write summary one paragraph for each article

A Nation of Immigrants: Assimilation and Economic Outcomes in the Age of Mass Migration

Ran Abramitzky

Stanford University and National Bureau of Economic Research

Leah Platt Boustan

University of California, Los Angeles, and National Bureau of Economic Research

Katherine Eriksson

California Polytechnic State University

We Searc

[ Journa © 2014

All us

During the Age of Mass Migration ð1850–1913Þ, the United States main- tained an open border, absorbing 30 million European immigrants. Prior cross-sectional work finds that immigrants initially held lower- paid occupations than natives but converged over time. In newly assem- bled panel data, we show that, in fact, the average immigrant did not face a substantial occupation-based earnings penalty upon first arrival and experienced occupational advancement at the same rate as natives. Cross-sectional patterns are driven by biases from declining arrival co- hort skill level and departures of negatively selected return migrants. We show that assimilation patterns vary substantially across sending coun- tries and persist in the second generation.

I. Introduction

We study the assimilation of European immigrants in the US labor mar- ket during the Age of Mass Migration ð1850–1913Þ, one of the largest mi-

are grateful for access to census manuscripts provided by Ancestry.com and Family h.org. We benefited from the helpful comments of participants at the University of

l of Political Economy, 2014, vol. 122, no. 3] by The University of Chicago. All rights reserved. 0022-3808/2014/12203-0004$10.00

467

This content downloaded from 132.174.255.086 on October 20, 2016 18:06:23 PM e subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).

468 journal of political economy

All

gration episodes in modern history. Almost 30 million immigrants moved to the United States during this period; by 1910, 22 percent of the US labor force was foreign-born, compared with only 17 percent today. At the time, US borders were completely open to European immigrants. Yet, much like today, contemporaries were concerned about the ability of mi- grants to assimilate into the US economy. Congress fought with various presidential administrations about whether to tighten immigration poli- cies for over 30 years before finally imposing strict quotas in 1924, put- ting an end to the era of open borders. Our paper challenges conventional wisdom and prior research about

immigrant assimilation during this period.1 The common view is that Eu- ropean immigrants held substantially lower-paid occupations than natives upon first arrival but that they converged with the native-born after spend- ing some time in the United States.2 Using newly assembled panel data for 21,000 natives and immigrants from 16 sending European countries, we instead find that, on average, long-term immigrants from sending coun- tries with real wages above the European median actually held significantly higher-paid occupations than US natives upon first arrival, while immi- grants from sending countries with below-median wages started out in

1 For example, politicians and commentators often assume that even uneducated Eu- ropean immigrant groups were able to achieve economic success within a generation or two. Sowell ð1996Þ gives expression to this view, writing that “although notoriously uned- ucated and illiterate . . . Southern and Eastern Europeans eventually became . . . well- represented in occupations requiring education” ð48Þ. In contrast, observers often divide contemporary immigrant groups into those that assimilate quickly and those that do not, as typified by Huntington’s ð2004Þ assessment that “immigrants from India, Korea, Japan and the Philippines, whose educational profiles more closely approximate those of native Americans, have generally assimilated rapidly . . . ½while� Latin American immigrants, par- ticularly those from Mexico, and their descendants have been slower in approximating American norms” ð187–88Þ.

2 Following the economics literature, we focus on immigrants’ labor market perfor- mance rather than on other measures of cultural or social assimilation. In particular, we discuss occupational attainment because individual earnings were not recorded in popu- lation censuses before the mid-twentieth century.

California, Davis, Interdisciplinary Conference on Social Mobility, the Agence Française de Développement–World Bank Migration and Development Conference, the Labor Markets, Families and Children Conference at the University of Stavanger, the Economic History Association, and the National Bureau of Economic Research Development of the Ameri- can Economy Summer Institute. We also thank participants of seminars at Berkeley, Cal- tech, Chicago, Duke, Hebrew University, Northwestern, Norwegian School of Economics, Stanford, Tel Aviv, University of Caifornia at Davis and Los Angeles, and University of Texas at Austin. We learned from conversations with Manuel Amador, Attila Ambrus, Pat Bayer, Doug Bernheim, Tim Bresnahan, Marianne Bertrand, David Card, Greg Clark, Dora Costa, Pascaline Dupas, Liran Einav, Joseph Ferrie, Erica Field, Doireann Fitzgerald, Bob Gor- don, Avner Greif, Hilary Hoynes, Nir Jaimovich, Lawrence Katz, Pete Klenow, Pablo Kurlat, Aprajit Mahajan, Robert Margo, Daniel McGarry, Roy Mill, Joel Mokyr, Jean-Laurent Ro- senthal, Seth Sanders, Izi Sin, Yannay Spitzer, Gui Woolston, Gavin Wright, and members of the UCLA KALER group. Roy Mill provided able assistance with data collection. We ac- knowledge financial support from the National Science Foundation ðgrant SES-0720901Þ, the California Center for Population Research, and UCLA’s Center for Economic History. Data are provided as supplementary material online.

This content downloaded from 132.174.255.086 on October 20, 2016 18:06:23 PM use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).

nation of immigrants 469

equal or lower-paid occupations.3 We find little evidence for the commonly held view that immigrants converged with natives but rather document substantial persistence of the initial earnings gap between immigrants and natives ðwhether positive or negativeÞ over the life cycle. In other words, re- gardless of starting point, immigrants experienced occupational upgrad- ing similar to that of natives, thereby preserving the initial gaps between immigrants and natives over time. Furthermore, this gap persisted into the second generation; when migrants from a certain source country out- performed US natives, so did second-generation migrants, and vice versa.4

Prior studies used cross-sectional data, which confound immigrant con- vergence to natives in the labor market with immigrant arrival cohort ef- fects and the selection of return migrants from the migrant pool. Indeed, when we use cross-sectional data, we confirm the findings of prior studies. When contrasting these findings with those using our panel data, we con- clude that the apparent convergence in a single cross section is driven by a decline in the quality of immigrant cohorts over time and the departure of negatively selected return migrants ðsee Borjas ½1985� and Lubotsky ½2007� for discussions of these sources of bias in contemporary dataÞ.5 We conclude that the notion that immigrants faced a large initial oc-

cupational penalty during the historical Age of Mass Migration is over- stated. Even when US borders were open, the average immigrant who ended up settling in the United States long-term held occupations that commanded pay similar to that of US natives upon first arrival.6 These findings suggest that migration restrictions or selection policies are not necessary to ensure strong migrants’ performance in the labor market. At the same time, the notion that European immigrants converged

with natives after spending 10–15 years in the United States is also ex-

3 We follow individuals between the 1900, 1910, and 1920 US censuses by using their name, age, and place of birth. Assembling such panel data is possible because US Census Bureau policy makes complete individual records ðincluding namesÞ publicly available after 72 years. In particular, we link immigrants and US natives from the 1900 census manuscripts to the 1910 and 1920 census manuscripts using the genealogy websites Ancestry.com and FamilySearch.org.

4 Occupational differences may have persisted over generations because children of mi- grants grew up in migrant enclaves, inherited skills from their parents, or used their par- ents’ networks to find jobs.

5 Over 25 percent of migrants returned to Europe during this era ðGould 1980; Ban- diera, Rasul, and Viarengo 2013Þ. Return migrants may have been negatively selected be- cause those who were unsuccessful in the United States returned home. In addition, many migrants in this era employed a deliberate strategy of temporary migration to the New World ðPiore 1980; Wyman 1996Þ. These temporary migrants will likely appear to be nega- tively selected using our occupation-based measures if they remained in low-paid occupa- tions during their short sojourn ðDustmann 1993Þ. We note that, conceptually, return mi- grants could be positively selected if more productive migrants reached their “target savings” goal faster ðGalor and Stark 1990Þ.

6 Immigrants were more likely than natives to settle in states with a high-paying mix of occupations; location choice was an important strategy that immigrants used to achieve occupational parity with natives.

This content downloaded from 132.174.255.086 on October 20, 2016 18:06:23 PM All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).

470 journal of political economy

All

aggerated, as we find that initial immigrant-native occupational gaps per- sisted over time and even across generations. This pattern casts doubt on the conventional view that, in the past, immigrants who arrived with few skills were able to invest in themselves and succeed in the US economy within a single generation. The remainder of the paper proceeds as follows. Section II discusses

the historical context and related literature. Section III reviews methods used to infer immigrant assimilation in cross-sectional and panel data settings and the biases associated with each. In Section IV, we describe the data construction and matching procedures. Section V presents our empirical strategy and main results on immigrant assimilation and the selection of return migrants. Section VI contains country-by-country re- sults on assimilation and return migration. In Section VII, we assess the robustness of our main findings and present occupational transition matrices that provide more detail about how immigrants and natives moved up the occupational ladder over time. Section VIII rules out other sources of selective attrition from the panel sample beyond return migration, including selective mortality or name changes. Section IX analyzes the performance of second-generation immigrants relative to their parents, and Section X presents conclusions.

II. Immigrant Assimilation in the Early Twentieth Century: Historical Context and Related Literature

The United States absorbed 30 million migrants during the Age of Mass Migration ð1850–1913Þ. By 1910, 22 percent of the US labor force—and 38 percent of workers in nonsouthern cities—were foreign-born ðcom- pared with 17 percent todayÞ.7 Over the period, migrant-sending countries shifted toward the poorer regions of southern and eastern Europe ðHat- ton and Williamson 1998Þ. Many contemporary observers expressed con- cerns about the concentrated poverty in immigrant neighborhoods and the low levels of education among immigrant children, many of whom left school at young ages in order to work in textiles and manufacturing ðMuller 1993; Moehling 1999Þ. Prompted by these concerns, progressive reformers championed a series of private initiatives and public legislation, including child labor laws and compulsory schooling requirements, to fa- cilitate immigrant absorption ðLleras-Muney 2002; Carter 2008; Lleras- Muney and Shertzer 2011Þ, while nativists instead believed that new arrivals would never be able to fit into American society ðHigham 1988; Jacob- son 1999Þ. Fears about immigrant assimilation encouraged Congress to convene

a special commission in 1907 to study the social and economic condi-

7 Authors’ calculations using the 1910 Integrated Public Use Microdata Series ðIPUMSÞ.

This content downloaded from 132.174.255.086 on October 20, 2016 18:06:23 PM use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).

nation of immigrants 471

tions of the immigrant population. The resulting report concluded that immigrants, particularly from southern and eastern Europe, would be unable to assimilate, in part because of high rates of temporary and re- turn migration.8 The Immigration Commission report provided fuel for legislators seeking to restrict immigrant entry ðBenton-Cohen 2010Þ. In 1917, Congress passed a literacy test, which required potential immigrants to demonstrate the ability to read and write in any language ðGoldin 1994Þ. In 1924, Congress further restricted immigrant entry by setting a strict quota of 150,000 arrivals per year, with more slots allocated to northern and western European countries. Since the publication of the Immigration Commission report, gen-

erations of economists and economic historians have assessed the labor market performance of this large wave of immigrant arrivals.9 The ear- liest studies in this area ðre-Þanalyzed the aggregate wage data published by the Immigration Commission and found that, contrary to the initial conclusions of the commission, immigrants caught up with the native- born after 10–20 years in the United States ðHiggs 1971; McGouldrick and Tannen 1977; Blau 1980Þ. Related work examined individual-level wage data from surveys conducted by state labor bureaus ðHannon 1982; Eichengreen and Gemery 1986; Hanes 1996Þ. Although early studies of these sources found no wage convergence, Hatton ð1997Þ argues that this discrepancy is due to specification choice. He reanalyzes the state data with two simple modifications and finds that immigrants who ar- rived at age 25 fully erased the wage gap with natives within 13 years in the United States.10

More recent studies on immigrant assimilation incorporate data from the federal census of population. The census offers complete industrial and geographic coverage but contains information only on occupation rather than on individual wages or earnings. Relying on the 1900 and

8 Two authors of the report, Jeremiah Jenks and W. Jett Lauck, later summarized this view of temporary migrants, writing, “if an immigrant intends to remain permanently in the US and become an American citizen, he naturally begins at once . . . to fit himself for the conditions of his new life. . . . If, on the other hand, he intends his sojourn in this coun- try to be short . . . the acquisition of the English language will be of little consequence. . . . The chief aim of a person with this intention is to put money in his purse . . . not for in- vestment here but for investment in his home country” ðquoted in Wyman 1996, 99–100Þ.

9 In a related body of work, Ferrie ð1997, 1999Þ measures immigrant assimilation in the antebellum period. Lieberson ð1980Þ and Alba and Nee ð2003Þ are two core references in the sociological literature on immigrant assimilation.

10 In particular, Hatton ð1997Þ allows for differences in the return to experience for younger and older workers and separates immigrants who arrived as children from those who arrived as adults. The convergence figure reported in the text is based on Hatton’s ta- ble 4 ðcols. 1, 3Þ. Because he estimates different returns to experience parameters for im- migrants and the native-born, the size of the initial wage gap varies by age. For this calcula- tion, we consider an immigrant who arrives at age 25, at which point the implied wage gap with natives is 0.275, a gap that is erased after the immigrant spends 13 years in the United States.

This content downloaded from 132.174.255.086 on October 20, 2016 18:06:23 PM All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).

472 journal of political economy

All

1910 census cross sections, Minns ð2000Þ finds partial convergence be- tween immigrants and natives outside of the agricultural sector.11 Im- migrants eliminate 30–40 percent of their ðbetween-occupationÞ earn- ings deficit relative to natives after 15 years in the United States. Overall, across three different data sets, the existing literature suggests

that immigrant workers experienced substantial occupational and earn- ings convergence with the native-born in the early twentieth century. How- ever, all these analyses compare earnings in a single cross section, a method that suffers from two potentially important sources of bias: selective re- turn migration and changes in immigrant cohort quality over time.12

III. Inferring Immigrant Assimilation from Cross-Sectional and Panel Data

Imagine that the researcher has only a single cross section of data, say the 1920 census, from which to estimate the pace of convergence between immigrants and the native-born in the labor market. In this case, she may compare the earnings of a long-standing immigrant who arrived in the United States in 1895 to that of a recent immigrant who arrived in 1915. For illustration, let the mean earnings of a native-born worker be $100. If the immigrant who arrived in 1895 also earned $100 in 1920 while the immigrant who arrived in 1915 earned $50, the researcher could conclude that, upon arrival, migrants faced an earning penalty relative to natives that is completely erased after 25 years in the United States. However, this conclusion might mistake differential skills across arrival cohorts for true migrant assimilation; this point was first made by Doug- las ð1919Þ and was developed by Borjas ð1985Þ.13 If, for example, the long- standing migrant was a literate craftsman from Germany whereas the recent arrival was an unskilled common laborer from Italy, the differ- ence in their earnings in 1920 may reflect permanent gaps in their skill levels rather than temporary gaps due to varying time spent in the United States. This bias can be addressed with repeated cross-sectional observations

on arrival cohorts, say by observing 1895 immigrant arrivals in both the

11 Consistent with our results, Minns finds that the full immigrant population actually earn as much as ðor more thanÞ natives. The immigrant deficit explored in his paper is present only outside of the agricultural sector.

12 Minns ð2000Þ acknowledges the potential bias from changes in the quality of immi- grant arrival cohorts. Hatton ð1997Þ partially addressed the shift in sending countries by separately analyzing assimilation profiles by country of origin for three sending countries ðBritain, Ireland, and GermanyÞ.

13 In an early paper in this literature, Chiswick ð1978Þ found that immigrants in the 1970 cross section experienced faster wage growth than the native-born and overtook natives within 15 years of arrival. Borjas ð1985Þ demonstrated that, in this period, half of the ap- parent convergence in a cross section is driven by changes in cohort quality over time.

This content downloaded from 132.174.255.086 on October 20, 2016 18:06:23 PM use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).

nation of immigrants 473

1900 and 1920 censuses. However, in this case, inferences on migrant as- similation may still be inaccurate because of selective return migration; this point was first made by Jasso and Rosenzweig ð1988Þ and was investi- gated empirically by Lubotksy ð2007Þ.14 In the 1900 census, the 1895 mi- grant arrival cohort contains both temporary arrivals who will return to their home country before the 1920 census and longer-standing immi- grants who will remain in the United States in 1920. By 1920, only the long- standing immigrants remain. If the temporary migrants have lower skills or exert less effort in moving up the occupational ladder in the United States, this compositional change in the repeated cross section will gen- erate the appearance of wage growth within the cohort over time as the lower-earning migrants return to Europe.15

We emphasize that in our panel data we estimate an assimilation pro- file for immigrants who were in the United States in both 1900 and 1920, that is, those who remained in the United States for at least 20 years. These immigrants are of particular interest because they participate in the US labor market for many years and are more likely to raise children in the United States who then contribute to the labor force in the next generation. However, to understand the experience of the typical mi- grant in the United States at a point in time, a group that includes both permanent migrants and migrants who will later return to their home country, the assimilation patterns in the repeated cross sections are also of interest. For a set of earlier papers, we compiled a panel data set of immigrants

that matched individuals from their childhood household in Europe to their adult outcomes; in this case, we were able to focus on only a single sending country, Norway ðAbramitzky, Boustan, and Eriksson 2012, 2013Þ. These data allowed us to analyze the selection of who migrates from Eu- rope to the United States and the economic return to this migration. We found evidence of negative selection in the sense that men whose fathers did not own land or whose fathers held low-skilled occupations were more likely to migrate. We also estimated a return to migration free from between- household selection by comparing brothers, one of whom migrated and one of whom stayed in Norway; using this method, we found a return to migration of around 70 percent. In this paper, we use panel data to com- pare immigrants to US natives ðassimilationÞ rather than to compare im-

14 During the Age of Mass Migration, some immigrants engaged in circular migration, migrating to the United States and returning to Europe multiple times ðPiore 1980; Wy- man 1996Þ. Circular migrants will enter the panel sample only if they happen to live in the United States in the census years; otherwise, they will be treated as temporary migrants.

15 In addition to Lubotsky ð2007Þ, other panel analyses of immigrant assimilation in the contemporary period include Borjas ð1989Þ, Edin, LaLonde, and Aslund ð2000Þ, Hu ð2000Þ, Duleep and Dowhan ð2002Þ, Constant and Massey ð2003Þ, Eckstein and Weiss ð2004Þ, and Kim ð2011Þ. Zakharenko ð2009Þ provides descriptive evidence that return migrants leaving the United States are negatively selected.

This content downloaded from 132.174.255.086 on October 20, 2016 18:06:23 PM All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).

474 journal of political economy

All

migrants to the European sending population ðselectionÞ. Furthermore, we move beyond our focus on a single country and assemble new panel data for immigrants from 16 sending countries.

IV. Data and Matching

A. Matching Men between the 1900, 1910, and 1920 US Censuses

Our analysis relies on a new panel data set that follows native-born work- ers and immigrants from 16 sending countries through the US censuses of 1900, 1910, and 1920. We match individuals over time by first and last name, age, and country or state of birth; details on the matching proce- dure are provided in Appendix A. We restrict our attention to men be- tween the ages of 18 and 35 in 1900, an age range in which men are both old enough to be employed in 1900 and young enough to still be in the workforce in 1920. We further limit the immigrant portion of the sample to men who arrived in the United States between 1880 and 1900. For comparability with the foreign-born, 95 percent of whom live outside of the South, we exclude native-born men residing in a southern state and all black natives regardless of place of residence.16 We compare results in this panel data set to similarly defined cross sections of the population drawn from the census public use samples of 1900, 1910, and 1920 ðRug- gles et al. 2010Þ. Table 1 presents match rates and final sample sizes for each sending

country and for native-born men in the panel sample. Our matching pro- cedure generates a final sample of 20,225 immigrants and 1,650 natives. We can successfully match 16 percent of all native-born men forward from 1900 to both 1910 and 1920. For the foreign-born, the average forward match rate across countries is lower ð12 percentÞ, which is expected given that a sizable number of migrants return to Europe between 1900 and 1920. These double match rates are similar to those in Ferrie ð1996Þ and Abramitzky et al. ð2012Þ.17

16 In a robustness exercise, we included native-born men living in the South in the sam- ple. Because men who live in the South held lower-paid occupations, the earnings pre- mium enjoyed by long-term immigrants increases to $4,000 ðcompared to only $450 in the nonsouthern sampleÞ. Yet the extent of convergence in both samples and the comparison between immigrants in the cross section and panel ðrelative to nativesÞ are preserved. Re- sults are presented in online App. B.

17 Our iterative matching procedure can produce false matches if there are two indi- viduals with the same name and similar ages who then misreport their ages on the next census. We also use a more conservative matching strategy that requires all matches to be unique by name and age within a 5-year age band. This procedure results in fewer matches ð8,806 casesÞ that appear to be somewhat positively selected from the population perhaps because entry into this sample requires a very uncommon name. We discuss results from this alternative sample in n. 25.

This content downloaded from 132.174.255.086 on October 20, 2016 18:06:23 PM use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).

TABLE 1 Sample Sizes and Match Rates by Place of Birth

Country

1900 Number in Universe

ð1Þ

Number Matched

ð2Þ

Match Rate, Total ð3Þ

1900 Number, Unique

ð4Þ

Match Rate Unique

ð5Þ A. 1900 Source: IPUMS

Austria 4,835 339 .070 4,677 .072 England 7,438 664 .089 6,175 .107 France 11,615 728 .063 9,139 .079 Germany 19,855 2,248 .113 16,733 .134 Ireland 9,737 861 .088 6,323 .136 Italy 7,624 811 .106 7,042 .115 Norway 3,541 425 .120 2,822 .151 Russia 5,804 644 .111 5,203 .124 Sweden 6,164 559 .091 4,070 .137 US natives 10,000 1,650 .165 8,345 .197

B. 1900 Source: Ancestry.com

Belgium 6,060 545 .090 5,962 .091 Denmark 34,594 1,980 .058 17,425 .114 Finland 23,843 828 .035 22,197 .037 Portugal 12,585 584 .046 8,362 .070 Scotland 53,091 4,349 .082 15,529 .280 Switzerland 22,276 3,311 .149 20,588 .161 Wales 17,767 1,342 .076 9,876 .135

Note.—The sample universe includes men between the ages of 18 and 35 in 1900. Im migrants must have arrived in the United States between 1880 and 1900. We exclude al blacks and native-born men living in the South. For large sending countries and the native born, we start with the 1900 IPUMS sample ðpanel AÞ. For smaller sending countries, we begin with the complete population in 1900. The text describes our matching procedure The number of matched cases refers to men who match to both the 1910 and 1920 cen suses. We report the number of unique cases by first name, last name, age, and country o birth and the match rate for this group in cols. 4 and 5.

nation of immigrants 475

This content downloaded from 132.174.255.086 on October 20, 2016 18:06:23 PM All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchic

,

- l -

. - f

Despite the fact that men with uncommon names are more likely to match between census years, our matched sample is reasonably represen- tative of the population. Appendix table A1 compares the occupation- based earnings of men in the matched sample to men in the full popula- tion in 1920 ðthe earnings measure is described in the next sectionÞ. By definition, men in both the panel data and the 1920 cross section must have survived and remained in the United States until 1920. Thus, by 1920, up to any sampling error, differences between the panel and the repre- sentative cross section must be due to an imperfect matching proce- dure. Among natives, the difference in the mean occupation score in the matched sample and the population in 1920 is small ð$37Þ and statistically indistinguishable from zero. In contrast, immigrants in the matched sam- ple have a $300 advantage over immigrants in the representative sample. Therefore, up to $300 of the occupation-based earnings differential be- tween immigrants and natives in the panel data could be due to sample selection induced by our matching procedure.

ago.edu/t-and-c).

476 journal of political economy

All

B. Occupation and Earnings Data

We observe labor market outcomes for our matched sample in 1900, 1910, and 1920. Because these censuses do not contain individual infor- mation about wages or income, we assign individuals the median income in their reported occupation.18 Table 2 reports the 10 most common occupations for our sample of matched natives and foreign-born work- ers. Although the top 10 occupations are similar for both groups, mi- grants to the United States were less likely to be farmers ð18 vs. 26 per- centÞ and more likely to be managers or foremen ð14 vs. 10 percentÞ. The native-born were also more likely to be salesmen and clerks, two occu- pations with high returns to fluency in English. Other common occu- pations in both groups include operatives and general laborers.19

Our primary source of income data is the “occupational score” vari- able constructed by IPUMS. This score assigns to an occupation the me- dian income of all individuals in that job category in 1950. For ease of interpretation, we convert this measure into 2010 dollars. Using this mea- sure, our data set contains individuals representing around 125 occupa- tional categories. Occupation-based earnings are a reasonable proxy for “permanent” income, by which we can measure the extent to which im- migrants assimilate with natives in social status. One benefit of matching occupation to earnings in a single year is that

our measure of movement up the occupational ladder will not be con- founded by changes in the income distribution. Butcher and DiNardo ð2002Þ, for example, point out that much of the growth in the immigrant- native wage gap between 1970 and 1990 was due to widening income in- equality ðsee also Lubotsky 2011Þ. Given that immigrants today are clus- tered in low-skill jobs, their wages stagnated while the wages of some natives grew. Although the growth in the immigrant-native wage gap is “real” in the sense that immigrants had lower purchasing power in 1990 than they did in 1970, it does not necessarily reflect a decline in immi- grants’ social standing or ability to assimilate into the US economy. Yet our reliance on occupation-based earnings prevents us from mea-

suring the full convergence between immigrants and natives. In par- ticular, we are able to capture convergence due to advancement up the

18 For observations taken from the 1900 IPUMS ðthe native-born and immigrants from large sending countriesÞ, we use the occupation recorded in the digitized microdata. For the remaining countries in 1900 and for all countries in 1910 and 1920, we collect the oc- cupation string by hand from the historical manuscripts on Ancestry.com. We then stan- dardize occupation titles to match those identified in the 1900 IPUMS. Our final sample has 1,193 native-born men and 16,962 immigrants with nonmissing occupation data.

19 Men who were not employed at the time of the survey reported their last-held occu- pation. The 1910 census was the only one in our time period to ask about unemployment. In that year, native-born men of native parentage ðaged 18–60Þ had an unemployment rate of 4.4 percent, while 5.7 percent of the foreign-born were unemployed. This differential un- employment likely contributed to the true earnings gap between immigrants and natives.

This content downloaded from 132.174.255.086 on October 20, 2016 18:06:23 PM use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).

T A B L E 2

C o m m o n O c c u p a t i o n s f o r N a t i v e s a n d F o r e i g n - B o r n i n M a t c h e d S a m p l e , 1 9 2 0

N a t i v e s

F o r e i g n -B o r n

O c c u p a t i o n

F re q u e n cy

P e rc e n t

H IS C L A S S

O c c u p a t i o n

F re q u e n cy

P e rc e n t

H IS C L A S S

F ar m e r

2 8 2

2 5 .9 9

8 F a rm

e r

3 ,1 4 1

1 8 .4 3

8 M an

a g e r

8 9

8 .2 0

9 M a n a g e r

1 ,8 6 3

1 0 .9 3

9 L a b o re r

8 3

7 .6 5

1 2

L a b o re r

1 ,6 1 6

9 .4 8

1 2

S al e sm

a n

5 8

5 .4 4

5 O p e ra ti ve

1 ,0 2 0

5 .9 9

5 O p e ra ti ve

5 4

4 .9 8

9 F o re m a n

5 7 6

3 .3 8

3 C le ri ca l

4 3

3 .9 6

5 M in e o p e ra ti ve

5 7 5

3 .3 7

9 C ar p e n te r

3 6

3 .3 2

7 M a ch

in is t

5 5 0

3 .2 3

9 M ac h in is t

3 5

3 .2 3

9 C a rp e n te r

4 9 3

2 .8 9

7 F ar m

la b o re r

3 3

3 .0 4

1 2

S a le sm

an 4 4 4

2 .6 1

5 F o re m a n

2 1

1 .9 4

3 C le ri ca l

3 1 4

1 .8 4

5 T o ta l ðto

p 1 0 Þ

7 1 4

6 5 .8 1

1 0 ,2 7 8

6 0 .3 1

O u ts id e to p 1 0

3 7 1

3 4 .1 9

6 ,7 6 3

3 9 .6 9

N o t e . — S e e th e n o te

to ta b le

1 fo r sa m p le

re st ri ct io n s. H IS C L A S S is a 1 2 -p a rt

cl a ss ifi ca ti o n sy st e m

in d ic at in g th e

so ci a l cl a ss

o f e ac h o cc u p a ti o n ðv a n L e e u w e n a n d M a a s 2 0 0 5 Þ.

This content downloaded from 132.174.255.086 on October 20, 2016 18:06:23 PM All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).

478 journal of political economy

All

occupational ladder ðbetween-occupation convergenceÞ, but we cannot measure potential convergence between immigrants and natives in the same occupation. To assess the extent of this bias, we use data from the 1970 and 1980 IPUMS samples, the first census years to record both wage data and year of immigration for the foreign-born. We find that occupation-based earnings capture around 30 percent of the initial earn- ings penalty in the cross section and 65 percent in repeated cross sec- tions.20 Similarly, occupation-based earnings account for 30 percent of total convergence between immigrants and natives in the cross section and can explain all of the ðmuch lowerÞ earnings convergence in the re- peated cross sections.21 It is reasonable to conclude, then, that our mea- sure is able to capture at least 30 percent of true earnings convergence ðal- though note that inferring occupational advancement from cross-sectional data suffers from the biases described aboveÞ. A further concern with the IPUMS occupation score variable is its an-

choring to occupation-based earnings in the year 1950. The 1940s–1950s was a period of wage compression ðGoldin and Margo 1992Þ. If immi- grants were clustered in low-paying occupations, the occupation score variable may understate both their initial earnings penalty and the con- vergence implied by moving up the occupational ladder. We address this concern by using occupation-based earnings from the 1901 Cost of Living survey as an alternative dependent variable ðPreston and Haines 1991Þ.22 We also try extrapolating the 1950 occupation-based earnings back to the early 1920s using a time series of earnings by broad occupa- tion category ðclerical, skilled blue-collar, and unskilled blue-collarÞ re- ported in Goldin and Margo ð1992Þ.

20 We estimate two earnings equations for immigrants and natives in these years, first using the occupation-based earnings measure from the paper as a dependent variable and then using actual individual earnings. In the 1970 cross section, immigrants’ initial earn- ings penalty is 23 log points when using individual earnings and only 4 log points when instead focusing on occupation-based earnings. In 1980, the two earnings penalties are closer together ð26 and 10 log points, respectivelyÞ, and in the repeated cross-section spec- ification, the earnings penalty for the 1960s arrival cohort is 10 log points in individual earn- ings and 7 log points in occupation-based earnings.

21 In particular, we estimate two earnings equations for immigrants and natives in these years, first using the occupation-based earnings measure from the paper as a dependent variable and then using actual individual earnings. In the 1970 cross section, immigrants ap- pear to experience 29 log points of total wage convergence relative to natives after spending 30 years in the United States and only 8 log points of convergence when using an occupation- based measure of earnings, suggesting that occupation-based earnings measures capture only 30 percent of total convergence. If instead we follow arrival cohorts from the 1970 to the 1980 census, we observe much lower rates of total wage convergence ð1.5 log pointsÞ and cannot rule out that all of this convergence takes place through movement up the occu- pational ladder.

22 The 1901 Cost of Living survey has several disadvantages relative to the 1950 occu- pation score. First, the Cost of Living surveys were not nationally representative but in- stead focused on urban married households. Second, income in the surveys is missing for a number of occupations ðincluding farmers, which we instead infer from the US census of agricultureÞ.

This content downloaded from 132.174.255.086 on October 20, 2016 18:06:23 PM use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).

nation of immigrants 479

V. Immigrant Assimilation in Panel Data

A. Occupational Distribution of Immigrants and Natives in 1900

Before turning to occupation-based earnings measures, we illustrate our main findings in a series of charts in figure 1. These charts match indi- viduals’ reported occupations to social classes using the Historical Inter- national Social Class Scheme ðHISCLASSÞ developed by van Leeuwen and Maas ð2005Þ and then further group these codes into five categories: white-collar, skilled blue-collar, farmers, semiskilled blue-collar, and un- skilled. For reference, we also report the average earnings of these social classes in table 3. These results, and all others, are reweighted so that the panel sample reflects the actual distribution of country of origin in the 1920 population. Each panel of figure 1 graphs the occupational distributions of differ-

ent groups of men in 1900, either in the representative cross section or in the panel sample. Figures 1A and 1B compare immigrants to the native- born. Although, on average, immigrants and natives held similarly paid occupations ðsee table 3Þ, the native-born were more likely to hold white- collar positions ðsuch as salesmenÞ and to be farmers, while immigrants were more likely to engage in skilled or semiskilled blue-collar work ðcar- penter, machinistÞ. Immigrants and natives were roughly equally likely to be unskilled. These occupation distributions suggest that whether or not immigrants faced a wage penalty or a wage premium relative to natives may be sensitive to the placement of farmers in the earnings distribution. Comparing the full population in the cross section ðfig. 1AÞ with the

panel sample ðfig. 1BÞ is also informative. First, long-term immigrants were less likely than the typical immigrant in 1900 to hold unskilled po- sitions ð25 percent vs. 34 percentÞ. The difference in the probability of engaging in unskilled work is made up by the fact that long-term im- migrants are more likely to be farmers or to hold white-collar or skilled blue-collar positions. These occupational differences suggest that there was negatively selected attrition from the cross section consisting of un- skilled temporary migrants who returned to Europe. Second, beyond be- ing slightly more likely to be farmers, there are no other notable differ- ences between the natives in the cross section and those in the panel, which is consistent with a lack of other forms of selective attrition in the data ðe.g., due to mortalityÞ. Section VIII discusses other sources of poten- tial selective attrition in more detail. Earlier- and later-arriving immigrants are compared in figures 1C and

1D. Immigrants who arrived in the 1890s are substantially more likely than immigrants who arrived in the 1880s to be unskilled workers in 1900 ð41 percent vs. 26 percentÞ. Much of this difference is due to the lower skills of this later cohort and does not disappear with age. The gap between these arrival cohorts is smaller but still apparent among long- term immigrants in the panel sample.

This content downloaded from 132.174.255.086 on October 20, 2016 18:06:23 PM All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).

480 journal of political economy

All

B. Estimating Equation

Our main analysis compares the occupational mobility of native-born and immigrant workers. We estimate

Occupation scoreijmt 5 gt2m 1 mm 1 vt 1 aj 1 b1ageit 1 b2age 2 it

1 b3age 3 it 1 b4age

4 it 1 εijmt;

ð1Þ

where i denotes the individual, j denotes the country of origin, m is the year of arrival in the United States, t is the ðcensusÞ year, and t 2 m is thus the number of years spent in the United States. Occupation score is a proxy for labor market earnings that varies between ðbut not withinÞ

FIG. 1.—Occupational distribution of natives and immigrants in cross section and panel in 1900. A, Cross section, immigrants and natives. B, Panel, immigrants and natives.

This content downloaded from 132.174.255.086 on October 20, 2016 18:06:23 PM use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).

TABLE 3 Mean Earnings by Nativity and Social Class in 1901

and 1950 Data in Panel Sample

1901 ð$Þ 1950 ð$Þ All immigrants 17,939 22,698 All natives 18,106 21,357 White-collar 24,337 30,906 Skilled blue-collar 19,058 26,604 Farmers 21,324 12,609 Semiskilled 15,757 23,085 Unskilled 8,583 13,554

Note.—Figures are reported in 2010 dollars. Occupations are clas- sified according to the HISCLASS rubric: HISCLASS 1–5 5 white- collar; HISCLASS 6–7 5 skilled blue-collar; HISCLASS 8 5 farmers; HISCLASS 9 5 semiskilled; HISCLASS 10–12 5 unskilled.

FIG. 1 ðContinuedÞ.—Occupational distribution of natives and immigrants in cross sec- tion and panel in 1900. C, Cross section, immigrants in early and late arrival cohorts. D, Panel, immigrants in early and late arrival cohorts.

This content downloaded from 132.174.255.086 on October 20, 2016 18:06:23 PM All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).

482 journal of political economy

All

occupations. The coefficients b1 through b4 relate years of labor market experience to the worker’s position on the occupational ladder.23

A vector of indicator variables gt2m separates the foreign-born into five categories according to time spent in the United States ð0–5 years, 6– 10 years, 11–20 years, 21–30 years, and 30 or more yearsÞ, with the native- born constituting the omitted category. The sign and magnitude of the coefficient on the first dummy variable ð0–5 yearsÞ indicate whether im- migrants received an occupation-based earnings penalty ðor premiumÞ upon first arrival to the United States, whereas the remaining dummy variables reveal whether immigrants eventually catch up with or surpass the occupation-based earnings of natives. Our main specification divides the foreign-born into two year-of-arrival cohorts indicated by mm ðarrivals before and after 1890Þ to allow for differences in occupation-based earn- ings capacity by arrival year; Section VII explores the sensitivity of the results to the choice of the number of arrival cohorts. Observations are weighted to reflect the actual distribution of country of origin in the 1920 population.24

We begin by estimating two versions of equation ð1Þ using pooled data from the 1900, 1910, and 1920 IPUMS samples. The first specification omits the arrival cohort dummy ðmmÞ, thereby comparing immigrants in the United States for various lengths of time both between and within arrival cohorts. We refer to this specification as the “cross-section” model. We then add the arrival cohort dummy and reestimate equation ð1Þ. We refer to this specification as the “repeated cross-section” model because it follows arrival cohorts across census waves. Comparing the cross section and the repeated cross section allows us to infer how much of the initial occupational penalty can be attributed to differences in the quality of arrival cohorts. Note that, because we include country fixed effects, we measure differences in arrival cohorts within sending countries over time. Finally, we compare the repeated cross-section results with estimates

of equation ð1Þ in the panel sample. The panel data follow individuals, rather than arrival cohorts, across census waves. Therefore, comparing the estimates in the repeated cross section and the panel allows us to infer whether and to what extent return migrants were positively or negatively selected from the immigrant population. If we observe more ðlessÞ convergence in the repeated cross section than in the panel, we can infer that the temporary migrants are drawn from the lower ðupperÞ

23 The rates of convergence for immigrants in the cross section and the panel are similar if, instead, as in Hatton ð1997Þ, we allow the slope of the experience profile to vary by age to account for steep returns to labor market experience for young workers in the early twentieth century ðsee App. BÞ.

24 We need to reweight the matched sample because our universe of potential matches is drawn from 5 percent samples for large countries and from 100 percent samples for smaller countries. We weight according to the 1920 cross section to reflect the fact that migrants in the panel sample remain in the United States until 1920.

This content downloaded from 132.174.255.086 on October 20, 2016 18:06:23 PM use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).

nation of immigrants 483

end of the occupation-earnings distribution, thereby leading their de- parture to increase ðdecreaseÞ the immigrant average.

C. Occupational Convergence in Cross-Section and Panel Data

In this subsection, we estimate equation ð1Þ using occupation-based earn- ings, first using data from the 1950 census and then using data from the 1901 Cost of Living survey. We show that, with both earnings measures, ð1Þ in the cross section, immigrants initially hold lower-paid occupations but converge on natives over time; ð2Þ following arrival cohorts from 1900 to 1920 in the repeated cross sections reduces the initial migrant disad- vantage; and ð3Þ long-term immigrants in the panel data look even closer to natives upon first arrival, closing the earnings gap completely when us- ing the 1950 occupation-based earnings data and drawing closer to but not completely converging with natives in the 1901 earnings data. That is, the apparent immigrant disadvantage in a single cross section is driven by the lower quality of later arrival cohorts ð1890s vs. 1880sÞ and the negative se- lection of temporary migrants who eventually return to Europe. We begin by discussing the results when occupations are matched to

1950 earnings, as presented in table 4. In the cross section, new immi- grants hold occupations that earn $1,200 below natives of similar age and appear to completely make up this gap over time ðcol. 1, in 2010 dol- larsÞ. Columns 2 and 3 pool data from the cross section and panel and report the interactions between being in the cross section ðor the panelÞ and the indicators for years spent in the United States and for arrival co- hort.25 When we simply control for arrival cohort in column 2, the occu- pation score gap between recently arrived immigrants and natives shrinks to $400. In other words, even within sending countries, around three- quarters of the initial gap in the pooled cross section is due to the lower occupational skills of immigrants who arrived after 1890.26 Indeed, im- migrants who arrived after 1890 had significantly lower occupation-based earnings than earlier arrivals, receiving an arrival cohort penalty of $750. Coefficients for the panel data are reported in column 3. In this case,

occupation-based earnings gaps between immigrants and natives are found by subtracting the coefficient for the native-born from each co- efficient for years spent in the United States. For this subsample of long- term migrants, we find no initial occupation score gap between immi-

25 Note that, by pooling the two data sources, we constrain the year, country of origin, and age effects to be common across the two samples. We do allow the coefficients on the fixed effects for arrival cohort ðmmÞ and years spent in the United States ðgt2mÞ to vary by sample. Results are similar when we run eq. ð1Þ separately for the panel and the repeated cross section ðsee App. BÞ.

26 The decline in cohort quality within countries of origin over time is consistent with the idea that “pioneer” migrants are more skilled than migrants who follow their friends and family to the United States.

This content downloaded from 132.174.255.086 on October 20, 2016 18:06:23 PM All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).

TABLE 4 Ordinary Least Squares Estimates, Age-Earnings Profile for Natives and

Foreign-Born, 1900–1920: 1950 Occupation-Based Earnings in 2010 Dollars

Pooled Cross Section

and Panel

Right-Hand-Side Variable Cross Section

ð1Þ

Cross Section Coefficients

ð2Þ

Panel Coefficients

ð3Þ 0–5 years in US 21,255.73 2384.49 293.51

ð143.44Þ ð187.30Þ ð237.96Þ 6–10 years in US 2734.51 22.89 467.64

ð147.44Þ ð172.05Þ ð213.61Þ 11–20 years in US 2352.93 173.83 329.38

ð131.27Þ ð134.02Þ ð150.49Þ 21–30 years in US 2294.87 128.44 74.34

ð142.10Þ ð138.93Þ ð150.33Þ 30 years in US 22.41 155.77 231.90

ð184.65Þ ð178.49Þ ð186.55Þ Arrive 18911 . . . 2739.18 2232.77

ð106.99Þ ð160.58Þ Native-born . . . . . . 2153.83

ð176.14Þ Observations 205,458 259,093

Note.—See the table 1 note for sample restrictions. Columns report coefficients from estimation of eq. ð1Þ. Column 1 pools three cross sections ð1900–1920Þ; the regression in cols. 2 and 3 adds the matched panel sample. The coefficients in col. 2 are interactions between the right-hand-side variables listed and a dummy for being in the cross section, and col. 3 reports interactions between the right-hand-side variables and a dummy for being in the panel. The omitted category is native-born men in the cross section. Coeffi- cients on age, census year dummies, and country-of-origin fixed effects are not shown.

484 journal of political economy

All

grants and natives. If anything, immigrants start out about $450 ahead of natives ð5 293 1 153Þ, although a gap of this size may be partially due to differential selection into the matched sample ðsee Sec. IV.AÞ.27 The immigrant-native occupation-based earnings gaps in the repeated cross section and the panel are statistically different from each other for im- migrants who arrived between 0–5 and 6–10 years ago. This comparison suggests that the observed occupation-based earnings gap in the repeated cross section is capturing the negative selection of immigrants who end up returning to Europe. Note that the difference between the 0–5 years in the United States co-

efficients in the panel and repeated cross section reflects the occupation- based earnings gap between long-term migrants and a weighted average of temporary and longer-term migrants. This gap can be used to back out

27 Results are qualitatively similar in the restricted sample that contains only those in- dividuals with a unique match by name and age within a 5-year age band ðsee App. BÞ. Long-term immigrants experience a $900 premium relative to natives upon first arrival in the restricted sample, compared to a $450 premium in the main sample. As in the main results, long-term immigrants in the restricted sample experience a negligible amount of convergence relative to natives after 30 years in the United States.

This content downloaded from 132.174.255.086 on October 20, 2016 18:06:23 PM use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).

FIG. 2.—Convergence in occupation score between immigrants and native-born workers by time spent in the United States, cross-sectional and panel data, 1900–1920. The graph plots coefficients for years spent in the United States indicators in equation ð1Þ. Note that for the panel line, we subtract the native-born dummy from the years in the United States indicators ðbecause the omitted category in that regression is natives in the panel sampleÞ. See table 4 for coefficients and standard errors.

nation of immigrants 485

the differential in occupation-based earnings between long-term and tem- porary migrants. For a country experiencing a 25 percent return migra- tion rate ðsee n. 5Þ, the gap of $678 ð5 293 1 384Þ implies that the typical return migrant held an occupation that earned $2,700 ðor 12 percentÞ less than the average migrant who remained in the United States.28

The differences in the initial immigrant-native gaps and implied rates of convergence between the cross-section and panel samples are un- derscored in figure 2. This figure graphs the coefficients on the 5 years in the United States dummy variables in the pooled cross section and the repeated cross sections and the difference between the native-born dummy and the years in the United States indicators for the panel sam- ple. In graphical form, it is even easier to see that, in the cross section,

28 The difference between the panel and repeated cross section coefficients on the 0–5 years in the United States indicator can be written as ðearnings of permanent migrants who have been in the United States for 0–5 yearsÞ 2 ½0.75ðearnings of permanent migrants who have been in the United States for 0–5 yearsÞ � 0.25ðearnings of temporary migrants who have been in the United States for 0–5 yearsÞ�. This expression simplifies to 0.25ðearnings of permanent migrants 2 earnings of temporary migrantsÞ, which equals the observed differ- ence of $678. Therefore, the differential between the earnings of permanent and temporary migrants is $678/0.25, or $2,700. Note that regardless of how long a migrant stays in the United States, he will contribute to the 0–5 year coefficient in the cross section.

This content downloaded from 132.174.255.086 on October 20, 2016 18:06:23 PM All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).

486 journal of political economy

All

immigrants appear to face an occupation score gap relative to natives upon first arrival but are able to erase this gap over time. In contrast, ac- cording to the repeated cross section line, immigrants in the pre-1890 ar- rival cohort experienced a much smaller occupation score gap relative to natives upon first arrival. Finally, permanent immigrants in the panel data hold slightly higher-paying occupations than natives, even upon first ar- rival, and retain this advantage over time. Of the $1,600 difference between the immigrant occupation-based earnings penalty observed in the cross section and the immigrant earnings premium in the panel, around 55 per- cent can be attributed to arrival cohort skill level ð5 2$384 2 ½2$1,255�Þ and the remaining 45 percent can be attributed to the negative selection of return migrants ð5 ½$293 2 $153� 2 ½2$384�Þ. Table 5 repeats the analysis using occupation-based earnings from

the 1901 Cost of Living survey. When occupations are matched to the 1901 earnings in panel A, immigrants in the cross section appear to have a much larger initial occupation-based earnings gap with natives ð$4,200 in 2010 dollars vs. $1,200 when matched to the 1950 occupation-based earnings data in table 4Þ. Yet, despite differences in the size of the initial gap between the two data sources, we continue to find here that a large portion of the observed convergence in the cross section is driven by bi- ases due to changes in arrival cohort skill level and negatively selected return migration. Panel B of table 5 explores the source of the larger initial occupation-

based earnings gap between immigrants and natives in the 1901 data. In particular, we make three adjustments to the 1950 occupation-based earnings to match the attributes of the Cost of Living survey: using only urban workers to calculate occupation-based earnings; using mean, rather than median, earnings by occupation; and using the 1900 Census of Agriculture rather than the 1950 Census of Population to infer earn- ings of farmers. Together, these three adjustments can account for 70 per- cent of the difference in the estimated coefficients generated by these two income sources.29 We favor the 1950 occupation-based earnings because it covers the entire population, both rural and urban, and because it places farmers below the median of the income distribution, which is consistent with the fact that, as a profession, farming was declining in earning power and social status over the early twentieth century.30

29 We compare the coefficients on the years in the United States indicators across spec- ifications with different dependent variables. The average difference in the coefficients between the 1950 occupation-based earnings ðtable 4Þ and the 1901 Cost of Living survey ðpanel A of table 5Þ is $2,300, while the average difference in the coefficients between the adjusted 1950 earnings and the Cost of Living survey ðpanels A and B of table 5Þ is only $700. Therefore, we conclude that these three simple adjustments can account for 70 per- cent ð5 1,600/2,300Þ of the difference between the two income sources.

30 In an alternative approach to adjust for changes in the wage structure over time, we use time series of earnings by broad occupation category—clerical, skilled blue-collar, and

This content downloaded from 132.174.255.086 on October 20, 2016 18:06:23 PM use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).

TABLE 5 Ordinary Least Squares Estimates, Age-Earnings Profile for Natives and

Foreign-Born, 1900–1920: 1901 Occupation-Based Earnings in 2010 Dollars

Pooled Cross

Section and Panel

Pooled Cross

Section and Panel

Cross

Section

ð1Þ

Repeated Cross Section ð2Þ

Panel ð3Þ

Cross

Section

ð4Þ

Repeated Cross Section ð5Þ

Panel ð6Þ

A. 1901 Income B. 1950 Income with Adjustments

0–5 years in US 24,176.52 23,286.33 22,558.65 23,186.66 22,364.00 22,354.07 ð122.47Þ ð150.51Þ ð200.05Þ ð138.04Þ ð175.45Þ ð240.66Þ

6–10 years in US 23,433.90 22,723.76 21,900.42 22,450.13 21,797.87 21,521.15 ð130.80Þ ð144.10Þ ð174.11Þ ð144.89Þ ð165.09Þ ð206.19Þ

11–20 years in US 22,670.61 22,200.14 21,859.93 21,783.47 21,361.64 21,241.91 ð117.84Þ ð115.74Þ ð124.76Þ ð131.08Þ ð131.68Þ ð145.40Þ

21–30 years in US 22,402.06 22,032.18 21,896.77 21,540.32 21,227.39 21,127.69 ð124.08Þ ð117.95Þ ð124.79Þ ð139.89Þ ð135.54Þ ð146.22Þ

30 years in US 21,906.83 21,773.97 21,634.05 21,146.96 21,107.02 2814.98 ð148.13Þ ð139.57Þ ð144.52Þ ð175.02Þ ð168.41Þ ð177.33Þ

Arrive 18911 . . . 2740.37 2284.08 2745.86 220.58 ð82.96Þ ð127.89Þ ð97.89Þ ð150.86Þ

Native-born . . . . . . 580.02 28.22 ð200.05Þ ð145.85Þ

Observations 204,134 261,079 204,134 261,079

Note.—Columns 1–3 follow the format of table 4 using income from the 1901 Cost of Living survey. Columns 4–6 adjust the 1950 occupation-based earnings to match the Cost of Living survey in three ways: using only urban workers to calculate occupation-based earnings, using mean rather than median earnings by occupation, and using the census of agriculture rather than the census of population to infer earnings of farmers.

nation of immigrants 487

D. Geographic Location in the United States

In table 6, we adjust for aspects of immigrants’ location choices within the United States, first by controlling for state of residence and then by separately considering the urban subsample. Controlling for state of res- idence raises concerns about endogenous location choice; however, we

unskilled blue-collar—in Goldin and Margo ð1992Þ to “back-cast” what earnings in each occupation was likely to have been in the early 1920s. Goldin and Margo report that clerical earnings increased by 37 percent over this period, while both skilled and unskilled blue- collar earnings increased by 75 percent. We assume that all white-collar occupations in our sample ðprofessional, managerial, clerical, and salesÞ grew at the clerical rate while all blue- collar earnings grew at the skilled/unskilled rate. In doing so, we find that the occupation- based earnings gap between immigrants and natives upon first arrival is $500–$900 larger because immigrants are less likely than natives to hold these ðnow-higher-paidÞ white-collar jobs. In this case, long-term immigrants in our panel sample start out with a $650 deficit ðrather than a $450 premiumÞ relative to natives on a base of $23,000. However, we con- tinue to find that much of the observed gap between immigrants and natives in the cross section is due to the two sources of bias highlighted in the paper and that immigrants in the panel sample experience very little convergence relative to natives over time. The re- sults are presented in App. B.

This content downloaded from 132.174.255.086 on October 20, 2016 18:06:23 PM All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).

TABLE 6 Ordinary Least Squares Estimates, Age-Earnings Profile for Natives and

Foreign-Born, 1900–1920, Taking Geography into Account

Pooled Cross

Section and Panel

Pooled Cross

Section and Panel

ight-Hand-Side ariable

Cross

Section

ð1Þ Cross Section ð2Þ

Panel ð3Þ

Cross

Section

ð4Þ Cross Section ð5Þ

Panel ð6Þ

A. Add State Fixed Effects B. Urban Residents Only

–5 years in US 22,679.42 21,696.95 2955.81 24,209.72 23,355.25 21,983.01 ð154.57Þ ð198.43Þ ð242.28Þ ð179.89Þ ð232.60Þ ð304.03Þ

–10 years in US 22,125.78 21,323.73 2817.13 23,450.40 22,749.18 21,567.89 ð158.64Þ ð182.91Þ ð216.28Þ ð180.64Þ ð210.87Þ ð266.92Þ

1–20 years in US 21,662.49 21,123.49 2802.01 22,752.16 22,240.59 21,708.59 ð141.99Þ ð143.23Þ ð155.94Þ ð161.14Þ ð165.61Þ ð187.95Þ

1–30 years in US 21,473.55 21,045.93 21,024.76 22,329.79 21,913.40 22,188.47 ð153.48Þ ð149.28Þ ð155.92Þ ð173.94Þ ð170.52Þ ð191.53Þ

01 years in US 21,059.34 2955.81 2804.77 21,867.16 21,748.92 22,069.28 ð195.52Þ ð187.71Þ ð189.56Þ ð224.75Þ ð216.93Þ ð236.96Þ

rrive 18911 . . . 2907.09 2309.26 . . . 2770.35 2365.21 ð110.94Þ ð161.97Þ ð126.49Þ ð199.18Þ

ative-born . . . . . . 2156.52 . . . . . . 2258.68 ð172.08Þ ð238.42Þ

bservations 194,383 247,378 110,934 137,726

Note.—Regressions follow the format of table 4. Panel A adds a vector of state fixed ffects and panel B limits the sample to men living in urban areas in 1900. We define urban reas as counties in which at least 40 percent of the county’s residents lived in a town with population of 2,500 or more.

488 journal of political economy

All

R

V

0

6

1

2

3

A

N

O

e a a

believe that these specifications shed light on the mechanism underlying the occupation-based earnings difference between immigrants and na- tives. Panel A of table 6 adds state fixed effects, implicitly comparing immi-

grants and natives who settled in the same state. This adjustment doubles the immigrant occupation-based earnings penalty in the cross section and converts the small occupation-based earnings premium into an earnings penalty of $800 for long-term immigrants in the panel sample. In a com- parison of these findings to the main results in table 4, it appears that im- migrants achieved earnings parity with natives by moving to locations with a well-paid mix of occupations ðBorjas 2001Þ.31 The immigrant-native occupation-based earnings gap varies by state; immigrants outearn natives in the industrial states of the Midwest ðe.g., Ohio, Illinois, and MichiganÞ and underperform natives in industrial New England ðMassachusetts, Con-

31 Alternatively, immigrants may settle in states with higher costs of living. In this case, higher nominal wages may not translate into higher real wages. Comparing immigrants to natives within the same state may be closer to picking up differences in real wages.

This content downloaded from 132.174.255.086 on October 20, 2016 18:06:23 PM use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).

nation of immigrants 489

necticut, and Rhode IslandÞ and the Great Plains ðthe Dakotas, Iowa, Ne- braska, and MinnesotaÞ.32 Furthermore, we find an even larger occupation- based earnings gap between immigrants and natives when restricting the sample to urban residents in panel B ð$4,200 upon first arrival in the cross section and $1,700 ½5 21,983 1 258� upon first arrival in the panelÞ, perhaps because less productive immigrants settled in cities to take ad- vantage of the larger ethnic networks and the presence of immigrant aid societies.33 Note that although the immigrant occupation-based earnings penalty is larger in both cases, we continue to find that immigrants and natives experience little convergence in the panel sample and that long- term immigrants held occupations that pay more than those of the av- erage immigrant, consistent with negatively selected return migration.

VI. Heterogeneity by Sending Country

We have argued thus far that the typical long-term immigrant in the panel sample holds a slightly higher-paid occupation than the average native, even upon first arrival. However, this pattern masks substantial heterogeneity across sending countries. Figure 3 illustrates cross-country variation in the occupation-based earnings of immigrants relative to the native-born, both upon first arrival and after 30 or more years in the United States. The black bars indicate that immigrants from six of the 16 sending countries held occupations that paid significantly less than the native-born upon first arrival, immigrants from five countries held occu- pations that paid significantly more, and immigrants from the five re- maining countries exhibited little difference in earnings power relative to natives upon first arrival. We use data on real wages in the sending countries in 1880 from

Williamson ð1995Þ to subdivide the sample into richer ðabove-median real wagesÞ and poorer ðbelow-medianÞ sending countries.34 On average, long-term immigrants from poorer sending countries started out $1,700

34 Poorer countries include Denmark, Finland, Ireland, Italy, Norway, Portugal, and Sweden. Richer countries include Austria, Belgium, England, France, Germany, Scotland, Switzerland, and Wales. We assign real wages for Great Britain to England, Scotland, and Wales. Williamson ð1995Þ does not report wage data for Finland, Switzerland, or the Rus- sian Empire. We assign the Norwegian real wage to Finland and the German real wage to Switzerland. Results would not change if we chose other region-appropriate proxies ðsuch as Sweden or France, respectivelyÞ. Even if we did have real wage data for the Russian Empire,

32 These estimates are based on the 1900 cross section because the panel sample does not have enough cases to estimate state-specific premia. The underlying regression controls for immigrants’ country of origin, so they do not simply reflect the sending country mix in each state. Results are available in App. B.

33 We define an individual as urban if 40 percent or more of the county’s residents lived in a town with a population of 2,500 or more in the year 1900. This classification divides our sample roughly in half. We use this method because, in the panel sample, which was col- lected by hand from census manuscripts, we do not have information on the exact town or city in which an individual resided.

This content downloaded from 132.174.255.086 on October 20, 2016 18:06:23 PM All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).

FIG. 3.—Earnings gap between the native- and foreign-born in the panel sample: native versus immigrants upon first arrival ð0–5 years in the United StatesÞ and after time in the United States ð301 years in the United StatesÞ, by country of origin. The graph reports co efficients on the interaction between country-of-origin fixed effects and dummy variables for being in the United States for 0–5 years or for 301 years from regression of equation ð1Þ in the panel sample. All coefficients for the 0–5 year interaction are significant except those for Austria, Germany, Ireland, Italy, and Sweden. None of the differences between the 0–5 year and 301 year coefficients are significant except for those of Finland and Ireland.

35 We do not include Germany and Switzerland in these calculations as their popula tions were relatively evenly divided between Catholics and Protestants.

36 More formally, we tried regressing the earnings penalty ðor premiumÞ of recently ar rived immigrants on a set of economic characteristics for the sending country in 1880 and on measures of the linguistic, cultural, and religious difference between the source country and the United States. We find that immigrants from countries with a higher share of the labor force working in agriculture or a lower real wage hold lower-paid occupations relative to natives when they arrive in the United States. In contrast, immigrants from countries tha

it is not clear that these wages would have applied to the immigrants in our sample, many o whom were Russian Jews living in the western part of the empire. Thus, we analyze the Russian case separately below.

490 journal of political economy

This content downloaded from 132.174.255.086 on October 20, 2016 18:06:23 PM All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.

s

-

behind natives, while immigrants from rich sending countries already held occupations that paid $800 more than those of natives upon arrival. Another potentially relevant division was between predominantly Catho- lic and predominantly Protestant countries. Long-term immigrants from the typical Catholic country started out $600 behind the native-born, while immigrants from Protestant countries arrived about even with na- tives.35 Other factors that predict occupation-based earnings upon arrival are the linguistic and cultural distances between the source country and the United States.36

Comparing the black to the white bars in figure 3 demonstrates that, on the whole, permanent immigrants experience little occupational

-

-

t

f

edu/t-and-c).

FIG. 4.—Changing quality of arrival cohorts, difference between immigrant penalty for early and late arrivals in the panel sample, by country of origin. Estimates are based on the version of equation ð1Þ that contains country fixed effects and dummy variables for four arrival cohorts ðsee table 7, panel BÞ. In addition, we interact the country fixed effects with the dummy variables for arrival cohort. The graph reports the difference between the dummy variable for arriving in the United States between 1880 and 1885 and the dummy variable for arriving in the United States between 1895 and 1900, separately by country. Differences that are significantly different from zero are in black. The sample includes ob- servations in the panel sample.

nation of immigrants 491

growth relative to natives after spending time in the United States. Mi- grants from 10 countries experience a small amount of convergence rel- ative to natives over this period, while migrants from six countries either diverge from or experience a small reversal relative to natives. Migrants from the typical poorer country, with the exception of Finland, experi- enced $750 of convergence with the native-born over 30 years, closing 40 percent of their initial gap, while those from the typical rich country widened their lead over natives by nearly $300.37

Countries also differ in the degree of change in the skills of their ar- rival cohorts over time and in the selectivity of their return migration. We start by examining heterogeneity in arrival cohort skill level. Figure 4 reports differences by country between immigrants who arrived between

share a language, cultural background, or religious affiliation with residents of the United States are more successful in their new destination. Population pressure and health con- ditions in the source country, as measured by the rates of natural increase and of infant mortality, have no relationship with subsequent immigrant outcomes. We emphasize that, because of the small sample size ð16 countriesÞ and lack of exogenous variation, these re- lationships are merely suggestive. Results are available in App. B.

37 Finland is an outlier here, experiencing substantial divergence from the native-born over time. Including Finland in the calculation would lead us to conclude that those from the typical poor country experienced only $250 of convergence relative to natives over time. Although we hesitate to speculate about why Finland is so unique, one possible reason is that the country went through an extreme famine in 1868–69, in which 15 percent of its population perished. Early Finnish migrants to the United States may have been particu- larly negatively selected, moving simply to escape starvation.

This content downloaded from 132.174.255.086 on October 20, 2016 18:06:23 PM All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).

FIG. 5.—Implied selection of return migrants, difference between estimated convergence in panel and repeated cross-section data, by country of origin. The figure reports the differ ence between immigrants’ occupational upgrading relative to natives ðdefined as the differ ence between occupation-based earnings after 21–30 years and after 0–5 yearsÞ in the pane sample versus the cross section, by sending country. Results are from a regression of equa tion ð1Þ that pools the panel and cross-section samples. Coefficients that are significantly different from zero are in black.

38 We compare the difference in the 0–5 years in the United States indicators to the 21– 30 years in the United States variables rather than to the 301 years in the United State

492 journal of political economy

This content downloaded from 132.174.255.086 on October 20, 2016 18:06:23 PM All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.

- - l -

1880 and 1884 and those who arrived between 1895 and 1900. Coun- tries such as Russia and Italy, whose immigration waves began in large numbers only in the early 1880s, are among those with the largest de- cline in arrival cohort skill level over this period, perhaps because posi- tively selected “pioneer” migrants are replaced by the more typical mi- grant over time. Old immigrant groups such as the English and the Irish experience smaller declines in arrival cohort skill level ðor no decline at allÞ during this time. Figure 5 explores heterogeneity in the implied selection of return mi-

grants by sending country. In particular, we report the difference be- tween the coefficients on the 0–5 years in the United States indicators in the cross section versus the panel sample by sending country; recall that a negative value indicates that return migrants are negatively se- lected. We normalize the differences for the 0–5 year indicators by the difference between the cross section and the panel for the 21–30 years in the United States indicators to account for potential biases in match quality.38 The figure reveals statistically significant negative selection in the

s

edu/t-and-c).

nation of immigrants 493

return migration flow back to five sending countries ðDenmark, England, Italy, Russia, and SwitzerlandÞ and sizable positive but statistically insig- nificant selection to one country ðBelgiumÞ. The return migrant flow to the remaining 10 countries is neutral.39 The declining cohort quality among Italian immigrants, coupled with a negatively selected set of tem- porary migrants from Italy, may explain why the perception of Italian im- migrants to the United States was so poor by the 1910s despite the fact that, as we estimate in figure 3, long-standing Italian immigrants who ar- rived before 1900 held occupations quite similar to those of natives. Russia is another particularly interesting case. Figure 3 shows that Rus-

sian migrants performed well in the United States upon first arrival, and figure 5 suggests that return migrants to Russia were particularly nega- tively selected. These patterns can be explained by the ethnic composition of the Russian migration. The Russian migrant flow is made up of two groups, Jews and non-Jews, who were primarily Poles and other noneth- nic Russians. The Jewish immigrants were both higher skilled and less likely to return to Russia than their non-Jewish counterparts ðPerlman 1999Þ. In fact, only 7.1 percent of Russian Jews returned to Europe com- pared with 87 percent of Russian non-Jews ðGould 1980Þ. Therefore, the return migrant flow is made up primarily of low-skilled, non-Jewish Rus- sians.

VII. Alternative Specifications

A. Modifications to the Main Specification

Table 7 assesses the sensitivity of our main findings to a series of alter- native specifications. In each case, we continue to find ð1Þ limited con- vergence between immigrants and natives in the panel sample ð$300 or less after 30 years in the United StatesÞ and ð2Þ higher occupation-based earnings for long-term migrants in the panel than for the weighted aver- age of long-term and temporary migrants in the cross sections.

39 The height of the bars in fig. 5 represents the product of the return migration rate and the earnings gap between permanent and temporary migrants ðsee n. 28Þ. In a sepa- rate analysis, we use return migration rates by country reported either in Gould ð1980Þ or in Bandiera et al. ð2013Þ to back out the gap between permanent and temporary migrants by country. Gould reports return migration rates for Russian Jews and non-Jews separately ð7.1 percent and 87 percentÞ; we use the weighted average. Because there is little cross- country variation in the rates of return migration, the resulting ordering is nearly identical to the pattern reported in fig. 5 in both cases ðsee App. BÞ. The one exception is that re- turn migrants to Russia look even more negatively selected when we use the Bandiera et al. return migration rates.

variable because, in our 20-year sample, immigrants who are observed in their first 5 years in the United States are never observed at 301 years in the United States. In other words, the coefficients on the 0–5 year and 301 year indicators are derived from different arrival cohorts ðimmigrants who arrived before/after 1890Þ.

This content downloaded from 132.174.255.086 on October 20, 2016 18:06:23 PM All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).

TABLE 7 Robustness for Age-Earnings Profile in Panel Sample, 1900–1920

Cross Section ð1Þ

Repeated Cross Section ð2Þ

Panel ð3Þ

Cross Section ð4Þ

Repeated Cross Section ð5Þ

Panel ð6Þ

A. Without Country Fixed Effects B. Four Arrival Cohorts

0–5 years in US 2888.67 2216.01 558.44 21,255.73 256.53 579.92 ð115.93Þ ð169.86Þ ð226.70Þ ð143.44Þ ð219.51Þ ð275.95Þ

6–10 years in US 2239.95 290.97 717.56 2734.51 11.56 391.35 ð108.00Þ ð144.83Þ ð193.05Þ ð147.44Þ ð196.99Þ ð239.13Þ

11–20 years in US 164.98 507.08 557.89 2352.93 250.91 364.76 ð74.66Þ ð92.39Þ ð115.39Þ ð131.27Þ ð155.94Þ ð173.98Þ

21–30 years in US 177.53 424.65 283.32 2294.87 176.77 93.22 ð92.56Þ ð100.28Þ ð113.96Þ ð142.10Þ ð157.54Þ ð170.67Þ

301 years in US 373.49 339.10 430.04 22.41 211.37 260.04 ð142.90Þ ð148.39Þ ð155.31Þ ð184.65Þ ð187.14Þ ð196.94Þ

Arrive 18911 . . . 2660.11 2137.18 . . . . . . . . . ð107.32Þ ð163.46Þ

Native-born . . . . . . 2154.12 . . . . . . 2154.09 ð176.14Þ ð176.14Þ

C. lnðOccupation ScoreÞ D. Drop Child Migrants 0–5 years in US .006 .047 .084 21,288.05 2522.29 115.65

ð.008Þ ð.010Þ ð.012Þ ð144.40Þ ð192.51Þ ð245.32Þ 6–10 years in US .028 .063 .083 2842.87 2201.99 170.87

ð.007Þ ð.008Þ ð.010Þ ð149.46Þ ð178.79Þ ð222.92Þ 11–20 years in US .041 .065 .070 2535.17 230.39 42.86

ð.006Þ ð.006Þ ð.007Þ ð135.42Þ ð146.12Þ ð167.39Þ 21–30 years in US .034 .053 .059 2328.99 88.91 2174.59

ð.006Þ ð.006Þ ð.006Þ ð147.93Þ ð151.20Þ ð169.82Þ 301 years in US .041 .046 .061 71.42 188.59 183.61

ð.008Þ ð.008Þ ð.008Þ ð206.36Þ ð203.89Þ ð221.55Þ Arrive 18911 2.034 2.009 2671.59 284.89

ð.004Þ ð.007Þ ð114.15Þ ð172.86Þ Native-born 2.005 2153.59

ð.008Þ ð176.14Þ Note.—All regressions follow the specification in table 4 with the exception of the mod

ification listed in panel titles. In panel B, the four arrival cohorts are 1880–85, 1886–90 1891–95, and 1896–1900. Panel D drops immigrants who arrived in the United States before age 10 or after age 40. Standard errors are in parentheses. Sample size for panels A–C i 259,093; panel D has 243,671 observations.

494 journal of political economy

This content downloaded from 132.174.255.086 on October 20, 2016 18:06:23 PM All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.

- ,

s

Thus far, we have emphasized the changes in cohort quality that occur within a sending country as the immigrant flow increases over time. More broadly, the set of sending countries contributing to US immigration may have shifted over this period, leading the skill level of entrants to decline with arrival year. We assess this possibility in panel A, which omits country-of-origin fixed effects from the regression. This modification barely alters the comparison between the coefficients in the cross sec- tion and the repeated cross section, suggesting that declines in arrival cohort skill level occur primarily within sending countries, at least for

edu/t-and-c).

nation of immigrants 495

immigrants arriving before 1900. In particular, compare a difference be- tween single and repeated cross section coefficients for recent arrivals of $870 ðtable 4, with country fixed effectsÞ and $670 here ðwithout coun- try fixed effectsÞ. Panel B includes indicators for a series of finer arrival cohorts ðarrival 1886–90, 1891–95, and 1896–1900; arrival before 1885 is the omitted categoryÞ. These controls completely eliminate the occupation- based earnings gap between immigrants and natives in the repeated cross section, implying that the $1,200 earnings penalty in the cross section is entirely due to changes in arrival cohort skill level. Panel C replaces the dependent variable with the logarithm of our

1950 occupation-based earnings measure. In this case, immigrants in both the repeated cross section and the panel outearn natives upon first arrival, by 4.5 percent and 9.0 percent ð5 8.5 1 0.5Þ, respectively. Dif- ferences between the logarithm and levels specifications are driven by the concentration of natives at the top end of the occupation-based earn- ings distribution ðsee white-collar workers in fig. 1Þ; these lucrative occu- pations are more heavily weighted in the levels specification. Panel D excludes the 20 percent of the migrant sample who arrived in

the United States as children before the age of 10.40 Young immigrants may experience systematically different rates of assimilation as a result of heightened fluency in English or education in the US school system ðFriedberg 1993; Bleakley and Chin 2010Þ. We find that excluding child immigrants has little effect on the results.

B. Occupational Transition Matrices, 1900 –1920

The main results demonstrate that long-term immigrants moved up the occupation ladder at the same rate as natives. Table 8 examines these oc- cupational transitions directly, presenting transition matrices between 1900 and 1920 for natives and immigrants in the panel sample. As in figure 1, we use the HISCLASS classification collapsed into five categories to observe transitions between white-collar, skilled blue-collar, semiskilled blue-collar, farm, and unskilled work. The occupational transitions reveal a series of interesting patterns.

First, we find that, even though immigrants and natives experience sim- ilar occupation-based earnings growth, immigrants are more likely than natives to move both up and down the occupational ladder over time. As seen in the diagonal entries, immigrants are less likely to remain in the same occupational category in 1920 that they inhabited in 1900. This pattern is true both at the top of the occupational distribution ðwhite-

40 We choose the age of 10 because it is an age at which most people did not work, even in this historical period. Results are similar at cutoffs of age 12 or 14 as well.

This content downloaded from 132.174.255.086 on October 20, 2016 18:06:23 PM All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).

TABLE 8 Occupational Mobility from 1900 to 1920, Immigrants and Natives in Panel

1920

1900 White-Collar

Skilled Blue- Collar Farmer Semiskilled Unskilled

Row

Total

A. Native-Born

White-collar 66.71 8.10 6.67 17.14 2.38 100 ð138Þ ð17Þ ð14Þ ð36Þ ð5Þ ð210Þ

Skilled blue-collar 19.44 49.07 10.19 15.74 5.56 100 ð21Þ ð53Þ ð11Þ ð17Þ ð6Þ ð108Þ

Farmer 12.63 4.12 68.95 4.74 9.47 100 ð24Þ ð8Þ ð131Þ ð9Þ ð18Þ ð190Þ

Semiskilled 25.74 9.41 9.41 47.52 7.92 100 ð52Þ ð19Þ ð19Þ ð96Þ ð16Þ ð202Þ

Unskilled 17.77 11.15 33.80 13.24 24.04 100 ð51Þ ð32Þ ð97Þ ð38Þ ð69Þ ð287Þ

Column total 28.69 12.94 27.28 19.66 11.43 100 ð286Þ ð129Þ ð272Þ ð196Þ ð114Þ ð997Þ

B. Immigrants

White-collar 55.34 12.49 7.83 14.50 9.84 100 ð1,045.44Þ ð235.96Þ ð148.00Þ ð273.89Þ ð185.84Þ ð1,889.15

Skilled blue-collar 27.37 35.84 9.12 16.81 10.86 100 ð611.59Þ ð800.95Þ ð203.86Þ ð375.59Þ ð242.57Þ ð2,234.57

Farmer 12.81 6.88 63.20 7.89 9.23 100 ð155.32Þ ð83.45Þ ð766.28Þ ð95.61Þ ð111.87Þ ð1,212.55

Semiskilled 29.49 13.16 8.29 34.41 14.64 100 ð1,048.43Þ ð467.72Þ ð294.08Þ ð1,223.25Þ ð520.45Þ ð3,554.67

Unskilled 21.40 12.30 22.69 20.16 23.45 100 ð791.17Þ ð454.81Þ ð838.81Þ ð745.14Þ ð867.13Þ ð3,697.06

Column total 29.01 16.23 17.89 21.56 15.32 100 ð3,651.95Þ ð2,042.91Þ ð2,251.76Þ ð2,713.50Þ ð1,927.88Þ ð12,588Þ

Note.—Occupations are classified according to the HISCLASS rubric. HISCLASS 1–5 5 white-collar; HISCLASS 6–7 5 skilled blue-collar; HISCLASS 8 5 farmers; HISCLASS 9 5 semiskilled; HISCLASS 10–12 5 unskilled. Each cell reports the share of immigrant ðnativesÞ in a certain occupation class in 1900 ðrowÞ and in 1920 ðcolumnÞ. In parenthese is the number of cases underlying each percentage. Because the immigrant figures are weighted to reflect population shares in 1920, the numbers of cases in panel B are non integer.

496 journal of political economy

This content downloaded from 132.174.255.086 on October 20, 2016 18:06:23 PM All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.

Þ

Þ

Þ

Þ

Þ

s s

-

collar positionsÞ and at the lower end of the occupational scale ðsemi- skilled blue-collarÞ. Second, immigrants and natives use different rungs to move up the

ladder. For example, 32 percent of immigrants who held unskilled jobs in 1900 ascend into skilled or semiskilled blue-collar work by 1920, com- pared with only 24 percent of similarly positioned natives. In contrast, 34 percent of formerly unskilled natives move into owner-occupier farm- ing by 1920, compared with only 23 percent of unskilled immigrants; some of the transitions into farming for the native-born could be driven by inheriting a family farm.

edu/t-and-c).

nation of immigrants 497

As we saw in figure 1 above, natives were more likely to work in farm- ing in 1900, while immigrants were more likely to hold skilled blue-collar positions. Over the next 20 years, natives and immigrants continue to follow these divergent strategies to get ahead. On average, though, these different paths lead to equal occupation-based earnings growth ðespe- cially if farming is treated as an occupation with below-median earnings, as in the 1950 earnings distributionÞ.

VIII. Ruling Out Other Sources of Selective Attrition

Our empirical approach infers the direction of selection of return mi- grants relative to long-term migrants indirectly, by comparing occupa- tional upgrading patterns in the repeated cross section versus the panel data. Yet, more generally, any differences between the repeated cross sections and the panel are due to selective attrition from the cross sec- tions, which can be due to selective return migration but could also be due to selective mortality or selective name changes. We argue here that mortality and name changes are not likely to be driving the results. Selective mortality is not a likely concern. Mortality in 1900 for this

age group ðages 15–45Þ was fairly low and uniform across sending coun- tries. Although the Irish were slightly more likely to die ðeight per 1,000Þ and the Russians were slightly less likely to die ðthree per 1,000Þ, mortal- ity rates for members of other nationalities and for US natives were all around five to six per 1,000 ðfigures by Merriam ½1903�, based on the 1900 censusÞ. Furthermore, note that selective return migration is not an issue for the native-born because few US natives emigrated from the country. Therefore, one way to test for the presence of selective mortality is to compare the occupation-based earning patterns of native-born men in the repeated cross section versus the panel data. For natives, any difference between these samples can be due to selective mortality but not to out- migration. We find that the occupation-based earnings of natives are sim- ilar in the repeated cross sections and the panel in all years, suggesting that selective mortality is a nonissue ðat least for the native-bornÞ.41 We note that this test for selective mortality relies on the assumption that na- tive and foreign-born men were subject to the same mortality process. Likewise, we do not expect that selective name changes by immigrants

will bias the data. First, most name changes occurred upon entry to the United States ðe.g., at Ellis IslandÞ. Any such change would have taken

41 We regress the occupation-based earnings score on a dummy for being in the panel sample for the native-born. In 1900, e.g., the coefficient on this dummy variable is 20.212 ðstandard error 5 0.294Þ. After adjusting for age differences between the two samples, the difference falls further to 20.130 ðstandard error 5 0.288Þ. This finding is consistent with the presence of a minimal relationship between socioeconomic status and health in the early twentieth century ðFrank and Mustard 1994; Hummer and Lariscy 2011Þ.

This content downloaded from 132.174.255.086 on October 20, 2016 18:06:23 PM All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).

498 journal of political economy

All

place before we first observe migrants in the 1900 census and would thus affect neither data source. Second, men who changed their name be- tween censuses are not likely to affect the results because name changers cannot be matched over time, so they are never included in the panel sample; name changing has nothing to do with enumeration in the cen- sus, and so name changers will always be present in the cross sections. That is, unlike men who die between census waves, name changers do not drop out of the cross-section data over time. Third, we find that, al- thoughimmigrants inthepanelsample haveslightly more“foreign”names than their counterparts in the cross section, the small observed differ- ence in the “foreignness” index is associated with only a $60 difference in occupation-based earnings ðin 2010 dollarsÞ and so is not quantitatively large enough to affect the results.42

IX. Second-Generation Migrants in the US Labor Market

Occupational convergence between immigrants and natives may take more than one generation. On the one hand, second-generation mi- grants may attain equal standing with natives because they were educated in the United States and, therefore, were likely fluent in English and may have been exposed to US norms and culture. On the other hand, occupational differences could persist over generations if, for example, second-generation migrants grew up in migrant enclaves or inherited occupational skills from their parents.43

We compare the occupation-based earnings of US-born men whose parents were born abroad to US-born men whose parents were born in the United States ðhereafter referred to as US natives, even though second- generation immigrants are also born in the United StatesÞ. In particular, we use our panel sample to compare first-generation immigrants to US natives and supplement this with the 1 percent IPUMS samples of the US census from 1920 to 1950, which we use to compare US natives to the co- hort of children born to these first-generation immigrants.44

42 Following Fryer and Levitt ð2004Þ, the foreignness index is constructed by first calcu- lating the probability of being foreign-born conditional on having a given first name ðand, separately, a given last nameÞ in the 1900–1920 IPUMS samples. The foreignness index is then the sum of the two probabilities; the index varies between zero and two. Foreign-born men in the cross section ðpanel sampleÞ have an index value of 1.13 ð1.23Þ.

43 Borjas ð1994Þ and Leon ð2005Þ examine the effect of parental literacy and “ethnic capital,” or the average skills in one’s ethnic group, on the literacy, school attendance, and wages of the second generation during the Age of Mass Migration. They document that both within-household and within–ethnic group transmission are important for the skill development and, therefore, for the persistence of skill differentials between groups.

44 We draw the sample of second-generation immigrants, defined as men with two par- ents from the same country of origin, from the censuses of 1920–50 and compare them to US natives in those years. We focus on nonsouthern men between the ages of 20 and 60. Because census records are made public only after 72 years, we are unable to construct a

This content downloaded from 132.174.255.086 on October 20, 2016 18:06:23 PM use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).

FIG. 6.—Convergence in occupation-based earnings across immigrant generations: first- generation and second-generation migrants versus natives, by country of origin. We esti- mate the regression in equation ð2Þ separately for each group and for each country: im- migrants ðfirst generationÞ, US natives in the same censuses and ages as the immigrants, sons of immigrants ðsecond generationÞ, and US natives in the same censuses and ages as the second-generation sample. The bars for the first generation represent the difference in the predicted occupation-based earnings of an immigrant who came in 1890 and is 35 years old in 1910 relative to a 35-year-old native. The bars for the second generation represent the difference in the predicted occupation-based earnings of a man born in the United States to immigrant parents relative to a man born in the United States to native parents, both of whom were 35 years old in 1930. First-generation immigrants are taken from the panel sample. Natives and second-generation immigrants come from IPUMS data in the respective census year.

nation of immigrants 499

We estimate the following age-earnings profile separately for each group and for each country of origin:

Yit 5 at 1 baAgeit 1 ba2Age 2 it 1 ba3Age

3 it 1 ba4Age

4 it

1 Migranti � o k

gkYearsIUSitk 1 εit: ð2Þ

As before, our outcome variable is occupation-based earnings converted to 2010 dollars. In figure 6, we illustrate the results from equation ð2Þ for a person who is 35 years old in either 1910 ðfirst generation vs. na- tivesÞ or 1930 ðsecond generation vs. nativesÞ. We assume that the first- generation migrant moved to the United States in 1890. Figure 6 suggests strong evidence of persistence across generations. If

the first-generation immigrants outperformed natives ðe.g., Russia, Scot- land, EnglandÞ, so did the second generation and vice versa ðe.g., Norway,

panel sample that matches children to their parents in this period. Note that second- generation migrants are not subject to the two sources of bias that affect the first gener- ation—namely, changes in arrival cohort skill level and selective return migration—and so following birth cohorts through repeated cross sections provides an accurate measure of occupational progress.

This content downloaded from 132.174.255.086 on October 20, 2016 18:06:23 PM All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).

500 journal of political economy

All

PortugalÞ.45 A notable exception is Finland, in which first-generation mi- grants held lower-paid occupations but second-generation migrants held higher-paid occupations.

X. Conclusion

We construct a new panel data set of native- and foreign-born men in the US labor market at the turn of the twentieth century, an era in which US borders were open to all European migrants. This Age of Mass Mi- gration not only is of interest in itself, as one of the largest migration waves in modern history, but also is informative about the process of im- migrant assimilation in a world without migration restrictions. Most of the previous research on this era relies on a single cross section of data and finds that immigrants started with lower-paid occupations than natives but caught up with natives after spending some time in the United States. In our panel data set, we instead find that the average immigrant who

settled in the United States long-term did not hold lower-paid occupa- tions than US natives, even upon first arrival, and moved up the occu- pational ladder at the same rate as natives. We conclude that the appar- ent convergence in a single cross section reflects a substantial decline in the quality of migrant cohorts over this period as well as a change in the composition of the migrant pool as negatively selected return migrants left the United States over time. Our paper further demonstrates the importance of accounting for differences in migration patterns across sending countries. Long-term migrants from highly developed sending countries performed better than natives upon first arrival, while long- term migrants from poorer sending countries performed worse. Yet im- migrants from all countries, regardless of their starting position, experi- enced little occupational convergence with natives. Contemporaries questioned the ability of European immigrants to

assimilate in the US economy and called for strict migration restrictions. Our results indicate that these concerns were unfounded: the average long-term immigrants in this era arrived with skills similar to those of natives and experienced identical rates of occupational upgrading over their life cycle. These successful outcomes suggest that migration re- strictions are not necessary to ensure migrant assimilation. Yet at the same time, we also note that migrants who arrived with low skill levels did not manage to close their skill gap with natives over time. This finding undercuts the view, which is commonly held today, that past waves of European immigrants, even those who arrived with limited skills and

45 The magnitudes of the occupation-based earnings gap between first-generation im- migrants and natives reported here differ from those in fig. 3 because, here, we are looking at the average earnings of a 35-year-old, whereas in fig. 3 we report mean earnings for recent arrivals ðin United States 0–5 yearsÞ.

This content downloaded from 132.174.255.086 on October 20, 2016 18:06:23 PM use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).

nation of immigrants 501

without the ability to speak English, were able to quickly catch up with natives.

Appendix A

Data

This appendix describes the procedure by which we match men from the 1900 census to the 1910 and 1920 censuses. We begin by identifying a sample of men in the base year from two census sources. For large sending countries ðlisted in table 1, panel AÞ, we rely on the 1900 5 percent Integrated Public Use Microdata Series ðIPUMS; Ruggles et al. 2010Þ to find immigrants from large sending countries and to randomly select a sample of 10,000 native-born men. To en- sure a sufficient sample size for smaller sending countries ðtable 1, panel BÞ, we instead compile the full population in the relevant age range in 1900 from the genealogy website Ancestry.com. Altogether, we identify immigrants from 16 send- ing countries.46

We search for viable matches for these men in 1910 and 1920 using the itera- tive matching strategy developed by Ferrie ð1996Þ and employed more recently by Abramitzky et al. ð2012Þ and Ferrie and Long ð2013Þ. More formally, our matching procedure proceeds as follows:

1. We begin by standardizing the first and last names of men in our 1900 samples to address orthographic differences between phonetically equiv- alent names using the NYSIIS algorithm ðsee Atack and Bateman 1992Þ. We restrict our attention to men in 1900 who are unique by first and last name, birth year, and place of birth ðeither state or countryÞ in the 1900 census. We do so because, for nonunique cases, it is impossible to deter- mine which of the records should be linked to potential matches in 1910 and 1920. Table 1 presents information about the number of potential matches by country.

2. We identify potential matches in 1910 and 1920 by searching for all men in our 1900 sample in the 1910 and 1920 census manuscripts available from Ancestry.com. The Ancestry.com search algorithm is expansive and returns many potential matches for each case, which we cull using the iterative match procedure described in the next step.47

46 We include men from all European sending countries with at least 10,000 migrants living in the United States in 1900, with the exception of Czechoslovakia, Hungary, the Netherlands, Poland, and Romania. Altogether, our sample represents 91 percent of Eu- ropean immigrants living in the United States in 1900. Poland and Czechoslovakia were removed entirely from the 1910 census as options for the birthplace variable and then re- stored in 1920 after both countries gained their independence in 1918. Although Hungary and Romania remained birthplace options in all years, their borders changed substantially between 1910 and 1920. Migrants from the Netherlands reported varied ðoften provincial- levelÞ birthplaces on the census, making it difficult to find these individuals in Ancestry.com and follow them over time.

47 The Ancestry.com search engine aims to maximize potential “hits” under the assump- tion that individual users can identify their relatives from a longer list by hand. To this end, it uses many approaches to convert names into their phonetic equivalents and applies a very

This content downloaded from 132.174.255.086 on October 20, 2016 18:06:23 PM All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).

TABLE A1 Comparing Matched Panel Sample with Population,

1920 Occupation-Based Earnings in 2010 Dollars

Mean,

Panel Sample

ð1Þ

Difference, Panel

Sample 2 Population

Levels ð2Þ

Logs ð3Þ

Native-born $23,185 37.64 .009 ð316.41Þ ð.013Þ

Foreign-born $23,405 303.63 .019 ð131.81Þ ð.006Þ

Note.—Occupation-based earnings is based on 1950 medians, con- verted into 2010 dollars. Regressions in cols. 2 and 3 pool the 1920 IPUMS cross section with our matched sample and regress occupation-based earnings on a dummy variable for being in the matched sample. Standard errors are in parentheses.

lax matching rule. For small sending countries, we instead match the complete 1900 popu lation to the complete 1910 and 1920 populations obtained from Ancestry.com.

502 journal of political economy

This content downloaded from 132.174.255.086 on October 20, 2016 18:06:23 PM All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.

3. We match observations forward from 1900 to either the full population ðfor small countriesÞ or the set of potential matches ðfor large countriesÞ in 1910 and 1920 using an iterative procedure. We start by looking for a match by first name, last name, place of birth ðeither state or countryÞ, and exact birth year. There are three possibilities: ðaÞ if we find a unique match, we stop and consider the observation “matched”; ðbÞ if we find multiple matches for the same birth year, the observation is thrown out; ðcÞ if we do not find a match at this first step, we try matching within a 1-year band ðolder and youngerÞ and then with a 2-year band around the reported birth year; we accept only unique matches. If none of these at- tempts produces a match, the observation is discarded as unmatched.

4. After matching each sample in 1900 separately to 1910 and 1920, we create our final data set by restricting to men who were located in both 1910 and 1920.

Our matched sample may not be fully representative of the immigrant and native-born populations from which they are drawn. In particular, men with un- common names are more likely to be successfully linked between censuses, and the commonness of one’s name could potentially be correlated with socioeco- nomic status. We assess this possibility by comparing men in the cross-sectional and panel samples in 1920. By definition, men in both the panel and repeated cross sections must have survived and remained in the United States until 1920. Thus, by 1920, up to sampling error, differences between the panel and the re- peated cross sections must be due to an imperfect matching procedure.

Table A1 compares the mean occupation score of men in our cross-section and panel samples in 1920. We consider natives and the foreign-born separately

-

edu/t-and-c).

nation of immigrants 503

and reweight the matched sample to reflect the distribution of country of origins in the 1920 population. Immigrants in the matched sample slightly outearn their native counterparts by 1920 ð$23,400 vs. $23,200Þ. Among natives, the difference in the mean occupation score in the matched sample and the population in 1920 is small ð$37Þ and statistically indistinguishable from zero. In contrast, im- migrants in the matched sample have a $300 advantage over immigrants in the representative sample.48 Therefore, up to $300 of earnings differential between immigrants and natives in the panel data can be due to sample selection induced by our matching procedure.

References

Abramitzky, Ran, Leah Boustan, and Katherine Eriksson. 2012. “Europe’s Tired, Poor, Huddled Masses: Self-Selection and Economic Outcomes in the Age of Mass Migration.” A.E.R. 102 ð5Þ: 1832–56.

———. 2013. “Have the Poor Always Been Less Likely to Migrate? Evidence from Inheritance Practices during the Age of Mass Migration.” J. Development Econ. 102:2–14.

Alba, Richard, and Victor Nee. 2003. Remaking the American Mainstream: Assimi- lation and Contemporary Immigration. Cambridge, MA: Harvard Univ. Press.

Atack, Jeremy, and Fred Bateman. 1992. “‘Matchmaker, Matchmaker, Make Me a Match’: A General Personal Computer–Based Matching Program for Histori- cal Research.” Hist. Methods 25 ð2Þ: 53–65.

Bandiera, Oriana, Imran Rasul, and Martina Viarengo. 2013. “The Making of Modern America: Migratory Flows in the Age of Mass Migration.” J. Development Econ. 102 ðMayÞ: 23–47.

Benton-Cohen, Katherine. 2010. “The Rude Birth of Immigration Reform.” Wil- son Q. 34 ð3Þ: 16–22.

Blau, Francine. 1980. “Immigration and Labor Earnings in Early Twentieth Century America.” Res. Population Econ. 2:21–41.

Bleakley, Hoyt, and Aimee Chin. 2010. “Age at Arrival, English Proficiency, and Social Assimilation among U.S. Immigrants.” American Econ. J.: Appl. Econ. 2 ð1Þ: 165–92.

Borjas, George. 1985. “Assimilation, Changes in Cohort Quality, and the Earn- ings of Immigrants.” J. Labor Econ. 3 ð4Þ: 463–89.

———. 1989. “Immigrant and Emigrant Earnings: A Longitudinal Study.” Econ. Inquiry 27 ð1Þ: 21–37.

———. 1994. “Long-Run Convergence of Ethnic Skill Differentials: The Chil- dren and Grandchildren of the Great Migration.” Indus. and Labor Relations Rev. 47 ð4Þ: 553–73.

———. 2001. “Does Immigration Grease the Wheels of the Labor Market?” Brookings Papers Econ. Activity ð1Þ: 69–133.

Butcher, Kristin F., and John DiNardo. 2002. “Immigrant and Native-Born Wage Distributions: Evidence from United States Censuses.” Indus. and Labor Rela- tions Rev. 56 ð1Þ: 97–121.

Carter, Linda. 2008. “A Hard Day’s Night: Evening Schools and Child Labor in the United States, 1870–1910.” Manuscript, Vanderbilt Univ.

48 Country-by-country comparisons reveal that this gap is generated by five sending countries: Belgium, France, Ireland, Italy, and Norway. Results are robust to dropping these five countries from the analysis.

This content downloaded from 132.174.255.086 on October 20, 2016 18:06:23 PM All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).

C

C

D

D

D

Ec

Ed

Ei

Fe

—

—

Fe

Fr

Fr

Fr

G

G

G

G

H

H

H

504 journal of political economy

All us

hiswick, Barry R. 1978. “The Effect of Americanization on the Earnings of Foreign-Born Men.” J.P.E. 86 ð5Þ: 897–921. onstant, Amelie, and Douglas S. Massey. 2003. “Self-Selection, Earnings, and Out-Migration: A Longitudinal Study of Immigrants to Germany.” J. Population Econ. 16 ð4Þ: 631–53. ouglas, Paul. 1919. “Is the New Immigration More Unskilled than the Old?” Pubs. American Statis. Assoc. 16 ð126Þ: 393–403. uleep, Harriett O., and Daniel J. Dowhan. 2002. “Insights from Longitudinal Data on the Earnings Growth of U.S. Foreign-Born Men.” Demography 39 ð3Þ: 485–506. ustmann, Christian. 1993. “Earnings Adjustment of Temporary Migrants.” J. Population Econ. 6 ð2Þ: 153–68. kstein, Zvi, and Yoram Weiss. 2004. “On the Wage Growth of Immigrants: Israel, 1990–2000.” J. European Econ. Assoc. 2 ð4Þ: 665–95. in, Pers-Anders, Robert J. LaLonde, and Olof Aslund. 2000. “Emigration of Immigrants and Measures of Immigrant Assimilation: Evidence from Swe- den.” Swedish Econ. Policy Rev. 7 ð2Þ: 163–204. chengreen, Barry, and Henry A. Gemery. 1986. “The Earnings of Skilled and Unskilled Immigrants at the End of the Nineteenth Century.” J. Econ. Hist. 46 ð2Þ: 441–54. rrie, Joseph P. 1996. “A New Sample of Males Linked from the Public Use Micro Sample of the 1850 U.S. Federal Census of Population to the 1860 U.S. Federal Census Manuscript Schedules.” Hist. Methods 29 ð4Þ: 141–56. ——. 1997. “The Entry into the U.S. Labor Market of Antebellum European Immigrants, 1840–1860.” Explorations Econ. Hist. 34 ð3Þ: 295–330. ——. 1999. Yankees Now: Immigrants in the Antebellum United States, 1840–1860. New York: Oxford Univ. Press. rrie, Joseph P., and Jason Long. 2013. “Intergenerational Occupational Mo- bility in Great Britain and the United States since 1850.” A.E.R. 103 ð7Þ: 1109– 37. ank, John W., and Fraser Mustard. 1994. “The Determinants of Health from a Historical Perspective.” Daedalus 123 ð4Þ: 1–19. iedberg, Rachel M. 1993. “The Labor Market Assimilation of Immigrants in the United States: The Role of Age at Arrival.” Manuscript, Brown Univ. yer, Roland, and Steven D. Levitt. 2004. “The Causes and Consequences of Distinctively Black Names.” Q. J.E. 119 ð3Þ: 767–805. alor, Oded, and Oded Stark. 1990. “Migrants’ Savings, the Probability of Return Migration and Migrants’ Performance.” Internat. Econ. Rev. 31 ð2Þ: 463–67. oldin, Claudia. 1994. “The Political Economy of Immigration Restriction in the United States, 1890 to 1921.” In The Regulated Economy: A Historical Approach to Political Economy, edited by Claudia Goldin and Gary D. Libecap, 223–57. Chi- cago: Univ. Chicago Press. oldin, Claudia, and Robert A. Margo. 1992. “The Great Compression: The Wage Structure in the United States at Mid-Century.” Q. J.E. 107 ð1Þ: 1–34. ould, John D. 1980. “European Inter-continental Emigration. The Road Home: Return Migration from the U.S.A.” J. European Econ. Hist. 9 ð1Þ: 41–112. anes, Christopher. 1996. “Immigrants’ Relative Rate of Wage Growth in the Late 19th Century.” Explorations Econ. Hist. 33 ð1Þ: 35–64. annon, Joan. 1982. “Ethnic Discrimination in a 19th-Century Mining District: Michigan Copper Mines, 1888.” Explorations Econ. Hist. 19 ð1Þ: 28–50. atton, Timothy. 1997. “The Immigrant Assimilation Puzzle in Late Nineteenth- Century America.” J. Econ. Hist. 57 ð1Þ: 34–62.

This content downloaded from 132.174.255.086 on October 20, 2016 18:06:23 PM e subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).

H

H

H

H

H

H

Ja

Ja

Ki

Le

Li

Ll

Ll

Lu

—

M

M

M

M

M

Pe

Pi

Pr

nation of immigrants 505

Al

atton, Timothy J., and Jeffrey G. Williamson. 1998. The Age of Mass Migration: Causes and Economic Impact. New York: Oxford Univ. Press. iggs, Robert. 1971. “Race, Skills, and Earnings: American Immigrants in 1909.” J. Econ. Hist. 32 ð2Þ: 420–28. igham, John. 1988. Strangers in the Land: Patterns of American Nativism, 1860– 1925. 2nd ed. New Brunswick, NJ: Rutgers Univ. Press. u, Wei-Yin. 2000. “Immigrant Earnings Assimilation: Estimates from Longitu- dinal Data.” A.E.R. Papers and Proc. 90:368–72. ummer, Robert, and John T. Lariscy. 2011. “Educational Attainment and Adult Mortality.” In International Handbook of Adult Mortality, vol. 2, edited by R. Rog- ers and E. M. Crimmins, 241–61. New York: Springer. untington, Samuel. 2004. Who Are We? The Challenges to America’s National Iden- tity. New York: Simon & Schuster. cobson, Matthew F. 1999. Whiteness of a Different Color: European Immigrants and the Alchemy of Race. Cambridge, MA: Harvard Univ. Press. sso, Guillermina, and Mark R. Rosenzweig. 1988. “How Well Do US Immigrants Do? Vintage Effects, Emigration Selectivity, and Occupational Mobility.” Res. Population Econ. 6 ð3Þ: 229–53. m, Seik. 2011. “Economic Assimilation of Foreign-Born Workers in the United States: An Overlapping Rotating Panel Analysis.” Manuscript, Univ. Washington. on, Alexis. 2005. “Does ‘Ethnic Capital’ Matter? Identifying Peer Effects in the Intergenerational Transmission of Ethnic Differentials.” Manuscript, Univ. Pitts- burgh. eberson, Stanley. 1980. A Piece of the Pie: Blacks and White Immigrants since 1880. Berkeley: Univ. California Press. eras-Muney, Adriana. 2002. “Were Compulsory Attendance and Child Labor Laws Effective? An Analysis from 1915 to 1939.” J. Law and Econ. 45 ð2Þ: 401–35. eras-Muney, Adriana, and Allison Shertzer. 2011. “The Americanization Move- ment, 1910–1930.” Manuscript, Dept. Econ., Univ. California, Los Angeles. botsky, Darren. 2007. “Chutes or Ladders? A Longitudinal Analysis of Immi- grant Earnings.” J.P.E. 115 ð5Þ: 820–67. ——. 2011. “The Effect of Changes in the U.S. Wage Structure on Recent Immigrants’ Earnings.” Rev. Econ. and Statis. 93 ð1Þ: 59–71. cGouldrick, Paul F., and Michael B. Tannen. 1977. “Did American Manu- facturers Discriminate against Immigrants before 1914?” J. Econ. Hist. 37 ð3Þ: 723–46. erriam, William R. 1903. Twelfth Census of the United States, Taken in the Year 1900: Statistical Atlas. Washington, DC: US Census Office. inns, Chris. 2000. “Income, Cohort Effects and Occupational Mobility: A New Look at Immigration to the United States at the Turn of the 20th Century.” Explorations Econ. Hist. 37 ð4Þ: 326–50. oehling, Caroline M. 1999. “State Child Labor Laws and the Decline of Child Labor.” Explorations Econ. Hist. 36 ð1Þ: 72–106. uller, Thomas. 1993. Immigrants and the American City. New York: New York Univ. Press. rlman, Joel. 1999. “Selective Migration as a Basis for Upward Mobility? The Occupation of the Jewish Immigrants to the United States, ca. 1900.” Manu- script. http://econpapers.repec.org/paper/wpawuwpma/9805023.htm. ore, Michael J. 1980. Birds of Passage: Migrant Labor and Industrial Societies. Cambridge: Cambridge Univ. Press. eston, Samuel H., and Michael R. Haines. 1991. Fatal Years: Child Mortality in Late Nineteenth-Century America. Princeton, NJ: Princeton Univ. Press.

This content downloaded from 132.174.255.086 on October 20, 2016 18:06:23 PM l use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).

506 journal of political economy

All

Ruggles, Stephen, J. Trent Alexander, Katie Genadek, Ronald Goeken, Matthew B. Schroeder, and Matthew Sobek. 2010. Integrated Public Use Microdata Series: Version 5.0 ½machine-readable database�. Minneapolis: Univ. Minnesota.

Sowell, Thomas. 1996. Migrations and Cultures: A World View. New York: Basic Books.

van Leeuwen, Marco, and Ineke Maas. 2005. “A Short Note on HISCLASS.” http://historyofwork.iisg.nl/docs/hisclass-brief.doc.

Williamson, Jeffrey. 1995. “The Evolution of Global Labor Markets since 1830: Background Evidence and Hypotheses.” Explorations Econ. Hist. 32 ð2Þ: 141–96.

Wyman, Mark. 1996. Round Trip to America: The Immigrants Return to Europe, 1880– 1930. Ithaca, NY: Cornell Univ. Press.

Zakharenko, Roman. 2009. “Return Migration: An Empirical Investigation.” MPRA Paper no. 13755, Univ. Library Munich.

This content downloaded from 132.174.255.086 on October 20, 2016 18:06:23 PM use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).