U N I V E R S I T Y O F O X F O R D Discussion Papers in Economic and Social History Num ber 161, January 2018 Skill Selection and American Immigration Policy in the Interwar Period Alexander A. J. Wulfers Skill Selection and American Immigration Policy in the Interwar Period Alexander A. J. Wulfers∗ January 24, 2018 Abstract The Age of Mass Migration came to an end in the interwar period with new American immigration restrictions, but did this end affect some potential migrants more than others? I use previously unanalysed data from passenger lists of ships leaving Bremen, one of the major European ports of emigration, between 1920 and 1933, to identify occupations and skill levels of individual migrants. The main focus of the paper is on the role that policy played in influencing the selection of migrants. I study the American quota laws of 1921, 1924, and 1929, and find that increasingly strict quotas led to an increase in the skill level of migrants as well as a shift from agricultural to manufacturing workers first, and from manufacturing to professional workers later. JEL Codes: J15, K37, N32, N34. Keywords: immigrationpolicy,skillselection,quotas,UnitedStates, Bremen, interwar period. ∗St John’s College, University of Oxford, contact: [email protected] 1 Introduction The Age of Mass Migration in the Atlantic economy came to an end in the interwar period when the United States gradually restricted migration from Europe. Before World War I, European emigrants faced few entry restriction policies in the Americas.1 This changed with the literacy test in 1917, followed bythenationalquotalawsin1921, 1924, and1929. EmigrationfromtheNorth and West of Europe had at this point decreased significantly while emigration from the South and East had grown a lot up to the war. Nonetheless, absolute migration rates were still high in the “old” migrant source countries by modern standards and were affected far less than in the newer ones by the new restric- tions.2 Europe remained a continent of emigration rather than immigration in the interwar years, with 1.46 million people leaving the continent between 1921 and 1939.3 The characteristics of these emigrants constitute the focus of this article. I analyse specifically how the skill composition of the migrant movement changed as a consequence of tightening American immigration laws. The quotas set an upper bound for each country’s number of migrants but did not explicitly discriminate within countries. The question therefore is whether they had any impact on skill levels and whether they disproportionally hurt the low-skilled. In this endeavour I draw upon a new source that has previously not been used for econometric analysis. I have compiled all remaining passenger lists of the port of Bremen through which a large share of Europeans left the con- tinent for the United States and other destinations, mostly in the Western hemisphere. These lists have been digitised, transcribed and published online by the Bremen Genealogical Society.4 Until a few years ago research on histor- 1. T. Hatton and J. Williamson, The Age of Mass Migration: Causes and Economic Impact (Oxford University Press, 1998). 2. A. McKeown, “Global Migration, 1846-1940,” Journal of World History 15, no. 2 (2004): 155–189. 3. D. Kirk, Europe’s population in the interwar years (Taylor & Francis, 1969), 279. 4. K. Wesling, ed., “Bremer Passagierlisten,” 2009, http://www.passengerlists.de. 2 ical immigration patterns to the United States had to rely on rough estimates and macro data, and most literature relies on one central source for migration statistics, compiled by Ferenczi and Willcox.5 Only recently has there been a revival of research on historical migration with the surge in digitisation of previously unavailable micro data collected from arrival port statistics such as Ellis Island,6 census data,7 or subsamples of passenger lists.8 This article will continue this trend and make use of the detailed micro data available in the Bremen passenger lists. The four major developments in US immigration policy, the 1917 literacy test, the introduction of the quota system in 1921, and its subsequent reforms in1924and1929, allconstitutedsevereexogenousshocksforEuropeanemigra- tion, the effect of which varied a lot across countries. Since the quota laws were non-discriminatory within source countries, they raise the question of whether the composition of migrants was affected by this overall reduction. The large cross-country variation allows me to use those less or not at all affected by the quotas as a control group and thus analyse the effect of the policies in a difference-in-differences framework. I will show that all four restrictions led to an increase in migrant skill levels, but at a more disaggregated level the nature of the change differed between the early 1921 quota and the later adjustments. Section 2 sums up the economic literature on migrant selection with a special focus on policy. Section 3 presents the historiography of the American 5. I. Ferenczi and W. F. Willcox, International Migrations: Statistics, vol. 1 (National Bureau of Economic Research, Inc, 1929). 6. O. Bandiera et al., “The Making of Modern America: Migratory Flows in the Age of Mass Migration,” Migration and Development, Journal of Development Economics 102 (2013): 23–47. 7. R. Abramitzky et al., “Europe’s Tired, Poor, Huddled Masses: Self-Selection and Eco- nomic Outcomes in the Age of Mass Migration,” American Economic Review 102, no. 5 (2012): 1832–56; R. Abramitzky et al., “Have the poor always been less likely to migrate? Evidence from inheritance practices during the Age of Mass Migration,” Journal of De- velopment Economics 102 (2013): 2–14; R. Abramitzky et al., “A Nation of Immigrants: Assimilation and Economic Outcomes in the Age of Mass Migration,” Journal of Political Economy 122, no. 3 (2014): 467–506. 8. G. Roseman, “An analysis of changes in the characteristics of emigrating Farmers and Laborers from the Hamburg Passenger lists: 1855–1857, 1875–1877, and 1880,” The Social Science Journal 41, no. 3 (2004): 409–421. 3 immigration policy in the interwar era. Section 4 contextualises the passenger lists that constitute my core data source, section 5 describes the specification of the diff-in-diff model I use, and section 6 contains the empirical results. Section 7 sums up and concludes. 2 Literature Review on the Economics of Mi- grant Selection Economictheorydoesnotmakeaclearpredictionabouthowthequotasshould affect migrant flow composition. Standard models can be applied under the assumption that quotas affect selection through a rise in migration costs, but even then the predictions are ambiguous. In the standard Roy-Borjas model9 where cost is independent of skill level, an existing selection would be rein- forced; i.e. if migrants are positively selected from the population of origin they will become even more so after the quota. The model predicts that, if migrants move from a country with relatively low returns to skills, implying relatively lower inequality, to one with relatively high returns and higher in- equality, themigrantswillbepositivelyselectedbecausehigh-skilledhavemore towinfrommigration, assumingasufficientlyhighcorrelationoftheindividual skill level in the host and source country. Conversely, migrants from a country with relatively high inequality compared to the destination will be negatively selected since those on the lower end of the skill distribution have more to gain. For instance, Norway had a more unequal income distribution than the United States in the 19th century and Norwegian migrants were therefore negatively selected from the source population.10 If we now increase the cost of migra- tion by enforcing a non-discriminatory restriction on the number of migrants, 9. A.D.Roy,“SomeThoughtsontheDistributionofEarnings,” OxfordEconomicPapers 3,no.2(1951):135–146; G.J.Borjas,“Self-SelectionandtheEarningsofImmigrants,” The American Economic Review 77, no. 4 (1987): 531–553. 10. Abramitzky et al., “Europe’s Tired, Poor, Huddled Masses: Self-Selection and Eco- nomic Outcomes in the Age of Mass Migration,” 1834. 4 the group that can profit from migration gets smaller as the migration return shrinks, and since this group is found only on one end of the skill distribution, only the least-skilled or the highest-skilled remain in the migrant pool.11 This simple prediction is complicated by a series of confounding factors. First of all, the model assumes that migration costs are identical for each migrant and do not cause a bias in the self-selection. A series of studies has incorporated a more complex understanding of migration costs. Chiswick,12 and Chiquiar and Hanson13 incorporate a skill-dependent migration cost into the model. Empirical research, e.g., on modern Mexico-US migration, has been inconclusive about the model’s predictions.14 Belot and Hatton15 find that the skill premium is only an accurate predictor of selection if poverty constraints are accounted for, while factors such as the variables found in common gravity frameworks often play a more important role than policy or wage incentives. Migration costs in the early 20th century, including the interwar period, have been found to have a negative effect on migrant quantity butnoeffectonaverageskilllevel,whileoutputpercapitainthesourcecountry affected the liquidity constraint for the lower end of the income spectrum, thus 11. Borjas, “Self-Selection and the Earnings of Immigrants”; C. G. Massey, “Immigration quotas and immigrant selection,” Explorations in Economic History 60 (2016): 21–40. 12. B. R. Chiswick, “Are Immigrants Favorably Self-Selected?,” The American Economic Review 89, no. 2 (1999): 181–185. 13. D. Chiquiar and G. H. Hanson, “International Migration, Self-Selection, and the Dis- tribution of Wages: Evidence from Mexico and the United States,” Journal of Political Economy 113, no. 2 (2005): 239–281. 14. ChiquiarandHanson, “InternationalMigration, Self-Selection, andtheDistributionof Wages: Evidence from Mexico and the United States”; P. M. Orrenius and M. Zavodny, “Self-selection among undocumented immigrants from Mexico,” Journal of Development Economics 78, no. 1 (2005): 215–240; D. McKenzie and H. Rapoport, “Network effects and the dynamics of migration and inequality: theory and evidence from Mexico,” Journal of development Economics 84, no. 1 (2007): 1–24; R. Kaestner and O. Malamud, “Self- selection and international migration: New evidence from Mexico,” Review of Economics and Statistics 96,no.1(2014):78–91;J.W.AmbrosiniandG.Peri,“TheDeterminantsand the Selection of Mexico–US Migrants,” The World Economy 35, no. 2 (2012): 111–151. 15. M.V.BelotandT.J.Hatton,“ImmigrantSelectionintheOECD*,” TheScandinavian Journal of Economics 114, no. 4 (2012): 1105–1128. 5 being negatively correlated with migrant selectivity.16 These findings are based on occupation in the source country rather than occupation in the US, which, the authors argue, would be an insufficient measure for a migrant’s skill level at the time of migration. Compared to modern immigration policy, early attempts by the United Statestoregulateimmigrationwere, withtheexceptionoftheliteracytest, not explicitlyskill-selective. Itisthereforenotclearfromtheoutsetwhateffectthe quotas had on selection. If we use a model that incorporates varying migration costs as a function of skill, as in Chiquiar and Hanson,17 and apply it to the case of tightening policy, the selection effect of rising costs will still be positive if the selection was already positive before, but not necessarily negative in an initially negatively selected migrant pool.18 Migration policy, in particular, can make immigration more costly for low-skilled migrants if the destination country only accepts more educated people. Empirical research shows that the Roy model’s predictions are correct in scenarios with free migration, e.g. most19 (though not all20) cases of internal migration in the modern United States or of pre-WWI transatlantic migration.21 Possible reasons for a skill- based cost bias are, apart from discriminatory policy, the ability of the better 16. M. Covarrubias et al., “Who comes and Why? Determinants of Immigrants Skill Level intheEarlyXXthCenturyUS,” Journal of Demographic Economics 81,no.01(2015):115– 155. 17. ChiquiarandHanson, “InternationalMigration, Self-Selection, andtheDistributionof Wages: Evidence from Mexico and the United States.” 18. Massey, “Immigration quotas and immigrant selection.” 19. G. B. Dahl, “Mobility and the return to education: Testing a Roy model with multiple markets,” Econometrica 70, no. 6 (2002): 2367–2420; G. J. Borjas, Issues in the Economics of Immigration (University of Chicago Press, 2008). 20. R.Akee,“WhoLeaves?DecipheringImmigrantSelf-SelectionfromaDevelopingCoun- try,” Economic Development and Cultural Change 58, no. 2 (2010): 323–344. 21. S. A. Wegge, “Occupational self-selection of European emigrants: Evidence from nineteenth-century Hesse-Cassel,” European Review of Economic History 6, no. 3 (2002): 365–394; J. Long and J. Ferrie, “Intergenerational occupational mobility in Great Britain and the United States since 1850,” The American Economic Review 103, no. 4 (2013): 1109–1137; J. Mokyr and C. Ó. Gráda, “Emigration and poverty in prefamine Ireland,” Explorations in Economic History 19, no. 4 (1982): 360–384; R. L. Cohn, “Occupational Evidence on the Causes of Immigration to the United States, 1836-1853,” Explorations in Economic History 32, no. 3 (1995): 383–408; Abramitzky et al., “Europe’s Tired, Poor, Huddled Masses: Self-Selection and Economic Outcomes in the Age of Mass Migration.” 6 educated to quickly adapt to a new environment, and the inability of the poor to finance their trip through credit.22 Research on changing selection over time has, e.g., been conducted by Jasso et al.23 for the post-war US where they find an increasingly positive selection. Most recently, Massey24 looks at the effect of the United States’ national quota system of 1921 and finds a positive effect of restrictive migration policy on selection. Her research provides the methodological basis for this arti- cle. Massey exploits the fact that immigrants from the Western hemisphere were not affected by the American quota system. She applies a difference- in-difference regression model where European immigrants are the treatment group that are affected by the quota, and Canadian immigrants are the control group. She then examines the introduction of the 1921 quota and its effect on selection in the two groups. Since she uses a dummy variable approach, every European emigrant is assumed to experience the same treatment affect of being subject to a quota.25 3 American Immigration Policy The freedom of entry to the United States for Europeans during the “Age of Mass Migration” came to an end in 1917.26 This was when the US Congress 22. R.AbramitzkyandL.P.Boustan,“ImmigrationinAmericanEconomicHistory,” Work- ing Paper, Working Paper Series 21882 (National Bureau of Economic Research, January 2016). 23. G. Jasso et al., “The Changing Skill of New Immigrants to the United States: Recent Trends and Their Determinants,” in Issues in the Economics of Immigration, ed. G. Borjas (University of Chicago Press, January 2000), 185–226. 24. Massey, “Immigration quotas and immigrant selection.” 25. Massey shows her results to be robust when she specifies the quota restriction in a continuous way.ibid. 26. For other ethnicities, notably the Chinese, it ended much earlier, first with the 1875 PageActtargetedatAsianprostitutesandforcedlabourers,thenwiththeChineseExclusion Actof1882,whichprohibitedtheentryofChineselabourersingeneralanddeniedanyperson of Chinese heritage US citizenship.E. Lee, “A Nation of Immigrants and a Gatekeeping Nation:AmericanImmigrationLawandPolicy,” inACompaniontoAmericanImmigration, ed. R. Ueda (Wiley-Blackwell, 2006), 10 A series of more general immigration restrictions in the late 19th century then attempted to target undesirable individuals such as criminals, the mentally ill, or beggars, but no systematic attempt at overall migration restriction was 7 passed an Immigration Act that required a literacy test from every potential immigrant and excluded the inhabitants of a newly defined “Asiatic Barred Zone” stretching from the Arab peninsula to New Guinea.27 In addition to illiterates,theactallowedthedenialofentrance“ongroundsofphysical,mental or moral defects, or for political and economic reasons”, with the last category including “contract labourers, unaccompanied children under sixteen, and (...) those ‘likely to become a public charge’.”28 The literacy requirement was the first major policy that left a notable mark on European migration. A second restrictive measure followed with the Quota Act of 1921. It first established a system of national quotas that would be the core of American immigration law for more than four decades. It limited annual overall migration to 355,000 and, more importantly, restricted the number of migrants for each nationality to three per cent of the number of foreign-borns from that respective country in the US at the time of the 1910 census. Since the existing migration stock was disproportionally Northern and Western European, as these made up the bulk of migrants in earlier decades, the policy effectively discriminated against the South and East and shifted the composition of migrants back towards its 19th century pattern.29 Germans, as the biggest ethnic group in the United States, were scarcely affected, while the inflow of Poles, Czechs and Russians that often left through German ports, plummeted.30 made.Lee,“ANationofImmigrantsandaGatekeepingNation: AmericanImmigrationLaw and Policy,” 11 27. Ibid., 12. 28. B.-A.Zucker,“AmericanRefugeePolicyinthe1930s,” inRefugees from Nazi Germany and the liberal European states, ed. B. Caestecker Frank & Moore (2010), 153. 29. Lee, “A Nation of Immigrants and a Gatekeeping Nation: American Immigration Law and Policy,” 12. 30. D.Hoerder,“ThetrafficofemigrationviaBremen/Bremerhaven:Merchants’interests, protective legislation, and migrants’ experiences,” Journal of American Ethnic History 13, no. 1 (1993): 68–101. 8 In 1924, the legislation became even more restrictive: The quotas were loweredfromthreetotwopercentandtheoverallnumberreducedto165,000.31 Furthermore, the reference census was moved back from 1910 to 1900, leading to a further shift in favour of Northwestern Europeans.32 Hatton and Williamson33 describe a shift from “new”, predominantly East- ern European source countries to the “old”, Western and Northern European sources, as intended by the quotas, which induced a positive selection effect as these old countries were more developed. From 1911-16 to 1936-40 the share of “skilled, professional, or commercial occupations” among immigrants rose from 17.9 per cent to 30.6 per cent. A rise in the share of female migrants is attributed by Hatton and Williamson34 to the rise in family reunification. This meant that the high labor force participation and low dependency rates of previous decades could not be upheld in the interwar years. Meanwhile, po- tential overseas migrants from Eastern Europe who were deterred by the quota laws, instead often chose second-best destinations within Europe and ended up in, e.g., France. Hatton and Williamson35 argue that transatlantic migration would have dried up in the interwar years without the quotas. As migration follows a life cycle pattern, driven by industrial catch-up, demographic tran- sitions and network effects, the peaking Eastern European exodus would have probably ended sooner or later, just as the Northwestern one did. This, how- 31. Lee, “A Nation of Immigrants and a Gatekeeping Nation: American Immigration Law and Policy,” 12. 32. Immigrants who were exempt from the quota included (a) close relatives of American citizens, their wife/ husband and their unmar- ried children under twenty-one years of age; (b) returning resident aliens; (c) natives of Western-hemisphere countries, their wives and unmarried children; (d) ministers and professors, their wives and unmarried children under eigh- teen years of age; (e) bona fide students; and (f) American women who had lost their United States citizenship by marrying aliens. Zucker, “American Refugee Policy in the 1930s,” 157 33. T.HattonandJ.Williamson,GlobalMigrationandtheWorldEconomy:Twocenturies of policy and performance (MIT Press, 2005). 34. Ibid., 259. 35. Ibid., 199. 9