ (5.94)

0.005∗∗∗ (3.50)

-0.002 (-1.25)

-0.002 (-1.37)

(8)

Manual (9)

0.038∗∗∗

0.026∗∗∗ (3.19)

-0.005∗∗ (-2.43)

0.031 (1.53)

0.026∗∗∗ (6.53)

0.014∗∗∗ (4.81)

0.012∗∗∗ (3.80)

0.009∗∗∗ (5.86)

0.008∗∗∗ (5.18)

0.005∗∗∗ (3.15)

-0.002 (-1.04)

-0.004∗∗∗ (-2.69)

Table 14: OLS Regressions, Task Usages, with Language Proficiency and Group Shares, Men Only

44

8479052 0.357

8479052 0.358

8479052 0.320

8479052 0.321

(3.71)

8479052 0.321

0.121∗∗∗ (7.16)

0.042 (1.30)

(3.61)

Notes: Dependent variables are analytical (1-3), interactive (4-6), and manual (7-9) task usages. All regressions include a constant term, survey year dummies, cohort dummies, age, age squared, age cubed, and five education dummies, and the education and age effects are interacted with survey year. Country (language) shares are the fraction of state residents from the same country of origin (speak the same language). Standard errors clustered at the immigrant year of arrival level. Estimation includes 1980, 1990, 2000, and 2010 Census samples, where 2010 sample is composed of the three-year 2009-2011 ACS sample. Sample consists of natives and immigrants. t statistics in parentheses. ∗ p < 0.10, ∗∗ p < 0.05, ∗∗∗ p < 0.01

8479052 0.358

8479052 0.283

8479052 0.283

Observations R2

8479052 0.281

-0.004 (-0.40)

-0.142∗∗∗ (-13.25)

Language Share

(-4.30) -0.150∗∗∗ (-9.12)

(-4.21)

-0.013 (-0.70)

(-6.90)

Country Share

(-6.98)

45

0.002 (1.32) -0.001 (-1.11)

0.003 (1.25) -0.000 (-0.16) -0.010∗∗∗ (-4.31) -0.010 (-1.59)

0.003∗∗ (2.06) -0.002∗ (-1.74) -0.000 (-0.28) -0.002 (-0.65) -0.005∗∗ (-2.06) -0.016∗∗∗ (-7.49) -0.014∗∗ (-2.09) -0.028 (-1.57)

YSM*50/10

YSM*70/10

YSM*75/10

YSM*80/10

YSM*85/10

YSM*90/10

YSM*95/10

YSM*00/10

-0.048∗∗∗ (-4.29) -0.079∗∗∗

-0.056∗∗∗ (-4.29) -0.091∗∗∗

Well

Not Well

-0.017 (-1.22)

-0.008 (-1.31)

-0.009∗∗∗ (-4.09)

0.002 (1.30)

-0.019 (-1.29)

-0.010∗∗∗ (-3.57)

0.004∗ (1.90)

-0.041∗∗ (-2.23)

-0.050∗∗ (-2.36)

-0.104∗∗∗

-0.068∗∗∗ (-3.75)

0.005 (0.77)

-0.019∗∗ (-2.27)

-0.013∗∗∗ (-4.79)

-0.009∗∗∗ (-2.89)

-0.005∗∗ (-2.14)

-0.004∗∗∗ (-3.43)

-0.003∗∗ (-2.16)

0.002 (1.33)

0.012∗∗∗ (10.61)

(5)

Interactive

-0.024∗∗∗ (-2.65)

-0.020∗∗∗ (-7.23)

-0.015∗∗∗ (-4.38)

-0.007∗∗∗ (-5.80)

0.003∗∗ (2.14)

0.001 (0.53)

-0.004∗∗ (-2.40)

0.003∗ (1.83)

0.021∗∗∗ (14.31)

(4)

-0.001 (-0.80)

0.002 (1.13)

0.009∗∗∗ (7.74)

(3)

0.002 (0.73)

Very Well

0.007∗∗∗ (6.42)

0.014∗∗∗ (17.11)

YSM/10

0.002 (1.38)

(2)

(1)

Analytical

-0.096∗∗∗

-0.062∗∗∗ (-3.67)

0.005 (0.88)

-0.039∗∗ (-2.25)

-0.018∗∗ (-2.14)

-0.012∗∗∗ (-4.77)

-0.008∗∗∗ (-2.77)

-0.004∗∗ (-2.12)

-0.003∗∗∗ (-2.91)

-0.003∗ (-1.83)

0.002 (1.07)

0.014∗∗∗ (11.48)

(6)

0.043∗ (1.94)

0.018∗∗∗ (2.78)

0.016∗∗∗ (10.64)

0.013∗∗∗ (11.14)

0.013∗∗∗ (6.25)

0.009∗∗∗ (5.59)

0.001 (0.53)

-0.001 (-0.49)

-0.014∗∗∗ (-9.69)

(7)

0.065∗∗∗

0.064∗∗∗

0.039∗∗∗ (3.90)

-0.012∗∗∗ (-2.76)

-0.012∗∗ (-2.64) 0.040∗∗∗ (3.95)

0.037∗ (1.83)

0.015∗∗ (2.49)

0.012∗∗∗ (8.72)

0.009∗∗∗ (8.61)

0.009∗∗∗ (5.92)

0.006∗∗∗ (4.21)

-0.000 (-0.18)

-0.000 (-0.08)

-0.009∗∗∗ (-10.39)

(9)

0.037∗ (1.82)

0.015∗∗ (2.49)

0.012∗∗∗ (8.42)

0.009∗∗∗ (8.40)

0.009∗∗∗ (5.71)

0.007∗∗∗ (4.41)

0.000 (0.01)

-0.000 (-0.20)

-0.008∗∗∗ (-9.25)

(8)

Manual

Table 15: OLS Regressions, Task Usages, with Language Proficiency and Group Shares, Women Only

46

7277633 0.300

7277633 0.303

7277633 0.104

7277633 0.106

(5.08)

7277633 0.106

-0.052∗∗∗ (-3.81)

0.261∗∗∗ (11.02)

(4.95)

Notes: Dependent variables are analytical (1-3), interactive (4-6), and manual (7-9) task usages. All regressions include a constant term, survey year dummies, cohort dummies, age, age squared, age cubed, and five education dummies, and the education and age effects are interacted with survey year. Country (language) shares are the fraction of state residents from the same country of origin (speak the same language). Standard errors clustered at the immigrant year of arrival level. Estimation includes 1980, 1990, 2000, and 2010 Census samples, where 2010 sample is composed of the three-year 2009-2011 ACS sample. Sample consists of natives and immigrants. t statistics in parentheses. ∗ p < 0.10, ∗∗ p < 0.05, ∗∗∗ p < 0.01

7277633 0.303

7277633 0.256

7277633 0.256

Observations R2

7277633 0.253

-0.106∗∗∗ (-5.05)

-0.247∗∗∗ (-12.09)

Language Share

(-4.65) -0.160∗∗∗ (-5.54)

(-4.91)

0.061∗∗ (2.23)

(-5.37)

Country Share

(-5.69)

References ` , A. and A. Ferrer (2014a): “The effect of linguistic proximity on the occupaAdsera tional assimilation of immigrant men,” CLSRN Working Paper No. 144. ——— (2014b): “The Myth of Immigrant Women as Secondary Workers: Evidence from Canada,” American Economic Review Papers and Proceedings, 104, 360–364. Autor, D. H. (2013): “The ”task approach” to labor markets: an overview,” Journal for Labor Market Research. Autor, D. H. and D. Dorn (2013): “The Growth of Low-Skill Service Jobs and the Polarization of the US Labor Market,” American Economic Review, 103, 1553–1597. Bleakley, H. and A. Chin (2004): “Language Skills and Earnings: Evidence from Childhood Immigrants,” Review of Economics and Statistics, 86, 481–496. Borjas, G. J. (1985): “Assimilation, changes in cohort quality, and the earnings of immigrants,” Journal of Labor Economics, 3, 463–489. ——— (2015): “The Slowdown in the Economic Assimilation of Immigrants: Aging and Cohort Effects Revisited Again,” Forthcoming, Journal of Human Capital. Chiswick, B. R. (1978): “The effect of Americanization on the earnings of foreign-born men,” The Journal of Political Economy, 86, 897–921. Chiswick, B. R. and P. W. Miller (2010): “Occupational language requirements and the value of English in the US labor market,” Journal of Population Economics, 23, 353– 372. Cohen-Goldner, S. and Z. Eckstein (2010): “Estimating the return to training and occupational experience : The case of female immigrants,” Journal of Econometrics, 156, 86–105. Dustmann, C. and A. V. Soest (2002): “Language and the Earnings of Immigrants,” Industrial and Labor Relations Review, 55, 473–492. ¨ nberg (2010): “How General Is Human Capital? A TaskGathmann, C. and U. Scho Based Approach,” Journal of Labor Economics, 28, 1–49. Green, D. (1999): “Immigrant occupational attainment: Assimilation and mobility over time,” Journal of Labor Economics, 17, 49–79. Imai, S., D. Stacey, and C. Warman (2014): “From Engineer to Taxi Driver? Language Proficiency and the Occupational Skills of Immigrants,” Working Paper. Isphording, I. E. and S. Otten (2013): “The Costs of Babylon-Linguistic Distance in Applied Economics,” Review of International Economics, 21, 354–369.

47

——— (2014): “Linguistic barriers in the destination language acquisition of immigrants,” Journal of Economic Behavior and Organization, 105, 30–50. Lessem, R. and C. Sanders (2013): “Decomposing the Native-Immigrant Wage Gap in the United States,” Working Paper, 1–27. Ottaviano, G. I. P., G. Peri, and G. C. Wright (2013): “Immigration, Offshoring and American Jobs,” American Economic Review, 103, 1925–59. Peri, G. and C. Sparber (2009): “Task Specialization, Immigration, and Wages,” American Economic Journal: Applied Economics, 1, 135–169. Poletaev, M. and C. Robinson (2008): “Human Capital Specificity: Evidence from the Dictionary of Occupational Titles and Displaced Worker Surveys, 1984-2000,” Journal of Labor Economics, 26, 387–420. Ruggles, S., J. T. Alexander, K. Genadek, R. Goeken, M. B. Schroeder, and M. Sobek (2010): Integrated Public Use Microdata Series: Version 5.0 [Machine-readable database], Minneapolis: University of Minnesota. Scotese, C. A. (2012): “Wage inequality, tasks and occupation,” Working Paper.

48

A

Appendix: Task Usage Assignment

In this section I describe the procedure for matching task data from the O*NET with data from the Census. Additional details and code used are available upon request. Version 17.0 of the O*NET is used in this paper. Occupations in the O*NET are coded with a version of the Standard Occupation Classification (SOC) system. While several years in the Census, including the 2010 and 2011 ACS, also use include SOC codes, there is not a perfect match between the two coding schemes. In order to assign task usages to all observations in the Census beyond the 2010 and 2011 ACS samples, I begin by merging the task usages from the O*NET for each SOC code to the 2010 and 2011 ACS using the full six-digit occupation codes (see occsoc code in the IPUMS). I include both men and women who are employed, and I adjust their survey weights based on the number of weeks worked. The Census data often censors the the SOC occupational code by omitting between the last digit and the last four digits. Those workers with censored SOC occupational codes are thus unmatched during the first round of merging between the O*NET and Census data. For these individuals, I create a new occupational code by dropping the final digit of their SOC occupation code. In the O*NET, the occupational characteristics of new new occupational code is calculated by averaging over all of the occupations contained in that code.28 I then match individuals in the Census left unmatched by the first round using this new occupation code. This process matches a number of additional individuals, but again leaves some unmatched, so the process is repeated, i.e. another digit is dropped, until all individuals are properly matched. This approach allows for the maximum amount of variation to be maintained but, for those individuals in the Census whose occupation codes to not match perfectly to the O*NET codes, I can still assign task usages at a more 28

For example, if a worker’s occupation code in the Census is 51609X, they will be unmatched initially. Thus, I create a new occupation code, 51609. The occupational characteristics for the new occupation coded as 51609 is the average over all occupations in the range of 516090-516090, with equal weight applied to each occupation. In this example, I average over three occupations: 516091 (Extruding and Forming Machine Setters, Operators, and Tenders, Synthetic and Glass Fibers), 516092 (Fabric and Apparel Patternmakers), and 516093 (Upholsterers).

49

aggregate level. Once this merging process is complete, each individual in the 2010 and 2011 ACS samples has a vector of 52 characteristics based on their occupation. Each of these occupational characteristics is rescaled so that they represent the individual’s percentile distribution in the population. So, for example, if 30% of workers in the 2010 Census have lower “Oral Comprehension” scores than a given worker, that worker is assigned a value of 0.3 to their “Oral Comprehension” characteristic. These percentiles also account for the survey weights. Once the individual characteristics are rescaled, I take the average over the characteristics in each task group - analytical, interactive, and manual. Table 1 describes the grouping of occupational characteristics into task bundles. At this point, each individual in the 2010 and 2011 ACS samples has an analytical, interactive, and manual task usage. In order to assign task usages to all survey years, I create a new occupation code similar to the occ1990 variable provided in the Census. While the occ1990 variable allows for consistent comparison across survey years, it is not a balanced panel, since some occupations are used in previous Census years but not used in later periods. For example, prior to the 2000 Census, a high school teacher’s subject (e.g. biology, chemistry) were recorded in a separate occupation code, while starting in the 2000 Census, these occupations are aggregated into the “Subject instructors (HS/college)” occupation. So instead of relying on the occ1990 variable, I construct an alternative occupation variable similar to the one used in Autor and Dorn (2013) that is balanced across the Census samples.29 I refer to this variable as occ1990alt.30 I then take the mean task usage, weighted by survey weights, within each occ1990alt occupation code. The result of this procedure is that each occ1990alt occupation is now assigned an analytical, interactive, and manual task usage, which allows me to merge task usages across the surveys. This procedure assumes that changes in task usage over time 29

My thanks to David Dorn for providing access to the crosswalks. Small modifications were made to the original crosswalks download from Dorn’s website. In addition, I create a 2010 ACS crosswalk, since the website provides only a 2005 ACS crosswalk, and there were small changes in the ACS occupation codes starting with the 2010 ACS. 30

50

result from changes in distribution of workers across occupations, and not changes in task content for each occupation.

51