The Impact of Immigration on Wages and Labor Markets in the US (1890-1921), Study Guides, Projects, Research of Political Economy

Download The Impact of Immigration on Wages and Labor Markets in the US (1890-1921) and more Study Guides, Projects, Research Political Economy in PDF only on Docsity! This PDF is a selection from an out-of-print volume from the National Bureau of Economic Research Volume Title: The Regulated Economy: A Historical Approach to Political Economy Volume Author/Editor: Claudia Goldin and Gary D. Libecap, editors Volume Publisher: University of Chicago Press Volume ISBN: 0-226-30110-9 Volume URL: http://www.nber.org/books/gold94-1 Conference Date: May 20-21, 1993 Publication Date: January 1994 Chapter Title: The Political Economy of Immigration Restriction in the United States, 1890 to 1921 Chapter Author: Claudia Goldin Chapter URL: http://www.nber.org/chapters/c6577 Chapter pages in book: (p. 223 - 258) 7 The Political Economy of Immigration Restriction in the United States, 1890 to 1921 Claudia Goldin It does not matter in the least what the favored classes of the coun- try think about immigration; the doors of this land will never be closed except upon the initiative and the imperative of the laboring classes, looking to their own interests, and to the heritage of their children. Francis A. Walker, Discussions in Economics and Statistics 7.1 Introduction With the passage of the Emergency Quota Act in May 1921 the era of open immigration to the United States came to an abrupt end.' The American policy of virtually unrestricted European immigration was transformed, almost over- night, to a quota system that would last, virtually unchanged, until 1965. The ultimate switch in policy is not hard to explain. The perplexing part of the legislative history of immigration restriction is its timing. More astonishing than the closing of the door in 1921 is that it remained open despite twenty- five years of assault during which 17 million immigrants from among the poor- est nations in Europe found refuge in America. This paper details the remark- able set of events that propped the door open and the forces that eventually slammed it shut. Because the story of immigration restriction is a legislative one, its main players will be representatives, senators, and presidents. But behind the legisla- tive tale are the shifting interests of various groups. The first is organized labor, Claudia Goldin is professor of economics at Harvard University, director of the Development of the American Economy Program of the National Bureau of Economic Research, and a research associate of the National Bureau of Economic Research. The author thanks Lisa Kao, Boris Simkovich, and Marian Valliant for providing superb re- search assistance. Helpful comments were provided by Stanley Engerman, Zadia Feliciano, Law- rence Katz, Robert Margo, Jeffrey Williamson, the members of the Harvard Economic History Workshop, and participants at the NBER-DAE conference on the Political Economy of Regula- tion, particularly the discussant, Joseph Ferrie. Shawn Kantor supplied the wage data by city for the union sample, 1907 to 1923, and Howard Rosenthal provided the congressional districts for the 1915 vote. The author thanks them both. This research has been funded by National Science Foundation grant SES-9122782. I . As 1 will argue later, the abrupt end should more accurately date with the final passage of the literacy test in 1917. since it was a simple step to move from the test to a quota. 223 226 Claudia Goldin posed, only two could have significantly restricted immigration-the financial and literacy tests. Only the literacy test received serious deliberation.h By the time the literacy test finally passed, it was not as restrictive a measure as when it was first proposed because literacy rose rapidly in Europe. Thus the quotas of 1921, 1924, and 1929 quickly followed. The forces that prompted these more restrictive measures were the same as those that led to the passage of the literacy test. Thus most of this paper is concerned with the passage of the literacy test, since the quotas were its logical extension. The literacy test was not merely given careful consideration in Congress from 1897 to 1917. It passed the House on five separate occasions and passed the Senate on four. Further, the House overrode presidential vetoes of the bill twice and on two occasions failed to override by fewer than seven votes. The Senate overrode a presidential veto once, when the test became law in 1917. The literacy test was to be administered to physically capable adults to as- sess their ability to read. The test was well-defined, although it varied some- what across proposed immigration legislation. It generally consisted of reading several sentences of the Constitution in any language chosen by the potential immigrant, including recognized dialects. Some of the proposed legislation also required that immigrants be capable of writing the sentences they could read. Close relatives of an adult male immigrant who was literate were often exempted. Because the shipping companies that brought immigrants across the ocean were responsible for the return voyage of any who did not meet U.S. immigration standards, it is likely that these companies would have adminis- tered a literacy test of their own, in the same way that they screened for health violations in European p 0 m 7 immigration for at least two years (Higham [ 19551 1981). During the debates over the quota Icgis- lation in the aftermath of World War I, several bills were introduced that would have suspended immigration for periods of from three to five years (Hutchinson 198 I , 17 I ) . Of the many possible means of restricting and regulating immigration contained in the Reports of the Immigrarion Com- mission of 1910, none was a blanket quota of the type eventually adopted in I92 I, 1924, and 1929. One suggested means would have limited “the number of each race arriving each year to a certain percentage of the average of that race arriving during a given period of years” (Senate 191 I a, 747). 6. Section 39 of the immigration bill introduced in 1906 contained a financial test that would have required, among other things, that all male immigrants over sixteen years old (or the male head of the household) have $25 or its equivalent (Hutchinson 1981, 139). The final version of the 1907 act did not contain the provision. An amount of $25 was 2.4 weeks of income for lower- skilled manufacturing labor in America in 1906 and about 9 weeks of income for an equivalent worker in southern and eastern Europe at the time (U.S. Bureau of the Census 1975, aeries D 778; Simkovich, Taylor, and Williamson 1992). 7. The literacy test was put in place in 1917 and remained after the quotas were passed. The experience with the literacy test immediately following its passage, and prior to the quotas, can be seen in U.S. Department of Labor 1918,23. The 1917 act allowed for a fine of $200 per alien to be assessed against any transportation company bringing an alien excludable by the literacy test. The fine and the passage home may have been sufficiently steep to give shipping companies an incentive to screen aliens prior to passage, although I do not know whether or how they accom- plished that task. In 1917 fines were levied for only 192 excludable illiterate aliens out of a total of almost 300,000 aliens. 227 The Political Economy of Immigration Restriction in the U.S., 1890 to 1921 The literacy test first came to a vote in Congress in 1897 and was over- whelmingly passed by the House and cleared a majority in the Senate (see the chronology in table 7.1). At least one other bill was proposed during the debate in the House that could have been even more restrictive and that would have restricted immigration from any port in Europe not having a consular inspec- tion station. Several factors operated in the mid- 1890s to create a short-lived coalition, yet one that would resurface in another form, around regulating and restricting Table 7.1 Immigration Restriction Chronology: Votes on the Literacy Test Date Branch of Government Vote Notes 2/9/97 2/17/97 3/2/97 3/3/97 3/3/97 111 7/98 1 21 14/98 512712 6/25/6 411 911 2 1211 811 2 21 1411 3 211 911 3 1/2/15 1/15/15 2/41 1 5 3/30/16 1211 411 6 2/1/17 2/5/17 House Senate President Cleveland House Senate Senate House House House Senate House President Taft House Senate House President Wilson House House Senate President Wilson House Senate 217-36- 1 O P 34-3 1-25 Veto 195-37- 123 45-28-16 101 -1 04-150 No vote found 128-116 9-56-30 179-52 Veto 213-114-54 227-94- I03 Veto 307-87-39 Veto 50-7-39 261-136-26 64-7-25 287-106-40 62- 19-5 Affirmative vote on bill Affirmative vote on bill Ovenides Presidential veto Takes no action, bill dies Affirmative vote on bill Negative vote on consideration of bill Affirmative vote on bill, literacy test dropped in House-Senate conference Vote to remove literacy test from immigration bill and to set up Immigration Commission Vote to strike the literacy test from the bill; affirmative vote on bill, sent to conference conference Affirmative vote on bill, sent to Fails to override Affirmative vote on bill Affirmative vote on conference report of bill Fails to override Affirmative vote on bill Affirmative vote on bill Overrides veto Overrides veto Sources: Hutchinson 1981; Congressional Record, 62d, 63d, and 64th Cong. Note: Roll call votes count those not voting, whereas non-roll call votes have only pro and con. aHutchinson reports those not voting as 125, not 102. 228 Claudia Goldin immigration. The leadership and members of the AFL and the Knights of La- bor came out strongly in favor of the literacy test in 1897, but had not done so before. The depression of the 1890s, with its extremely high rates of unem- ployment, particularly in the manufacturing sector, appears responsible for the change of heart.* But capital, too, turned against immigration. Industry had depended on immigrant labor. Thus the restrictionist sentiment of certain associations of capitalists may seem inexplicable. The labor unrest of the 1880s and early 1890s, fresh in the minds of many, may have been a deciding factor. In addition to a rash of strikes there were particularly odious events, such as the Homestead Strike of 1892 and the Haymarket Riot of 1886. The business faction that united against immigration in the last two decades of the nineteenth century is not easily categorized, but it disintegrated rapidly once economic conditions improved, labor unrest subsided, and wage de- creases from immigration were more apparent (Heald 1953; Wiebe 1962). The face of immigration changed rapidly in the 1890s, moving from north- ern and western Europe to southern, central, and eastern Europe. Whereas the new immigrants were 35 percent of the total flow in 1890, they were 56 percent in 1896, although the flow was of comparatively modest size in the mid-1 890s, a product of economic depression (see figures 7.1 and 7.2).9 Some have claimed that the new immigrants were too recent and too few to motivate the wave of anti-immigrant sentiment in the 1890s (Higham 1955). A reading of the Congressional Record affords ample reason to disagree with this claim, but not with a related assertion that the new immigrants were too recent and too few to influence policy.“’ But they would be fortified by numbers and unified by fear very soon. President Cleveland vetoed the immigration legislation in 1897 because it contained the literacy test, and although the House voted to override his veto, the Senate took no action and the bill died. Just one year later, in 1898, a similar immigration law was proposed in Congress. In this case the bill cleared the Senate but failed by three votes to pass the House, which had just a year before given it overwhelming support.lI The flip-flopping that took place on this im- 8. The AFL letter to Congres, in I898 argued that “laborers are imported from other countries to reduce our wagez and thereby our standard of living” (Congressionnl Record 1898, 3 1686). The AFL, like others, was arguing against contract labor and shipping and railroad companies’ enticing people to emigrate to the United States. 9. New immigrants are those from southern, central, and eastern Europe. The countries (at various points in time) in the eastern, central. and southern European group include Bulgaria, Croatia, Czechoslovakia. Greece, Hungary, Italy, Montenegro, Poland, Portugal, Rumania, Russia, Serbia, Spain, Turkey (in Europe), Yugoslavia, and the Baltic republics. I have included non- German-speaking emigrants from Austria in eastern Europe. 10. According to Higham ( [ 19551 1981) the Immigration Protection League, organized primar- ily by the older immigrant groups in the late 189Os, led the defeat of the I898 literacy requirement in the House. 1 I . Of the 45 yeas in the Senate in 1898, 23 voted altirmatively in 1897, 6 had voted negatively. 9 had been recorded as absent. and 7 were new members of the Senate. Had all those present in both I897 and I898 voted as they did in 1898, the vote would have been 37 for and 22 againht in 231 The Political Economy of Immigration Restriction in the U.S., 1890 to 1921 Both the Senate and House passed the literacy test again in 1912, only to have it vetoed by lame duck President Taft. The House failed by just six votes to override the veto. Had it been able to override, the test would have become law, since the Senate vote was 86 percent in favor of the amendment on the literacy test. The literacy test was reintroduced in 1915, passing the Senate by a wide majority and the House by enough to override a veto but with a large segment not voting. President Wilson, an ardent Progressive remembering his promise to immigrants in the 1912 election, vetoed the legislation, and the House failed to override it, this time by just five votes.’7 In 1916 the House and Senate once again passed a bill containing the literacy test, and finally, in 191 7, both houses successfully overrode Wilson’s second veto. The literacy test had become law. The votes on the literacy test are evidence of the shifting coalitions men- tioned earlier. The first vote in the House, in 1897, brought southern and urban- northeastern interests together in opposition to the test, with virtually the rest of the country favoring it.’4 The overwhelmingly anti-immigrant vote in 1897 may have been a hysterical reaction to the prolonged economic downturn of the 189Os, although recovery was well under way by the date of the vote. A more sober view of the immigration issue may have been given to the vote in 1898, a very close one in the House. 7.2.2 Analysis of Votes on the Literacy Test by State As can be seen in table 7.2, New England, much of the Middle Atlantic, and about half of the midwestern region were in favor of the test in 1898. The South was generally against it, as it had been in the previous vote. The Mountain and Pacific states were not yet numerous enough to categorize. The next roll call vote on the literacy test was not for another fifteen years, in 1913. By that date the shifting coalitions mentioned earlier had become apparent. The Northeast was split, with the larger cities voting pro-immigrant and the rural areas voting anti. The Midwest was also split. Differences there may have been rooted in percent and 70 percent were, but that states often had residence requirements that the mobile foreign born often could not meet. The evidence presented here supports, in principle, the asser- tions of the older literature. The foreign born might have been an even more potent force had naturalization been faster and had various states had more lenient residency requirements. 13. Wilson’s veto of the 1915 act can be rationalized, after the fact, by his promise to the foreign born during the election, but it is not clear that it could have been predicted prior to the vote in the House. Only after the House passed the act did Wilson warn the Senate that he would veto the bill if the literacy test was not removed (Link 1954. 60-61). But there is no indication that Wilson explicitly stated that he would veto the bill prior to its passage in the House, although Link states that Wilson “intimated that he would.” In fact, the official magazine of the International Brother- hood of Teamsters predicted in August 1913, six months after Taft’s veto of the literacy act, that “any immigration law passed, carrying a literacy test in all probability, will be approved by Presi- dent Wilson” (International Brotherhood of Teamsters, August 191 3, 5). 14. Of the thirty-seven negative votes, twenty-five were cast by southerners. Three from New York City joined them together with eight others from urban areas in the Northeast. One additional representative, from Wisconsin, voted against the test (Congressional Record 1897, 29:2947). 232 Claudia Goldin Table 7.2 Proportion of House Voting for the Literacy Test or to Override a Presidential Veto of the Literacy Test, by State, 1898, 1913, 1915, 1917 Proportion to Override Number Voting For Test - 1898 1913 1915 1917 1898 1913 1915 1917 New England CT 1 .0 MA 0.82 ME 1 .o NH 1 .0 RI 1 .0 VT 1 .0 Middle Atlantic NJ I .0 NY 0.52 PA 0.85 East North Central IL 0.47 IN 0.69 MI 0.50 OH 0.65 WI 0.43 West North Central IA 0.33 KS 0.0 MN 0.33 MO 0.13 ND 1 .0 NE 0.40 SD 0.50 DE 0.0 FL 0.0 GA 0.0 MD 1.0 NC 0.60 sc 0.57 VA 0.44 WV 0.67 East South Central AL 0.43 KY 0.22 MS 0.0 TN 0.22 West South Central AR 0.80 LA 0.0 OK - TX 0.17 South Atlantic 0.40 0.33 0.50 I .0 0.0 1 .0 0.0 0.28 0.65 0.39 0.80 0.44 0.74 0.40 0.27 0.86 0.44 0.79 1 .o 0.50 0.50 1 .0 I .O I .0 0.83 I .0 I .o 1 .o 1 .0 1 .o 0.91 0.89 1 .0 1 .0 0.17 I .0 0.80 0.0 0.25 0.67 0.0 0.0 1 .0 0.45 0.19 0.64 0.54 0.67 0.69 0.67 0.40 0.60 0.88 0.57 0.67 0.67 0.50 1 .0 I .0 1 .0 0.83 0.67 1 .0 0.86 1 .0 1 .o 0.8 0.91 1 .0 1 .0 1 .0 0.43 1 .0 0.78 0.20 0.35 0.80 1 .0 0.0 1 .0 0.67 0.26 0.75 0.63 0.50 0.50 0.80 0.55 0.80 0.86 1 .0 0.8 I 0.67 0.80 1 .0 1 .0 1 .0 I .o 0.83 1 0 0.86 I .o I .o I .0 0.91 I .0 I .0 1 .0 0.63 1 .0 0.84 4 I 1 2 2 1 1 7 23 27 15 13 10 17 7 9 3 3 15 I 5 2 1 2 10 4 5 7 9 3 7 9 6 9 5 6 12 - 5 12 4 2 I 2 6 29 23 23 10 9 19 10 I I 7 9 14 I 6 2 1 1 10 6 10 6 9 2 9 I I 9 10 7 6 4 15 5 16 3 2 I 2 1 1 37 36 26 12 13 18 10 10 8 7 15 3 6 3 1 4 12 6 9 7 10 6 10 I I 8 10 8 7 7 18 5 17 5 2 3 2 12 42 36 27 12 14 20 11 10 7 1 1 16 3 5 3 I 4 12 6 10 7 10 6 10 I 1 8 9 7 8 6 19 233 The Political Economy of Immigration Restriction in the US. , 1890 to 1921 Table 7.2 (continued) Proportion to Override Number Voting For Test 1898 1913 1915 1917 1898 1913 1915 1917 Mountain Az CO ID MT NM NV UT WY Pacific CA OR WA Total 0.75 1 .0 0.0 I .0 1 .0 I .0 0.67 0.50 1.0 1 .o 1 .0 0.50 - 1.0 1 .0 0.0 0.0 1 .o 1 .0 1 .0 1 .0 0.0 0.50 0.0 I .o I .0 1 .0 0.71 0.91 0.90 1 .0 I .0 1 .0 I .0 1 .0 1 .0 4 1 I 28 I 1 3 1 2 1 1 1 - I I I 10 I 3 3 3 5 4 342 409 427 Notes: A vote to override was a vote against open immigration. “Paired’ votes (these were two to one for the override) are included with either the yeas or nays. Those not voting (and not paired) or absent are not included in the denominator. ‘Not yet a state. hNo votes were cast by representatives of this state. Source: Congressional Record, various years. the nativity of constituencies, as they were in the cities.I5 The South was firmly against open immigration, as were the Pacific region and most of the Mountain states. The 1915 and 1917 votes are similar to that in 1913 with an erosion of support in much of the Midwest and an increase in support in some large cities. A large segment of rural America was against open immigration at least by the first vote in 1897 and even in the first strongly contested vote in 1898. Why this was the case probably has more to do with the history of nativist sentiment in America than with the particulars of immigration restriction of concern here. It is important, however, that some parts of rural America were considerably less in favor of restriction than were others. Rural Minnesota, Wisconsin, Iowa, and Michigan can be easily contrasted with equally rural areas in Ohio, Indi- ana, and Kansas (see table 7.2). In general, those from countries whose popula- tions were still emigrating at high rates voted to keep the door open, while the native born and those from countries that were not active sending regions did 15. Ongoing research of mine on this issue indicates that those of German and British ancestry opposed open immigration, whereas those of Scandinavian and “new” ancestry supported it. I am also exploring the role of concentration. Areas with many foreign born of one nativity may have been pro-immigration. But areas with many foreign born of several nativities may have been less willing to keep the door open. 236 Claudia Goldin had blacks felt safe to leave, manufacturing interests in the North did entice blacks to emigrate during World War I and throughout the 1920s. Southern manufacturing interests may have recognized that their sole advan- tage was a low-wage, nonunion workforce, and that immigrants were providing the North with a similar workforce. If immigrants would not come South, the South would deprive the North of them. Yet another potential explanation is that the North was gaining power in Congress and that much of its population increase was in the form of the foreign born and their children.21 Although I cannot differentiate among these various hypotheses, each could have been reinforcing. By the early 1900s the South saw nothing to lose and much to gain from closing the door. 7.2.3 The Eventual Triumph of the Anti-immigration Forces The three votes on the literacy test by three successive seatings of the House enable one to see how the changed composition of the electorate altered the outcome and precisely which forces held the anti-immigrant forces at bay (see table 7.4).22 Comparing first those representatives who voted in both the 62d ( 19 12/13) and 63d (1 9 14/15) Congresses, 74 percent voted for the literacy test. Thus the incumbent members of the House were overwhelmingly in favor of restriction in 1915. The recently seated members of the House did amass a majority in favor of restriction, but they did so just barely. Only 54 percent voted for the test in 1915, clearly not enough to override a presidential veto. Thus it was the newly elected representatives who held the literacy test at bay, suggesting that big-city districts had changed composition. The new immi- grants themselves, it seems, managed to elect representatives who voted dis- proportionately against the literacy test. But if this were the only change in the House, the vote would have become less in favor of the act over time. Rather, the percentage voting in favor remained at 65 percent. Those who were voted out of office were in favor of the keeping the door open to the same degree as those who took their place. Thus the vote in 1913 would have cleared the two- thirds needed to override, had only those who kept their seats to 1915 voted. Those who were defeated in 1914 voted far more decidedly against restriction, although with a majority in favor of the literacy test. Those who remained seated from the 62d to the 64th Congresses voted dis- proportionately prorestriction in the 1913 and 1915 votes. Those newly elected and those who suffered defeat at the polls in 1914 were less restrictionist. The 21. The South had opposed cheap land, a half century before, on similar grounds. Cheap land meant more immigrants, and more immigrants meant greater political power for the North. There were additional reasons for southern reluctance to give land away. Cheap land also meant higher tariffs, and the South opposed both high tariffs and increased political power for the North. On the South’s opposition to free land, see Robbins [1942] 1976. 22. I am looking only at the voting record of the House because the Senate passed the test by wide enough margins in 1912/13, 1914/15, and 1916/17 to override a presidential veto. The Senate would be expected to be more supportive of restrictive immigration than the House, in which certain representatives were elected in districts populated by the new immigrants. 237 The Political Economy of Immigration Restriction in the U.S., 1890 to 1921 Table 7.4 Votes to Override the Presidential Veto on the Literacy Test in the 62d, 63d, and 64th Congresses, 1913,1915, and 1917 Number For Number Against Number Not Voting % For“ Vote in 62d Congress (1913) Those remaining in office to 63d Those defeated in 19 I4 Vote in 63d Congress (I 9 15) Incumbents only (62d and 63d) Nonincumbents only (63d not 62d) Those defeated in 1916 Vote in 64th Congress (1917) Incumbents only (63d and 64th) Nonincumbents only (64th not 63d) 213 I60 53 262 178 84 73 287 187 I00 I14 71 43 136 64 72 55 106 70 36 54 65.1 69.3 55.2 26 65.8 74.2 53.8 57.0 40 73.0 72.8 73.5 Sources: Congressional Record, various years; Congressional Directory, various years dTwo-thirds is necessary to override a presidential veto. new members hailed primarily from the large and industrial cities of the North- east and Midwest, whereas those defeated in 1914 came from small to middle- sized towns across America. Those suffering defeat, therefore, were replaced by representatives far less in favor of open immigration. But the newly elected group was able to make up the difference and prop the door open. America had become more bifurcated along the lines of open immigration, and it was redistricting in 19 14 that resuscitated the pro-immigration bloc.*’ Without it, the anti forces would have won. The increased population of the nation’s big and industrial cities, with its largely immigrant composition, was responsible for keeping the anti-immigrant forces just below the two-thirds majority needed to override. All that changed by 1917, however, when there was no relationship between incumbency and the vote on the literacy test. All in the House-save those whose districts were in the nation’s largest cities and a handful of others-voted overwhelmingly for it, regardless of time in office and party affiliation. 7.2.4 Restrictiveness of the Literacy Test The literacy test was an overture to the Emergency Quota Act passed in 1921, the Immigration Act of 1924, and, eventually, the National Origins Act 23. The possibility that it was redistricting is by inference only. There were forty-five more representative!, seated in the 63d Congress than in the 62d Congress, and there were forty-three more representatives present for the vote in the 63d than in the 62d Congress to override the president’s veto (see table 7.1 ). Much of the redistricting took place within states, it appears. A tabulation of representatives by state does not reveal much difference between the two Congresses. But New York City, for example, gained seven representatives. Among those who were not seated in the 62d Congress hut who voted in the 63d, there were nine from New York City who voted against the test. Two representatives from New York City were not reelected, one of whom was against and one of whom was for the test. Three of the newly elected representatives were from Philadelphia, which lost only one seat from the 62d to the 63d Congresses. Chicago, however, made no net gain. 238 Claudia Goldin passed in 1929. Although the quotas were plausibly more potent than the liter- acy test, the test could have imposed considerable constraints, particularly on the newer immigrant groups. How much of a constraint depended on the type of test, the sending country flows, and the period considered. As initially conceived in 1897, the literacy test involved reading and writing a short passage of the US. Constitution and barred illiterate adult males and their accompanying family members. At that time it was believed that the test would have checked the entry of 25 percent of all recent arrivals, although more than 40 percent of the newer groups would have been barred.24 More precise estimates were compiled for the Reports of the Immigration Commis- sion. According to the report, data collected by the U S . commissioner general of immigration from the self-reported statements of immigrants upon arrival indicated that 33.4 percent of eastern European and 44.9 percent of southern European immigrants (fourteen years and older) arriving from 1899 to 1910 were illiterate.25 Thus the test would have reduced the new immigrants by 37.4 percent in 1907 at the height of immigration. The constraint would have been less in the 1920s due to the rising literacy in eastern and southern Europe, although the test could have been made more difficukzh For the entire 1905 to 1914 period, a decade of immigrant flows of more than one million per year, the literacy test would have restricted immigration from southern and eastern Europe to about 445,000 annually when the flow was, in actuality, 7 1 2,000.27 But the eventual quotas were far more restrictive. The 1921 act limited southern and eastern Europeans to 156,000, and the 1924 24. During the debate on the immigration act of 1898, Senator Fairbanks of Indiana inserted data in the Congressiunal Record showing that about 25 percent of immigrants (fourteen years old and over) arriving from I895 to I897 were illiterate. Illiteracy was declared by the immigrant, and no official test was given (Cungressiorzal Record 1898, 3 1 :5 15). 25. Female immigrants were less literate than male immigrants. Because many versions of the literacy test allowed the illiterate family members of a literate adult male immigrant to emigrate, the constraint would have been less than calculated on the basis of the aggregate data. But younger adults were more literate than older adult immigrants, and since the Immigration Commission data group all ages, this factor would tend to bias the calculation in the other direction. The data from the U.S. commissioner general of immigration in the Reports of rhe Immigration Commission ( 1 9 1 I a, I :99) differ, often radically, by country from those reported in the Congressional Record ( 1898, 3 1 :5 16) for a somewhat earlier period of time. But the data in the report are consistent with estimates I have computed using the 1910 Public Use Micro-data Sample (PUMS). 26. Primary-school enrollment had been rising secularly in Italy, Spain, Yugoslavia, and Ruma- nia across the latter half of the nineteenth century and exploded in Russia after the revolution. See, for example, the data in Easterlin 198 1. 27. Emigration to the United States from Europe could have slowed in the 1920s as conditions improved in certain European countries relative to those in the United States. Wage data collected for a project on international economic convergence (Simkovich, Taylor, and Williamson 1992) indicate that Italy, the only new immigrant country in the data set, improved its real wage position relative to the United States during the 1900s to 1920s period. In 1910, for example, the ratio of Italian to American real wages for unskilled laborers was 0.29, but by 1925 it was 0.48. It should also be noted that even though gross immigration was 6.71 million from 1908 to 19 14, many immigrants returned home. The net immigration figure is 61 percent of the gross, or 4.07 million (Willcox 1931, 88). 241 The Political Economy of Immigration Restriction in the U S . , 1890 to 1921 Table 7.5 Percentage of “New” and Recent Immigrant Males in the Labor Force, by Industry and Selected Occupations in the Goods-Producing Sector, 1910 ( 3 ) (5) (1) ( 2 ) “New” (4) Relative %, Born Immigrant Immigrant4 Foreign Born” and Recent” Foreign “New” and Recent Relative %, “New” (%) (%) ([1]/32.9) ([3]/12.2) All employed males (2 14 years) Excluding those in agriculture In goods-producing sector Mining Building trades Laborers Painters Brick and stone masons Manufacturing Chemicals Clay, glass, and stone Clothing Food Bakeries Iron and steel Foundries Foundry laborers Machinists” Leather Liquor and beverages Lumber and furniture Metals (except iron and steel) Paper and pulp Printing and publishing Textiles Tobacco and cigars 21.0 25.9 32.9 42.3 27.2 29.9 22.6 33. I 31.9 30.6 30.8 67. I 40.7 53.7 36.9 34.9 54.5 25.7 35.8 41.3 22.6 33.9 31.0 20.0 31.0 36.2 8.7 11.7 16.5 29.3 10.6 15.6 8.2 10.9 15.1 14.8 18.4 53.6 16.0 21.0 20.5 18.2 37.2 5.6 18.7 8.9 8.3 17.1 13.1 4.1 12.3 15.5 5.8 8. I 12.2 21.7 7.4 12.1 5.2 7.1 11.2 12.4 14.2 32.3 12.2 15.4 16.9 15.2 32.2 3.6 15.2 5.6 6.1 13.3 11.2 2.4 9.1 8.3 1 .OO 1.29 0.83 0.91 0.69 I .OO 0.97 0.93 0.94 2.04 I .24 I .63 1.12 1.06 1.66 0.78 I .09 1.26 0.69 1.03 0.94 0.61 0.94 1.10 1 .00 1.78 0.6 1 0.99 0.43 0.58 0.92 I .02 1.16 2.65 1 .OO I .26 1.39 I .25 2.64 0.30 1.25 0.46 0.50 1.09 0.92 0.20 0.75 0.68 Source: 1910 PUMS, males fourteen years and older. Note: “New” and recent immigrants are eastern, central, and southern Europeans who emigrated during the ten years preceding the 1910 census. “he relative percentage divided by the percentage of all employed males (fourteen years and older) in the goods-producing sector for each of the two immigrant groups. hNot necessarily working in foundries or in the iron and steel industry. for example, 32 percent of the laborers were of the new and recent group of immigrants but only 4 percent of the machinists were. Immigrants went disproportionately to the nation’s largest cities, but so did all Americans during the period under study. Despite the notion that immi- grants, particularly from 1900 to 1914, crowded themselves into a handful of America’s urban centers, they were in fact extremely dispersed across all cities 242 Claudia Goldin regardless of size.31 Indeed, the change in the foreign-born population from 1900 to 1910 was, on average, the same across almost all deciles of the size distribution of cities in 1900. The fifteen cities with the largest and smallest increases in the proportion of foreign born in their populations are given in part A of table 7.6 for 1890 to 1900 and 1900 to 1910. No city in the top decile (decile = 10) is included in the fifteen having the largest increases from 1890 to 1900, and there are many small cities represented among the ranks of those accumulating the foreign born at a faster rate than they accumulated native- born residents. And while there is some repetition in the top and bottom lists across the decades, there is also a lot of movement. Immigrants went to differ- ent cities in different decades. They went where the jobs were, and, as will be demonstrated in table 7.7, they went where their earning power would be highest. Also of importance in assessing the political economy of immigration re- striction is whether immigrants went to areas already populated by immigrants. To the extent that “immigration begot immigration,” certain cities and congres- sional districts within them would have become even more disproportionately immigrant in makeup and thus more inclined to oppose immigration restric- tion. Part B of table 7.6 reports the results of the regression of the difference in the percentage foreign born across a decade on the percentage foreign born in the earlier year. That is, A[% Foreign Born,, r + , O ) ] is run on [% Foreign Born,]. Interestingly, the coefficient is negative for the 1890 to 1900 and 1910 to 1920 decades, but positive for the 1900 to 1910 decade.’? Immigration was reinforcing or concentrating in its impact from 1900 to 1910. Thus immigra- tion restriction was held at bay during the largest immigrant flows, in part be- cause the new immigrants were able to capture various congressional districts. By the 1910 to 1920 decade, however, the flows had a more diluting impact. Also note that only during the decade of the greatest immigration, from 1900 to 1910, did immigrants flow into America’s cities at the same rate that native- born Americans populated the same urban areas. The percentage foreign born actually fell during the 1890 to 1900 and 1910 to 1920 decades in the cities under study. Similar notions are apparent in part C of table 7.6. During the 1890 to 1900 and 1910 to 1920 decades, the percentage foreign born in the urban population declined where population grew, but the reverse occurred from 1900 to 1910. Only in the 1900 to 1910 decade did the fastest-growing cities also increase their population share of the foreign born. These bur- geoning urban areas gained representatives who held the prorestriction move- ment at bay, at least for a while. 31. The one exception-and it is an important one-is New York City. There are 142 cities in the 1890 to 1900 sample and 127 in the 1900 to 1910 sample. (These are the cities of the Bureau of Labor Statistics wages and hours studies for the various time periods.) The earlier sample in- cludes more small cities, although the deciles in table 7.6 are recomputed for each decade. 32. The same cities have been used for the 1890-1900 and 1900-I910 regressions. There are twelve fewer cities for the 1910-1920 regression. 243 The Political Economy of Immigration Restriction in the U.S., 1890 to 1921 Table 7.6 Changes in the Proportion of Foreign Born by City, 1890 to 1920 A. Chunges in proportion of foreign born in the populution (AFB)a Largest Increases AFB Decileb Smallest Increases AFB Decileb I890 to I900 New Bedford, MA Passaic, NJ Hartford, CT Bridgeport, CT Tampa, FL Middletown, CT Lincoln, NE Nashua, NH Providence, RI Pueblo, CO Lynn, MA New London, CT Somerville, MA Brockton, MA Schenectady, NY 1900 to 1910 Johnstown, PA Passaic, NJ Lynn, MA St. Joseph, MO Brooklyn, NY' Utica, NY Trenton, NJ Elizabeth, NJ Youngstown, OH Spokane, WA Bridgeport, CT Bayonne, NJ New Haven, CT Canton, OH New Bedford. MA ,056 6 .055 2 ,025 7 ,023 7 ,018 I ,018 I ,014 4 ,013 2 .011 9 .010 2 .009 7 .008 I ,008 6 .006 4 .005 3 ,072 3 ,056 3 .05 1 6 ,050 5 ,047 10 ,044 5 ,044 6 ,043 5 ,043 5 ,043 7 ,042 7 ,042 3 ,042 8 .04 1 2 .039 6 St. Paul, MN Spokane, WA Duluth, MN Portland, OR Milwaukee, WI Seattle, WA Davenport, IA Neenah, WI Tacoma, WA Saginaw, MI Minneapolis, MN Holyoke, MA Chicago, IL Dubuque, IA Cincinnati, OH Davenport, IA Fall River, MA Covington, KY Clinton, IA Saginaw, MI Fort Worth, TX Quincy, IL Troy, NY Oshkosh, WI Dubuque, IA Evansville, IN Peoria, IL Salt Lake City, U'I Louisville, KY St. Paul. MO -.I12 -.I03 -.093 - .08 1 -.077 -.076 .07 1 - .070 - ,069 - .067 - .067 -.065 p.064 - ,063 - ,063 ,052 -.050 - ,050 - .049 -.038 ,037 - ,037 - ,036 -.035 -.033 -.031 - .028 - .027 p.026 - ,023 9 5 5 8 10 7 3 1 4 4 9 5 10 3 10 2 7 3 1 2 3 1 5 1 1 4 4 6 9 8 B. Regression of difference in % foreign born between I and ( t i 10) on %foreign born in yeartd Dependent Variable Mean % Foreign Born N R' Unweighted Weighted Coefficient (t-stat.) on 1890 to 1900 -.I35 (-10.4) 127 .68 - ,0296 -.0373 1900 to 1910 ,192 (1.86) 127 .27 ,0045 ,0131 1910 to 1920 -.I 19 (- 11.2) 115 .52 p.0298 .0390 C. Regression of difference in %foreign born between t und (t i 10) on log ofpopulation in year'd Coefficient (t-stat.) on Log Population N R' 1890 to 1900 -.0041 (-3.20) 127 5 2 I900 to 1910 ,0053 (4.26) 127 .42 1910 to 1920 -.0057 (-5.31) 115 .I9 (continued) 246 Claudia Goldin reau of Labor Statistics (BLS) wages and hours series for nonunionized em- ployees that were used by Paul Douglas and Albert Rees, among others. As many as one hundred cities were surveyed for each of about twenty occupa- tions, with information on hourly earnings given annually. For the 1907-23 period the BLS wages and hours series covers unionized workers in thirteen occupations across sixty-six cities. In the data from I890 to 1907, two groups of occupations have been selected for study. The first includes four types of laborers-working in foundries, by contract on streets and sewers, in municipal street and sewer work, and in the building trades, as common laborers and as hod carriers. A second group in- cludes skilled workers-painters, bricklayers, plasterers, plumbers, and ma- chinists working in foundries and machine shops. The series through 1903 is contained in the Nineteenth Annual Report (U.S. Commission of Labor 1905) and is continued through 1907 in the subsequent BLS wages and hours series, although with a reduced number of cities. After 1907 the series covers only unionized employees by occupation. In the data from 1907 to 1923 there are only skilled workers and their helpers-bricklayers, carpenters, wiremen and their helpers, painters, steamfitters and their helpers, and iron finishers and their helpers. Both sets of data-those for the nonunionized sample and the unionized-contain hourly wages by year and occupation for a large number of cities. That for the nonunionized group contains the number of workers in the occupation-city cell, whereas that for the unionized group does not. Among the building tradesmen, laborers had about the same proportion of new and recent immigrants as did the entire goods-producing labor force. Painters and masons, however, were disproportionately native born (see table 7.5), although a large fraction of the masons were from older immigrants groups, such as Germans. Among street and sewer workers 22 percent were the new and recent immigrants, whereas only 12.2 percent of all in the entire goods-producing sector were, yielding a relative proportion of 1 .8.35 City-level earnings data can also be found in the censuses of manufacturing for 1899, 1904, 1909, and 1914. The data in this source are by industry, not occupation. ,211 employees, not just adult males, are covered, although for some of the industries men were the bulk of the labor force. Annual earnings per production workers, not hourly wages, are available for each of the four years considered. Four industries-men’s clothing, printing and publishing, bread and bakery products, and foundries-have been chosen to span the various characteristics of workers and products. The most serious constraint on the choice of indus- tries was that the number of cities represented had to be substantial, and not many industries were found in a large enough sample of cities. Further, the 35. The data on street and sewer workers are not included in table 7.5. Foreign-born workers were 49.4 percent of all street laborers, the new immigrants were 30.5 percent, and the new and recent immigrants wcre 22.0 percent. 247 The Political Economy of Immigration Restriction in the U.S., 1890 to 1921 choice of industries was governed by the skills and ethnic composition of workers. The nature of the product, as will be apparent soon, was also a consid- eration. Men’s clothing hired immigrant labor to a very large extent, particularly tai- lors who came to America with training and who worked in the production of coats that were traded nationally. Printing and publishing, at the other end of the spectrum, hired more highly educated laborers and very few immigrants- only 2 percent of its workforce were new and recent immigrants (table 7.5). The product was often locally consumed newspapers. Bread and other bakery products, like men’s clothing, had large numbers of immigrants among its workers and was found in virtually every city, and like printing and publishing, its product was generally nontraded. Foundries hired a mixture of skills and produced a nationally traded good. Although foundry laborers were dispropor- tionately new and recent immigrants, few machinists were. The impact of immigrants on the wages of workers already in an industry depends on the complementarity versus substitutability between the two la- boring groups in the production function. It also depends on how much immi- grants increase the demand for the good produced by the industry. Immigrants increase the demand for many types of goods, but their impact on local wages is greater and more positive if these goods are produced locally. In terms of the two main determinants of the impact of immigration on wages, the four industries considered here can be categorized using the following matrix: Immigrants as a Percentage of the Labor Force Below Average Above Average _ _ _ _ _ _ _ ~ Product Demand Local Printing National Foundries Bakeries Clothing 7.3.3 The Economic Impact of Immigration on Local Labor Markets The objective of this section is to estimate the impact of immigration on the wage outcomes of native-born workers, in part to assess whether immigration restriction was motivated by economic concerns. Immigration to particular cities, like that to particular countries, was not exogenous. Rather, immigrants went to cities that had high wages. Thus a simple cross-sectional regression of city-level wages on the percentage of immigrants yields a strong positive coefficient, as is apparent in the regression coefficients in table 7.7. But rather than indicating that immigrants caused wages to increase, the result suggests that immigrants sought out labor markets with high wages. Certain cities could have had higher demand curves for less-skilled labor than did others. If this higher demand were a permanent feature of the city, as opposed to one that was transitory, there is a simple way around simultaneity. 248 Claudia Goldin Table 7.7 Cross-sectional Relationship between Immigrant Flows and City Wages A. Regression of hourly wages o n fraction of immigrants. by city for various occupations, 1890-1 91 0 Using 1893 Wage, Using 1903 Wage, Occupations Elasticity" N Elasticityd N Laborers and hod carriers 0.094 192 0.135 192 Building trades and machinists 0.101 278 0.082 278 B. Regression of annual earnings on frucrion of immigrants, by city for various industries, 1900-191 0 Using 1904 Wage, Industries Elasticityd N Bakeries Clothing Foundries Printing 0. I26 108 0. I25 48 0.078 101 0.092 105 Sources: By occupation: U.S. Commissioner of Labor 1905; by industry: U.S. Bureau of the Cen- sus, Census of Manufactures, 1904; population: U.S. Bureau of the Census, Census of Populu- rion, 1890-1920. Note; Fraction of immigrants = [foreign horn in (f + 10) - foreign born in r]/[average population from r to ( t + lo)]. "he elasticities are evaluated at the means from a regression of the wage in the year given on the percentage of the city population that was immigrant, where immigrant = (foreign horn in year t + 10) - (foreign born in year t ) . The regressions are weighted by the number of workers in each occupation-city cell or in each city-industry cell. When the 1893 wage is used, the percentage immigrant is for 1890 to 1900; when the 1903 (or 1904) wage is used, the percentage immigrant is for 1900 to 1910. The method is to estimate a difference equation. The difference in the (log of) wages for a group of workers is regressed on the difference in the percentage of the population (or the labor force) that is immigrant. The procedure, which estimates a fixed-effect model, assumes that, for each city i, the (log) wage at time t , (wJ, is a function of the percentage foreign born, (TI), and an error term consisting of a portion that may be correlated with F,,, .sr or the fixed effect, and a portion that is not, (p,,,): (1) Mw,,) = P" + P, (TI) + + F,,. If equation (1) were estimated, the coefficient of interest, @,, would be biased because cities that have positive demand shocks will have both high wages and a high percentage foreign born. By first differencing (and dropping the i subscripts) we get 251 The Political Economy of Immigration Restriction in the U.S., 1890 to 1921 Table 7.8 Percentage Change in Wages with a Percentage Point Change in the Proportion of Foreign Born: City-Level Observations by Occupation or Industry, 1890 to 1914 P t-Statistics N Laborersb 1890 to 1897 1890 to 1903 1890 to 1907 1890 to 1897 1890 to I903 1890 to I907 Artisans' Artisans' I907 to 1915 1909 to 1915 1907 to 1923 1909 to 1923 By occupation, nonunion," hourly wage -0.010 (-0.053) - 1.02 (-2.98) -1.60 (-3.39) 0.679 (2.92) -0.539 (-1.88) -0.145 (-0.33) By occupution, union." weekly wage -1.44 (-3.27) -1.20 (-3.58) -1.60 (-2.81) -1.41 (-2.65) I92 I92 I60 278 278 162 223 223 225 225 By industqd annuul wage 1899 to 1914 Bread and bakery products 0.418 (0.69) 107 Clothing, men'sd -3.06 (-2.45) 27 Foundry -0.829 (- 1.92) 91 Printing and publishing 0.764 ( I .47) 104 Sources: By occupation, nonunion: 1890-1903, U.S. Commissioner of Labor 1905; 1907, Depart- ment of Commerce and Labor 1908. By occupation, union: data provided by Shawn Kantor, from U S . Department of Labor, 1907-23. By industry: U.S. Bureau of the Census, Census ofManufac- tures, 1899-1914. Population: US. Bureau of the Census, Census of Population, 1890-1920. Notes: Regressions are estimated for each group of occupations or each industry. The dependent variable is the difference in the log of wages between the end and beginning years. Percentage foreign born is (foreign born)/(total population). All regressions have been weighted by the average number of sample workers in the interval, except those for the union sample, where the weights are the log of city population in 1910. The growth rate of the population (difference in the log of the population between the end and beginning years) is also included as an independent variable in the regressions. "he change in the percentage foreign born is for 1890-1900 for the nonunion occupation data and for 1910-20 for the unionized occupation data. That for industry uses 1900-1910. bLaborers include laborers in building trades, in foundries, and in streets and sewer work (munici- pal city and contract) and hod carriers. 'Nonunion artisans include building tradesmen (bricklayers, carpenters, painters, plasterers, and plumbers) and machinists in foundries. Union artisans include bricklayers, carpenters, wiremen, painters, steamfitters, and structural-iron workers. dExcludes firms that do not remain in the sample to 1919 and the observation for New York City. 252 Claudia Goldin That they were perceived as a threat is clear in several labor union journals of the time. A mason in New York, for example, complained in 1906 that “emi- grants come [to New York City] with the intention of making big money. . . . By their killing work they drive down the American bricklayer, for if he does not follow suit he will have to join the great army of unemployed brickies that are now marching through this wonderful state” (Bricklayers and Masons International Union, September 1906). The industry results conform to the predictions regarding the roles of labor composition and product demand. In men’s clothing, which contained a large proportion of immigrants, wages were distinctly depressed in cities having an increase from 1899 to I909 in the percentage of their populations that was foreign born. The decrease is substantial: a 1-percentage-point increase in the fraction of the city’s population that was foreign born decreased wages by about 1.5 to 3 percent. Foundries also show negative coefficients. Because foundries hired both skilled (native) and unskilled (foreign-born) workers (see table 7 .3 , the results are even more supportive of the view that immigration severely depressed the wages of less-skilled labor. The other two industries considered show small, generally insignificant, if not positive, coefficients. The absence of a negative effect in printing and pub- lishing, indeed the presence of a positive effect, should not be surprising. Most printing establishments employed skilled and native-born labor and produced a locally consumed good the demand for which would have risen with immi- g ra t i~n .~ ’ The small, positive, but always statistically insignificant effects of immigration on the wages of workers in bakeries may, as in the printing and publishing case, be due to the positive demand effect of immigration on a lo- cally consumed good. Bread was, and is, the staff of life, but was even more so for immigrant and poor populations in America.44 It should be noted that the generally negative impact of immigration on the wages of both lower-skilled and higher-skilled workers could not be caused by the simple addition to the working population of lower-waged workers. The mean wage is no more than a simple average of the wages of native-born and (the union sample) there was wage convergence, but the addition of the initial wage for the subperi- ods in table 7.8 left unchanged the magnitude and significance of the coefficient on the change in the percentage foreign born. 43. The largest positive effect would occur in an industry hiring both skilled and unskilled (or native and immigrant) labor in which the two types of labor are complementary and the good is locally consumed in its entirety, if the wages of only the skilled workers were considered. The data. however, consist of a labor force-weighted average of the wages of all workers in the industry. 44. In a simple model of local labor markets the nature of product demand alone cannot generate a positive impact of immigration on wages; one needs complementarity of demand between immi- grants and the labor in the occupation or industry. In the case of printing and publishing there were probably both effects. In the case of bakeries i t is less likely that both effects operated, and thus the existence of a positive coefficient is curious. Even if the wages of (skilled) labor hired by an industry were unaffected by the increase in (primarily unskilled) immigrants, the coefficient would be zero, not positive. See, for example, Altonji and Card 1991. 253 The Political Economy of Immigration Restriction in the US. , 1890 to 1921 foreign-born workers. If immigrants earned less than natives by virtue of their lack of skill or by dint of labor market discrimination, then the mean wage would have decreased as foreign-born workers increased. But the depressing impact of the foreign born on wages in the difference regressions is found even for the artisan group, which included very few of the new and recently arrived foreign born (see table 7.5). The difference in wages between immigrants and natives in the same occupation would have to have been extremely high to account for the large negative impact of immigration on wages in general and even for those occupations in which the foreign born were a large percentage. 7.3.4 Explaining the 19 15 Vote to Override Wilson’s Veto The wage effect of the foreign born suggests a role for economic forces in the movement to restrict immigration. The underlying model is one in which constituents more vigorously urge their representatives to vote for restriction (that is, to pass the literacy test or to override a presidential veto) when the increase in wages is lower (or the decrease in wages is greater). The foreign born may be the cause of the wage change, or they may be the scapegoats for other economic influences. But at the same time, if a large enough fraction of the constituents were themselves foreign born, they would probably urge their representatives to vote against restriction. Table 7.9 explores these two factors in determining the House vote in 19 15 on the override of President Wilson’s first veto of the literacy test. The data are by city for the union-occupation sample, whereas the votes are by congressional district. I have matched the cities to the district in the 63d Congress. For those cities covering more than one congressional district, the dependent variable is the fraction of representa- tives who voted to override the veto. The estimation is performed for all city- occupation observations in the union data set and for the non-South subset as well. Southern cities were typically small and voted overwhelmingly to over- ride Wilson’s veto. In both samples (all cities and the nowSouth) an increase in the wage by occupation, from 1907 to 1915, decreased the proportion of votes for the over- ride. A vote for the override was a vote for closing immigration; thus the lower the wage increase, the more support for closing immigration. Increasing the wage change by one standard deviation in the non-South sample (a 13 percent increase) would have decreased the percentage voting against open immigra- tion by 12 percentage points. The percentage foreign born in the city was an even more powerful determinant of the vote. The proportion of foreign born is divided into four groups to evenly divide the cities. In the non-South group, however, very few cities are in the smallest class of percentage foreign born. With the exception of these few small cities, increasing the percentage foreign born would decrease the probability of voting against the override by a substan- tial amount. When the foreign born were about 30 percent of the total popula- tion, almost all representatives voted against the override, given the mean val- ues for all other variables in the non-South sample. 256 Claudia Goldin small. Even for rural Americans, the well-being of those in the cities may have been the litmus test for immigration restriction. Pro-immigration support even- tually faded in the midsection of America, the far West, and all but the largest cities. A regime change was i n e ~ i t a b l e . ~ ~ From the early 1900s to 1917 it was just a matter of waiting for some exogenous force-an economic downturn, a war, a rash of labor unrest-to close the door. That 17 million slipped through from 1897 is the miracle. References Altonji, Joseph G., and David Card. 1991. The Effects of Immigration on the Labor Market Outcomes of Less-Skilled Natives. In John Abowd and Richard Freeman, eds., Immigration, Trade, and the Labor Market, 407-21. Chicago: University of Chicago Press. Benhabib, Jess. 1992. A Note on the Political Economy of Immigration. C. V. Starr Center for Applied Economics, New York University, Working Paper RR 92-42. Borjas, George J., Richard B. Freeman, and Lawrence F. Katz. 1992. On the Labor Market Effects of Immigration and Trade. In George J. Borjas and Richard B. Free- man, eds., Immigration and the Work Force: Economic Consequences for the United States and Source Areas, 213-44. Chicago: University of Chicago Press. Bricklayers and Masons International Union. 1906. Official Journal of the Bricklayers and Masons International Union of America. Indianapolis. Congressional Directory. 1896-1 9 17. Washington, DC: GPO. Congressional Record. 1897-19 17. Washington, DC: GPO. Coombs, Whitney. 1926. The Wages of Unskilled Labor in Manufacturing Industries in Douglas, Paul H. 1930. Real Wages in the United States, 1890-1926. Boston: Easterlin, Richard. 198 1. Why Isn’t the Whole World Developed? Journal of Economic Hannon, Joan. 1982. Ethnic Discrimination in a Nineteenth Century Mining District: the United States, 1890-1 924. New York: Columbia University Press. Houghton. History (March): 41: 1-20. Michigan Copper Mines, 1888. Explorutions in Economic History 19 (January): 28-50. Hatton, Timothy J., and Jeffrey G. Williamson. 1992. International Migration and World Development: A Historical Perspective. Harvard Institute of Economic Re- search Working Paper no. 1606. Heald, Morrell. 1953. Business Attitudes toward European Immigration, 1880-1 900. Journal of Economic History 13 (Summer): 291-304. Higham, John. [I9551 1981. Strangers in the Land: Patterns of American Nativism, 1860-1925. Westport, C T Greenwood Press. 46. See Benhabib 1992 for a theoretical model of why regime changes may be inevitable and what might explain the 1965 regime change (and why there could be another quite soon). Immigra- tion restriction cycles, according to Benhabib’s model, are rooted in a median-voter model with wealth accumulation. If the median voter is rich in capital, immigration will beopen. When the median voter becomes poor in capital, immigration will be restricted by capital (human and/or physical). 257 The Political Economy of Immigration Restriction in the US . , 1890 to 1921 Hutchinson, E. P. 1981. Legislative History of American Immigration Policy, 1798- 1965. Philadelphia: University of Pennsylvania Press. International Brotherhood of Teamsters, Chauffeurs, Stablemen, and Helpers. 191 3. Oficial Magazine of the International Brotherhood of Teamsters, Chauffers, Sta- blemen, and Helpers of America. Indianapolis. Jones, Maldwyn Allen. 1992. American Immigration. 2d edition. Chicago: University of Chicago Press. Keyssar, Alexander. Forthcoming. The Free Gift of the Ballot. Tables provided by author. Link, Arthur S. 1954. Woodrow Wilson and the Progressive Era, 1910-1917. New York: Harper and Brothers. Rees, Albert. [1961] 1975. Real Wages in Manufacturing, 1890-1914. New York: Arno Press. Robbins, Roy Marvin. [1942] 1976. Our Landed Heritage: The Public Domain, 1776- 1970. Lincoln: University of Nebraska Press. Simkovich, Boris, Alan Taylor, and Jeffrey G. Williamson. 1992. The Evolution of Global Labor Markets: Appendix. Harvard University. Taylor, Philip. 1971. The Distant Magnet: European Emigration to the U.S.A. London: Eyre and Spottiswoode. U S . Bureau of the Census. 1899-1914. Census of Manufactures: Reports by States, with Statistics for Principal Cities. Washington, DC: GPO. . 1890-1920. Census of Population. Washington, DC: GPO. . 1975. Historical Statistics of the United States: Colonial Times to 1970. Wash- ington, DC: GPO. U S . Commissioner of Labor. 1905. Nineteenth Annual Report, 1904: Wages and Hours of Labor: Washington, DC: GPO. US. Department of Commerce and Labor, 1908. Wages and Hours of Labor. Bulletin of the Bureau of Labor no. 77. Washington, DC: GPO. U S . Department of Labor. 1907-1923. Union Scales of Wages and Hours of Labor Washington, DC: GPO. U.S. Department of Labor. Bureau of Immigration. 1918. Report of the Commissioner General of Immigration. Washington, DC: GPO. U S . Senate. 191 la . Reports of the Immigration Commission: Abstracts of Reports of the Immigration Commission. Vol. 1. 61st Cong., 3d sess. S. Doc. 747. Washington, DC: GPO. . 191 Ib. Reports of the Immigration Commission. 42 vols. Also known as the Dillingham Commission Reports. Washington, DC: GPO. Walker, Francis A. 1899. Discussions in Economics and Statistics. New York: Henry Holt and Company. Wiebe, Robert H. 1962. Businessmen and Reform. Cambridge: Harvard University Press. Willcox, Walter F. 1931. International Migrations. Vol. 2. Interpretations. New York: National Bureau of Economic Research. Williamson, Jeffrey G. 1982. Immigrant-Inequality Trade-offs in the Promised Land: Income Distribution and Absorptive Capacity prior to the Quotas. In Barry Chiswick, ed., The Gateway: U.S. Immigration Issues and Policies, 25 1-88. Washington, DC: American Enterprise Institute. This Page Intentionally Left Blank