How Their Laws Affect our Laws: Mechanisms of Immigration Policy Diffusion in the Americas, 1790–2010

Abstract

Why do laws become similar across countries? Is the adoption of similar laws and policies due to factors operating independently within each country? Do countries develop similar rules in response to similar challenges? Or is the similarity of laws and policies due to the interdependent responses that scholars have referred to as processes of policy convergence, transfer, and diffusion? We draw on an analysis of immigration and nationality laws of 22 countries throughout the Western Hemisphere from 1790 to 2010, and of seven case studies of national and international policymaking, to show that policies are often interdependent, even in the domain of immigration law, which scholars have presumed to be relatively immune to external influence. We argue that specific mechanisms of diffusion explain the rise of racist immigration policies in the Americas, their subsequent decline, and the rise of an anti-discriminatory norm for policies. Most striking among our findings is that at key junctures after 1940, weaker countries effectively advanced an anti-discriminatory policy agenda against the desires of world powers. We identify the conditions under which weaker countries were able to reach their goals despite opposition from world powers.

Scholars of policy convergence, transfer, and diffusion attempt to explain why laws become similar across countries. Immigration and nationality laws present a “hard case” to explain because these laws express a nation-state's sovereign ability to define its population and are presumed to be insulated from external influences. We draw on an analysis of immigration and nationality laws of 22 countries throughout the Western Hemisphere from 1790 to 2010, and of seven case studies of national and international policymaking, to show that even in this jealously guarded sovereign domain, policymaking is often interdependent. Three distinct mechanisms of diffusion explain the rise of racist immigration policies in the Americas, their subsequent decline, and the rise of an anti-discriminatory norm. Surprisingly, we find that weaker countries in the international system of states can exert leverage over the domestic laws of more powerful countries by collectively linking their emigration agenda to the core geo-political interests of dominant countries.

Reports of a “Muslim ban” in the United States, draconian border policies against refugees trying to enter Hungary, the closing of migrant camps in the French port of Calais, and Brexit-related animosity toward foreigners in the United Kingdom suggest a common turn toward restriction in the United States and Western Europe. What is known about how and why the adoption of restrictive policies in one country affects the spread of similar policies elsewhere? Are policy makers responding independently to similar challenges like migrant flows from neighboring regions or global economic forces? Or are governments' responses to migration conditioned by migration policies of other countries? Answers to such questions have immediate relevance to how we understand the rise and spread of different types of policy, including immigration ones that select by origin. A comparative and historical perspective offers some valuable lessons. In this article, we take such an approach and explore the spread of ethnic selection in the immigration and nationality policies of the Americas since the eighteenth century.

Hannah Arendt (1973: 278) noted that sovereignty has nowhere been more absolute than “in matters of emigration, naturalization, nationality, and expulsion.” The core principle of the nation-state is that its members belong to a particular group of people on a particular territory controlled by a particular government. As a consequence, the movement of people across jurisdictions poses a fundamental challenge to nation-states. A vast body of literature describes the struggles within each state over how to realize a particular vision of the nation by defining rules of admission and citizenship (Brubaker 1990; Castles and Davidson 2000; Joppke 1998, 2005; Massey 1999; Portes and Rumbaut 2014; Putnam 2014; Waldinger 2017; Zolberg 2012). International treaties concerning immigration and nationality law show that diplomats tread lightly around the sovereignty of nation-states to make their own rules. For this reason, immigration and nationality policy fields are strategic sites to examine because they represent “hard cases” of why policies become more similar across countries.¹ Are likenesses attributable solely to factors operating independently within a particular state's territory, such as when countries develop similar rules in response to similar challenges? Or is the similarity of laws and policies due to the interdependent responses that scholars have referred to as processes of policy “convergence” (Busch and Jörgens 2005), “transfer” (Dolowitz and Marsh 2000), and “diffusion” (Dobbin et al. 2007)?

We draw on a quantitative analysis of immigration and nationality laws of 22 countries throughout the Western Hemisphere from 1790 to 2010, and of seven qualitative case studies of national and international policymaking, to argue that countries adopt laws interdependently even in domains presumed to be sovereign. We chose these countries not because processes of diffusion were limited to the Western Hemisphere, but because policy makers in this region experimented extensively with means of selecting immigrants and citizens by ethnicity, and because these states adopted many similar policies despite variation by type of political regime (e.g., democratic, corporatist, oligarchic, and socialist).² We selected a subset of these countries and intergovernmental organizations for case study on the grounds of substantive and theoretical importance. Through a combination of quantitative and qualitative analysis, we find evidence of interdependence in policy adoption. In particular, we find extensive evidence that the adoption of blatantly racist immigrant selection criteria in one country increased the likelihood of the adoption of such criteria in the policies of other countries. Surprisingly, diffusion can constrain the policies of powerful countries when weaker countries ally and link to the geo-political interests of dominant countries a common reputational interest in protecting their emigrants abroad.

We also argue that specific mechanisms of diffusion explain the rise of racist immigration policies in the Americas, their subsequent decline, and the rise of an anti-discriminatory norm for policies. Most salient among our findings is that at key junctures after 1940, weaker countries effectively advanced an anti-discriminatory policy agenda against the desires of world powers. We specify when three distinct mechanisms of diffusion are at work and the conditions under which weaker countries exert leverage on more powerful ones. Our goal is not to assign a specific weight to the importance of diffusion relative to other underlying causal sources of policy like economic factors or domestic interest group politics, but rather to show that diffusion was a consequential source of change in domestic policies, and that diffusion followed different pathways about which we offer transportable lessons for researchers to test in other contexts.

In addition to offering a framework to test in other settings, our arguments and findings have significant implications for the analysis of the supposed global turn toward restrictive immigration policies.³ Will there be a return of racialized immigration law in addition to racist rhetoric and de facto ethnic selection? Do countries of emigration offer any recourse against ethnoracial selectivity that works against their citizens? We maintain that a return to blatant ethnic selection in immigration law is unlikely, but not impossible, because of the institutionalized constraints inherent in belonging to an international system of states. We also identify the circumstances under which such a return could happen. The conditions under which countries of origin have leverage over more powerful countries of destination—which we outline in this article—will shape to what extent countries of immigration adopt ethnically discriminatory policies and whether they do so openly or by subterfuge.

Why Are Policies Similar across Countries?

Theorists of why policies look similar take either a bounded or a systemic approach. Bounded approaches look for explanatory factors within a unit of analysis like the nation-state or compare several such units of analysis. Single-case national studies and comparative national studies look within the boundaries of the unit(s) of analysis to explain the formation of policies. These inward-looking approaches have considerable utility and are necessary to explain policy changes, which may wholly emanate from within a given jurisdiction. A rich tradition of studying immigration policy in the United States, for example, explains change over time as the product of struggles among labor and capital, other domestic interest groups, state incumbents, and competing ideologies (Freeman 1995; Higham 1994; Smith 1997; Tichenor 2002). These accounts may refer to foreign policy considerations, but typically do not ask whether immigration policies in other countries shape immigration policies at home. For example, Tichenor (2002) fully recognizes that foreign policy interests may generate pressure for immigration reform, but his authoritative historical analysis of U.S. immigration policy centers on factors within the United States rather than on how U.S. immigration policy interacts with the migration policies of other countries.

Comparative national studies seek to explain why policies vary across countries. They typically adopt the Millian method of agreement, in which similar outcomes across cases are attributed to similar conditions in each case, or the method of difference, in which different outcomes across cases are ascribed to different causal conditions in those cases. For example, Brubaker's (1992) classic study examined why nationality was based on jus soli in France and jus sanguinis in Germany, a difference he attributed to the French model of nationhood that was state-centered and assimilationist while the German version was more people-centered and differentialist. As with national case studies, comparative national studies typically look within the boundaries of each country to explain policy outcomes. Bounded approaches, of both the national case study and comparative national studies variety, have been criticized for not considering how each country's policies influence the other (FitzGerald 2012; Weil 2008).

One explanation of why policies might look similar across different cases is that each country's policy makers have independently arrived at a similar solution to a similar problem—a process known as parallel path development (Hansen and Weil 2001). Parallel path development is not a type of diffusion as we describe it below, but rather an alternative account of similarity in which causal processes in each unit of analysis are independent of each other. The possibility of parallel path development, rather than diffusion, can only be ruled out by using qualitative evidence to demonstrate that policy makers were reacting solely to internal processes or reacting without coordination to a common exogenous cause. Parallel path development and diffusion are mutually exclusive reasons why policy looks similar across countries.

By contrast, some sociologists and political scientists have developed explanations of policy similarity that take a systemic approach. This analytical approach examines explanatory factors by taking into account that states are located on a political field of other states and networks that cross state boundaries. From this political field perspective, factors within national political units may interact with others in the international system of states (Cook-Martín 2013). Systemic theorists may approach policy similarity in terms of policy convergence, transfer, or diffusion.

Convergence

Scholars of policy convergence study why policies sometimes become more similar or the same across units. The most notable form of policy convergence results in the nearly homogenous institutions of the “world polity,” in which templates of perceived modernity spread from the West to the rest (Strang and Meyer 1993). The contemporary political map of the planet is carved up into nation-states, rather than tribes, empires, or many other possible forms of government (Wimmer and Feinstein 2010). Meyer et al. (1997) has pointed out that, with minor exceptions, each of these nation-states has a standard template of symbols and institutions such as a rectangular flag, national currency, ministry of education, vehicle code, and so forth, which a newly independent nation-state adopts almost overnight. Convergence includes both superficial and core institutions.

Constitutions are another standard template of nation-states that have circulated among countries. Hirschl's (2004) work on juristocracy examines the mutual constitutional influence among Canada, Israel, New Zealand, and South Africa. International organizations and experts are carriers of theories about how the world should work, and these are propelled by the ideological power ascribed to them by political actors (Dobbin et al. 2007). Other explanations for convergence include coercive imposition by powerful states, uncoordinated modeling, and cooperative harmonization in which governments agree to align their policies (Bennett 1991; Busch and Jörgens 2005). Harmonization occurs extensively in the supranational institutions of the European Union (EU) due to intergovernmental decisions to adopt the same policy or their imposition by supranational organs such as the European Court of Justice. Within the EU, extensive harmonization also happens informally through expert networks and modeling on policies of other EU member states (Boswell and Geddes 2010; Lavenex 2001, 2014; Luedtke 2009).

Clustering and Transfer

Lawmaking is often influenced by laws elsewhere without a uniform outcome or even a converging trend, particularly in the absence of formal institutions to harmonize law. Elkins and Simmons (2005) note the temporal and spatial “clustering” of policies, which requires explanations of why they cluster without converging on one outcome. Dolowitz and Marsh (2000) similarly seek to explain the processes of “policy transfer” between units that do not necessarily imply convergence across all units. Sometimes laws are literally photocopied and transferred, complete with typographical errors. However, Peck and Theodore (2010) point out that laws usually are not transferred as a whole, but rather reworked in new settings as they spread. The transfer model also leaves out mechanisms by which laws “there” influence laws “here.” Knill (2005) notes that transfer studies typically focus on case studies of why a particular country adopted a law created elsewhere, rather than establishing the broad patterns that scholars of diffusion analyze.

Diffusion

Scholars of diffusion describe and explain a set of processes through which the adoption of a policy in one country increases the prospects of its adoption in another country (Strang 1991). Although authors define diffusion in a number of ways, they share several basic premises. Like scholars of convergence and transfer, diffusionists take a systemic approach and conceive of the site of social action as a political field on which states or other organizational units interact, rather than focusing on nationally bounded processes. However, compared to scholars of policy convergence and policy transfer, scholars of diffusion typically analyze a broader set of outcomes, including clustering as well as convergence and homogenous isomorphism, and a broader set of processes, such as the “intermestic” interaction between domestic and international developments (Peck and Theodore 2010).⁴ Notable exceptions to this assessment include the work of Dixon and Posner (2011) that examines constitutional convergence theories and Drezner (2005) that offers a general model of policy convergence. Jordan et al. (2003) identify the competitive arena in which ideas about new environmental policy instruments transfer from one political jurisdiction to another within the EU, and examine if transfer is driven by policy expert organizations or by market and harmonization pressures. In this study, we take a diffusionist perspective to examine causal factors in the interdependent relations among sovereign states.

Mechanisms of Legal Diffusion

Diffusion takes place through several distinct mechanisms. To understand how and under what conditions immigration and nationality policies are affected by diffusion, we focus on three different mechanisms—emulation, strategic adjustment, and leverage—that are characteristic of the immigration policy arena.

Emulation refers to policy makers in one country voluntarily modeling their policies on those of another country or institution. Constructivist accounts of international relations are sensitive to how norms spread without coercion (Arend 1997). Countries intentionally emulate one another when one of them represents a template of modernity or offers a “best practice” in a particular domain (Dobbin et al. 2007; Gilardi 2010). Intergovernmental organizations such as the United Nations have often played an important role in processes of global diffusion (Torfason and Ingram 2010), as have “elite networking” (Bennett 1991) and “epistemic communities” of experts that wield influence (Haas 1992). “Norm entrepreneurs” promote new policies through such channels (Dolowitz and Marsh 2000). These organizations, networks, and communities are well-known vehicles for diffusion by emulating the norms of the powerful, but as we show in what follows, under certain conditions, they also are sources of leverage from below.

Strategic adjustment occurs when actual or anticipated changes in the policies of other countries push a government to adapt accordingly. This mechanism emphasizes new elaborations of policy based on observable or anticipated changes to the status quo. Unlike the mechanism of emulation discussed above, in the strategic adjustment mechanism, policy makers in country B do not view policies in country A as a model. Rather, policy makers in B consider how A's policies change the conditions that B aims to manage. If emulation is about changing ideas, strategic adjustment is about reactions to changing incentives (Dobbin et al. 2007). In this mechanism, states are not applying pressure on each other with the intent of shaping policy elsewhere, but in a system of interacting states, the actions of one can influence the decisions of others (see Elkins and Simmons 2005).

Leverage refers to diplomatic, military, or economic pressure that one country puts on another to change its policies. It can work on both sides of a fulcrum such that even weak actors can exert effective pressure to achieve their goals against the wishes of more powerful actors, as we show in our findings. Following Dolowitz and Marsh (2000), we conceive of leverage as taking place on a continuum of coercion from military to diplomatic action.⁵ Under certain circumstances, leverage is the mechanism that allows less powerful actors to get their way against more powerful ones.

Conditions for Diffusion

We examined two major propositions about the conditions for diffusion and its direction. A first proposition, which we refer to as the unidirectional power assumption, is that countries with greater power and higher status create policies that weaker or lower status countries follow. The direction of influence runs from the more to less powerful, and over the last several centuries, from North to South, and West to East (Dobbin et al. 2007: 452). The neo-institutionalist literature suggests that diffusion cannot be explained solely by power or competitive advantage and that membership in an organizational field constrains individual organizations. However, organizational theory also strongly implies that some organizations adopt change because they depend on, and are culturally expected to follow, other leading organizations (DiMaggio and Powell 1983: 150; see also Haveman 1993). In the organizational field composed of states, the most powerful and high-status countries are frequently the policy leaders.

Based on the unidirectional power assumption, one would expect to find evidence that the United States was the policy leader in the Americas because of its status as an exemplar of modern nationhood since the mid-nineteenth century. By means of the mechanisms described earlier, policies would spread from the United States to other countries in the hemisphere.

A second assumption in the diffusion literature is that geographic or cultural proximity increases the likelihood that countries will adopt similar policies. Kopstein and Reilly (2000; 18) suggest, “spatial proximity permits a more extensive level of diffusion, which in turn, exercises a strong and independent effect on political and economic outcomes.” Similarly, Wejnert (2005), Weyland (2005), and Wimmer and Feinstein (2010) argue that geographic proximity to policy exemplars accelerates the speed of policy diffusion. Based on the geographic assumption, one would expect to find that neighboring or spatially proximate countries influenced each other's policies more than distant exemplars across the northern and southern hemispheres of the Americas. Policies would also diffuse among culturally similar countries that share “psychological proximity” (Rose 1993) regardless of the degree of spatial proximity. Simmons and Elkins (2004): 175–76) maintain “cultural similarity will be a positive predictor of policy diffusion among states.” On this view, English-speaking settler states should form one cluster of similar laws, while Spanish American states form another, because they are distinct communities with different (common vs. positivist) traditions of law, languages, and colonial backgrounds.

Methods

Our methodological strategy proceeds in two steps. First, we establish the broad patterns of ethnic selection policies by coding laws governing immigrant admissions and nationality from the year of a country's independence to 2010 in 22 countries of the Americas. This analysis shows which and how many countries had ethnically selective laws and of which kind, when they had these types of laws, and how long periods of selection lasted. The coding identifies patterns of clustering, convergence, or isomorphism. The coding of law does not, however, explain why ethnically selective policies occurred where and when they did, or whether processes of parallel development or diffusion shaped the policies. In a second step—on which we focus the bulk of our analysis in this article—qualitative case studies of six key countries in the Americas and of international organizations provide fine-grained evidence for when convergence or diffusion took place, applicable mechanisms, and the conditions under which each mechanism shaped policies. Our analysis demonstrates the powerful methodological combination of an exhaustive historical mapping of policy patterns with case studies that specify causal pathways.

Coding of Law in 22 Countries of the Americas, 1790–2010

We examine countries in the Americas because policy makers in this region experimented for decades with different strategies to shape their populations, including proactively recruiting immigrants deemed desirable and barring those who were not. The countries in our sample vary by colonial history (British, French, Spanish, and Portuguese), type of political regime, economic base, geographic proximity to core centers of power, and level of engagement with regional and global organizations. Despite these differences, we find clear patterns of similar immigration policies. The focus on the Americas does not mean that countries in other regions were not experimenting with such policies. On the contrary, we find that diffusion of ideas about immigration and ethnic selection linked countries in the Americas in different ways to Australia, South Africa, Japan, China, India, and Europe. As a practical matter, however, we restricted the quantitative analysis to a sample that is still unprecedented in the immigration policy literature for its historical span (more than two centuries) and breadth across countries. The sample includes all 22 major countries in the Western Hemisphere—Argentina, Bolivia, Brazil, Canada, Chile, Colombia, Costa Rica, Cuba, Dominican Republic, Ecuador, El Salvador, Guatemala, Haiti, Honduras, Mexico, Nicaragua, Panama, Paraguay, Peru, United States, Uruguay, and Venezuela. It does not include the 14 micro-states of the Caribbean Basin that gained independence after World War II, because they were not sovereign for most of the study's period.

To establish policy patterns, we first constructed a corpus of laws, regulations, court decisions, legislative debates, circulars, and other relevant official documents in which we could observe the logic by which state actors selected prospective immigrants and citizens. To build this corpus, we used indexes of legislation and executive orders in government bulletins, legal guides for each country, references to laws no longer in force, the regulations and debates related to relevant laws that often referred to other legal sources, and sources identified in the vast historiographical materials on the region.

We then selected all legal materials with ethnic selections from a corpus of 49,467 pdf-pages and systematically coded laws pertaining to immigration policy (the selection of who can enter a country and stay) and nationality laws passed at independence regulating who is a national and subsequent laws of naturalization.⁶ For reasons of feasibility, and following Hammar's (1989) typology of policy domains, we did not code immigrant policy that regulates the rights of immigrants vis-à-vis other citizens and how to integrate newcomers. Specifically, we coded for the presence or absence of positive preferences for, or negative discriminations against, 23 ethnic groups. Positive preferences refer to affirmative measures to foster immigration or facilitate naturalization among particular groups, frequently Northwestern Europeans, and included such policies as assisted passage, free land, higher immigration quotas, or exemptions from requirements enforced against other groups. Negative discriminations refer to measures to limit or preclude the immigration or citizenship of members of defined ethnic groups. These measures included outright bans on entry and citizenship, lower immigration quotas, or special entry taxes. We do not use “positive” and “negative” in a normative sense, but rather to capture the logic of particular tactics. In both instances, they were part of a larger strategy of shaping national populations on ethnoracial grounds.

For every country in our sample and for every year of the period studied, we coded for the legal selection of groups such as Spaniards, Jews, Chinese, African/blacks, Roma, and various groups of Europeans (see the Appendix for a list of codes). For example, the dataset contains a cell with a dichotomous variable for whether or not there was a positive preference in nationality law for Japanese in Paraguay in 1897. Another cell shows whether there was negative discrimination against Japanese in Paraguayan immigrant admissions law in 1897. The coding is repeated for each of the 23 ethnic groups for each year. For each country-year, aggregated data show whether there was a preference or discrimination against any ethnic group, which is the source of the data in Figure 1.

Figure 1. Number of Countries with Negative Immigration Discrimination against Select Ethnic Groups, 1790–2010. Source: Dataset on File with Authors.

Some laws categorized potential immigrants—particularly blacks and Asians—in strictly racial terms, in the sense of groups defined by phenotype and/or notions of immutable biological characteristics. Other laws based their categories on legal nationality or country of birth, or distinguished groups by their language, religion, or culture. We use “ethnicity” as an umbrella analytic term for all of these forms of categorization. To maintain consistency across cases, we coded only those laws that explicitly name an ethnic group, and those that were publicly available at the time that they were enacted, such as constitutions, statutes, published regulations of immigration and nationality, published bilateral and multilateral treaties, and court cases. In common law countries such as the United States and Canada, court cases were critical sources of selection law, particularly when it came to defining the racial boundaries of whiteness in the United States and thus which groups were eligible to naturalize. The relevant U.S. and Canadian court cases have been identified by previous scholars. The courts were not as relevant in this regard in Latin America, which has a Napoleonic tradition of positive statutory law.

Qualitative Cases Studies

We chose countries as case studies on the grounds of theoretical and substantive importance (Cook-Martín and FitzGerald 2010). Five of the cases in this study—the United States, Argentina, Canada, Brazil, and Cuba—received 92 percent of transoceanic European immigration in the period before World War II when policies of ethnic selection were first enacted. During the same period, up to 2.5 million Asians migrated across the Pacific. The United States, Cuba, Mexico, Peru, and Canada were the major destinations for 1.5 million Chinese. More than 600,000 Japanese migrated primarily to Brazil, Hawaii, the U.S. mainland, Canada, and Peru. While immigration to most of Latin America has fallen since the 1930s, flows of large and ethnically diverse migrations have continued to Argentina and Brazil in particular, and immigration to North America rebounded after World War II. The Western Hemisphere has been the destination of roughly a quarter of all international migrants since 1960, with most going to the United States. The case of Mexico tests whether ethnic restriction is linked to the magnitude of immigration. The Mexican government constructed an extensive system for ethnically selecting immigrants even though its immigrant population never surpassed 1 percent of the population.⁷

The six countries vary in power, and in geographic and cultural proximity to each other, allowing us to test assumptions about the directionality of diffusion by degree of power and cultural and geographical distances. The case selection strategy also enables us to trace different mechanisms for the diffusion of policies via emulation, strategic adjustment, and leverage.

A case study of international organizations reveals how ideas about immigration policy spread and the direction in which they diffused. These organizations were especially important for institutionalizing norms promoted by governments and quasi-official scientific experts acting as norm entrepreneurs (see Colyvas and Jonsson 2011). Intergovernmental organizations key to the immigration and nationality policy arena include the League of Nations, the International Labour Organization (ILO), the Pan-American Union, and the United Nations. Networks of experts like those who gathered at eugenics conferences were linked to government either because they served in official posts or made influential policy recommendations. We included the organizations most frequently cited in legislative policy debates about how to choose immigrants, although other expert networks also were interested in immigrant selection and the categorization of populations by race (see Loveman 2014).

To build on findings from our coding of formal law, we systematically analyzed materials related to each of our cases: transcripts of legislative debates and nongovernmental conference proceedings, public and secret internal memos, administrative decrees, and correspondence among policy experts and scientists. We coded materials from the United States, Argentina, Canada, Brazil, Cuba, and Mexico. We also coded materials from international organizations and triangulated our analysis with a close reading of the extensive secondary literature about immigration and nationality policy in our six country studies as well as the organizations and expert networks examined below.⁸

Our analysis yielded a detailed picture of the dynamics of policy development, including whether diffusion took place, its mechanisms, and the conditions under which specific types of mechanisms operated. Our understanding of the sources of policy did not a priori privilege the importance of either diffusion or internal factors. We used qualitative research software to code primary materials systematically for evidence of the motivations and social sources of the law. Specifically, we coded these materials for political and economic policy rationales, type of political regime, ostensibly neutral but de facto discriminatory policies, reference to foreign influences as well as foreign policy considerations, invocations of scientific expertise, and a number of other in vivo codes that emerged from reading the documents. For instance, our analysis of legal texts and conference proceedings yielded the in vivo category of assimilability—the perceived ability to integrate—which policy makers inferred from categorical group membership (FitzGerald et al. 2018). We follow Bennett's (1991: 224) prescription that “confirmation of the emulation hypothesis requires the satisfaction of a number of conditions: a clear exemplar (a state that has adopted an innovative stance); evidence of awareness and utilization of policy evidence from that exemplar; and a similarity in the goals, content or instruments of public policy.” We apply the same techniques of process tracing to identify mechanisms of strategic adjustment and leverage while drawing on the quantitative data to establish broad patterns of change.⁹

Findings and Discussion

The systematic coding of ethnic selection in immigration and nationality laws of the Americas shows convergence in two periods. From the late nineteenth century to about 1940, states increasingly adopted policies of explicit discrimination against particular ethnic groups and preferences for others. From 1940 until the turn of the twenty-first century, states eliminated their explicit negative ethnic discriminations. Convergence in policy patterns, however, is not enough to show that diffusion occurred. To make a case for diffusion, evidence would show that the adoption of a particular policy by one country increased the likelihood that other countries followed suit. Below we briefly review overall patterns of clustering and convergence in immigration and nationality policies in the Americas before and after 1940. We then examine evidence collected from case studies to argue that diffusion shaped key shifts in immigration law in sometimes surprising ways.

Pattern before 1940: Convergence on Negative Racial Discrimination

An analysis of immigrant admissions laws in the 22 countries shows that they converged on racist criteria for selecting prospective immigrants by the 1930s. This convergence happened despite differing levels of immigration, source countries, political systems, labor markets, and population size. Every country in the Americas passed some form of discriminatory immigration law after 1880—shortly after the publication of the 1874 Cuba Commission Report denouncing the Chinese “coolie” trade. Chinese exclusions were the most common in the hemisphere. The other most commonly targeted groups were Japanese, Roma (gitanos in Spanish-speaking countries, ciganos in Brazil), blacks, and Middle Eastern immigrants (see Figure 1). The broad pattern here is the clustering of discrimination against particular groups and a convergence in the principle of practicing some form of negative discrimination in immigrant admissions.

Nationality law shows clustering rather than complete convergence. Shortly after U.S. independence, a 1790 law reserved naturalization to free whites, and it specifically banned Chinese naturalization in 1882, although the foundational naturalization law already implied discrimination against all non-white groups. The only other country in the Americas to racialize nationality law to the same extent was Haiti, which after its slave-led revolution against the French, in 1816, banned the naturalization of whites and gave citizenship to any black or Amerindian who came to Haiti. The other countries with negative discrimination in their nationality law were Canada, Costa Rica, and Panama. There was greater convergence in positive preferences. Of the 22 countries in this study—18 of them were Latin American—all but Uruguay had positive ethnic preferences in their nationality law.¹⁰

Pattern after 1940: Convergence on Facially Neutral Immigration Policies

The analysis of immigration admission laws also shows that countries moved away from racially selective criteria after 1940 (see Figure 1). Eighteen countries in the Americas had laws with negative discrimination against Chinese immigrants in 1936; but the number dropped to 11 by 1945, 6 by 1955, and 2 by 1968. Our data show similar trends—although of smaller magnitude—for Japanese, Middle Eastern, black, “gitano,” and “unassimilable” immigrants thought to be incapable of becoming part of the national society. Canada and 11 Latin American countries developed different kinds of assimilability provisions, which in practice favored Western Europeans and discriminated against Middle Easterners and Asians. This subtle form of selection became widespread between the mid-1920s and through 1980, but eventually disappeared.

There is no uniform pattern of convergence in ethnic selection in naturalization law, but rather a cluster of countries in Latin America that retained preferences for different configurations of Ibero-American nationalities, while others in the hemisphere were ethnically neutral. In 2010, 16 countries in Latin America retained naturalization preferences for Spaniards, 10 for Latin Americans, and 3 for Portuguese.

The broad patterns in positive and negative selection of immigrants and nationals outlined above thus provide evidence of convergence and clustering, but not necessarily of diffusion. In the next section, we offer evidence of diffusion of immigration selection techniques in each of the periods studied.

Case Study Evidence: How Diffusion Shaped Immigration and Nationality Laws

Immigration and nationality laws in the Americas before and after World War II are more similar than one would expect if the determinants of these laws resulted from independent processes unfolding in countries with significant cultural, political, and demographic differences. Our case studies reveal the specific mechanisms through which diffusion affected policy and allow us to challenge conventional notions about how geographic and cultural proximity shape diffusion and how power affects the direction of influence. We provide representative evidence in support of our argument from those case studies to show that strategic adjustment and cultural emulation were the main mechanisms of diffusion before 1940, and leverage from below was an understudied but critical mechanism that led to the move away from racially selective immigration policy after 1940. International organizations and norm entrepreneurs played a critical role in both periods, but most surprisingly, they have been the means through which weaker countries cooperatively pressed anti-discriminatory immigration policy against resistance by stronger countries that created those very institutions during and after World War II.

Cultural Emulation and Strategic Adjustment before 1940: Diffusion across Culture and Distance

Geographic proximity did not strongly influence legal ethnic selection of immigrants in the Americas, unlike in other policy domains such as pension reform (Weyland 2005). Governments routinely adopted practices from countries thousands of miles away within the Americas and even from countries on the other side of the globe through the mechanism of cultural emulation. This mechanism was frequently an iterative process (see Dolowitz and Marsh 2000: 6). The diffusion of literacy requirements aimed at restricting southern and eastern European immigration to the United States and Asian immigration to Australia, Canada, New Zealand, and South Africa illustrates the iterative mode of emulation, and suggests the cultural and geographic conditions under which it takes place. The Natal Act (1897) required immigrants to the southern African colony to speak a European language. It was a modified version of the U.S. bill passed by Congress in 1897 requiring intending immigrants to prove their literacy in any language. Natal's Prime Minister Harry Escombe explicitly invoked the U.S. literacy bill when he urged the Natal assembly to pass the Act in March 1897 and attributed its origins to the “American Act” (Lake and Reynolds 2008: 130).¹¹

Although U.S. President Grover Cleveland vetoed the U.S. literacy requirement bill, the British government successfully promoted the “Natal formula” throughout its empire, which resulted in literacy acts in New South Wales, Western Australia, Tasmania, and New Zealand in 1898; Australia at federation in 1901; and the Union of South Africa in 1910 (Huttenback 1976; Martens 2006). The U.S. Congress finally passed a literacy test in 1917 over a presidential veto. The Canadian parliament then enacted a literacy provision in its 1919 amendments to the Immigration Act that was also explicitly linked to its U.S. precursor (McLean 2004). Intending immigrants to Canada could satisfy the requirement by demonstrating literacy in any language, thus reflecting the provisions of the 1917 U.S. law, rather than the versions in the other dominions that specified European languages only. Despite the great geographic distances between South Africa, Australia, and North America, anglophone countries openly emulated versions of each other's policies even as they were instantiated with minor national variations.

The mechanism of strategic adjustment also operated over great distances and across cultural divides. The magnitude of immigration to the United States meant that other countries, even if they were thousands of miles away, generally reacted to how U.S. policy shaped transoceanic migration (Timmer and Williamson 1998: 754). To compete with the United States in attracting desirable immigrants, countries such as Argentina developed more aggressive recruiting campaigns and positive ethnic preferences for Europeans, such as a constitutional mandate to “foster European immigration” (Argentine Constitution of 1853, Art. 25). On the negative side, many governments imposed ethnic discrimination because they feared that U.S. bans on Chinese labor migration in the nineteenth century and restrictions of southern Europeans in the 1920s would redirect those groups to other ports. Figure 1 shows the rapid growth of discrimination against Chinese around the Americas after the United States restricted Chinese indentured servant “coolie” migration in 1864 and all Chinese labor migration in 1882. The 1907 “Gentlemen's Agreement” between the U.S. and Japanese governments to restrict Japanese migration to the United States redirected Japanese migrant flows to Canada and Brazil. Canada almost immediately imposed restrictions modeled on the Gentleman's Agreement. Policy makers abroad sometimes even preemptively changed their ethnic selection policies in anticipation that proposed changes in the United States would redirect immigration flows. For example, in 1862, Costa Rica passed a law banning black immigration, because its leaders feared a U.S. plan backed by President Lincoln to send U.S. blacks to Central America (Kanstroom 2007: 88–89).

Intergovernmental organizations and networks of experts have bridged both cultural and geographic divides. Regional conferences of scientists, medical professionals, and policy makers, such as the Conferences on Eugenics and Homiculture of the American Republics, recommended racial selection of immigrants beginning in the 1920s. Delegates from 15 Latin American countries and the United States met in Havana (in 1927) and recommended, “American nations will draft and apply immigration laws designed to prevent the entry in their territories of the representatives of races whose association is considered biologically undesirable” (Primera Conferencia 1928: I, 163).

Intersecting networks of experts shared ideas about the scientific legitimacy and forms of racial selection. Eugenicist Harry Laughlin's congressional testimony on eugenics helped shape the U.S. quota system in 1924 (Tichenor 2002). He and his Cuban colleague, Domingo Ramos, advanced their immigration policy proposals in the eugenics conferences sponsored by the Pan-American Union in the late 1920s and 1930s. Ramos then invited Laughlin to Cuba to shape a proposed new Cuban immigration law in the 1930s (Primera Conferencia 1928; Conferencia Panamericana de Eugenesia y Homicultura de las Repúblicas Americanas 1934; The Harry H. Laughlin Papers, Truman State University, Box C-4-1:7). Non-state actors affected policy across multiple countries through their membership in these epistemic communities of transnational experts.

The work of transnational norm entrepreneurs was instantiated in national laws. A wave of countries in the 1930s adopted the language of ethnoracial purity. Mexico's 1930 Law of Migration, whose preamble stated that it was based on the recommendations of international conventions such as the migration conferences organized by the ILO, limited immigration to those “belonging to races, that, because of their conditions, are easily assimilable to our environment, with benefit for the species” (Preámbulo, Proyecto de Ley de Migración, 25 enero 1930, SRE IV-395-17). Nicaragua banned immigrants who were “dangerous for the existing social order” because of their ethnicity (La Gaceta no. 117 y 118 of May 30–31, 1930). Article 121 of Brazil's 1934 Constitution limited immigration in the interests of guaranteeing “ethnic integration.” Guatemala banned immigrants whose race would make them undesirable in the Law of January 25, 1936, and Peru's Supreme Decree of June 26, 1936 introduced immigration quotas that would safeguard the ethnic patrimony of the nation. Finally, Bolivia issued the Supreme Decree of January 28, 1937 regulating immigration in the interest of the “ethnic betterment” of the country.

In sum, international organizations and scientific elites helped spread policies across great geographic distances and cultural divides. Strategic adjustment to changes in immigration patterns caused by early adopters of discriminatory policies, most importantly the United States, further contributed to similarities in the pattern of ethnoracial discrimination in the selection of immigrants around the Americas before 1940.

Leverage from Below after 1940: The Rise and Consolidation of an Antiracist Norm

Can laws diffuse from less to more powerful countries in the international system of states? We argue they can when two conditions apply to the asymmetric relationships among countries. First, less powerful countries must have the organizational means and capacity to advance policy models that challenge those adopted by more powerful countries. Second, political opportunities must exist that allow less powerful countries to link proposed policy models to core strategic interests of more powerful countries. The diffusion of anti-discriminatory models of immigration and nationality policy ran counter to policies supported by the major world powers during and after World War II because these two conditions applied.

Accumulating the Organizational Means to Resist Powerful Countries

Why and how did Latin American governments build the capacity to use leverage against racist immigration norms? Latin American leaders had resented humiliations by the heavy-handed racism of U.S. policy makers since the turn of the twentieth century (Appelbaum et al. 2003; Loveman 2010, 2014). Prominent U.S. policy makers openly treated Latin Americans as inferiors and threatened to include them in the U.S. national-origins quotas enacted in the early 1920s from which independent Western Hemisphere countries had been exempted. In 1926, Rep. John Box (D-TX) introduced an unsuccessful bill to include Mexico and other Western Hemisphere countries in the quota system because the influx of Mexicans created “the most insidious and general mixture of white, Indian, and Negro blood strains ever produced in America” (70 Congressional Record, S2817–2818; Feb. 9, 1928). Eugenics expert Harry Laughlin testified to the House Immigration Committee that immigration from the Western Hemisphere should be restricted to whites, a position echoed by Princeton economist Robert Foerster (1925) in his report to the Secretary of Labor. At a Pan American conference, Laughlin and his boss, Charles Davenport, threatened to lobby against the Western Hemisphere exemption unless Latin American governments adopted a hemispheric eugenics code (Primera Conferencia 1928). United States public health agencies, patriotic societies, and the American Federation of Labor also supported restriction of Mexicans. Latin American governments in turn felt humiliated by U.S. treatment of Mexicans in the Southwest, Cubans in Florida, and Central Americans and Caribbean islanders in the racially segregated Panama Canal Zone (FitzGerald and Cook-Martín 2014).

To be sure, Latin American governments had emulated some of the racial exclusions of immigrants in the United States and created their own forms of exclusion, but they found it hard to accept the humiliation of U.S. racism directed at Latin Americans on a world stage and took policy steps in response (FitzGerald and Cook-Martín 2014; Stepan 1991).

By 1938, Latin American countries publically united against overt racial discrimination in law. The Eighth International Conference of American States in Lima (1938) recommended that member states:

…coordinate and adopt provisions concerning immigration, wherein no discrimination based on nationality, creed or race shall be made, inasmuch as such discrimination is contrary to the ideal of fraternity, peace and concord which they undertake to uphold without prejudice to each nation's domestic legislation. (8^th Pan American Conference, Resolution XLV, 1939: 268)

In the wake of the conference, countries such as Chile, Uruguay, and Paraguay removed their explicit discrimination against particular racial groups, followed by others such as Cuba, Argentina, and Mexico in the 1940s (recall Figure 1). National laws adopted collective statements against using race in immigration policy. Mexico's 1947 General Law of Population ended its earlier national quota system based on policy makers' notions of which groups represented an “undesirable race” (Yankelevich and Chenillo Alazraki 2009). The preamble of the 1947 law cited the conclusions and “promises made by Mexico” in the First Inter-American Demographic Congress held in Mexico City in 1943. Mexico and all of the other 22 countries involved, with the exception of Canada, explicitly rejected racial discrimination, denounced the doctrine of racial superiority as unscientific, and appealed to national governments to improve the biological and social characteristics of their populations regardless of race (Primer Congreso Demográfico Interamericano 1944).

The Collective Pursuit of Anti-Discrimination

The very organizational structures and procedures that world powers, including the United States, had created in previous decades gave weaker countries the capacity to collectively push an agenda of anti-discrimination. The League of Nations and associated conferences were important venues through which Latin American countries forged intergovernmental ties and solidarity. All Latin American countries belonged to the League at some point between 1920 and 1946, and nine were charter members (Thomas and Thomas 1963). The Pan-American Union was an organization built by the United States as a means to exert economic and political control over other countries in the region, but Latin American countries resisted these efforts and in the process learned to work cooperatively. The Union became a venue where each country in the organization had an equal vote and the collective action of Latin American countries gave them leverage against the United States. In Union conferences, Latin American countries passed early resolutions linking liberal principles of equality before the law with nondiscrimination based on race or religion (e.g., Eighth International Conference of American States, 1938).

The ILO also provided a forum for the building of Latin American solidarity against powerful countries, and it was as a nexus between the League—of which it was originally a subsidiary—and the Pan-American Union (Kelchner 1930). The ILO acted as a clearing house and publisher of information about labor and social legislation, and analyses of immigration policies in the region and the world. These data served as a basis for domestic policy discussions, but also for the ILO-sponsored meetings during which participants discussed labor issues and migration. The ILO's General Conference decided matters by a simple majority, which gave Latin American countries a bloc voting advantage if they wanted to challenge U.S. positions. Eighteen of 47 ILO member countries were in the Americas. The ILO was the organization to which Latin American countries presented complaints in the 1930s and 1940s about U.S. treatment of Mexicans in the Southwest, Cubans in Florida, and Central Americans and Caribbean islanders in the Panama Canal Zone.

The League, the Pan-American Union and the ILO had become politically significant organizations for Latin American countries by the eve of World War II. The organizations and their rules and procedures allowed these countries to seek cover from the interventionism of the United States, and afforded them a means to act collectively. In the following decades, the solidarity forged among Latin American countries in these organizational networks extended to postwar entities and included newly independent countries and sometimes ideologically strange bedfellows, but the effect was that weaker countries had the capacity to collaboratively pursue policies that the Great Powers opposed.

As Latin American and Asian countries demonstrated the capacity to act collaboratively when World War II ended, the winning powers outlined a vision of the postwar world order without consulting less powerful countries that had supported the war effort. Latin American countries felt especially betrayed by their exclusion from the Dumbarton Oaks conference in which the United States, the United Kingdom, and the Soviet Union met to discuss postwar political organization.¹² Latin American leaders feared a return to unilateral decisionmaking in the region, U.S. interventionism, and discrimination against Latin American immigrants in the United States. Twenty-two countries held an extraordinary meeting at Chapultepec Castle in Mexico City for the Inter-American Conference on Problems of War and Peace (February 1945). They examined the Dumbarton Oaks proposals in detail and presented a unified front that resolved to make “every effort to prevent racial or religious discrimination” (Lauren 2003: 171). These countries also agreed to put human rights and anti-discrimination on the agenda for discussion at the founding conference of the United Nations in San Francisco. Sensing an opportunity to use discrimination as a wedge between the West and countries in Africa, Asia, and Latin America, the Soviet Union surprised other Dumbarton Oaks participants by reversing its previous strong opposition to racial equality and anti-discrimination (Glendon 2003; Morsink 1999; Wright-Carozza 2003).

Latin American, Caribbean, and Asian delegates at the San Francisco conference contributed several drafts of an international bill of rights and anti-discrimination proposals. By the end of the conference in June 1945, delegates had inserted human rights language into the UN Charter in seven places (see UN Charter and Articles 1, 13, 55, 62, 68, and 76 of the initial draft United Nations (1945)). The Charter also provided for the creation of a Commission for the Promotion for Human Rights, which later prepared a draft of the Universal Declaration of Human Rights (Morsink 1999; Sikkink 2017).

Following the San Francisco conference, several Latin American and Asian countries pressed the issue of an international bill of rights, which had been resisted by the major powers. In his memoirs, Canadian John Humphrey—one of the authors of the foundational draft of the Universal Declaration of Human Rights—recognized his debt to a Panamanian and a Chilean draft presented at the San Francisco conference (Humphrey 1984: 32; Morsink 1999). Latin American delegates made substantial contributions to the UN Charter and to the draft of the Universal Declaration of Human Rights, often against the grain of what the Great Powers desired. In collaboration with delegates from Asia and networks of legal experts (e.g., the American Society of International Law), they leveraged moral claims made by powerful liberal democratic countries—that all peoples should unite to fight the threat of tyranny—to extract commitments for human rights and racial nondiscrimination.

The United Nations became the main organizational forum for the governments of most Latin American countries and all of the countries in Africa and Asia plus the Soviet Union to condemn racial discrimination. It was the venue where debates about how South Africa treated Indian immigrants became a wider attack against the entire system of apartheid and racism. The General Assembly discussed the question of Indians in South Africa from 1946 until 1994, when apartheid finally collapsed (Anderson 2003; Lauren 1996, 2003). Brazilian social scientists played a prominent role between 1950 and 1967 in crafting four UNESCO statements on race, which removed any hint of scientific authority from race as a biologically based category (Banton 2002). An incipient anti-racism norm became further consolidated through international conferences that fostered collaborative discussions on the challenges of race and human rights in the postwar era. African and Asian countries that met in Bandung, Indonesia, in 1955 joined several Latin American countries in advancing the human rights and anti-discrimination agenda (Burke 2010), and were another part of the feedback loop from the Global South to the Global North.

The admission into the United Nations of Asian and African countries that had formerly been colonies of some of the major powers transformed the drafting of rights declarations after 1960. A group of African governments concerned with apartheid forced the organization to hear individual petitions that previously had been ignored. The newly formed Special Committee on Decolonization and the Special Committee on Apartheid heard petitioners and launched investigations. Western powers resented what they perceived as the radical behavior of these Committees, but the United Nations gave them considerable leeway. The International Convention on the Elimination of All Forms of Racial Discrimination (ICERD) that passed in 1965 was a major human rights initiative of the Special Committee on Decolonization (Banton 2002).

Through the 1950s, the United States, Canada, Australia, South Africa, and New Zealand resisted the adoption of anti-racist provisions in agreements such as the 1948 Universal Declaration of Human Rights and what became the ICERD. Governments controlled by whites feared that the institutionalization of anti-racism would limit their sovereign right to select immigrants by race. Also, the South African and U.S. governments sought to avoid international scrutiny of their treatment of blacks. Eleanor Roosevelt, who represented the United States on the drafting committee, challenged the anti-racism clause proposed in the 1948 declaration because she was concerned that conservative Southern Democrats who controlled the Senate would never ratify a treaty containing such a provision (Anderson 2003: 131). In the debates about what became the ICERD, Western countries began to see support for individual petition as a way to embarrass African and Asian countries for the gap between what they advocated and their own discriminatory domestic practices (Burke 2010).

The organizational capacity in multilateral venues of less powerful countries was one condition that allowed them to exert leverage on stronger states, but it alone was insufficient to achieve deep change. There had been, after all, other instances in which countries had organizational capacity, but failed to get their way against the opposition of stronger countries. For example, at the drafting of the Covenant of the League of Nations in 1919, an imperial power like Japan was not able to advance a racial equality clause even with support from the majority of participating countries (Shimazu 1998). In the next section, we demonstrate that a second condition, when combined with organizational capacity, made it possible for norms and policies to diffuse against the wishes of powerful countries.

Cooperative Linking of Discrimination to Core Geo-Political Interests of the Powerful

Three geo-political conjunctures created opportunities for weaker countries to advance an agenda of human rights and anti-discrimination despite the opposition of powerful countries: the struggle against fascism before and during World War II, negotiation of a new global order after World War II, and processes of decolonization and the prosecution of the Cold War. At these conjunctures, Latin American, Asian, and African countries played important roles in linking an agenda of non-discrimination to core geo-political goals of more powerful countries. They did so by working together in the multilateral organizations examined in the previous section.

Immediately before and during World War II, Latin Americans linked racist immigration policy to key geo-political interests of the United States. By the time that Latin American countries joined forces against discriminatory immigration policies in 1938 in Resolution XLV of the Eighth Pan-American Conference, the United States had reversed its opposition to a Latin American initiative declaring the principle of non-intervention, pulled its troops out of Nicaragua and Haiti, revoked the interventionist Platt Amendment in Cuba, given up financial control in the Dominican Republic, and proclaimed the Good Neighbor policy in 1933 (Smith 2013). Latin American countries pressed nondiscrimination in immigration policy knowing that the United States wanted Latin American support in case of war in Europe, which seemed likely by the late 1930s.

After the United States entered World War II, it had an even greater need for the support of its Latin American and Asian allies on the battlefront and as sources of raw materials and labor. Large-scale U.S. agriculture in the 1940s relied heavily on Mexican braceros (temporary migrant workers), and congressional calls for the inclusion of Western Hemisphere countries in the discriminatory nationality quota system ended. Mexico promoted the Allied cause in Latin America and tried to persuade countries that resisted the role of the United States in hemispheric unity (FitzGerald and Cook-Martín 2014: 66). Aware of the mistreatment of Mexicans in the United States, Latin American countries used this geo-political opening to push the agenda of nondiscrimination against Latin Americans generally.

During World War II, the United States and some of its powerful allies lacked credibility among non-whites because of discrimination against some of the very people they tried to mobilize for the war effort. Enemies seized on this hypocrisy for propaganda purposes. Many of the Allies had rounded up and interned their citizens of Japanese descent. Allied military forces discriminated against their own citizens of color, including soldiers. The immigration and nationality laws of the United States, Canada, Australia, New Zealand, South Africa, and 15 Latin American countries discriminated against the immigration of people from key members of the “Grand Alliance” like China and India (Lake and Reynolds 2008; see also Figure 1).

The repeal of longstanding Chinese exclusions in the United States, Cuba, and Canada is attributable to this geo-political conjuncture. The Allies could not continue to validate Japanese propaganda about their hypocrisy by maintaining racial exclusions against Chinese nationals and Asians more generally. The 1943 Magnuson Act repealed the Chinese exclusion acts in the United States, established quotas for “persons of the Chinese Race”, and made “Chinese persons or persons of Chinese descent” eligible for naturalization (57 Stat. 600). A number of acts and bills followed that made persons of other previously excluded Asian nationalities eligible for citizenship and immigration (Tichenor 2002). The incremental lifting of anti-Asian restrictions during World War II was a response to geo-political concerns. By the 1960s, the geopolitics of decolonization and of the Cold War would altogether dismantle the “wall” against immigration from Asia.

Like Japan and Germany during World War II, Soviet propagandists exposed the racist policies of the United States on a world stage. From a U.S. perspective, the price of maintaining racist immigrant admissions policies was too high given its national security goals. Following the drafting of the UN Charter calling for racial equality, Soviet media relentlessly called out the United States for its racial discrimination. Secretary of State Dean Acheson wrote to President Truman that “our failure to remove racial barriers provides the Kremlin with unlimited political and propaganda capital for use against us in Japan and the entire Far East” (Chin 1996: 288). Truman moved civil rights reforms to the top of his domestic agenda. His 1947 Presidential Committee on Civil Rights report cited three critical reasons that the status quo needed to change: harm to U.S. foreign relations, morality, and economic efficiency. The report, titled To Secure These Rights, strongly recommended the elimination of racial prerequisites to naturalization and described the quotas for Japanese, Koreans, and other Asians and Pacific Islanders as unfair. Significantly, the Committee's report cited Article 55 of the UN Charter—“universal respect for, and observance of, human rights and fundamental freedoms for all without distinction as to race, sex, language, or religion”—as the warrant for congressional actions to achieve those goals. This is the very clause that Latin American and Asian UN delegates had advanced in 1945 against the U.S. delegation's attempts to quash it.

Rapid decolonization in Asia and Africa increased the number of countries with a vote in the United Nations and, in the escalating Cold War, created opportunities to further advance the agenda of human rights and anti-discrimination. By the early 1960s, these newly independent countries were on the verge of successfully passing the ICERD, and U.S. policies were among the main targets of sustained international condemnation. Elsewhere, we have argued that this geo-political dynamic led to passage of the 1965 immigration act that repealed laws that relied on racial criteria for the selection of immigrants (FitzGerald and Cook-Martín 2019). Thus, processes of decolonization and engagement in the Cold War over time created the political context in which less powerful countries could advance norms of anti-discrimination and human rights that eventually led to policy changes.

The demise of ethnic selection in the Americas was the result of weak states banding together through multilateral institutions to effectively apply diplomatic leverage on stronger states. This is not a story of parallel path development—uncoordinated but similar reaction to the same external cause. Each country did not independently develop a similar solution to the Cold War problem by deciding to stop selecting immigrants by race or national origin. Instead, this process was the result of years of strategic action on the part of governments in Asia, Latin America, and Africa to shift the consensus in powerful countries away from viewing racism as the natural basis of immigration policy to viewing racism as an abhorrent ideology whose expression in immigration policy required reform. In this instance, linking immigration to core geo-political interests reversed the typical direction of policy diffusion, against the more common pattern in which policies are transferred from the strong to the weak (cf. Halliday and Osinsky 2006: 455; Twining 2005).

Discussion and Conclusion

By mapping patterns in the immigration laws of the Americas, we have shown how these laws converged on negative discrimination before the late 1930s and subsequently moved away from discriminatory policies. A cluster of countries in the mid-twentieth century developed policies preferring “assimilable” immigrants, and all Latin American countries except Cuba and Haiti had ethnic preferences in naturalization policies as of 2010. These descriptive findings reveal previously unknown patterns, but to conclusively demonstrate the diffusion of selection criteria in immigration policy, they must be combined with qualitative process tracing.

An analysis of case studies shows how three mechanisms of diffusion operated and the conditions under which they were most consequential. In the period before 1940, strategic adjustment and emulation were the salient mechanisms of diffusion. International scientific organizations and intergovernmental institutions channeled policies across great geographic distances and cultural divides. Most significantly, after 1940, the mechanism of leverage challenges the common assumption that diffusion primarily flows in one direction, from the strong to the weak. Instead we have shown that diffusion can run against geo-political gravity, from global South to North. The conditions under which leverage from below operates include (1) the existence of the organizational means and capacity to advance policies in a collaborative way, and (2) geo-political conditions that allow less powerful countries to link their policy proposals to the fundamental strategic interests of more powerful countries. The end of overt ethnic selectivity in immigration law was not caused by U.S. emulation of weaker countries, but rather by collective leverage of weaker states through international institutions to take advantage of the change in opportunity structures provided by the U.S. Good Neighbor Policy toward Latin America in the 1930s, World War II, and decolonization during the Cold War.

This study demonstrates the utility of combining a comprehensive historical mapping of policies to document instances of clustering and convergence with case studies that specify causal pathways for the patterns observed in the related quantitative analysis. Attention to a time span that is longer than that adopted by typical studies of policy development makes it possible to identify patterns and the conditions under which diffusion happens, the pathways it follows, and the direction in which policy travels. Further research attending to other policy domains using these methodological strategies would help establish the generalizability of these findings for explaining the mechanisms and conditions under which diffusion occurs and its relative causal importance as a source of change.

This article makes two core theoretical contributions to an understanding of how policies become similar across countries. First, it demonstrates that diffusion can take place even in policy arenas that analysts presume are relatively insulated from the influence of other actors. In the arena of immigration and nationality policy, we show that sovereign states shaped each other's policies by means of specific mechanisms. We have not assumed that diffusion is always at work where policies are similar, but rather considered the possibility of parallel development. We also have not assumed that a single factor explains the patterns observed, a caution raised by Dobbin et al. (2007), or that diffusion is necessarily more important for explaining a particular policy than other factors.

Second, this article extends knowledge of the physics of organizational fields by identifying the conditions under which less powerful state actors achieve their aims against the policy agenda of more powerful actors. This advance is only possible when one makes explicit and tests the assumptions implicitly made by theorists about how power works in the organizational field constituted by the international system of states. One assumption has been that leaders in the international system of states play a determinative role in shaping and modeling policies that other countries then adopt. The direction of influence is from more to less powerful. A second assumption has been that countries influence other countries to which they are culturally or geographically close. There is evidence for both of these claims, but at key moments, the physics of international political fields shift so that (1) less powerful actors achieve their aims on meaningful policy points and (2) policy patterns traverse geographical and cultural gaps.

In addition to the theoretical implications discussed above, our arguments and findings are relevant to assessing the prospect of a return to policies that rely on overtly ethnic selection criteria. Our finding that international institutions can exercise leverage from below makes it less likely, although not impossible, that people would explicitly be barred from entry or naturalization because of ethnicity. The institutionalization of an international system of states acts as a strong deterrent against blatantly racialized discrimination. The cost of explicit bans against nationals of sovereign states is high in a world as interconnected as the one in the second decade of the twenty-first century. Yet at the time of writing, the specter of legal bans on immigrants by national origins and religion is being hotly debated in one of the key country cases in our study. In the United States, Donald Trump pledged during his 2016 campaign to ban the entrance of Muslims. His executive order of January 27, 2017 instituted a ban against refugees from Syria and prohibited the entry of nationals from seven primarily Muslim countries. A subsequent order on March 6, 2017 banned the entry of nationals from six primarily Muslim countries.

Do these developments demonstrate that we are mistaken in our assessment of the importance of the deterrent effects of international relations on blatant ethnic exclusions? As we have argued elsewhere, the international system of states is a strong constraint, but it is not absolute (FitzGerald and Cook-Martín 2014). There are conditions under which states could reinstate explicitly ethnic exclusions, mostly related to perceived existential threats: inter-state war or large-scale terrorist attacks blamed on a particular group, pandemics linked to particular countries, or groups that do not have the protection of a particular nation-state. In addition, the reaction to the U.S. executive orders has been swift and forceful in terms of judicial review, diplomatic protests, and public outcry and demonstration. Other countries have reacted strongly against the threat of exclusion, which suggests that the politics of avoiding humiliation on a world stage are still at work.

Finally, the return or continuation of ethnic selection will most likely be expressed in policies that are neutral on their face but which have an intentionally disparate effect on specific categories of people. Ostensibly neutral policies that assess prospects for assimilation, for instance, were first modeled in the Americas in the early 1900s and are now becoming common in Europe (FitzGerald et al. 2018). These assimilability provisions disproportionately affect prospective immigrants from primarily Muslim countries. Facially neutral policies are more difficult to examine but are an important focus for future research. The conditions under which sending countries have leverage over more powerful receiving countries and the extent to which they find multilateral organizational venues for collaboration will shape the success of efforts to contest ethnicized immigration policies.