Historical Immigration Policies: Trends and Lessons

Abstract

In recent years, scholars of migration have created several new immigration policy indexes, but most existing databases have limited temporal scope. They also focus, to a large extent, on the Global North. In this research note, we introduce the Historical Immigration Policy dataset (HIP), which begins to fill these gaps. We first provide an overview of the data and then describe how they offer new insights into immigration policy. We make three empirical observations. (1) On average, democracies are less open to immigration than authoritarian states but grant resident migrants more rights. (2) European states were open to immigration earlier than standard accounts of global migration assume. (3) Historically, openness to immigration and inclusive rights for resident migrants have often been complements, not substitutes.

En los últimos años, los académicos en el campo de la migración han creado varios índices nuevos referentes a políticas de inmigración. Sin embargo, la mayoría de las bases de datos existentes tienen un alcance temporal limitado. Además, los académicos tienden a centrarse, mayoritariamente, en el norte global. En esta nota de investigación, presentamos el conjunto de datos de la Política Histórica en materia de Inmigración (HIP, por sus siglas en inglés), que representa un primer paso para llenar estos vacíos. En primer lugar, proporcionamos una visión general de los datos. A continuación, describimos cómo estos ofrecen nuevas perspectivas sobre la política de inmigración. Realizamos tres observaciones empíricas: (1) Como norma general, las democracias están menos abiertas a la inmigración que los Estados autoritarios, pero otorgan más derechos a los migrantes residentes. (2) Los Estados europeos estaban abiertos a la inmigración antes de lo que suponen los relatos habituales en materia de migración mundial. (3) Históricamente, la apertura a la inmigración y los derechos inclusivos de los migrantes residentes han sido, con frecuencia, complementarios y no sustitutivos entre ellos.

Ces dernières années, les chercheurs qui travaillent sur les migrations ont créé plusieurs nouveaux index de politiques d'immigration, mais la portée temporelle de la plupart des bases de données est limitée. Elles se concentrent aussi en grande partie sur les pays du Nord. Dans cette note de recherche, nous présentons l'ensemble de données Historical Immigration Policy (HIP, ou Politique d'immigration historique), qui commence à combler ces lacunes. D'abord, nous proposons un aperçu des données, puis nous décrivons comment celles-ci offrent de nouveaux renseignements concernant la politique d'immigration. Nous formulons trois observations empiriques. (1) En moyenne, les démocraties sont moins ouvertes à l'immigration que les États autoritaires, mais accordent davantage de droits aux migrants résidents. (2) Les États européens se sont ouverts à l'immigration plus tôt que le supposent les récits habituels sur les migrations mondiales. (3) D'un point de vue historique, l'ouverture à l'immigration et les droits inclusifs des migrants résidents se complètent souvent, au lieu de se substituer.

Introduction

In recent years, migration scholars have created several new immigration policy indexes. But most existing databases have limited temporal scope—11 years on average, according to Solano and Huddleston’s recent review (2021). The scarcity of historical data impedes quantitative analyses of the causes and consequences of immigration policy in the long run. It also limits our understanding of how contemporary immigration policies differ from the past.

A second limitation of existing datasets is that they focus mainly on the Global North, especially Europe (Solano and Huddleston 2021). This precludes many interesting research questions. Since most states in existing datasets are consolidated, wealthy democracies, it is currently difficult to examine the effects of political and economic development on immigration policy with quantitative methods. Focusing on majority-white and European countries also overlooks the increasing importance of the Global South as a migrant destination and how entrenched racial or ethnic relations have shaped the politics of migration. While countries in the Global North have become more ethnically and racially diverse over time, they remain relatively homogeneous compared with many countries in the Global South.

To fill some of these gaps, we introduce the Historical Immigration Policies Database (HIP). HIP covers the following 31 states either from 1789 or from independence until the 2010s: Argentina, Australia, Austria, Belgium, Botswana, Brazil, Canada, Chile, Denmark, Finland, France, Germany, Hong Kong, Ireland, Italy, Japan, Kuwait, the Netherlands, New Zealand, Norway, Saudi Arabia, Singapore, South Africa, South Korea, Spain, Sweden, Switzerland, Taiwan, the United Kingdom, the United States, and Venezuela.

HIP allows scholars to re-examine long-taken-for-granted ideas about the historical evolution of immigration policies, evaluate new arguments with longitudinal data, and study the relationship between immigration policies and slow-moving, rarely-changing variables, both at the domestic and international system levels. The HIP sample varies widely across many variables of interest to political scientists, such as regime type, wealth (including natural resource wealth), and type of economy. We code different dimensions of immigration policy separately—including rules of entry, rights, and enforcement—which allows scholars to examine how policies have co-evolved over long periods. HIP speaks to a broad audience in international relations, and the data can be used to investigate the relationship between immigration policy and broader international-relations topics such as the evolution of North–South relations, patterns of democratization and autocratization, and the rise and spread of far-right ideologies and populism. HIP promises to be especially useful in the fast-growing sub-field of historical IR, with its focus on the evolution of states, state systems, and international ties.

In this paper, we explore how the data can be used to provide new insights on three specific topics. First, we examine the correlation between democratization and immigration policy. We find that many countries in our sample began to limit immigration during the first wave of democratization. The extension of the franchise likely provided a fertile institutional setting for anti-immigrant groups to influence policymaking. Second, we examine the widely held belief that European states had relatively closed immigration policies during the formative period of nation-building in the long nineteenth century (Freeman 1995). We show that this is not correct: immigration policies in Europe were fairly similar to policies in the colonial settler states in the New World during the long nineteenth century, and they were typically more open to immigration in the early twentieth century. Third, we examine whether states trade off openness and rights (Ruhs and Martin 2008; Ruhs 2013). We find a negative correlation between openness and rights from the 1950s onward, but no such correlation exists in earlier periods.

Quantitative Data on Immigration Policies

In recent years, there has been an increase in the number of datasets focusing on immigration, citizenship, and integration; however, most of them are limited in geographic or temporal scope. Most of the 67 indexes that are included in Solano and Huddleston’s review of immigration-policy datasets (2021) focus on Europe, and only a handful cover the period before 1970. Table 1 provides an overview of four indexes with a relatively long timespan and shows how they compare to HIP regarding country coverage, temporal coverage, and policy areas covered.

HIP has several advantages over existing datasets. It covers three key policy areas—entry regulations, immigrant rights, and enforcement—for a greater variety of countries and over a longer period than all existing datasets.¹ No other dataset ranges from the long nineteenth century to the modern era. The only other dataset that goes back to the nineteenth century, Timmer and Williamson (1998), only covers the period from 1860 to 1930, and their index codes immigration policy on a single dimension, which masks significant variation among countries.

There is a trade-off between the breadth of the time period covered and the granularity of the coding scheme. IMPIC (Helbling et al. 2017), a well-cited immigration policy database that covers 33 OECD countries between 1980 and 2010 (soon to be updated with data for 2011–2018), provides fine-grained scoring, including, for example, information about whether there is a fee for obtaining a work permit and the sum that must be paid. It is impossible to get systematic information at that level of detail for historical cases. Another challenge when coding HIP is finding criteria that make sense over long periods. For example, IMPIC uses legal provisions concerning “safe third countries” as an indicator of asylum-policy restrictiveness. While today, this is an indicator that provides important information about a country’s asylum policies, for most of the period covered by HIP, the use of this criterion would be anachronistic since it is a relatively recent immigration-control invention.

The DEMIG POLICY database (De Haas, Natter, and Vezzoli 2015) covers a longer period than IMPIC (Helbling et al. 2017),1945 to the present, and includes longer time series for a few countries. However, like Mayda (2010), DEMIG POLICY codes annual changes in the restrictiveness of policy, not levels. This makes it hard to compare policies across countries. Since HIP codes the level of openness (or restrictiveness) of different dimensions of immigration policy, it allows for comparisons both between and within countries.

HIP also has a relatively broader geographic scope than most existing indices. In addition to all the classic immigration countries and several European countries, it includes countries in Latin America, Africa, the Middle East, and East Asia. That said, the coverage of countries in the Global South is considerably patchier than that of the Global North. To our knowledge, the database with the best coverage of the Global South is the Developing World Refugee and Asylum Policy (DWRAP) dataset, which codes policies towards refugees and asylum seekers (and their families) in 92 countries in the Global South from 1952 (Blair, Grossman, and Weinstein 2022). Due to its focus on the Global South, the overlap with the other indexes is limited; however, DWRAP does not include labor migrants or family migrants (other than the families of refugees and asylum seekers). As we discuss in greater detail below, the geographic and temporal scope of HIP allows scholars to examine questions that are currently impossible to examine using existing datasets.

Concepts and Scope

We began collecting data on historical immigration laws to answer questions about the long-run evolution of immigration policies. Thus, we set out to create indicators of immigration policy that could travel across time and space. We conceptualize immigration policy as a set of policies that define who can and cannot enter a state (entry regulations), the rights enjoyed by resident migrants (immigrant rights), and the enforcement procedures of public authorities (enforcement). Immigration policy is clearly a matter of who can enter the state. But rights matter, too. Ruhs (2013), Fitzgerald, Leblang, and Teets (2014), Leblang and Helms (2022), and others have shown that rights influence the likelihood that immigrants choose one destination state over another. Finally, immigration laws need to be enforced if they are to be effective. Otherwise, it would be possible for a policymaker to pass a restrictive immigration law, to get credit with voters, and then not enforce it to win support from interest groups.

Governments have relied on different definitions of “immigrants,” and their policies have distinguished between high- and low-wage migrants, permanent and temporary migrants, and refugees and asylum seekers, as defined by international and national law. For example, in the United States, only permanent stayers are immigrants; temporary workers belong to the “non-immigrant” category. By contrast, Canada distinguishes between permanent and temporary immigrants. States at times refer to humanitarian entrants as “refugees” and not as “immigrants.”

But all these definitions are modern inventions. Even though there has been temporary migration since the early-modern period (see, for example, Altman 1989 and Lucassen 1994), until the rise of the railroad and the steamship, most people who moved long distances had plans to move permanently because of the costs and dangers of travel. Immigration laws were structured on the assumption that everyone who came was there to stay and everyone was simply an “immigrant” regardless of their reason for leaving their home.

We employ a comprehensive definition of “immigrant,” aligning with the views of the United Nations. It includes individuals who permanently or temporarily change their country of residence, whether motivated by economic, family, or asylum-related considerations, regardless of their documentation status. Our definition excludes frontier workers, tourists, and short-term travelers. In practical terms, however, differentiating between immigrants and travelers can be challenging. While states often attempt to create clear distinctions through legislation, these efforts are frequently inadequate in practice.

Historically, states have primarily aimed to deter poor and unskilled immigrants from entering their territories (Hammar 1985; Borrie 1991; Neuman 1993; Kelley and Trebilcock 1998; Lynch and Simon 2003; Hirota 2020). Although debates over “high-skill” immigration have deep historical roots, notably in countries like Britain and its white colonies, states did not enact special policies for “high-skill” immigrants until the 1990s, at the end of our time period. Instead, they often sought to limit the entry of migrants deemed “low-skill.” We create a variable to capture policy restrictions against “low-skill” migrants. We have not established a separate variable to capture policies favoring “high-skill” migrants since such policies typically emerged quite late. For scholars with a specific interest in high-skill immigration policy, we recommend consulting alternative datasets that address this particular aspect of immigration policy more comprehensively.

Table 2 lists the cases selected for inclusion in HIP. The cases were chosen to have large variation on key variables in political science: we include democracies and autocracies, open and closed economies, states with large and small social-welfare systems, states in the Global North and South, colonial settler states and non-colonial settler states, and so on. Ideally, we would have wanted to code more states, but for some—especially autocracies like Qatar or Brunei—it was extremely difficult to find information on immigration policy. Nonetheless, the variety within the HIP sample allows scholars to examine many different hypotheses that other datasets, with their focus on the Global North (Adamson and Tsourapas 2020), cannot be used to test. Moreover, HIP is the culmination of of our individual data collection as outlined in Peters (2015, 2017), Shin (2017, 2019), and Boräng, Kalm, and Lindvall (2020, 2022), employing a consistent coding scheme established by Peters (2015, 2017). Throughout the data collection phase, the authors maintained communication and prioritized expanding the dataset with additional country-year observations, while preserving the integrity of the original coding scheme.

As discussed in the introduction, HIP’s emphasis on longitudinal historical data is motivated by two concerns: the authors’ research questions and the slow-moving nature of immigration policy. Our study goes back to 1789, a year that marked the beginning of the long nineteenth century. The sample period encompasses the onset of the modern state system, stabilized borders, and the intensified regulation of cross-border population movements.² Toward the end of the eighteenth century, many governments began to expand their control over migration, with notable immigration laws enacted in France (1790), England (1792), and the newly established United States (1790). States not in existence in 1789 are included in the dataset upon their formation, following the Correlates of War Project’s criteria (Correlates of War Project 2017). Former British dominions such as Australia, Canada, New Zealand, South Africa, and Hong Kong were included when British authorities delegated control over immigration to them.

Data Collection and Coding

The data collection relied on several types of primary and secondary sources, including legislative texts, scholarly studies, and descriptions of laws by journalists and non-governmental organizations.³ Further back in time and in non-democracies, finding the text of the law was often difficult, so our analysis, in many cases, relies on extant scholarly literature and other secondary accounts. For more recent periods and in democracies, legislative texts were usually more readily available. For these observations, we therefore relied more on the actual text of the laws. Except when we examined the exact text of the relevant statutes, we endeavored to find multiple sources from different traditions of historical scholarship to minimize bias.

One type of bias that we probably did not eliminate stems from the fact that small changes in immigration policy are less likely to be reported in newspapers or government reports, which also means that they are less likely to appear in secondary sources. We are, therefore, more confident in our ability to capture significant overhauls in immigration policy than in our ability to capture all the minor changes in between—especially further back in time.

It is important to note that our dataset provides information about policies adopted by legislatures, not how those policies were applied in practice by immigration authorities. Concentrating on de jure policies was a necessity since it would have been impossible to develop consistent measures of how policies were implemented de facto over the more than 200 years our dataset covers. Studying de jure policies is also valuable in itself, as it reflects policymakers’ intentions. Even our “enforcement” indicator, designed to assess states’ efforts to implement their admission policies, relies on policy decisions regarding the establishment of immigration authorities and their funding, ultimately capturing “policy on paper.”

We classify immigration policy into three main categories: entry regulations, immigrant rights, and enforcement. Our aim in creating the dataset was to facilitate comparisons across time and regions. To achieve this, we have chosen indicators that allow for meaningful comparisons while avoiding excessive complexity. The indicators were derived through an inductive process as we considered common policies such as nationality-based entry rules, restrictions on the poor and those with criminal records, developmental/physical disabilities, or mental/physical health issues, as well as citizenship policies. Less common policies vis-à-vis resident migrants, such as state funding of churches, were grouped under a single “rights” indicator to streamline the dataset.

We created twelve indicators in total, which are detailed in Table 3 (comprehensive coding criteria are provided in Online Appendix A). These indicators are scored on a scale from 1 (reflecting the most restrictive policies in the category) to 5 (representing the most open policies in the category), with increments as small as 0.05 and as large as 4. For example, in nationality restrictions, a 1 means that only descendants of nationals were allowed in, while a 5 means that there were no exclusions based on nationality. In between, a 4 represents the exclusion of 1 nationality (such as the Chinese Exclusion Act in the US), while a 3 represents the exclusion of several nationalities (such as the prohibitions against all Asian and African immigrants in the 1921 US Quota Act).⁴ For special carve-out categories such as refugees, asylum seekers, and family members, 1 represents no specific policy, and 5 suggests very inclusive policies. The absence of a carve-out often implies restrictions, as was evident when the United States created its first asylum policy in 1875, allowing exceptions for those convicted of “political crimes” to grant asylum seekers the right to enter the country.

The entry regulations category contains eight indicators, reflecting the diversity of policies used in this area. The first indicator focuses on nationality-based regulations, encompassing both prohibitions on specific nationalities (e.g., the Chinese Exclusion Act) and special access based on nationality (e.g., European Union policies). These nationality restrictions are commonly employed for the practical reason that they are easy to enforce. We measure how many nationalities are allowed entry for two reasons. First, many states lack reliable records of annual immigrant arrivals, making it impossible to create a weighted measure based on immigrant numbers. Second, policymakers often anticipate growth in migration from certain countries when they make policy.

The second indicator is skill or wealth restrictions. Skill and wealth restrictions explicitly limit immigration based on skill level and are often implicitly used to restrict immigration based on ethnocentrism and racism. A factor analysis on all the indicators of our dataset reveals that the two variables we have discussed so far—entry requirements based on nationality and skills (or wealth)—are highly correlated and have the highest factor loadings for the first dimension in the data, which captures the variation in admission policies across countries. Figure 1 shows how those policies have evolved over time.

The longitudinal data in HIP reveal important patterns. The figure shows that policies became gradually stricter in the 1850s, which is earlier than suggested by some scholars (e.g., Hatton and Williamson 1998 and Timmer and Williamson 1998). After World War I, governments introduced many new restrictions: between the 1910s and 1940s, the mean policy score declined from about 4.5 to approximately 3.5 on our 1–5 scales. This restrictive phase was followed by slightly less stringent policies in the 1950s and 1960s, during the Golden Age of Capitalism and the era of the Gastarbeiter (guest workers). Around 1970, with unemployment increasing among the wealthy democracies, governments reverted to stricter policies. While these restrictions persisted among affluent European democracies, many autocracies became wealthier and opened up to immigration, which is why the line in Figure 1 remains relatively flat.

An important lesson from this figure is that immigration policy exhibits significant long-term stability. By examining the entire period covered by HIP, we find that in democracies, the post-1980 era has been marked by relative policy stability compared with earlier periods of substantial change, such as the interwar period or the immediate post-WWII era (see Figure 4 below as well). A narrow time frame could lead to misconceptions about the extent of change during that period.

The third indicator is quotas. This measure is based on what percentage of the population (total quota divided by population) the country is willing to allow each year. Quotas are easy to code because they are numeric, leading to fewer judgment calls, but they have rarely been used outside the United States. The fourth indicator is recruitment policies. States have, at times, gone out to recruit immigrants to increase the number of immigrants coming to their country. In this case, the policymaker has decided that the “natural rate” of immigration is too low and wants to increase it.

The fifth indicator is work prohibitions.One way to regulate immigration is to deter people from migrating by prohibiting them from working in specific industries or limiting the number of immigrant workers employers can use. In the interwar period, work prohibitions were used throughout Europe and in the colonial settler countries to reduce unemployment pressures, or at least the political pressure from unemployment. We could also think of these policies in the rights category, but because the ability to work plays such a significant role in an individual’s decision to migrate, we choose to include it in the entry category. Next, we have family reunification policy. Family migration constitutes a large share of immigration in many countries. These policies determine if family members get special treatment—and, if so, which categories of family members do—or if they have to enter under the same procedures as others.

Finally, we include refugee and asylum policies. Refugee policy refers to policies for resettling individuals already determined to be legal refugees. Asylum policy refers to policies that adjudicate whether someone who is in the country or has presented themselves at the border is entitled to refugee status. Figure 2 describes the policy trends for asylum seekers and refugees over time. When studying these trends, it is important to consider that refugee and asylum policies are only relevant when the overall policy stance—as captured by our other indicators—is restrictive. States admitted those we would now call asylum seekers in the nineteenth and early twentieth centuries, but they typically did not enter as asylum seekers—policies were generally quite open, so there was less need for special provisions for asylum seekers in particular.

We divide rights into two different indicators: rights to citizenship and other rights. Citizenship is a foundational right since it allows individuals to make claims on a state, grants immigrants all rights given to native-born citizens, and protects them from deportation (in most cases). There are two main aspects of citizenship: who is granted citizenship at birth—including via which parent—and how hard it is to naturalize. Our citizenship measure incorporates both. Citizenship can also provide a privileged pathway to move to another state. For example, after World War II, West Germany gave citizenship to all people with German ancestry living in the Eastern Bloc countries, allowing for easy entrance to West Germany. The German government later curtailed this practice, making immigration for these groups much more difficult.

The variable “other rights” captures all other rights that immigrants are granted, which are defined in relation to citizens’ rights. This coding was chosen for methodological reasons. States have generally given citizens more and more rights over the last two hundred years, so a nonrelational measure would lead us to code immigrants as having more rights over time and automatically code all democracies as granting more rights than autocracies. Thus, we use a relational measure.

Immigrants tend to be granted fewer rights than citizens. A 1 on this score captures cases where immigrants have almost no rights. For example, the Dutch and Eastern European “immigrants” to Nazi Germany were slave laborers, and Germany scored a 1 during the Nazi period. Countries such as Kuwait and Saudi Arabia, which provide few rights for migrants, score 2–3. Most democracies score about 4–5 today. In the nineteenth century, many states offered immigrants almost all rights, except the right to vote or run for office in national elections, scoring a 5. Again, one must keep in mind that the rights variable is relative: the high score meant that immigrants’ rights were similar to citizens’ rights, but in most countries, citizens did not have as many rights back then as they do now.

All other rights, such as access to the social welfare system, the ability to own land or a business, and the ability to vote in local elections, are included in this category. The category could have been broken down further, but for those primarily interested in immigration policy—rather than immigrant rights—this categorization should suffice. In practice, most of the legislative changes in the modern era that we capture with our measure of other rights concern social-welfare rights.

A factor analysis suggests that the citizenship and general rights indicators are highly correlated with the second dimension in the data. The evolution of citizenship and rights policies is described in Figure 3. We find that governments were quite generous in their citizenship policy and in the rights given to migrants in the nineteenth century, as many states were trying to attract immigrants to their country. As policymakers increasingly wanted to deter migrants in the Interwar Period, they curtailed migrants’ rights and access to citizenship. But access to citizenship and rights for migrants have gradually increased since their nadir during World War II.

Finally, we coded two indicators of enforcement. We code policy rather than implemented enforcement for three reasons: first, we were interested in policymakers’ intentions; second, all our other indicators concern policy, not implementation, and we want to compare like actions; and third, it is hard to find cross-national data on implementation.⁵ We split enforcement into (1) deportation, which allows states to control who can remain in the country, and (2) “other enforcement,” which primarily means border enforcement, including externalization and internalization of the border (Menjívar 2014; FitzGerald 2019)—but it also includes employer sanctions and amnesties.

The data show that the types of policies a state uses tend to be path-dependent. States rarely overhaul their immigration systems completely; instead, they tend to use one type of policy and make changes to it. For example, the United States relied on a quota system to enact its system of racial preferences in immigration in 1921, while Australia first enacted a skills test based on literacy to enact the White Australia Policy. Both countries have since sought to do away with their racist policies, but they continued to rely on the same kinds of tools, with a new quota system in the United States and a more complicated skills-based policy in Australia. Understanding the sources of this path dependency—whether it is the product of bureaucratic incentives, interest group activity, or some other factor—is an important topic for future research.

We end this section by discussing three challenges we faced when we compiled the data. First of all, as we mentioned in the previous section, governments in the past did not categorize migrants as we do today. In particular, policies for refugees, asylum seekers, and family reunification often only emerged in the twentieth century, with some governments yet to implement such policies. However, these groups often entered countries using existing channels. For instance, the “48-ers,” persecuted German liberals after the 1848 revolution, could be considered refugees under the 1951 Convention, but they entered the United States alongside other immigrants under the same general immigration policy (Dippel and Heblich 2021).

The challenge arises when a country lacks specific policies for certain immigrant groups, but those groups can easily enter through other means. Returning to our example of the “48-ers,” should the absence of an asylum policy in United States in 1848 be coded as 1 (no policy) or 5 (easy entry for asylum seekers)? For family reunification, refugees, and asylum policies, we applied a “bright-line” rule: if no policy existed, we coded it as 1. However, we also introduced a “(policy) exists” variable, taking the value 0 when there is no policy and 1 when a policy is first implemented.

In the primary coding, these indicators were set to 1 until a policy was introduced. During the nineteenth century, when immigration regulations were limited, refugees, asylum seekers, and family members often migrated as regular immigrants. As an alternative coding approach, we marked these policies as 5 for years without a policy and 1 when nationality or skill coding dropped below 4. Typically, at this point, laws began excluding refugees, asylum seekers, or family-reunification immigrants unless specific policies were enacted. The “(policy) exists” variable offers flexibility for scholars to adapt these variables based on their research questions.

Second, in most states, laws are interpreted by the executive branch when they are implemented. For example, in Canada, the law allows the government to set the immigration target each year after a period of consultation. We do not code such administrative law changes but instead focus on the policy passed by the legislature. Administrative law changes are both challenging to find information about and are numerous, making their collection difficult. There is one deviation from this rule: Saudi Arabia. This is our only case of an autocracy in which the autocrat is both the legislative and executive branches. In all other cases of autocracies, there is at least a nominal legislature that passes laws. Our choice to code the laws and not how the authorities implemented them reflects our general emphasis on de jure instead of de facto policies.

Third, and finally, we needed to decide how to handle federal states. All federal states are coded according to the policy of the most open member until the federal government takes sole responsibility for immigration policy. At this point, the national policy is coded. Most federal states allow the free movement of people among the members of the federation; therefore, an immigrant who can come to one of the members can have access to all.

Among member states of the European Union, the immigration policies of the most liberal countries are not coded because the freedom-of-movement policies of the European states do not extend to third-country nationals. When European Union policy does affect some or all of the nations—either because they implement the regulations with national legislation, such as the Dublin Convention on Asylum Seekers, or have delegated control to the EU, as in Schengen—the EU policy is coded as the policy for each nation it affects.

Insights From Historical Data

In this section, we highlight the advantages of the historical data contained in HIP, challenge conventional wisdom on immigration policies, and conduct preliminary tests of some key hypotheses from the scholarly literature. It is important to note that our results here are preliminary, and we have not comprehensively addressed all relevant aspects of these important questions. Additionally, we note that we have not applied weighting to country-level data based on population or migration flows for the analyses in this section. While practical challenges and data limitations play a role, our focus is on mapping the policies adopted by each country, treating each state as a unique immigration policy experiment. Weighting would be more relevant if our aim were to characterize the policy environment faced by the average migrant in each historical period.

Observation 1: On Average, Democracies Are Less Open Than Autocracies, But Resident Migrants Have More Rights

There is a growing literature on the relationship between political regimes and immigration policies (e.g., Breunig, Cao, and Luedtke 2012; Joo 2022; Mirilovic 2010), and HIP can help to shed light on this important problem. Figure 4 and Figure 5 illustrate the average policies on entry and rights separately for democracies and authoritarian states. As Figure 4 shows, democracies and authoritarian states have tracked each other fairly closely on entry policies (with the exception of the Second World War, when the remaining democracies in the world closed their borders while authoritarian states in Latin America and elsewhere did not change their policies much), until the late twentieth century. At this point, authoritarian states adopted significantly more liberal policies, while democracies maintained the restrictive policies they had implemented toward the end of the post-war boom.

When we examine the rights of resident migrants, however, democracies have consistently gone further than authoritarian states. Interestingly, Figure 5 suggests that the gap between democracies and authoritarian states opened up during the first wave of democratization. It has kept growing in modern times, as the tendency in the democratic states ever since the Second World War has been to grant resident migrants more rights, reducing the differences between the legal status of citizens and resident migrants.

Observation 2: Many European States Were Open to Immigration Earlier Than Previously Thought

Within immigration historiography, there has been an idea that states have identities based on their history with immigration. For example, Freeman (1995) famously categorized states into English-speaking settler societies, European states with postcolonial and Gastarbeiter migration, and new countries of immigration. He argued that because settler states were built by immigrants, the immigrant experience is part of their national identity, leading them to be more open to immigration. In contrast, for the European states, “their modern experience of mass immigration occurred when they were already fully developed national states” (Freeman 1995, 889). Similarly, Zolberg (1989) argues that homogeneous states (those with one ethnicity, as in some European countries) will be wary of accepting immigrants who will change the political culture. In contrast, heterogeneous states (such as the settler countries) have a longer tradition of integrating minorities without too much disruption.

We can examine this argument using HIP. In Figure 6, we compare entry requirements based on nationality and skills in the European countries in our sample on the one hand and the English-speaking colonial settler countries Australia, Canada, New Zealand, and the United States on the other hand. The database reflects a great deal of synchronous variation among countries, especially in the twentieth century. Until the late nineteenth century, there were few entry requirements in either region, with the European states being slightly more restrictive. This all began to change around 1880 (e.g., the Chinese Exclusion Act in the US in 1882), when entry requirements were put in place in several English-speaking settler colonies. Since then, there have been interesting and significant differences between these regions of the world. In the interwar period, admission policies became more restrictive in both groups of countries. Interestingly, the European countries overtook the English-speaking countries in North America and Oceania until the post-war boom ended around 1970; since then, the European states have been significantly more restrictive.

The data suggest that the relationship between national identity and immigration policy is much more complex than previously thought. Both the English-speaking offshoots and European states had very open immigration policies for much of the long nineteenth century. Europe remained open to immigration far longer into the long nineteenth century than the English-speaking offshoots. A lot of the literature on international migration seems to be based on the assumption that immigration was simply not an issue in Europe because the continent did not experience large-scale inward migration from other parts of the world. But intra-European migration was an issue. During this period, we saw large movements of Irish and others into Great Britain; Belgians, Swiss, Italians, Poles, and North Africans into France; and Dutch and Poles into Germany (Lucassen and Lucassen 2009). It is therefore important to examine immigration policies in European countries, as we do here. With the data HIP provides, we can revisit our assumptions about migration history.

Observation 3: Historically, Openness to Immigration and Inclusive Rights for Resident Migrants Were Often Seen as Complements, Not Substitutes

Ruhs and Martin (2008) and Ruhs (2013) argue that there is a trade-off between the number of immigrants that a country allows in and the rights it gives.⁶ Wealthy democracies typically offer many rights but greatly restrict immigration. Wealthy autocracies, on the other hand, allow a large amount of immigration but significantly limit rights. Yet, because all types of countries are desperate for high-skill migrants, they offer this group substantially more rights and allow in proportionately greater numbers.

Figure 7 visualizes the correlations between the 12 dimensions of immigration policy across all country-year observations. We first turn to the two rights variables: Citizenship and Other Rights. Only Citizenship, Work Prohibitions, and Deportation exhibit strong correlations to Other Rights, which is the only variable that is highly correlated with Citizenship. However, Family Reunification, another important immigrant right, shows strong negative correlations with some of the entry dimensions, most notably Quotas, Other Enforcement, and Skill. Based on the correlation matrix, evidence for the numbers vs. rights trade-off remains mixed.

But the figures we examined earlier help shed more light on these relationships. Figure 3 delves into the temporal variation of the average of two indicators: rights and citizenship. High values mean that migrants enjoyed most of the rights that citizens enjoyed and that there was a clear path to citizenship for migrants; low values indicate that immigrants had few rights and that it was difficult to qualify for citizenship. The broad trends in Figure 3 are a little different than the trends described in Figure 1: there was a gradual decline in the rights enjoyed by resident migrants during the nineteenth century and the first part of the twentieth century. Countries became more inclusive from the 1940s to the 2000s. In other words, if one compares the world of 2010 to the world of 1950, there are more restrictions on the mobility of labor migrants today, but the migrants who were nevertheless admitted enjoyed more rights.

Yet, if we go back further in time to the long nineteenth century, states typically had both open entry policies and gave substantial rights to migrants. During this period, states wanted to attract immigrants and often competed with other states for immigrants. To be competitive, states offered more extensive rights. We see this in our data: the correlation between rights and entry is positive throughout the period at 0.84. Yet, this overall correlation masks changes in the relationship over time. Before the 1950s, the relationship was close to 1 (0.9) but beginning in the 1950s, the relationship turned negative.⁷ Thus, our understanding of how different policies relate to each other depends on the period we examine.

Using the Data

As the previous section showed, the HIP data can be used in a variety of ways. For example, researchers interested in the evolution of asylum policies can utilize the asylum measure as a dependent variable to track changes over time and across different countries.⁸ For insights into the entry dynamics for asylum seekers specifically, they should combine the asylum measure with the asylum exists indicator.⁹

Scholars examining the variation among multiple measures might explore their covariance—as seen in the interplay between nationality and skill restrictions—analyzing their concurrent movements and divergences. For analyses within a specific category such as entry restrictions, rights, or enforcement, as referenced in Boräng, Kalm, and Lindvall (2022), a simple average of the measures should be used. There is also the potential to redefine or expand categories—for instance, by integrating work prohibitions and family reunification policies into a broader rights category, despite their traditional classification as entry criteria.

For comprehensive studies, scholars may construct an index from all measures, employed either as a dependent variable, as in Peters (2015, 2017), Peters and Shin (2023), and Shin (2017, 2019  , 2023), or as an explanatory variable, exemplified by Kalm and Lindvall (2019). We recommend principal component analysis for creating such indices, as detailed in the appendix.

Conclusion

While there has been much scholarship on immigration policy across the last two centuries, we have lacked comparable quantitative data that can help us, as a field, to understand the main shifts over time. HIP provides those data. It allows scholars to test their theories with data on the longue duree of immigration policy in countries in both the Global North and the Global South. It also helps us better understand the evolution of different policies that have been used to control people’s movement. In our work, we have examined the interaction of different policies with immigration, including trade, foreign direct investment, natural resources, and social welfare, Nonetheless, there are many other areas to explore.

Author Biography

Margaret E. Peters is a Professor of Political Science at UCLA. She works on immigration policy, focusing on business support for immigration and globalization, emigration and authoritarianism, and the politics of refugee and asylum seeker resettlement.

Frida Boräng is an Associate Professor of Political Science at the University of Gothenburg. She works on immigration policy from a comparative and historical perspective, and migrants’ political integration.

Sara Kalm is an Associate Professor of Political Science at Lund University. Her research focuses on historical and current transformations of immigration and citizenship policy and contemporary authoritarian political movements.

Johannes Lindvall is the August Röhss Professor of Political Science at the University of Gothenburg. He works on comparative politics, especially the evolution of political institutions and public policies in the long run.

Adrian J. Shin is an Assistant Professor of Political Science at the University of Colorado Boulder. He specializes in the political economy of international migration, inequality, and international organizations.

Notes

Authors’ note: We thank David Leblang, Sara Wallace Goodman, and the participants of the DEMSCORE Assembling the Wheel of Comparative Migration Policy Research workshop for their helpful feedback. We thank Klara Eitrem Holmgren, Johan Ekstedt, Moa Olin, Vinnie Intersimone, Minh Dan Vuong, Karine Hoffman, Roxana Moussavian, Lucia Hennelly, Pauline Hilmy, Meredith Garry, and Jessica Soley, for their research assistance. Funding came from the National Science Foundation Graduate Student Research Fellowship (DGE 0718128), NSF Doctoral Dissertation Research Improvement Grants (SES-1559661), Riksbankens Jubileumsfond (programme grant M14–0087:1), and the Swedish Research Council (grant 2017-01644). All replication files can be found at https://dataverse-harvard-edu.libproxy.ucl.ac.uk/dataverse/isq and https://dataverse-harvard-edu.libproxy.ucl.ac.uk/dataset.xhtml?persistentId=doi:10.7910/DVN/F7V8YL. All questions regarding the data and replication should be sent to the authors.

Footnotes

References