Skip to navigation
Skip to main content
Immigration, Politics, and the American Labor Market: An Historical Perspective -- Alan Kessler
An Historical Perspective
Department of Political Science
University of California, Los Angeles
Los Angeles, CA 90095-1472
(310) 825-4331/ 825-0778 fax
Prepared for delivery at the Comparative Immigration and Integration Program Winter Workshop, University of California, San Diego, Thursday-Friday, February 19-20, 1998. Preliminary draft. Do not cite without author's permission. Comments and suggestions are greatly appreciated.
The nineteenth century is often characterized as the "liberal epoch" of international migration. Somewhat unexpectedly, however, the liberal epoch came to an abrupt end when the largest of the migrant receiving nations, the United States, adopted restrictions on immigration in the early 1920s. Despite the persistent efforts of anti-immigrant forces at the turn of the century, the United States had remained a country remarkably receptive to immigration. With war in Europe, adverse economic conditions at home, and fears of renewed mass migration, however, the coalition in support of open immigration ultimately unraveled. This paper develops a factor endowment based model of immigration policy to account for the variation in migration flows and policies over time. The basic insights of the Heckscher-Ohlin model are extended to the case of labor mobility, and hypotheses regarding the preferences of unskilled labor, skilled labor, and capital owners over alternative immigration policies derived. Evidence in support of the proposition that the distributional consequences of immigration drive group demands for particular policies is adduced and the welfare implications of immigration for various domestic factor owners in sending and receiving states are then explored.
I draw on international trade theory in an effort to provide one such account below. In short, I couple a simple trade model with a stylized model of domestic politics in order to generate insights into a country's choice of migration policy. I argue that the welfare effects of immigration on domestic factors of production, particularly labor, are key determinants of a country's migration policy. Where immigrants and domestic workers are substitutes, or competitors in the labor market, I expect "labor" to lobby more vociferously for protection, in this case immigration restriction. Whether the demand for immigration restriction is met, however, depends upon supply considerations as well. Thus, where organized labor and pro-labor governments coexist, I expect national policy to exhibit a greater tendency toward restriction. On the other hand, where labor exercises limited influence, governments beholden to capital or land owners have little to gain from restriction and liberal policies are more likely to prevail.
While a simple model of immigration policy by no means adequately explains the variety and complexity of national policies, it does offer helpful insights into the politics of immigration control. In fact, after a brief review of the literature in section one, I present the basic model and illustrate how it explains aspects of immigration flows and policies during the "liberal epoch" of European inter-continental emigration in sections two and three. I then turn to one of the most persistent puzzles in the migration literature in the fourth section; namely, why the United States, after almost a century of openness, closed the gates to immigrants in the 1920s. I consider how the model might be extended and applied to migration in more recent times in the fifth section and conclude in the sixth.
Scholars and policy makers have long been preoccupied with explaining the determinants and consequences of migration. Classical economists, such as J.R. Hicks, contended in the 1930s that "differences in net economic advantage, chiefly differences in wages, are the main cause of migration." Since then additional variables, such as distance, search and information costs, unemployment, and a variety of non-economic factors have been advanced as key determinants of migration. Yet, despite a host of theoretical and empirical work, the "sizable literature dealing with the determinants of migration is," in the words of one observer, "almost completely devoid of direct policy implications."
Accordingly, economists and other social scientists interested in migration policy have turned to the consequences of migration, in particular, the welfare effects of labor flows between countries. As Borjas notes, recent economic research addresses the impact of immigration on the economies of sending and receiving countries and investigates the effect of immigration on national labor markets. The policy implications of this line of research are, in theory, more straightforward. Where immigrants exercise an innocuous effect on the labor market, open migration policies are to be encouraged. Conversely, where immigrants adversely affect the earnings of domestic labor, restrictive policies are more likely to be advocated.
The economic consequences of immigration thus play a crucial role in shaping national migration policies. Yet exactly how migration affects the welfare of citizens in the sending and receiving states is unclear, and the political implications of immigration, although the subject of heated debate, remain largely unexplored in the economic literature.
Political scientists, in contrast, directly confront the political implications of immigration for receiving states and have offered a number of insightful contributions to the contemporary immigration debate. Foremost among these contributions are arguments identifying political parties, economic interests, and spatial concentration as the key components of national migration policies.
Political Parties. A frequent explanation for variation in the national migration policies of receiving states emphasizes the role of political parties and electoral politics. In this line of argument, party platforms and positions on immigration are advanced as key determinants of national policies, with shifts in policy presumably reflecting changes in the composition of national governments. The rise of right-wing parties and contemporary politicization of immigration figure prominently in many analysts' explanations of the turn toward restriction in Western Europe. From this perspective, the extreme right has successfully exploited anti-migrant sentiment for electoral gain, driving mainstream parties and national policies to the (political) right in the process.
In persuasively documenting the role of political parties, particularly the extreme right, in national migration debates, these accounts offer an important, but ultimately incomplete, explanation for the supply of restrictive immigration policies. Partisan efforts to reap electoral rewards through anti-immigrant politics, although increasingly apparent in the United States and much of Western Europe, have been surprisingly infrequent in the postwar era. Even in Britain, France, and Switzerland, countries often beset by divisive national controversies over migration, immigration policies exhibit notable stability regardless of public opinion and changes in national governments. Similarly, postwar U.S. immigration policy, despite heated partisan debates, remains an area governed by "benign neglect" under both Democratic and Republican administrations. Empirically, the persistence of similar migration policies across diverse governments poses a significant challenge to purely party-based accounts of immigration policies, suggesting that additional factors are at play.
Interest Groups. A prominent alternative to the party-based approach contends that the demands of influential interest groups underlie variation in national immigration policies. In this line of argument, consensus management among major parties implies that governments respond not to the general public but, instead, to organized interests. Immigration policy making is characterized as a form of client politics, dominated by employers, and policy is thus held to exhibit an expansive bias. Because other societal interests experience costs and benefits from immigration as well, policy is viewed, in pluralist fashion, as the result of political bargaining. Restrictive policies are expected infrequently, when adverse economic conditions and organized opposition transform client politics into more competitive interest group politics, thereby prompting vote-maximizing politicians to reconsider the electoral benefits of open policies.
In examining how economic interests identify political actors preferences over alternative migration policies and considering politicians' incentives to comply, interest group accounts lay the groundwork for a persuasive account of variation in national migration policies. In fact, the approach offers insights into migration politics and the policies of many postwar advanced industrial states. Yet, as Freeman explains, the framework "needs to be supplemented by a theoretical statement of the distribution and incidence of the costs and benefits of immigration." In the absence of such a statement, the assertion that policy should exhibit an expansive bias remains unclear. Why, for instance, do politicians privilege the policy demands of employers as opposed to other organized societal interests? Greater attention to both the distributional impact of immigration for various socioeconomic actors and politicians' incentives to respond is required in order to identify political cleavages and conflicts over migration.
Spatial Concentration. An alternative, but closely related, approach to explaining variation in national migration policies emphasizes how the costs and benefits of immigration at the local level create a spatial context for immigration control at the national level. In this perspective, the geographical concentration of immigrant communities and competition between native and immigrant groups over scarce resources are posited as the forces underlying the demand for local migration policies. Local pressures are then catapulted to the national political agenda when political parties view these constituencies as crucial to creating or maintaining a national political coalition.
In focusing on how politicians respond to local constituencies in building national coalitions, the spatial approach makes a novel contribution to the study of immigration politics and policy. Yet, by rooting the political response to migration at the local level in the competition between competing interests, spatial arguments mirror those of economic interests without offering additional insights into how the costs and benefits from immigration drive demands for political action. Spatial concentration increases the intensity with which groups who gain or lose from migration might prefer particular policies, but says little about the actual incentives of various interests, their role in the policy making process, or why politicians might favor the demands of one group over those of another. The bridge between local and national party positions is thus tenuous and the position of the local median voter ambiguous a priori.
Theories positing political parties, economic interests, and spatial concentration as the main determinants of immigration politics and policies offer broad insights into the sources of supply and demand for alternative policies. At the same time, much about the preferences of socioeconomic actors in favor of liberal or restrictive regimes and subsequent policy outcomes remains unexplained. Why do particular firms favor liberal migration, as proponents of these arguments maintain, and under what conditions might policy exhibit an expansive bias? What underlies labor organizations' possible objection to immigration and when will politicians respond in kind? In order to convincingly answer these questions and to explain the coalitions and conflicts that arise over immigration, I contend we must consider how the distributional effects of variation in international labor mobility drive socioeconomic actors to pursue their interests in the policy making process.
The "standard trade model" of international economics begins with a few simple assumptions about endowments, technology, and preferences and, from there, derives a number of interesting propositions about the patterns of trade between and the distribution of income within countries. Trade theory assumes that labor is immobile across countries, but nonetheless yields results of great relevance to the study of migration: the by now familiar Heckscher-Ohlin, Stolper-Samuelson and factor-price equalization theorems. These theorems imply that labor abundant countries, which export goods that are relatively intensive in the production of labor, in essence export labor. The export of labor intensive goods, in this framework, leads to the equalization of wage rates across countries even in the case of labor immobility as trade in goods substitutes for trade in factors. Because factor movements can substitute for trade, as Krugman and Obstfeld note, "it is not surprising that international migration of labor is similar in its causes and effects to international trade based on differences in resources." Accordingly, I turn to a simple model of international trade and explore the consequences of immigration within the context of the model.
Consider a standard two-country world consisting of A(nonymous) and W(orld), each endowed with two factors of production, land and labor. Furthermore, assume for the moment, that the two countries produce only one good, "output." If both countries produce the same good, there is little scope for trade in goods in the model; thus trade in factors represents the only route to exchange at this point. Because land is immobile, this scenario represents a simple model of exchange through international labor mobility.
Further assume that countries A and W possess similar levels of technology but differ in their endowments of land and labor (land-labor ratios). An incentive for migration now exists. If W is the labor-abundant country, workers there will earn less than those in A, while land in W will earn more than in A creating an incentive for factors of production to move. Workers in W would clearly welcome the opportunity to earn higher wages in A, while landowners in A desire the higher rents in W. With international migration, workers move from W to A reducing the labor force in W, thus raising real wages there, while increasing the labor force in A and reducing wages there. In the absence of institutional impediments and high transportation or other costs, this process should continue until the marginal product of labor is the same in both countries.
Yet, institutional impediments to labor mobility often exist. Despite the increase in world output accompanying migration (illustrated below), restrictions on the free flow of labor are the rule rather than the exception. As one noted observer remarks:
B. The Gains and Loses from Migration
Figure 1 illustrates the causes and effects of international migration graphically. The left panel depicts the supply of labor in country W, assumed to be perfectly elastic at equilibrium wage We. The right panel shows the labor market in the country of immigration labeled A. In the absence of immigration restrictions or other barriers to entry (high transportation or information costs, for example), profit minded workers migrate from W to A. In this case, L3L4 (= L'1L'4) laborers migrate to A, increasing A's labor supply from SA to S"A and decreasing its wage rate from WA, the closed economy rate, to the world equilibrium rate.
Suppose, however, that for some reason (political or other) country A imposes restrictions on immigration. If A limits the inflow of migrants to some arbitrary amount, say L1L4 (=L'1L'4), its labor supply rises from SA to S'A. This increase in the labor supply has important distributional repercussions. Notably, the domestic wage rate falls from WA to W'A and total employment rises from OL5 to OL2. At the same time, however, domestic employment falls by L1L5 from OL5 to OL1 indicating that, to some extent, immigrants displace domestic workers in the aggregate. In addition, the income accruing to domestic workers falls with immigration. Note that when wages fall from WA to W'A, labor's income changes from OxzL5 to OcaL2, of which domestic workers earn OcbL1 and immigrants L1baL2. Domestic workers thus lose the equivalent of cxzb in wages. Concurrently though returns to nonlabor factors of production rise from xyz to cya, indicating that certain groups do indeed gain while others lose as the quote from Krugman and Obstfeld above suggests.
International migration also involves a redistribution of income between sending and receiving states. While of less concern in the discussion to follow, these distributional consequences are readily apparent in Figure 1. Migration appears to clearly benefit the receiving state, which gains L5zaL4 in aggregate output. Furthermore, as long as positive wage differentials between states drive migration, the gain in output of the receiving state outweighs the loss of the sending nation and global output rises. The effects on the sending state are not quite as positive, with output falling in accordance with the number of unemployed and characteristics of the migrants.
While undoubtedly highly stylized, the simple two country, two factor, one good model of international labor mobility offers insights into why factor movements occur and their likely distributional effects. "Extending the simple model," as Krugman and Obstfeld demonstrate, "does not change its basic message. The main point is that trade in factors is in purely economic terms very much like trade in goods, occurring for much the same reasons and producing similar results." Migration in the model, like trade in the Heckscher-Ohlin world, is driven by differences in endowments. Similarly, migration, like trade, is welfare improving in the aggregate sense that world output increases, but, at the same time, has strong distributional effects complicating the realization of such increases. From the perspective of national welfare, immigration restriction thus represents a puzzle analogous to that of protectionism, an understanding of which must be sought in the distributional struggle between domestic winners and losers.
The effects of immigration on domestic factors of production in the receiving state should by now be clear. Immigration, in the short run, increases the relative rates of return of factors that are complementary with the migrant's factor endowments but decreases the rates of returns of factors that are substitutable for those of the migrant. Thus, if migrants and domestic workers are substitutes in production, the relative earnings of domestic landholders (or capitalists) will rise, while those of domestic labor will fall. Consequently, labor has a clear political incentive to lobby for immigration restriction whereas land (or capital) has none.
Yet, the demand for immigration policy is only part of the story. Even if one assumes that factor owners demand policies that increase the returns to their factor and lobby politicians accordingly, as I do, migration policy remains uncertain without greater attention to supply-side considerations. I therefore examine how the role of factor owners in the political process affects a government's choice of migration policy before turning to the historical record below.
The discussion of Figure 1 to this point has involved little mention of the state's role in setting migration policy. Implicit in the underlying model, however, is the notion that government maximizes aggregate returns to factors dominant in the political process. In the simple two factor, one good case, government aims to augment aggregate output by maximizing the trade-off between wages and rents. Where labor is the dominant force in the political process, re-election minded politicians maximize returns to labor. In the case above, where immigrants and domestic labor are substitutes in production, the expectation is thus that politicians beholden to labor enact restrictive immigration policies. Conversely, where land holders (or capital owners) predominantly shape policy, politicians aim to enhance returns to land. The expectation here would thus be a liberal immigration policy aimed at increasing rents and depressing wages. Extending this logic to the incentives of political parties, viewed as the representatives of particular coalitions of factor owners, yields the additional expectation that labor or left parties, whose core constituency is composed largely of domestic labor, are more likely to advocate restriction than conservative or right parties, whose base of support typically lies elsewhere.
The supply of policy (to be overly economistic) is thus a constrained optimization problem for the state, in which the government maximizes an objective function, with factor prices as arguments and endowments as constraints, that best ensures it continuation while meeting the demands of its constituents. The solution to such a problem mirrors and extends the analysis of Figure 1 and has recently been provided by Foreman-Peck. Much like in the analysis above, migration harms domestic workers who compete with immigrants in the labor market while broadly benefiting other factors. Coupling the distributional effects of immigration with this idealized notion of the political process thus yields several seemingly robust propositions about the likely course of migration policy. Notably, states in which labor plays a significant role in the political process are likely to pursue restrictive immigration policies when immigrants and domestic workers are substitutes in production, but less likely to do so where immigrant and domestic labor are complements. Conversely, states dominated by landowners (or capital holders) are likely to pursue liberal immigration policies regardless of the impact on domestic labor. Finally, more politically competitive states are likely to fall in the intermediate range, with migration policies less predictable and more dependent upon aggregate economic conditions. With these expectations in mind, a glance at the historical record of international migration is now in order.
III. The Political Economy of Immigration Restriction
How does one account for the U.S. turn from an open to a restrictive migration policy? Does the experience of the 1920s hold lessons for policy makers today? Before suggesting possible answers, I illustrate how the simple model of the preceding section broadly captures significant aspects of international migration flows and policies in the pre-World War I international economy.
A. International Migration Flows, 1870-1914
If "the years from 1870 to the first World War were," as Green and Urquhart suggest, "a time of mass migration of a kind not experienced before or since," the factor based model of migration should offer some insights into labor movements and migration policies during this time, and indeed it does. A glance at Green and Urquhart's migration data, reproduced in Table 1, reveals that the "movements of population were from resource scarce (densely populated) areas to resource abundant (sparsely populated) regions," with the U.S., Canada, Australia, Argentina, and Brazil receiving the bulk of intercontinental migrants.
In addition, the potential rewards to migrants, gauged in terms of either gross domestic product per capita or wage differentials, explain a large component of the observed pattern of migration. As the figures in Table 2 reveal, European sending countries, with the exception of the United Kingdom, possess relatively low per capita incomes in comparison with those of the receiving countries. The real wage data in Table 3 tell essentially the same story. Wage and income differentials thus appear to play a key (but by no means exhaustive) role in explaining the pattern of nineteenth century migration. In fact, a naive regression of migration against income, accounts for roughly 40 percent of the variation in decadal migration rates from 1870-1920.
Table 2. Gross Domestic Product Per Capita, 1870-1910 (in 1970 dollars) Country 1870 1880 1890 1900 1910
on Great Britain 1905=100)
Country 1870 1880 1890 1900 1910
Source: Williamson (1995).
Table 4. Determinants of International Migration Flows, 1870-1920
(1) IMigr = -7.44 + 0.80 GDPpcap
R2 = 0.41
(-3.29) (5.34) (2.85)
R2 = 0.47
a OLS estimates of pooled cross-sectional analysis for 1870s through 1910s. t-values in parentheses.
Note: The dependent variable, IMigr, is decadal migration as a percent of initial population from Green and Urquhart(1976), pp.224-225. GDPpcap is gross domestic product per capita in 1970 dollars as defined by Maddison(1979), pp.425-429. PDens is population density per square mile from Banks (1971). Countries included in the analysis include: Great Britain, France, Germany, Italy, Ireland, Spain, Sweden, Portugal, U.S., Canada, and Australia.
Number of Cases
*Represents significant at the 5 percent level. t-statistics in parentheses.
Yet, while these pedestrian accounts do generate some insights into the understanding of international migration patterns, they say little about actual policies, or of related interest, variations in policies across similarly endowed countries. To get at the matter of policy choice, it is thus necessary to move from the aggregate flow of labor to its factor market consequences and to investigate whether or not the simple factor endowment based model is of any help on this score. This is, however, a task complicated by both the small number of immigrant receiving countries and the lack of adequate data for the time frame in general. Accordingly, I merely illustrate how the model captures aspects of migration policies of the major immigrant receiving states during the late nineteenth and early centuries in a cursory manner before turning to a more detailed examination of the U.S. case below.
At first glance, the international migration policies of the major receiving states conform quite well to those predicted in terms of the model. On the mere basis of factor endowments alone, one would expect land-abundant, labor-scarce countries to attract migrants and, indeed, they do. The land-abundant dominions of the British empire in the Americas and Australasia are uniformly populated by migrants in the nineteenth and early twentieth centuries. Canada, Australia, New Zealand, Argentina, and Brazil alone attract over seven million European migrants between 1870 and 1920. Each country, unsurprisingly, pursued relatively liberal immigration policies, some at times actively subsidizing immigration. Endowments alone, however, present only a partial portrait of a nation's choice of migration policy. As Denoon notes in a related context:
Adding political considerations to the endowment based model, in fact, helps account for the variation in policies between similarly endowed states. As noted above, where landed interests dominate the political process of the receiving state, a bias against wage earners inheres and liberal migration policies are to be expected. This characterization offers a plausible (if admittedly impressionistic) explanation for the divergent migration policies of Latin American and Australasian land-abundant states. Whereas the landed oligarchies of Argentina and Brazil, committed equally to unrestricted exports and migration, resisted calls for restriction, relatively weaker landed interests in Canada and Australia were unable to completely resist the demands of labor. "The option of cheap, migrant, or coerced labour was simply not available for Australian rural producers." The difference between Uruguay and New Zealand's policies can be explained in similar terms. Landed interests in both countries stood to gain, but only in Uruguay where they arguably were the state, did liberal immigration prevail. While these examples are but caricatures of historically fascinating cases, the outcomes of migration policies fall within the expectations of the model. In the land-abundant and land-owner dominated Latin American cases, open migration policies result. Conversely, in the land-abundant but more politically competitive political systems of Australia, New Zealand, and Canada, policies of "managed" migration ensue.
In this light, the U.S. represents an interesting apparent anomaly. Why does the United States, a land-abundant and politically competitive nation, suddenly adopt immigration restrictions in the early 1920s? I contend that the answer lies in the labor market consequences of immigration and the electoral incentives of American politicians.
The endowment based model above suggests an immediate explanation, namely, the end of land-abundance, or in the U.S. context, the waning of the frontier. Land in nineteenth century America, as Briggs argues, was often equated with opportunity. By 1890, however, "the era of free land was essentially over as the frontier had virtually disappeared and with it went any rural settlement option that might have existed for subsequent immigrant streams."
Yet if the frontier had been the sole draw for migrants and land-abundance the mainstay of liberal migration policy, a reduction in immigration and/or a trend toward restriction should have ensued before the turn of the century, not in the 1920s. A glance at the annual rate of immigration data presented in Table 6 reveals a decline but then a resurgence of immigration at the turn of the century, with a precipitous decline awaiting the onset of the First World War. The evidence on U.S. immigration legislation is, however, more ambiguous and worthy of note.
After a brief flirtation with the regulation of immigration in 1798, no federal restrictions on immigration existed for nearly a century. "Throughout its first 133 years as an independent nation (1788-1921)," as Briggs slightly overstates, "there were no limits on the number of immigrants who could enter the United States per year." Only with the rising tide of immigrants in the wake of the Civil War did immigration become a divisive national issue. From 1860-1890, roughly ten million immigrants poured into the U.S., nearly twice as many as in the previous three decades. Between 1890 and 1914 the numbers were even greater, almost fifteen million additional immigrants arrived, with over one million alone in 1905, 1906, 1907, 1910, 1913, and 1914.
This gradual movement away from open immigration might thus constitute partial support for the waning of the frontier thesis (allowing for considerable lag time in curtailing immigration). The explanation becomes more credible, however, once the role of various factors in the political process and the labor market consequences of migration are examined. Viewing the U.S. as a competitive democracy during this period, I argue that the impact of immigration on domestic labor and its concomitant concern with immigration policy implied the turn toward restriction at the close of the nineteenth century, and, in combination with the impact of World War I, the adoption of the national quota system in the 1920s. While this argument is undoubtedly controversial and incomplete, I demonstrate that labor consistently supported restrictive legislation and that politicians responded in only a palliative manner until other actors joined the restrictionist bandwagon in the wake of the war. I then turn to a cursory review of more recent developments in immigration policy to suggest how the model might offer insights into contemporary policy both here and abroad.
Regardless of the overall impact of immigrants on the economy, American wage-earners during the late nineteenth century "eyed the foreigner for what he was at the moment -- a cheap competitor, whose presence undoubtedly held down wages and bred unemployment." Much like import competing industrialists who hoped to gain from the tariff, workers hoped to gain through anti-foreigner discrimination. In the wake of the depression of 1877, the California labor movement became local and then national "champions of restrictions on Chinese and other Oriental immigrants, regarding these workers as a threat to wages, working conditions, and [the] general welfare of all workers." In fact, labor leaders in California secured the aid of contemporaries in other parts of the country in their campaign against the Chinese. As Taft notes, opposition to Chinese immigration became an article of faith among organized labor, whose persistent agitation led to the enactment of the first Chinese exclusion law in 1882. That year also saw the introduction of a 50 cent head tax on each immigrant and excluded various classes of undesirable migrants.
The restrictionist turn did not end there however. The dramatic influx of immigrants after 1882 and the industrial downturn of 1883-1886 "persuaded many American wage-earners that the whole incoming stream directly threatened their own livelihood." Against this backdrop, it is hardly surprising that the nascent Knights of Labor succeeded in gaining the adoption of the Alien Contract Act by Congress in 1885. Unlike earlier laws, such as those of 1875 and 1882 excluding criminals and paupers, the "alien contract labor law was the first enactment to exclude able and industrious Europeans." These laws were renewed and extended in subsequent sessions of Congress as labor and even segments of the broader public fell victim to nativism in the late 1880s.
Yet, despite the continuous arrival of the immigrants and the worsening economic conditions of the 1890s, the pace of both immigration and restrictive legislation slowed. While a number of state level initiatives sought to curtail immigration, for all the hysteria and hatreds of the decade, federal policy proved difficult to change. In fact, with the Immigration Act of 1891, the course of legislation turned from the xenophobia of the era to preserving and institutionalizing the provisions of the 1880s.
Only with the resurgence of mass immigration after the turn of the century, apparent in Figure 2 below, did the proponents of restriction find renewed, albeit limited, success in the legislature. Following the influx of just over 1.2 million immigrants in 1907, Congress reaffirmed earlier restrictive measures and raised the head tax in the Immigration Act of 1907. This act also authorized the creation of a commission for the study of immigration, whose forty two volume final report of 1911, known as the "Dillingham Reports," provided ample ammunition for the restrictionists. The scope and coverage of the volumes were truly imposing, yet the conclusions (while empirically suspect) were politically quite clear. Jenks and Lauck convey the views of the Immigration Commission nicely:
The commission endorsed the literacy test, favored a reduction in the supply of unskilled migrants, and cast aspersions on both the character and abilities of the new immigrants. Despite widespread support in Congress, however, the commission's recommendations were repudiated, at least for the moment, by President Taft's veto of the literacy test in 1913 and again by Wilson's veto in 1915.
Sources: Historical Statistics (1976), Series A-29 and C-89 for
immigration data and Weir (1992) for unemployment.
War proved more effective than legislation in stemming the tide of migration after 1914. Yet the old xenophobia surfaced during the period of U.S. neutrality and concerns over the loyalty of "hyphenated-Americans" reinvigorated the restrictionists. The sentiment conveyed in the Dillingham reports was not easily dispelled. "After a quarter of a century of agitation and over a score of favorable votes in Congress on literacy bills," as Higham details, "a provision excluding from the United States adult immigrants unable to read a simple passage in some language was enacted on February 5, 1917." The premise that the "national origin of an immigrants was a reliable indication of his capacity for Americanization" inhered in the immigration legislation of 1917 as restrictionists employed the literary test as a device "to eliminate the 'new' while perpetuating the 'old' immigrants." The embrace of the literacy test after some twenty odd years of debate, in combination with a host of restrictive measures prompted by the Red Scare and labor unrest during the war, signaled the impending end of open immigration.
The revival of immigration after the war again coincided with adverse economic conditions. As over five thousand immigrants a day poured into Ellis Island, the daily papers "teemed with hostile comments on the relations between the torrential influx and the worsening unemployment problem." The American Federation of Labor now championed the cause of immigration restriction with renewed vigor. In addition to organized labor, still reeling from its wartime association with socialism and radicalism, "even big business conceded the value of a 'selective' immigration policy" and "stringent restrictions seemed instantly imperative." A series of new legislation resulted. The most significant of which, the Quota Act of 1921, established the principle of national numerical limits in American immigration law. Meant as a temporary measure, the 1921 statute set annual admissions based on the preexisting composition of ethnic groups in the U.S. population at the time of the 1910 census. European immigration was thus limited to three percent of its current share of the population for an annual maximum quota of roughly 350,000 (most of which from Northwestern Europe) as ethnic affiliation became the mail criteria for admissions to the United States.
Yet restrictionists remained dissatisfied, complaining that "too great a number of Italians, Slavs, east European Jews, and Greeks were eligible for admission." Indeed, the AFL and its chief arbiter in Congress at the time, Albert Johnson, chair of the House Committee on Immigration, vociferously campaigned for a permanent restriction policy. With the upturn of the economy in 1922 though the restrictionist coalition began, temporarily, to unravel. As the economy began to expand and unemployment diminished, industrialists who had acquiesced to restrictions in 1921 reconsidered as wages spiraled. Unskilled labor in particular, in short supply after restriction and the demands for intensive training during the war, commanded higher wages prompting a concerted campaign by heavy industry to block further restrictions. Surprisingly, however, by the following year "industrialists looked on passively, with little complaint, as the politicians proceeded to raise the barriers against European immigration still higher" in the Johnson-Reed Act of 1924.
The 1924 law, in fact, institutionalized the quota system of 1921. A long act, in thirty-two sections, it, in Hutchinson's account:
By the late nineteenth century, opposition to immigration in the U.S. was widespread, yet significant restrictions would await the 1920s. Does the endowment based model help to account for this apparent anomaly? I maintain that it does and suggest that a closer look at the labor market consequences of immigration and the electoral incentives of politicians helps to explain why.
While few econometric studies directly address the impact of immigrants on the earnings of domestic workers prior to 1924, those that do focus narrowly on the question of whether or not immigrants were discriminated against in the labor market. Many immigrants were undoubtedly discriminated against (as were many domestic workers), but in the aggregate, immigrants appear to have fared just as well as natives earning roughly equivalent pay for equivalent work. In fact, the majority of studies "suggest that we treat new immigrants and resident Americans (or given skill) as perfect substitutes in production."
Accordingly, in terms of the model above, the incentive for labor to protest increasing immigration, as it did, until the quotas. The large jump in immigration following the turn of the century intensified the competition between immigrants and natives. While the conventional story of stagnant or declining real wages prior to W.W.I, discernible in Figure 3, has been the subject of much debate, it remains "reasonable that the assimilation of the massive immigration of 1900-1914 should retard the growth of real wages." Following the war, American wage earners feared renewed competition and the concomitant potential loss of income. With the economic downturn of 1920 and unemployment levels at twenty some odd percent in sectors likely to be swamped by incoming new immigrants, the incentives of many domestic workers, unsurprisingly, were to support restriction.
Figure 3. Trends in Relative Wages, 1890-1926.
Sources: Douglas (1930), Tables 24 and 61, deflated with cost of
living index on p.60.
The nativism of the war and economic downturn of 1920, in a sense, resolved the immigration dilemma for the government. The question of which factor dominated politics at the time had, in essence, become moot. For a brief moment, with the fear of hordes of radical European immigrants overcrowding a taut labor market, proponents of restriction fashioned a compromise acceptable to both business and labor. While economic recovery by 1922 threatened the fragile consensus, capital investment during the war allowed industry to thrive without surplus labor. By the time Congress reconvened in 1924, the NAM and other business groups were recommending that the 1921 quotas be retained, not reduced. As Higham notes, "industrialists looked on passively, with little complaint, as the politicians proceeded to raise the barriers against European immigration still higher." As well they could, since restrictions applied only to immigration from Asia and Southern and Eastern Europe, leaving the Western Hemisphere largely unfettered. The 1924 act, while completely satisfying no one, thus managed to meet the needs of both industrialists and laborers. Given that the old immigrants had by the turn of the century become a significant force in the electorate, the restrictions were not a bad deal for politicians either.
Implicit in the narrative of U.S. migration policy above is the notion that the factor market consequences of immigration drive migration policy. Yet aside from intermittent references to the gains and loses to labor, the determinants of restriction remain unclear. Accordingly, I turn to the likely motives for restriction in the receiving state and attempt to empirically assess the variation in U.S. migration policy over time. A simple regression framework provides a useful starting point for analysis. I estimate a migration equation of the form Mt = f(Yrt, Urt, Art) where Mt represents various measures of immigration at time t, Yrt represents a measure of income per capita in the U.S. at time t, Urt represents the unemployment rate, and Art captures additional factors expected to influence the expected costs and benefits of migration. Migration is expected to be a positive function of income and a negative function of unemployment. The role of relative wages, labor's factor reward, assumes particular importance, directly linking the empirical analysis with the trade based model proffered above.
Because my interest is in immigration restriction, I define the dependent variable in terms of changes in immigration levels and, more appropriately, changes in immigration policy. Changes in immigration levels, operationalized as the difference between annual migration flows, arguably captures the intent behind administrative or legislative efforts to curb or promote migration and serves as a rough proxy for policy change. As a more direct gauge, I also employ a dichotomous measure of immigration policy, coded 1 in years in which Congress passes restrictive legislation and 0 otherwise. I then analyze the impact of a variety of economic and political independent variables, such as income, unemployment, volume of immigration, union density, and partisanship, among others (detailed in the Appendix) on the dependent variables for the years 1870-1925 using appropriate regression techniques.
Table 7 presents preliminary results of regression analysis of changes in economic, demographic and political explanatory variables on U.S. immigration rates over the 1870-1926 period. Model 1 investigates the impact of economic variables on the immigration rate. As expected, unemployment is negatively related to the immigration rate. In addition, the measure of labor's factor rewards, the ratio of wages to GDP per capita is significant and of expected sign. When labor's share of income rises, the country appears to be more open to immigration. Aggregate economic conditions, however, while not significant are of the opposite sign, suggesting that contemporaneous growth may not suffice to draw migrants. The impact of trade is also examined as a way to determine if migration and trade are compliments or substitutes. Unfortunately, the result is ambiguous as well.
Model 2 attempts to assess whether the ethnic composition of the immigration stream affects migration policy. The choice of variable, new immigrants as a percent of total immigration, however fails to adequately capture ethnic concerns as its impact simply reflects the underlying increase in immigration during the later part of the 1890-1926 period.
Efforts to incorporate political variables also require additional work. The sign on the House variable, which is coded 1 when Democrats are the majority party, is plausible given the party's early opposition to Chinese immigration. Nonetheless, the finding more likely reflects the Democrat's control of the House following the depression of 1893 and during W.W.I, years in which the immigration rate was quite low to start. The party of the President also appears unrelated to the immigration rate, as do various measures of union density (not reported).
Table 7. Effects of Annual Economic, Demographic, and Political Variables on U.S. Immigration Rates, 1875-1926
(6.65) (3.18) (5.59) (2.69) (5.91) (2.82)
(-1.89) (-2.04) (-3.83) (-1.86) (-3.55) (-2.04)
(-1.35) (-1.82) (-1.19) (-1.47) (-1.33) (-1.14)
(4.58) (2.07) (2.12) (1.82) (2.47) (1.88)
(1.21) (1.24) (1.12) (0.94) (1.33) (0.898)
R2-adj. .457 .243 .444 .289 .367 .242
D.W. 1.257 1.86 0.948 1.61 1.01 1.76
N 55 54 55 54 55 54
Note: Entries for first model specification are unstandardized regression coefficients. Coefficients in the second specification are corrected for serially correlated residuals using the Cochrane-Orcutt iterative process. t-statistics are in parentheses.
*p .10; **p .05.
The quantitative inquiry into the origins of immigration restriction, unfortunately, raises as many questions as it answers. What other variables might underlie restriction? Does partisanship of the legislature matter? Might the U.S. have maintained a liberal immigration policy if Wilson, not Harding, had been president in 1924? Do the same results hold over time and across nations? While I have no answers to these and other countless questions at this point, the factor based approach offers a useful first cut into the politics of immigration restriction.
Upon cursory inspection, in fact, the approach appears to hold great promise in accounting for migration flows in the contemporary U.S. and post-war Western Europe. While the details must await separate treatment, Freeman and Money's recent analyses of the politics of immigration restriction in advanced industrial states are consistent with the factor based argument above. Furthermore, Kindleberger's study of post-war Europe, a recovering region in short supply of labor, emphasizes migration as a key component of growth in the years where migrants and domestic laborers complement each other in the labor market. With the transformation of guest workers into de facto settlers by the late 1960s, however, the competition between foreign and domestic workers intensified. Rising job competition and unemployment following the first oil price shock prompted virtually all European receiving countries to either stop recruiting foreign labor of to sharply increase selectivity and controls on labor migration. The cross-national variation in the strength of unions in Europe should offer some interesting insights into variations in migration policy. The incentives of politicians to court labor's vote and the labor market effects of immigration should explain the turn to restriction in the mid-1970s and today.